Carbon 44 (2006) 3102–3112 www.elsevier.com/locate/carbon
The influence of various factors on aqueous ozone decomposition by granular activated carbons and the development of a mechanistic approach P.M. Alva´rez a
a,*
, J.F. Garcı´a-Araya a, F.J. Beltra´n a, I. Gira´ldez a, J. Jaramillo b, V. Go´mez-Serrano b
Departamento de Ingenierı´a Quı´mica y Energe´tica, Universidad de Extremadura, Badajoz 06071, Spain b Departamento de Quı´mica Inorga´nica, Universidad de Extremadura, Badajoz 06071, Spain Received 22 December 2005; accepted 10 March 2006 Available online 5 May 2006
Abstract The decomposition of aqueous ozone in the presence of various granular activated carbons (GAC) was studied. The variables investigated were GAC dose, presence of tert-butyl alcohol (TBA), aqueous pH as well as textural and chemistry surface properties of GAC. All the GAC tested enhanced the rate of ozone decomposition to some extent. From the analysis of experimental results it was deduced that ozone transformation into HO radicals mainly occurred in the liquid bulk through a radical chain reaction initiated by OH and HO 2 ions. Hydroperoxide ions arise from the formation of H2O2 on surface active sites of GAC and its further dissociation. No direct relationship between textural properties of GAC and the rate of ozone decomposition was found. However, a multiple regression analysis of data revealed that basic and hydroxyl surface oxygen groups (SOG) of GAC favor the kinetics of the ozone decomposition process. It is thought that these groups are the active sites for ozone transformation into H2O2. Repeated used of GAC in ozonation experiments resulted in loss of basic and hydroxyl SOG with formation of carboxyl, carbonyl and lactone-type groups. Then, pre-ozonation of GAC reduces its ability to enhance the aqueous ozone transformation into hydroxyl radicals. 2006 Elsevier Ltd. All rights reserved. Keywords: Activated carbon; Oxidation; Reaction kinetics; Surface oxygen complexes
1. Introduction Catalytic ozonation is a promising technology for the effective removal of water and wastewater contaminants that are refractory to conventional oxidation treatments. The main advantages of the catalytic processes with respect to traditional non-catalytic ozonation are better ozone utilization, increased contaminant removal efficiency and greater degree of organic matter mineralization [1]. Heterogeneous catalytic ozonation, though less studied than the homogeneous processes, is more attractive since oxidation efficiency and selectivity can be improved by proper selec*
Corresponding author. Tel./fax: +34 924 289385. E-mail address:
[email protected] (P.M. Alva´rez).
0008-6223/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2006.03.016
tion or modification of the catalysts. Moreover, when a homogeneous catalysts is used (i.e., metal ion) an additional separation process is required to remove the toxic metal ions from the treated water. The use of a solid as catalyst overcomes this problem as it can be easily recovered from the aqueous medium. In addition to metal based heterogeneous catalysts (metal oxides and metals or metal oxides on supports) the use of different types of activated carbons to promote ozone reactions in water has arisen a great deal of interest in the last few years. Thus, some dissolved organic water pollutants have been effectively transformed into final products (CO2 and H2O) through ozone reactions enhanced by powered and granular activated carbons e.g., [2–8]. The effect of the presence of activated carbon
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
in the ozonation process is reported to be mainly the acceleration of the decomposition of dissolved ozone into free hydroxyl radicals (HO), which are highly strong oxidant species able to react rapidly with most of the water pollutants in such an advanced oxidation process (AOP) first named as Carbozone [9]. The overall effectiveness of this type of AOP will depend on both the activated carbon properties and the water constituents. In a recent work on this subject, Sa´nchez-Polo et al. [10] have addressed that the efficiency of a granular activated carbon (GAC) to transform ozone into HO radicals is mainly related to the surface area, pore volume and basicity of the GAC. They also found that the natural organic matter of water from the Lake Zurich (Switzerland) consumed HO but did not negatively affect to the yield of ozone transformation into these free radical species. Although there is good agreement in that activated carbon enhances the aqueous ozone decomposition, its role as a catalyst, initiator or promoter of the radical-type chain reaction is still not well understood. On one hand, some authors refer to the process as a catalytic system. Thus, for example, Guiza et al. [11] reported that the rate of ozone decomposition in an aqueous synthetic solution containing given concentrations of sodium acetate (inhibitor of ozone decomposition by scavenging HO) and methanol (promoter of ozone decomposition by transforming HO into O2 H= O 2 that in turn react with molecular ozone) was two folded by adding just 15 mg/L of a powdered activated carbon prepared from olive stones. Moreover, these authors did not observe any loss of activity upon pre-ozonation of the activated carbon so they attributed a catalytic role to the activated carbon in the ozone decomposition reaction. On the other hand, other researchers e.g., [10] have observed changes in both textural and chemical surface properties of GAC upon exposition to high ozone dosages. This suggests that activated carbon cannot really act as a catalyst for ozone reactions but as initiator or promoter. To clarify whether or not activated carbon is a catalyst for aqueous ozone decomposition into HO radicals the understanding of the exact reaction mechanism would be helpful. Also, this would provide the basis for a sound engineering development of water treatment processes based on ozone reactions in the presence of activated carbons. However, though different mechanism approaches have been proposed up to date, this point has not yet been clearly elucidated. Beltra´n et al. [12] showed that the aqueous ozone decomposition in the presence of a GAC was greatly affected by the solution pH. Thus, we suggested that at pH from 2 to 6 the reaction of the ozone on the carbon surface is not of radical-type, being molecular oxygen the one by-product while at pH > 6 the generation of HO radicals comes in addition from the direct reaction between molecular ozone and OH-containing active sites of the activated carbon. Sa´nchez-Polo and Rivera-Utrilla [13] proposed that the sites of activated carbon for ozone decomposition are mainly graphene layers with unpaired p electrons and functional groups of basic nature (e.g., pyrone, and chro-
3103
mene structures) that would act as Lewis bases that reduce ozone into OH and H2O2, respectively. These species initiate the radical-type chain reaction in the aqueous medium through the well known mechanism of Staehelin and Hoigne´ [14]. However, they proposed the formation of H2O2 in hydrochloric acid solution, conditions at which its dissociation to the initiator/promoter HO 2 is not favored. In another contribution of these authors [10] they reported that pyrrolic groups of urea-functionalized GAC react with ozone to directly yield hydroperoxide radicals (O2H) which also can enhance the rate of ozone transformation into HO radicals [14]. In this study the behavior of various GAC of different nature in decomposing aqueous ozone is examined. The influence of the main variables affecting the rate of ozone depletion (i.e, GAC dose, presence of the HO scavenger tert-butyl alcohol (TBA), pH as well as textural and chemical properties of the GAC) are evaluated, thus providing insights into the reaction mechanism. 2. Experimental 2.1. Activated carbons A lab-prepared GAC produced from cherry stones by the two-stage thermal method using carbon dioxide as activating agent was primarily used in this work. This GAC is referred to as CS30, being 30% the burn-off percentage achieved during the activation. Other GAC produced in a similar way were CS10 and CS20 with 10% and 20% burn-off, respectively. Details of the preparation method have been reported previously [15]. In addition to CS carbons, four commercial GAC manufactured from specific grades of bituminous coal, namely Hydraffin P110 (Donau Carbon GmbH & Co, KG, Germany), Darco 20–40 mesh (Norit Americas Inc, USA), Chemviron SS4P and AQ40 (Chemviron Carbon, Belgium) were used in this work. The lab-prepared and the as-received commercial GAC were sieved (the particles with size in the range 1.0– 1.6 mm being chosen), thoroughly washed with boiling Milli-Q water, dried at 110 C for 12 h and stored in a desiccator until use in ozone decomposition experiments. Table 1 shows textural and chemical properties of the GAC as analyzed in this work. Textural characterization of GAC was mainly accomplished by adsorption of nitrogen at 77 K using a Quantachrome Autosorb-1 automated gas adsorption system. From the nitrogen isotherm the BET surface area (SBET), external surface area (Sext) and micropore volume (V1) were determined using the BET and t-plot methods, respectively. The volume of pores with diameter larger than 3.5 nm (V2) was determined by mercury porosimetry with a Quantachrome Austoscan-60 apparatus. The surface oxygen groups (SOG) content of GAC was analyzed by the Boehm’s titration method following the recipe given by Vidic et al. [16]. Accordingly, alkaline solutions of different base strength (NaHCO3 0.05 N, Na2CO3
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
3104
Table 1 Textural and chemical properties of the GAC used in this work GAC
CS10 CS20 CS30 Hydraffin P110 Chemviron SS4P Chemviron AQ40 Darco 20 · 40
SBET (m2 g1)
368 574 658 972 1193 863 703
Sext (m2 g1)
137 107 163 208 335 80 337
V1 (cm3 g1)
0.127 0.252 0.288 0.395 0.435 0.374 0.162
V2 (cm3 g1)
0.589 0.555 0.608 0.328 0.330 0.535 0.784
Ash %
0.33 0.35 0.42 6.30 0.58 6.92 16.25
0.05 N, NaOH 0.25 N and NaOH 0.05 N) were used for the individual analysis of the various acidic groups and HCl 0.05 N for the determination basic groups as a whole. The ash content of GAC was determined after subjecting samples to combustion with air at 850 C. 2.2. Ozone decomposition experiments Ozone decomposition experiments were carried out at 20 C in a thermo-regulated batch glass reactor (volume 500 mL). Ozone was produced from a 20 L/h dry oxygen flow in a laboratory ozone generator (Sander model 301.7). The outlet gas of the ozone generator, containing about 60 mg O3/L as measured by an Anseros Ozomat GM19 analyzer, was bubbled through a porous plate into the reactor previously loaded with 300 mL of 3–10 mM phosphate-buffered organic-free water (Milli-Q system). Gaseous ozone was fed to the reactor for 2 h to saturate the water with ozone and, at the same time, remove any trace of organic matter that could further affect the ozone stability significantly. The dissolved ozone concentration after this stage was about 7–8 mg/L. Then, the gas flow was stopped and the reactor was loaded with a given mass of GAC. In some cases, the required amount of TBA was also added to the water to achieve the desired concentration of the HO scavenger. Mechanical agitation (i.e., typically 200 r.p.m.) was provided so that the vessel operated as a slurry reactor with the carbon particles suspended in the aqueous medium. At regular intervals of time, aqueous samples were withdrawn from the reactor through a dispenser and immediately analyzed for the dissolved ozone concentration following the indigo method [17]. Also, hydrogen peroxide in solution was sometimes analyzed. For that, immediately after collecting, samples were placed in vials containing indigo solution to quench the residual ozone. Then, hydrogen peroxide concentration was determined by the peroxidase-DPD method [18]. In addition to experiments in the presence of GAC, ozone alone and O3/H2O2 runs were conducted for comparative purposes. 2.3. TBA adsorption experiments Equilibrium adsorption isotherms at 20 C of TBA on GAC were determined using the bottle point method.
Acidic SOG (lmole g1)
Basic SOG (lmole g1)
Carboxyl
Lactone
Hydroxyl
Carbonyl
Total
Total
12 16 10 0 828 17 139
15 6 16 26 93 9 61
58 74 77 26 262 131 166
39 42 53 87 388 70 235
124 138 156 139 1571 227 601
347 385 420 648 0 392 290
Phosphate buffered (pH 7) aqueous solutions containing TBA were the adsorption solution media while CS30 and AQ40 were the adsorbents. About 100 mg of a GAC and 25 ml of the aqueous solutions with different TBA concentrations in the 1–10 mM range were placed in sealed bottles and left in a thermo-regulated shaking bath for 15 days to reach equilibrium. After this time, GAC particles were filtered out and the concentration of residual dissolved TBA was indirectly determined by measuring the total organic carbon (TOC) with a Dorhman DC-190 analyzer. To evaluate the kinetics of the adsorption of TBA on CS30 and AQ40 batch experiments were conducted in the same contactor as ozone decomposition experiments. About 1 g of GAC was contacted with 300 mL of pH 7 buffered, 5 mM TBA aqueous solution. Temperature was kept at 20 C and 200 rpm stirring was provided. At given time intervals, liquid samples were withdrawn and analyzed for TOC, thus following TBA concentration. After 240 h of processing, the GAC sample was filtered out, dried under vacuum and labeled as CS30* or AQ40*. They were kept in a desiccator to be further used in ozone decomposition experiments. The amounts of TBA on GAC were estimated through a mass balance to be 13.0 and 13.5 mg/g on CS30* and AQ40*, respectively. 3. Results and discussion 3.1. Application of the first order kinetic model To quantitatively evaluate the effect of some major variables affecting the ozone decomposition rate, the first order kinetic approach was considered. Accordingly, the observed rate coefficient of ozone decay was obtained from the slope of the semi-log plot of the residual aqueous ozone concentration against the reaction time. As it is apparent from Fig. 1, where some examples are presented, data from the ‘‘ozone alone’’ experiment fit well the first order kinetic model over the entire course of the reaction (i.e., 30 min). On the contrary, the plots corresponding to the O3/GAC systems show an initial stage of rapid disappearance of ozone followed by a slower kinetic stage. This behavior is reported to be also observed for ozone decomposition in natural waters [19] and in pure water in the presence of soil
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
3.2. Effect of the GAC dose
1 0.9 0.8 0.7 0.6 0.5 0.4
CO3 /CO30
3105
0.3
0.2
0.1 0
5
10
15
20
25
30
Time, min Fig. 1. First order plots for aqueous ozone decay. Conditions of experiments: pH = 7; T = 20 C; CO30 = 7–8 mg/L; agitation speed: 200 rpm; GAC dose (if applied): 0.5 g/L. Symbols: j ozone alone; d O3/CS30; m O3/Hydraffin P110.
First, a series of ozone decomposition experiments was carried out at constant temperature and pH but varying GAC dose in the range of 0–5 g/L. In Fig. 2 the ozone decomposition rate coefficient, k O3 , is plotted against the doses of CS30. As seen, the ozone decay rate was enhanced by using increasing doses of this GAC. The first part of the plot (i.e., low GAC doses) is quite linear but for CS30 doses higher than 1 g/L the k O3 /GAC dose ratio substantially decreases. In an experimental approach similar to ours [10], Sa´nchez-Polo et al. using another GAC (Filtrasorb 400, Calgon Corp., USA) also observed an optimal GAC dose range leading to the highest specific ozone transformation rate. Therefore, from a practical point of view the GAC dose is an important factor to be considered in designing systems aimed to achieve high performance in ozone decomposition. All further experiments of this work were carried out at a GAC dose of 0.5 g/L since for all the GAC used this dose was shown to be within the optimal range. 3.3. Effect of the presence of TBA In the presence of GAC, aqueous ozone is transferred to the GAC surface while simultaneously decomposes in the liquid bulk through a free radical-type chain reaction [12]. The ozone on GAC, in turns, may remain physically adsorbed or react with surface functionalities (mainly surface oxygen groups, SOG) and inorganic constituents of GAC. The surface reaction is likely to be mainly of radical-type as well [20,21]. To asses the importance of free radical reactions in the kinetics of ozone decomposition, inhibitors can be used. The choice of TBA as free radical scavenger in ozonation processes is the most usual approach as it does not react with ozone (k 103 M1 s1) but it 0.30
0.25
k O3, min-1
slurry [20]. In these cases the first stage, the so-called instantaneous ozone demand (IOD), is explained by the fast consumption of ozone by natural organic matter of natural waters and by metal oxides and soil organic matter of slurries to be transformed into various free radical species, which in turn might contribute to the overall ozone consumption. Therefore, in our case IOD is expected to be due to a rapid ozone uptake and its reaction with organic functional groups and inorganic constituents on the GAC surface. In natural water and soil slurry the IOD takes place within less than 1 min but in the presence of GAC, as seen in Fig. 1, it lasts for about 5–10 min. The longer duration of this phase in the presence of GAC is likely due to the porous nature of GAC that may make the process to kinetically develop under pore-diffusion control. As seen in the examples of Fig. 1, the data of the initial phase (c.a. 5–10 min) can be fairly fitted with first order kinetics (R2 > 0.9) to derive an observed ozone decomposition rate coefficient (k O3 ). According to the above discussion this coefficient accounts for the ozone uptake by GAC but also for the ozone that simultaneously decomposes in the liquid bulk. Therefore, k O3 comprises the combined influence of all the pathways consuming ozone within the first minutes of batch experiments. A detailed kinetic study of the ozone decomposition in the presence of GAC will be presented in a forthcoming paper. Nevertheless, though the application of the first order kinetic model is not a rigorous method, for the sake of comparison of the efficiency of various GAC and to evaluate the influence of some variables on the ozone decomposition rate, k O3 values thus obtained have been used in this study. Therefore, further discussion will focus mainly on the initial reaction phase.
0.20
0.15
0.10
0.05 0
1
2
3
4
5
GAC dose, g/L Fig. 2. Effect of GAC dose on the observed rate coefficient of ozone decomposition in the presence of CS30. Conditions of experiments: pH = 7; T = 20 C; CO30 = 7–8 mg/L; agitation speed: 200 rpm.
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
1.00
TBA uptake (ne ), mg/g
20
CTBA /CTBA0
0.95
Freundlich model ne=KfCe1/n
15
10
5
0 0
2
4
6
8
10
TBA concentration (Ce), mM
0.90
0.85 0
50
100
150
200
250
Adsorption time, h Fig. 3. TBA adsorption behavior on GAC. Conditions of experiments: pH = 7; T = 20 C; agitation speed: 200 rpm. Symbols: j CS30; d AQ40.
0.15
0.10
AQ-40*(TBA)
AQ-40*
AQ-40(TBA)
AQ-40
CS-30*(TBA)
CS-30*
CS-30(TBA)
CS-30
0.00
Blank(TBA)
0.05
Blank
does with HO radicals (k = 6 · 108 M1 s1) not leading to the formation of the chain carrier pair ðHO2 =O 2 Þ [22]. Prior to ozone decomposition experiments, the adsorption behavior of TBA on GAC was assessed. Fig. 3 shows the kinetics and equilibrium isotherms at 20 C of TBA adsorption onto CS30 and AQ40. As seen in this figure, both GAC exhibited a quite similar behavior though TBA adsorption onto AQ40 was somewhat slower likely due to its narrower micro and mesoporosity. From the kinetic curves it is also apparent that the concentration of aqueous TBA decreased slowly with time up to achieve 11.7% and 12.2% removal after 10 days when using CS30 and AQ40, respectively. These results mean that the adsorption of TBA on the GAC is rather slow as compared to the ozone decomposition rate (i.e., half-life 5–8 min as can be seen in Fig. 1) but some amount of TBA can eventually be adsorbed. The equilibrium isotherms (inset of Fig. 3) fitted well the Freundlich equation (i.e., ne ¼ K f C e1=n ). The adsorption capacity, given by the Freundlich parameter Kf, is rather similar for CS30 and AQ40 (5.25 ± 0.15 and 5.19 ± 0.24, respectively). The very similar adsorption behavior of TBA on CS30 and AQ40 enables the comparison of the ability of both GAC to produce ozone decomposition through molecular or free radical mechanisms by ozone experiments in the presence of TBA in solution and/or adsorbed. On the basis of the above mentioned observations, ozone decomposition experiments were further carried out in the presence of 1 mM TBA with fresh and TBA pre-loaded GAC. Fig. 4 shows the first order rate coefficients obtained from this series of experiments. The most evident feature of Fig. 4 is that whenever TBA was added to the solution the rate coefficient significantly decreased with regard to that obtained in the corresponding experiment in the absence of TBA. Although such a decrease is greater in the presence of CS30 and CS30* than when using
kO3, min -1
3106
Fig. 4. Effect of the presence of TBA on the observed rate coefficient of ozone decomposition. Conditions of experiments: pH = 7; T = 20 C; CO30 = 7–8 mg/L; agitation speed: 200 rpm; GAC dose (if applied) 0.5 g/ L; TBA concentration (if applied): 1 mM. Patterns: h ozone alone (blank runs); Q O3/GAC (CS30 or CS30*); P O3/GAC (AQ40 or AQ40*). The asterisk (*) means TBA-preloaded GAC.
AQ40 and AQ40*, the k O3 values from all experiments in the presence of dissolved TBA and GAC are relatively close to that of the blank experiment. This indicates that both GAC enhanced ozone decomposition primarily through the acceleration of the HO radical chain reaction in the liquid bulk. As we detected noticeable concentrations of dissolved H2O2 (i.e., 1–3 lM) during the course of the experiments, a plausible argument for the enhancement of ozone decomposition by GAC is the formation of H2O2 from the interaction between ozone and GAC surface functionalities. Some H2O2 would be released to the liquid bulk and, depending on pH, dissociated to HO 2 which is a strong initiator of the transformation of aqueous ozone into hydroxyl radicals [23]. The evidence of H2O2 formation in our study will be shown in the next section. Another interesting feature of Fig. 4 can be noted when comparing k O3 values derived from experiments with and without TBA deposited on GAC (i.e., CS30* and AQ40* versus CS30 and AQ40, respectively). The observed decrease in k O3 when both GAC were preloaded with TBA might be due to one or the two following reasons: (i) TBA could block the access of ozone to reactive sites of the GAC surface; (ii) surface free radicals, including HO radicals, could significantly contribute to the overall ozone decomposition on the GAC surface. Surface hydroxyl radicals might arise from direct O3 surface reactions and from H2O2 decomposition on reducing sites of GAC. In this sense, Oliveira et al. [24] identified basic SOG and p electron rich regions within the basal planes as sites for H2O2 decomposition. Ash content of GAC may also play an important role on direct O3 transformation and H2O2
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
surface decomposition into HO through a Fenton-like process [20]. 3.4. Evidence of the formation of H2O2 from the ozone–GAC reaction In Fig. 5 the profiles of aqueous H2O2 concentration during the course of ozone decomposition experiments in the presence of three GAC are presented. The concentration of H2O2 in the blank experiment (ozone alone) was not detectable at any reaction time but when any of the GAC was used the H2O2 concentration reached a maximum value (c.a. 2–3 lM) after the first 1–2 min of reaction. The H2O2 concentration steadily decreased with time thereafter. This corroborates that a fraction of the dissolved ozone is fast transferred to the GAC surface where decomposes to yield H2O2, which at least partially is released to the liquid bulk. The appearance of H2O2 in solution from the beginning of the ozone experiments suggests that when the driving force for the heterogeneous ozone decomposition reaction is maximum (i.e., difference in ozone concentration), both diffusion of ozone towards the GAC surface and surface reaction steps are faster than the liquid bulk ozone decomposition initiated by the hydroxide ion. As the reaction progresses the gradient of ozone concentration between the liquid and solid interfaces diminishes which slows the kinetics of the heterogeneous process and, therefore, the formation of H2O2. Moreover, the rapid appearance of H2 O2 =HO 2 in solution strongly accelerates the transformation of ozone into hydroxyl radicals through a H2O2 consuming process. The resulting effect is that after
5
the first minutes of reaction a net disappearance of dissolved H2O2 is continuously observed. From Fig. 5 the H2O2 exposure of waterR over the first 10 10 min reaction period was calculated as 0 CH2 O2 dt to be 15.2, 17.4 and 19.7 lM min for AQ40, CS30 and Hydraffin P110 experiments, respectively. As it is apparent from the inset of Fig. 5, the higher the H2O2 exposure the higher the k O3 value. This again suggests that ozone decomposition through the chain reaction initiated by HO 2 in the liquid bulk is the major pathway contributing to the overall ozone transformation. To definitively confirm whether this is true, a series of O3/H2O2 experiments in the absence of GAC were carried out with different initial H2O2 concentrations to obtain H2O2 exposures within the same range as in O3/GAC systems. In the inset of Fig. 5 it can be observed that for a given H2O2 exposure the rate of ozone removal achieved with the O3/H2O2 system is only slightly lower than those of the O3/GAC systems. This clearly demonstrates that such a little fraction of ozone is removed on the surface of the GAC compared to that consumed in the radical chain reaction that develops in the liquid phase. 3.5. Effect of the solution pH The pH is a variable of prime importance in ozone chemistry in aqueous solution. In order to study the effect of the pH on the decomposition of aqueous ozone in the presence of GAC, a series of experiments was carried out at varying pH from 3 to 9. Presented in Fig. 6 are the first order rate coefficients of ozone decomposition obtained from runs without carbon (blank runs) and with CS30 and Hydraffin P110 GAC. As expected, the ozone decay rate increased with the pH in blank experiments due to the increasing concentration of OH which is the only
0.15 0.14
4
0.5
k O3, min-1
0.13
3
0.12
0.4 0.11 0.10
2
C H O , μM
3107
15
20
25
H2O2 exposure, μMmin
kO3, min-1
2
0.3 0.09 10
2
0.2
1 0.1
0 0
5
10
15
20
25
30
t, min
0.0 2
Fig. 5. Evolution of the dissolved H2O2 concentration during O3/GAC experiments. Inset: effect of the H2O2 exposure on the observed rate coefficient of ozone decomposition. Conditions of experiments: pH = 7; T = 20 C; CO30 = 7–8 mg/L; agitation speed: 200 rpm; GAC dose (if applied) 0.5 g/L; initial H2O2 concentration (if applied): 1–5 lM. Symbols: j O3/CS30; d O3/Hydraffin P110; m O3/AQ40; h O3/H2O2.
3
4
5
6
7
8
9
10
pH Fig. 6. Effect of GAC dose on the observed rate coefficient of ozone decomposition. Conditions of experiments: T = 20 C; CO30 = 7–8 mg/L; agitation speed: 200 rpm; GAC dose (if applied): 0.5 g/L. Symbols: j ozone alone (blank runs); d O3/GAC (CS30); m O3/GAC (P110).
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
3108
species that could initiate the ozone decomposition in this event [14]. On comparing experiments in the presence of GAC, a positive effect of pH was also observed with both GAC. Moreover, at any pH, the ozone decomposition rate was enhanced by the presence of GAC. Thus, for example, in experiments using CS30 the ratio between rate coefficients in the presence and absence of GAC were 1.29, 1.32, 2.17 and 2.39 for pH 3, 5, 7 and 9, respectively. These figures also mean that the positive effect of the GAC was more pronounced at neutral and basic pH than at acidic conditions. At pH 3 and 5 the role of GAC in initiating the liquid phase radical chain reaction through HO 2 is unimportant as this strong initiator is not appreciably formed in aqueous solution given the pKa = 11.6 of the H2O2 acid–base equilibrium. Therefore, at acidic pH the ozone decomposition is enhanced just by the ozone uptake on GAC. At neutral and basic pH, in addition to the decomposition of ozone on the GAC surface, H2O2 dissociates in water appreciably into HO 2 thereby accelerating the aqueous ozone decomposition.
Two examples of the proposed ozonation reaction pathway are shown below for 4-pyrone (7) and 4Hchromene (8) O
O
O
O3
O
OH2
O O
O
O
O
O
O
O OH OH
O
-H2O2 O
O
ð7Þ O3
O
OH2
O O
O
O
O
O OH
O
OH
O
-H2O2 O
O
ð8Þ
3.6. Proposed reaction mechanism pathways Based on the general knowledge of the ozone chemistry a possible reaction mechanism explaining the results can be tentatively proposed. Steps of the mechanisms are the following: (i) A fraction of dissolved ozone would be transferred from the liquid bulk to the external surface (Sex) of the GAC particles and from there to the internal surface (Sin) through inside the pores K O3 a
O3 ðlÞ ! O3 ðSex Þ Deff;O3
O3 ðSex Þ ! O3 ðSin Þ
ð1Þ ð2Þ
(ii) Physical adsorption of ozone on the GAC pore sites K ads
O3 ðSin Þ ! O3 ðadsÞ
ð3Þ
(iii) Reaction of adsorbed and/or non-adsorbed ozone, O3(s), with some surface functionalities and inorganic matter of GAC. The complex chemical nature of the GAC surface makes possible a number of reactions but the direct transformation of O3 into HO radicals, the formation of O2H radicals as suggested by Sa´nchez-Polo et al. [10] and the generation of hydrogen peroxide can be highlighted O3 ðsÞ þ GAC ! ! HO ðsÞ O3 ðsÞ þ GAC ! ! O2 HðsÞ
O3 ðsÞ þ GAC ! ! H2 O2 ðsÞ
ð4Þ ð5Þ ð6Þ
Plausible reactions (6) are those of some basic SOG of GAC (pyrone-like and chromene-like structures) following cyclo-addition according to Criegee mechanism similarly to the ozonolysis of dihydropyran [25].
Not only basic SOG but also some types of acidic SOG can react with ozone to produce H2O2. Thus, phenolic structures react with ozone both by electrophilic attack and through cycloaddition leading to dicarboxylic acids while yielding H2O2 as by-product [26]. (iv) Because its high oxidizing power, HO radicals could also react with hydroxyl and carboxyl SOG attached to aromatic rings of the basal planes of GAC leading to the formation of olefins that further would react with ozone to yield H2O2 [27] HO ðsÞ þ GAC ! ! H2 O2 ðsÞ
ð9Þ
(v) Transformation of H2O2 into surface HO radicals catalyzed by p electrons of basal planes, basic SOG and some metal oxides (i.e., ash content) [20,24,28] H2 O2 ðsÞ þ GAC ! ! HO ðsÞ
ð10Þ
(vi) Surface free radicals, SFR (i.e., HO (s) and O2H(s)) might also react with ozone and H2O2 to propagate a surface radical chain reaction which termination steps would lead to GAC gasification: SFRðsÞ þ GAC ! ! Gaseous products
ð11Þ
(vii) Diffusion of hydrogen peroxide from the inside of the GAC to the liquid bulk where it dissociates to HO 2 to an extent that depends on the pH of solution KH2 O2 a
H2 O2 ðsÞ ! H2 O2 ðSex Þ Deff;H2 O2
H2 O2 ðSex Þ ! H2 O2 ðlÞ pK a ¼11:6
þ H2 O2 ðlÞ ! HO2 ðlÞ þ H ðlÞ
ð12Þ ð13Þ ð14Þ
It is worth mentioning at this point that the transfer of free radicals from GAC surface to water is though
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
to be rather unlikely because of their short lifetimes and extremely high reactivity with the GAC surface itself [29]. (vii) Radical type chain reaction of ozone decomposition in the liquid bulk initiated by OH and HO 2 [22] O3 ðlÞ þ OH ðlÞ
HO 2 ðlÞ ! ! HO ðlÞ
and
ð15Þ 3.7. Comparison between various GAC. Effect of textural and chemical properties To address the impact of textural and chemical surface properties of GAC on the aqueous decomposition of ozone the next series of experiments was designed to compare the various GAC selected in this work. In Fig. 7 the first order rate coefficients computed from these experiments are plotted against the four textural properties of GAC presented in Table 1 (i.e., SBET, Sext, V1 and V2). Graphs A and B of Fig. 7 contrast with the findings of earlier studies where the rate of ozone decomposition was related to the surface area and pore volume of activated carbons [10,11]. First, contrary to the reported by Guiza et al. [11] graph A does
GRAPH B
0.14 P110
CS30
0.12
P110 CS30
0.12 CS20 CS10
0.10
AQ40
Darco SS4P
0.08
0.06
CS10
0.10
0.06
200
400
600
800
1000
1200
Blank 0.0
0.2
0.4
0.6
GRAPH C
0.14
GRAPH D
P110 CS30
0.12
0.10
kO3, min-1
kO3, min-1
CS20
CS10
Darco SS4P
0.08
0.06
50
0.10
0.06 100
150
200
Sext, m2/g
250
300
CS10
AQ40
Darco SS4P
0.08
Blank
0
P110
CS30
0.12
CS20
AQ40
0.8
Meso- and macropore volume (V2), cm3/g
SBET, m2/g
0.14
Darco
SS4P
0.08
Blank 0
CS20
AQ40
kO3, min-1
kO3, min-1
not show a linear relationship between the ozone decay rate coefficients and the BET surface areas of the GAC used. Secondly, Sa´nchez-Polo et al. [10] concluded that the ozone decomposition on GAC may be greatly affected by the volume of meso- and macropores as the process may be controlled by pore diffusion. In our case study, however, as can be seen in the graph B, this volume does not affect the rate of ozone decomposition likely due to the fact that all the used GAC exhibited a volume of meso- and macropores large enough to facilitate the diffusion of ozone and hydrogen peroxide through the pore network. Graphs C and D are also inconclusive as data are rather scattered and therefore no direct relationship can be deduced between textural properties of GAC and rate of ozone decomposition. On the basis of the reaction pathways discussed in the previous section it can be expected that if no serious diffusion limitations develop and ozone is weakly adsorbed onto GAC by physical interactions the presence of sites on GAC able to react with ozone and/or HO(s) to produce H2O2 would control the kinetics of the process. Therefore, SOG and ash content of GAC might have a joint effect on the rate of ozone decay. To asses what is the relative impact of each variable the following were analyzed as predictors
GRAPH A
0.14
3109
350
Blank 0.0
0.1
0.2
0.3
0.4
0.5
Micropore volume (V1), cm3/g
Fig. 7. Effect of textural properties of GAC on the observed rate coefficient of ozone decomposition. Textural properties: (A) BET surface; (B) meso and macropore volume; (C) external surface; (D) micropore volume. Conditions of experiments: pH = 7; T = 20 C; CO30 = 7–8 mg/L; agitation speed: 200 rpm; GAC dose (if applied) 0.5 g/L.
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
3110
Table 2 Summary of multiple regression (Eq. (16)) and statistical analysis Coefficient value (ai)
Variable
Standard error
2
2
t-value
p-value
13.139 13.702 4.424 1.752
<0.0005 <0.0005 0.011 0.150
Intercept Basic SOG, lmole g1 Hydroxyl SOG, lmole g1 %Ash
5.559 · 10 1.182 · 104 1.071 · 104 5.791 · 104
0.425 · 10 0.086 · 104 0.242 · 104 3.408 · 104
Overall significance
R-square = 0.9804
Adj. R-square = 0.9658
Analysis of variance
Degrees of freedom
Sum of squares
Mean square
F statistic
Model Error Total
3 4 7
0.00389 7.769 · 105 0.00397
0.0013 1.942 · 105
66.82
by a forward stepwise multiple linear regression method: carboxyl, hydroxyl, lactone, carbonyl and basic groups as well as ash content. By considering all these six variables, the model failed to accurately predict the ozone decay rate constant due to the low significance of carboxyl, carbonyl and lactone-type SOG in the model (p 0.5). Then, a model with just three variables as predictors (Eq. (16)) was eventually used for the linear regression fitting procedure
0.14
kO3, min-1 (Experimental)
0.12
0.10
Root-MSE = 0.0044
kO3 ¼ a1 Cb þ a2 COH þ a3 %Ash
ð16Þ
where Cb, and COH are the concentration of basic and hydroxyl SOG, respectively and %Ash is the percentage of ash in GAC. The results of multiple regression analysis and statistical calculations are presented in Table 2. As seen in this table the confidence level was of 85% (i.e. p < 0.15). To check for the accuracy of the model in predicting k O3 , in Fig. 8 a plot of the calculated values against the experimental ones is presented. As seen all data are close to the diagonal which gives validity to the model. From the values of the regression coefficients (ai) of Table 2 it is concluded that both basic and hydroxyl SOG positively affect the ozone decay rate while the ash content exerts a negative influence which suggests that the role of metal oxides is the decomposition of H2O2 rather than direct ozone transformation. 3.8. Repeated use of GAC
0.08
0.06 0.06
0.08
0.10
0.12
0.14
kO3, min -1 (Computed with the regression model) Fig. 8. Experimental rate coefficient of ozone decomposition against the rate coefficient predicted by the best-fit of Eq. (16). Symbols: j Fresh GAC; h Re-used GAC.
The activities of CS30 and Hydraffin P110 GAC towards ozone decomposition were tested by completing a series of consecutive ozone decomposition runs re-using the GAC up to 200 times (i.e., overall ozone dose of about 2.8 g O3/g GAC). As a result of repetitive ozonation the individual SOG contents of GAC noticeably varied and continuous decreases in k O3 were observed as shown in Table 3. Thus, after applying an ozone dose of about 2.8 g O3/g GAC the rate of ozone decomposition drastically decreased to be only c.a. 40% higher than that achieved in the process carried out in the absence of
Table 3 First order rate coefficient of ozone decay and SOG content of GAC throughout repeated use of GAC in ozone experiments GAC
Times used
Overall ozone dose applied (mg O3/g GAC)
k O3 (min1)
Acidic SOG (lmole g1) Carboxyl
Lactone
Basic SOG (lmole g1) Hydroxyl
Carbonyl
Total
Total
CS30
1 50 100 200
14.4 730 1410 2816
0.1193 0.0991 0.0904 0.0822
18 72 96 127
12 18 24 34
81 57 51 43
65 99 161 224
176 246 332 428
417 339 243 161
Hydraffin P110
1 200
11.8 2764
0.1271 0.0798
0 117
28 54
28 189
222 278
278 638
651 0
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
GAC. The apparent SOG changes induced by ozonation were the disappearance of basic and hydroxyl groups with the formation of carboxyl, carbonyl and lactone-type groups. These results agrees well with the reaction mechanism proposed in this work as basic and hydroxyl groups are pointed out as primary active sites of GAC for ozone decomposition with formation of acidic SOG (i.e., carboxyl, carbonyl and lactone) and H2O2. The decrease in k O3 as a result of pre-ozonation is thought to be a direct consequence of the loss of GAC active sites. In fact, as it is shown in Fig. 8, Eq. (16) was successfully applied to predict the ozone decay rate coefficient for re-used GAC. 4. Conclusions The results from this investigation give reasoning to the positive effect of GAC on decomposing aqueous ozone into hydroxyl radicals. This effect depends mainly on the GAC dose, pH and the chemical surface of GAC. There is an optimum GAC dose range that leads to the highest specific ozone decay rate. The pH exerts a positive influence on the ozone decomposition kinetics due to the formation of H2O2 from surface ozone reactions that at neutral and basic pH partially dissociates to HO 2 which is a strong initiator of the ozone decomposition in the liquid bulk. H2O2 was detected at significant concentration in the aqueous solution during ozone experiments in the presence of a GAC. It was found that the rate of ozone decomposition was related to the amount of H2O2 released from the GAC surface to the liquid rather than to the direct ozone surface decomposition. Therefore, the GAC is not a catalyst for ozone decomposition but an initiator of the O3/ H2O2 system. First-order rate coefficients of ozone decay could not be directly related to some textural properties of GAC (i.e., surface area and pore volume) as far as the GAC used in this work is concerned. However, surface chemistry plays a key role in promoting surface ozone decomposition to yield H2O2. By applying a multiple regression model it was demonstrated that basic and hydroxyl SOG of GAC are the groups that most influence the kinetics of the process. Although the statistical significant of this result was rather low (i.e., 85% confidence level) it seems that the ash content of GAC has a negative impact on ozone decomposition as it may decompose H2O2 on the GAC surface thereby decreasing the net formation of the HO 2 in aqueous solution. As a result of the ozone-GAC interaction the removal of basic and hydroxyl SOG takes place with formation of carboxyl, carbonyl and lactonetype groups. As a consequence, after applying such an ozone dose the rate of ozone decomposition decreases drastically. The results obtained in this work clearly show that GAC enhances the aqueous ozone decomposition into HO radicals through H2O2 formation and deserve further studies to address a more detailed kinetic description of the surface reactions (i.e., ozone–GAC and H2O2–GAC interactions) as an useful tool for the selection and application of specific
3111
GAC in such a promising AOP aimed to the removal of specific water pollutants (targeted compounds). Acknowledgements The authors thank the C.I.CY.T of Spain and European Feder Funds for the economic support through the project PPQ2003/00554. Ms. Gira´ldez also thanks the Spanish Ministry of Education for providing her a FPU grant. References [1] Legube B, Leitner NKV. Catalytic ozonation: a promising advanced oxidation technology for water treatment. Catal Today 1999;53(1): 61–72. [2] Zaror CA. Enhanced oxidation of toxic effluents using simultaneous ozonation and activated carbon treatment. Chem Technol Biotechnol 1997;70(1):21–8. ´ lvarez PM, Montero de [3] Beltra´n FJ, Rivas JF, Ferna´ndez LA, A Espinosa R. Kinetics of catalytic ozonation of oxalic acid in water with activated carbon. Ind Eng Chem Res 2002;41(25):6510–7. [4] Beltra´n FJ, Acedo B, Rivas FJ, Gimeno O. Pyruvic acid removal from water by the simultaneous action of ozone and activated carbon. Ozone Sci Eng 2005;27(2):159–69. [5] Rivera-Utrilla J, Sanchez-Polo M. Ozonation of naphthalenesulphonic acid in the aqueous phase in the presence of basic activated carbons. Langmuir 2004;20(21):9217–22. [6] Ma J, Sui MH, Chen ZL, Wang LN. Degradation of refractory organic pollutants by catalytic ozonation – activated carbon and Mnloaded activated carbon as catalysts. Ozone Sci Eng 2004;26(1):3–10. [7] Faria PCC, Orfao JJM, Pereira MFR. Mineralisation of coloured aqueous solutions by ozonation in the presence of activated carbon. Water Res 2005;39(8):1461–70. [8] Sa´nchez-Polo M, Leyva-Ramos R, Rivera-Utrilla J. Kinetics of 1,3,6naphthalenetrisulphonic acid ozonation in the presence of activated carbon. Carbon 2005;43(5):962–99. [9] 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. [10] 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. [11] Guiza M, Ouederni A, Ratel A. Decomposition of dissolved ozone in the presence of activated carbon: an experimental study. Ozone Sci Eng 2004;26(3):299–307. [12] Beltra´n FJ, Rivas J, Alvarez P, Montero-de-Espinosa R. Kinetics of heterogeneous catalytic ozone decomposition in water on an activated carbon. Ozone Sci Eng 2002;24(4):227–37. [13] Sa´nchez-Polo A, Rivera-Utrilla J. Effect of the ozone–carbon reaction on the catalytic activity of activated carbon during the degradation of 1,3,6-naphthalenetrisulphonic acid with ozone. Carbon 2003;41(2): 303–7. [14] Staehelin J, Hoigne´ J. Decomposition of ozone in water: rate of initiation by hydroxide ions and hydrogen peroxide. Environ Sci Technol 1982;16(10):676–81. ´ lvarez PM, Beltra´n FJ, Go´mez-Serrano V, Jaramillo J, Rodrı´guez [15] A EM. Comparison between thermal and ozone regenerations of spent activated carbon exhausted with phenol. Water Res 2004;38(8): 2155–65. [16] Vidic RD, Tessmer CH, Uranowski LJ. Impact of surface properties of activated carbons on oxidative coupling of phenolic compounds. Carbon 1997;35(9):349–59. [17] Bader H, Hoigne´ J. Determination of ozone in water by the indigo method. Water Res 1981;15(4):449–56. [18] Bader H, Sturzenegger V, Hoigne´ J. Photometric-method for the determination of low concentrations of hydrogen peroxide by the
3112
[19]
[20]
[21]
[22]
[23]
P.M. Alva´rez et al. / Carbon 44 (2006) 3102–3112
peroxidase catalyzed oxidation of N,N-diethyl-p-phenylenediamine (DPD). Water Res 1988;22(9):1109–15. Hoigne´ J, Bader H. Characterization of water-quality criteria for ozonation processes. 2. Lifetime of added ozone. Ozone Sci Eng 1994; 16(2):121–34. Lim HN, Choi H, Hwang TM, Kang JW. Characterization of ozone decomposition in a soil slurry: kinetics and mechanism. Water Res 2002;36(1):219–29. Chiang HL, Huang CP, Chiang PC. The surface characteristics of activated carbon as affected by ozone and alkaline treatment. Chemosphere 2002;47(3):257–65. Staehelin J, Hoigne´ J. Decomposition of ozone in water in the presence of organic solutes acting as promoters and inhibitors of radical chain reactions. Env Sci Technol 1985;19:1206–13. Glaze WH, Kang JW. Advanced oxidation processes. Description of a kinetic model for the oxidation of hazardous materials in aqueous media with ozone and hydrogen peroxide in a semibatch reactor. Ind Eng Chem Res 1989;28(11):1573–80.
[24] Oliveira LCA, Silva CN, Yoshida MI, Lago RM. The effect of H2 treatment on the activity of activated carbon for the oxidation of organic contaminants in water and the H2O2 decomposition. Carbon 2004;42(11):2279–84. [25] Thomson QE. Ozonolysis of dihydropyrane. Reactions of 4-hydroperoxy-4-methoxybutyl formate. J Org Chem 1962;27:4498–502. [26] Yamamoto Y, Niki E, Shiokawa H, Kamiya Y. Ozonation of organic-compounds. 2. Ozonation of phenol in water. J Org Chem 1979;44:2137–41. [27] Pi YZ, Schumacher J, Jekel M. Decomposition of aqueous ozone in the presence of aromatic organic solutes. Water Res 2005;39(10):83–8. [28] Huang HH, Lu MC, Chen JN, Lee CT. Catalytic decomposition of hydrogen peroxide and 4-chlorophenol in the presence of modified activated carbons. Chemosphere 2003;51(9):935–43. [29] Voudrias EA, Larson RA, Snoeyink VL. Importance of free radicals in the reactivity of granular activated carbon under water treatment conditions. Carbon 1987;25(4):503–15.