Removal of Ammonia from Air on Molybdenum and Tungsten Oxide

Mar 13, 2008 - Microporouscoconut-basedactivatedcarbonwasimpregnated with solutions of ammonium metatungstate or ammonium molybdate and then ...
3 downloads 0 Views 799KB Size
Download PDF
Environ. Sci. Technol. 2008, 42, 3033–3039

Removal of Ammonia from Air on Molybdenum and Tungsten Oxide Modified Activated Carbons CAMILLE PETIT AND TERESA J. BANDOSZ* Department of Chemistry, The City College of New York and The Graduate School of the City University of New York, 160 Convent Ave, New York, New York 10031

Received December 7, 2007. Revised manuscript received January 28, 2008. Accepted January 30, 2008.

Microporous coconut-based activated carbon was impregnated with solutions of ammonium metatungstate or ammonium molybdate and then calcined in air in order to convert the salts into their corresponding oxides. The surface of those materials was characterized using adsorption of nitrogen, potentiometric titration, Fourier-transform infrared spectroscopy, X-ray diffraction, and thermal analysis. The results indicated a significant increase in surface acidity related to the presence of tungsten or molybdenum oxides. On the materials obtained, adsorption of ammonia from either dry or moist air was carried out. The oxides distributed on the surface provided Lewis and/or Brønsted centers for interactions with ammonia molecules or ammonium ions. Water on the surface of carbon or in the gas phase increased the amount of ammonia adsorbed via involvement of Brønsted-type interactions and/or by leading to the formation of molybdate or tungstate salts on the surface. Although the amount of ammonia adsorbed is closely related to the number of moles of oxides and their acidic centers, the carbon surface also contributes to the adsorption via providing small pores where ammonia can be dissolved in the water film.

Introduction Ammonia removal is still a current and important issue. This gas not only represents a real air pollutant via particulate matter formation (1), but it is also harmful for human beings (2). All of this, added to a continuous reinforcement of environmental measures and the wide use of ammonia in industries, has motivated research activities in the field of ammonia removal via adsorption process. So far, various types of adsorbent have been used in ammonia retention, such as zeolites, alumina, or activated carbons (3–11). Because ammonia is a basic gas, its adsorption requires the use of an acidic adsorbent with pores similar in size to that of the molecule. That is why activated carbons, due to their high surface area and pore volume, are considered the most efficient adsorbents in pollutants removal (12). Nevertheless, this type of adsorbents still demonstrates two main drawbacks. First, it does not provide enough very small pores that are comparable to the size of ammonia (about 3Å (13)). Moreover, a virgin activated carbon usually does not provide a very acidic surface. All of this limits ammonia * Corresponding author phone: (212)650-6017; fax: (212) 650-6107; e-mail: [email protected]. 10.1021/es703056t CCC: $40.75

Published on Web 03/13/2008

 2008 American Chemical Society

adsorption and facilitates its desorption from the surface when the adsorbent is purged with air (14). That is why ways to improve ammonia adsorption and to provide specific interactions between the gas and the adsorbent’s surface have been proposed. The most common process is carbon oxidation (9, 11). It leads to the formation of new acidic functional groups that strongly interact with ammonia via acid–base reaction and/or via hydrogen bonding. Other methods such as impregnation with metal chlorides (4, 14) or oxides (10) lead to complexation or hydrogen bonding, respectively. Besides carbon’s modifications, the presence of water can improve ammonia uptake via the formation of ammonium ions that bond to the carbon’s acidic centers (9). Several papers show the importance of acid–base reactions in ammonia adsorption that occur via interactions with Brønsted and/or Lewis acidic centers. So far, the former ones have been considered as the most efficient (6, 15). Molybdenum and tungsten oxide have been extensively used in catalysis applications due to their acidic properties (16, 17). Most of the time, tungsten or molybdenum oxides are supported on alumina, silica, zirconia, or zeolites. Nevertheless, an increasing number of papers show the use of activated carbon as support for those catalytic species (17). The objective of this paper is to analyze the effects of the changes in the surface of a microporous carbon after its impregnation with either tungsten or molybdenum oxide, in terms of their impact on ammonia adsorption. Taking into account the high acidity of these two oxides (18, 19), owing to both Lewis and Brønsted acidic sites, an improvement in ammonia uptake is expected. The role of the metal, along with its type, amount, and dispersion, the influence of water, and the impact of the nature of the interactions involved are discussed.

Experimental Section Materials. Four different carbon samples loaded with either tungsten oxide or molybdenum oxide were prepared by impregnation on a coconut shell-based carbon, S208 (Calgon Carbon Corporation). The carbon samples were impregnated with solutions of ammonium metatungstate ((NH4)6H2W12O40) or ammonium molybdate ((NH4)6Mo7O24 4H2O), leading to tungsten or molybdenum loading of either 5 or 10 wt% (metal content). The applied treatment involved incipient impregnation by adding to the carbon the volume of solution equal to its pore volume. The samples were dried overnight at 120 °C and then calcined at 500 °C for three hours in an air atmosphere to remove ammonia and to form tungsten or molybdenum oxides as described in eqs 1 and 2, respectively. (NH4)6H2W12O40 f 12WO3 + 6NH3 + 4H2O

(1)

(NH4)6Mo7O24·4H2O f 7MoO3 + 6NH3 + 7H2O

(2)

The four samples prepared as described are referred to as S208-W5, S208-W10, S208-Mo5, and S208-Mo10 (“W” and “Mo” standing for tungsten and molybdenum, “5” and “10” indicating the metal loading). Another carbon sample, namely S208–500, was prepared by calcination of the initial S208 carbon, at 500 °C for three hours, in order to assess the influence of oxidation of the carbon support surface on ammonia retention. Methods. Ammonia Breakthrough Capacity. Adsorption capacity for removal of ammonia was measured in dynamic conditions, at room temperature. In this process, a flow of VOL. 42, NO. 8, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3033

ammonia diluted in air went through a fixed bed of a carbon sample. The total flow rate of the inlet gas was 450 mL/min with an ammonia concentration of 1000 ppm. The detailed conditions of the experiments are listed in the Supporting Information. To evaluate the influence of water, the experiments for all carbon samples were performed with a flow of ammonia gas diluted either in dry air (ED) or in moist air (70% humidity) (EM). Moreover, the experiments were run with and without a 2 h prehumidification (70% humidity). On all samples, the desorption of ammonia was evaluated when exposed to 360 mL/min of dry air. The combination of all these experimental parameters led to four different experiments for each carbon sample. In two experiments, carbon samples were exposed to a flow of ammonia diluted in dry air with, or without, prehumidification. In these cases, the references of the exhausted samples are respectively: -EPD and -ED (D- dry, P- prehumidification). For the two other experiments, ammonia gas was diluted in moist air with, or without, prehumidification. In these cases, the references of the exhausted samples are, respectively: -EPM and -EM (M- moisture). Surface pH. The pH of the initial carbon samples and the exhausted carbon samples was measured after an overnight stirring of a solution containing 0.4 g of carbon sample powder added to 20 mL of distilled water. Thermal Analysis. Thermogravimetric (TG) curves were obtained using a TA instrument thermal analyzer. About 30 mg of carbon sample were submitted to a regular increase of temperature, from 30 to 1000 °C, with a heating rate of 10 °C/min, while the nitrogen flow rate was 100 mL/min. Textural Characterization. Nitrogen isotherms were measured at -196 °C using an ASAP 2010 (Micromeritics). Prior to each measurement, samples were outgassed at 120 °C. The apparent surface area, SBET; the total pore volume, Vt (calculated from the volume of N2 adsorbed at p/po ) 0.0985); the microporous volume, Vmic (DR method (20)); and the mesoporous volume, Vmes (Vt - Vmic) were obtained from the isotherms. The pore size distributions (PSDs) were calculated using DFT method (21). Potentiometric Titration. Potentiometric titration measurements were performed with a DMS Titrino 716 automatic titrator (Metrohm). The experiments were done in the pH range of 3–10 (22). The surface properties were first evaluated using potentiometric titration experiments (22, 23), assuming that the population of sites can be described by a continuous pKa distribution, f(pKa). The details of the method are described in the Supporting Information. X-ray Diffraction (XRD). X-ray diffraction measurements were conducted using a standard powder diffraction procedure. Adsorbents were ground with methanol in a small agate mortar. The mixture was smear-mounted onto the zerobackground quartz window of a Phillips specimen holder and allow to air-dry. Samples were analyzed by Cu KR radiation generated in a Phillips XRG 300 X-ray diffractometer. A quartz standard slide was run to check for instrument wander and to obtain accurate locations of 2θ peaks.

Results and Discussion As seen from Table 1, the surface pH decreases with the impregnation and the metal loading as a result of the introduction of new acidic sites via metal deposition. S208 carbons loaded with molybdenum, and especially S208Mo10, demonstrate the greatest acidity with a 4-pH-unit decrease compared to the virgin carbon. Calcination without impregnation is accompanied by a slight increase in the pH value. This result can be an apparent effect of an increase in the ash content of basic nature as a result of partial gasification of carbonaceous phase and introduction of oxygen in the form of basic groups. The metal content of the impregnated samples (Table 1) was calculated based on the 3034

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 8, 2008

TABLE 1. Surface pH, Tungsten and Molybdenum Content (M), Ash Content and Number of Functional Groups for the Virgin and Modified Carbons sample

pH

S208 S208–500 S208-W5 S208-W10 S208-Mo5 S208-Mo10

8.66 9.98 7.54 6.75 5.28 4.47

M (%)

ash content (%)

number of groups (mmol/g)

3.6 8.9 5.0 10.7

1.80 3.70 6.38 13.06 9.31 18.03

0.302 0.295 0.749 0.653 1.321 1.773

ash content values and assuming that MO3 (M ) Mo or W) was the only specie formed. Results were corrected considering the ash content in the initial sample. The values obtained are in good agreement with the expected content of metal except for S208-W5 carbon. In this latter case, the presence of other tungsten oxides with a lower O/W molar ratio, as for instance W2O5 and WO2 (24) can explain this apparent discrepancy. Changes in the surface acidity of the samples after modification are also seen on proton binding curves (Figure 1). In agreement with the surface pH measurements, both S208 and S208–500 are basic because only a proton uptake is noticed (Q > 0). Heating in air leads to an increase in the basic character of the sample, as already observed with the pH values. On the other hand, both proton release and proton uptake are detected on impregnated samples, except for S208-W5, which has the least acidic surface. This increase in the acidic character, due to the presence of metal oxides, reveals a significant alteration of the surface acidity, especially for the molybdenum modified samples. The distributions of acidity constants (Figure 1) show five well-defined peaks representing various oxygen-containing functional groups on the surface of S208 carbon, mainly corresponding to phenolic (pKa > 8) and carboxylic groups (pKa < 8) (25). After heating in air, almost no changes in distributions of acidity constants are noticed, indicating the resistance of coconut shell-based carbon for surface oxidation. As expected, impregnation with ammonium tungstate or molybdate leads to an increase in the amount of functional groups that can represent a valuable asset in ammonia retention (Table 1, Figure 1). For tungsten loaded samples, the new pKa distribution, compared to the one of S208, results from the combined effects of the functional groups initially present and those linked to tungstate, which were described by Contescu and co-workers (26). Following their analysis, the peak at pKa between 4 and 5 is assigned to metatungstate ([H2W12O40]6-) and at pKa about 6, to paratungstate ([HW6O20(OH)2]5-). At pKa about 7 and 10, the acidity can be linked to protons associated with other W(O5)O- entities and W(O6)structures, respectively (26) (O atoms in parenthesis being located in W-O-W or W-OH bridges). Despite its lower amount of tungsten, S208-W5 shows a greater amount of groups than S208-W10. This difference suggests differences in the nature of the acidity provided by tungsten species, probably related to their distribution on the surface. Indeed, high contents of metals on the surface may lead to clustering and thus ineffective use of the impregnant. Nevertheless, the sample with the higher content of tungsten has a more heterogeneous surface (more peaks) from the point of view of surface acidity. For molybdenum modified samples, mainly three acidic centers exist, with that at pKa about 5 having a prominent influence on the acidity. Even though there is no certitude regarding the nature of each peak, this new pKa distribution is likely related to combined effects of the functional groups initially present and those linked to molybdenum oxide, as for tungsten-impregnated samples.

FIGURE 2. DTG curves in nitrogen for the virgin and modified carbons.

FIGURE 3. XRD spectra of the virgin and impregnated carbons.

FIGURE 1. Proton binding curves (A) and distribution of acidity constants (B and C) for the virgin and modified carbons. Considering the work of Hu and co-workers, we can assume the formation of the following species on the carbon’s surface: MoO42-, Mo7O246-, and Mo8O264- (27). The pKa distribution can then originate from the presence of the corresponding inorganic acids and the functional groups of the carbon’s surface. Molybdenum-impregnated samples have twice more acidic groups on the surface than their tungsten counterparts, which is consistent with their lower pH values (Table 1). DTG curves (Figure 2) show that the S208 sample has a very small number of surface groups, as suggested by the potentiometric titration results. They decompose at temperature higher than 800 °C, which supports their basic character (28). The peak below 120 °C is related to the removal of physically adsorbed water. A similar trend is noticed for S208–500; however, slightly more groups decompose. The

apparent discrepancy with potentiometric titration results for this sample can be explained by a predominantly chemically neutral nature of the oxygen-containing groups formed on the surface as a result of oxidation. In the cases of S208-W5 and S208-W10, broad peaks around 850 °C might be related to the reduction of tungsten trioxide, WO3. For S208-Mo5 and S208-Mo10, sharp and well-defined peaks can be attributed to MoO3 reduction (24, 29). The results also suggest that partial reduction of these two species start to take place at 600 °C. Peaks are bigger for molybdenum-containing samples than for their tungsten counterparts, probably because, even though weight % is similar, the molar ratio is more favorable for molybdenum, whose molar mass is almost half of that of tungsten (96 g/mol vs 184 g/mol). Thus, reduction of more moles of molybdenum oxides results in more oxygen, thus more extensive effect of gasification. XRD spectra (Figure 3) provide a more accurate insight of the inorganic matter deposited. For S208-W10, peaks at 2θ around 26.6, 38.5, and 44.7° might be attributed to WO3, whereas the one at 2θ around 27.5° is likely due to the presence of W6O16(OH) (30). This last compound supports our previous hypothesis regarding the presence of oxides with a lower O/W molar ratio compared to WO3. For the S208-Mo10 sample, evidence of the presence of Mo5O7(OH)8 is seen via peaks at 2θ around 26.0, 37.0, and 44.7°. Peaks at 37.4 and 53.7° might be related to MoO2 (31). S208-W5 and VOL. 42, NO. 8, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3035

TABLE 2. Textural Parameters Calculated Form Nitrogen Adsorption Isotherms at -196 °C for the Virgin and Modified Carbons. sample

SBET [m2/g]

Vt [cm3/g]

Vmeso [cm3/g]

Vmic [cm3/g]

Vmic/Vt

S208 S208–500 S208-Mo5 S208-Mo10 S208-W5 S208-W10

1069 987 850 684 908 878

0.55 0.51 0.43 0.36 0.48 0.44

0.01 0.01 0.00 0.02 0.03 0.01

0.54 0.50 0.43 0.34 0.45 0.43

0.98 0.98 1.00 0.94 0.94 0.98

S208-Mo5 samples do not reveal any diffraction peak due to their low metal content and high dispersion. The textural parameters were also analyzed (Table 2). The virgin carbon S208 has a high apparent surface area and is mainly microporous. Calcination of that sample leads to a slight decrease in Vt and Vmic, due to the partial gasification and destruction of pore walls caused by the oxidation. Although one could expect an increase in the volume of mesopores, their unchanged values are likely the result of compensation caused by deposition of surface functional groups. The observed decrease in structural parameters for impregnated samples can be explained by both a pore blocking effect and by the metal deposition and can also be detected on the pore size distributions (Figure 4). An increase in the metal content results in more pronounced changes in the porosity. Impregnation with molybdenum affects the porosity to a greater extent than impregnation with tungsten. For the former samples, 15 and 30% decreases in the apparent surface area are observed, in comparison with about 10 and 20% for the latter. This might be once again related to the twice greater content of molybdenum than tungsten in terms of moles of impregnant. Based on the NH3 adsorption capacity values (Table 3), the virgin carbon S208 does not appear to be an efficient adsorbent for ammonia, and its average capacity is about 1.3 mg/g. Nevertheless, it is still interesting to notice that the efficiency of this sample increases 2-fold when water is preadsorbed on its surface, whereas no improvement is observed when water is present in the challenging gas. This is related to the surface chemistry of this material. Despite the fact that when water is present in the challenging gas and ammonium ions are formed (9), those ions cannot react efficiently with the acidic groups present on the surface of the carbon. On the contrary, at 70% humidity, micropores should be filled with water to a significant degree (31), and in that film, the small amount of carboxylic groups can dissociate and strongly interact with ammonium ions dissolved there. Similar behavior is observed for the S208–500 sample. For this sample, as expected, only a slight change/ increase in the adsorption capacity is noticed after oxidation, due to the limited effect of exposure to air at elevated temperature. Introduction of metal oxides improves ammonia adsorption, but the extent of this improvement is apparently linked to the type and the amount of metal. Indeed, although the highest adsorption capacity obtained with tungsten-containing samples is 6.6 mg/g, 26.6 mg/g is reached with the S208-Mo10 sample, which is an improvement of almost 1 order of magnitude compared to the virgin carbon. Several hypotheses might explain these differences between tungsten and molybdenum-containing samples. First, the initial molybdenum-loaded samples are more acidic than their tungsten-loaded counterparts, which represents an asset in the retention of a basic molecule (15). Nevertheless, we are aware that the acidity is not the only parameter that governs 3036

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 8, 2008

FIGURE 4. Pore size distributions (PSDs) curves in nitrogen for the virgin and modified carbons. ammonia uptake. The nature of the support, the metal loading, and the presence of humidity in the preparation of supported molybdenum or tungsten trioxide catalysts can also affect the form of the oxide, its interaction with the carbon surface, and thus the accessibility of the adsorption centers. Besides the above-mentioned factors, the amount and nature of the deposited species able to interact with ammonia should influence the amount adsorbed. In dry conditions, ammonia is expected to interact with surface acidic centers of both Lewis and Brønsted types coming from introduced oxides/hydroxides, and in wet conditions, the chemistry should be totally different. Indeed, in wet conditions, following previous observations (26, 27, 32), taking into account the origin of the oxides present on the surface, and assuming that their valency does not change, one would expect formation of the corresponding heteropolyacids previously mentioned. Thus, when exposed to ammonia, and especially to ammonium ions, those heteropolyacids can react and form the ammonium salts deposited on the surface. Although the more likely salts seem to be (NH4)6H2W12O40 and (NH4)6Mo7O24, the following compounds are also expected: (NH4)2MoO4, (NH4)4Mo8O26, (NH4)2WO4, and (NH4)5HW6O21 (27). Considering the twice greater molar amount of molybdenum deposited compared to tungsten (due to molar mass values) and the formulas of the ammonium salts most likely formed, about four times more ammonia should be adsorbed on molybdenum modified samples than on their tungsten counterparts, at wet conditions. Following the above, we also evaluated the efficiency of each impregnation in terms of the amount of ammonia retained on the surface (Table 3). Assuming that ammonium molybdate and tungstate of the initial formulas used for impregnation are formed, the molar ratio of ammonia to metal is expected to be a maximum of 0.5 in the case of tungsten and 0.9 in the case of molybdenum. Analyzing the results after prehumidification, a relatively good agreement is found between the results obtained and the theoretical calculations. Yet, in some cases, the amount of ammonia retained in wet conditions is even higher than expected, and that “excess” reaches 50%. This indicates that another mechanism exists and that ammonia is also likely dissolved in water present in carbon pores, in the form of ammonium hydroxide. This kind of ammonia should be easily removed with the dry gas stream used in the desorption experiment. This explains high concentration of ammonia detected on the desorption run for those samples. Considering now adsorption in dry conditions, different interactions should be considered. At such conditions,

TABLE 3. Ammonia Breakthrough Capacity, Molar Ratio Ammonia to Metal, Amount of Water Preadsorbed and the Changes in the pH after Ammonia Adsorption NH3 breakthrough capacity sample

[mg/g of carbon]

[mg/cm3 of carbon]

S208-ED S208-EM S208-EPD S208-EPM S208–500-ED S208–500-EM S208–500-EPD S208–500-EPM S208-W5-ED S208-W5-EM S208-W5-EPD S208-W5-EPM S208-W10-ED S208-W10-EM S208-W10-EPD S208-W10-EPM S208-Mo5-ED S208-Mo5-EM S208-Mo5-EPD S208-Mo5-EPM S208-Mo10-ED S208-Mo10-EM S208-Mo10-EPD S208-Mo10-EPM

0.9 0.8 1.9 1.6 1.3 0.8 3.5 2.4 1.0 1.2 4.4 3.0 1.4 1.7 6.6 4.9 2.1 3.5 12.5 13.8 5.7 17.9 26.6 23.2

0.5 0.4 1.0 0.8 6.201 0.347 1.523 1.099 0.468 0.582 2.064 1.470 0.685 0.845 3.122 2.465 0.989 1.677 6.000 6.587 3.540 9.899 14.550 12.945

molar ratio ammonia to metal

water adsorbed [mg/g]

256 205 260 219 0.216 0.260 0.952 0.647 0.151 0.184 0.713 0.529 0.238 0.395 1.411 1.524 0.321 1.011 1.502 1.310

358 324 363 314 346 358 282 286

pH initial

exhausted

8.66 8.66 8.66 8.66 9.98 9.98 9.98 9.98 7.54 7.54 7.54 7.54 6.75 6.75 6.75 6.75 5.28 5.28 5.28 5.28 4.47 4.47 4.47 4.47

9.23 8.91 9.74 8.13 9.95 9.57 9.31 8.45 7.53 8.58 8.22 8.07 6.78 7.38 6.86 6.98 5.32 6.09 5.37 5.50 4.98 4.99 5.38 5.23

FIGURE 5. Ammonia breakthrough curves with their desorption part for the metal loaded carbons. ammonium ions are not formed, and ammonia can interact either with Brønsted and/or Lewis acidic sites. In the latter case, the lone pair of electrons is involved, and in the case

of Brønsted interactions, ammonia accepts a proton from the surface sites. So far, Brønsted interactions are considered as the strongest ones (6, 15). VOL. 42, NO. 8, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3037

Some conclusions about the strength of surface interactions can be derived from the desorption curves (Figure 5). Lack of desorption curves in some cases, as for the S208-Mo10 run after prehumidification, indicates that, for three hours, the concentration of ammonia in the effluent gas was higher than 100 ppm (sensor limit), indicating weak adsorption forces. An accurate discussion on the desorption curves remains difficult at this stage of our study, owing to the fact that several parameters have to be taken into account and that the relative importance of each of them is still unknown (amount of adsorbed ammonia, ratio of ammonia interacting with Brønsted acid sites to ammonia interacting with Lewis acid sites, metal loading, and dispersion). Nevertheless, a semiqualitative interpretation can be proposed. On samples S208-Mo5-ED, S208-Mo5-EM, and S208-Mo10-ED, ammonia concentration decreases quickly due to the low amount of preadsorbed ammonia in those cases. An interesting comparison can be made between S208-Mo5-EPM and S208-Mo5-EPD. Indeed, the amount of adsorbed ammonia is similar for both samples, but the concentration of desorbed gas remains higher than 100 ppm in the case of the sample run in dry air. This means that interactions between NH3 and the carbon’s surface are different and/or that the properties of the system change. Running experiments in dry air likely causes a gradual removal of preadsorbed water. Thus, the ratio of ammonia retained by dissolution in water to ammonia adsorbed on Brønsted sites decreased. The results suggest that ammonia adsorbed on Brønsted sites, probably in larger pores, is easier to remove than that dissolved in water. Generally speaking, when the size of pores in which water is retained are relatively big (which happens at high humidity level), ammonia is detected earlier during desorption experiments because of the weaker adsorptive forces for water with adsorbed ammonia in those pores compared to forces in smaller pores. Thus, in the complexity of ammonia adsorption, not only does the type of adsorption sites contribute but also their location in the pore system. When more molybdenum is present in the system, a noticeable decrease in the degree of microporosity is observed. This causes, for both samples run after prehumidification in dry and moist air, the concentration of ammonia higher than 100 ppm to be detected at the desorption run. Because of the changes in the distribution of adsorption sites, this weak adsorption can also be explained by the above presented hypothesis (about the role of water, its location, and location of Brønsted centers). Thus, for the interpretation of results in such complex systems, one has to consider the nature of the interactions involved in the different experimental conditions. The small amount of ammonia adsorbed on the surface and the significant amount of ammonia desorbed during air purging make the formation of ammonium salts on the surface of impregnated carbons, in presence of water, quite difficult to detect. Nevertheless, for S208-Mo10 exhausted samples, an indication of these species is seen on the DTG curves (Figure 6) between 150 and 200 °C, which is in agreement with decomposition of ammonium molybdate (24). The results presented in this paper show that molybdenum and tungsten oxides distributed on the surface of activated carbon provide Lewis and/or Brønsted centers for interactions with ammonia molecules or ammonium ions. Water on the surface of carbon or in the gas phase increases the amount of ammonia adsorbed via involvement of Brønsted interactions and/or via leading to the formation of molybdate or tungstate salts on the surface. Although the amount of ammonia retained on the surface is closely related to the numbers of moles of oxides and their acidic centers, the 3038

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 8, 2008

FIGURE 6. DTG curves in nitrogen for S208-Mo10 exhausted samples. carbon surface also contributes to the adsorption by providing small pores where ammonia can be dissolved in the water film.

Acknowledgments This work was supported by (Army Research Office) ARO grant No. W911NF-05-1-0537. The authors are grateful to Dr. Jacek Jagiello for SAIEUS software.

Supporting Information Available The details of the NH3 breakthrough measurements, potentiometric titration, and FTIR analysis of the initial samples: method, spectra, interpretation. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Manahan, S. E. Environmental Chemistry 7th Edition; Lewis: Boca Raton, 1999. (2) Issley, I.; Toxicity, Ammonia. eMedecine Journal; 2001, 2; www.emedecine.com/EMERG/topic846.htm. (3) Helminem, J.; Helenius, J.; Paatero, E. Adsorption Equilibria of Ammonia Gas on Inorganic and Organic Sorbents at 298.15 K. J. Chem. Eng. 2001, 46, 391–399. (4) Sharanov, V. E.; Aristov, Y. I. Ammonia adsorption by MgCl2, CaCl2 and BaCl2 confined to porous alumina: the fixed bed adsorber. React. Kinet. Catal. Lett. 2005, 85, 183–188. (5) Park, So.-J.; Kim, B.-J. Ammonia removal of activated carbon fibers produced by oxyfluorination. J. Colloid Interface Sci. 2005, 291, 597–599. (6) Yin, X.; Han, H.; Gunji, I.; Endou, A.; Ammal, S. S. C.; Kubo, M.; Miyamoto, A. NH3 adsorption on the Brönsted and Lewis acid sites of V2O5(010): A periodic density functional study. J. Phys. Chem. B 1999, 103, 4701–4706. (7) Ellison, M. D.; Crotty, M. J.; Koh, D. K.; Spray, R. L.; Tate, K. E. Adsorption of NH3 and NO2 on single-walled carbon nanotubes. J. Phys. Chem. B 2004, 108, 7938–7943. (8) Valyon, J.; Onyestak, G.; Rees, L. V. C. Study of the dynamics of NH3 adsorption in ZSM-5 zeolites and the acidity of the sorption sites using the frequency-response technique. J. Phys. Chem. B 1998, 102, 8994–9001. (9) Le Leuch, L. M.; Bandosz, T. J. The role of water and surface acidity on the reactive adsorption of ammonia on modified activated carbons. Carbon 2007, 45, 568–578. (10) Stoeckli, F.; Guillot, A.; Slasli, A. M. Specific and non-specific interactions between ammonia and activated carbons. Carbon 2004, 42, 1619. (11) Mangun, C. L.; Braatz, R. D.; Economy, J.; Hall, A. J. Fixed bed adsorption of acetone and ammonia onto oxidized activated carbon fibers. Ind. Eng. Chem. Res. 1999, 38, 3499–3504. (12) Bandosz, T. J. Desulfurization on activated carbons. In Activated Carbon Surfaces in Environmental Remediation, Bandosz T. J. Ed.; Elsevier: Oxford, 2006; pp 231–292.

(13) Thompson, J. C. Compressibility of metal-ammonia solutions. Phys. Rev. A 1971, 4, 802–804. (14) Petit, C.; Karwacki, C.; Peterson, G.; Bandosz, T. J. Interactions of ammonia with the surface of microporous carbon impregnated with transition metal chlorides. J. Phys. Chem. C 2007, 111, 12705–12714. (15) Petit, C.; Bandosz, T. J. The role of water and surface acidity on the reactive adsorption of ammonia on modified activated carbons. Phys. Chem. Chem. Phys. Submitted. (16) Wivel, C.; Clause, B. S.; Claudia, R.; Mørup, S.; Topsøe, H. Mössbauer emission studies of calcined Co-Mo/Al2O3 catalysts: Catalytic significance of Co precursors. J. Catal. 1984, 87, 497– 513. (17) Ma, X.; Gong, J.; Yang, X.; Wang, S. A comparative study of supported MoO3 catalysts prepared by the new “slurry” impregnation method and by the conventional method: their activity in transesterification of dimethyl oxalate and phenol. App. Catal. A 2005, 280, 215–223. (18) Freedman, M. L. The surface acidity of tungsten (VI) and molybdenum (VI) oxides. Anal. Chem. 1960, 32, 637–639. (19) Bernholc, J.; Horsley, J. A.; Murrell, L. L.; Sherman, L. G.; Soled, S. Brønsted acid sites in transition metal oxide catalysts: Modeling of structure, acid strengths, and support effects. J. Phys. Chem. 1987, 91, 1526–1530. (20) Dubinin, M. M. Porous structure and adsorption properties of active carbons. In Chemistry and Physics of Carbon, Walker, P. L., Ed.; M. Dekker: New York, 1966; Vol 2, pp. 51–120. (21) Lastoskie, C.; Gubbins, K. E.; Quirke, N. Pore size distribution analysis of microporous carbons: a density functional theory approach. J. Phys. Chem. 1993, 97, 4786–4796. (22) Jagiello, J.; Bandosz, T. J.; Schwarz, J. A. Carbon surface characterization in terms of its acidity constant distribution. Carbon 1994, 32, 1026–1028.

(23) Jagiello, J. Stable numerical solution of the adsorption integral equation using splines. Langmuir 1994, 10, 2778–2785. (24) Weast, R. C.; Melvin, J. A. Handbook of Chemistry and Physics 62nd Edition; CRC Press: Boca Raton, Florida, 1981. (25) Kortum, G.; Vogel, W.; Andrusso, K. Dissociation Constants of Organic Acids in Aqueous Solutions; Butterworth: London, 1961. (26) Contescu, C.; Jagiello, J.; Schwarz, J. A. Chemistry of surface tungsten species on WO3/Al2O3 composite oxides under aqueous conditions. J. Phys. Chem. 1993, 97, 10152–10157. (27) Hu, H.; Wachs, I. E.; Bare, S. R. Surface structures of supported molybdenum oxide catalysts: Characterization by raman and Mo L3-Edge XANES. J. Phys. Chem. 1995, 99, 10897–10910. (28) Papirer, E.; Bantzer, J.; Sheng, L.; and Donnet, J. B. Surface groups on nitric acid oxidized carbon black samples determined by chemical and thermodesorption analyses. Carbon 1991, 29, 69– 72. (29) Shaheen, W. M. Thermal behaviour of pure and binary Fe(NO3)3 · 9H2O and (NH4)6Mo7O24 · 4H2O systems. Mater. Sci. Eng., A 2007, 445–446, 113–121. (30) Hanawalt, J. D. Inorganic Index to the Powder Diffraction File; Berry L.G., 1971. (31) McCallum, C. L.; Bandosz, T. J.; McGrother, S. C.; Muller, E. A.; Gubbins, K. E. A molecular model for adsorption of water on activated carbon: Comparison of simulation and experiment. Langmuir 1999, 15, 533–544. (32) Ostromecki, M. M.; Burcham, L. J.; Wachs, I. E.; Ramani, N.; Ekerdt, J. G. The influence of metal oxide additives on the molecular structures of surface tungsten oxide species on alumina: I. Ambient conditions. J. Mol. Catal. A: Chem. 1998, 132, 43–57.

ES703056T

VOL. 42, NO. 8, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3039