1 Introduction

EU directive 2010/75/EU [1] requires the drastic reduction of NO_x emission in flue gases from the 1st of January 2016. Until now the norms could have been fulfilled in many cases with appropriate primary measures, such as e.g., flue gas recirculation, air staging, low excess air technique, low NO_x burners, etc. The BAT methods for coal fired power stations, allowing the fulfilment of the new norms, require now (with the exception of fluidized bed combustion) that either selective non-catalytic reduction SNCR or selective catalytic reduction SCR be introduced in large stationary sources. Selective catalytic reduction (SCR) of NO_x with ammonia is the most efficient method of NO_x emissions reduction and it is widely used in EU, USA and Japan.

The requirements for in SCR-NH₃ are as follows: a catalyst should:

• • be active in reduction of NO_x in the oxidizing atmosphere and not sensitive to the changes in oxygen content (flue gases usually contain ca. 6% O₂);
• • have a good selectivity to N₂ and not lead to a side product N₂O, as it is a greenhouse gas ca. 300 times more harmful then CO₂;
• • be stable, i.e. not undergo structural or chemical changes under operation conditions;
• • have low activity towards oxidation of ammonia, especially to dinitrogen oxide;
• • have a broad temperature window;
• • not deactivate with H₂O, which is always present in flue gases;
• • be resistant towards typical poisons present in flue gases, especially SO₂.

Additionally, it would be of advantage if the catalysts could be installed at end-tail configuration where the flue gas has already been cleaned of fly ash and most of sulfur dioxide.

A catalyst most often applied in industry applied catalyst is a monolith with vanadium as active component and titanium as a support with the addition of tungsten oxide or molybdenum oxide promoter to increase the catalyst lifetime by decreasing the harmful influence of sulfur oxides. The temperature window of such catalysts is fairly broad in comparison to e.g., Pt-containing catalysts, but optimum operation temperature is from ca. 573 K to 673 K. This catalyst may be installed either (i) upstream of the electrostatic precipitator (ESP) or (ii) downstream of the ESP and upstream of the desulfurization unit (FGD). In the former case, its life may be shortened by fly ash and sulfur dioxide, and in the latter one by SO₂. The operation of SCR downstream of FGD unit requires the reheating of flue gas. This is the main reason why low-temperature catalysts (below 523 K) are searched. Among the best candidates are carbonaceous materials, which, in contrast to V₂O₅/TiO₂ systems, are active at temperature as low as 393 K. Moreover, it was found in many cases that the same active component showed higher NO conversion when supported on active carbon than on Al₂O₃ or TiO₂ [2,3]. There are examples in the literature showing that V₂O₅/carbonaceous materials are more active than commercial V₂O₅/WO₃/TiO₂ [4]. Actually, there exists a DeSO_x–DeNO_x technology based on activated carbons–BF-Mitsui technology [5,6] which is considered a BAT method under appropriate circumstances. In this method, SO₂ is removed from flue gases by adsorption on activated carbon, and NO_x are reduced with ammonia, but, due to insufficient activity of carbonaceous materials, the technology has to apply moving beds. It was observed, however, that different types of modification might result in considerable increase in NO conversion.

The main focus of this work is to discuss, based on published literature and the own results of the authors, the possibilities to enhance the catalytic performance of carbonaceous catalysts and to shed light both on the possible preparation procedures, and open questions concerning the design of successful carbonaceous catalysts. Special attention has been paid to carbons containing oxygen and/or nitrogen surface functionalities, used both as catalysts and as supports for active components.

2 The possibilities of the modification of carbon materials

Carbon materials, especially activated carbons and carbon fibres, are characterized by their textural, as well as chemical and surface properties. They are very important both for adsorption and catalysis but S_BET may not always be correlated with catalytic performance. It was recognized that for carbon materials, the decisive role was played by surface chemistry [7,8]. The change in surface chemistry affects adsorption and catalytic properties of the modified carbonaceous materials. Figueiredo and Pereira [7] gave an overview of the reactions catalyzed by carbon materials, together with the type of sites necessary for these processes. Among the reactions important for environmental protection, the following should be mentioned: selective catalytic reduction of NO_x with NH₃, for which acidic and basic sites are of importance, NO reduction with propene [9], NO or N₂O reduction by carbon [10], oxidation of NO, SO₂ or H₂S [7], which require basic sites, or hydrodehalogenation where pyridinic nitrogen sites were identified as being of primary importance. Apart from those, the influence of surface chemistry of carbonaceous materials was recognized for numerous other catalytic reactions, such as e.g., CO oxidation over carbon-supported Wacker-type catalysts [10], Fischer–Tropsch reaction on Co/carbon nanotubes [11–13], oxidative dehydrogenation [7,8], dehydration of alcohols [7,8], etc. Additionally, the surface chemistry of carbonaceous supports influences the catalytic performance of MeO_x/catalysts as it will be discussed later.

In this short review, only carbonaceous materials for the DeNO_x process are addressed in more detail. The main focus is on the improvement of their catalytic performance, which arises from chemical surface modification. The specific surface area S_BET is believed to be of less importance. Admittedly, Teng et al. [14] found for a series of carbons obtained from the same precursors that there was a correlation between activity and S_BET, but some surface in micropores was not totally utilized. Several other authors did not find such correlation, possibly also due to the complicated character of activated carbons porosity and internal area being only partly utilized. It must be mentioned, however, that porosity may play an important role in case bigger pores are present. Such conclusions may be drawn from the works of:

• • Loiland and Lobo [15], who worked on NO oxidation at low temperature over different materials, stated that carbon samples dictated catalytic activity more than the zeolite and metal-organic framework materials, this being the result of irregular pore structure of these materials;
• • García-Bordejé et al. [16], who observed differences in stability for V₂O₅/AC (micro- versus mesoporous) for a SCR mixture containing SO₂. No plugging by accumulated NH₄HSO₄ was observed for mesoporous carbon in contrast to catalysts based on microporous support;
• • Huang et al. [17] reported that NO conversion for V₂O₅ supported on carbon nanotubes (CNT) increased with the mean diameter of CNT. The effect was explained by differences in V₂O₅ dispersion.

On the other hand, Su et al. [18] studied carbon nanotubes with MnO_x supported on inner or outer surface of nanotubes. Taking into account that the former showed higher NO conversion than the latter, it seems to suggest that internal diffusion is not the key influencing factor, at least for carbonaceous support with such large porous system.

From the point of view of DeNO_x applications, the most interesting modifications are those leading to surface groups containing oxygen or nitrogen surface functionalities. Surface oxygen species present on carbonaceous materials include (i) acidic species, such as carboxylic acids and anhydrites, lactones, lactols and phenols; (ii) neutral species, such as carbonyl or ether species; and (iii) quinone, chromene or pyrone groups whose basic character is still under discussion [7]. A review of types of surface groups on carbonaceous materials, their stability as a function of temperature and the analytical tools used for their characterization was given by Figueiredo and Pereira [7]. To summarize shortly, several types of groups may either form spontaneously on the surface of active carbons or be introduced by different modification techniques [7]. Fig. 1a summarizes the methods resulting in the formation of oxygen or nitrogen surface functionalities. Fig. 1b shows typical oxygen- or nitrogen-containing groups formed on carbonaceous materials.

3 The influence of oxygen surface groups on the catalytic performance in DeNO_x

Several authors paid attention to the role of oxygen-containing surface species in SCR-NH₃ [22,24,79–83]. The general conclusion was that the presence of such species led to the increase in activity. The observed gain in conversion was in most cases rather small, generally of the order of 10% or lower, depending on the used type of oxidative pretreatment [e.g., 20,84,85]. The dependence of NO conversion on temperature showed in most cases a minimum both for non-oxidized and oxidatively treated activated carbons at around 423–473 K (or 523 K) [83,84]. Thus, the positive effect of oxygen functionalities usually decreased over 453 K. Additionally, a deterioration of catalytic performance with time on stream was observed [83], most probably resulting from the removal of acidic surface sites.

Table 2 summarizes most important experimental information on catalytic activity of carbonaceous materials with the increased number of oxygen functionalities. The increase in SCR activity was found by the following.

Ahmed et al. [85] for commercial active carbons (AC) modified by concentrated H₂SO₄ at room temperature, followed by treatment in N₂ flow at 373, 473 or 573 K. There was a correlation between NO conversion and oxygen bulk content. H₂SO₄ treatment led to the formation of acidic surface functionalities, which increased NH₃ sorption capacity and played a positive role in SCR.

Mochida et al. [86] for polyacrilonitrile-based activated carbons oxidized by H₂SO₄ and subsequently heat-treated at 623–873 K. There was an optimum temperature of heat treatment which left surface oxidized but removed sulfur species.

Ku et al. [87] for activated carbon modified by the solution of sulfuric acid, or first ammonium sulfate and then sulfuric acid, followed by heat treatment in nitrogen at 673 K. This sort of treatment, as previously found in many studies, eliminates carboxylic species, and this was then rationalized in the mechanism, as carboxylic species not being the main reaction sites.

Teng et al. [80], who studied activated carbon fibres (ACF) oxidized by HNO₃ or H₂SO₄. Depending on the presence or absence of oxygen in reaction mixture, activity formed a sequence:

• • with O₂: ACF oxidized by HNO₃ > ACF oxidized by H₂SO₄ > ACF;
• • without O₂: ACF oxidized by HNO₃ > ACF > ACF oxidized by H₂SO₄.

Marbán et al. [22] oxidized activated carbon fibres with HNO₃ solutions of different concentrations at 323 or 363 K for 1 or 2 h and with air at 573 K. Activity sequence was the same as that of the acidity of the obtained catalysts.

Szymański et al. [83] who also found the positive affect of HNO₃ treatment.

Sousa et al. [70] confirmed the positive effect of HNO₃ treatment on catalytic properties. Nitric acid modified samples exhibited a ca. 10% higher NO oxidation ability versus untreated samples.

Huang et al. [20] handled the case of viscose-based activated carbon fibres treated with HNO₃ or oxygen plasma and untreated, as discussed in Section 2.1. NO conversion in SCR increased by ca. 10–15% after HNO₃ treatment and by ca. 5% after oxygen plasma treatment. This was ascribed to the formation of oxygen functionalities, which was higher in the former than in the latter case.

Higher influence of pre-oxidation by HNO₃ in comparison to that carried out by oxygen plasma was also registered by Zhu et al. [88], who compared the performance of DeNO_x for CeO₂ supported on viscose-based activated carbon fibres so pretreated. It was observed that NO conversion was higher for the former than for the latter catalysts, which was explained by the positive role of oxygen functionalities in active phase dispersion and the formation of adsorption sites for NH₃.

Boyano et al. [24], who studied the effect of oxidation of carbon layer formed on ceramic honeycomb. They noticed that the acidic character of the surface is of key importance for achieving a high NO reduction efficiency.

The mechanism of SCR with NH₃ was discussed based both on DeNO_x results, studies of NO interaction with carbonaceous materials (i.e. the reaction of direct reduction of NO by carbon) and NO sorption studies [89].

Sorption of NO and NH₃ on the surface was connected to acidic–basic character of the surface. NO sorption was ascribed to the presence of basic sites. Ahmed et al. [85] observed that oxidation of AC with H₂SO₄ led to the decrease in adsorption of NO from a mixture containing 15% O₂. This may be attributed to the increase in the amount of acidic groups with the simultaneous depletion of basic groups.

Xue et al. [89] studied (by TPD) NO sorption on activated carbon untreated or oxidized by HNO₃, H₂O₂, and H₂SO₄. The sequence of amount adsorbed was the same as the sequence of the number of basic sites.

Mochida et al. [86] studied reduction of NO by carbon oxidatively pretreated. The increase in NO reduction was explained by the increased adsorption of NO on surface vicinal O-species. The reaction between such two groups was found to be a rate-determining step. N₂ desorbed from the surface and the initial sites were recovered when CO₂ desorbed from the surface.

There is a general agreement that acidic functionalities are sites for ammonia adsorption, although, as discussed below in more detail, the appropriate surface species were not yet identified without doubt.

Zawadzki and Wiśniewski [90] investigated by IR sorption of NH₃ on carbon films prepared by carbonization of cellophane. NH₃ was chemisorbed as NH₄⁺ on carboxylic species. At higher temperatures (473–673 K), additionally amide and imide structures were formed.

Marbán et al. [91] analyzed the mechanism of low-temperature selective catalytic reduction of NO with NH₃ on Mn₃O₄/carbon–ceramic monolith by step-response and temperature-programmed desorption experiments. Although most conclusions concern the role of the active component, NH₃ interaction with AC was also considered. AC surface devoid of carboxylic groups due to pretreatment in N₂ at 673 K showed the following NH₃ surface species:

• • NH₃ adsorbed through hydrogen band to anhydrides;
• • NH₃ adsorbed as NH₂ (with H abstraction);
• • NH₄⁺ ions linked to phenolic surface groups.

Teng et al. [80] proposed the overall mechanism for DeNO_x reaction, basing on the comparison of activity order for carbonaceous materials pretreated with HNO₃ and H₂SO₄, as described above. The number of species evolving CO₂ in TPD formed the same sequence as NO conversion for a mixture NO + NH₃. The number of species decomposing to CO formed a sequence that was the same as for NO conversion from a NO + NH₃ + O₂ mixture. Thus, two types of sites for NH₃ sorption, different for a mixture with or without oxygen were postulated: phenolic groups for the latter and OH part of carboxylic group for the former. According to the authors, adsorption of NO occurs either (i) in the nearest vicinity of NH₄⁺ species i.e. on the CO part of carboxylic group (for a mixture without oxygen) or on basic species of pyrone-like type groups (for a mixture with oxygen). The two types of mechanisms are illustrated by the reactions in Fig. 3a–c.

The dependence of SCR activity on the presence of acidic phenolic species rather than carboxylic moieties may be also derived from the results of Ku et al. [87]. Two important findings should be mentioned in this case: (i) the number of CO₂-evolving groups decreased and those evolving CO increased after treatments with H₂SO₄ or (NH₄)₂SO₄·H₂SO₄ while SCR activity increased, and (ii) NO conversion depended linearly on the amount of NaOH used for neutralization i.e. on acidity. Thus, acidic phenolic groups which evolve CO on decomposition were identified as the most likely sites in the reaction.

On the other hand, the study of Szymański et al. [83] indicates that mechanism (b) may not be the only possible route for a reaction mixture containing oxygen. The studies were carried out on active carbon oxidized with HNO₃, which leads predominantly to the formation of acidic (carboxylic, anhydrides, lactones and phenolic species). The reaction procedure was: the measurement of NO conversion in an increasing temperature sequence (413, 453, 493, 533 and 573 K) and afterwards at a decreasing temperature sequence (533 and 493 K). A clear decrease in activity for the same temperature was observed for the “downward” curve in the comparison to the “upward” one. According to the literature [25,92,93], carboxylic groups decompose at 373–673 K and anhydrites and lactones around 883 K (both to CO₂), while phenol, carbonyl, quinone and pyrone groups decompose to CO at respectively 923, 973, 1073 and 1123 K. Taking into account that the highest temperature in the experiment of Szymański et al. [93] was 573 K, the decrease in activity observed on the “downward” curve should be ascribed rather to the removal of carboxylic species than phenol groups, which are much more stable. Possibly both routes (a) and (b) are valid in the presence of oxygen in an SCR mixture, the former being the secondary one and the latter the primary one.

4 The influence of nitrogen functionalities or catalytic performance in DeNO_x

4.1 Activity and selectivity

Apart from oxygen functionalities, surface groups containing nitrogen have been considered as beneficial for DeNO_x reaction [32,34,70,80,87,94–96]. Generally, it was found that the modification with N-compounds resulted in a much higher increase in activity than that observed for oxidatively pretreated activated carbons, as illustrated by Szymański et al. [83] who compared NO conversion for activated carbons treated under vacuum at high temperature (DHT), activated carbon oxidized with HNO₃ (D-ox) and DHT and D-ox treated with NH₃ at 1173 K. For low temperature (413 K) and high temperature (573 K), NO conversion formed a sequence: D-ox-N > D-HT-N > D-ox > DHT. At medium temperatures, D-ox was slightly better than DHT-N but still less active than D-ox-N.

On the whole, however, the direct comparison of the effects of modification procedures and compounds is rather difficult, because of varying DeNO_x testing conditions. The procedures used for modification and their influences on catalytic performance in SCR-NH3 are summarized in Table 2.

One of the first articles published by Singoredjo et al. [94] described DeNO_x properties of carbonaceous materials obtained by co-carbonization of sucrose with glucosamine, uracil, picolylamine or 2-hydroxy-methyl-pirydine or impregnation of activated carbons with glucosamine. The latter led to the highest SCR activity.

Ku et al. [87] compared NO conversion for activated carbons modified either by O₂–NH₃, or H₂SO₄, or (NH₄)₂SO₄/H₂SO₄. The last of the mentioned proved the most efficient.

When activated, carbon composites were treated under vacuum at high temperature (1073 K, sample HT), with NH₃ at 1073 K (sample AM), with HNO₃ and subsequently with NH₃ at 523 K (sample ONA) or ammoxidized at 723 K (sample AO), NO conversion at 373–673 K formed a sequence: AO > ONA > AM > untreated sample > HT [84].

Grzybek et al. [34,97] and Szymański et al. [83] compared, under the same conditions, the behavior of N-promoted activated carbons obtained by the treatment with (i) NH₃ at 1073 K [34,83], (ii) aqueous solution of urea followed by curing in air [34] and ammoxidation at 623 K [98]. The latter samples were prepared in the following way: activated carbon N either untreated or pre-oxidized with concentrated HNO₃ at 363 K for 1 h were ammoxidized with a mixture of NH₃ + O₂ (1:3) at 623 K for 4 h, then cooled under argon (respectively further addressed as C/350/NH₃-O₂ or C90/350/NH₃-O₂). These samples were characterized by low temperature argon sorption (S_BET), XPS (N/C ratio and chemical state of N-functionalities and TPD of surface oxides). Fig. 2 compares DeNO_x performance of selected so-modified active carbons [34,97–101]. Oxidative pretreatment prior to the modification with urea or NH₃ at 1073 K [83] or ammoxidation [98] resulted in more active catalysts than the same procedures applied directly to untreated AC. The beneficial effect was explained by the formation of acidic surface sites, which preferentially interacted with N-precursors. The formed N functionalities were believed to be sorption sites for NO molecules [96], thus leading to the improvement in NO conversion. However, urea treatment and ammoxidation resulted in considerable increase in NO conversion between 413 and 573 K, while high-temperature NH₃ treatment was less efficient in this temperature range. No direct correlation with the amount of introduced nitrogen was observed at low DeNO_x temperatures. The urea-treated sample (ca. 3.7 wt% N) showed activity only slightly higher than the ammoxidized sample C90/350 NH₃-O₂ (ca. 7 wt% N) while AC oxidized with NH₃ at 1073 K with ca. 1.4 wt% N was much less active at lower temperatures. A possible explanation may be derived from quantum mechanics calculations of Strelko et al. [102] who suggested that N-doping changes the catalytic properties of carbon by modifying electron mobility and semiconductor properties, to an extent depending on the amount of N, with a certain optimum content. Additionally, apart from N-species, oxygen functionalities present on carbonaceous materials whose DeNO_x performance is given in Fig. 2 may play a positive role. Such oxygen groups are depleted on the surface during the high-temperature NH₃ treatment in contrast to ammoxidation or urea/air treatment.

The second type of procedures leading to N-containing activated carbons was based on carbonization and activation of N-containing precursors: polymers [103] or resins [96].

Muñiz et al. [103] studied N-promoted activated carbons obtained by carbonization and activation of polyaryloamide fibers (Nomex and Kevlar). The applied carbonization and activation conditions leading up to 25% burn-off influenced NO conversion only to a small extent.

Huang and Teng [63] used phenol–formaldehyde resin as the starting material to prepare N-containing activated carbons. Then, 10, 20 or 30 wt% of m-phenylenediamine solution in alcohol was added to the resin, which was subsequently carbonized in nitrogen by heating from room temperature to 1073 K and then gasified in oxygen at 673 K, resulting in carbonaceous materials containing 3.7 to 8.8 wt% N (elemental analysis) and S_BET of ca. 309 to 372 m²/g. Pyridinic and pyrrolic or pyridonic groups accounted for the majority of nitrogen functionalities. SCR-NH₃ experiments carried out between 383 and 503 K revealed a considerable increase in NO conversion in comparison to activated carbon obtained from the resin without the modification with the above-mentioned N-compound. Again no direct correlation between the N content (as determined by elemental analysis) and activity was observed.

5 O- or N-modified carbonaceous materials as supports for transition metal oxides/hydroxides in DeNO_x

Transition metal oxides supported on typical carriers, such as Al₂O₃, SiO₂ or TiO₂ have maximum activity at medium temperature range around 523 to 723 K. Silica-aluminas, such as zeolites or layered clays, promoted with transition metal oxides/hydroxides, may be applied at higher temperatures, over 573 to ca. 873 K, depending on the material [109–112]. In contrast, promoted activated carbons proved to have high activity in the low temperature region, below 573 K. This is illustrated by the article of Yoshikawa et al. [3], who compared MnO_x supported on activated carbon and alumina. At temperatures between 320 and 523 K, NO conversion was several times higher for the former than the latter catalyst. Similarly, Chuang et al. [2] proved that Cu supported on activated carbon was more active than on Al₂O₃. The effect was ascribed to a better distribution of Cu. Gálvez et al. [4] compared commercial V₂O₅/WO₃/TiO₂ catalysts with active carbon briquettes differently pretreated, with petroleum coke ash (main component V) as the active component. Under 573 K, NO conversion was 3–5 times lower for commercial catalyst. The extent of the improvement for carbon-based catalysts depended on the type of their pretreatment.

Several types of active components supported on activated carbons or other carbonaceous materials, such as e.g., nanotubes or semi-cokes were proposed as DeNO_x catalysts. The most often studied were oxides/hydroxides of: vanadium [16,23,27,113–119], manganese [3,18,22,97,113,120–123], iron [3,81–83,121,124] or copper [122,125–127]. Rare earth (Ce, La, Pr, Nd) oxides supported on activated carbon fibres or semi-coke were also tested [88,128]. Several facts for MeO_x supported on carbonaceous materials were well established. One of them is that the chemistry of surface carbon is of primary importance for catalytic performance of the MeO_x-AC system. Table 3 summarizes the most important experimental information on MeO_x/carbon-supported DeNO_x catalysts. In this work, only catalysts with carbonaceous supports with the chemistry modified by the change in oxygen or nitrogen surface functionalities are considered. This choice was made based on the fact that the treatments leading to such changes resulted in improvements of SCR-NH₃ catalytic performance.

6 H₂O and SO₂ influence on catalytic performance

The presence of water in SCR mixtures may lead to a larger or smaller decrease in activity, depending on the catalyst type. Only a few studies address the problem of water for carbon-based catalysts [16,97,113,123,136–138].

At low temperatures, H₂O had an inhibitory effect on DeNO_x performance even when water was not added to the reaction mixture, as found for SCR reaction carried out for ca. 8 h at 413 K for N-promoted activated carbons prepared by urea treatment [97] or ammoxidation [98] (cp. Fig. 2f), as well as for doubly promoted (N and MeO_x) carbons [97]. This is illustrated by Fig. 2 g for N- and MnO_x-promoted active carbon (cp. Fig. 2g [97]). At 453 K, all the above-mentioned activated carbons proved stable without water online and this, in connection with TPD showing the extensive peak of water desorption up to ca. 423–453 K, suggests that probably NH₃ (substrate) and H₂O (product of the reaction) compete for adsorption on hydrophilic sites at lower temperatures of SCR process.

On addition of 10% water to SCR mixture, Muñiz et al. [103] registered that activated carbons obtained by carbonization and activation of Nomex fibers underwent deactivation at temperatures below 523 K. At 373 K, the decrease in NO conversion was ca. 80%. Over 573 K, there was no inhibitory effect [103]. Similar temperature dependence was observed for active carbon fibres oxidized and then treated with NH₃ at 523 K, although in this case some small deactivation was registered, even at higher temperatures [131]. Deactivation upon addition of H₂O was also found at 453–533 K for active carbons doubly promoted with urea and MnO_x [96] or FeO_x [139], but the extent of deactivation was smaller than, e.g., for Mn/Al₂O₃ [138]. Moreover, it was totally reversible after removing H₂O from the feed, similarly as found for vanadium on AC [137], or V, Fe or Cr oxides supported on mesoporous carbon covered cordierite monolith [121]. For V₂O₅/AC, that inhibition effect was reported to increase with increasing H₂O content [16].

It should be stressed, however, that the presence of water in the feed positively influenced selectivity in SCR reaction. For Mn/N/AC catalysts [97], the amount of N₂O in the products diminished, especially at higher temperatures (493, 533 K) to a much higher degree, than would be expected from the decrease in NO conversion. Similar results were registered by some authors for other (non-carbon-based) catalysts: MnO_x supported on γ-Al₂O₃ at 598 K [138], iron-vanadyl phosphate catalyst [141] or on a commercial V₂O₅/WO₃/TiO₂ catalyst [142].

A consistent mechanism including the role of H₂O has not been offered as yet. As discussed in detail in [97], it is not even concluded without doubt if NH₃ or NO sorption sites are affected.

Huang et al. [137] showed for V₂O₅/AC that H₂O did not influence NO sorption, while sorption of NH₃ was even higher in the presence of water due to the increased Brønsted acidity. Bagnasco [141] also reported increased NH₃ sorption at 423 K in the presence of H₂O for iron-vanadyl phosphate SCR catalysts and attributed it to the formation of new Brønsted acid sites.

Valdés-Solís et al. [113] attributed the inhibitory effect of H₂O to its competition for adsorption sites with NH₃ (Mn/carbon–ceramic catalysts). Similar explanation was offered by Amiridis et al. [143] and Klijstra et al. [140] for non-carbon-based catalysts: MnO_x supported on V₂O₅/TiO₂ [143] and MnO_x/γ-Al₂O₃ [140]. Amiridis et al. [143] confirmed this based on kinetic measurements and Raman spectra. Klijstra et al. [140] offered for MnO_x/γ-Al₂O₃ a more complicated mechanism explaining H₂O inhibition. The model takes into account molecular adsorption of H₂O and the formation of stable surface hydroxyls. TPD studies showed that H₂O competed for sites with both NH₃ and NO in the presence of O₂. However, kinetic measurements revealed that the reaction order in NH₃ was not affected, in contrast to reaction order in NO. This was interpreted as inhibition of NO sorption being relevant for SCR. NH₃ was sorbed on the surface in excess and thus its sorption inhibition was believed as not important for the overall NO reduction.

The explanation of the decrease of N₂O formation is also not totally understood. Taking into account that: (i) the increase in selectivity to N₂ was more pronounced for catalysts with high amounts of labile oxygen [97] and (ii) at least part of N₂O originates from oxidation of ammonia, it may be speculated that the decrease in N₂O formation is due to the limitation of available oxygen by H₂O sorption. This may be supported, e.g., by Curtin et al. [144] who claimed that water decreased the number of labile oxygen species on the surface and thus restricted the access of NH₃ molecules to oxidation sites.

The presence of SO₂ in flue gases may also lead to deactivation. On the whole, limited effort has been made to address the problem of sulfur tolerance. Liu et al. [145] published an extensive review on the subject of deactivation of different types of catalysts by SO₂, which shows that there are still a lot of open questions in this matter and in some cases the experimental information is contradictory, even for similar catalysts.

SO₂ interaction with active carbon-based catalysts was considered only in a few publications. The influence of SO₂ on activated carbon modified by N-species was discussed by Samojeden et al. [100,146]. Their research showed that the decrease of NO conversion on catalysts prepoisoned by SO₂ containing mixture [100]. There was a possibility of partial or total recovery of activity after regeneration by heating at helium [146].

Valdés-Solís et al. [113] reported the influence of SO₂ on MnO_x or VO_x supported on carbon–ceramic monolith. First, SCR was carried out on a fresh catalyst, which was then subjected to a stream of SO₂ (350 ppm) and O₂ (3%) in nitrogen, followed again by SCR measurement at 423 K. The poisoning procedure was repeated for different periods of time until no change in the following SCR reaction was observed. Higher deactivation was found for Mn than for V-containing catalysts, due to the formation of manganese sulfates on the surface. For V/AC, the loss of catalytic activity was ca. 20% of the initial value. As a possible reason for the relatively small loss in activity, the formation of ammonium sulfate NH₄HSO₄ or (NH₄)₂S₂O₇, which may react with NO_x, was claimed.

On the other hand, Zhu et al. [115,116] reported that V/AC catalysts showed an increase in activity in the presence of SO₂ at temperatures over 453 K. This was explained by (i) the decomposition of NH₄HSO₄ being promoted by AC and only slightly by V₂O₅ [115], or (ii) the increase of surface acidity [116]. The positive effect was observed for low V contents only. The authors proposed thus two different mechanisms for SO₂ promotional effect of SO₂ for vanadium supported on mesoporous carbon-coated monoliths. Similar behavior was also observed by García-Bordejé et al. [16]. The so-increased activity was stable for at least 20 h at 423 and 453 K and was ascribed to the increased acidity. Additionally, mesoporosity of carbon material played a positive role; for typical carbon supports there is an accumulation of NH₄HSO₄ in the pores while no plugging was found for mesoporous carbons.

These observations were only partly confirmed by Hou et al. [114] who studied the influence of different loadings of V₂O₅ on the extent of deactivation by SO₂. No deactivation by SO₂ was observed for active carbon support and activity in SCR reaction increased for ca. 50 minutes to reach a steady state of ca. 90% for 0.5 or 2 wt% V₂O₅ AC. For higher vanadium loading (5 wt%), a small negative effect was observed. The introduced SO₂ showed a promoting effect even after the removal of sulfur dioxide from the feed, suggesting a positive role of S containing surface species (most probably sulfate anions) on the surface, which form additional sites of acidic character and thus may promote ammonia adsorption.

Certain additives were found to improve catalytic performance in the presence of SO₂. As an example an industrial V₂O₅/TiO₂ catalyst with WO₃ additive is most often quoted. Zhu et al. [147] studied the influence of oxides of W, Mo, Zr and Sn on the catalytic activity of V₂O₅/AC. In the absence of SO₂, NO conversion formed a sequence: V₂O₅/AC > V–Zr/Ac > V–W/AC > V–Sn/AC > V–Mo/AC. When SO₂ was present in the feed, all additives, except MoO₃, increased SCR activity. For Sn addition, the increase was considerable and stable for several hours at temperature as low as 473 K, although SnO₂/AC showed no activity in SCR. NO conversion for V–Sn/AC was stable for several hours, depending on the details of the preparation procedure, which was attributed to different interactions between V and Sn species. Deactivation which followed after some time on stream (20 h to ca. 50 h, depending on the preparation technique used) was explained by pore plugging by deposited sulfates.

Tang et al. [148] compared the deactivation of Mn, Cu, Fe and V oxides supported on ceramic support covered with activated carbon, either promoted or unpromoted. All catalysts showed irreversible deactivation after exposure at 473 K to SO₂ containing feed. The loss in activity formed a sequence: MnO_x > CuO > FeO_x> V₂O₅. Doping of MnO_x with cerium oxide had a negative effect in contrast to Mn/ZSM-5 catalyst. The latter upon contact with SCR feed containing H₂O and SO₂ was reported to show only minimal decrease in DeNO_x activity [127]. On the other hand, doping of MnO_x-based monolithic catalysts with Fe or V decreased the extent of deactivation to reach ca. 51 and 56% of the initial value, respectively [148]. SEM showed that deactivation was due to the formation of oxides of active material and the deposition of ammonium sulfate on the surface, the former effect being irreversible and the latter possible to remove by washing.

The simultaneous presence of H₂O and SO₂ was studied for V₂O₅/AC by Huang et al. [138] who reported a strong deactivation of V₂O₅/AC catalysts, in contrast to the effects observed when only H₂O or only SO₂ was present in feed. H₂O deactivated the catalysts at 523 K only slightly while the presence of SO₂ led to the increase in activity, as discussed above. Two reasons for deactivation by H₂O + SO₂ were given: (i) competitive adsorption of H₂O and NH₃ and/or NO, and (ii) pore blocking by NH₄HSO₄ and (NH₄)₂S₂O₇ whose deposition rate is increased in the presence of water. In the absence of H₂O, those salts could react with NO, which resulted in stable catalytic activity. Deactivation strongly increased with V₂O₅ loading, which was due to the increased SO₂ oxidation.

The simultaneous addition of SO₂ and H₂O to the reaction mixture was also studied by Pourkhalil et al. [123] for 12% MnO_x loaded carbon nanotubes. The decrease in NO_x conversion was observed which was partly reversible after the removal of these poisons. The deactivation was ascribed to the formation of ammonium sulfate salts while no MnSO₄ was registered by XRD. The full initial activity could be regained after heating of the catalyst in helium at 350 °C for 2 h.

7 The application of monolithic supports

Industrial application of DeNO_x catalysts requires an appropriate configuration due to high flows of flue gases. The preferred solution, realized on industrial scale in case of V₂O₅/WO₃/TiO₂ catalysts, is the application of a monolithic or a plate catalyst. For carbonaceous catalysts, the following possibilities are taken into account: (i) carbon honeycombs [24,121,122], or carbon–ceramic monoliths [16,24,117,150], or (ii) carbonaceous materials fixed on plates [149], or (iii) activated carbon briquettes (for smaller stationary sources) [4,132,152,153].

The preparation of carbon monoliths was described e.g. by Marbán et al. [154]. The monoliths were prepared from Nomex rejects fibres carbonized in N₂ at 1123 K, milled and dry-mixed with powdered phenolic resin, formed in a mold, cured and carbonized again, followed by activation in steam at 1023 K, resulting in composites with specific surface area up to ca. 1300 m²/g. Other support-forming carbon fibres (Rayon, phenolic resin-, pitch and PAN-derived) were also studied [121]. As active materials oxides of manganese, iron, vanadium, chromium or nickel were chosen. The DeNO_x activity order was found to be: Fe > Mn > V > Cr > Ni [121].

Ouzzine et al. [122] reported the preparation of carbon monoliths formed from volatile bituminous coal with the following additives: silicate clay, glycerin, methylcellulose, and aluminum phosphate dissolved in o-phosphoric acid. The extrudates were dried, carbonized and activated. As active components, Mn or Cu was used. High NO conversion over a fairly broad temperature window was registered.

Carbon–ceramic materials are classified into two types: integral and coated. The former are obtained by mixing the carbon precursor with binders and other substances but they often have low mechanical strength, while the latter are prepared by dipping of a ceramic honeycomb in a polymeric solution, followed by the procedures of active carbon layer formation: carbonization and activation [132]. Valdés-Solís et al. [132] used cordierite monoliths and phenolic resin, furanic resin and polysaccharides in appropriate solvent as activated carbon precursors. Carbonization was carried out in nitrogen stream up to 973 K (heating rate 5 °C/min) and then physical (steam/nitrogen mixture at 973 K) or chemical activation followed (zinc chloride). Up to 14 wt% C could be introduced, deposited carbon was found to be uniformly distributed and carbon-coated monoliths were claimed to have higher mechanical strength than the uncoated materials [132]. Carbonized monoliths activated by gasification in steam/nitrogen stream, with MnO_x as active components were tested in SCR-NH₃ [150]. The best catalytic results were obtained for catalyst support with 40% burn-off, oxidized with nitric acid and Na-exchanged prior to impregnation with Mn precursor. NO_x reduction was 73% at 423 K at space velocity of 4000 h⁻¹. At higher temperatures loss of selectivity was observed [150].

Cordierite monoliths covered with a blend of two polymers (Furan resin and polyethylene glycol) were proposed by García-Bordejé et al. [16,117]. As active component vanadium oxide was used. Dispersion and structure of vanadium species determined activity and selectivity, respectively. Loading higher than a monolayer resulted in a considerable decrease of both [117]. The addition of a second metal (Fe, Mn, Cu or Cr) led to a small increase in specific activity which was attributed to better reducibility of bimetallic catalyst in comparison to a monometallic one [16]. Boyano et al. [24] studied the influence of oxidative pretreatment of polymeric carbon covered monoliths with VO_x as active component. NO reduction activity followed the trend: V/AC/ox HNO₃ > V/AC/H₂SO₄ > V/AC/H₂O₂ and was related to the type of oxygen surface species created in the oxidation process. Li et al. [151] tested carbonaceous catalysts modified with Cu, which were pasted on aluminum plates with organoclay, and inserted into a fixed-bed flow reactor.

A different approach to the problem was presented by Boyano et al. [152], Lázaro et al. [132,153], and Gálvez et al. [4]. Carbon-supported catalysts were prepared in the form of briquettes from Spanish low-rank coal, which was first pyrolized, then blended with tar pitch and pressed into briquettes. Pressed briquettes were pyrolized and activated by steam or carbon dioxide at different temperatures and residence times [3,153]. Vanadium was chosen as the active component. As one of the possible sources of V, petroleum coke ash was considered [4]. The so-prepared catalysts were active in SCR-NH₃ at low temperatures (423 K) [4,134]. Catalytic briquettes showed also good thermal stability while mechanical strength depended strongly on activation and oxidation processes [134].

8 Conclusions

Carbonaceous materials (activated carbons, carbon fibres, carbon cloth or nanotubes) are promising catalysts for low-temperature SCR with ammonia. Several facts have been well established.

The surface chemistry of carbonaceous material plays a decisive role in the preparation of an efficient catalyst. The most important preparative factors are:

• • the presence of oxygen functionalities; they increase activity only to a small extent, but more importantly, influence the distribution of active component for MeO_x/C systems, thus improving both NO conversion, and in some cases also selectivity. Two effects play a role:
  • ∘ the increase in polarity of carbonaceous materials which allows a better contact with aqueous solution of the precursor,
  • ∘ the direct interaction (exchange) of Meⁿ⁺ cations with acidic surface groups.

On carbonaceous materials whose surface is characterized by higher acidity, smaller aggregates of active component are formed in contrast to bigger aggregates deposited mainly on the outer surface of unoxidized support particles. Smaller aggregates result in lower amounts of N₂O in the product mixture.

The introduction of nitrogen functionalities considerably increases activity, as well as the stability of carbon materials towards oxidation. The extent of the improvement depends on both the type of N-compound used and the pretreatment (oxidative treatment prior to the introduction of N-species is beneficial). The appropriate materials may be also obtained starting from N-containing polymers or other carbon precursors modified by N-species, followed by carbonization and activation.

Several transition metal oxides/hydroxides supported on activated carbons result in the increased activity at low temperature region of SCR-NH₃.

Vanadium, manganese, iron and copper were identified as the most efficient oxides/hydroxides.

The performance depends strongly on the chemistry of the support, as well as on other preparation details, such as the method of introduction of the active material, MeO_x loading and calcination temperature. It is especially important to control the type and amount of oxygen surface functionalities of the support, as explained above. In that respect, wet oxidation, especially HNO₃, proved to be the best choice. However, the overall effect depends also on the concentration of the oxidant, the duration of the process and its temperature.

The introduction of MeO_x usually results in higher conversion, but also in decreased selectivity and stability. The latter two effects depend on temperature and are not very important at low temperatures (T < 473 K).

Methods were established to prepare carbonaceous systems in the form of monoliths – either integral or carbon–ceramic, or carbon briquettes, thus making it possible to use such materials in SCR-NH₃ on an industrial scale.

There are full or partial mechanistic models for modified activated carbons. It was recognized that NH₃ is adsorbed on acidic sites, either phenolic, or carboxylic and lactonic. There are some arguments that phenol groups form the main site, but other mentioned groups also play some role. The sites for NO are less certain. It is agreed that they are basic in character. For unmodified or oxidized carbon materials, CO surface groups are believed to form the reaction sites. The increase in basicity by N-functionalities is responsible for increased NO conversion. On the other hand, some models assume that SCR on carbon materials follows Eley–Rideal mechanism with NO interacting from gas phase.

There are still several open questions that need to be considered.

• • a detailed analysis of the relation between the type of the formed nitrogen functionalities and DeNO_x improvement is necessary, in order to optimize catalytic performance. Especially, the role of such functionalities has to be further studied. The experimental information suggests that they increase NO sorption due to their basic character. Special role of pyrrolic/pyridinic group is suggested. However, there are also indications that N-functionalities, due to their electron donor properties, may be sites for NO oxidation to NO₂. In turn acidic NO₂ molecule sorbs on basic sites;
• • more experimental data on H₂O and SO₂ are needed. In most cases, H₂O leads to deactivation, the higher, the lower the temperature. Upon removal of H₂O, full activity is usually recovered. Additionally, N₂O formation is decreased. It was not decided, however, without doubt, which adsorption sites are influenced, those for NO or NH₃. SO₂ may influence DeNO_x performance in several ways: (i) by the formation of new Brønsted sites, leading to the increase in activity (as found for V₂O₅/AC catalysts), (ii) by poisoning of active component, decreasing conversion, or (iii) by the formation of ammonium sulfates which plug the porous system. The latter may be decreased by changes in porosity, which was proven by higher stability of mesoporous carbons derived from polymer blends in comparison to typical microporous activated carbons;
• • although mechanistic models have been elaborated for carbonaceous materials, only partial solutions exist for more complicated catalysts, e.g., MeO_x/AC or V2O5/activated semi-coke. The most detailed model of Marbán et al. [91] for manganese oxides on carbon support may not necessarily be applicable for typical DeNO_x case, as it concerns temperatures as low as 398 K. The model of Wang, on the other hand, suggests both Langmuir–Hinshelwood and Eley–Rideal mechanisms in SCR-NH₃ on V₂O₅/activated semi-coke. For higher temperatures, direct reduction of NO by carbon cannot also be totally excluded. Thus, more mechanistic studies are necessary. The established fact for MeO_x/C systems is that the active component may increase the acidity and/or the reducibility of the system. Thus (i) more NH₃ is sorbed on the surface, and (ii) NO may dissociatively sorb on MeO_x sites, which are then replenished by oxygen from gas phase. NO may also form nitrosyls. From the studies of Wang et al. [119,127], it may be also derived that NO oxidation on the support may play an important role.

Acknowledgments

The work was supported by AGH grant No. 11.11.210.213.