Introduction

Over more than a decade, use of titanium dioxide (titania, TiO2) nanoparticles for catalysis and in photo-electrochemical reactions has become increasingly important, owing to their high chemical and thermal stability [1–3]. Anatase titanium dioxide is widely used as preferred support material for selective catalytic reduction of NO x by ammonia (NH3-SCR). The overall NH3-SCR process is promoted by the presence of abundant Lewis acidic sites on the surface of anatase titanium dioxide [4–6]. Many researchers have successfully used anatase titanium dioxide to improve catalytic efficiency for NH3-SCR of NO x [7–9] and have reported that the presence of abundant Lewis acidic sites on the surface of TiO2 is directly responsible for its high catalytic efficiency. Vanadia–titania catalyst is one of the best combinations, because of its high catalytic activity (at approx. 300 ºC) and resistance to SO2 poisoning [10, 11].

Among the most common TiO2 minerals are rutile, anatase, brookite, ilmenite, and leucoxene [1, 12]. Among these, rutile is the richest in titanium dioxide (it contains 95 % TiO2) and is also the easiest to mine and concentrate. However, the rutile content of titanium deposits has decreased, because of their continuous exploitation [13, 14]. Thus extensive research has been conducted on the production of synthetic TiO2 from its other mineral resources. Ilmenite (FeTiO3), which contains 40–65 % TiO2, has attracted much attention because of its huge reserves (1,300 million tons) [1].

Synthetic TiO2 can be produced by either thermo/electrochemical or hydrometallurgical processes. Because the first of these is poorly efficient and energy consumption is high, it is not economically feasible. The hydrometallurgical process requires either sulfate or chloride leaching to manufacture TiO2 pigment. Although chloride processes have advantages over sulfate processes of low-cost and easy waste management [1], they require high-grade rutile as raw material; this is produced by means of expensive upgrading processes. Also, the final product of the chloride processes is not anatase TiO2, which is potentially resistant to SO2 poisoning [15]. Sulfate processes accept low-grade titanium as raw material and produce more than 98 % anatase TiO2.

We have previously reported use of an SO2-resistant, antimony-promoted V2O5–TiO2 catalyst for NH3-SCR of NO at low temperatures [11]. In this study we focused on synthesis of TiO2 slurry as a support, to reduce the overall cost of the catalyst while maintaining consistent catalytic performance. In fact, use of TiO2 slurry excludes the requirement for other additives (for example aqueous ammonia and chlorine) and extra drying and calcination steps (Fig. 1).

Fig. 1
figure 1

Schematic diagram of sulfate reduction process for production of TiO2 slurry from Ilmenite

Full size image

Experimental

Preparation of TiO2

Titania (TiO2) slurry was produced from ilmenite (FeTiO3) by a sulfate reduction process. A schematic diagram of the sulfate reduction process is displayed in Fig. 1. Finely powdered ilmenite (40–65 % TiO2) was digested with sulfuric acid (Table 1) to form water-soluble sulfates (titanyl and iron sulfates). Metallic iron scrap was used to reduce the ferric iron to ferrous iron; otherwise iron will co-precipitate with the titania and contaminate the final product. The reduced solution was further clarified and filtered to separate the soluble titanyl sulfates from the solid impurities in the sludge (iron sulfate). Subsequently, the solution was cooled to 17 ºC to crystallize the iron, resulting in ferrous sulfate hydrate (FeSO4·7H2O). Until this step, the process is in the “black stage”. Further, this solution was subjected to hydrolysis and nucleation with three different concentrations of titania (0.2, 0.3, and 0.6 %; samples 1, 2, and 3), as shown in Table 1. The resulting solutions of the three different samples were filtered separately and then washed with water to separate titania crystallites. All the titania crystallite samples were separately collected in different beakers and are referred to as solution “A” series. Approximately half of the each solution “A” was independently subjected to a second filtration to reduce the Fe content (<0.3) and then collected in different beakers (referred to as solution “B” series). These six solutions (A and B series) were subjected to drying followed by calcination at 400 ºC for 4 h. These calcined titania (TiO2) samples are referred to as A-1, A-2, and A-3 for the “A” series and B-1, B-2, and B-3 for the “B” series.

Table 1 Specifications of TiO2 samples prepared by use of the sulfate reduction process
Full size table

Deposition of vanadia and antimony

The antimony and vanadia were deposited on the titania slurry support by impregnation. A solution of 2 wt% antimony (Sb) from antimony acetate (Alfa Aesar, 99.5 %), and 2 wt% vanadia (V2O5) from reaction of NH4VO3 (Junsei, 99 %) with oxalic acid (Junsei, 99 %) was prepared in a beaker (solution “C”). Six batches of solution “C” were prepared in six different beakers. Subsequently, three of these were independently mixed with appropriate amounts of three different “A” series solutions prepared as described above. Mixing was followed by dehydration in a rotary evaporator, drying, and calcination at 400 ºC for 4 h. The three samples obtained are referred to as SbVA-1, SbVA-2, and SbVA-3. Similarly, rest of the three solutions of ‘C’ were independently mixed with desired amounts of three different solutions of above prepared ‘B’ series and then followed the similar procedure as explained above. These final three samples are referred as SbVB-1, SbVB-2 and SbVB-3.

Catalyst characterization

The physicochemical characteristics of all the catalysts were determined by nitrogen adsorption–desorption at −196 ºC by use of an ASAP 2020 V3.00 Q facility (Micromeritics). Each sample was degassed for 2 h at 200 ºC before BET (Brunauer–Emmett–Teller) measurement. The average pore size distribution of the calcined samples was calculated by use of the Barrett–Joyner–Halanda (BJH) model from a 60 point BET surface area plot. Adsorption isotherms were used to calculate mean pore diameters and distributions. A Perkin–Elmer 2400 series II ICP (inductively coupled plasma) analyzer was used for elemental analysis of all the samples. The morphology of fresh samples was observed through a JEM 2100F (Jeol) transmission electron microscope (TEM). X-ray diffraction (XRD) spectra were recorded using Ni-filtered Cu Kα radiation at 40 kV and 20 mA. NH3-TPD (temperature programmed desorption) was performed on the prepared catalysts by use of an Autochem 2920 analyzer (Micromeritics) equipped with a thermal conductivity detector (TCD). Before each experiment, samples (0.1 g) were pre-heated to 200 ºC under helium (He) flow, followed by NH3 adsorption at 80 ºC and at a flow rate of 50 cm3 min−1 for 60 min to achieve adsorption equilibrium. He flow was then continued for 30 min, followed by heating at 100 ºC for 1 h to remove physically adsorbed excess NH3. Finally, NH3 desorption was performed from 50 to 600 ºC at a constant heating rate of 10 ºC min−1.

Activity measurements

SCR (selective catalytic reduction) of NO over the synthesized samples was evaluated with ammonia as reductant. Activity tests were conducted using a fixed-bed quartz reactor (10 mm i.d., 50 cm length) placed vertically in a tubular furnace. In all experiments, the sample (0.5 mL catalyst sieved to 40–50 mesh size) was preheated with an O2–N2 gas mixture at 400 ºC for 1 h, followed by activity measurements in the temperature range 180–400 ºC. A simulated flue gas consisting of 800 ppm NO, 800 ppm NH3, 3 % O2, and N2 balance was fed into the reactor at a gas hourly space velocity (GHSV) of 60,000 h−1. The concentrations of NO in the inlet ([NO]in) and outlet ([NO]out) of the reactor were measured by use of an online infrared gas analyzer (NDIR Fuji Electric Systems). The ammonia concentration was measured before ([NH3]in) and after ([NH3]out) the reaction by use of a Gastec model GV-100 detector. Finally, NO and NH3 conversions (%) were calculated as follows:

$$ {\text{NO conversion}} = \left( {\left[ {\text{NO}} \right]_{\text{in}} -\left[ {\text{NO}} \right]_{\text{out}} } \right)/\left[ {\text{NO}} \right]_{\text{in}} \times 100 $$
$$ {\text{NH}}_{ 3} {\text{conversion}} = \left( {\left[ {{\text{NH}}_{ 3} } \right]_{\text{in}} -\left[ {{\text{NH}}_{ 3} } \right]_{\text{out}} } \right)/\left[ {{\text{NH}}_{ 3} } \right]_{\text{in}} \times 100 $$

Results and discussion

Physicochemical characteristics

BET surface area and BJH pore size distribution results were investigated for all the samples by nitrogen adsorption and desorption using an ASAP 2020 at liquid nitrogen temperature. The BET surface area, pore diameter, and pore volume of different samples after the first and second filtrations are presented in Table 2. The isotherms of samples A-1 and B-1 (Fig. 2) are classified as type II according to IUPAC [16]. The isotherms for the first-time filtered sample (A-1) contain hysteresis loops (H4 type) around partial pressures (P/P 0) from 0.65 to 0.95, indicating the presence of mesopores. The mesoporous structures of first-time filtered samples with different concentrations of titania nuclei (0.2–0.6 %) are indicative of low surface area. The second-time filtered samples give hysteresis loops starting at a partial pressure of 0.5 and ending at 0.97, suggesting the presence of more micropores and mesopores. As a result, these samples had a high specific surface area owing to the formation of additional micropores. The samples with the lowest amount of nuclei (0.2 %) contained pores of larger diameter and smaller volumes after the first-time filtration. Increasing titania nuclei concentration (0.3 and 0.6 %) led to formation of smaller agglomerates with a slight decrease in pore diameter and increased pore volume (Table 2). Formation of typical mesoporous structures by first-time filtered samples (A-1, A-2, A-3) was because of removal of iron from the larger agglomerates during the calcination. This results in larger pore sizes of 7.7, 5.7, and 6.5 nm (Table 2) for the A-1, A-2, and A-3 samples, as indicated in Table 2. Subjecting the samples to a second filtration resulted in a decrease in pore diameter and an increase in pore volume, owing to the formation of micro and mesopores. It is possible that most of the iron present in the bulk was removed during calcination, leading to the formation of additional micro pores, which apparently increases the specific surface area for second-time filtered samples (B-1, B-2, and B-3). It was also observed that the Fe was greatly reduced by 80–90 % by second filtration (Table 1). Li et al. [17] reported that the N2 adsorption–desorption isotherms of milled ilmenite contained no hysteresis loop, indicative of a non-porous structure with extremely low surface area. This confirms the sulfate reduction process is highly effective in production, from ilmenite, of TiO2 with increased surface area and pore size distribution.

Table 2 Physicochemical properties of TiO2 samples prepared by the sulfate reduction process
Full size table
Fig. 2
figure 2

N2 adsorption–desorption isotherm

Full size image

Morphology and structure

Figure 3a, c shows the TEM images of samples with different concentrations of nuclei (A-1 with 0.2 % TiO2 and A-3 with 0.6 %) after one filtration. Addition of a low concentration (0.2 %) of TiO2 nuclei led to formation of bigger aggregates, as observed in Fig. 3a compared with the sample containing the high concentration of titania nuclei (Fig. 3c). This increased size of the primary agglomerates and aggregates was attributed to the presence of residual iron sulfate and/or iron oxide after calcination [18]. The second filtration then calcination resulted in a significant reduction in the iron content (Table 1). However, a small amount of iron remained, bridging the particles and/or primary agglomerates of the samples filtered a second time. Figure 3b, d clearly shows the smaller agglomerates in second-time filtered samples. As a result, the surface area of these samples was significantly increased (Table 2). Figure 4 depicts the XRD patterns of six (“A” and “B” series) different samples. The XRD spectra of all the samples confirmed the presence of distinctive peaks corresponding to anatase TiO2 (PDF-21-1272) obtained from sulfate reduction of ilmenite. However, no significant change in 2θ could be observed for six samples prepared by varying the TiO2 nuclei concentration (0.2, 0.3, or 0.6 %) with two-time filtration. No significant peaks corresponding to vanadia and antimony was observed for samples SbVA-1, SbVB-1, SbVA-2, SbVB-2, SbVA-3, and SbVB-3 (figure not shown). This is possibly because of efficient dispersion of vanadia and antimony over the surface of the TiO2 support.

Fig. 3
figure 3

TEM images of samples A-1 a, B-1 b, A-3 c, and B-3 d

Full size image
Fig. 4
figure 4

XRD patterns of freshly prepared TiO2 samples after first and second filtration

Full size image

Catalytic activity

Activity was measured for samples SbVA-1, SbVB-1, SbVA-2, SbVB-2, SbVA-3, and SbVB-3. The results are shown in Figs. 5 and 6. Figure 5a, b shows NO and NH3 conversion by first-time filtered samples with different nuclei concentrations (0.2, 0.3, and 0.6 %). Samples prepared with the low concentration of nuclei (0.2 %) led to high NO and NH3 conversion at low temperature. Increasing the nuclei concentration resulted in a decrease of overall NO and NH3 conversion at low temperature. This may suggest the presence of iron inside the large pores of support A-1 which can interact with antimony and vanadia to increase the overall surface acidity of sample SbVA-1. This increases ammonia adsorption by the vanadia and antimony-impregnated surface for reaction with NO. This leads to conversion of NO with NH3 to N2 and H2O at low temperatures. However, the SbVA-2 and SbVA-3 samples resulted in low conversion compared with SbVA-1 at low temperature. It is possible most of the iron in the support of samples SbVA-2 and SbVA-3 is present in the bulk, leading to reduced surface acidity of the samples which results in less activity at low temperatures. NO and NH3 conversion by first-time filtered samples containing different concentrations of nuclei follows the order SbVA-1 > SbVA-2 > SbVA-3. In the samples subjected to second-time filtration the iron content was reduced by approximately 80–90 % and the surface area was simultaneously increased by formation of additional micro pores. In fact, by performing the second filtration step the order of NO and NH3 conversion efficiency was reversed. These conversions are shown in Fig. 6a, b. It seems that the small amount of bridging iron (0.6 %) in sample SbVB-3, as a result of the second filtration and the calcination, might have led to meso/micro pores on the surface compared with the other two samples (SbVB-1 and SbVB-2). This may lead to increased surface acidity of the samples. Thus, the overall NO conversion efficiency was increased. The results from determination of BET pore size distribution clearly reveal larger pore diameter and large pore volume for sample B-3 (Table 2) compared with samples B-1 and B-2. The activity data clearly show that the NO conversion by SbVB-1 decreased slightly compared with that by SbVA-1, whereas for the other two samples, subjected to second filtration, NO conversion at low temperatures was increased. Among all the samples, the larger amount of iron in sample SbVA-1 resulted in the best catalytic activity for NO conversion at low temperatures, owing to the bigger pore size of the support.

Fig. 5
figure 5

a Effect of TiO2 nuclei concentration on NO conversion (reaction conditions: [NO] = [NH3] = 800 ppm, [O2] = 3 vol%, N2 balance, GHSV = 60,000 h−1), b Effect of TiO2 nuclei concentration on NH3 conversion (reaction conditions: [NO] = [NH3] = 800 ppm, [O2] = 3 vol%, N2 balance, GHSV = 60,000 h−1)

Full size image
Fig. 6
figure 6

a NO conversion of second-time filtered samples (reaction conditions: [NO] = [NH3] = 800 ppm, [O2] = 3 vol%, N2 balance, GHSV = 60,000 h−1), b NH3 conversion of second-time filtered samples

Full size image

NH3-TPD

NH3-TPD profiles for the samples is shown in Figs. 7 and 8. Two NH3 desorption peaks were observed for all the samples: one from 80 to 340 ºC and the other from 350 to 500 ºC. These two peaks are ascribed to ammonia desorbed from acidic sites. The low-temperature broad peak is attributed to desorption of physisorbed NH3 and partially ionic NH4 + bound to Brønsted-acidic sites [4]. The high-temperature sharp peak is attributed to desorption of coordinated NH3 bound to Lewis-acidic sites [19]. NH3-TPD profiles for samples SbVA-1, SbVA-2, and SbVA-3 are compared in Fig. 7. Among these samples, the amount of NH3 desorbed from the Lewis-acidic sites at low temperatures was higher for SbVA-1 than for the other two samples (SbVA-2 and SbVA-3) owing to the large pore size of the support (Table 2). Figure 8a, b, c illustrates the effect of iron content on the NH3-TPD profiles for each sample. Ammonia adsorption was greater for samples containing larger amounts of iron (SbVA-1, SbVA-2, and SbVA-3). The second filtration step reduced the amount of iron from 2.5 to 0.2 wt% (Table 1), as a consequence reducing the total acidity of all second-time filtered samples. Comparison of the amount of ammonia adsorbed by SbVA-1 with that adsorbed by the other samples (SbVA-2, SbVA-3, SbVB-1, SbVB-2, and SbVB-3) reveals that sample SbVA-1 contains abundant ammonia species in the form of ionic (NH4 +) and coordinated NH3 at low temperatures. Both of these can participate in the NH3-SCR reaction [20]. This led to the highest NO conversion by sample SbVA-1.

Fig. 7
figure 7

NH3-TPD profiles for samples containing different concentrations of TiO2 nuclei

Full size image
Fig. 8
figure 8

Comparison of NH3-TPD profiles a 1-1 and 1-2, b 2-1 and 2-2, c 3-1 and 3-2

Full size image

Conclusions

A porous slurry of TiO2 containing different concentrations of TiO2 nuclei was prepared from ilmenite by use of a sulfate reduction process. Subsequent filtration steps were used to leach iron. TEM images showed that samples with a low concentration of TiO2 nuclei (0.2 %) formed larger agglomerates. The second filtration reduced the iron by leaching, and the formation of micropores led to increased surface area and pore volume. The NH3-TPD profiles clearly revealed that the presence of excess iron resulted in greater total acidity of the first-time filtered antimony and vanadia-doped samples. The presence of iron inside the large pores of support of SbVA-1 enhanced the adsorption of ammonia species in the form of ionic NH4 + and coordinated NH3 at low temperatures. Therefore apparent NO conversion was enhanced, resulting in the best catalytic activity of SbVA-1 among all the samples.