**Mechanism and Performance of the SCR of NO with NH3 over Sulfated Sintered Ore Catalyst**

**Wangsheng Chen 1,2, Fali Hu 1,2, Linbo Qin 1,2, Jun Han 1,2,\*, Bo Zhao 1,2, Yangzhe Tu 1,2 and Fei Yu <sup>3</sup>**


Received: 25 December 2018; Accepted: 14 January 2019; Published: 16 January 2019

**Abstract:** A sulfated sintered ore catalyst (SSOC) was prepared to improve the denitration performance of the sintered ore catalyst (SOC). The catalysts were characterized by X-ray Fluorescence Spectrometry (XRF), Brunauer–Emmett–Teller (BET) analyzer, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and diffuse reflectance infrared spectroscopy (DRIFTS) to understand the NH3-selective catalytic reduction (SCR) reaction mechanism. Moreover, the denitration performance and stability of SSOC were also investigated. The experimental results indicated that there were more Brønsted acid sites at the surface of SSOC after the treatment by sulfuric acid, which lead to the enhancement of the adsorption capacity of NH3 and NO. Meanwhile, Lewis acid sites were also observed at the SSOC surface. The reaction between <sup>−</sup>NH2, NH<sup>+</sup> <sup>4</sup> and NO (E-R mechanism) and the reaction of the coordinated ammonia with the adsorbed NO2 (L-H mechanism) were attributed to NOx reduction. The maximum denitration efficiency over the SSOC, which was about 92%, occurred at 300 ◦C, with a 1.0 NH3/NO ratio, and 5000 h−<sup>1</sup> gas hourly space velocity (GHSV).

**Keywords:** sintered ore catalyst; sulfate; In-situ DRIFTS; SCR

#### **1. Introduction**

Nitrogen oxide (NOx) is one of the major atmospheric pollutants and is mainly generated from the combustion of fossil fuel, which has serious harmful effects on human health and the ecological environment. In 2016, the total emission of NOx from the iron and steel industry was about 1.04 million tonnes [1]. The sintering process in iron-making plants uses coal or coke as fuel, which is a major emission source of NOx. About 35–50% of the total NOx emission from the iron and steel industry is attributed to the sintering process [2–4]. The new ultra-low emission standard of air pollutants for the iron and steel industry will be issued by Chinese government and require that NOx concentration in the sintering flue gas should be below 50 mg/m3. Hence, it is urgent to address the treatment of sintering flue gas.

Selective catalytic reduction (SCR) over V2O5-WO3 (MoO3)/TiO2 catalyst has been widely applied in power stations because of its high denitration efficiency [5–7]. However, there are still shortcomings such as the toxicity of the catalyst, and the high operation costs [8]. In particular, the reaction temperature of the current commercial catalysts is higher than the temperature of the sintering flue gas [9]. Therefore, the sintering flue gas must be heated to the reaction temperature of the catalyst by using additional fuels.

Many researchers paid more attention to developing the low temperature catalysts or the catalysts prepared by the non-noble metals for sintering flue gas. Zhang et al. reported that the Fe–based catalysts exhibited a high catalytic activity for NO reduction [10]. Wang [11] also stated that the Fe2O3 particles had a good performance during NOx elimination, where the highest NOx conversion reached 95%. Yang et al. [12] claimed that α–Fe2O3 had a poor SCR activity, while γ–Fe2O3 had an excellent SCR activity at 200–350 ◦C. Meanwhile, it was also found that the further increase reaction temperature from 350 to 500 ◦C would suppress NOx conversion over γ–Fe2O3 [13].

It was well known that the sintered ore was one of the raw materials in the sintering plant, its main component was Fe2O3. In our previous study, Han et al. and Chen et al. [14,15] proposed that the hot sintered ore was used as catalysts for NOx removal from the sintering flue gas. The experimental results illustrated that the hot sintered ore had a good denitration performance, the denitration efficiency was about 60% at 300 ◦C, with a 1.0 NH3/NO ratio, and 1000 h−<sup>1</sup> GHSV. However, the denitration efficiency was too low and the NOx concentration from the sintering flue gas could reach the limit after SCR over the sintered ore.

Ciambelli [16] found that introduction of SO2<sup>−</sup> <sup>4</sup> into SCR catalysts can promote the surface acidity of the catalysts. Thus, the catalytic activity was promoted. Khodayari [17] and Xu [18] thought that the sulfation of the catalysts would decrease the oxidization ability of Fe3<sup>+</sup>, and separated the active sites of adsorbing –NH2 and the active sites of oxidizing –NH2. Therefore, the catalytic oxidization of NH3 over γ–Fe2O3 was suppressed. This process resulted in an obvious promotion of NOx conversion. Zhang [19] also investigated the sulfation of CeO2-ZrO2, and their experiment results demonstrated that the sulfated CeO2-ZrO2 provided more surface acidities and acidic sites, and Brønsted acid sites were also increased.

In order to improve the denitration efficiency, SSOC was prepared and the influence of the acidification on SCR performance was investigated. At the same time, the reaction mechanisms of SCR over SSOC were also discussed.

#### **2. Results and Discussion**

#### *2.1. Characterizations*

The main components of the two catalysts were shown in Table 1. It presented that the main components of both two catalysts were Fe2O3 and CaO. After the introduction of sulfuric acid, the proportion of iron oxide in the catalyst was decreased and sulphate content was significantly increased.


**Table 1.** Chemical compositions of catalysts (%).

The BET surface area, pore-size and pore volume of the catalysts were presented in Table 2. The specific surface area of the SOC was 3.684 m2/g, the total pore volume was 0.00714 cm3/g, and the average pore diameter was 6.3 nm. The BET surface area, pore size and pore volume of the SSOC were 4.734 m2/g, 5.9 nm and 0.00964 cm3/g, respectively. It was indicated that the introduction of sulfuric acid would open some micro-pores, which lead to the increase of the total pore volume. At the same time, the sulphate would block some pores. The BET of SSOC was higher than that of SCO, which meant that the sulfation had a positive effect on the pore structure of SOC.


**Table 2.** Textural properties of the catalysts.

The XRD patterns of the catalysts were demonstrated in Figure 1. Several sharp diffraction peaks at 24.1◦, 33.2◦, 35.6◦, 49.6◦, 54.2◦, 57.1◦ and 62.6◦ were observed, which were assigned to α-Fe2O3 (JCPDS NO. 33-0664), and the characteristic peaks of α-Fe3O4 at 30.1◦, 33.2◦, 49.6◦, 54.2◦ and 57.1◦ were also detected in Figure 1. The results indicated that the main components of SOC were α-Fe2O3 and a small amount of α-Fe3O4. After the acidification with sulfuric acid, it could be seen that the peak of α-Fe2O3 was decreased. Meanwhile, the characteristic peaks (25.4◦, 31.4◦, 38.7◦, 52.2◦) of Fe2(SO4)3 were detected, and a small amount of FeSO4 and CaSO4 also appeared.

**Figure 1.** XRD patterns of the catalysts: SOC (**a**), SSOC (**b**) and SSOC after tested (**c**).

X-ray photoelectron spectroscopy (XPS) was used to study the valence states of the catalysts, as shown in Figure 2. It could be seen from Figure 2 (I) that there were obvious peaks of C (1s), O (1s), and Fe (2p) in both of the two catalysts. Moreover, there were obvious S (2p) peaks after the introduction of sulfuric acid. As shown in Figure 2 (II), the binding energies of Fe 2p 2/3 and Fe 2p 1/2 on SOC were mainly centered at about 710.8 and 724.3 eV, which were indications that the iron species were mainly in the form of Fe3<sup>+</sup> in the SOC [20]. At the same time, a small number of Fe2<sup>+</sup> existed, which was also consistent with the XRD results. The binding energies of Fe 2p 2/3 and Fe 2p 1/2 of SSOC were higher than those of SOC. The peak of Fe 2p 2/3 appeared at 711.8 eV and the peak of Fe 2p 1/2 appeared at 724.9 eV. After the acidification with sulfuric acid, both Fe2(SO4)3 and FeSO4 appeared in the SSOC, which contributed to Fe 2p peaks shifting to the higher position. In addition, in Figure 2 (III), the peak of S 2p appeared at 169.3 eV, indicating that sulfur compounds existed in the form of SO2<sup>−</sup> <sup>4</sup> [21].

**Figure 2.** XPS spectra of the catalysts: SOC (**a**) and SSOC (**b**).

#### *2.2. In-Situ DRIFTS Studies*

#### 2.2.1. Adsorption of NH3

Figure 3 showed the FTIR spectra of NH3 adsorption over the SSOC with 1000 ppm NH3/Ar under different temperatures (250–350 ◦C). Thirty minutes after NH3/Ar introduction, the bands at 1690, 1525, 1460 and 1408 cm−<sup>1</sup> were observed. The bands at 1690, 1460 and 1408 cm−<sup>1</sup> were symmetric bending vibration of NH<sup>+</sup> <sup>4</sup> species at Brønsted acid sites [20,22]. The bands at 1525 and 3434 cm−<sup>1</sup> were assigned to the intermediate species, such as ammonium or amide species (–NH2) [20]. Meanwhile, the bands at, 3242 and 3120 cm−<sup>1</sup> could be ascribed to the N–H stretching vibration of coordination NH3, and the bands at 1259 cm−<sup>1</sup> (symmetric bending vibration of NH3 on Lewis acid sites) and 1102 cm−<sup>1</sup> (NH3 species adsorbed on Lewis acid sites) also appeared [20,23,24]. The intensities of the peaks of NH<sup>+</sup> <sup>4</sup> species (1690, 1460 and 1408 cm−1) and –NH2 species (1525 cm−1) first increased and then decreased in the temperature range of 250–350 ◦C. There were peaks at 960 and 930 cm−<sup>1</sup> (NH3 of gaseous state or weak adsorption state) under 300 ◦C. Yu et al. [25] stated that the following reactions probably took place in the reaction:

$$\text{NH}\_3(\text{g}) \xrightarrow{} \text{NH}\_3(\text{a}) \tag{1}$$

$$\mathrm{NH\_3(a) + O(a)/O^{2-} \to NH\_2(a) + OH(a)}\tag{2}$$

$$\text{NH3(g)} + \text{H}^+ \rightarrow \text{NH}\_4^+ \tag{3}$$

According to Equations (1)–(3), the more functional groups formed at the surface of the catalyst, the higher the reaction activity. Thus, the optimum adsorption activity of NH3 over the SSOC was 300 ◦C, as shown in Figure 3.

The dependence of FTIR spectra over the SOC and SSOC on the reaction time at 1000 ppm NH3/Ar and 300 ◦C was presented in Figure 4. Figure 4a demonstrated that there were 6 bands in the range of 1690–1100 cm−1. The bands at 1690, 1405 and 1454 cm−<sup>1</sup> were related to the symmetric bending vibration of NH<sup>+</sup> <sup>4</sup> species, and the band at 1525 cm−<sup>1</sup> was ascribed to the intermediate products of ammonium or –NH2 species. Moreover, the band at 3242 cm−<sup>1</sup> (N–H stretching vibration of coordinated NH3) appeared at 10 min. With the feed of NH3, the adsorption band at 1259 cm<sup>−</sup><sup>1</sup> (symmetric bending vibration of NH3 on Lewis acid sites) appeared at 30 min, and the band at 1107 cm−<sup>1</sup> (NH3 of gaseous state or weak adsorption state) appeared at 60 min. Hence, there were both Lewis acid sites and Brønsted acid sites on the surface of SOC. The FTIR spectra over SSOC dependence of the reaction time was presented in Figure 4b, and the bands of NH3 adsorption were basically the same as those of SOC. Moreover, the bands at 3434 cm−<sup>1</sup> (–NH2 groups) and 3128 cm−<sup>1</sup> (coordination NH3 on Lewis acid sites) were observed because the surface acidity of SSOC was strengthened. There were two new weak bands that appeared at 965 and 927 cm<sup>−</sup>1, where NH3 was in a gaseous state or weak adsorption state. These weakly adsorbed or gaseous NH3 could rapidly adsorbed on the acid sites once the adsorbed NH3 species was consumed. It could also be seen that the strength of NH3 adsorption peak was enhanced after SOC acidification with sulfuric acid, which was the reason that the surface acidity of SSOC was enhanced after the acidification. The influence of the acidity of SSCO on the dentiration efficiency is presented in Figure 5. The different acidity of SSOC was achieved by sulfating with 1, 3 and 5 mol/L sulfuric acid. In this experimental run, the mass of catalyst and the volume of sulfuric acid were the same. Only the concentration of sulfuric acid was varied. It was demonstrated the maximum denitration efficiency occurred at the catalyst treated by 5 mol/L sulfuric acid, and its denitration efficiency was 92.3% at 300 ◦C. At the same time, the denitration efficiency of the catalysts sulfated by 1 and 3 mol/L was 56.6% and 68.5% at the same reaction temperature. Moreover, the experimental results proved that the optimum reaction temperature for all catalysts was 300 ◦C. These sulfates on the surface of SSOC provided more Brønsted acid sites (Peaks at 1690, 1525 cm−<sup>1</sup> were increased), which promoted the adsorption capacity for NH3. Hence, the catalyst denitration performance was also improved [26,27].

**Figure 3.** DRIFT spectra of SSOC in the condition of 1000 ppm NH3 at 250–350 ◦C for 30 min.

**Figure 4.** DRIFT spectra of SOC (**a**) and SSOC (**b**) exposed to 1000 ppm NH3 at 300 ◦C at different time.

**Figure 5.** The influence of acidity of SSOC on SCR performance.

#### 2.2.2. Co-Adsorption of NO and O2

Figure 6 indicated the DRIFTS spectra of co–adsorption of NO and O2 under 250–350 ◦C, and 1000 ppm NO + 15% O2/Ar. After 30 min, there were 4 bands appeared in the range of 1620–1290 cm−<sup>1</sup> and a weak band appeared at 1014 cm<sup>−</sup>1. The band at 1607 cm−<sup>1</sup> was ascribed to bridged nitrate species and adsorbed NO2 molecules [28,29]. The bands at 1485 and 1417 cm−<sup>1</sup> were assigned to bidentate nitrates and monodentate nitrates respectively [30–32]. In addition, the bands at 1293 and 1014 cm−<sup>1</sup> were related to nitro compounds [31]. The variation trend of the intensities of these bands were same as NH3 adsorption in the temperature range of 250–350 ◦C. The adsorption process can be explained by the following formula:

$$\text{NO}(\text{g})\text{NO} + \text{O}(\text{a}) \leftrightarrow \text{NO}\_2(\text{a}) - \text{Bridgednittites} \tag{4}$$

$$\rm NO(g) + O(a) \leftrightarrow NO\_2(a) - Monddotntatedmittites \tag{5}$$

$$\rm NO\_2(g) + O(a) \leftrightarrow NO\_3(a) - Bidenatentrittes \tag{6}$$

$$\text{NH}\_2(\text{a}) + \text{NO}(\text{g}) \rightarrow \text{NH}\_2\text{NO} \rightarrow \text{N}\_2 + \text{H}\_2\text{O} \tag{7}$$

$$\rm NH\_4^+ (a) + NO (g) \rightarrow \langle NH\_3NO \rightarrow NH\_2NO + H\_2O \rangle \rightarrow N\_2 + H\_2O \tag{8}$$

$$\rm NO\_2(a) + NH\_3(a) \to NO\_2[NH\_3]\_2(a) + NO \to 2N\_2 + 3H\_2O \tag{9}$$

$$\text{NO}\_2(\text{a}) + 2\text{NH}\_4^+(\text{a}) \to \text{NO}\_2\text{NH}\_4^+(\text{a}) + \text{NO} \to \text{N}\_2 + 2\text{H}\_2\tag{10}$$

According to Equations (4)–(10), all the peaks in Figure 6 were important to the SCR reaction. The intensity of peaks at 300 ◦C was the highest, which mean that the optimum adsorption temperature for NO was also 300 ◦C.

Figure 7 showed the DRIFTS spectra over SOC and SSOC at different reaction times under 300 ◦C, 1000 ppm NO + 15% O2/Ar. Similar with the spectra in Figure 6, after 10 min, there were 4 bands that appeared in the range of 1620–1300 cm−<sup>1</sup> and a weak band appeared at 1022 cm−1. The bands intensities were gradually increased with the adsorption time. It could be seen that in Figure 7b, the adsorption intensity of SSOC was obviously higher than that of SOC, especially for the nitro compounds (1290 cm<sup>−</sup>1) and the nitrate species (1490 cm−1). The results indicated that the adsorption capacity of NO was improved after the acidification with sulfuric acid. It had been shown that the introduction of SO2<sup>−</sup> <sup>4</sup> enhanced the adsorption of NO [26].

**Figure 6.** DRIFT spectra of SSOC in the condition of 1000 ppm NO and 15% O2 at 250–350 ◦C for 30 min.

**Figure 7.** DRIFT spectra of SOC (**a**) and SSOC (**b**) exposed to 1000 ppm NO and 15% O2 at 300 ◦C for a different time.

The results showed that the peak intensities of the adsorbed species related to NO2 molecules and bridged nitrate species (1607 and 1618 cm−1) were decreased obviously with purging by Ar, which indicated NO2 molecules and bridged nitrate species absorbed at the surface of catalysts were unstable. At the same time, it was found that the intensities of bidentate nitrates, monodentate nitrates and nitro compounds were stable even after an Ar purge.

#### 2.2.3. Reaction between NH3 and NO

The reaction at the surface of catalyst by introducing NH3 and NO + O2 into an in-situ reactor was also studied. Figure 8 shows that NH<sup>+</sup> <sup>4</sup> species (1695 cm<sup>−</sup>1), amide species(–NH2) or ammonium (1533 cm−1), and several weak adsorption bands at 1252, and 3242 cm−<sup>1</sup> (NH3 adsorbed on Lewis acid site) were detected on the surface of SOC after NH3 introduction at 30 min. On the surface of SSOC catalyst, NH<sup>+</sup> <sup>4</sup> species at 1695, 1427 and 1454 cm−<sup>1</sup> and weakly adsorbed NH3 or gaseous NH3 (966, 925 cm<sup>−</sup>1) were also detected. There were two bands at 3434 cm−<sup>1</sup> (–NH2 groups) and 3127 cm−<sup>1</sup> (coordination NH3 on Lewis acid sites) that appeared at the same time. After switching to NO + O2, the adsorbed species of NH3 over SOC and SSOC gradually disappeared and the adsorbed species of NOx appeared. These bands were ascribed to NO2 molecules and bridged nitrate species (1602 and 1618 cm<sup>−</sup>1), bidentate nitrates (1488 and 1498), monodentate nitrates (1413 and 1419 cm−1) and nitro compounds (1290, 1295, 1020 and 1011 cm<sup>−</sup>1). Comparing Figure 8a with Figure 8b, it could be seen that there were more nitrite species and NH3 species adsorbed at SSOC than those at SOC. This could be explained that the SSOC contained more active sites, which resulted from the sulfates on SSOC, and Equations (7)–(10) occurred [33]. The reaction of amide species (–NH2) and NH<sup>+</sup> <sup>4</sup> species on the surface of catalysts with the gaseous NO was E-R mechanism, and the reaction between adsorbed state NO2 and adsorbed NH3 followed L-H mechanism. Therefore, the reaction between NO and NH3 over SOC and SSOC had two mechanisms.

**Figure 8.** DRIFT spectra of SOC (**a**) and SSOC (**b**) successively exposed to 1000 ppm NH3, 1000 ppm NO + 15% O2 for different time under 300 ◦C.

#### 2.2.4. SCR Performance

SCR performance were carried out in a fixed-bed reactor, which was made of quartz with an inner diameter of 20 mm and a length of 1000 mm. In this experiment, the mass of the catalysts samples was 13.49 g, the flow rate of the flue gas which simulated sintering flue gas was about 600 mL/min, the simulated sintering flue gas contained 300 ppm NO, 15% O2 and balance N2. The temperature in the reactor was kept at 100 ◦C–350 ◦C, with the condition of a 0.5–1.0 NH3/NO ratio and 5000 h−<sup>1</sup> GHSV. The NOx concentrations in simulated flue gas at the inlet and outlet of the reactor were continuously recorded by a gas analyzer (PG-350, Horiba, Kyoto, Japan) with an accuracy of ±1.0%. The NOx conversion was calculated according to the following equations:

$$\text{NO}\_{\text{x}}\text{ conversion} = \frac{[\text{NO}\_{\text{x}}]\_{\text{in}} - [\text{NO}\_{\text{x}}]\_{\text{out}}}{[\text{NO}\_{\text{x}}]\_{\text{in}}} \times 100\text{\%} \tag{11}$$

Figure 9 presented the denitration performance of SOC and SSOC in the temperature range of 100 ◦C–350 ◦C. It was found that the reaction temperature had a great effect on the NH3-SCR denitration performance of SOC and SSOC. The optimum reaction temperature was 300 ◦C, which conformed well with the in-situ DRIFTS results. The NOx conversion of SOC was only 27% at 300 ◦C, 1.0 NH3/NO ratio and 5000 h−<sup>1</sup> GHSV, and the NOx conversion of SSOC was 92% at the same condition. It was found that the denitration performance of SSOC was greatly improved and the optimum reaction temperature was 300 ◦C.

**Figure 9.** Effect of temperature on SCR activity on SOC and SSOC. 300 ppm NO, 15% O2, Ar as balance gas, NH3/NO ratio: 1.0, GHSV: 5000 h<sup>−</sup>1.

The stability test of SSOC was shown at Figure 10. The reaction continued for 24 h at 300 ◦C, 1.0 NH3/NO ratio, GHSV = 5000 h<sup>−</sup>1. It could be seen that the NOx conversion was stable at about 92%, which indicated that SSOC has a good denitration stability. Compared with the SOC, the denitration performance had been greatly improved after the acidification with sulfuric acid. The adsorption of NH3 and NO on SSOC was obviously improved, which subsequently promoted the NH3-SCR reaction.

**Figure 10.** The stability test of SSOC. 300 ppm NO, 15% O2, Ar as balance gas, NH3/NO ratio: 1.0, GHSV: 5000 h<sup>−</sup>1.

#### **3. Experimental Process**

#### *3.1. Catalyst Preparation*

In this experiment, the sintered ore was sampled from a sintering workshop in Wuhan. The sintered ore was dried, milled and sieved to 0.15–0.25 mm, which was denoted as SOC. The SSOC was prepared by using an impregnation method. Firstly, 100 g sintered ore (0.15–0.25 mm) was weighed and put into a beaker, then 50 mL sulfuric acid solution with a concentration of 5 mol/L was added and stirred simultaneously for 30 min. After filtration and washing with the deionized water, the mixture was dried at 105 ◦C and then calcined at 500 ◦C for 3 h in the air atmosphere. Finally, the prepared SSOC were naturally cooled to the room temperature, then crushed and sieved to 0.15–0.25 mm.

#### *3.2. Catalyst Characterization*

The main chemical composition of the SOC and SSOC were analyzed by X-ray fluorescence spectroscopy (XRF) (ARL SMS-XY, Thermo Fisher Scientific Corp., Waltham, MA, USA). The specific surface area, pore volume and pore size distribution of the catalysts were measured by an automated adsorption analyzer (Micromeritics ASAP 2020, Micromeritics Corp., Norcross, GA, USA). The catalysts samples were firstly degassed at 240 ◦C for 4 h before the test, and the adsorption medium was liquid nitrogen. The nitrogen adsorption and desorption were analyzed at −196 ◦C using BET analyzer (Micromeritics ASAP 2020, Micromeritics Corp., Norcross, GA, USA). The surface area was calculated by using the BET method according to nitrogen adsorption data in the relative pressure (P/P0) range of 0.01–1. X-ray diffraction (XRD; Rigaku RINT2000, Tokyo, Japan) was performed using CuKα radiation (λ = 1.54056 Å) to detect the crystalline phases of the samples. The analysis of XRD was referred to International Centre for Diffraction Data (ICDD). X-ray photoelectron spectroscopy (XPS, AXIS Ultra DLD, Shimazu Corp., Kyoto, Japan) was used to determine the valence states of the surface atoms of the catalysts with Al Kα radiation.

In-situ DRIFTS experiments were carried out in a FTIR spectrometer (FT-IR, Bruker Tensor II, Bruker optics Corp., Karlsruhe, Germany) equipped with an in-situ cell and a mercury-cadmium-telluride detector [34–37]. For the adsorption of NH3 (or NO + O2), the catalyst was exposed to a 20 mL/min NH3 (or NO + O2), which resulted in the variation with adsorption time of the DRIFT spectra, and argon purging was subsequently performed. In the reaction mechanism studies, the catalyst was pretreated in a flow of 20 mL/min NH3 for 40 min, then was shifted NO + O2 at 300 ◦C to get the DRIFT spectra. All spectra were recorded by accumulating 100 scans at a spectra resolution of 4 cm<sup>−</sup>1.

#### **4. Conclusions**

In this paper, SSOC displayed an excellent catalytic denitration activity. There were some sulfates that appeared in SSOC after acidification with sulfuric acid solution, which provided more Brønsted acid sites. In-situ DRIFTS results demonstrated that there were Lewis and Brønsted acid sites simultaneously on the surfaces of SOC and SSOC. The acidification contributed to the increase of the Brønsted acid sites at SSOC, which improved the adsorption capacity of NH3 and NO. NH3 and NO were adsorbed on the surface of catalysts to form amide species (–NH2), NH<sup>+</sup> <sup>4</sup> species, NO2 molecules in gaseous or weakly adsorbed state and nitrate species. Meanwhile, it could be seen that the NH3-SCR process of SOC and SSOC followed E-R and L-H mechanisms. Moreover, the optimum reaction temperature of catalysts was 300 ◦C and maximum NOx conversion fraction over SSOC was 92% at 1.0 NH3/NO ratio and 5000 h−<sup>1</sup> GHSV. It was observed that the NOx conversion could be steadily maintained.

**Author Contributions:** Data curation, F.H.; Investigation, Y.T.; Methodology, L.Q.; Project administration, J.H.; Software, B.Z.; Writing—original draft, W.C.; Writing—review & editing, F.Y.

**Funding:** The present work was partly supported by the National Natural Science Foundation of China (Grant No. 51476118 and 51576146).

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Promotional Effect of Cerium and/or Zirconium Doping on Cu/ZSM-5 Catalysts for Selective Catalytic Reduction of NO by NH3**

#### **Ye Liu, Chonglin Song \*, Gang Lv, Chenyang Fan and Xiaodong Li**

State Key Laboratory of Engines, Tianjin University, Tianjin 300072, China; liuyeskle@tju.edu.cn (Y.L.); lvg@tju.edu.cn (G.L.); fanchenyang@tju.edu.cn (C.F.); linahaifeng@tju.edu.cn (X.L.)

**\*** Correspondence: songchonglin@tju.edu.cn; Tel.: +86-22-27406840-8020

Received: 7 July 2018; Accepted: 24 July 2018; Published: 28 July 2018

**Abstract:** The cerium and/or zirconium-doped Cu/ZSM-5 catalysts (CuCe*x*Zr1−*x*O*y*/ZSM-5) were prepared by ion exchange and characterized by X-ray diffraction (XRD), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), temperature-programmed reduction by hydrogen (H2-TPR). Activities of the catalysts obtained on the selective catalytic reduction (SCR) of nitric oxide (NO) by ammonia were measured using temperature programmed reactions. Among all the catalysts tested, the CuCe0.75Zr0.25O*y*/ZSM-5 catalyst presented the highest catalytic activity for the removal of NO, corresponding to the broadest active window of 175–468 ◦C. The cerium and zirconium addition enhanced the activity of catalysts, and the cerium-rich catalysts exhibited more excellent SCR activities as compared to the zirconium-rich catalysts. XRD and TEM results indicated that zirconium additions improved the copper dispersion and prevented copper crystallization. According to XPS and H2-TPR analysis, copper species were enriched on the ZSM-5 grain surfaces, and part of the copper ions were incorporated into the zirconium and/or cerium lattice. The strong interaction between copper species and cerium/zirconium improved the redox abilities of catalysts. Furthermore, the introduction of zirconium abates N2O formation in the tested temperature range.

**Keywords:** selective catalytic reduction; nitric oxide; ammonia; Cu/ZSM-5; cerium; zirconium

#### **1. Introduction**

Metal-based ZSM-5 catalyst has received considerable attention due to its excellent performance in the low-temperature selective catalytic reduction (SCR) reaction [**????** ]. As a crystalline inorganic polymer, ZSM-5 consists of a three-dimensional network of SiO4 and AlO4 tetrahedra linked by interconnecting oxygen ions. When transition metal is introduced in ZSM-5, the catalytic performance can be enhanced through different types of active species, including isolated ions on the exchange sites within the ZSM-5 structure, metal oxide clusters, and metal oxide particles in the appearance of ZSM-5 [**? ?** ]. Among all the relevant catalysts studied, Cu/ZSM-5 is of particular interest because it exhibits excellent SCR activity in NO abatement [**? ?** ]. However, it has been reported that the catalytic activity of Cu/ZSM-5 catalysts is decreased, after an initial increase, with increasing copper content [**?** ], which suggests that it is difficult to improve catalytic performance by merely increasing the copper content.

Fabrication of copper catalysts can be implemented through incorporating more reducible promoters, such as ceria, lanthana, and zirconia [**???** ]. Cerium-containing materials are promising candidates for the application of NO abatement due to their high oxygen storage capacity and unique redox properties [**? ?** ]. Nevertheless, a major drawback of cerium lies in its poor thermal and hydrothermal stability, especially under a practical diesel exhaust atmosphere [**? ?** ]. It is well known that the incorporation of zirconia in cerium-based materials via high-temperature calcination can enhance the thermal stability and dispersion of the active component on the support surface [**? ?** ]. Furthermore, the addition of zirconia to ceria introduces structural defects through substitution of Ce4+ by Zr4+, which further enhances the oxygen storage capacity of ceria, the oxygen mobility in the lattice, the redox property, and thermal resistance [**???** ]. Consequently, cerium and zirconium materials have the potential to improve the catalytic activity of Cu/ZSM-5 catalyst. Based on this background, the present work attempts to address the effects of the cerium and/or zirconium addition on Cu/ZSM-5 catalysts for SCR of NO by NH3. A series of CuCe*x*Zr1−*x*O*y*/ZSM-5 catalysts (*x* = 0, 0.25, 0.50, 0.75 and 1) were synthesized using a conventional ion-exchange method, and the effects of adding cerium/zirconium metal ions into Cu/ZSM-5 catalysts on the SCR reaction were investigated using X-ray diffraction (XRD), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and temperature-programmed reduction by hydrogen (H2-TPR). The correlation between the structural characteristics, dispersion, reduction, and activity for the SCR process are discussed. The purpose of this work is to establish the relationships between structure and catalytic performance, which will be beneficial for the design and rationalization of the practical diesel catalysts.

#### **2. Results and Discussion**

#### *2.1. Structure and Morphology*

Figure **??** shows the XRD patterns of ZSM-5, Cu/ZSM-5, and CuCe*x*Zr1−*x*/ ZSM-5 (*x* = 0, 0.25, 0.5, 0.75 and 1). As expected, the catalysts still retain the ordered microstructure after the addition of copper, and cerium and/or zirconium because the inherent MFI structure of ZSM-5 (2*θ* = 7.8◦, 8.7◦, 24.5◦, 24.9◦, PDF 44-003) can be observed for all the catalysts. However, the incorporation of copper, and cerium and/or zirconium leads to a decrease of the intensity of the ZSM-5 principal diffraction peaks, which can be attributed to the higher absorption coefficient of metal compounds for X-ray radiation [**?** ]. Crystallized CuO (2*θ* = 35.5◦, 38.6◦, PDF = 02-1041) nanoparticles are detected for the Cu/ZSM-5 catalysts. For the CuZr1/ZSM-5, CuCe0.25Zr0.75/ZSM-5, CuCe0.5Zr0.5/ZSM-5, and CuCe0.75Zr0.25/ZSM-5 catalysts, no diffraction peaks for metal or metal oxide clusters were observed, indicating that the copper, zirconium, and cerium oxides are well dispersed as amorphous metal species, or aggregated into mini-crystals that are too small (<4 nm) to be detected by XRD [**?** ]. For CuCe1/ZSM-5, the observation of a small peak of CeO2 (2*θ* = 28.2◦, PDF = 34-0394) suggests that the extra-framework cerium is prone to agglomerating into cerium oxide clusters.

**Figure 1.** XRD patterns of CuCe*x*Zr1−*x*/ZSM-5 catalysts with different loading ratios: (a) ZSM-5, (b) Cu/ZSM-5, (c) CuZr1/ZSM-5, (d) CuCe0.25Zr0.75/ZSM-5, (e) CuCe0.5Zr0.5/ZSM-5, (f) CuCe0.75Zr0.25/ ZSM-5, and (g) CuCe1/ZSM-5.

To validate the XRD results, for each catalyst sample, more than 200 metal oxide particles from different TEM images were randomly chosen to determine the size distribution of the metal oxide particles. The size distributions of the metal oxide particles for CuCe0.25Zr0.75/ZSM-5, CuCe0.5Zr0.5/ZSM-5, CuCe0.75Zr0.25/ZSM-5, and CuCe1/ZSM-5 are shown in Figure **??**. The metal oxide particles in the d, e and f samples are less than 4 nm in size, while most metal oxide particles in the g sample are larger than 4 nm in size. These results are consistent with the XRD data.

**Figure 2.** Size distributions of metal oxide particles for (**d**) CuCe0.25Zr0.75/ZSM-5, (**e**) CuCe0.5Zr0.5/ ZSM5, (**f**) CuCe0.75Zr0.25/ZSM-5, and (**g**) CuCe1/ZSM-5.

The typical TEM images of pure ZSM-5 and CuCe*x*Zr1−*x*/ZSM-5 catalysts (*x =* 0, 0.5 and 1) are shown in Figure **??**. The typical interference fringes of the ZSM-5 crystal structure are found in Figure **??**a. Upon zirconium addition, small aggregates are observed for the CuZr1/ZSM-5 catalyst (see Figure **??**b), and most of them are less than 2 nm in size. The subsequent energy-dispersive X-ray (EDX) analysis for Point 1 in Figure **??**b indicates a copper, zirconium, and oxygen-rich phase for these small aggregates. With increasing cerium content, the sizes of aggregates become larger, as shown in Figure **??**c for CuCe0.5Zr0.5/ZSM-5 and Figure **??**d for CuCe1/ZSM-5. These results reveal that the addition of cerium decreases the dispersion of metal ions. The EDX analysis for Point 2 in Figure **??**d indicates copper and cerium enrichment for the large black aggregates.

**Figure 3.** TEM images and energy-dispersive X-ray (EDX) quantitative analyses for (**a**) ZSM-5, (**b**) CuZr1/ZSM-5, (**c**) CuCe0.5Zr0.5/ZSM-5, and (**d**) CuCe1/ZSM-5.

#### *2.2. XPS Analysis*

The chemical state and surface composition of the elements in the catalysts were characterized by XPS. The Cu 2p spectra of the CuCe*x*Zr1−*x*/ZSM-5 catalysts (*x =* 0, 0.5 and 1) in Figure **??**a show two main peaks: a peak at 933.3 eV that can be attributed to Cu 2p3/2, and a peak at about 953 eV that can be attributed to Cu 2p1/2. The intense satellite peak located at approximately 943 eV confirms the existence of divalent copper. Peak deconvolution and fitting to experimental data indicate that the Cu 2p3/2 peak can be well fitted by two peaks at 932.5 and 933.6 eV, corresponding to the Cu+ and Cu2+ ions, respectively [**???** ].

The XPS spectra of O 1s for the CuCe*x*Zr1−*x*/ZSM-5 catalysts (*x =* 0, 0.5, and 1) in Figure **??**b show two primary peaks. The peak at a higher binding energy (BE) of 531.9 eV can be assigned to regular lattice oxygen from the ZSM-5 zeolite structure [**???** ], while the shoulder peak at about 529.6 eV corresponds to characteristic lattice oxygen bound to metal (copper, cerium and zirconium) cations. As the cerium content increases by 6.5% in CuCe1/ZSM-5, the shoulder peak corresponding to the lattice oxygen slightly shifts to a lower BE value (529.3 eV). Considering the findings from XRD that no metal oxide crystals are observed except for the CuCe1/ZSM-5 catalyst, the existence of lattice oxygen implies that the oxides are well dispersed on the ZSM-5 support as the mini-crystal form.

The XPS doublet spectra of Zr 3d obtained from the CuCe*x*Zr1−*x*/ZSM-5 catalysts (*x =* 0 and 0.5) are shown in Figure **??**c. The peaks centered at 181.9 and 184.2 eV correspond to Zr 3d5/2 and Zr 3d3/2, respectively. The BE of Zr 3d5/2 is about 0.9 eV, higher than that reported for zirconium metal and 0.5 eV lower than that reported for zirconia [**?** ]. This phenomenon probably originates from the contribution of electrons from the surrounding species. Wang et al. [**?** ] also found that the zirconium cations in Cu/ZrO2 catalysts possess a lower BE of Zr 3d5/2 than zirconia, and they ascribed it to copper oxides located near oxygen vacancies on the exterior of the zirconia.

Figure **??**d shows the XPS spectra of Ce 3d for the CuCe*x*Zr1−*x*/ZSM-5 catalysts (*x =* 0.5 and 1). The complex spectrum of Ce 3d with respect to Ce 3d3/2 and Ce 3d5/2 ionization features are deconvoluted into eight components. The peaks at about 882.5, 889.3, and 898.5 eV are ascribed to the peak of Ce 3d5/2 and two "shake-up" satellite peaks, respectively. The peaks at approximately 901.1, 908.1, and 916.6 eV can be assigned to Ce 3d3/2 and two shake-up satellite peaks, respectively [**? ?** ]. These features are viewed as the fingerprints for the existence of Ce4+. The inconspicuous peaks at about 885.5 and 904.3 eV represent the initial state of Ce3+ [**? ?** ]. It is evident from Figure **??**d that cerium is mostly in a quadrivalent oxidation state, and a small quantity of Ce3+ co-exists. The presence of Ce3+ can be attributed to the relative homogeneous Ce*x*Zr1−*<sup>x</sup>*O2 or the substitution of Ce4+ by Zr4+ ions [**?** ].

XUH D **Figure 4.** X-ray photoelectron spectra for (**a**) Cu 2p, (**b**) O 1s, (**c**) Zr 3d, and (**d**) Ce 3d over (1) CuZr1/ ZSM-5, (2) CuCe0.5Zr0.5/ZSM-5 and (3) CuCe1/ZSM-5.

For the catalysts, the atomic ratios of Cu/Si, Ce/Si, and Zr/Si obtained from AAS and XPS are listed in Table **??**. The atomic ratios determined from XPS are larger than those obtained from AAS, suggesting that the metal oxides become enriched on the surface of ZSM-5 grains. The atomic ratios of Cu+/Cu2+ and Ce3+/Ce4+ calculated from XPS spectra are also listed in Table **??**. As the cerium content increases from 0% for CuZr1/ZSM-5 to 3.4% for CuCe0.5Zr0.5/ZSM-5, the Cu+/Cu2+ ratio decreases from 0.64 to 0.35, while the Ce3+/Ce4+ ratio increases from 0 to 0.198. Upon further increasing the cerium content, the Cu+/Cu2+ ratio increases to 0.71 for the CuCe1/ZSM-5 catalyst, while the Ce3+/Ce4+ ratio decreases after an initial increase.


**Table 1.** Surface compositions of CuCe*x*Zr1−*x*/ZSM-5 catalysts (*x =* 0, 0.25, 0.5, 0.75, 1) obtained from X-ray photoelectron spectroscopy (XPS) analysis.

#### *2.3. H2-TPR*

Figure **??** shows the H2-TPR profiles of ZrO2, CeO2, and the Cu/ZSM-5 and CuCe0.75Zr0.25/ZSM-5 catalysts, and the relevant data are listed in Table **??**. Three main reduction peaks are evident for the catalysts: the α-peak (182–213 ◦C) is ascribed to the reduction of the copper species dispersed on the ZSM-5 support, the β-peak (260–264 ◦C) is associated with the reduction of the copper oxide adhering to the external surface of zeolite crystallites, and the γ-peak (406 ◦C) is generally considered to be due to the reduction of bulk and unsupported copper oxide. Some of the copper species have incorporated into the vacant sites of cerium and/or zirconium oxides to form a coordinated surface structure with capping oxygen [**?** ], corresponding to the δ peak (164–178 ◦C) and the ε peak (315–342 ◦C). As the cerium content increases, the δ peak shifts from 171 ◦C for CuZr1/ZSM-5 to 164 ◦C for CuCe1/ZSM-5, indicating that the composite oxides of copper and cerium are more readily reduced than those of copper and zirconium.

**Figure 5.** Temperature-programmed reduction by hydrogen (H2-TPR) profiles of (a) ZrO2, (b) CeO2, (c) Cu/ZSM-5 and (d) CuCe0.75Zr0.25/ZSM-5.


**2.** H2-TPR results from Cu/ZSM-5 and CuCe*x*Zr1−*x*/ZSM-5 catalysts (*<sup>x</sup> =* 0, 0.25, 0.5, 0.75, 1).

**Table** 

#### *Catalysts* **2018** , *8*, 306

From the H2-TPR quantitative data in Table **??**, it can be seen that as the cerium content increases from 0% for CuZr1/ZSM-5 to 5.2% for CuCe0.75Zr0.25/ZSM-5, the hydrogen consumption for the δ and α peak monotonously increases, while the hydrogen consumption for the β and γ peak gradually decreases. As mentioned above, the δ and α peak are related to highly dispersed copper species. Therefore, this increase in hydrogen consumption for the δ and α peak indicates that the addition of cerium improves the copper dispersion. For the CuCe1/ZSM-5 catalyst, although the reduction temperatures of the δ and α peaks are lower than those for the other catalysts, the H2 consumption of the δ and α peaks is not the maximum. This can be attributed to the formation of ceria, which is evidenced by the above XRD and TEM results. The ceria formed over the catalyst leads to the reduction not only extending deeply into the bulk of crystalline ceria but also being confined to its surface, so that the CuCe1/ZSM-5 catalyst may consume less hydrogen than the CuCe0.75Zr0.25/ZSM-5 catalyst. Moreover, because the total H2 consumption shows an increase with the cerium content increasing from 0% (CuZr1/ZSM-5) to 5.2% (CuCe0.75Zr0.25/ZSM-5), the hydrogen uptake can be attributed not only to the copper reduction but also to a partial reduction of the cerium surface for the cerium-containing catalysts.

#### *2.4. Catalytic Activity Test*

Figure **??** shows the NO conversion in the SCR reaction on all the catalysts from 50 to 600 ◦C. All of the catalysts exhibit excellent SCR activity. For the Cu/ZSM-5 catalyst, the temperature range for 95% NO conversion is 209–406 ◦C. After the addition of cerium and/or zirconium to Cu/ZSM-5, the temperature range for 95% NO conversion extends toward both lower and higher temperatures, corresponding to active window broadening. The temperature ranges for 95% NO conversion are 190–436 ◦C for the CuZr1/ZSM-5, 202–456 ◦C for CuCe0.25Zr0.75/ZSM-5, 203–460 ◦C for CuCe0.5Zr0.5/ZSM-5, 175–468 ◦C for CuCe0.75Zr0.25/ZSM-5, and 179–435 ◦C for CuCe1/ZSM-5. Among the catalysts tested, the CuCe0.75Zr0.25/ZSM-5 catalyst has the highest catalytic activity for NO conversion. The improvements in the SCR activity after the addition of cerium and/or zirconium can be accounted for by three factors:


Furthermore, as the temperature increases from 406 to 600 ◦C, a clear decrease of NO conversion for all the tested catalysts is observed in Figure **??**. This decrease is mainly due to the occurrence of non-selective NH3 oxidation by the reaction 4NH3 + 5O2 → 4NO + 6H2O at high temperatures [**???** ]. The consumption of NH3, together with the production of new NO, limits NO conversion.

The SCR process induced by all the catalysts inevitably produces N2O and NO2 as byproducts. In the whole temperature range considered in this study, the NO2 concentration is below 3 ppm for all the catalysts (not shown). Figure **??** shows the yield of N2O plotted against reaction temperature. For all the catalysts, the yield of N2O increases with increasing temperature up until a certain temperature and then starts to decrease. The N2O produced in the low temperature range mainly arises from side reactions between NO and NH3: 4NO + 4NH3 + 3O2 → 4N2O + 6H2O [**? ?** ]. Upon further increase of the reaction temperature, the formation of N2O is mainly from NH3 oxidation via the

reaction 2NH3 + 2O2 → N2O + 3H2O [**? ?** ], and the amount of N2O rapidly increases with the maximum yield at reaction temperatures <450 ◦C. In the high temperature range, non-selective NH3 oxidation takes place by the reaction 4NH3 + 5O2 → 4NO + 6H2O [**???** ], and thus the amount of N2O rapidly decreases as the temperature increases. Moreover, close inspection of Figure **??** shows that the zirconium-containing catalysts yield less N2O than the Cu/ZSM-5 and CuCe1/ZSM-5 catalysts, which means that the introduction of zirconium abates N2O formation in the tested temperature range.

**Figure 6.** Catalytic activities for NO*<sup>x</sup>* reduction by NH3 for Cu/ZSM-5 and CuCexZr1−*x*/ZSM-5 catalysts (*x =* 0, 0.25, 0.5, 0.75 and 1).

**Figure 7.** Yield of N2O during the selective catalytic reduction (SCR) process as a function of reaction temperature for Cu/ZSM-5 andCuCe*x*Zr1−*x*/ZSM-5 catalysts (*x* = 0, 0.25, 0.5, 0.75 and 1).

#### **3. Experimental**

#### *3.1. Synthesis of Catalysts*

H/ZSM-5 with an atomic Si/Al ratio of 25 was provided by Nankai University, Tianjin, China. A series of CuCe*x*Zr1−*x*O*y*/ZSM-5 catalysts (*x =* 0, 0.25, 0.5, 0.75, and 1) were synthesized by an aqueous ion-exchange technique. A desired amount of copper acetate, and zirconium and cerium nitrate was added to deionized water and mixed with 20 g of H/ZSM-5 powder at room temperature. The resulting solution was stirred at 80 ◦C for 24 h. The samples were filtered and dried by evaporation in air and then calcined at 550 ◦C for 4 h. The copper content of the CuCe*x*Zr1−*x*O*y*/ZSM-5 catalysts was fixed at 3 wt. %, and the molar ratio of Cu/(Ce + Zr) was 1:1. These catalysts are denoted as CuZr1/ZSM-5, CuCe0.25Zr0.75/ZSM-5, CuCe0.5Zr0.5/ZSM-5, CuCe0.75Zr0.25/ZSM-5, and CuCe1/ZSM-5. The copper,

cerium, and zirconium contents of each calcined catalyst were determined by atomic absorption spectroscopy (AAS) using a PerkinElmer AAnalyst 300 spectrometer, and the results are shown in Table **??**. To evaluate the catalytic activity, the catalysts were compressed by a pressure of 20 MPa, then granulated and screened to the size of a 20–40 mesh.


**Table 3.** Element compositions of all catalysts.

#### *3.2. Characterization*

Powder X-ray diffraction (XRD) was performed using a Rigaku D/MAC/max 2500 v/pc instrument with Cu Kα radiation (40 kV, 200 mA, λ = 1.5418 Å) (Jananese Science, Tokyo, Japan). The scan was performed with a 2*θ* rate of 0.02◦/min from 5–80◦. Transmission electron microscopy (TEM) images of the catalysts were obtained using a Philips Tecnai G2 F20 microscope operating at 200 kV equipped with an Oxford-1NCA energy-dispersive X-ray detector (EDX) (FEI, Hillsboro, OR, USA). Prior to measurement, the catalysts were dispersed in pure ethanol through sonication, and then mounted on nickel grids with a carbon film. X-ray photoelectron spectroscopy (XPS) was recorded using a Perkin-Elmer PHI-1600 ESCA spectrometer with a Mg Kα X-ray source (Perkin Elmer, Wellesley, MA, USA). The binding energies (BEs) were calibrated by referencing the C 1s at 284.8 eV. A ChemBet Pulsar system was used to perform temperature-programmed reduction by hydrogen (H2-TPR) (Quantachrome Instruments, Boynton Beach, FL, USA). For the H2-TPR experiments, 100 mg of the catalyst was exposed to pure Ar at a flow rate of 30 mL/min for 1 h at 300 ◦C. Then, the same gas was used to cool the catalyst down to 50 ◦C. The H2-TPR measurements were carried out by increasing the temperature to 700 ◦C at a heating rate of 10 ◦C/min in 5% hydrogen at a constant flow rate of 30 mL/min. The hydrogen consumption was monitored and quantified using a thermal conductivity detector.

#### *3.3. Catalytic Activity*

The activities of the catalysts were measured in a continuous flow apparatus at atmosphere pressure. Before each test run, the catalyst powder was first pressed into a wafer and sieved into 20–40 meshes, and then 0.5 g of catalyst was set into a fix-bed reactor made of a quartz tube with an internal diameter of 10 mm. The reaction was carried out in the temperature range 50–600 ◦C, and a K-type thermocouple was located inside the catalyst bed to monitor the reaction temperature. The feed gas was controlled by calibrated electronic mass flow controllers, and contained 1000 ppm NO, 1000 ppm NH3, and 10% O2 and N2 as the balance gas. The space velocity was set at 15,000 h<sup>−</sup>1. An online mass spectrometry (Dycor LC-D200) was used to monitor the effluent NO, NO2, N2O, and NH3. From the concentration of the gases at steady state, the NOx conversion was defined as:

$$\text{NO}\_{\text{x}}\text{ conversion} \left( \% \right) \ = \frac{\left[ \text{NO}\_{\text{x}} \right]\_{\text{in}} - \left[ \text{NO}\_{\text{x}} \right]\_{\text{out}}}{\left[ \text{NO}\_{\text{x}} \right]\_{\text{in}}} \times 100 \text{, } \left[ \text{NOx} \right] \ = \left[ \text{NO} \right] + \left[ \text{NO} \right] \ $$

#### **4. Conclusions**

The present study has highlighted the effects of cerium and/or zirconium incorporation into Cu/ZSM-5 on the SCR activity in NO abatement. The CuCe*x*Zr1−*x*O*y*/ZSM-5 catalysts used in this study present a more than 95% NO conversion rate in a wide temperature range (175–468 ◦C), which is an improvement relative to the Cu/ZSM-5 catalyst (209–406 ◦C) in terms of SCR activity. The zirconium doping improves copper dispersion, and the introduction of zirconium promotes surface copper enrichment and prevents copper crystallization, which favors catalytic NO reduction. Partial substitution of cerium and zirconium ions by copper ions increases the reactive lattice oxygen content and reducibility, and thus improves the low-temperature activity. Because of the existence of the Ce3+/Ce4+ redox couple in the cerium-containing catalysts, the SCR activities of cerium-rich catalysts are higher than those of zirconium-rich catalysts. Moreover, formation of N2O is suppressed in the range of test temperatures due to the presence of zirconium in the catalysts.

**Author Contributions:** This study was conducted through contributions of all authors. C.S. and G.L. designed the study and wrote the manuscript. Y.L., C.F. and X.L. designed and performed the experiments and analyzed the data. Y.L. checked and corrected the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This study was supported by the National Natural Science Foundation of China (No. 51476116).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **A CeO2/ZrO2-TiO2 Catalyst for the Selective Catalytic Reduction of NO***<sup>x</sup>* **with NH3**

### **Wenpo Shan 1,2, Yang Geng 3, Yan Zhang 1,2, Zhihua Lian <sup>1</sup> and Hong He 1,2,4,\***


Received: 30 September 2018; Accepted: 27 November 2018; Published: 30 November 2018

**Abstract:** In this study, CeZr0.5TiaO*<sup>x</sup>* (with a = 0, 1, 2, 5, 10) catalysts were prepared by a stepwise precipitation approach for the selective catalytic reduction of NO*<sup>x</sup>* with NH3. When Ti was added, all of the Ce-Zr-Ti oxide catalysts showed much better catalytic performances than the CeZr0.5O*x*. Particularly, the CeZr0.5Ti2O*<sup>x</sup>* catalyst showed excellent activity for broad temperature range under high space velocity condition. Through the control of pH value and precipitation time during preparation, the function of the CeZr0.5Ti2O*<sup>x</sup>* catalyst could be controlled and the structure with highly dispersed CeO2 (with redox functions) on the surface of ZrO2-TiO2 (with acidic functions) could be obtained. Characterizations revealed that the superior catalytic performance of the catalyst is associated with its outstanding redox properties and adsorption/activation functions for the reactants.

**Keywords:** Ce-based catalyst; stepwise precipitation; selective catalytic reduction; diesel exhaust; nitrogen oxides abatement

#### **1. Introduction**

NO*<sup>x</sup>* (mainly NO and NO2) in the atmosphere plays critical roles in the formation of severe air pollution problems, such as haze, acid rain, and photochemical smog. In the last few decades, great efforts have been devoted to the development of NO*<sup>x</sup>* emission control technologies [1–3]. Selective catalytic reduction of NO*<sup>x</sup>* with NH3 (NH3-SCR) has been widely applied for the removal of NO*<sup>x</sup>* generated from stationary sources for many years, and it has also been used for the control of NO*<sup>x</sup>* emission from diesel vehicles [2,4].

Catalysts play an important role in the development of NH3-SCR technology [5,6]. Vanadium-based catalyst (especially V2O5-WO3/TiO2), with excellent SO2 resistance, is the most widely used NH3-SCR catalyst for NO*<sup>x</sup>* emission control from power plants, and it was also applied on diesel vehicles as the first generation of SCR catalyst [4]. However, this catalyst system still has some problems, including the toxicity of active V2O5, narrow temperature window, and low thermal stability [2].

There has been strong interest in developing a vanadium-free catalyst that can be used on diesel vehicles [5–11]. Ce is a key component in three-way catalysts for emission control in automobiles for gasoline. CeO2 provides an oxygen storage function through redox cycling between Ce3+ and Ce4+. In recent years, Ce has also attracted great attention for applications as a support [12,13], promoter [14–18], or main active component [19–26] for NH3-SCR catalysts.

Pure Ce oxide is not suitable for use as an NH3-SCR catalyst [27,28]. When Zr oxide was introduced into Ce oxide, the thermal stability and the oxygen storage capacity of the oxide could be significantly improved. Therefore, Ce-Zr oxide was investigated for NH3-SCR [12,13,29–34]. In the NH3-SCR reaction, both redox functions and acidic functions of the catalyst are needed [4,35]. Therefore, a high dispersion of active sites and close coupling of redox with acid sites is the way to design a highly efficient NH3-SCR catalyst.

In this study, starting from a preparation of Ce-Zr oxide by the co-preparation method, we developed a Ce-Zr-Ti oxide catalyst using a stepwise precipitation approach, under the theoretical guidance of the close combination of the Ce-Zr oxide with strong redox functions and Ti oxide with excellent acid properties [4,5]. This obtained catalyst showed superior catalytic performance for NH3-SCR.

#### **2. Results and Discussion**

#### *2.1. NH3-SCR Activity*

Figure 1A presents the NO*<sup>x</sup>* conversion over the catalysts with different Ti contents under a relatively high gas hourly space velocity (GHSV) of 200,000 h−1. The CeZr0.5O*<sup>x</sup>* just exhibited over 50% NO*<sup>x</sup>* conversion in a narrow temperature range of 350–425 ◦C. When Ti was introduced, all of the Ce-Zr-Ti oxide catalysts exhibited much better activities. With the increase in Ti content, the low temperature firstly increased and then decreased. As a result, the CeZr0.5Ti2O*<sup>x</sup>* catalyst presented the best activity in a low temperature range, together with a high NO*x* conversion in a wide temperature range. On the other hand, the variation in high temperature activity with Ti content was contrary to that of low temperature activity, with the activity of CeZr0.5Ti2O*<sup>x</sup>* slightly lower than those of the other Ce-Zr-Ti oxide catalysts in a high temperature range. In addition, adding Ti to the catalyst also enhanced the N2 selectivity, and the Ce-Zr-Ti oxide catalysts all presented higher N2 selectivity than CeZr0.5O*<sup>x</sup>* (Figure 1B).

**Figure 1.** (**A**) NO*<sup>x</sup>* conversions and (**B**) N2 selectivity over the CeZr0.5O*<sup>x</sup>* and Ce-Zr-Ti oxide catalysts. Reaction conditions: [NO] = [NH3] = 500 ppm, [O2] = 5 vol.%, N2 balance, and GHSV = 200,000 h<sup>−</sup>1.

The influences of H2O and space velocity on the NO*<sup>x</sup>* conversion over CeZr0.5Ti2O*<sup>x</sup>* were tested and the results are shown in Figure 2. The existence of 5% H2O in the flow gas decreased the low temperature activity, but enhanced the high temperature activity. As a result, over 80% NO*<sup>x</sup>* conversion could still be achieved from 250 to 450 ◦C. When the GHSV was decreased from 200,000 h−<sup>1</sup> to 100,000 h<sup>−</sup>1, the activity of the catalyst at low temperatures was obviously improved.

**Figure 2.** NO*<sup>x</sup>* conversion over CeZr0.5Ti2O*<sup>x</sup>* catalyst under different reaction conditions. Reaction conditions: [NO] = [NH3] = 500 ppm, [O2] = 5 vol.%, [H2O] = 5 vol.% (when used), N2 balance, and GHSV = 100,000 or 200,000 h<sup>−</sup>1.

#### *2.2. Separated NO/NH3 Oxidation*

To analyze the effects of Ti on the catalyst, separated NO oxidation and NH3 oxidation tests were carried out for the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* (Figure 3). The NO2 production during NO oxidation over the CeZr0.5Ti2O*<sup>x</sup>* was clearly higher than that over CeZr0.5O*<sup>x</sup>* at a low temperature. Since the presentation of NO2 in the reaction gas could promote the SCR reaction at a low temperature by accelerating the fast SCR process (2 NH3 + NO + NO2 → 2N2 + 3H2O), the enhanced low-temperature activity by the introduction of Ti should be associated with the promoted oxidation of NO to NO2 over CeZr0.5Ti2O*<sup>x</sup>* [10,35]. In addition, the introduction of Ti also promoted NH3 oxidation over the catalyst at a high temperature. The NH3-SCR reaction route at a high temperature mainly follows the Eley-Rideal mechanism, and the activation of NH3 to form NH2 species by oxidation plays the key role for the reaction with NO to form N2 and H2O, owing to NH2 + NO(g) → N2 + H2O. Therefore, promoted NH3 oxidation would be beneficial for the improvement of high temperature activity.

**Figure 3.** (**A**) NO2 productions during separate NO oxidation reaction and (**B**) NH3 conversions during separate NH3 oxidation reaction over the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts. Reaction conditions: (A) [NO] \_ENREF\_30= 500 ppm, (B) [NH3] = 500 ppm, [O2] = 5 vol.%, N2 balance and GHSV = 200,000 h<sup>−</sup>1.

#### *2.3. XRD*

The X-ray diffraction (XRD) results of the CeZr0.5O*<sup>x</sup>* and Ce-Z-Ti oxide catalysts are presented in Figure 4. Both CeO2 and ZrO2 were detected in CeZr0.5O*x*. With the increase of Ti, the peaks for CeO2 and ZrO2 became more and more weak, and only anatase TiO2 was observed for CeZr0.5Ti10O*x*. Only weak peaks for CeO2 with cubic fluorite structures (PDF# 43-1002) were observed in the CeZr0.5Ti2O*x*, indicating that the introduction of Ti had induced the structural change of the CeZr0.5O*x*, and the crystallizations of Ce, Zr and Ti oxides in CeZr0.5Ti2O*<sup>x</sup>* were significantly inhibited. As a result, the CeZr0.5Ti2O*<sup>x</sup>* (165.1 m2/g) showed a higher Brunauer–Emmett–Teller (BET) surface area than CeZr0.5O*<sup>x</sup>* (113.5 m2/g).

**Figure 4.** XRD patterns of the CeZr0.5O*<sup>x</sup>* and Ce-Z-Ti oxide catalysts.

#### *2.4. H2-TPR*

The H2 temperature-programmed reduction (H2-TPR) profiles of CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* are presented in Figure 5. The CeZr0.5O*<sup>x</sup>* exhibited two peaks at 496 and 755 ◦C due to the surface and bulk reductions of CeO2 (as detected by XRD), respectively [31,36–38]. During the test, coordinatively unsaturated surface oxygen anions are easily reduced by H2 in the low temperature region, while the bulk oxygen species are reduced only after the transportation to the surface [39]. With the addition of Ti, a sharp H2 consumption peak appeared at 567 ◦C, which indicates that another type of Ce species might be formed. Considering the XRD results, this sharp peak might be associated with the reduction of the highly dispersed Ce species from Ce4+ to Ce3+ [22,34]. In addition, the H2 consumption of CeZr0.5Ti2O*<sup>x</sup>* was much higher than that of CeZr0.5O*<sup>x</sup>* at a low temperature. The H2-TPR results clearly indicated the enhancement of redox functions for CeZr0.5Ti2O*x*.

**Figure 5.** H2-TPR profiles of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts.

Previous studies have indicated that the redox properties of NH3-SCR catalyst play a dominant role in the low temperature activity [35,40,41]. Therefore, the enhanced redox function of CeZr0.5Ti2O*<sup>x</sup>* would beneficial for low temperature activity.

#### *2.5. NOx/NH3-TPD*

To investigate the NO*<sup>x</sup>* and NH3 adsorption/desorption properties of CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*x*, NO*<sup>x</sup>* temperature-programmed desorption (NO*x*-TPD) and NH3 temperature-programmed desorption (NH3-TPD) were performed for the catalysts (Figure 6).

**Figure 6.** (**A**) NO*x*-TPD and (**B**) NH3-TPD profiles of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts.

The NO*x*-TPD profiles are presented in Figure 6A. The first NO*<sup>x</sup>* peak of CeZr0.5Ti2O*<sup>x</sup>* was at ca. 110 ◦C, mainly due to the desorption of physisorbed NO*x*, while the other NO*x* peak was at ca. 300 ◦C and was associated with the decomposition of chemsorbed NO*x* species [42,43]. On the other hand, two weak peaks were observed for CeZr0.5O*<sup>x</sup>* at ca. 270 ◦C and ca. 410 ◦C, respectively, which were due to the decomposition of different types of chemsorbed NO*x* species. With the addition of Ti, the adsorbed NO*<sup>x</sup>* on CeZr0.5Ti2O*<sup>x</sup>* was obviously more than that of CeZr0.5O*x*. Particularly, the desorbed NO2 of CeZr0.5Ti2O*<sup>x</sup>* was much higher, owing to the enhanced low-temperature activity for NO oxidation (as shown by the separated NO oxidation results), which could facilitate the conversion of NO*<sup>x</sup>* in NH3-SCR.

Surface acidity plays a dominate role in the high-temperature SCR activity due to its effects on the adsorption and activation of NH3 [35,41]. Previous studies have revealed that Ti species of NH3-SCR catalysts mainly act as acid sites in the reaction for NH3 adsorption [4]. Therefore, the adsorbed NH3 of CeZr0.5Ti2O*<sup>x</sup>* was much more than that of CeZr0.5O*x*, which might be an important reason for the better NH3-SCR activity of CeZr0.5Ti2O*<sup>x</sup>* at high temperatures.

#### *2.6. XPS*

The X-ray photoelectron spectroscopy (XPS) results for Ce 3d of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* are shown in Figure 7. The sub-bands labeled with u'/v' and u0/v0 represent the 3d104f1 initial electronic state corresponding to Ce3+ and the 3d94f2 state of Ce3+, respectively [44]. The sub-bands labeled with u"' and v"' represent the 3d104f0 state of Ce4+, and the sub-bands labeled with u, u", v and v" represent the 3d94f1 state corresponding to Ce4+ [44]. The presence of Ce3+ would induce a charge imbalance, which could lead to unsaturated chemical bonds and oxygen vacancies. The calculated Ce3+ ratio of CeZr0.5Ti2O*<sup>x</sup>* (36.0%) was higher than that of CeZr0.5O*<sup>x</sup>* (33.8%), indicating that more surface oxygen vacancies presented in CeZr0.5Ti2O*x*. In addition, the Ce3+ ratio of the catalyst could influence the redox ability and reactant adsorption and activation functions, and thereby contribute to NH3-SCR performance.

**Figure 7.** XPS results of Ce 3d of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts.

The surface oxygen vacancies of the catalysts might generate weakly-adsorbed oxygen species or additional chemisorbed oxygen on the surface of the catalyst [27,45]. The XPS results of O 1s of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* are shown in Figure 8. The O 1s peak was fit into two sub-bands. The sub-bands at 531.2–531.5 eV and 529.1–529.6 eV were assigned to the surface adsorbed oxygen (Oα), such as the O2 <sup>2</sup><sup>−</sup> and O<sup>−</sup> belonging to defect-oxide or a hydroxyl-like group, and the lattice oxygen O2<sup>−</sup> (Oβ), respectively [46]. The O<sup>α</sup> ratios of the catalysts were calculated by Oα/(O<sup>α</sup> + Oβ), and the CeZr0.5Ti2O*<sup>x</sup>* showed higher O<sup>α</sup> ratio than CeZr0.5O*x*. The results confirmed that the addition of Ti indeed induced more surface-adsorbed oxygen, which would facilitate NO oxidation to NO2 (as shown by the separated NO oxidation and NO*x*-TPD results), and thus facilitates the conversion of NO by fast SCR effects.

#### *2.7. Formation Process Analysis of the CeZr0.5Ti2Ox Catalyst*

Figure 9 shows the pH variations of the mixed solutions for the preparation of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts. During the preparation of CeZr0.5O*x*, the initial pH value of the solution was 1.6. With the hydrolysis of urea, the pH increased gradually to be 7.6 after heating for 12 h. Due to the increase in pH, suspended particles began to appear in the solution in the second hour. The particles with the precipitation time of 2 h, 4 h, 6 h, and 12 h were collected and then calcined to be catalyst samples. The activity tests of these samples showed similar NO*x* conversions with each other.

**Figure 8.** XPS results of O 1s of the CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts.

**Figure 9.** The pH variation of the mixed solution during the preparation of the (**A**) CeZr0.5O*<sup>x</sup>* and (**B**) CeZr0.5Ti2O*<sup>x</sup>* catalysts, and the NO*<sup>x</sup>* conversions of the obtained samples at different precipitation time.

Due to the acidity induced by the added Ti(SO4)2, the initial pH value of the mixed solution during the preparation of CeZr0.5Ti2O*<sup>x</sup>* dropped to be 1.1. With the hydrolysis of urea, the pH increased gradually after heating, and some white particles generated in the first hour and suspended in the solution. With the increase of time, the particles gradually turned yellow. The pH reached ca. 7.0 after 12 h of reaction. The particles with the precipitation times of 1 h, 4 h, 6 h, and 12 h were collected and then calcined to be catalyst samples. Interestingly, the activity test showed a remarkable enhancement of NO*x* conversions for the four samples with the increase in precipitation time.

The surface metal atomic concentrations of the CeZr0.5Ti2O*<sup>x</sup>* samples with different precipitation times were analyzed using XPS, and the variations in Ce, Zr, and Ti concentrations with precipitation time are shown in Figure 10. For the 1-h precipitation sample, only Ti and Zr, without Ce, were detected. With the increase in precipitation time, surface Ce concentration increased gradually in the samples. At the same time, Ti and Zr concentrations gradually decreased with the increase in precipitation time. A TEM-EDS mapping image showed that Ce was highly dispersed in the CeZr0.5Ti2O*<sup>x</sup>* catalyst (Figure 11).

**Figure 10.** Surface metal atomic concentrations of the CeZr0.5Ti2O*<sup>x</sup>* samples with different precipitation times.

**Figure 11.** TEM image (A) and the corresponding EDS mapping (B) for the Ce of the CeZr0.5Ti2O*<sup>x</sup>* catalyst.

Considering the variations in the solution pH value when preparing the CeZr0.5Ti2O*x*, the formation process of the catalyst can be proposed as follows: The Ti and Zr species were first co-precipitated with the increase in solution pH. Then, the Ce species uniformly precipitated onto the precipitated Zr-Ti species with the further increase in pH. Finally, a CeZr0.5Ti2O*<sup>x</sup>* catalyst with a higher surface Ce concentration than Ti and Zr was obtained. Through control of the hydrolysis of urea, the variations in the solution pH can be controlled, and then we can control the precipitation process of

the catalyst, which is very important for the formation of highly-dispersed CeO2 on ZrO2-TiO2. Thus, the obtained catalyst can present excellent NH3-SCR performance.

#### **3. Experimental Section**

#### *3.1. Catalyst Preparation and Activity Test*

The CeZr0.5TiaO*<sup>x</sup>* (a = Ti/Ce molar ratio = 0, 1, 2, 5, 10), with a Zr/Ce molar ratio fixed to be 0.5, was prepared using a precipitation method. Desired precursors of Ce(NO3)3·6H2O (>99%, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China), Zr(NO3)4·5H2O (>99%, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) and Ti(SO4)2 (>98%, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) were dissolved together in distilled water, and urea (>99%, Sinopharm Chemical Reagent Co., Ltd., Shanghai, China) was added to the mixed solution as a slowly-releasing precipitator. Then, the solution was heated to 90 ◦C to facilitate the release of NH3 and thereby raise the pH value gradually. The temperature of the mixed solution was held at 90 ◦C for 12 h under vigorous stirring (some samples with shorter precipitation times were also prepared). After that, the precipitated powders were collected via filtration, washed using distilled water, and dried for 12 h at 100 ◦C. Finally, the catalyst was obtained after calcination at 500 ◦C for 5 h.

The SCR activity of the catalysts (40–60 mesh) were tested in a fixed-bed quartz flow reactor. The reaction conditions were controlled as follows: 500 ppm NO, 500 ppm NH3, 5 vol.% O2, N2 balance, and 400 mL/min total flow rate. Different gas hourly space velocities (GHSVs) were obtained by changing the volume of catalysts, i.e., 0.24 mL catalyst for a GHSV = 100,000 h−<sup>1</sup> and 0.12 mL catalyst for a GHSV = 200,000 h<sup>−</sup>1. The concentrations of effluent N-containing gases (NO, NH3, NO2 and N2O) were continuously measured by an online FTIR gas analyzer (Nicolet Antaris IGS analyzer, Thermo-Fisher Scientific, Waltham, MA, USA). NO*<sup>x</sup>* conversion and N2 selectivity were calculated using the following equations, respectively:

$$\text{NO}\_{x}\text{ conversion} = (1 - \frac{[\text{NO}]\_{out} + [\text{NO}\_{2}]\_{out}}{[\text{NO}]\_{in} + [\text{NO}\_{2}]\_{in}}) \times 100\%$$

$$\text{N}\_{2}\text{ selectionity} = (1 - \frac{2[\text{N}\_{2}\text{O}]\_{out}}{[\text{NO}\_{x}]\_{in} + [\text{NH}\_{3}]\_{in} - [\text{NO}\_{x}]\_{out} - [\text{NH}\_{3}]\_{out}}) \times 100\%$$

#### *3.2. Characterizations*

X-ray diffraction (XRD) measurements were carried out on a computerized AXS D8 diffractometer (Bruker, GER), with Cu Kα (λ = 0.15406 nm) radiation, from 20 to 80◦ at 8◦/min.

Surface areas were tested using an ASAP 2020 (Micromeritics, Norcross, GA, USA) at −196 ◦C by N2 adsorption/desorption and calculated using a BET equation in the 0.05–0.35 partial pressure range.

The X-ray photoelectron spectroscopy (XPS) results of Ce 3d and O 1s were measured on an ESCALAB 250Xi Scanning X-ray Microprobe (Thermo-Fisher Scientific, Waltham, MA, USA) using Al Ka radiation (1486.7 eV) andaC1s peak, with BE = 284.8 eV as the calibration standard.

The transmission electron microscopy (TEM) image and energy-dispersive X-ray spectroscopy (EDS) mapping of Ce were obtained using a JEM-2100F equipment (JEOL, Tokyo, Japan), combined with a specimen tilting beryllium holder for energy dispersive spectroscopy. The accelerating voltage was 200 kV.

The H2 temperature-programmed reduction (H2-TPR) was tested using an AutoChem\_II\_2920 chemisorption analyzer (Micromeritics, Norcross, GA, USA), and the temperature-programmed desorption of NH3 and NO*<sup>x</sup>* (NO*x*-TPD and NH3-TPD) were tested using the same reaction system as the activity tests. Experiment details can be found in Reference [42].

#### **4. Conclusions**

A series of Ce-Zr-Ti oxide catalysts were prepared using a stepwise precipitation approach for NH3-SCR. CeZr0.5O*<sup>x</sup>* without Ti just showed a relatively low NO*<sup>x</sup>* conversion. When Ti was introduced, Ce-Zr-Ti catalysts showed much better activities and N2 selectivity. A CeZr0.5Ti2O*<sup>x</sup>* catalyst, which contains moderate Ti amounts, showed the best performance, which is associated with its optimal ratios for the redox (CeO*x*) and acidic (TiO2) components.

CeZr0.5O*<sup>x</sup>* and CeZr0.5Ti2O*<sup>x</sup>* catalysts were characterized using various methods and the formation process during preparation was investigated. CeZr0.5Ti2O*<sup>x</sup>* catalyst showed superior redox properties (by H2-TPR), good adsorption and NO*x*/NH3 activation functions (by NO*x*-TPD and NH3-TPD, respectively), and enhanced charge imbalance (by XPS).

During preparation, the Ti and Zr species were first co-precipitated with an increase in solution pH. Then, the Ce species uniformly precipitated onto the precipitated Zr-Ti species with the further increase in pH. As a result, CeZr0.5Ti2O*<sup>x</sup>* catalyst with a surface Ce concentration higher than those of Ti and Zr was obtained. This preparation process resulted in the formation of highly-dispersed CeO2 on ZrO2-TiO2, and thus the catalyst can present excellent NH3-SCR performance.

**Author Contributions:** W.S. and H.H. conceived the project; Y.G. and Y.Z. performed the experiments; W.S. and Z.L. carried out the data analysis; W.S. and Y.G. wrote the paper; H.H. supervised the study.

**Funding:** This work was supported by the National Key R&D Program of China (2017YFC0212502, 2017YFC0211101), the National Natural Science Foundation of China (201637005), and the Key Research Program of the Chinese Academy of Sciences (ZDRW-ZS-2017-6-2-3).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Gas-Phase Phosphorous Poisoning of a Pt/Ba/Al2O3 NOx Storage Catalyst**

**Rasmus Jonsson 1, Oana Mihai 1, Jungwon Woo 1, Magnus Skoglundh 1, Eva Olsson 1, Malin Berggrund <sup>2</sup> and Louise Olsson 1,\***


**\*** Correspondence: louise.olsson@chalmers.se; Tel.: +46-31-772-4390

Received: 9 March 2018; Accepted: 7 April 2018; Published: 11 April 2018

**Abstract:** The effect of phosphorous exposure on the NOx storage capacity of a Pt/Ba/Al2O3 catalyst coated on a ceramic monolith substrate has been studied. The catalyst was exposed to phosphorous by evaporating phosphoric acid in presence of H2O and O2. The NOx storage capacity was measured before and after the phosphorus exposure and a significant loss of the NOx storage capacity was detected after phosphorous exposure. The phosphorous poisoned samples were characterized by X-ray photoelectron spectroscopy (XPS), environmental scanning electron microscopy (ESEM), N2-physisorption and inductive coupled plasma atomic emission spectroscopy (ICP-AES). All characterization methods showed an axial distribution of phosphorous ranging from the inlet to the outlet of the coated monolith samples with a higher concentration at the inlet of the samples. Elemental analysis, using ICP-AES, confirmed this distribution of phosphorous on the catalyst surface. The specific surface area and pore volume were significantly lower at the inlet section of the monolith where the phosphorous concentration was higher, and higher at the outlet where the phosphorous concentration was lower. The results from the XPS and scanning electron microscopy (SEM)-energy dispersive X-ray (EDX) analyses showed higher accumulation of phosphorus towards the surface of the catalyst at the inlet of the monolith and the phosphorus was to a large extent present in the form of P4O10. However, in the middle section of the monolith, the XPS analysis revealed the presence of more metaphosphate (PO3 −). Moreover, the SEM-EDX analysis showed that the phosphorous to higher extent had diffused into the washcoat and was less accumulated at the surface close to the outlet of the sample.

**Keywords:** LNT; NSR; NOx storage; phosphorous; deactivation; poisoning

#### **1. Introduction**

In diesel and lean burn gasoline engines, the engine operates with a large excess of air, and this results in an increased fuel economy, and thereby reduced emissions of CO2, which is a greenhouse gas. However, it is critical to remove NOx, which can be done by different catalytic aftertreatment techniques. Emission standards have therefore been implemented in large parts of the world and the emission levels have been significantly reduced over the years [1].

Since the traditional Three-way catalyst (TWC), requires stoichiometric air:fuel ratio, the Lean NOx Trap (LNT) was introduced as a catalytic concept for exhaust gas treatment in diesel vehicles [2–5]. In the LNT, usually barium is used as a storage component, and the nitrogen oxides during lean conditions are stored on barium as barium nitrates. Moreover, noble metals, such as platinum, play a key role to oxidize the NO to NO2, which facilities the NOx storage [2,6–10]. Under short periods of time the engine is switched from operating under lean to rich conditions. Under rich

conditions, barium nitrates decompose and are reduced over the noble metal sites to mainly N2 and H2O. The reducing agents in the exhaust gas are hydrocarbons (HC), carbon monoxide (CO) and hydrogen (H2), which react with the NOx to produce CO2, H2O and N2. Nitrogen is the desired product from the reduction of stored nitrates. However, there are other possible bi-products from the reduction process such as ammonia (NH3) and nitrous oxide (N2O) [3].

A vehicle is expected to be in service for a long time, which requires the catalyst to be durable. Over time, the catalytic properties of the catalyst deteriorate. There are a few mechanisms that contribute to degradation of catalysts, such as sintering of catalytic particles as a result of high-temperature exposure [11]. The catalytic washcoat can also undergo mechanical tear and poisoning of active sites on the catalyst caused by chemisorption or reactions of catalytically active sites with poisons in the exhaust gas [12]. The poisons are often introduced to the system through the gasoline or diesel fuel, which is the case for SO2, or through the lubricant of the engine. Lubricants, such as zinc dialkyldithiophosphate (ZDDP), are commonly used in vehicles and are a source for accumulation of both sulfates, zinc and phosphorus in the catalyst [13]. The deactivation effect of sulfur has extensively been studied over LNTs [12,14–16], and also how to regenerate the catalyst from sulfur [17].

There are only a few studies available in the open literature where phosphorus poisoning has been studied over noble metal catalysts used for emission control. Bunting et al. [13] studied a diesel oxidation catalyst (DOC) connected to an engine rig where the fuel was doped with ZDDP, which showed accumulation of foreign compounds such as phosphates, zinc and sulfates on the catalyst. Also, results from characterization of field-aged TWCs showed accumulation of elements found from ZDDP poisoning of catalysts [18–20]. Moreover, it has been shown that the distribution of phosphorus as it accumulates in the catalytic washcoat exhibits a gradient, with higher concentration at the inlet and lower along the washcoat length [19–23]. Compared to sulfur, which has a more even distribution over the washcoat and penetrates deep into the washcoat, phosphorus tends to be more concentrated close to the washcoat surface [13]. A study performed by Galisteo et al. [24], where a Pt/Ba/Al2O3 catalyst was impregnated with (NH4)2HPO4, showed that the amount of phosphorus in the catalyst correlated with the loss of NOx storage capacity.

However, there are, to our knowledge, no studies in the open literature that have examined phosphorus poisoning of LNT catalysts using vapour phase exposure that is closer to realistic poisoning conditions, which is the objective of the present work. In this study, we have examined a Pt/Ba/Al2O3 catalyst exposed to phosphorous in the gas phase using different phosphorous concentrations. The catalyst samples were studied in a flow reactor to determine changes in NOx storage capacity, NOx reduction activity and bi-product formation. The phosphorous-exposed samples were also characterized by X-ray photoelectron spectroscopy (XPS), nitrogen-physisorption, Inductive coupled plasma atomic emission spectroscopy (ICP-AES) and Environmental Scanning Electron Microscopy (ESEM) with energy dispersive X-ray (EDX) analysis.

#### **2. Results and Discussion**

#### *2.1. Activity Measurements*

Lean/rich cycling experiments were conducted in which the system was exposed to 400 vol.-ppm NO, 5 vol.-% CO2, 5 vol.-% H2O and 8 vol.-% O2 during the lean phase. In this phase, NO is first oxidized over the platinum sites, and thereafter storage of NOx species occurs as nitrates through a disproportionation mechanism, but also via nitrite formation at low temperature [5]. The NO oxidation capacity of the catalyst is clearly visible by the NO2 formation seen in Figure 1. In the initial part of the lean phase, all NOx is stored and after a certain amount of time, breakthrough of NOx, due to saturation of the storage sites, is seen. Most NOx is stored onto the barium sites in form of Ba(NO3)2, but some NOx is also stored onto alumina sites [3]. At lower temperature, storage of NOx occurs over

both the barium and alumina sites, but at higher temperature the barium sites dominate the NOx storage [25], due to the lower stability of alumina nitrates.

**Figure 1.** Measured outlet NO, NO2, N2O and NH3 concentrations for the Pt/Ba/Al2O3 sample "100 ppm, 34 h" before and after exposure to phosphorous (100 vol.-ppm H3PO4, 8 vol.-% O2, 5 vol.-% H2O at 200 ◦C) in a flow-reactor experiment at 400 ◦C. Lean phase; 400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and 8 vol.-% O2. Rich phase; 400 vol.-ppm NO, 5 vol.-%H2O, 5 vol.-% CO2 and 1 vol.-% H2.

As the system is switched from lean to rich phase, a NOx peak is visible, which is related to release of stored NOx in the catalytic material [26]. Directly after the switch from the lean to the rich (1 vol.-% H2, 5 vol.-% CO2 and 5 vol.-% H2O) phase, minor formation of N2O is seen and it is during this short period of time, that most of the N2 formation is suggested to occur [27–29]. After the N2O peak a large ammonia peak is observed, which is consistent with the work by Lindholm et al. [25]. Most of the formed NH3, early in the rich phase, is consumed by a Selective Catalytic Reduction (SCR) reaction with the stored NOx, according to Lietti et al. [27] and Lindholm et al. [6]. Hence, the NH3 peak is delayed and will dominate more as the rich phase proceeds and this is also seen in Figure 1. Indeed, spatially resolved MS measurements performed by Partridge et al. [10] showed a clear axial consumption of ammonia.

The effect of phosphorus exposure was studied by exposing the Pt/Ba/Al2O3 catalyst to 100 vol.-ppm H3PO4, 8 vol.-% O2 and 5 vol.-% H2O at 200 ◦C for 34 h and thereafter repeating the same experiment, see Figure 1. In addition, a second catalyst was used, which was exposed to 50 vol.-ppm H3PO4 instead. These catalysts were studied under lean/rich cycling experiments, which were conducted both at 300 and 400 ◦C, using 400 vol.-ppm NO, 5 vol.-% CO2, 8 vol.-% O2, and 5 vol.-% H2O for the lean conditions and 1 vol.-% H2 or 1000 vol.-ppm C3H6, 400 vol.-ppm NO, 5 vol.-% CO2 and 5 vol.-% H2O during the rich conditions, and the resulting outlet NOx concentrations are shown in Figure 2. The results show that exposure to phosphorus causes a significant loss of NOx storage capacity for the Pt/Ba/Al2O3 catalyst. The loss of NOx storage capacity is summarised in Table 1. It is clear that the loss in storage capacity is similar for both poisoning levels and, moreover, that the effect is more pronounced after exposure at 300 compared to 400 ◦C.

Our results are in line with those in the study by Galisteo et al. [24], where the effect of phosphorous on the NOx storage capacity of Pt/Ba/Al2O3 was studied by impregnation of the catalyst with (NH3)3PO4 dissolved in water. The authors also observed a decreased NOx storage capacity after phosphorous exposure. At 300 ◦C, the NOx reduction properties of C3H6 are poor compared to 400 ◦C (compare graph B and D in Figure 2), which previously has been observed by Olsson et al. [30]. However, loss in NOx storage capacity is still observed for this case after phosphorus exposure. For exposure at 400 ◦C, the phosphorus poisoning is similar for both H2 and C3H6, which indicates that as long as the reductant can remove the NOx, the phosphorus poisoning is related to poisoning of the reactions under lean conditions. For phosphorous exposure at 300 ◦C on the other hand, the reduction with propene is poor, which results in low NOx storage and therefore the effect of phosphorus poisoning is not seen equally clear.

**Figure 2.** Measured outlet NOx before and after exposing Pt/Ba/Al2O3 to 50 or 100 vol.-ppm H3PO4, 8 vol.-% O2 and 5 vol.-% H2O for 34 h. The lean phase (400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and 8 vol.-% O2) starts at time = 0. The rich phase consist of 400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and H2 or C3H6, where (**A**) 300 ◦C with 1 vol.-% H2 in the rich phase, (**B**) 300 ◦C with 0.1 vol.-% C3H6 in the rich phase, (**C**) 400 ◦C with 1 vol.-% H2 in the rich phase, and (**D**) 400 ◦C with 0.1 vol.-% C3H6 in the rich phase.


**Table 1.** NOx storage capacity for fresh and phosphorous exposed samples.

The resulting N2O formation during these experiments is shown in Figure 3, and it is observed that the selectivity towards N2O is higher at low temperatures, which is in accordance with the literature [31]. Most of the N2O is formed in the first part of the rich period, but also a peak in the beginning of the lean phase is observed. Choi et al. [32] found that the second N2O peak relates to reactions with NH3. When shifting from rich to lean phase, the formation of NH3 ceases due to the replacement of H2 with O2 in the feed. Therefore, it is plausible that the second N2O peak relates to consumption of stored NH3 to form N2O.

**Figure 3.** Measured outlet N2O before and after exposing Pt/Ba/Al2O3 to 50 or 100 vol.-ppm H3PO4, 8 vol.-% O2 and 5 vol.-% H2O for 34 h. The rich phase (400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and H2 or C3H6) starts at time = 0. The lean phase consist of 400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and 8 vol.-% O2, where (**A**) 300 ◦C with 1 vol.-% H2 in the rich phase, (**B**) 300 ◦C with 0.1 vol.-% C3H6 in the rich phase, (**C**) 400 ◦C with 1 vol.-% H2 in the rich phase, and (**D**) 400 ◦C with 0.1 vol.-% C3H6 in the rich phase.

In addition, more N2O is formed when using propene as the reducing agent compared to hydrogen. The N2O formation was previously studied by Olsson et al. [30] where a kinetic model for the reduction of NOx by C3H6 was developed. The reason for the N2O formation is the reaction between the hydrocarbon and NOx to form N2O [33]. Furthermore, in the beginning of the lean phase after propene reduction, there is an increase in the N2O formation. This is due to reaction between NOx and HC species, stored during the rich phase. In addition, the results clearly show that the formation of N2O in the rich phase decreases after exposure to phosphoric acid. The decrease in N2O formation occurs at both 300 and 400 ◦C for both H2 and C3H6 as the reducing agent. When comparing the ratio of (N2O formed)/(NOx stored) at 300 ◦C with propene as reductant, the ratio clearly decreases after phosphorous exposure (0.015, 0.0084, 0.020 and 0.011 for first sample before and after exposure for 100 vol.-ppm P, and for second sample before and after exposure for 50 vol.-ppm P, respectively). Thus, it is clear that the N2O formation not decreases only due that less NOx is stored, but also due to that the selectivity changes because of the poisoning. De Abreu Goes et al. [11] studied thermal exposure of LNT catalysts, which resulted in sintering of the catalytic particles that in turn resulted in decreased formation of N2O. Since the catalyst in the present study was not exposed to high temperature again after the degreening step, sintering as a cause for the loss of N2O formation is not likely. Therefore, the reduced N2O formation after phosphorous exposure is more likely to relate to poisoning of the platinum sites.

With H2 as the reducing agent, the formation of NH3 is higher compared to C3H6 as reductant, which can be seen in Figure 4. At 300 ◦C, the formation of NH3 is even negligible for the C3H6 case. For both samples, the NH3 formation is slightly higher after the exposure to phosphorus when H2 is used as the reducing agent. The ammonia formed over the LNT depends on several reactions occurring simultaneously: (i) ammonia production from inlet NOx; (ii) ammonia production from stored NOx; and (iii) reaction of the formed ammonia with stored NOx in an SCR process [6]. An increase in the

ammonia production has also been observed for field-aged, oven-aged and sulphur-poisoned LNT catalysts [11,34]. The reason for the increase in ammonia formation after phosphorus exposure could be related to the loss of NOx storage capacity depicted in Figure 2. Since less NOx is available in the form of nitrates during the lean phase, less ammonia can in turn react in the SCR reaction with the stored nitrates [6], and thereby the ammonia release is higher. However, less ammonia will also be produced from the stored NOx, thus this means that the SCR reaction is more influenced by the phosphorous poisoning compared to the ammonia production. The increased ammonia production during rich conditions can be beneficial for the cases where an SCR catalyst is placed downstream of the LNT [8], in the so-called passive SCR technique.

**Figure 4.** Measured outlet NH3 concentration before and after exposing Pt/Ba/Al2O3 to 50 or 100 vol.-ppm H3PO4, 8 vol.-% O2 and 5 vol.-% H2O for 34 h. The rich phase (400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and H2 or C3H6) starts at time = 0. The lean phase consist of 400 vol.-ppm NO, 5 vol.-% H2O, 5 vol.-% CO2 and 8 vol.-% O2, where (**A**) 300 ◦C with 1 vol.-% H2 in the rich phase, (**B**) 300 ◦C with 0.1 vol.-% C3H6 in the rich phase, (**C**) 400 ◦C with 1 vol.-% H2 in the rich phase, and (**D**) 400 ◦C with 0.1 vol.-% C3H6 in the rich phase.

The outlet N2 concentration was not measured due to IR spectroscopy was used for the gas phase analysis. By assuming that all nitrogen leaving the catalyst are either in form of NO, NO2, NH3, N2O or N2, the remaining N2 can be determined. The quantities of NOx, NO, NO2 NH3 and N2O were estimated by integration of the data from the FTIR analysis. The results from these calculations are shown in Figure 5 and it is clear that the formed amount of N2 is quite low. The reason for this is that we intentionally used long lean and rich cycles to come closer to saturation and also more complete regeneration. In a real application, the cycle times would be different, giving a high amount of N2. Moreover, the results clearly show a minor N2O formation in all conditions except with C3H6 as reducing agent at 300 ◦C. Hence, the main difference between the results from the different experimental conditions is the formation of NH3, N2 and release of NOx. The phosphorous-exposed samples tend to release higher amounts of NOx and NH3 at the cost of formation of the desired product; N2. According to the ICP-AES results shown in Table 2, the amount of accumulated phosphorous is similar for both H3PO4 concentrations. Thus the H3PO4 concentration does not seem to be such a critical factor for the poisoning at these levels. Therefore, the difference before and after phosphorous exposure between the two cases appears similar. For H2 as the reducing agent, exposure to phosphorus has a more pronounced effect at 300 than 400 ◦C. The reason for this could be that since the ammonia

production is higher at lower temperature, and is to a large extent influenced by the phosphorous poisoning, it results in a more pronounced effect on the N2 selectivity. However, for C3H6 as the reducing agent, there is no clear trend regarding the selectivities between the fresh and phosphorous exposed samples at 300 and 400 ◦C. The reason for this could be that the amounts of NH3 and N2O formed in these experiments are low, so most of the species are NOx and N2.

**Figure 5.** Balance over outlet nitrogen atoms over a lean/rich cycle before and after exposing Pt/Ba/Al2O3 to 50 and 100 vol.-ppm H3PO4 for 34 h, assuming that all nitrogen leaving the catalyst is either in form of NOx, NH3, N2O or N2. (**A**) 300 ◦C with 1 vol.-% H2 in the rich phase, (**B**) 300 ◦C with 0.1 vol.-% C3H6 in the rich phase, (**C**) 400 ◦C with 1 vol.-% H2 in the rich phase, (**D**) 400 ◦C with 0.1 vol.-% C3H6 in the rich phase.

#### *2.2. Characterization of the Phosphorous Exposed Samples*

One channel of the phosphorous-exposed monolith sample "50 ppm, 34 h" was cut out and the surface composition in the inlet, middle, and in the outlet section of the channel was analyzed by XPS. The overall spectra (see Figure 6A), show that the magnitude of the phosphorus peaks is highest from the surface of the inlet section and lowest from the outlet section of the phosphorous exposed sample. The P 2p binding energy region measured with higher resolution for the inlet and outlet sections for the phosphorous exposed sample are shown in Figure 6B. Due to different signal intensities from the two samples, the spectra are normalized to facilitate the interpretation of the peak positions when comparing different spectra with each other. The intensity of the P 2p peak from the middle section of the sample is significantly lower compared to the peak from the inlet section. In addition, the results show that the position of the P 2p peak shifts towards lower binding energy for the middle section compared to the inlet section of the sample. Deconvolution of the P 2p peak for the inlet and middle section of the sample can be seen in Figure 7. For the inlet section of the sample the deconvolution shows a higher presence of P4O10 with a binding energy ranging between 135.0 and 135.5 eV [35]. However, for the middle region of the sample, the deconvolution shows that metaphosphates, PO3 −, dominate, which have a binding energy ranging between 134.0 and 134.5 eV [35]. Formation of

metaphosphates has previously been observed after exposing Cu/BEA [36] as well as Fe/BEA [37] for phosphoric acid in gas phase. In addition, formation of P4O10 was also found on Fe/BEA [37], which is in line with our observations. Our results show that the phosphorus-containing surface species mainly are in the form of P4O10 in the inlet section of the sample, where the phosphorus concentration and deposition is higher, while in the middle section more metaphosphates are seen.

**Figure 6.** (**A**) Normalized X-ray photoelectron spectroscopy (XPS) spectra from the inlet, middle and outlet section of the Pt/Ba/Al2O3 sample exposed to 50 vol.-ppm H3PO4 for 34 h. (**B**) Corresponding XPS spectra for the P 2p region from the inlet and middle section of the sample.

**Figure 7.** Deconvolution of the P 2p XPS spectra from the inlet (**A**) and middle (**B**) section of the Pt/Ba/Al2O3 sample exposed to 50 vol.-ppm H3PO4 for 34 h. The considered phosphorous species are P4O10, PO3 − and PO4 <sup>3</sup>−.

In order to achieve further insight into the effect of phosphorous exposure in the axial direction of the Pt/Ba/Al2O3 catalyst, the specific surface area and pore volume, and elemental composition of the inlet, middle and outlet sections of the samples exposed for 100 and 50 vol.-ppm phosphorous for 34 h were analysed together with a degreened sample. In Table 2, the phosphorus content from ICP-AES analysis is presented along with the specific surface area and pore volume. The results show that phosphorus has a gradient distribution, from the inlet to the outlet section of the two phosphorous-exposed samples, which is in line with other studies of both field-aged catalysts and experiments in engine rigs [13,19–23,38,39]. Moreover, it is found that the effect of the phosphorus concentration is minor, which is in line with the flow reactor experiments. Glisteo et al. [24] impregnated a Pt/Ba/Al2O3 catalyst with aqueous solutions of phosphorous of different concentrations. Their results showed that the specific surface area decreased more by higher phosphorous concentration accumulated on the catalyst, which is in line with our results. We find that the phosphorus content is the highest, and the surface area and pore volume have lowest values, for the inlet section of both catalysts exposed to phosphorous. Close to the outlet section, the phosphorus content is much lower for both samples, whereas the specific surface area and pore volume are higher and more similar to the degreened sample.


**Table 2.** Phosphorous content, specific surface area and pore volume of the fresh Pt/Ba/Al2O3 sample and of the Pt/Ba/Al2O3 samples exposed to 50 and 100 vol.-ppm phosphorous for 34 h. The phosphorus contents are measured by inductive coupled plasma atomic emission spectroscopy (ICP-AES).

<sup>1</sup> Estimated based on washcoat amount.

The results from ESEM and EDX analyses acquired from two different positions; 2 mm from the front and 2 mm from the back of the monolith sample exposed for 50 vol.-ppm phosphorous for 34 h are presented in Figures 8 and 9, respectively. Besides phosphorus, other elements (Al and Ba) that are present in the washcoat are depicted in the images. Moreover, silicon, which is one of the major compounds of the cordierite substrate, is also shown. The analysis 2 mm from the front of the sample (Figure 8) indicates that phosphorus is located towards the surface of the washcoat, which previously has been shown for catalysts exposed to ZDDP doped fuel in engine rigs [13]. Closer to the outlet of the sample, less phosphorus is present and the distribution of the phosphorus is more uniform in the washcoat. Accumulation of phosphorus towards the front of the catalyst has been shown for field-aged catalysts [21]. The ESEM-EDX results in the present study can be related to the XPS results (Figure 7), which showed more P4O10 in the inlet section of the sample, while more metaphosphates in the middle section of the phosphorous exposed sample. Finally, in the outlet section of the monolith sample, only small amounts of phosphorus are observed with EDX, but it should be noted that phosphorus is still visible, which means that in a catalytic aftertreatment system it is possible that a catalyst placed downstream the NOx storage catalyst also can be exposed to phosphorus.

**Figure 8.** Environmental scanning electron microscopy (ESEM) and energy dispersive X-ray (EDX) images from the cross-section 2 mm from the front of the Pt/Ba/Al2O3 sample exposed to 50 vol.-ppm H3PO4 for 34 h.

**Figure 9.** ESEM and EDX images from the cross-section 2 mm from the back of the Pt/Ba/Al2O3 sample exposed to 50 vol.-ppm H3PO4 for 34 h.

To summarize, the combined XPS, EDX, BET and ESEM-EDX data suggest that during phosphorus exposure in the vapour phase more phosphorus is deposited in the inlet of the monolith sample and this phosphorous is mainly located at the surface of the washcoat and contains a large fraction of P4O10. While the middle section of the monolith contains less phosphorus and has a higher fraction of metaphosphates. Moreover, closer to the outlet the phosphorus is more evenly distributed.

#### **3. Materials and Methods**

#### *3.1. Catalyst Synthesis*

The Pt/BaO/Al2O3 catalyst was synthesized using incipient wetness impregnation. To enhance the NOx storage capacity of the chosen model catalyst, the impregnation of platinum on the support material was performed before incorporation of barium, which previously has shown beneficial to achieve high NOx storage capacity of the catalyst [25]. Prior to the impregnation, the γ-alumina (Sigma-Aldrich) support was calcined for 2 h at 600 ◦C. Thereafter the γ-alumina support was impregnated with a platinum nitrate precursor (Heraeus platinum (II) nitrate; solution type K, batch: 12207), diluted with Milli-Q water (18 MΩ·cm) of the estimated volume targeting 2 wt.-% Pt. The impregnated support material was dried in air at 100 ◦C for 2 h and subsequently calcined in air at 550 ◦C for 2 h, starting from room temperature with a temperature increase of 5 ◦C/min. The calcined Pt/Al2O3 sample was then impregnated with a barium precursor (Sigma-Aldrich barium acetate, 99% A.C.S reagent, batch: 11415KA), dissolved in Milli-Q water in two steps also by incipient wetness impregnation, targeting a total of 16 wt.-% of BaO. The catalytic material was dried and calcined after each impregnation step under same conditions as for the impregnation of platinum.

#### *3.2. Monolith Preparation*

The catalytic material was coated on honeycomb-shaped ceramic monolith substrates (20 mm in length, 21 mm in diameter, a wall thickness of 0.106 mm and with 400 cpsi). To improve the adhesiveness, 10 wt.-% boehmite (Sasol disperal P2, product code: 538116) was mixed with the catalyst in a solution of 50 vol.-% Milli-Q water and 50 vol.-% ethanol. The monolith substrates were stepwise dipped in the solution and dried at 90 ◦C until the target washcoat weight of 700 mg per monolith sample was reached. Thereafter, the catalysts were calcined at 550 ◦C for 2 h, starting from room temperature with a temperature increase of 5 ◦C/min.

#### *3.3. Flow-Reactor Experiments*

The flow-reactor experiments were carried out in a synthetic gas bench (SGB) reactor. The coated monolith samples were placed in a quartz tube, which was 750 mm in length and with an inner diameter of 22 mm. To prevent by-pass, quartz wool was wrapped around the monoliths. The gas flow and gas concentrations were regulated using mass flow controllers from Bronkhorst and water vapor was dosed using a Bronkhorst Controlled evaporation system (CEM) system. The quartz tube was placed in an electric heating coil and covered with insulation material, and the temperatures were measured by two thermocouples, one measuring and regulating the gas temperature before the catalyst and one measuring the catalyst temperature in a center channel of the monolith. An MKS Multigas 2030 FTIR spectrometer provided by MKS instruments, Andover, MA, USA was used to monitor the gas phase concentrations of NO, NO2, N2O, NH3 and H2O, and the total flow was 3500 mL/min, corresponding to 30,300 h−<sup>1</sup> space velocity, with argon as carrier gas for all measurements. All lines and connections prior and after the reactor were heated to 200 ◦C, to prevent condensation.

To avoid sintering of platinum particles under the experiments, the coated monoliths were first H2-treated in 1 vol.-% H2, 5 vol.-% CO2 and 5 vol.-% H2O for 20 min at 500 ◦C and thereafter degreened at 600 ◦C, according to the following procedure. First the catalyst was exposed to the rich phase (400 vol.-ppm NO, 5 vol.-% CO2, 1 vol.-% H2 and 5 vol.-% H2O with argon as carrier gas) for 60 min, followed by exposure to the lean phase (400 vol.-ppm NO, 5 vol.-% CO2, 8 vol.-% O2, and 5 vol.-% H2O in argon) for 15 min. Subsequently, the catalyst was reduced in 1 vol.-% H2, 5 vol.-% CO2 and 5 vol.-% H2O in Ar for 20 min at 600 ◦C.

The NOx storage and reduction performance of the catalysts was studied at 300 and 400 ◦C using lean/rich cycling. At each temperature, two cycle segments consisting of five cycles in each segment were conducted. Cycle number 4 for all sequences is shown in the Figures. The gas composition of the lean phase was 400 vol.-ppm NO, 5 vol.-% CO2, 8 vol.-% O2, and 5 vol.-% H2O. In the first lean/rich phase cycle segment, hydrogen was used as the reductant (1 vol.-% H2, 400 vol.-ppm NO, 5 vol.-% CO2 and 5 vol.-% H2O) and in the second cycle segments propene was used as the reducing agent (0.1 vol.-% C3H6, 400 vol.-ppm NO, 5 vol.-% CO2 and 5 vol.-% H2O). For all cycles, the duration of the lean and rich phase were 4 and 1 min, respectively. Prior to the experiment at each temperature, the catalyst was pre-treated with 1 vol.-% H2, 5 vol.-% CO2 and 5 vol.-% H2O for 15 min at 400 ◦C. This pre-treatment was also conducted for the phosphorus-exposed catalysts.

#### *3.4. Phosphorous Exposure*

The NOx storage capacity of the coated monoliths was measured in the flow reactor before and after the exposure to phosphorus, using the procedure described in the previous section. The same reactor as described above was used for the phosphorous exposure, however using a separate reactor tube. In order to reduce surfaces for the evaporated phosphoric acid to stick to, the monolith was not wrapped in quartz wool during the phosphorous exposure. Phosphoric acid was introduced to the gaseous flow by a syringe pump connected to a Teflon pipe disposing the acid inside the quarts tube. The composition of the gas phase during the phosphorous exposure was 50 or 100 vol.-ppm evaporated phosphoric acid, 5 vol.-% H2O, 8 vol.-% O2 in argon as carrier gas, and the temperature was 200 ◦C. Two different phosphorous exposure conditions were studied in this work; in the first case the catalyst was exposed to 50 vol.-ppm phosphoric acid for 34 h, and in the second case the sample was exposed to 100 vol.-ppm for phosphoric acid for 34 h. This corresponds to an exposure of 1.7 and 3.4 g P/l catalyst, respectively. Note that high amounts of phosphorus passed through the catalyst without adsorption. These samples are denoted "50 ppm, 34 h" and "100 ppm, 34 h", respectively. Note that new monolith samples were used for each poisoning condition.

#### *3.5. XPS*

To evaluate the oxidation state of phosphorus accumulated on the catalytic washcoat, an entire channel from the coated monolith was placed on a holder to measure the surface species of the washcoat. The use of a whole channel also enabled the identification of the axial position where measurements took place. The instrument used for the XPS analysis was a Perkin Elmer PHI 5000 ESCA system provided by PerkinElmer, Waltham, MA, USA equipped with an EDS elemental mapping system. The X-ray source for the XPS measurements was monochromatic Al Kα radiation at 1486.6 eV. Correction for charging was performed by normalizing the spectra using the C 1s peak at 284.6 eV [40] as reference. Furthermore, since normalizing will cause the largest peak being equal to 1, all spectra were divided by 1.1 for better visualization. Deconvolution of the P 2p peak from each sample was performed by fitting a Gaussian function to the experimental data, with subtraction of a linear baseline under both peaks. The peak positions were optimized according to the same procedure for both samples to achieve the lowest standard deviation and found to be 134.25, 135.5 and 132.5 eV for PO3 −, P4O10 and PO4 <sup>3</sup>−, respectively.

#### *3.6. N2-Physisorption and ICP-AES*

The specific surface area and pore volume of the samples were measured by N2-physisorption isotherms at −195 ◦C and were collected using a TriStar 3000 gas adsorption analyzer. The inlet-, middle- and outlet section of the coated monoliths were crushed and grinded to powder form. Approximately 300 mg sample from each section was thermally dried at 90 ◦C in N2 gas flow for 4 h and used for specific surface area and pore-volume measurements.

Powder samples from each section were used for elemental analysis by inductive coupled plasma atomic emission spectroscopy (ICP-AES), and the measurements were performed by ALS Scandinavia AB, Luleå, Sweden. The concentration of the measured phosphorus is reported as weight percentages.

#### *3.7. ESEM*

Environmental scanning electron microscopy (ESEM) was used to acquire an overview of the location of different elements over the catalytic washcoat. The instrument used to gain these images was a Quanta200 ESEM FEG from FEI, Hillsboro, OR, USA. An energy dispersive X-ray (EDX) system from Oxford Inca was equipped to the instrument to map the elements over scanned areas.

#### **4. Conclusions**

In the present study, a Pt/Ba/Al2O3 NOx storage catalyst coated on a ceramic monolith substrate was exposed to two different phosphorus containing atmospheres for 34 h at 200 ◦C; 50 vol.-ppm H3PO4 + 8 vol.-% O2 + 5 vol.-% H2O and 100 vol.-ppm H3PO4 + 8 vol.-% O2 + 5 vol.-% H2O. The catalysts were characterized by N2-physisorption, XPS, ESEM and ICP-AES, and the NOx storage capacity was measured in flow reactor experiments.

The specific surface area and pore volume measurements, together with the elemental analysis by ICP-AES, showed a clear gradient distribution of phosphorous along the axial direction of the monolith. The phosphorus concentrations close to the inlet section of the sample were significantly higher compared to the middle and outlet section. The specific surface area and pore volume were found to be lower in the sections where the concentration of phosphorus was higher. At the outlet section, both the specific surface area and the pore volume were found to be comparable with those for the fresh sample, and the accumulated amount of phosphorus was only about 0.1 wt.-%, which can be compared with 2 wt.-% at the inlet section of the phosphorous exposed catalyst. Moreover, there was no significant difference between the different phosphorous concentrations, which indicates that the phosphorus concentration is not limiting the deposition of phosphorous on the catalyst, due to long exposure time.

Flow-reactor experiments showed that the NOx storage capacity of Pt/BaO/Al2O3 decreased with presence of phosphorous species on the catalyst. The results also showed that the formation of N2 decreased after the exposure to phosphorous, in favour of ammonia production. Furthermore, the formation of N2O decreased after the phosphorus exposure. The increased ammonia production can be explained by the fact that the outlet ammonia concentration is dependent on three reactions over the catalyst: (i) the formation of ammonia from NO in the inlet gas; (ii) formation of ammonia from stored NOx; and (iii) the consumption of ammonia through the selective catalytic reduction by the reaction with the stored nitrates. Since the amount of stored nitrates is lower for the phosphorus exposed sample, it means that less ammonia is produced in reaction (ii), but simultaneously less ammonia is consumed in reaction (iii). Since there is more ammonia in total, it means that the SCR reaction is more influenced by the phosphorus poisoning than the ammonia formation from the stored nitrates.

The XPS analysis showed that phosphorus exposure resulted in the formation of different species depending on the axial location of the washcoat. At the inlet section of the catalyst, where most of the phosphorus was accumulated, the dominating phosphorous species was found to be P4O10 while further along the axial direction metaphosphates (PO3 −) was the dominating species. From the ESEM and EDX analysis it could be concluded that most phosphorus, close to the inlet of the catalyst, was located at the washcoat surface. At the outlet, only a small amount of phosphorus could be detected by SEM-EDX, however, it was more diffused into the washcoat. To summarize, the results from XPS and SEM indicate that the accumulation of phosphorus differs between the front and the back of the monolith catalyst.

**Acknowledgments:** This work was conducted at the Competence Centre for Catalysis (KCK) hosted by Chalmers University of Technology and financially supported by the Swedish Energy Agency, Chalmers and the member companies AB Volvo, ECAPS AB, Haldor Topsøe A/S, Scania CV AB, Volvo Car Corporation AB, and Wärtsilä Finland Oy. KCK is greatly acknowledged for their financial support. Lars Ilver is gratefully acknowledged for the assistance with XPS measurements.

**Author Contributions:** R.J. and L.O. conceived and designed the experiments; R.J. performed the experiments; R.J., O.M., M.S., E.O., M.B. and L.O. analyzed the data; L.I. and J.W. contributed analysis tools; R.J. wrote the paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **"PdO vs. PtO"—The Influence of PGM Oxide Promotion of Co3O4 Spinel on Direct NO Decomposition Activity**

#### **Gunugunuri K. Reddy \*, Torin C. Peck and Charles A. Roberts \***

Toyota Research Institute of North America, Ann Arbor, MI 48105, USA; torin.peck@toyota.com **\*** Correspondence: krishna.gunugunuri@toyota.com (G.K.R.); charles.roberts@toyota.com (C.A.R.)

Received: 17 November 2018; Accepted: 5 January 2019; Published: 9 January 2019

**Abstract:** Direct decomposition of NO into N2 and O2 (2NO→N2 + O2) is recognized as the "ideal" reaction for NOx removal because it needs no reductant. It was reported that the spinel Co3O4 is one of the most active single-element oxide catalysts for NO decomposition at higher reaction temperatures, however, activity remains low below 650 ◦C. The present study aims to investigate new promoters for Co3O4, specifically PdO vs. PtO. Interestingly, the PdO promoter effect on Co3O4 was much greater than the PtO effect, yielding a 4 times higher activity for direct NO decomposition at 650 ◦C. Also, Co3O4 catalysts with the PdO promoter exhibit higher selectivity to N2 compared to PtO/Co3O4 catalysts. Several characterization measurements, including X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), H2-temperature programmed reduction (H2-TPR), and *in situ* FT-IR, were performed to understand the effect of PdO vs. PtO on the properties of Co3O4. Structural and surface analysis measurements show that impregnation of PdO on Co3O4 leads to a greater ease of reduction of the catalysts and an increased thermal stability of surface adsorbed NOx species, which contribute to promotion of direct NO decomposition activity. In contrast, rather than remaining solely as a surface species, PtO enters the Co3O4 structure, and it promotes neither redox properties nor NO adsorption properties of Co3O4, resulting in a diminished promotional effect compared to PdO.

**Keywords:** direct NO decomposition; PGM oxide promotion; PdO vs. PtO; in-situ FT-IR; NO adsorption properties; redox properties

#### **1. Introduction**

Nitrogen oxides (NOx) formed by combustion from fixed and mobile sources cause severe detrimental environmental problems, such as acid rain and photochemical smog [1–3]. Effectively controlling the emission of NOx is the topic of much research and has led to the introduction of many new catalyst technologies, such as three-way catalysts (TWC), NOx storage-reduction (NSR), and selective catalytic reduction (SCR) for NOx gas removal from mobile sources, and SCR and selective non-catalytic reduction (SNCR) for NOx gas removal from fixed sources [4–6]. Among various deNOx strategies, direct decomposition of NO (NO→1/2O2 + 1/2N2) has been considered to be the most desirable method because this reaction is thermodynamically favorable at low temperatures and does not need any reductants, such as NH3, H2, CO, or hydrocarbons. However, kinetic studies have indicated that the reaction needs to overcome a large activation energy (~335 kJ mol−1) barrier [4–15]. Accordingly, there is an apparent need for a suitable catalyst to decompose NOx at a given temperature, and therefore, significant research has been undertaken towards development of active and stable catalysts.

Since the pioneering work of Jellinek on the catalytic decomposition of NO in 1906, much research has been reported on NO direct decomposition over several materials, including perovskites, rare earth oxides, and Cu-zeolites [2–7]. Numerous metal oxides have also been examined as candidates for NO decomposition catalysts [16] and Co3O4 is often recognized as a significant component in many active catalysts at higher reaction temperatures [17–21]. However, Haneda et al. recently reported that NO decomposition takes place slowly, if at all, over pure Co3O4 at temperatures below 650 ◦C [18]. They reported that the presence of small amounts of alkali metals were essential to activate NO decomposition over Co3O4 oxide by enhancing NO adsorption [18–20]. This interesting effect of alkali metals, particularly Na, was also reported by Kung et al. [21], but dependence on alkali metals is not feasible for practical applications due to their volatile nature at temperatures above 600 ◦C.

Metal oxide supported platinum group metals (PGM metals) were also one of the earliest types of NO decomposition catalysts studied, and the results have been widely reported; mainly Au, Pt, Pd, and Ir at temperatures higher than 700 ◦C [22,23]. Suzuki et al. [24] synthesized a porous CaZrO3/MgO/Pt composite and found that this catalyst could obtain a NO conversion rate of about 52% at 900 ◦C in the absence of O2. Haneda et al. [25] found that the addition of Pt improved the direct NO decomposition performance of rare earth oxides. They [26,27] also compared the activity of [Pd(NH3)4] (NO3)2, Pd(NO3)2, Pd(CH3COO)2, and (NH4)2-[PdCl4] as palladium precursors for NO decomposition in a Pd/Al2O3 catalyst at 700 ◦C, and the activity was found to decrease in the order of Pd(NO3)2 > [Pd(NH3)4] (NO3)2 > Pd(CH3COO)2 >> (NH4)2[PdCl4]. Almusaiteer et al. [28] reported that compared to Pd/Al2O3, the Pd/C (activated carbon) catalyst was found to be more beneficial for O2 desorption, but both have similar activity. Oliveira et al. [29] investigated the catalytic performance of palladium and copper catalysts loaded in mordenite (MOR) and found that these catalysts were more active for NO decomposition than alumina supported catalysts. However, the reports on supported PtO catalysts for direct NO decomposition at temperatures below 700 ◦C are very limited in the literature. Similarly, metallic Pd catalysts always deactivate over time due to oxidation of Pd metal to PdO at temperatures below 650 ◦C [30].

To the best of our knowledge, Co3O4 supported PGM catalysts have never been explored for direct NO decomposition, likely due to the inactivity of the individual components at lower temperatures. However, the need for enhanced NO adsorption on Co3O4 suggests that the addition of PGM promotion can lead to increased low temperature activity. Hence, the present study aims to investigate the promotional effect of PdO vs. PtO on the Co3O4 for direct NO decomposition. The activity measurements show that the optimum PdO/Co3O4 catalyst exhibits 4 times higher activity than PtO/Co3O4 catalysts. Several characterization techniques such as X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), H2-Temperature programmed reduction (H2-TPR), and *in situ* FT-IR are employed to understand the influence of PdO and PtO on the structural and surface properties of Co3O4.

#### **2. Results and Discussion**

#### *2.1. Direct NO Decomposition Activity Measurements*

To qualify as direct NO decomposition, the catalyst must decompose NO into just two products: N2 and O2. The possibility for unwanted N2O and NO2 formation as side products cannot be neglected during the analysis of this reaction. Therefore, high NO conversion is desired, but not sufficient; it is also very important to maximize selectivity towards N2 rather than N2O or NO2. Considering all possible products, the reaction can be written as:

$$\begin{aligned} \text{2 NO} &\rightarrow \text{N}\_2 + \text{O}\_2\\ \text{4 NO} &\rightarrow 2\text{N}\_2\text{O} + \text{O}\_2\\ \text{NO} + [\text{O}] &\rightarrow \text{NO}\_2 \text{ ([O]: catalyst lattice oxygen)} \end{aligned}$$

The selectivity to N2 can be defined as:

$$\text{N}\_2\text{ selectionity (\%)} = 2 \times \text{[N}\_2\text{]}/\text{(2} \times \text{[N}\_2\text{]} + 2\text{[N}\_2\text{O]} + \text{[NO}\_2\text{]})$$

Also, when NO dissociates on the catalyst surface, it is possible for N2 production to occur, but without simultaneous release of the stoichiometric amount of the O2 product. This situation can occur quite frequently and is the topic of a previous study from some of us [30]. Rather than desorb as the O2 product, oxygen atoms either remain strongly adsorbed on the surface or chemically react with catalyst material, oxidizing both surface and bulk, and changing the catalyst composition. In either case, the catalyst deactivates over time. Hence, it is very important to confirm that the catalyst releases both N2 and O2 as products.

Direct NO decomposition measurements were performed over the pure spinel oxide Co3O4 and over the PdO- and PtO-promoted Co3O4 catalysts, denoted PdO/Co3O4 and PtO/Co3O4, respectively. Activity was measured for 2 h at 400, 450, 550, and 650 ◦C. The direct NO decomposition activity to N2 of PdO/Co3O4 and PtO/Co3O4 with varying PGM loading is presented in Figure 1 as a function of temperature and compared to the pure Co3O4 spinel. All numeric values of NO conversion and N2, N2O, NO2 ppm product concentrations of the Co3O4, PdO/Co3O4, and PtO/Co3O4 catalysts at various reaction temperatures are presented in Tables S1 and S2. The raw NO conversion profiles (NO converted to all the products) of Co3O4, 3PdO/Co3O4, and 4 PtO/Co3O4 during the steady state direct NO decomposition obtained from FT-IR detector in the temperature region 400 to 650 ◦C are presented in Figure S1. High values for NO conversion at lower temperatures may seem counterintuitive, however, most of the NO conversion at low temperature is simply due to thermodynamically favorable NO oxidation to NO2. For example, as shown in Table S1, the NO conversion values of Co3O4, 3 PdO/Co3O4, are 3 PtO/Co3O4 are 3.15, 2.92, and 7.66, respectively, giving the impression that 3 PtO/Co3O4 is the most active catalyst at 400 ◦C. However, for 3 PtO/Co3O4, the NO conversion specifically to N2 is lower than that of 3 PdO/Co3O4 because most of the NO conversion is due to NO oxidation. Hence, for NO decomposition to N2, it is more instructive to consider NO converted into N2 rather than total NO conversion. The activity values were calculated in this manner and are presented in Figure 1 in units of micromoles of NO converted to N2 per gram per second.

NO and Ar partial pressure values obtained by mass spectrometry (MS) during the steady state measurements are presented in Figure S2 and compared with data obtained in the absence of the catalyst, which serves as a baseline. Inert Ar gas was introduced as a tracer to monitor for potential systematic variation in signal intensity during the experiment. As shown in Figure S2, no change in Ar signal intensity was observed during the experiment, however, the intensities of NO signal changed based on catalyst identity and reaction temperature. These measurements confirm the change in the NO signal is due to catalytic conversion of NO and not due to artifacts. Also, the MS signal for NO (*m*/*z* = 30) tracks with the conversion reported by the FT-IR measurements (Figure S1). Figures S3–S5 present the MS partial pressure values of the N2, N2O, and NO2 products, respectively, during direct NO decomposition and are also compared to the MS partial pressures obtained in the absence of the catalyst. As shown in Figure S3, the N2 and O2 signal intensities are higher compared to the background signals, which confirms the simultaneous release of N2 and O2 as expected for NO decomposition. Furthermore, the MS intensity of the N2 signal qualitatively tracks with the N2 concentrations calculated by nitrogen mass balance from the FTIR measurements (Tables S1 and S2), lending additional confidence in the activity results. Similarly, good correlation between the FT-IR detection of N2O and NO2 and the MS signals was observed (compare Tables S1 and S2 to Figures S3 and S4). Finally, the release of oxygen as a product (Figure S3) and stable NO conversion (Figures S1 and S2) during the steady state measurements suggest that the catalysts were not poisoned by the irreversible chemical adsorption of oxygen. Thus, the good correlation between the FT-IR data and MS signals suggests that the calculation of N2 production from the FT-IR is reliable and can be used for the calculation of activity from NO conversion to N2. As shown in Figure 1b, NO decomposition activity to N2 increases slightly with temperature up to 550 ◦C for the pure Co3O4 catalyst. Further increase in the temperature to 650 ◦C results in decreased activity. This result suggests that the Co3O4 spinel is not a good catalyst for NO decomposition at temperatures below 650 ◦C, as indicated by Hamada et al. [21]. The addition of PdO and PtO to the Co3O4 spinel improves the direct NOx decomposition activity of

Co3O4. Direct NOx activity increases with temperature for all palladium and platinum loadings, but unlike the pure Co3O4, deactivation was not observed above 550 ◦C for the any of PdO or PtO catalysts.

The direct NO decomposition activity of PdO/Co3O4 and PtO/Co3O4 to N2 at various temperatures is presented as a function of the weight percent loading of the respective PGMs in Figure 2. For PdO/Co3O4, activity increases with palladium loading up to 3 wt%, but further increase in the palladium loading leads to decreased activity (Figure 2a). For PtO/Co3O4, activity increases with PtO loading up to 4 wt%, however, the overall effect on activity is significantly diminished compared to PdO/Co3O4 (Figure 2b). The optimum loading of each PGM was found to be 3 wt% PdO/Co3O4 and 4wt% PtO/Co3O4. Interestingly, PdO-promoted catalysts exhibit higher activity than the PtO-promoted catalysts. At 650 ◦C, the optimum PdO/Co3O4 catalyst exhibits 4 times higher activity compared to the optimum PtO/Co3O4 catalyst.

To confirm the reaction is indeed direct NO decomposition to N2 rather than the unwanted production of N2O or NO2, the selectivity to N2 was calculated. The N2 selectivity is presented as a function of PGM loading from 400–650 ◦C for PdO/Co3O4 and PtO/Co3O4 in Figure 3a,b, respectively. As expected, pure Co3O4 (0 wt% PGM loading) exhibited relatively low selectivity to N2 (≤20%) at 400 and 450 ◦C. The N2 selectivity increased to 80% at 550 ◦C and to 100% at 650 ◦C. Formation of N2O was not observed and only N2 and NO2 products are detected during direct NO decomposition over the PtO- and PdO-promoted Co3O4 catalysts (Tables S1 and S2). Thus, the product distribution measurements suggest NO oxidation (NO2 formation) is more favorable at lower reaction temperatures (≤450 ◦C), and at higher reaction temperatures, NO decomposition (N2 formation) is predominant. Remarkably, the addition of 1 wt% PdO to the Co3O4 improves the N2 selectivity from 1 to 40% at 400 ◦C and from 21 to 70% at 450 ◦C. The N2 selectivity further increased to 50% at 400 ◦C and 75% at 450 ◦C with increasing PdO loading to 3 wt%. Increasing the PdO loading from 3 wt% to 4 wt% lead to only a slight decrease in the N2 selectivity. When also considering the activity measurement in Figure 2, the N2 selectivity measurements confirm 3 wt% PdO as the optimum loading on Co3O4 for direct NO decomposition.

Regarding N2 selectivity as a function of PtO loading (Figure 3b), the addition of 1 wt% PtO yielded less improvement at 400 ◦C (1 to 16%) compared to the addition of 1 wt% PdO (1 to 40%). Furthermore, the N2 selectivity decreases with increasing PtO loading, dropping to 9% for the 4wt% PtO/Co3O4 catalyst. Moreover, there is no improvement in the selectivity observed at reaction temperatures at and above 450 ◦C. Therefore, the direct NO decomposition measurements show that the addition of PdO to Co3O4 improves the decomposition activity and N2 selectivity and 3wt% is the optimum Pd loading over Co3O4, whereas PtO loaded on Co3O4 leads to only slight improvement in the activity and almost no influence on the N2 selectivity.

**Figure 1.** Direct NO decomposition activity to N2 as a function of reaction temperature for (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts with varying PGM loading. The pure Co3O4 support is included for comparison. (Gas hourly space velocity (GHSV) = 2100 h<sup>−</sup>1, 1% NO/He).

**Figure 2.** Direct NO decomposition activity to N2 as a function of Pd and Pt loading for (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts at varying temperature. The pure Co3O4 support is included for comparison. (GHSV = 2100 h<sup>−</sup>1, 1% NO/He).

**Figure 3.** N2 selectivity as a function of PdO and PtO loading for (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts at varying temperature. The pure Co3O4 support is included for comparison. (GHSV = 2100 h<sup>−</sup>1, 1% NO/He).

#### *2.2. Catalyst Characterization*

#### 2.2.1. Structural and Textural Properties

These catalysts have been evaluated using several characterization techniques, like XRD, XPS, H2-TPR, BET surface area, and *in-situ* FT-IR during NO adsorption, to understand the influence of PdO and PtO on the structural and surface properties of Co3O4 and to explain the greater promoter effect of PdO on Co3O4 compared to PtO during NO decomposition. The palladium and platinum loadings of the studied catalysts prepared by impregnation were verified with XRF spectrometry. For the nominal 1.0, 2.0, 3.0, and 4.0 wt% of PdO, the experimental values were 0.83, 1.94, 2.80, and 4.15, respectively (Table 1). The experimental values for PtO doped Co3O4 catalysts were 0.93, 2.12, 3.23, and 4.02, respectively. Differences may be due to surface heterogeneity or incomplete precursor dispersion during the impregnation procedure or the inherent uncertainty related to the employed XRF method, which did not utilize a standard material to aid the data analysis.

The BET surface area values of PdO/Co3O4 and PtO/Co3O4 catalysts and the pure Co3O4 are presented in Table 1. The pure Co3O4 catalyst exhibits a BET surface area of 36 m2/g. Little change in the surface area is observed after impregnating Co3O4 with 1, 2, and 3 wt% Pd. However, increasing Pd loading from 3 to 4 wt% on to Co3O4 lead to a decrease in the surface area from 33 to 26 m2/g. These values suggest that PdO dispersed very well on the surface of Co3O4 until 3wt% and further

increase in the loading to 4wt% likely leads to surface agglomeration, which can block access to the active surface. The BET surface area measurements are corroborated by the activity measurements, which showed that the activity of Co3O4 increases with increasing palladium doping only until 3 wt%. Further increase in the Pd loading to 4 wt% lead to a decrease in the activity. Little change in surface area was observed for the Co3O4 with platinum impregnation (Table 1), as only a slight decrease in the surface area was observed at the highest loading. As suggested above, the blockage of the active surface may be the cause of the decrease in the activity for the 4 wt% PdO/Co3O4 catalyst. It is hypothesized that the formation of PdO crystallites is responsible for this behavior, and this hypothesis will be investigated below.


**Table 1.** Co/M (M = Pd, Pt) and BET surface area values of PdO/Co3O4 and PtO/Co3O4 catalysts.

\* As measured by XRF.

The X-ray diffraction (XRD) patterns of the fresh PdO/Co3O4 and PtO/Co3O4 catalysts are shown in Figure 4a,b, with the pattern of the pure Co3O4 for reference. The X-ray diffraction lines characteristic of the cubic cobalt spinel structure were indexed within the Fd3m space group (JCPDS card no. 01-080-1533) in the case of the pure Co3O4 catalyst [31]. As shown in Figure 4a, all the PdO/Co3O4 catalysts exhibit peaks due to Co3O4. The diffractograms provide evidence that the spinel structure was preserved after Pd impregnation, revealing no observable structural changes compared to the pure Co3O4 carrier. Diffraction peaks related to Pd or PdO were not detected on samples up to a nominal Pd loading of 3 wt%. For the 4 wt% Pd sample, a low-intensity diffraction peak indicative of tetragonal PdO (0 0 2) (JCPDS card no. 75-584) was visible at 2θ of 33.6◦ [32]. These measurements suggest that PdO is well-dispersed on the Co3O4 support up to 3 wt%, and above this loading, crystalline PdO forms on Co3O4 surface. As stated above, the BET surface decreases from 33 to 26 m2/g with increasing Pd loading from 3 wt% to 4 wt%. XRD measurements confirm that the decrease in the surface area is due to the formation of crystalline PdO on the surface of Co3O4 and blocking of the active surface. The X-ray diffraction patterns of PtO/Co3O4 catalysts are shown in Figure 4b and only peaks due to Co3O4 are present at all Pt loadings, suggesting the absence of bulk metallic Pt or PtO with long-range order. However, unlike PdO/Co3O4, the XRD peaks of the PtO/Co3O4 samples were shifted to higher values relative to the pure Co3O4 spinel for all Pt loadings. The shift in the peak position to higher values indicates that Pt is likely incorporating into the Co3O4 spinel structure in contrast to PdO/Co3O4, where the absence of the peak shift indicates that Pd remained dispersed on the Co3O4 surface.

Figure 5a,b display X-ray diffraction patterns of the spent PdO/Co3O4 and PtO/Co3O4 catalysts after direct NO decomposition. As shown in Figure 5a, the spent Co3O4 exhibits only peaks due to spinel structure. There are no peaks due to either CoO or metallic Co, suggesting the Co3O4 spinel is structurally stable during direct NO decomposition. In addition to the Co3O4 spinel peaks, the spent PdO/Co3O4 catalysts with 1, 2, and 3 wt% Pd loading also exhibit peaks at 2θ values of 40.3, 46.79, and 68.4◦. These peaks are due to the (1 1 1), (2 0 0), and (2 2 0) facets of Pd metal (JCPDS no: 46-1043). The XRD measurements of the spent catalysts show that the dispersed PdO reduced to Pd metal during direct NO decomposition. Similar to the XRD pattern for the fresh 4 wt% PdO/Co3O4, the spent XRD pattern also exhibits peaks due to PdO along with the Pd metal and Co3O4 peak, which suggests that the crystalline PdO remains even after direct NO decomposition. The X-ray diffraction patterns of the spent PtO/Co3O4 catalysts after direct NO decomposition are displayed in Figure 5b. The spent

PtO/Co3O4 catalysts exhibit peaks due only to Co3O4 after direct NO decomposition. In contrast to the metallic phase observed in the spent PdO/Co3O4 catalysts, there are no peaks due to metallic Pt observed in the spent PtO/Co3O4 catalysts.

The XRD measurements show that in the case of PdO/Co3O4 catalysts, the PdO reduced to metallic Pd during direct NO decomposition and promotes the activity of Co3O4 catalysts. On the other hand, no metallic Pt formation occurred in the PtO/Co3O4 catalysts, leading to a greatly diminished promoter effect compared to the PdO/Co3O4 catalysts. Also, the NO decomposition measurements show that the catalytic activity decreases with increasing Pd loading from 3 wt% to 4 wt%. The formation of crystalline PdO and decrease in the surface area (blocking of the active surface) explains the lower activity of 4 wt% sample compared to 3 wt% sample. Hence, XRD and BET surface area measurements corroborate with the activity measurements. The spent PtO/Co3O4 catalysts also exhibit a shift in the peak positions to higher 2θ values compared to the pure Co3O4 catalyst, which suggests the incorporation of Pt into the Co3O4 spinel structure even after direct NO decomposition.

**Figure 4.** X-ray diffraction patterns of fresh (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts. The pattern for the fresh pure Co3O4 support is included for reference.

**Figure 5.** X-ray diffraction patterns of spent (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts. The pattern for the spent pure Co3O4 support is included for reference.

#### 2.2.2. Redox Properties

The influence of PdO and PtO on the redox properties of Co3O4 are investigated using H2-temperature programmed reduction (H2-TPR) measurements. The H2-TPR profiles of PdO/Co3O4 and PtO/Co3O4 catalysts are presented in Figure 6a,b, along with that of the pure Co3O4 for comparison. Several authors reported that the reduction behavior of Co3O4 is strongly dependent on the preparation method, catalyst composition, and dispersion on a support [33,34]. The reduction behavior of Co3O4 was widely accepted as a stepwise process, including the reduction of Co3+ to Co2+ and Co2+ to metallic Co. There are three well-defined reduction peaks in the TPR profile of Co3O4 (Figure 6). The peak at 235 ◦C is attributed to the reduction of surface oxygen species. The other two peaks are for the stepwise reduction of Co3O4 to metallic cobalt. According to the literature, the second reduction peak centered at 275 ◦C is due to the reduction of Co3O4 to CoO, and the third peak at the region of 305 ◦C is due to the reduction of CoO to metallic cobalt [33,34].

$$\text{Co}\_3\text{O}\_4 + \text{H}\_2 \rightarrow 3\text{CoO} + \text{H}\_2\text{O}$$

$$\text{CoO} + \text{H}\_2 \rightarrow \text{Co} + \text{H}\_2\text{O}$$

The addition of 1 wt% PdO to the Co3O4 leads to a drastic change in the redox profile of Co3O4 (Figure 6). No peaks were observed in the 250–310 ◦C temperature region. Both PdO and Co3O4 were reduced at much lower temperature and all reduction events completed below 150 ◦C. These measurements show that PdO promotes the reduction of Co3O4. Two reduction peaks were observed in the 1 wt% PdO/Co3O4 H2-TPR profile at 79 and 104 ◦C. The first reduction peak at 79 ◦C is due to the reduction of PdO to metallic Pd and the second reduction peak is due to the reduction of Co3O4. The H2-TPR profiles for 1, 2, and 3 wt% PdO/Co3O4 were all very similar (Figure 6). The promotion of Co3O4 reduction by Pd observed in H2-TPR is possibly ascribed to hydrogen spillover and the synergistic effect between Pd species and Co3O4. The synergistic effect can weaken the Co-O bond. Chen et al. [35] also reported a similar promotional effect for PdO impregnated on Co3O4 catalysts with different morphologies, and the synergistic effect between Pd and Co existed, regardless of Co3O4 morphology. In the present study, the intensity of the first reduction peak increases with increasing PdO loading from 1 to 3 wt%, and the increase is accompanied by a slight shift in the peak temperature from 79 to 85 ◦C. This may be due to the increase in the PdO loading on the Co3O4 surface. The reduction profile of 4 wt% PdO/Co3O4 is slightly different from the PdO promoted catalysts of lower loading. Along with the peaks due to Co3O4 and PdO, a small additional peak is observed at 220 ◦C. Given the identification of crystalline PdO in the XRD pattern of the 4 wt% PdO/Co3O4, it is reasonable to assign this peak to reduction of crystalline PdO.

The H2-TPR profiles of PtO/Co3O4 catalysts are presented in Figure 6b. Two types of reduction features were observed in the case of PtO promoted Co3O4 catalysts, one from 130 to 190 ◦C and another from 200 to 325 ◦C. The first feature corresponds to reduction of PtO to metallic Pt and the second is reduction of Co3O4. Unlike PdO/Co3O4, little to no shift in the Co3O4 reduction temperature of PtO/Co3O4 catalysts was observed relative to the pure Co3O4. The reduction of Co3+ to Co2+ occurred in the 260–275 ◦C temperature region for Co3O4 and PtO/Co3O4 catalysts, irrespective of PtO loading, and the reduction of PtO occurred separately at a distinctly lower temperature. Yang et al. [36] observed similar behavior for Pt promoted Co3O4/Al2O3 catalysts, wherein both PtO and Co3O4 reduced separately in distinct temperature regions. Even though the Co3O4 reduction shifted to slightly lower temperatures at higher Pt loadings in their study, a synergistic effect, similar to Pd and Co in PdO/Co3O4, was not observed between Pt and Co. This lack of synergistic effect by Pt on the reduction of Co3O4 is consistent with the smaller promotional effect of Pt on direct NO decomposition activity compared to Pd promotion. Conversely, the decreased reduction temperature of Co3O4 observed in H2-TPR measurements of PdO/Co3O4 illustrates how Pd can promote direct NO decomposition by enhancing the reducibility of the catalyst.

**Figure 6.** H2- Temperature programmed reduction profiles of fresh (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts. The pure Co3O4 profile is shown for comparison.

#### 2.2.3. Surface Properties

The X-ray photoelectron spectroscopy (XPS) was used to investigate the surface elemental compositions, metal oxidation states, and adsorbed oxygen species of the as-prepared and spent samples. The O*1s* XPS spectra of fresh PdO- and PtO-promoted Co3O4 catalysts are presented in Figure 7, with that of the pure Co3O4 for comparison. The pure Co3O4 exhibits two peaks in the O*1s* spectra. The large peak at lower binding energy (BE = 530.2–530.7 eV) is attributed to the surface lattice oxygen in Co3O4 (denoted as Olat) [37]. The shoulder at higher BE (532.0–532.7 eV) is associated with oxygen atoms present as surface adsorbed oxygen or surface hydroxyl groups or defect oxide (denoted as Oad). The PdO- and PtO-promoted Co3O4 samples also exhibit two peaks in their O*1s* spectra due to the Olat and Oad species., however, little difference in the peak energies is observed. This is may be due to lower loadings of promoters.

**Figure 7.** O*1s* XPS spectra of fresh (**a**) PdO- and (**b**) PtO-promoted Co3O4 catalysts. Co3O4 spectrum is shown for reference.

The fitted Co*2p* XPS spectra of the fresh PdO- and PtO-promoted Co3O4 catalysts are presented in Figure 8, with that of the pure Co3O4 for comparison. In the pure Co3O4 XPS spectrum, the main peak in the BE range of 780.7–782.2 eV is assigned to Co*2p*3/2, and the shoulder at 795.9–797.9 eV is attributed to Co*2p*1/2. Pure Co3O4 exhibits peaks due to both Co3+ and Co2+ and their satellites. The main Co*2p*3/2 feature can be further resolved into two components, with BE values centered at 778.7–780.4 eV and 779.8–781.6 eV, and corresponding to Co3+ and Co2+, respectively [38]. Furthermore, the presence of the satellite peaks also confirms the presence of Co2+ in the catalysts. As expected for samples containing Co3O4 spinel, all catalysts exhibited peaks and satellites due to both Co3+ and Co2+, irrespective of Pd or Pt promoter loading. Also, no significant change in the position of the peaks was observed upon impregnation of Co3O4 with PdO or PtO. The spent catalysts also exhibit peaks due to the Co3+ and Co2+ ions, irrespective of promoter identity or loading. The Co*2p* XPS measurements show that the Co3O4 spinel is very stable during direct NO decomposition, which agrees with XRD measurements.

**Figure 8.** Co*2p* XPS profiles of fresh (**a**) PdO- and (**b**) PtO-promoted Co3O4. The pure Co3O4 spectra is presented for reference.

The Pd*3d* XPS spectra of the fresh and spent 2, 3, and 4 wt% PdO/Co3O4 are presented in Figure 9a,b. In general, Pd may exist as Pd0 (335.1–335.4 eV [39]), Pd2+ (336.8.1–337.2 eV or 336.3–336.8 eV [40–44]), Pd4+ (337.8–339.3 eV), or a combination thereof. All the PdO/Co3O4 catalysts exhibit peaks due to the Pd2+ and Pd4+ after calcination at all PdO loadings. However, XRD measurements show no peaks corresponding to PdO or PdO2 up to 3 wt% loading, which indicates that the PdO present on Co3O4 is in amorphous form and dispersed very well on the surface. In agreement with the above XRD analysis of the spent samples (see Figure 6a), XPS indicates PdO/Co3O4 catalysts exhibit peaks due to PdO and metallic Pd after direct NOx decomposition. These results suggest that some of the PdO reduced to metallic Pd during direct NOx decomposition, which corroborates the evidence from H2-TPR and XRD of the promotional effect of Pd on the activity of Co3O4 spinel catalysts. The intensity of the metallic Pd increases with increasing PdO loading from 2 to 3 wt%, however, the intensity of the metallic Pd peak decreases drastically with further PdO loading from 3 to 4 wt%. This is due to the formation of the separate bulk PdO phase in the spent 4 wt% PdO/Co3O4 sample, which is clearly observed in the spent 4 wt% PdO/Co3O4 Pd*3d* spectrum. The formation of a separate PdO phase leads to less reduction of PdO to metallic Pd during direct NOx decomposition and is the likely cause of the lower activity compared to the 3 wt% PdO/Co3O4 catalyst. These results agree with the conclusions made based on the spent XRD patterns (see Figure 5a), further strengthening the evidence that 3 wt% Pd is the optimum loading for promoting activity.

**Figure 9.** Pd3d XPS spectra of (**a**) fresh and (**b**) spent PdO-promoted Co3O4 catalysts.

Figure 10a,b present the Pt*4f* XPS spectra of the fresh and spent 2, 3, and 4 wt% PdO/Co3O4. All fresh and spent PtO/Co3O4 catalysts only exhibit peaks due to Pt2+ (72.3, 74.1 eV) at all Pt loadings [45]. There are no observed peaks due to either Pt4+ or metallic Pt0 in contrast to the

PdO/Co3O4 catalysts, wherein the PGM underwent significant changes in oxidation state with exposure to reaction conditions. Overall, XPS measurements show that the support Co3O4 is very stable during the promoter impregnation, as well as during the direct NO decomposition. PdO reduced to metallic Pd during the direct NO decomposition and improves the direct NO decomposition activity of Co3O4. On the other hand, PtO stays in an oxidized state (no metallic Pt formation) during the direct NO decomposition and exhibits less promotional effects compared to PdO.

#### 2.2.4. NO Adsorption Properties

The adsorption of NO and formation of surface intermediates is essential to establishing activity for direct NO decomposition. In situ FT-IR spectroscopy was performed on pure Co3O4 and the PdOand PtO-promoted Co3O4 catalysts to understand the interaction of NO with the catalyst during adsorption. In situ FT-IR measurements were collected during NO adsorption over the catalysts at 300 ◦C. Before NO adsorption, all the catalysts were pretreated at 350 ◦C in the presence of 10% O2 in a helium balance and cooled to 300 ◦C in the presence of helium. All the spectra collected were normalized with respect to the gas phase NO peak at 1874 cm−1. The in situ FTIR spectra of Co3O4, PdO- and PtO-promoted Co3O4 catalysts during NO adsorption are presented in Figure 11a,b. As shown in Figure 11a, little to no NO adsorption occurs over the pure Co3O4 spinel oxide at 300 ◦C, as no clear peaks were present relative to the noise level. Interestingly, impregnating PdO over Co3O4 leads to a formation of chelating surface nitrate intermediates (1577 and 1254 cm−1) during the NO adsorption [46]. The intermediate formation was observed for all the catalysts irrespective of loading and the intensity of the peak increases with PdO loading. On the other hand, PtO-promoted Co3O4 catalysts do not produce spectroscopically relevant amounts of intermediates during the NO adsorption and exhibit spectra similar to the pristine Co3O4 catalyst. The catalysts in the current study exhibit activity at temperatures ≥400 ◦C, however, at these temperatures, no spectroscopically relevant surface NOx species were observed by in situ FTIR (not shown). This observation indicates that neither the surface chelating nitrate nor any other surface NOx species is the most abundant reactive intermediate in the direct NO decomposition mechanism. The in situ FTIR results at 300 ◦C are, therefore, interpreted as a probe of the affinity of the NO reactant molecule to interact with the catalyst surface. In this interpretation, it is concluded that the presence of PdO improves the affinity of the catalyst to interact with NO compared to PtO or the pure Co3O4 support.

The direct NO decomposition measurements show that PdO promotes direct NO decomposition activity of Co3O4 much better compared to the PtO. The Co3O4 is a normal spinel with AB2O4 formula, where A (Td) sites are occupied by Co2+ ions and B (Oh) sites are occupied by Co3+ ions. According to the general mechanism proposed by Haneda et al. [18], initially NO adsorbs on the surface and decomposes into N and O. The oxygen atoms adsorb on Co2+ ions and are oxidized to Co3+. Then, Co3+ ions reduce back to Co2+ upon release of the oxygen as a product. Hence, NO adsorption and oxygen release (redox) properties are very important for direct NO decomposition. The *in-situ* FT-IR results reveal that PdO increases the affinity of the catalyst to form surface NOx species compared to PtO or a pure Co3O4 support. The H2-TPR studies in our work show that the Co3O4 reduction temperature is significantly decreased by the presence of dispersed PdO, thus suggesting a more facile reduction of Co3+ to Co2+ to release O2 during direct NOx decomposition is possible. The improvement in the NOx adsorption properties and ease of cobalt reduction explains the better direct NO decomposition activity of PdO catalysts compared to PtO catalysts.

**Figure 10.** Pt*4f* XPS profiles of (**a**) fresh and (**b**) spent PtO-promoted Co3O4 catalysts.

**Figure 11.** *In-situ* FT-IR spectra of Co3O4, PdO (**a**), and PtO (**b**) promoted Co3O4 catalysts during NO adsorption at 300 ◦C.

#### **3. Materials and Methods**

#### *3.1. Catalyst Synthesis*

Palladium and platinum promoted Co3O4 catalysts were synthesized using the wet impregnation method. Commercial Co3O4 was purchased from Sigma-Aldrich (St. Louis, MO, USA) (99.5% trace metal basis) and used as received without any further modification for the synthesis. In a typical synthesis procedure, 5 g of commercial Co3O4 were mixed with 50 mL of water. Then, the required quantity of palladium nitrate hydrate (Sigma-Aldrich), or tetraamine platinum (II) nitrate (Sigma-Aldric, 99.995% trace metal basis) was dissolved separately in deionized water and combined with the Co3O4 suspension. The mixture was heated to 80 ◦C with continuous stirring. The obtained powder was then dried in an oven at 120 ◦C for 12 h under air. Finally, the catalyst was calcined

at 400 ◦C for 4 h with a 1 ◦C/min ramp. Different loadings of palladium and platinum on Co3O4 (nominally 1 to 4 wt%) catalysts were prepared by varying the amount of palladium nitrate or platinum nitrate. For reference, the commercial Co3O4 support was also calcined at 400 ◦C for 4 h.

#### *3.2. Catalyst Characterization*

*X-ray diffraction*: X-ray powder diffraction (XRD) patterns were obtained using a Rigaku SmartLab X-ray diffractometer (Rigaku, The Woodlands, TX, USA) using Cu Kα radiation (1.5405 A). A glass holder was used to support the sample. The scanning range was from 10 to 80◦ (2θ) with a step size of 0.02◦ and a step time of 1 s. The XRD phases present in the samples were identified with the help of ICDD-JCPDS [31] data files.

*BET Surface Area Measurements*: The surface area of the PdO and PtO promoted Co3O4 materials were measured using a Micromeritics 3Flex surface characterization instrument (Micromeritics, Atlanta, GA, USA). N2 physisorption isotherms was conducted at −196 ◦C, and the surface area was measured by the BET method. Prior to the analyses, the samples were outgassed at 300 ◦C under vacuum (5 × <sup>10</sup>−<sup>3</sup> Torr) for 3 h.

*X-ray Fluorescence Measurements*: XRF was collected using a Rigaku ZSX, primus II X-ray spectrometer (Rigaku, The Woodlands, TX, USA). Impurities in the crystals were gained by X-ray fluorescence in operation of spectrometer in standard fewer modes with coverage of a full element. The amount of any elements and oxides particles was detected by the XRF experiment.

*H2-Temperature Programmed Reduction (H2-TPR) Measurements*: The redox properties of the PtO/Co3O4 and PdO/Co3O4 catalysts were studied using H2 temperature programmed reduction (H2-TPR) experiments. H2-TPR experiments were performed using a Micromeritics 3Flex surface characterization instrument (Micromeritics, Atlanta, GA, USA) equipped with a thermal conductivity detector. Before the experiment, the catalysts were preheated to 300 ◦C in the presence of 20% O2/He (30 mL/min). After the pretreatment, the temperature was decreased to 50 ◦C. The H2-TPR measurements were performed by heating the catalyst from 50 to 600 ◦C in the presence of 10% H2/Ar (30 mL/min).

*X-ray Photo Electron Spectroscopy:* The XPS measurements were performed using a PHI 5000 Versa Probe II X-ray photoelectron spectrometer (Physical Electronics, East Chanhassen, MN, USA) using an Al Kα source. Charging of the catalyst samples was corrected by setting the binding energy of the adventitious carbon (C*1s*) to 284.6 eV [47]. The XPS analysis was performed at ambient temperature and at pressures typically on the order of 10<sup>7</sup> Torr. Prior to the analysis, the samples were outgassed under vacuum for 30 min.

*In Situ FTIR Spectroscopy Measurements during NO Adsorption:* The NO adsorption properties were measured using in situ Fourier transform infrared (FTIR) spectroscopy. The Harrick High Temperature Cell with environmental (gas flow) and temperature control was used for in situ diffuse-reflectance FTIR spectroscopy. Spectra were recorded using a Thermo Scientific Nicolet 8700 Research FT-IR Spectrometer (Thermo Scientific Fidher, Waltham, MA USA) equipped with a liquid N2 cooled MCT detector. Spectra were obtained with a resolution of 2 cm−<sup>1</sup> and by averaging 64 scans. In situ diffuse-reflectance FTIR spectra were collected during NO adsorption at 300 ◦C. Prior to NO adsorption, the sample was first pretreated at 350 ◦C in 30 mL/min of 10% O2/He. The background spectrum (64 scans) was of the catalyst after cooling to 300 ◦C in 30 mL/min UHP He. Adsorption of NO was achieved by flowing 30 mL/min of 10,000 ppm NO over the catalyst for 25 min. Adsorption of NO proceeded for 25 min, while spectra were obtained every minute using a series collection. To compare peak intensities among different catalyst samples, the adsorption spectra were normalized to the NO gas phase peak at ~1876 cm<sup>−</sup>1.

#### *3.3. Direct NO Decomposition Measurements*

Direct NO decomposition measurements were performed in a fixed bed flow reactor. A gas mixture of ~1% NO in helium balance was used with a gas hourly space velocity of ~2100 h−<sup>1</sup> and in the temperature region of 450–650 ◦C. Before the reaction, catalysts were pretreated at 500 ◦C in the presence of 20% O2/He. After pretreatment, the bed temperature was decreased to 400 ◦C and direct NO decomposition measurements were collected. The measurements were performed at 400, 450, 550, and 650 ◦C, with 2 h of steady state at each temperature. The NO, N2O, and NO2 concentrations were analyzed with a FTIR detector (CAI 600 SC FTIR California Analytical Instruments, Inc., Orange County, CA, USA). The N2 concentration was calculated by mass balance of the total nitrogen species. The raw NO conversion (NO converted to all the products) during the steady state measurements are presented in Figure S1 of the supporting information and activity to N2 was reported in Figure 1 of the manuscript. The steady state direct NO decomposition measurements were also performed in a reactor, which was equipped with the mass spectroscopy (MKs, Cirrus 2). The changes in the NO, N2, O2, N2O, and NO2 signal intensities were monitored during the reaction (Figures S2–S5). The inert Ar gas was introduced as a tracer to monitor for potential systematic variation in signal intensity during the experiment (Figure S2).

#### **4. Conclusions**

The direct NO decomposition measurements show that PdO promotes the activity of Co3O4 and is 4 times more active compared to PtO at 650 ◦C. Also, the activity increases with increasing PdO loading until 3 wt% and further increase in the loading leads to a decrease in the activity. On the other hand, only a slight increase in the activity was observed with increasing PtO loading up to 4 wt%. Surface area measurements indicated that both PdO and PtO have little to no influence on the surface area of Co3O4, except for a decrease in surface area for 4 wt% PdO/Co3O4. The X-ray diffraction measurements show that Pt incorporated into the Co3O4 structure during the synthesis and PdO stays mostly on the surface. The diffraction measurements also suggested that PdO is in an amorphous form up to 3 wt% over Co3O4 surface and crystalline PdO forms at 4 wt% loading, whereas PtO mostly stays as amorphous from or incorporated into Co3O4 structure until 4 wt%. Due to the synergistic effect between Pd species and Co3O4, an improvement in the redox properties of Co3O4 was observed in the case of PdO/Co3O4 catalysts. Conversely, PtO do not have any influence on the redox properties of Co3O4. The X-ray photo electron spectroscopic measurements reveal that PdO reduced to Pd metal during the direct NO decomposition reaction and Pt was in 2+ oxidation state before and after the direct NO decomposition reaction. In situ NO adsorption measurements show that PdO improve the NO adsorption properties of Co3O4 by forming the nitrate ion intermediates, whereas PtO/Co3O4 do not form any intermediates during the NO adsorption at 300 ◦C. Overall, PdO ease the redox properties of Co3O4 and forms surface adsorbed species during NO adsorption and improves the direct NO decomposition activity of Co3O4. On the other hand, PtO do not have any influence on the redox or NO adsorption properties of Co3O4 and exhibits lesser promotional effects compared to PdO. For PdO/Co3O4 catalysts, the PdO remains in amorphous form until 3PdO/Co3O4 and improves the activity of Co3O4 with loading. However, further increase in the loading to 4 wt% leads to formation of crystalline PdO, which reduces separately during H2-TPR and inhibits the PdO reduction to metallic Pd during direct NO decomposition and exhibits lesser activity compared to 3 wt% PdO/Co3O4.

Supporting information: The NO conversion, and N2, N2O, and NO2 ppm values of various PdO/Co3O4 and PtO/Co3O4 catalysts are presented in Tables S1 and S2. The total NO conversion profiles of the Co3O4, 3PdO/Co3O4, and 4PtO/CoO calculated from the FT-IR detector during the steady state direct NO decomposition measurements are presented in Figure S2. NO, Ar, N2, O2, N2O, and NO2 partial pressure values (obtained from mass spectroscopy) of Co3O4, 3 wt% PdO/Co3O4, and 3 wt% PtO/Co3O4 during steady state direct NO decomposition in the 400 to 650 ◦C temperature region are presented in Figures S3 and S4.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/9/1/62/s1. Table S1: NO conversion, N2, N2O, NO2 ppm values of the Co3O4 and various PdO/Co3O4 catalysts in the temperature region 400–650 ◦C. Table S2: NO conversion, N2, N2O, NO2 ppm values of the Co3O4 and various PtO/Co3O4 catalysts in the temperature region 400–650 ◦C. Figure S1: Steady state NO conversion values of

the Co3O4, 3PdO/Co3O4, and 4PtO/Co3O4 catalysts during the direct NO decomposition in the temperature region 400 to 650 ◦C. Figure S2: NO and Ar M.S. partial pressures of the (**a**) Co3O4, (**b**) 3PdO/Co3O4, and (**c**) 3PtO/Co3O4 catalysts during the steady state direct NO decomposition in the temperature region 400 to 650 ◦C. Figure S3: N2 and O2 M.S. partial pressures of the (**a**) Co3O4, (**b**) 3PdO/Co3O4, and (**c**) 3PtO/Co3O4 catalysts during the steady state direct NO decomposition in the temperature region 400 to 650 ◦C. Figure S4: N2O M.S. partial pressures of the (**a**) Co3O4, (**b**) 3PdO/Co3O4, and (**c**) 3PtO/Co3O4 catalysts during the steady state direct NO decomposition in the temperature region 400 to 650 ◦C. Figure S5: NO2 M.S. partial pressures of the (**a**) Co3O4, (**b**) 3PdO/Co3O4, and (**c**) 3PtO/Co3O4 catalysts during the steady state direct NO decomposition in the temperature region 400 to 650 ◦C.

**Author Contributions:** This study was conducted through contributions of all authors. G.K.R. Reddy designed the study, performed the experiments, and wrote the manuscript. T.C.P. was involved in performing the experiments. C.A.R. checked and corrected the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors thank Hongfei Jia from Toyota Research Institute—North America and Naoto Nagata from Toyota Motor Corp. for their support.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Investigation of Various Pd Species in Pd/BEA for Cold Start Application**

#### **Beibei Zhang 1, Meiqing Shen 1,2,3, Jianqiang Wang 1, Jiaming Wang <sup>4</sup> and Jun Wang 1,\***


Received: 16 February 2019; Accepted: 4 March 2019; Published: 7 March 2019

**Abstract:** A series of Pd/BEA catalysts with various Pd loadings were synthesized. Two active Pd2+ species, Z−-Pd2+-Z<sup>−</sup> and Z−-Pd(OH)+, on exchanged sites of zeolites, were identified by in situ FTIR using CO and NH3 respectively. Higher NO*<sup>x</sup>* storage capacity of Z−-Pd2+-Z<sup>−</sup> was demonstrated compared with that of Z−-Pd(OH)+, which was caused by the different resistance to H2O. Besides, lower Pd loading led to a sharp decline of Z−-Pd(OH)+, which was attributed to the 'exchange preference' for Z<sup>−</sup>-Pd2+-Z<sup>−</sup> in BEA. Based on this research, the atom utilization of Pd can be improved by decreasing Pd loading.

**Keywords:** Pd/BEA; Cold start; Pd species; NOx abatement

#### **1. Introduction**

The exhaust regulation on NO*<sup>x</sup>* emissions is getting more stringent [1]. Presently, NH3 selective catalytic reduction (SCR) [2] and NO*<sup>x</sup>* storage reduction (NSR) [3] are widely used for NO*<sup>x</sup>* removal. However, standard NO*<sup>x</sup>* aftertreatment technologies fail to function efficiently at low temperatures, which results in a large proportion of the tailpipe NO*x* emission [4]. Meanwhile, high efficiency internal combustion engines require new and/or improved technologies which specifically address their low exhaust temperatures. In response to difficulties of low temperature emissions control, numerous efforts are underway to develop catalysts that light-off at temperatures below 150 ◦C. Passive NO*<sup>x</sup>* adsorbers (PNAs) could play a critical role in enabling high efficiency advanced combustion systems.

Recently, Pd/zeolite serving as passive NO*x* adsorbers (PNAs) was first proposed by Chen et al. [4]. This catalyst is emerging as effective passive NO*<sup>x</sup>* adsorbent technology because of its NO*<sup>x</sup>* storage/release capabilities, resistance to sulfur poisoning and hydrothermal deactivation [4–8]. Due to these excellent characteristics, Pd/zeolite has attracted great interest recently and has been further optimized [6–18].

Several recent studies show that isolated Pd ions in Pd/zeolite are the main active sites for NO*<sup>x</sup>* trapping [6,7,11,13,16]. It is reported that there are nine skeletal T sites with various chemical environments in BEA [19]. So, isolated Pd ions located in various positions of zeolite framework are likely to be formed. As reported by Gao et al. and Giordanino et al. [20–22], two kinds of isolated Cu species (Cu2+ and [Cu(OH)]+) are identified on various framework positions of Cu/SSZ-13, which indicates that species of isolated cations may be influenced by their locations in zeolites. They also mentioned that Cu species were significantly affected by Cu loading. So, various isolated Pd species may co-exist in Pd/BEA, and Pd loading may be capable of influencing the content of them. Actually, two kinds of active isolated Pd ions (bare Pd2+ and Pd(OH)+) have been observed by Zheng et al. [10]

and Khivantsev et al. [12]. However, they did not point out the difference in adsorption between these two species.

Therefore, these two kinds of isolated Pd ions were further studied in this research. A series of in situ FTIR experiments in CO and NH3 were carried out, and a semiquantitative method was adopted to distinguish these two species. Based on this method, the NO*<sup>x</sup>* storage capacity of each isolated Pd species was compared. Further, by increasing the proportion of isolated Pd species with higher NO*<sup>x</sup>* storage capacities, the atom utilization of Pd can be improved.

#### **2. Results**

#### *2.1. Ex-Situ FTIR*

Ex-situ FTIR spectra of each sample are exhibited in Figure 1. Compared with 0-Pd (Figure 1a), an extra vibration at 926 cm−<sup>1</sup> is observed on the FTIR spectrum of 1-Pd. This peak is attributed to the vibration distortion of the skeletal T–O–T bond due to the strong interaction of Pd ions [4]. This is an obvious piece of evidence for the existence of Pd ions on the exchange sites of zeolites. Peaks at 926 cm−<sup>1</sup> appear on spectra of 0.2-Pd, 0.5-Pd and 2-Pd, too (Figure 1b). So, the existence of Pd ions on exchange sites in these samples is confirmed.

**Figure 1.** (**a**) Ex-situ FTIR spectra of 0-Pd, 1-Pd, Na-1-Pd; (**b**) Ex-situ FTIR spectra of 2-Pd, 1-Pd, 0.5-Pd, 0.2-Pd (Temperature: 200 ◦C; Flow: N2, 1 L/min).

Compared with 1-Pd (Figure 1a), the peak at 926 cm−<sup>1</sup> disappears in the spectrum of Na-1-Pd, which indicates the elimination of isolated Pd ions on exchange sites. This is firm evidence of which isolated Pd ions in Na-1-Pd have been completely exchanged by Na<sup>+</sup> ions through the titration process. So, Pd loaded on Na-1-Pd mainly exists in form of Pd oxidations.

#### *2.2. Na<sup>+</sup> Titration*

Na+ titration was adopted to measure the content of isolated Pd ions in each sample [23], and the proportion of isolated Pd ions in total Pd loading (marked as Isolated Pd/Pd loading in Table 1) was further calculated. Considering that Pd is sensitive to Cl [24,25], the titration process was carried out in a NaNO3 solution. Besides, this measurement is believed to have no negative effect on zeolite structures since the titration process was carried out in mild conditions and no calcination was done. As shown in Table 1, lower Pd loading leads to less isolated Pd2+ content, whereas the proportion of isolated Pd ions in total Pd loading becomes larger. This phenomenon indicates that the increase of Pd loading leads to the formation of more Pd oxidations (PdOx), which is consistent with Jaeha Lee et al.'s study [11].


**Table 1.** The content of Pd ions on exchange sites of each sample.

#### *2.3. CO In Situ FTIR*

Pd species were further probed by CO using in situ FTIR, and the result is displayed in Figure 2. Peaks below 2100 cm−<sup>1</sup> are attributed to CO signals on metallic Pd (Pd0) formed via CO reduction [10,26], among which Pd0−CO (atop) are found at 2098 cm<sup>−</sup>1, 2076 cm−<sup>1</sup> and Pd2 <sup>0</sup>−CO (bridging) are found at 1951 cm−1. Besides, the peak at 2117 cm−<sup>1</sup> is attributed to the C-O vibration on Pd<sup>+</sup> [10,27]. As reported by Vu et al. [9], Pd<sup>+</sup> is formed by the CO reduction of ion-exchanged Pd species [9]. CO signals on isolated Pd2+ species are observed above 2100 cm<sup>−</sup>1. The peak at 2152 cm−<sup>1</sup> with a shoulder peak at 2137 cm−<sup>1</sup> is attributed to the C-O vibration on isolated Pd2+ bonded with the hydroxy of zeolites (marked as Z<sup>−</sup>-Pd2+-Z<sup>−</sup>) [10]. The peak at 2179 cm−<sup>1</sup> is attributed to the vibration of C-O adsorbed by another kind of isolated Pd2+ [10], which was first determined as Z<sup>−</sup>-Pd(OH)+ by Okumura et al. [28].

**Figure 2.** CO in situ FTIR spectra of 2-Pd, 1-Pd, 0.5-Pd, 0.2-Pd (Temperature: 80 ◦C; Flow: 1000 ppm CO, balanced with N2, 500 mL/min).

In short, two kinds of isolated Pd2+ (Z−-Pd2+-Z<sup>−</sup> and Z−-Pd(OH)+) were identified in 0.2-Pd, 0.5-Pd, 1-Pd and 2-Pd. Note that all spectra in Figure 2 were obtained when the steady state had been achieved, and corresponding peaks of Z<sup>−</sup>-Pd2+-Z<sup>−</sup> and Z<sup>−</sup>-Pd(OH)+ were still observed. So, these two

isolated Pd2+ species cannot be completely reduced by CO at this temperature, which indicated that the reduction reaction is a reversible one. As shown in Figure S1, the existence of Z<sup>−</sup>-Pd2+-Z<sup>−</sup> in 1-Pd-80 is also demonstrated by CO whereas no Z−-Pd(OH)<sup>+</sup> is observed. Since the complete reduction of Z−-Pd(OH)<sup>+</sup> cannot be achieved by CO, there is only one isolated Pd2+ species, Z−-Pd2+-Z<sup>−</sup>, in 1-Pd-80 (see Figure S1).

#### *2.4. Catalyst Evaluation*

Profiles of the NO adsorption stage are shown in Figure 3. NO*x* storage capacities are calculated by the integration of negative peaks on these profiles, and the dead volume has been subtracted. The result is displayed in Table 2. Note that samples with the same Pd loading as much as 1 wt % (see Figure S2) exhibit entirely different NO*<sup>x</sup>* storage capacities (53.3 μmol/gcat and 9.9 μmol/gcat respectively), which indicates that only part of Pd loaded on samples is efficient. As shown in Figure 3, NO*x* storage capacities of 0-Pd and Na-1-Pd are very low. So, Brønsted hydroxyl group and PdOx are inefficient active centers for NO*x* storage, whereas they do trap NO*x* in this condition as reported [10,27]. Since the NO*<sup>x</sup>* storage capacity of 1-Pd is much higher, isolated Pd2+ ions are likely to be the main active sites for NO trapping.

**Figure 3.** NO*<sup>x</sup>* adsorption profiles (Temperature: 80 ◦C; Flow: NO*x*, CO, H2O, O2, CO2, balanced with N2, 1 L/min).

**Table 2.** NO*<sup>x</sup>* storage capacities in NO*x*, CO, H2O, O2 and CO2.


#### *2.5. NH3 In Situ FTIR*

Acidity over samples is probed by NH3 with in situ FTIR, and the result is exhibited in Figure 4. The peak at 1463cm−<sup>1</sup> observed in both spectra of 1-Pd and 0-Pd (Figure 4a) is attributed to the vibration of NH4+ in Brønsted hydroxyl groups (NH4+-B) [29]. In the spectrum of 1-Pd, two additional peaks at 1625 cm−<sup>1</sup> and 1313cm−<sup>1</sup> corresponding to the vibration of NH3 on Lewis acid (NH3-L) [30,31] is observed. Nevertheless, these two peaks do not exist in Na-1-Pd in which isolated Pd2+ ions (Z−-Pd2+-Z<sup>−</sup> and Z−-Pd(OH)+) are replaced by Na+ completely. So, it is reasonable to believe that Lewis acid is generated from Z<sup>−</sup>-Pd2+-Z<sup>−</sup> and Z−-Pd(OH)<sup>+</sup> species.

**Figure 4.** (**a**) NH3 in situ FTIR spectra of 0-Pd, 1-Pd, Na-1-Pd; (**b**) NH3 in situ FTIR spectra of 2-Pd, 1-Pd, 0.5-Pd, 0.2-Pd (Temperature: 80 ◦C; Flow: 500 ppm NH3, balanced with N2, 500 mL/min).

As shown in Figure S3, there is only one peak at 1625cm−<sup>1</sup> corresponding to the vibration of NH3-L observed in the spectrum of 1-Pd-80. Since there is only one kind of isolated Pd2+, Z<sup>−</sup>-Pd2+-Z<sup>−</sup>, in this sample as discussed above, the peak at 1625cm−<sup>1</sup> should be assigned to the vibration of NH3-L originated from Z−-Pd2+-Z<sup>−</sup>. In this case, the peak at 1313cm−<sup>1</sup> should be attributed to the vibration of NH3-L originating from Z<sup>−</sup>-Pd(OH)+.

Since Z−-Pd2+-Z<sup>−</sup> and Z<sup>−</sup>-Pd(OH)+ are capable of being probed by NH3, the ratio of the height of the peaks at 1625 cm−<sup>1</sup> (PZ−-Pd2+-Z<sup>−</sup>) and 1313 cm−<sup>1</sup> (PZ−-Pd(OH)+) in Figure 4b can be defined as the relative content between these two isolated Pd2+ species. Considering that extinction coefficients for NH3 molecules adsorbed on Z−-Pd2+-Z<sup>−</sup> and Z<sup>−</sup>-Pd(OH)+ are both constants, they have no effect on the tendency of peak height ratios. So, the extinction coefficient is not considered in this part [32–34]. The result is shown in Figure 5a. It is worth nothing that PZ−-Pd(OH)+/PZ−-Pd2+-Z<sup>−</sup> rises in parallel with the increase of isolated Pd2+, which means that the higher content of isolated Pd2+ leads to a much more obvious increase of Z−-Pd(OH)+ than that of Z−-Pd2+-Z−. So, isolated Pd2+ in 0.2-Pd mainly exists in the form of Z−-Pd2+-Z−, whereas a large amount of Z−-Pd(OH)+ ions are formed in 2-Pd.

**Figure 5.** (**a**) tendency between isolated Pd2+ and PZ−-Pd(OH)+/PZ−-Pd2+-Z−; (**b**) tendency between PZ−-Pd(OH)+/PZ<sup>−</sup>-Pd2+-Z<sup>−</sup> and NO*x*/Pd2+.

Mole ratios of NO*<sup>x</sup>* adsorbed to isolated Pd2+ for each sample (marked as NO/Pd2+) are calculated via data in Tables 1 and 2. Figure 5b is plotted by PZ−-Pd(OH)+/PZ−-Pd2+-Z<sup>−</sup> on the horizontal axis and NO*x*/Pd2+ on the vertical. As reported, NO*<sup>x</sup>* is believed to be trapped in the form of Pd +-NO [12,15], so the maximum NO*x*/Pd2+ ratio is 1 in theory. However, the downward trend between PZ−-Pd(OH)+/PZ<sup>−</sup>-Pd2+-Z<sup>−</sup> and NO*x*/Pd2+ means that higher PZ−-Pd(OH)+/PZ<sup>−</sup>-Pd2+-Z<sup>−</sup> leads to lower NO*<sup>x</sup>* storage capacity of isolated Pd2+, which indicates that the NO*<sup>x</sup>* storage capacity of Z<sup>−</sup>-Pd(OH)+ is obviously lower than that of Z<sup>−</sup>-Pd2+-Z<sup>−</sup> in this condition.

A further discussion is given below so as to give a reasonable explanation from the perspective of the structure-function relationship.

In BEA zeolite, nine skeletal T sites with various chemical environments are determined [19]. Among these T sites, T5 and T6 are demonstrated to have the lowest Al substitution energy [35]. Thus, Al substitution on these two sites (marked as AlT5 and AlT6) is preferential, and the amount of AlT5 and AlT6 is larger than that of Al atoms on the other T sites in BEA. Besides, exchange sites formed on these two Al atoms are more active due to the lowest deprotonation energy [36] of them. Meanwhile, AlT5 and AlT6 are located in the meta-position of the same five-membered ring [37], and the co-ion-exchange of protons on these two exchange sites formed on AlT5 and AlT6 is feasible. These characteristics of BEA can explain the 'exchange preference' for Z−-Pd2+-Z<sup>−</sup> formed in Pd/BEA with lower content of isolated Pd ions to some extent. Besides, with the increase of Pd loading, more exchange sites are taken up. Limited by the amount of protons which are capable of being exchanged, Z<sup>−</sup>-Pd(OH)+ ions, which take up less exchange sites than Z−-Pd2+-Z<sup>−</sup>, are preferred to be formed. So, higher content of isolated Pd2+ leads to much more obvious increase of Z<sup>−</sup>-Pd(OH)+ than that of Z−-Pd2+-Z<sup>−</sup>.

As discussed above, the maximum NO*x*/Pd2+ ratio is 1 in theory. As reported by Khivantsev et al. [12], H2O molecules can occupy NO*<sup>x</sup>* storage sites due to strong hydration of isolated Pd2+ species. Since isolated Pd2+ is the main active site for NO*<sup>x</sup>* storage, the NO*x*/Pd2+ ratio of Pd/zeolite-based PNA materials will significantly smaller than 1 in the presence of H2O. However, as shown in Figure 5b, NO*x*/Pd2+ of 0.2-Pd is as much as 1, which indicates that isolated Pd2+ in this sample is not occupied by H2O. As discussed above, isolated Pd2+ in 0.2-Pd mainly exists in the form of Z−-Pd2+-Z−. Thus Z−-Pd2+-Z<sup>−</sup> may be insensitive to H2O in this condition. Furthermore, Z−-Pd(OH)<sup>+</sup> probably tends to hydrate due to the hydrogen bond interaction between H2O and hydroxy. Accordingly, the difference in NO*<sup>x</sup>* storage capacities of Z−-Pd(OH)+ and Z−-Pd2+-Z<sup>−</sup> is probably caused by the different resistance to H2O in this condition.

The atom utilization of Pd can be represented by mole ratios of NO*x* adsorbed to total Pd loading. According to Figure 6, the Pd utilization of 0.2-Pd is as much as 100% whereas that of 2-Pd is only 25%. So, it is obvious that lower Pd loading benefits to the increase of Pd utilization. Besides, Khivantsev et al. [13] have reported that the Pd utilization of the 1 wt% Pd/SSZ-13 (Si/Al ratio = 6) sample is 100%, which is much higher than that of 1-Pd (Si/Al ratio = 16) using BEA as a support. Accordingly, zeolite structure and Si/Al ratio can also influence the Pd utilization. In this part, only the influence of Pd loading will be discussed.

As discussed above, lower Pd loading leads to lower content of isolated Pd2+, whereas the proportion of isolated Pd ions in total Pd loading becomes larger. On the on hand, a larger proportion of isolated Pd ions in total Pd loading is beneficial for the improvement of Pd utilization, since isolated Pd ions are identified as the main active center for NO*<sup>x</sup>* storage. On the other hand, less isolated Pd2+ results in a preference for the formation of Z−-Pd2+-Z<sup>−</sup> which exhibits higher NO*<sup>x</sup>* storage capacity. Thus, the atom utilization of Pd can be improved by decreasing Pd loading.

Nevertheless, lower Pd loading also leads to less NO*x* storage capacities of unit mass of Pd/BEA. Note that the coating amount of catalysts is limited in order to obtain acceptable back pressure. Accordingly, to determine optimum Pd loading for low temperature NO*x* adsorption, the NO*x* storage capacity and the atom utilization of Pd should be both considered. As shown in Figure 6, 0.2-Pd and 0.5-Pd exhibits much higher Pd atom utilization than that of the other two samples. Meanwhile, Pd atom utilization of 0.5-Pd is 92%, which is only slightly lower than that of 0.2-Pd. However, the NO*<sup>x</sup>* storage capacity of 0.5-Pd is larger than twice of that of 0.2-Pd. So, 0.5wt% should be determined as the optimum Pd loading for Pd/BEA (Si/Al ratio = 16) served as PNA material.

**Figure 6.** Tendency between Pd loading and the atom utilization of Pd.

#### **3. Materials and Methods**

#### *3.1. Catalyst Preparation*

H-BEA zeolite (Si/Al = 16.2) was supplied by Novel Chemistry. Pd/BEA samples with various Pd loading were prepared by incipient wetness impregnation, and Pd(NO3)2 solution (15.47wt% Pd, Heraeus Materials Technology Shanghai Ltd., Shanghai, China) was used. Then, the powders obtained were dried under ambient temperature followed by a 4-h calcination at 550 ◦C in air. Finally, samples were stabilized at 750 ◦C for 12 h in air with 10% H2O. Pd loading of each sample was detected by inductively coupled plasma (ICP) analysis (USA Agilent 5100 ICP-OES, Santa Clara, CA, USA). Si/Al ratios were measured by X-ray fluorescence (XRF, ThermoFisher PERFORM'X, Waltham, MA, USA) analysis. Detailed information is exhibited in Table 3. In the following text, samples are abbreviated as *x*-Pd, where "*x*" represents the Pd loading of corresponding samples. Besides, the sample treated by Na<sup>+</sup> titration was named as Na-1-Pd.


**Table 3.** Pd loading of samples.

#### *3.2. Catalyst Characterization*

Na<sup>+</sup> titration was used to quantify isolated Pd ions as reported by Ogura et al. [23]. Each sample was mixed with NaNO3 (Purity above 99.0%, Tianjin Jiangtian Chemical Technology Co., Ltd., Tianjin, China) solution (0.1M) at 80 ◦C. The solution was stirred for 4 h followed by suction filtration. At last, the powders obtained were washed by deionized water and dried at 100 ◦C for 6 h. The whole process above was repeated three times. By analyzing the change of Pd loading in each sample after the titration process, the content of isolated Pd can be measured.

NO*x* storage capacities were measured by standard catalyst evaluation tests carried out in a plug flow reactor system. 0.25 g sample (60–80 mesh) mixed with 0.75 g quartz (60–80 mesh) was loaded in a quartz reactor with a thermocouple. The sample was oxidized in the flow of 10% O2/N2 for 30 min at 500 ◦C and was cooled to 80 ◦C in N2. Then the flow (200 ppm NO*x*, 200 ppm CO, 5% CO2, 5%H2O, 10% O2, balanced with N2) mixed in the bypass in advance was introduced into the reactor. The NO*<sup>x</sup>*

adsorption stage continued for 3 min. After that, the sample was heated to 500 ◦C with a ramping rate of 10 ◦C/min in the flow of N2. Gas concentrations were measured by an online MKS MultiGas 2030 FTIR gas analyzer in the whole process. Besides, a space velocity of 28,800 h−<sup>1</sup> was adopted.

Ex-situ FTIR was carried out on a Nicolet iS10 FTIR spectrometer equipped with a liquid N2 cooled mercury cadmium telluride (MCT) detector to identify the existence of Pd ions. The sample (20 mg) was pressed into a self-supporting wafer with a diameter of 13 mm and was inserted into a cell sealed with ZnSe windows connected with a gas manifold. KBr was used to obtain background spectra. All samples, as well as KBr, were oxidized in the flow of 10% O2/N2 for 30 min at 500 ◦C in advance. Spectra were obtained at 200 ◦C in N2.

In situ FTIR was carried out on a Nicolet iS10 FTIR spectrometer, too. Test temperatures and feed compositions varied according to the needs of different experiments, and detailed information will be reported below.

#### **4. Conclusions**

In this work, two isolated Pd2+ ions, Z−-Pd2+-Z<sup>−</sup> and Z−-Pd(OH)+, on exchange sites of zeolites are confirmed as the main active sites for NO trapping in cold-start applications. Lower Pd loading leads to a lower content of isolated Pd2+ whereas the proportion of isolated Pd ions in total Pd loading becomes larger. Lower content of isolated Pd2+ further leads to a sharp decline of Z<sup>−</sup>-Pd(OH)+, which is attributed to the 'exchange preference' for Z−-Pd2+-Z<sup>−</sup> in BEA. Besides, the higher NO*<sup>x</sup>* storage capacity of Z−-Pd2+-Z<sup>−</sup> is demonstrated compared with that of Z<sup>−</sup>-Pd(OH)+, which is caused by the different resistance to H2O. In conclusion, the atom utilization of Pd can be improved by using lower Pd loading. 0.5wt% should be determined as the optimum Pd loading for Pd/BEA (Si/Al ratio = 16) serving as PNA material among all samples.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/9/3/247/s1, Table S1: Detailed information of 1-Pd-80, Figure S1: CO in situ FTIR spectra of 1-Pd-80, Figure S2: NO*x* adsorption profiles of 1-Pd and 1-Pd-80, Figure S3: NH3 in situ FTIR spectra of 1-Pd-80.

**Author Contributions:** Conceptualization, B.Z.; methodology, J.W. (Jun Wang), M.S., and J.W. (Jianqiang Wang); software, B.Z.; validation, J.W. (Jianqiang Wang); formal analysis, J.W. (Jianqiang Wang); investigation, B.Z and J.W. (Jiaming Wang).; writing—original draft preparation, B.Z.; writing—review and editing, B.Z.; visualization, B.Z.; supervision, J.W. (Jun Wang); resources, M.S.; project administration, M.S.; funding acquisition, M.S.

**Funding:** This research was funded by National Key Research and Development Program, grant number 2017YFC0211002; State Key Laboratory of Advanced Technologies for Comprehensive Utilization of Platinum Metals, grant number SKL-SPM-2018017.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Optimization of Ammonia Oxidation Using Response Surface Methodology**

#### **Marek Inger \*, Agnieszka Dobrzy ´nska-Inger, Jakub Rajewski and Marcin Wilk**

New Chemical Syntheses Institute, Al. Tysi ˛aclecia Pa ´nstwa Polskiego 13a, 24-100 Puławy, Poland; agnieszka.dobrzynska-inger@ins.pulawy.pl (A.D.-I.); jakub.rajewski@ins.pulawy.pl (J.R.); marcin.wilk@ins.pulawy.pl (M.W.)

**\*** Correspondence: marek.inger@ins.pulawy.pl; Tel.: +48-(81)-473-1415

Received: 22 December 2018; Accepted: 4 March 2019; Published: 9 March 2019

**Abstract:** In this paper, the design of experiments and response surface methodology were proposed to study ammonia oxidation process. The following independent variables were selected: the reactor's load, the temperature of reaction and the number of catalytic gauzes, whereas ammonia oxidation efficiency and N2O concentration in nitrous gases were assumed as dependent variables (response). Based on the achieved results, statistically significant mathematical models were developed which describe the effect of independent variables on the analysed responses. In case of ammonia oxidation efficiency, its achieved value depends on the reactor's load and the number of catalytic gauzes, whereas the temperature in the studied range (870–910 ◦C) has no effect on this dependent variable. The concentration of nitrous oxide in nitrous gases depends on all three parameters. The developed models were used for the multi-criteria optimization with the application of desirability function. Sets of parameters were achieved for which optimization assumptions were met: maximization of ammonia oxidation efficiency and minimization of the N2O amount being formed in the reaction.

**Keywords:** ammonia oxidation; response surface methodology; desirability function; Box-Behnken design

#### **1. Introduction**

Nitric acid is mainly used for producing nitrogen fertilizers: ammonium nitrate (AN) and calcium ammonium nitrate (CAN) which constitute 75–80% of its entire production. The remaining amount of nitric acid is used in other industrial applications for example as a nitration agent for the production of explosives and other semi-organic products (aliphatic nitro compounds and aromatic nitro compounds) for the production of adipic acid, for metallurgy (etching steel) [1].

The industrial production of nitric acid is based on Ostwald process [1] which involves three basic stages: the catalytic oxidation of ammonia to nitrogen oxide (NO) with the use of oxygen from air, oxidation of nitrogen oxide (NO) to nitrogen dioxide (NO2) and absorption of nitrogen oxides in water with the formation of HNO3.

Ammonia consumption depends on the selectivity of the applied ammonia oxidation catalyst and on the process conditions. Among numerous catalysts [2–8], packages of gauzes made of noble metal alloys such as platinum and rhodium are most commonly applied in industrial practice [7–10]. Properly selected catalyst package allows to obtain ammonia conversion to main product (NO) in the range of 90–98% depending mainly on oxidation pressure [1,7,8]. Oxidation pressure has an inversely proportional effect on ammonia oxidation efficiency. In order to alleviate this effect, the temperature of reaction should be higher. However, this leads to the increased platinum losses and as a consequence, shortens the lifetime of the catalytic gauzes. For example, platinum losses are six times higher after increasing the temperature of reaction from 820 to 920 ◦C [1,2]. Therefore, both these aspects should be taken into account to determine the temperature of reaction.

The application of medium pressure in the oxidation unit (0.35–0.55 MPa) and high pressure in the absorption unit (0.8–1.5 MPa) is optimal for specific ammonia consumption and efficient energy use. Therefore, modern nitric acid plants are dual-pressure ones. The average pressure in the oxidation unit is a kind of trade-off between the capacity that is possible to achieve per 1 m<sup>2</sup> of catalytic gauzes, oxidation efficiency, number of gauzes in package, lifetime of gauzes and noble metals losses during exploitation [1,2].

In the context of global warming and climate changes, a very important issue related to ammonia oxidation process is the amount of the by-product formed that is nitrous oxide (N2O). In Kyoto Protocol, N2O was qualified as a greenhouse gas with a very high global warming potential, about 300 times higher than CO2 [11]. At room temperature, N2O is a colourless, non-flammable gas with a delicate pleasant smell and sweet taste [12]. Since it was isolated at the end of 17th century and because of its pain-relieving and anaesthetic properties, it has been widely applied in dentistry and surgery. Currently, due to some concerns, there is an ongoing discussion on its safe use which has the effect of decreasing the N2O application in medicine [12,13]. At the same time, the increasing trend of its use for recreational purposes is observed. Inhaling the 'laughing gas' causes euphoria and hallucinations [13].

Microbial nitrification and de-nitrification in land and aqueous eco-systems are the natural sources of N2O in environment. The anthropogenic sources are cultivated soils fertilized intensely with nitrogen fertilizers and industrial processes such as burning fossil fuels and biomass as well as the production of adipic acid and nitric acid with the last one being regarded as the biggest source of N2O in the chemical industry [14,15]. Nitrous oxide formed in nitric acid plant does not undergo any conversions and it is released to atmosphere. Currently, the emission of this gas is monitored and industry is obliged to reduce it. Pursuant to BAT requirements, concentration of this gas in outlet gases cannot exceed 20–300 ppm depending on the type of nitric acid plants [16,17]. However, due to the battle against climate change and global warming, further restrictions in emission limits can be expected.

There are several methods of limiting N2O emissions from nitric acid plants [14]. Generally, they can be classified as primary and secondary methods. Primary methods involve preventing the formation of N2O during ammonia oxidation. They include modification of catalytic gauzes (so-called low-emissions systems) and parameters optimization of ammonia oxidation process. Secondary methods involve the removal of N2O. At the temperature over 800 ◦C, thermal decomposition of N2O occurs but the efficient decomposition requires ensuring adequately long residence time at high temperatures [14,15].

The achievement of low level of N2O emissions requires the application of the catalytic methods such as high temperature N2O decomposition from nitrous gases, low- or middle temperature N2O decomposition or reduction from tail gases. High temperature method is more common. In some cases, the combination of primary method (application of modified catalytic gauzes packages and/or optimization of ammonia oxidation parameters) and high temperature N2O decomposition ensures meeting the emission standards.

Optimization of production process requires extensive knowledge and understanding the effect of particular parameters on the process. Until recently, the most commonly applied approach of researchers to study simple and complex processes was 'one-factor-at-a-time' (*OFAT*), which is time consuming and ineffective method for processes with multiple complicated dependencies between parameters. Over the last years, mathematical and statistical methods for design of experiments and parameters optimization have been applied more frequently [18]. Because of its usability, this method is applied for the design, improvement and optimization of production processes and products [19–21]. It is a widely applied method in research of various processes [22–27] and approx. 50% of all applications is in medicine, engineering, biochemistry, physics and computer science [28]. In this method, reaction kinetics equations and process mechanism are not taken into account and they are regarded as a 'black-box' [29] (Figure 1).

**Figure 1.** "Black–box" model of the research issue in design of experiments methodology.

The choice of experiment plan depends mainly on the issue which is the subject matter of investigations as well as on objectives which are set. The most commonly applied experiments plans include: full or fractional factorial, Plackett-Burman, central composite, Box-Behnken and Taguchi designs.

As a result of modelling of the data obtained, empirical equations with statistically significant importance are received which describe the effect of process variables (independent variables) on the process result (response variable).

Desirability function (*DF*) can be applied in search for optimal operational parameters. The method proposed by Derringer and Suich [18] involves the construction response surface model and then finding the values of independent variables which ensure the most desirable value. The objective of the presented studies was the analysis of the impact of reactor's operational parameters on ammonia oxidation reaction. To the best of our knowledge, the approach presented here to describe ammonia oxidation process is published for the first time.

#### **2. Results and Discussion**

#### *2.1. Design of Experiments*

Ammonia oxidation reaction depends on a few process variables. In this study, the effect of the reactor's load (*X*1), the temperature of nitrous gas specifying the temperature of reaction (*X*2) and the number of catalytic gauzes (*X*3) on ammonia oxidation reaction was investigated. The oxidation efficiency of NH3 to NO (*R*1) and N2O concentration in nitrous gases (*R*2) were selected as measures for ammonia oxidation reaction. The matrix of 15 experiments including particular levels of coded variables and achieved values of response variables *R*<sup>1</sup> and *R*<sup>2</sup> are presented in Table 1. In the regarded experimental area of independent variables, ammonia oxidation efficiency ranged from 91.4% to 96.4%, whereas N2O concentration in nitrous gases ranged from 1011 to 1762 ppm.


**Table 1.** The Box-Behnken design matrix and experimental data.

#### *2.2. Model Fitting*

The first task was to find out which equation would allow to obtain the best correlation between independent variables and responses. Analysis of Variance (ANOVA) was carried out for most frequently applied equations: linear, two-factor interaction (2FI), quadratic and cubic. Table 2 includes the summary statistics of both responses for different mathematical equations.


**Table 2.** Model summary statistics for response variables *R*<sup>1</sup> and *R*2.

\* - case(s) with leverage of 1.0000; PRESS statistic not defined.

Based on the achieved results, it was found that the experimental data is described best with quadratic and cubic equations. For both responses, high values of *R*<sup>2</sup> and adjusted *R*<sup>2</sup> were achieved. The number of conducted experiments caused that the cubic model was aliased. It means that the experimental matrix contains an insufficient number of experimental points for independent estimation of all effects for these models. Therefore, quadratic equation was selected for further analysis.

The statistical significance of these equations and their particular terms was specified based on Analysis of Variance (ANOVA). Results of this analysis are presented in Tables 3 and 4, for response variables *R*<sup>1</sup> and *R*<sup>2</sup> respectively. Large *F*-value indicates that most changes of independent variable can be explained with the developed regression equation. The correlated probability *p*-value is used to estimate whether *F*-value is large enough to show statistical significance.


**Table 3.** ANOVA results for response variable *R*1.

The probability *p*-value for the achieved model of variable *R*<sup>1</sup> is 0.0001. It means that the model is statistically significant but some terms of equation are statistically not significant. Coefficients: *R*2, adjusted R<sup>2</sup> and predicted R2 are very high: 0.9908, 0.9744 and 0.8577, respectively. There is also high compliance between coefficients: predicted *R*<sup>2</sup> and adjusted *R*<sup>2</sup> (difference <0.2). The achievement of statistically significant value lack of fit (0.0256) is the incompliance of this model as this parameter should be statistically not significant.



In case of response variable *R*2, the probability *p*-value (0.0012) indicates that the assumed quadratic equation is statistically significant but some of its terms are statistically not significant. High coefficients *R*2, adjusted *R*<sup>2</sup> and predicted *R*<sup>2</sup> are also achieved for the second response variable and they are: 0.9784, 0.9396 and 0.6574, respectively. However, the difference between predicted *R*<sup>2</sup> and adjusted *R*<sup>2</sup> is larger than the recommended one (>0.2). This may demonstrate a large block effect or problems with model or data. This model is also characteristic of statistically significant parameter lack of fit (*p* = 0.0119).

At a further stage of analysis, statistically not significant terms of initial equation were eliminated from the analysis. The reduction was made using step-by-step method (from the most insignificant term). For both these response variables, only statistically significant terms were left and higher *R*2, adjusted *R*<sup>2</sup> and predicted *R*<sup>2</sup> coefficients were achieved. Results of Analysis of Variance (ANOVA) are presented in Tables 5 and 6.


**Table 5.** ANOVA results for reduced model of the response variable *R*1.

The obtained mathematical model for response *R*<sup>1</sup> is highly significant (*p*-value < 0.0001). The dependence on linear terms *X*1, *X*3, interaction *X*1*X*<sup>3</sup> and quadratic term *X*<sup>3</sup> <sup>2</sup> are significant. High determination coefficients are obtained l (*R*<sup>2</sup> = 0.9835, adjusted *R*<sup>2</sup> = 0.9768, predicted *R*<sup>2</sup> = 0.9550).

The final model is presented in Equation (1).

$$R = 96.04 - 0.4125X\_1 + 1.88X\_3 + 0.575X\_1X\_3 - 1.72X\_3^2 \tag{1}$$


**Table 6.** ANOVA results for reduced model of the response variable *R*2.

The obtained mathematical model for response *R*<sup>2</sup> is highly significant (*p*-value < 0.0001). The dependence on linear terms *X*1, *X*2, *X*<sup>3</sup> and quadratic term *X*<sup>3</sup> <sup>2</sup> are significant. High determination coefficients (*R*<sup>2</sup> = 0.9513, adjusted *R*<sup>2</sup> = 0.9318, predicted *R*<sup>2</sup> = 0.8743) were achieved for the model. The final model is presented in Equation (2).

$$R = 1260 + 60.37X\_1 - 217.75X\_2 - 100.88X\_3 + 123.37X\_3^2 \tag{2}$$

#### *2.3. Model Diagnostics*

Before the process optimization, the model diagnostics for both equations was performed because of occurrence of statistically significant *Lack of fit* parameter. Results of model diagnostics: a normal probability of the residuals, residuals analysis and actual data versus predicted values plots were analysed.

Figure 2 presents model diagnostics for response variable *R*1, whereas Figure 3 presents model diagnostics for response variable *R*2. Normal plot of studentised residuals should be approximately a straight line, whereas studentised residuals versus predicted response values and versus run should be a random scatter. Points in plots of real response values with reference to predicted response values line up accurately along the axis at the angle of 45◦.

**Figure 2.** Model diagnostics for response variable *R*<sup>1</sup> (**a**) A normal probability plot of the residuals; (**b**) Residuals versus predicted value of *R*1; (**c**) Residuals versus run number; (**d**) Predicted versus actual value of response variable *R*1. Points on graphs correspond to results of particular experiments and colour points correspond to the value in accordance with the scale.

These diagnostics show that despite the fact that *Lack of fit* parameter is statistically significant, experimental and predicted points for both equations correlate well with each other.

**Figure 3.** Model diagnostics for response variable *R*<sup>2</sup> (**a**) A normal probability plot of the residuals; (**b**) Residuals versus predicted value of *R*2; (**c**) Residuals versus run number; (**d**) Predicted versus actual value of response variable *R*2. Points on graphs correspond to results of particular experiments and colour points correspond to the value in accordance with the scale.

#### *2.4. The Effect of Independent Variables*

In case of response variable *R*<sup>1</sup> (ammonia oxidation efficiency), the mathematical model shows a strong linear effect of the reactor's load (*X*1) and the number of catalytic gauzes (*X*3) and interaction between these two variables (*X*1*X*3) and the quadratic term number of catalytic gauzes (*X*<sup>3</sup> 2) on the achieved response variable. The temperature of reaction in the studied range does not affect the ammonia oxidation efficiency. The effect of *X*<sup>1</sup> and *X*<sup>3</sup> on response *R*<sup>1</sup> were shown as contour plot (Figure 4). According to the presented plot of variable of *R*1, a small number of catalytic gauzes causes lower ammonia oxidation efficiency for the entire range of studies reactor's load. The increase in the number of catalytic gauzes to *X*<sup>3</sup> = 0 causes increase in oxidation efficiency within the entire range of studied reactor's load.

Studies related to dependency of N2O concentration in nitrous gases on operating parameters are relatively new research issue. Therefore, there is a lack of scientific reports dedicated to systematic studies in this field. In case of the N2O concentration in nitrous gases, the achieved mathematical model demonstrates a significant effect of the selected process variables (*X*1, *X*2, *X*3) and the quadratic term number of catalytic gauzes (*X*<sup>3</sup> 2) on the achieved response variable. This effect is illustrated in Figures 5–7. The analysis of Equation (2) and Figures 5–7 indicates that the temperature of reaction has the biggest quantitative effect on N2O concentration in nitrous gases. From the comparison of plots (Figure 5a–c) it can be concluded that despite the presence of statistically significant terms of equations derived from variable *X*3, plots of contour line corresponding to levels 0 and 1 are similar. Only for *X*<sup>3</sup> = −1, higher values of *R*<sup>2</sup> are achieved. Profiles of response variable *R*<sup>2</sup> presented in Figures 6a–c and 7a–c confirm the effect of the number of catalytic gauzes. Both these figures show that the number of catalytic gauzes has little effect on the amount of N2O being formed. For the level of *X*<sup>3</sup> = 0–0.4 (10–12 gauzes), the local optimum is observed. For this number of gauzes, increasing the reactor's load (*X*<sup>1</sup> *=* 1) at the fixed reaction temperature (Figure 6a–c) and decreasing the reaction temperature at the fixed reactor's load (Figure 7a–c) does not cause a significant decrease in N2O concentration in nitrous gases.

**Figure 4.** Interaction effect between the reactor's load (*X*1) and the number of catalytic gauzes (*X*3) on ammonia oxidation efficiency (*R*1) by contour plot.

#### *2.5. Multi-Response Desirability Optimization*

The major optimization task is to find the number of catalytic gauzes and the permissible reactor's load ensuring the maximization of ammonia oxidation efficiency (*R*1) and the minimization of N2O concentration in nitrous gases. Results of experiments discussed in Section 2.4. indicate that statistically, the temperature has no significant effect on ammonia oxidation efficiency but on the other hand, the amount of N2O formed is reversely proportional to the temperature of reaction. For desirability function, it was assumed that independent variables are in the variability range. Assumptions for the optimization are presented in Table 7.

**Table 7.** Assumptions for the optimization of the ammonia oxidation process using desirability function. Variables symbol identification according to the Table 1.


\* Weight: 1—linear change of values in the range from 0 to 1; \*\* Importance: 5—high, 3—medium, 1—low.

For such optimization assumptions, the area of detailed set of parameters was achieved. It confirms that optimization assumptions are met. Desirability functions for three temperature levels are presented as contour plot in Figure 8.

**Figure 5.** Interaction effect between the reactor's load (*X*1) and the temperature (*X*2) at fixed number of catalytic gauzes (*X*3) on N2O concentration in nitrous gases (*R*2) by contour plot. (**a**) *X*<sup>3</sup> = −1; (**b**) *X*<sup>3</sup> = 0; (**c**) *X*<sup>3</sup> = 1.

**Figure 6.** Interaction effect between the reactor's load (*X*1) and the number of catalytic gauzes (*X*3) at fixed temperature (*X*2) on N2O concentration in nitrous gases (*R*2) by contour plot. (**a**) *X*<sup>2</sup> = −1; (**b**) *X*<sup>2</sup> = 0; (**c**) *X*<sup>2</sup> = 1.

**Figure 7.** Interaction effect between the temperature (*X*2) and the number of catalytic gauzes (*X*3) at fixed reactor's load (*X*1) on N2O concentration in nitrous gases (*R*2) by contour plot. (**a**) *X*<sup>1</sup> = −1; (**b**) *X*<sup>1</sup> = 0; (**c**) *X*<sup>1</sup> = 1.

**Figure 8.** Desirability function plots. Effect of the reactor's load (*X*1) and the number of catalytic gauzes (*X*3) at three levels of temperature (**a**) *X*<sup>2</sup> = −1; (**b**) *X*<sup>2</sup> = 0; (**c**) *X*<sup>2</sup> = 1.

High values of desirability function (*DF* > 0.9) at 910 ◦C are described with dependency according to which for the load of 456 kg NH3/(m2h), the sufficient number of catalytic gauzes is 8. However, for the maximum load studied, 10 catalytic gauzes should be applied. At the temperature of 910 ◦C and when all optimization criteria are met, the expected value of N2O concentration ranges from

1000 ppm to 1100 ppm (Figure 8a). Lowering the reaction temperature to 890 ◦C means that desirability function *DF* > 0.8 is within the region where the minimum catalyst gauzes is 9 for the loading not higher than 480 kg NH3/(m2h) and 12 gauzes for load of 645 kg NH3/(m2h) (Figure 8b). At this temperature, the expected concentration of N2O in nitrous gases ranges from 1180 to 1200 ppm. At the lowest temperature (within the studied range) of 870 ◦C, the highest value of desirability function is 0.69. For 12 gauzes and the load of 456 kg/(m2h), the expected concentration of N2O in nitrous gases is 1400 ppm.

Taking into account the amount of the primary emissions of N2O (the environmental aspect), it is favourable to conduct the reaction at the temperature of 910 ◦C. However, this leads to the increased platinum losses. Platinum losses at 910 ◦C are higher by approx. 25% as compared to losses at 890 ◦C and by 45% as compared to losses at 870 ◦C [1,2]. Lowering the reaction temperature to 890 ◦C with maintaining the optimal range of other parameters causes the increase of N2O concentration in nitrous gases by 100–200 ppm.

The assumption of other values of 'weight' and 'importance' for particular variables leads to obtain other profiles of desirability function. Under industrial conditions, the assumed value of 'weight' and 'importance' should take into account the process economics with regard to platinum losses.

#### *2.6. Validation*

Validation of the developed optimization model should be carried out under conditions specified as optimal. Optimization results indicate a wide set of parameters for which desirability function achieved high values. Therefore, in order to carry out additional measurements, the point with the independent variables value of: 1, 1, 1 was selected. This point is in the range of high desirability function value. In Table 8 levels of independent variables, results of validation experiment and predicted mean values of response variables with standard deviation are presented.


**Table 8.** The assumed levels of independent variables in validation studies and predicted responses values.

For the assumed independent variables, the values of predicted mean with 95% two-sided confidence intervals met by 99% of population were estimated based on the achieved mathematical model. High conformity of results expected according to mathematical models with the obtained measurements results was achieved.

#### **3. Materials and Methods**

#### *3.1. Materials*

Standard knitted gauzes made of platinum alloy with the addition of 10% wt. Rh made of 0.076 mm wire and specific weight of 600 g/m2 were used for ammonia oxidation studies. The catchment gauzes which are most commonly used in industrial process were not used in studies. The prefiltered compressed air and gaseous ammonia were used as raw materials for ammonia oxidation reaction.

#### *3.2. Experimental Procedure*

Ammonia oxidation studies were performed in a pilot plant equipped with a flow reactor (inner diameter: 100 mm). The Pt-Rh catalytic gauzes were installed inside the basket. After initiating the reaction, the stable ammonia and air ratio was maintained in reaction mixture amounting to approx. 10.9% vol. The air–ammonia mixture temperature was controlled in such a manner as to obtain the temperature of nitrous gas as assumed in the experiment plan. Air–ammonia mixture temperature was variable in the range of 135–195 ◦C. The flow of the air-ammonia mixture was also controlled in order to obtain the assumed reactor's load. All experiments were conducted under the pressure of 0.5 MPa. The range of temperature of reaction at which studies were carried out is similar to that applied in industrial practice. The reactor's load was selected in such a manner as to ensure that the gas flow through the catalytic package in the range applied for medium-pressure industrial reactor namely 1–3 m/s. The scheme of the pilot plant is presented in Figure 9. For measuring ammonia oxidation efficiency, samples of ammonia-air mixture were taken at the inlet to the reactor and samples of nitrous gases were taken at the outlet of the reactor. For determination of N2O concentration in nitrous gases, only samples of nitrous gases were taken.

**Figure 9.** Scheme of the pilot plant. Symbols: C1—absorption column, C2—bleaching column, E1–E6—heat exchangers, M1—air–ammonia mixer, R1—ammonia oxidation reactor, R2—selective catalytic reduction reactor.

#### *3.3. Analytical Methods*

The ammonia oxidation efficiency was calculated based on concentrations of ammonia in the air-ammonia mixture at the inlet and concentrations of NO in nitrous gases at the outlet of the reactor. Both analysis were determined according to a titration method. Ammonia from air-ammonia mixture samples was absorbed in water with the formation of ammonia-water solution which was then titrated with sulphuric acid. The nitrous gases samples were absorbed in 3% water solution of hydrogen peroxide. After ensuring the sufficient period of time, NO oxidized completely to NO2 and then, it reacted with water to HNO3. The formed HNO3 was titrated with the sodium hydroxide solution in the presence of an indicator.

Ammonia oxidation efficiency (*R*1) was calculated according to the following formula:

$$R\_1 = \left(\frac{C\_2}{C\_1}\right) \cdot 100\% \tag{3}$$

where: *C*1—ammonia concentration in ammonia-air mixture, % *w*/*w*; *C*2—concentration of oxidized ammonia, % *w*/*w*.

The result of each measurement is an average value, calculated from 7 independent samplings. The difference in the extreme individual values were not greater than ±0.3% in comparison to the average one.

N2O concentration in nitrous gases (*R*2) was determined by gas chromatography using a Unicam 610 system with a discharge ionization detector. Gaseous samples were collected in the vacuum flasks containing 3% water solution of hydrogen peroxide. After the absorption of nitrous gases and water vapor condensation, exhaust gas from the flasks was injected to the gas chromatograph through 1 mL sample loop. The result of each measurement was an average value calculated from 3 independent samplings. The difference in the extreme individual values was not greater than ±35 ppm in comparison to the average one.

#### *3.4. Statistical Methods*

The experimental procedure was carried out according to Box–Behnken design matrix. The reactor's load (*X*1), the temperature of nitrous gas specifying the temperature of reaction (*X*2) and the number of catalytic gauzes (*X*2) were selected as independent variables. Ammonia oxidation efficiency (*R*1) and N2O concentration in nitrous gas (*R*2) were specified as response variables. Each level of independent variables were coded according to the Equation (4).

$$X\_i = \frac{\mathbf{x}\_i - \mathbf{x}\_0}{\Delta \mathbf{x}\_i} \tag{4}$$

where, *Xi* is the dimensionless, coded level of independent variable (−1, 0 or 1), *xi* is the actual value of the independent variable, *x*<sup>0</sup> is the value of the independent variable at the centre point, Δ*xi* is the step change in *xi.*

Ranges and levels of independent variables are presented in Table 9.


**Table 9.** Coded and uncoded levels of independent variables used in experiments.

The total number of the experiments (*N*) was calculated using the Equation (5).

$$N = 2k(k-1) + c\_0 \tag{5}$$

where, *k* is the number of independent variables, *c*0—number of the replicates run of the centre point (in our research *c*<sup>0</sup> = 3). For three independent variables, the total number of experiments assumed in the plan was 15.

The experiments were conducted in a randomized order to avoid the influence of uncontrolled variables on the dependent responses.

A mathematical relationship between the independent variables and response variables was determined by fitting the experimental data with second-order polynomial Equation (6).

$$R\_i = b\_0 + \sum\_{i=1}^3 b\_i X\_i + \sum\_{i=1}^3 \sum\_{j=1,\ i$$

where *Ri* is the estimate response variable, *b*0, *bi*, *bii*, *bij* are regression coefficients fitted from the experimental data, *Xi*, *Xj* are coded independent variables, listed in Table 9.

The significance of the model equation, individual parameters, were evaluated through ANOVA with the confidence interval (CI) of 95%. A simultaneous optimization of several dependent variables requires the application of multi-criteria methodology. In this case, the desirability function (*DF*) was used. The particular desirability functions are combined using the geometric mean which allows to achieve overall desirability function [16], according to Equation (7).

$$DF = \left( (d\_1)^{w\_1} \times (d\_2)^{w\_2} \times \dots \times (d\_n)^{w\_n} \right)^{1/\sum w\_i} \tag{7}$$

where *n* is the number of responses, *di* is an individual response desirability, *wi* is a response 'weight'.

The adjustment of the shape of particular desirability function can be performed by assigning the specified 'weight.' Setting a different 'importance' for each objective with respect to the remaining objectives is also possible. For these studies, identical 'weight' for all the independent variables and response variables was assumed. Desirability function assigns values from 0 to 1 where 1 means meeting all the optimization criteria. It is not always necessary to search for the solution aiming at achievement of the highest value of desirability function but it is vital to search for the set of parameters which would meet the optimization objectives to the particular extent (e.g., *DF* > 0.75). The statistical software used to experimental design and analysis was Design Expert 11.0.6.0 version (Stat-Ease, Inc., Minneapolis, MN, USA).

#### **4. Conclusions**

The conducted studies allowed us to develop statistically significant mathematical models describing the course of variables of ammonia oxidation efficiency and N2O concentration in nitrous gases depending on three selected independent variables.

The design of the experiment allowed the reduction of the costs of studies and to achieve a number of results accurate for modelling. It was found that, within the studied range of variability, the temperature of reaction has no significant effect statistically on the achieved ammonia oxidation efficiency, whereas it has the effect on the amount of N2O formed in the side reaction (primary emission of N2O).

The developed models were used to optimize the process. As a result of this optimization, the set of the independent variables was developed for which optimization assumptions are met, which are expressed as a high value of desirability functions. It is possible to specify the optimum number of gauzes with the determined reactor's load for the studied package of catalytic gauzes.

In validation experiments, the developed model of desirability function achieved the high conformity of experimental values with the expected ones.

The presented methodology can be used to minimize the primary N2O emission at high ammonia oxidation efficiency. It can be applied for optimization of operating parameters of ammonia oxidation reactor with two types of catalysts: catalytic gauzes and catalyst for high temperature of N2O decomposition. As a result, it is possible to obtain the set of independent variables ensuring low N2O emission and to meet the binding environmental regulations.

**Author Contributions:** Conceptualization, M.I.; Formal analysis, M.I. and A.D.-I.; Investigation, M.I. and J.R.; Methodology, M.I.; Supervision, M.W.; Visualization, A.D.-I.; Writing—original draft, M.I.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

1. Thiemann, M.; Scheibler, E.; Wiegand, K.W. Nitric Acid, Nitrous Acid, and Nitrogen Oxides. In *Ullmann's Encyclopedia of Industrial Chemistry*; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2012.


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **Tuning the Catalytic Properties of Copper-Promoted Nanoceria via a Hydrothermal Method**

### **Konstantinos Kappis 1, Christos Papadopoulos 1, Joan Papavasiliou 1,2, John Vakros 3, Yiannis Georgiou 4, Yiannis Deligiannakis <sup>4</sup> and George Avgouropoulos 1,\***


Received: 8 January 2019; Accepted: 21 January 2019; Published: 1 February 2019

**Abstract:** Copper-cerium mixed oxide catalysts have gained ground over the years in the field of heterogeneous catalysis and especially in CO oxidation reaction due to their remarkable performance. In this study, a series of highly active, atomically dispersed copper-ceria nanocatalysts were synthesized via appropriate tuning of a novel hydrothermal method. Various physicochemical techniques including electron paramagnetic resonance (EPR) spectroscopy, X-ray diffraction (XRD), N2 adsorption, scanning electron microscopy (SEM), Raman spectroscopy, and ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS) were employed in the characterization of the synthesized materials, while all the catalysts were evaluated in the CO oxidation reaction. Moreover, discussion of the employed mechanism during hydrothermal route was provided. The observed catalytic activity in CO oxidation reaction was strongly dependent on the nanostructured morphology, oxygen vacancy concentration, and nature of atomically dispersed Cu2+ clusters.

**Keywords:** copper-ceria catalysts; hydrothermal method; CO oxidation; copper clusters; nanoceria

#### **1. Introduction**

Carbon monoxide (CO) is a harmful, toxic gas that is present in many industrial processes. Due to its negative impact for both humans and the environment, the catalytic oxidation of CO into CO2 has always been a research topic of great interest [1–3]. Moreover, the catalytic oxidation of CO is an important reaction in the technological fields of fuel cells [4–6], gas sensors [7], and CO2 lasers [8].

Cerium oxide or ceria (CeO2) has been thoroughly studied as a catalyst or support in CO oxidation reaction due to its defective structure enriched with oxygen vacancies and high oxygen storage capacity (OSC) resulting from the interaction between Ce3+ and Ce4+ [9–12]. In the case of ceria synthesized in a nanosized form, more remarkable functions can be obtained due to the nanosize effects. For this reason, research has focused on the understanding of the properties of nanoceria as well as improving its OSC, surface to volume ratio, and redox properties [13–16]. Computational studies have shown that the catalytic activity of nanoceria is strongly associated with the exposed surface plane. Sayle et al. [17] predicted that the (110) and (100) surfaces are catalytically more active for CO oxidation than the (111) surface, due to more oxygen vacancies located in the former. According to Conesa [18], the formation of oxygen vacancies on the (110) and (100) surfaces requires less energy than the (111) surface. It is noteworthy that the exposition of the reactive surface plane is dependent on the morphology of the material at the nanoscale. Zhou et al. [19] have shown that CeO2 nanorods, which exposed the (110) and (100) planes achieved higher catalytic activity for CO oxidation than nanoparticles exposing the (111) planes. Wu et al. [20] studied the morphology dependence of CO oxidation over ceria nanocrystals. They discovered that the activity for CO oxidation of those CeO2 nanostructures follows the order: Rods > cubes > octahedra, whereas the activity of different planes follows the order: (110) > (100) > (111). These results were also confirmed by Tana et al. [21]. In order to prepare various shapes of nanoceria, a number of methods have been applied, such as sol-gel [22], precipitation [23], hydrothermal or solvothermal methods [24–27], and electrochemical deposition [28]. Among these methods, the hydrothermal method has attracted great interest because a desired morphology can be obtained via appropriate control of the hydrothermal parameters such as reaction time, temperature, and concentration [29–31].

Despite the attractive physicochemical properties of ceria, poor catalytic activity of pure ceria [32] can by highly promoted via doping with a series of metal ions, in order to change its surface chemistry and promote the active oxygen content [33]. It is well known that the reduction behavior of ceria can be rapidly altered by the addition of a minimal amount of Au, Pd, and Pt precious metals and/or transition metals [34–38]. While the activity of the catalysts is improved by the addition of precious metals, their high cost prohibits their application. Numerous reports have indicated that the activity of ceria in oxidation reactions is enhanced by transition metals like copper. The copper–ceria system presents a cost-effective material with unique catalytic properties, comparable to noble metal catalysts, in many catalytic reactions and especially in the CO oxidation and in the preferential oxidation of CO in excess of hydrogen [39–44]. The superiority of CuCeOx catalytic system has been attributed to a synergistic effect. Reports on the mechanism of CO oxidation reaction over these catalysts have demonstrated the significance of both copper and ceria species in the adsorption of CO and CO2 production, as the former takes place in the copper-ceria interface [45]. Particularly, the main reasons that trigger the highly catalytic performance of these Cu-Ce catalysts are the large amount of well-dispersed copper species in the ceria support, the creation of oxygen vacancies due to incorporation of Cu2+ ions into the ceria structure and the presence of high concentration of active lattice oxygen [45–47]. Several Cu2+ entities (e.g., amorphous clusters, isolated ions, dimers, and discrete crystallites) have been detected, which can take part in the catalytic mechanism, displaying high levels of activity [5,39,48–50]. In order to form these entities, conventional preparation methods have been used such as the deposition of Cu2<sup>−</sup> species onto pre-synthesized ceria support or the coprecipitation of Cu and Ce precursors. The obtained materials are calcined at high temperatures, which enable the dispersion of copper species and the chemical bonding between the Cu and Ce components. For example, Harrison et al. [39] prepared CuO/CeO2 catalysts via coprecipitation and impregnation routes, and tested in CO oxidation. After thermal pretreatment of materials at 400 ◦C, the copper content on the surface of ceria consisted of amorphous clusters of Cu2+ ions, which presented high catalytic activity for CO oxidation. In several cases, the high temperatures required for the activation of copper component (>600 ◦C for 4 h) may lead to particle sintering and phase segregation, which facilitates the formation of tenorite particles of CuO, which are inactive for CO oxidation [51].

In the present work, copper clusters were atomically dispersed in ceria nanostructures via a novel hydrothermal route, yielding highly active catalysts in CO oxidation reaction. Tuning of the physicochemical and catalytic properties was studied by varying the hydrothermal parameters (temperature and concentration). A set of analytical techniques such as electron paramagnetic resonance (EPR) spectroscopy, X-ray diffraction (XRD), N2 adsorption-desorption, scanning electron microscopy (SEM), Raman spectroscopy, and ultraviolet-visible diffuse reflectance spectroscopy (UV-Vis DRS) was used to assess the physicochemical characteristics of the materials and correlate with the catalytic performance.

#### **2. Results**

#### *2.1. EPR Measurements*

Figure 1 illustrates the EPR spectra of the hydrothermally prepared ceria-based catalysts, using various concentrations of NaOH. The concentrations of the detected species are shown in Table 1. In general, the EPR spectra present three characteristic signals; A signal at ca. 1580 Gauss (g = 4) is attributed to Ce3+ ions in high spin (S = 5/2) state, corresponding to reduced Ce3+ ceria centers located inside the lattice of ceria particles [15,52]. The sharp signal at around 3400 Gauss (visualized better in the right column spectra of Figure 1) is ascribed to [Ce3+-O−-Ce4+] (S = 1/2) units localized on the surface of the ceria particles [5,53,54]. According to the geometrical disposition of the two types of Ce3+ centers—lattice and surface—it is considered that surface Ce3+ ions can be correlated with high catalytic activity, i.e., since they may result to the formation of oxygen vacancies [55–57]. Regarding the Cu2+ signals, the EPR signals of the sample treated at 180 ◦C show a strong signal at 2600–3300 Gauss, which is attributed to Cu2+ (S = 1/2, I = 3/2) ions. The line shape of this EPR spectrum indicates that the copper atoms show dipolar Cu-Cu interactions, i.e., the Cu2+ is clustered within 10 <sup>±</sup> 3 Angstroms from each other [58]. The g-and A values of Cu2+, (g⊥= 2.035, g//= 2.305, <sup>A</sup>⊥= 15, A//=150) suggest that the Cu2+ ions are located in the octahedral sites in ceria with a tetragonal distortion [39,49,55,56,59–61].

**Table 1.** Concentrations of the species as detected by EPR spectroscopy.


<sup>1</sup> Not-detected.

**Figure 1.** *Cont*.

**Figure 1.** *Cont*.

**Figure 1.** EPR spectra of the catalysts prepared with 0–5 M NaOH. Magnification of the region in which the bulk Ce3+ ions were detected, is shown in the inset figure; the figures on the right side present an enlarged region where the Cu2+ species and the surface Ce3+ ions were detected.

Quantitate data on the Ce3+ and Cu2+, estimated from the EPR spectra, are listed in Table 1. Copper loading was also confirmed via XRF measurements (within an experimental error of ±10%; not shown here), in line with EPR results. Concerning the surface Ce3+ ions, there is a clear trend vs. the NaOH concentration: Their concentrations increases when concentrated solutions of sodium hydroxide were employed in the hydrothermal route. With regard to the copper entities, the combination of low concentrations of NaOH and high hydrothermal temperatures caused the formation of a high amount of copper species (especially Cu2+ clusters). On the other hand, the high concentrations of NaOH halted the dispersion of copper species in the ceria phase, due to the high basicity of NaOH resulting in Cu(OH)n clusters' formation in the CeO2 phase.

#### *2.2. XRD Measurements*

The XRD diffractograms of all the samples are shown in Figure S1. Noticeably, all peaks can be indexed to (111), (200), (220), (331), (222), (400), (331), (420), and (422) planes corresponding to the pure cubic phase [space group: Fm-3m, JCPDS: 00-043-1002, α = 0.54113 nm] of CeO2 [30]. No diffraction peaks of crystalline copper species can be observed, due to the presence of highly dispersed amorphous copper species or/and the low copper loading, which can be hardly detected at XRD [62,63]. However, as it was discussed in EPR section, the copper species might be atomically dispersed in the ceria matrix. The calculated average crystallite size and the lattice parameter of the ceria-based materials

are presented in Table 2. It has to be noted that there is no clear relationship between Cu content (see Table 1) and crystallite size or lattice parameter of the catalysts.


**Table 2.** Average crystallite size and lattice parameter of the catalysts based on the crystal plane of (111) of CeO2.

The dependence of the crystallite size from the hydrothermal parameters is depicted in Figure 2. It can be seen in Figure 2a that, for concentrations of NaOH ≤0.05 M, small-size crystallites were formed at temperatures of 120–150 ◦C, while at 180 ◦C, a dramatic increase of their size can be observed. A more rapid increase of crystallites size was obtained in the case of 0.1 M NaOH. For instance, the sample Ce-120-0.1 presents an average crystallite size at 10 nm, while at 150 ◦C, the average size is ca. 20 nm. In the range of 0.5–1 M NaOH, the crystallites size varies from ~10 to 15 nm following an increase in the hydrothermal temperature. Finally, for 5 M NaOH, the combination of high concentration and high temperatures caused the formation of large crystallites. In Figure 2b, at 120 and 150 ◦C, a fluctuation of the crystallites size is indicated as the concentration of NaOH increases. At 180 ◦C, a decrease of the crystallite size is observed for concentrations ≤0.5 M. Further increase of the concentration caused the rise of the crystallites size.

**Figure 2.** Dependence of the crystallites size of the catalysts with respect to the hydrothermal parameters: (**a**) Based on the temperature of the hydrothermal treatment; and (**b**) based on the concentration of NaOH.

Concerning the lattice parameter, the vast majority of the samples present higher values of lattice parameter than the pure ceria, suggesting the lattice expansion for the obtained materials. According to several studies, the lattice expansion is closely correlated with the presence of Ce3+ ions in the crystal lattice because the radius of Ce3+ ions (0.114 nm) is higher than the radius of Ce4+ ions (0.097 nm), inducing the lattice expansion [27,64–66]. On the other hand, only two samples (Ce-120-0 and CuCe-150-0) presented a smaller value of lattice parameter than the pure ceria. According to Pan et al. [30], the hydroxyl groups may stabilize the smaller nanoparticles resulting in the smaller value of lattice parameter.

#### *2.3. N2 Adsorption Measurements*

The N2 adsorption/desorption isotherms and the pore size distribution for the ceria-based materials are illustrated in Figure 3. Additionally, the specific surface areas (SSA), the pore volume and the pore size of all samples are shown in Table 3. For hydrothermal solutions of NaOH ≤0.1 M, the materials present type II isotherms with type H2 hysteresis loops, independently of the hydrothermal temperature. Materials that present type H2 hysteresis loops are often disordered without well-defined pore distribution. However, for NaOH concentrations ≥0.5 M, it can observed that the type of isotherm becomes type IV, which is characteristic of mesoporous materials [67] with type H2 hysteresis loops. It is worth mentioning that a different type of hysteresis loops is revealed for Ce-150-5 and Ce-180-5 samples. Specifically, the type of hysteresis loop is type H1, which is associated with well-defined cylindrical pores.


**Table 3.** Specific Surface Area, Pore Volume, and Pore Size of the catalysts.

The pore size distribution diagrams (Figure 3) denote that these ceria-based catalysts have not got a well-defined pore size. Indeed, the combination of hydrothermal parameters affected this distribution. Pores ranging in the mesoporous region were formed in the case of NaOH concentration ≥0.5 M, while different distributions can be observed with NaOH concentrations ≤0.1 M. For example, the CuCe-150-0 catalyst mainly presents mesopores, whereas the CuCe-180-0 sample mainly consists of macroporous.

The effect of the hydrothermal parameters on the specific surface area (SSA) of the obtained materials is illustrated in Figure 4. It can be seen in Figure 4a that an increase of the hydrothermal temperature from 120 to 150 ◦C resulted in higher SSA when the NaOH concentration was ≤0.5 M. Further increase of the temperature lowered the surface area of the catalysts. On the other hand, for higher NaOH concentration (≥1 M), the highest surface area was obtained at 120 ◦C. The highlight of

this trend was the SSA of the Ce-120-5 catalyst (137.1 m<sup>2</sup> g−1). A general trend depicted in Figure 4b, suggests that for elevated concentration of NaOH, higher surface areas can be obtained for the catalysts prepared hydrothermally at 120 ◦C. A similar trend can be seen at 150 and 180 ◦C, however a maximum of SSA corresponds to NaOH concentration of 0.5 M. Therefore, it can be concluded that the combination of high concentrations and high hydrothermal temperatures favors the formation of catalysts with poor surface area.

**Figure 3.** N2 adsorption/desorption isotherms and pore size distribution curves of CeO2 and Cu/CeO2 catalysts: (**a**,**c**,**e**) N2 adsorption/desorption isotherms; and (**b**,**d**,**f**) pore size distribution curves.

**Figure 4.** Dependence of the specific surface area of the catalysts with respect to the hydrothermal parameters: (**a**) Based on the temperature of the hydrothermal treatment; and (**b**) based on the concentration of NaOH.

#### *2.4. SEM Measurements*

Figure 5 illustrates representatives SEM images of the materials which were hydrothermally synthesized at 120 ◦C. A morphology of bulk rods with various aggregates onto their surface was obtained for the Ce-120-0 sample. In the presence of NaOH, the morphology changed and spheres with a diameter of 4–8 μm were formed for the CuCe-120-0.05 sample. Spherical aggregates with a size of few hundred nanometers were formed onto this material surface. Further increases of the NaOH concentration resulted in rods with various lengths (2–10 μm) for the Ce-120-0.1 sample. No particles or aggregates can be observed on the surface of these rods, a fact that confirms the high crystallinity of the rods. These rods disappeared at elevated concentrations of NaOH, and very big aggregates with non-defined morphology were formed. In the case of hydrothermal synthesis at 150 ◦C, a spherical morphology dominates (Figure 6), while at high NaOH concentrations (≥ 0.5 M), bulky aggregates with particles without well-defined geometry, were formed. Figure 7 illustrates representative SEM images of the materials, which were treated hydrothermally at 180 ◦C. The CuCe-180-0 sample maintained the spherical morphology. The spheres composed of particles without well-defined geometry with size of ca. 70 nm. Adding low amounts of NaOH (0.05 M) resulted in a mixed morphology with rods of few micrometers, spheres and to a smaller extent sheets. All these structures contained irregular particles with an average size of a few dozen nanometers. The CuCe-180-0.1 sample appears to possess a spherical morphology, while similar structures with the lower hydrothermal temperatures were obtained at elevated NaOH concentrations (≥0.5 M).

#### *2.5. Raman Measurements*

Raman spectra of all the catalysts are shown in Figure 8. Ceria presents a fluorite structure with only one allowed Raman mode, which has an F2g symmetry and can be viewed to the symmetrical stretching mode of oxygen ions around Ce4+ ions [68–70]. For bulk ceria, this band appears at 465 cm<sup>−</sup>1. However, a shift to lower frequencies (Table 4), together with a non-linear linewidth, can be viewed for all the samples. According to Spanier et al. [71], a large number of factors can contribute to the changes in the Raman peak position and linewidth of the 465 cm−<sup>1</sup> peak, including phonon confinement, broadening associated with size distribution, defects, strain, and variations in phonon relaxation as a function of particle size. Apart from this main band, several other bands can be clearly distinguished in the corresponding spectra. The band at ca. 265 cm−<sup>1</sup> is attributed to the tetrahedral displacement of oxygen from the ideal fluorite lattice [25,69,72]. The band at ca. 600 cm−<sup>1</sup> is assigned to the defect-induced mode (D), related to the presence of lattice defects, mostly oxygen vacancies [5,42,68,73]. Noticeably, the samples synthesized at 150 ◦C with low concentration of NaOH (CuCe-150-0, CuCe-150-0.05, and CuCe-150-0.1) exhibit one additional band at ca. 830 cm−1. According to Choi et al. [72], this band is associated with peroxo-like oxygen species adsorbed on the oxygen vacancies, in close relation with reduced ceria species. No separated copper phase (CuO) can be confirmed via the appearance of the corresponding Raman peaks at ca. 295 cm−<sup>1</sup> and ca. 350 cm−1. Therefore, both the XRD and Raman results clearly indicate that Cu species onto ceria are highly dispersed.

**Figure 5.** Catalysts prepared hydrothermally at 120 ◦C: (**a**) Ce-120-0, (**b**) CuCe-120-0.05, (**c**) Ce-120-0.1, (**d**) CuCe-120-0.5, (**e**) CuCe-120-1, and (**f**) Ce-120-5.

**Figure 6.** Catalysts prepared hydrothermally at 150 ◦C: (**a**) CuCe-150-0, (**b**) CuCe-150-0.05, (**c**) CuCe-150-0.1, (**d**) Ce-150-0.5, (**e**) CuCe-150-1, and (**f**) Ce-150-5.

The position and the FWHM for the F2g band, and the relative intensity ratio of ID/IF2g for the ceria-based catalysts are shown in Table 4. In general, the ratio of ID/IF2g, where IF2g and ID correspond to the maximum intensity of F2g and D bands, respectively, can roughly reflect the amount of lattice defects (oxygen vacancies) in the obtained materials [72,74,75]. With appropriate tuning of the employed hydrothermal parameters, materials with high concentration of oxygen vacancies are obtained. For instance, the Ce-120-5 sample presents the highest value of this ratio, i.e., 0.106, suggesting a material with high perspectives in catalytic CO oxidation reaction. According to previous studies, the FWHM of the F2g band is influenced to a great extent by the crystallite size of ceria and the concentration of oxygen vacancies [75–77]. However, there is no specific trend in this work and mainly the high FWHM may be closely correlated with the high concentration of oxygen vacancies, since more factors might also play a role in these features.

**Figure 7.** Catalysts prepared hydrothermally at 180 ◦C: (**a**) CuCe-180-0, (**b**) CuCe-180-0.05, (**c**) CuCe-180-0.1, (**d**) CuCe-180-0.5, (**e**) CuCe-180-1, and (**f**) Ce-180-5.

**Figure 8.** Raman spectra of the catalysts synthesized hydrothermally (**a**) 120, (**b**) 150, and (**c**) 180 ◦C.

**Table 4.** F2g peak position, full width at half maximum (FWHM) for the F2g band, relative intensity ratio of ID/IF2g, and energy band gap (Eg) of the catalysts.


#### *2.6. UV-Vis DRS Measurements*

Figure S2 presents the UV-Vis DRS spectra of the materials. Each spectrum indicates three different bands, which according to the literature correspond to different electronic transitions between cerium and oxygen ions. The band at ca. 360 nm is ascribed to the O2<sup>−</sup> → Ce4+ interband-transfer transition. The bands at ca. 270 and ca. 230 nm are attributed to the O2<sup>−</sup> → Ce4+ and O2<sup>−</sup> → Ce3+ charge-transition, respectively [25,78]. Interestingly, the CuCe-180-0, CuCe-180-0.05, CuCe-180-0.1, and CuCe-150-0.1 samples present one extra wide band at 670 nm. It is important to mention that there is a lot of controversy about this band. According to Rakai et al. [79], this band is correlated with a surface redox couple of cerium ions (Ce3+/Ce4+). A similar band was found from Bensalem et al. [80] and Binet et al. [81]. However, there is a number of studies which ascribe this band to d-d transitions of Cu2+ in an octahedral environment [2,82,83]. It also has to be noted that, for the CuCe-150-0.1, CuCe-180-0, CuCe-180-0.05, and CuCe-180-0.1, catalysts a shoulder appears in the spectra at ca. 400–500 nm. This feature represents the charge transfer between "Support ← Oxygen—Active Phase", suggesting strong interactions between the copper species and ceria [43,44].

The energy band gap (Eg) of the catalysts, calculated via the Tauc plots method, is summarized in Table 4. Taking into account the reference value for the bulk CeO2 as a direct band gap semiconductor (Eg = 3.19 eV), the obtained values of the ceria-based materials, especially the ones synthesized with low concentrations of NaOH (≤0.1 M), are smaller than the energy band gap of bulk ceria. On the other hand, the catalysts prepared with higher concentrations of NaOH (≥0.5 M) resulted in a higher value of Eg. The former values indicate the presence of defects and especially oxygen vacancies [26,84], while the latter values of Eg are closely related to the quantum confinement effect [77,85].

#### *2.7. Formation Mechanism*

Over the years, the utilization of organic additives such as polyvinylpyrrolidone (PVP) [86], cetyltrimethylammonium bromide (CTAB) [87] and oleic acid [88] has been established in the hydrothermal method in order to obtain well-defined particles and morphologies. However, due to adsorption effects on the surface of particles these additives lead to quenching because of their high-energy vibration [89]. Compared with the above surfactants, citric acid (CA) presents weaker morphology control ability. Therefore, when the citric acid is used as a chelating agent, products with various morphologies and sizes can be obtained [90,91].

According to Levien [92], citric acid separates into different ionic species:

$$\text{H}\_{3}\text{cit}\,(\text{citric acid}) \rightarrow \text{H}^{+} + \text{H}\_{2}\text{cit}^{-}\,,\,(\text{pK}\_{a} = \text{3.2}),\tag{1}$$

$$\text{H}\_2\text{cit}^- \rightarrow \text{H}^+ + \text{Hci}^{2-}, \text{ (p}\text{K}\_a = \text{ 4.9)}, \tag{2}$$

$$\text{HCl}^{2-} \rightarrow \text{H}^+ + \text{cit}^{3-}, \text{ (p} \text{K}\_a = \text{ 6.4),}\tag{3}$$

The concentrations of these ionic species are closely depended on the pH value. In an acid solution (ca. pH = 2) the citric acid is not effectively dissociated to citric ions. Further increase of pH at values of 4, 6, and 8, the main citric species are the H2cit−, Hcit2<sup>−</sup> and cit3−, respectively. The citric ions can form complexes with the cerium ions, which are depended on the pH of solution [93].

The pH values from each step of synthesis route are illustrated in Table 5. It can be observed that the low concentrations of NaOH (≤0.1 M) did not increase the pH of the final solution (acid solution). As a result, the citric acid was not effectively dissociated, as there were H3cit and H2cit− species in the final solution. Concerning the partial dissociation of citric acid, it is not possible to form several and stable complexes with the cerium ions under ambient conditions. On the other hand, the high concentrations of NaOH (≥0.5 M) caused the formation of an alkaline solution. Under these conditions, the citric acid has been completely dissociated and can form several and stable complexes with the cerium ions, even under ambient conditions.


**Table 5.** pH values from the different part of the catalyst's preparation.

<sup>1</sup> The sample 0M is referred to the sample that no addition of NaOH is occurred and so on; <sup>2</sup> pH of the solution that contains Ce3+ ions; <sup>3</sup> pH of the solution that contains citric ions; <sup>4</sup> pH during the mixing of the previous solutions; <sup>5</sup> pH of the NaOH solution; and <sup>6</sup> pH of the final solution before the hydrothermal treatment.

It is believed that the increase of the temperature (from room temperature to the desired hydrothermal temperature) ensures the complete complexation of citric ions with the cerium ions [93–95], while the formed complex is polymerized and becomes stable in the solution. As the hydrothermal treatment proceeds the elevated pressure and temperature triggers the appearance of two events:


The final catalyst is obtained after the steps of filtration, drying, and calcination.

Concerning the various morphologies which are illustrated in Figures 5–7, it is proposed that the successful combination of hydrothermal parameters (temperature and concentration of NaOH) is the main reason behind these morphologies. It has been mentioned in Section 2.4 that the combination of low concentrations (≤0.1 M) and hydrothermal temperatures caused the formation spheres and/or rods. Given that the pH of the solution at these specific concentration was acidic, Ostwald Ripening seems to be the most dominant particle's formation mechanism [96–98]. Once the particles have been formed, a process of self-organization starts to happen, which results in a spherical and/or rod-like morphology depending on the hydrothermal parameters. On the other hand, the high concentrations of NaOH (≥0.5 M) and the hydrothermal temperatures resulted in the formation of bulky aggregates consisting of particles with a spherical and/or irregular geometry. At these conditions, the pH of the solution was alkaline and so the proposed formation mechanism of particles appears to be the oriented attachment [87,99]. Once the formation of particles has been completed, self-organization process initiates, but simultaneously the excess of OH− groups and the hydrothermal temperatures cause the formation of bulky aggregates.

#### *2.8. CO Oxidation Catalytic Studies*

The conversion of CO to CO2 as a function of the temperature of the reaction over the ceria-based catalysts is shown in Figure 9. As a general trend, it can be commented that the catalysts prepared with low concentrations of NaOH (≤0.1 M) and contained Cu2+ species onto their surface, presented better catalytic behavior than the catalysts which were hydrothermally treated with high concentrations of NaOH (≥0.5 M). Noticeably, the latter catalysts illustrated a significant amount of surface Ce3+ ions (see Table 1), which is usually correlated with high catalytic activity. However, this is not the case in this work and other factors control the catalytic activity, as will be discussed in the next section.

An indicator of the catalytic activity behavior are the temperatures where 50% and 90% CO conversion is achieved (T50 and T90, respectively) (Figure 9). For the samples treated hydrothermally at 120 ◦C, it can be seen that the most active catalysts are the CuCe-120-0.05 and the Ce-120-0, presenting T50 equal to 194 ◦C and 224 ◦C, respectively, and T90 equal to 272 ◦C and 263 ◦C, respectively. Additionally, the above samples achieved full removal of CO. The less active sample was the Ce-120-0.1, which revealed a T50 at 306 ◦C and T90 at 351 ◦C. One possible explanation for the poor catalytic

activity of this sample is its morphology, which presented rods with high crystallinity. Increasing the concentration of NaOH (≥0.5 M), the catalytic activity of the samples was improved, but it cannot be comparable with the activities of the CuCe-120-0.05 and Ce-120-0. The morphology of the samples prepared with high concentrations of NaOH (see Figure 5) seems to influence the catalytic activity in a negative way. Moreover, although the sample Ce-120-5 illustrated the highest SSA among all catalysts (see Table 3), this is not related to the achievement of high catalytic activity.

**Figure 9.** (**a**,**c**,**e**) CO conversion diagrams of the catalysts which were treated hydrothermally at 120, 150 and 180 ◦C, respectively; and (**b**,**d**,**f**) Bar graphs which are shown the temperatures where 50% and 90% conversion of CO to CO2 is achieved.

The catalytic activity of the samples synthesized at 150 ◦C, was dramatically improved, especially for NaOH concentrations ≤0.1 M, reaching in some cases 100% of CO conversion. For instance, the CuCe-150-0 sample presents a T50 at 257 ◦C and the sample CuCe-150-0.1 shows T50 at 165 ◦C. It should be mentioned that the former sample contained Cu2+ isolated ions dispersed in ceria, as confirmed via EPR spectroscopy, while the latter sample presented Cu2+ clusters onto ceria surface. Higher reactivity of Cu2+ clusters than Cu2+ isolated ions, was also reported by Harrison et al. [39]. Higher concentrations of NaOH (≥0.5 M) resulted in catalytic materials with poor activity.

Similar trends can be also depicted in the catalytic behavior of the materials synthesized with a hydrothermal temperature of 180 ◦C. Overall, the highest catalytic activity was achieved over the CuCe-180-0.05 sample which present in Figure 9, T50 and T90 at 132 ◦C and 180 ◦C, respectively. The rod-like morphology, the presence of high-content Cu2+ clusters (see Table 1) and high concentration of oxygen vacancies (see Table 4) seems to be the crucial reasons behind this extraordinary catalytic activity. It is noteworthy that the CuCe-180-0.05 as both CuCe-180-0 and CuCe-180-0.1 illustrated 100% conversion of CO. On the other hand, the samples that were prepared with high concentrations (≥0.5 M) presented similar catalytic behavior with the samples hydrothermally treated at 120 ◦C and 150 ◦C.

#### **3. Discussion**

#### *Correlation of the Physicochemical Characteristics with the Catalytic Activity*

Taking into consideration both the physicochemical characterization and catalytic evaluation, it can be proposed that the high catalytic activity is controlled by the successful combination of specific materials characteristics such as the morphology, oxygen vacancies, and type of copper entities. The specific surface area seems to be the feature with less impact in order to achieve high activity. More specifically:


measurements showed that the high concentration of oxygen vacancies promoted the buildup of adsorbed carbonates that prohibit the adsorption and activation of CO, but not O2. According to several studies, oxygen vacancies tend to form oxygen vacancy clusters [100,101]. Wang et al. [102] showed that oxygen vacancy clusters with suitable size and distribution are responsible for high activity. Thus, it is believed that the various morphologies, which were obtained in this study, can present oxygen vacancies with different size and distribution.

#### **4. Materials and Methods**

#### *4.1. Catalysts Preparation*

All the chemicals used in this work were of analytical reagent grade. Cerium(III) nitrate hexahydrate Ce(NO3)3 · 6H2O (purity 99.99%, Sigma-Aldrich) and a pure metallic copper ring were used as precursors for the preparation of ceria and copper-promoted ceria nanomaterials. Moreover, citric acid monohydrate C6H8O7 · H2O (purity 99.5–101.0%, Ing. Petr Švec PENTA, Prague, Czech Republic) and sodium hydroxide NaOH (purity ≥98.0%, Ing. Petr Švec PENTA, Prague, Czech Republic) were also employed in the synthesis procedure as a chelating agent and precipitating agent, respectively.

At first, 3.7834 gr of Ce(NO3)3 · 6H2O and 1.8314 gr of C6H8O7 H2O were dissolved under continuous stirring into 15 mL of triple distilled (3D) water, respectively. The molar ratio of citric acid to metal nitrate was adjusted according to the stoichiometry of the reaction (citric acid/Ce = 1/1). When the dissolution of the compounds was completed, the two aqueous solutions were mixed under continuous stirring, while at the same time, a 150 mL aqueous solution of NaOH (CNaOH = 0–5 M) was prepared. Both solutions were mixed in a Teflon beaker, and this mixture was kept under stirring for 20 min. The Teflon beaker was placed in a lab-made stainless-steel autoclave (chamber volume of 200 mL), where the copper ring was placed over the beaker (not immersed into the solution). The autoclave was sealed tightly and heated at various temperatures (120, 150, and 180 ◦C) for 24 h. After the hydrothermal treatment, the autoclave was opened, excess water was decanted, and the precipitates were filtered, washed several times with triple distilled water (until pH = 7), and dried under vacuum at 70 ◦C overnight. Finally, the obtained powders were calcined at 400 ◦C for 2 h (heating ramp = 2 ◦C min<sup>−</sup>1).

In order to facilitate the presentation of results, the following encoding of catalysts is used: Ce-T-M and CuCe-T-M, where T represents the temperature (◦C) of the hydrothermal reaction and M the molarity of NaOH (M). For example, the Ce-120-5 catalyst was synthesized hydrothermally at 120 ◦C, using 5 M NaOH.

#### *4.2. Catalysts Characterization*

An X-ray powder diffractometer (Bruker D8 Advance, Bruker, Birmingham, UK) employing Cu Ka radiation (λ = 0.15418 nm) at 40 kV and 40 mA was used to analyze the crystalline structure of the catalysts.

The specific surface area (SSA), the pore volume and the pore size distribution of the materials were determined from the adsorption and desorption isotherms of nitrogen at −196 ◦C using a TriStar 3000 Micromeritics instrument (Norcross, GA, U.S.A). Prior to the measurements, the samples were outgassed at 150 ◦C for 1 h, under N2 flow.

The morphology of the obtained materials was observed with scanning electron microscopy (SEM, Leo Supra 35VP (Carl Zeiss SMT AG Company, Oberkochen, Germany).

Raman spectra were accumulated with the 441.6 nm laser line as the excitation source emerging from a He–Cd laser (Kimon). The scattered light was analyzed by the Lab-Ram HR800 (Jobin-Yvon, Horiba, Montpellier, France) micro Raman spectometer at a spectral resolution of about 2.0 cm−1. A microscope objective with magnification 50× was used to focus the light onto a spot of ~3 μm in

diameter. Low laser intensities were used (∼0.37 mW on the sample) to avoid spectral changes due to heat-induced effects. The Raman shift was calibrated using the 520 cm−<sup>1</sup> Raman band of crystalline Si.

The diffuse reflectance spectra of the obtained materials were recorded in the range 200–800 nm at room temperature using a UV-vis spectrophotometer (Varian Cary 3; Varian Inc. Palo Alto, CA, USA) equipped with an integration sphere. The DR spectra were collected on calcined samples with PTFE disks. The powder samples were mounted in a quartz cell, which provided a sample thickness >3 mm to guarantee the "infinite" sample thickness.

Ceria-based nanomaterials were also characterized by Electron Paramagnetic Resonance (EPR) spectroscopy employing a Bruker ER200D spectrometer (Billerica, MA, USA) equipped with an Agilent 5310A frequency counter (Agilent Technologies, Santa Clara, CA, USA). EPR spectra were recorded at 77 K in suprasil-quarz tubes (3 mm inner diameter; Willmad Glass). 10 mg of nano-powders were introduced into the sample EPR tube followed by outgassing at 300 K for 10 min under 10-4 bar vacuum. The spectrometer was running under home-made software based on LabView. Numerical simulation of experimental EPR spectra was performed with EasySpin 5.2.21 software (The MathWorks Inc., Natick, MA, USA) [103]. Quantification of Ce3+-O–Ce4+ (S = 1/2) was performed using 2,2-diphenyl-1-picrylhydrazyl (DPPH) [104] as a spin standard (Sigma Aldrich). Quantification of Ce3+ (S = 5/2) was performed using FeIII (S = 5/2)-EDTA complex [105,106], while the quantification of the Cu2+ centers was done using a Cu(NO3)2 standard. Copper content was also determined via WDXRF analysis (Wavelength Dispersive X-Ray Fluorescence; ZSX PRIMUS II, RIGAKU, Austin, TX, USA).

#### *4.3. Catalytic Studies*

Activity measurements for the catalytic oxidation of CO were conducted in a conventional fixed-bed reactor (described in detail elsewhere [107]) at atmospheric pressure, in the temperature range of 30–400 ◦C. The catalyst was in the form of powder with a mass of 120 mg and the total flow rate of the reaction mixture was 25 cm<sup>3</sup> min<sup>−</sup>1, yielding a contact time of W/F = 0.288 g s cm−3, where W is the weight of catalyst and F the total flow rate of the reactant gas. The reaction feed stream contained 1 vol.% CO, 20 vol.% O2 and He as balance. Product and reactant analysis was carried out by a gas chromatograph (Shimadzu GC-14B) equipped with a thermal conductivity detector. The CO conversion calculation was based on the CO2 formation or CO consumption:

$$\text{CO conversion} \left( \% \right) = \frac{[\text{CO}\_2]\_{\text{OUT}}}{[\text{CO}]\_{\text{IN}}} \times 100\% \tag{4}$$

#### **5. Conclusions**

In this study, highly active ceria and copper-promoted ceria catalysts were synthesized via a novel hydrothermal method and evaluated in CO oxidation reaction. The physicochemical characteristics, and as a consequence, the catalytic properties were controlled via appropriate combination of hydrothermal parameters (temperature and concentration of the precipitating agent). EPR spectroscopy demonstrated the presence of different copper species (isolated ions and amorphous clusters), dispersed in ceria nanostructrures (mainly nanorods and nanospheres, depending on the pH of the hydrothermal solution). Elevated hydrothermal temperatures and NaOH concentrations favored the formation of isolated copper ions, which resulted to be less active, as compared with copper clusters. Moreover, the catalysts prepared with high concentrations of NaOH (≥0.5 M) presented a significant amount of surface Ce3+ ions, while Raman spectroscopy and UV-Vis DRS measurements revealed the presence of lattice defects and especially oxygen vacancies. Overall, ceria-based catalysts prepared at elevated temperatures with low concentrations of NaOH (≤0.1 M) were more active in CO oxidation, and this behavior can be mainly related with the obtained morphology and the nature of oxygen vacancies and dispersed copper species, and to a lesser extent, with the specific surface area and the concentration of defects.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4344/9/2/138/s1, Figure S1. XRD diffractograms of CeO2 and Cu/CeO2 catalysts: (**a**) Catalysts prepared hydrothermally at 120 ◦C; (**b**) Catalysts prepared hydrothermally at 150 ◦C; (**c**) Catalysts prepared hydrothermally at 180 ◦C, Figure S2. UV-Vis DRS spectra of CeO2 and Cu/CeO2 catalysts: (**a**) Catalysts prepared hydrothermally at 120 ◦C; (**b**) Catalysts prepared hydrothermally at 150 ◦C; (**c**) Catalysts prepared hydrothermally at 180 ◦C.

**Author Contributions:** Conceptualization, G.A.; investigation and formal analysis, K.K., C.P., Y.G., J.V., J.P. and Y.D.; writing—original draft preparation, K.K.; writing—review and editing, G.A.

**Funding:** This research was funded by Research Committee of the University of Patras via "K. Karatheodori" program, grant number (E.610). This research was also co-financed by Greece and the European Union (European Social Fund- ESF) through the Operational Programme "Human Resources Development, Education and Lifelong Learning" in the context of the project "Strengthening Human Resources Research Potential via Doctorate Research" (MIS-5000432), implemented by the State Scholarships Foundation (IKΥ).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

## **Catalytic Properties of Double Substituted Lanthanum Cobaltite Nanostructured Coatings Prepared by Reactive Magnetron Sputtering**

**Mohammad Arab Pour Yazdi 1, Leonardo Lizarraga 2,3, Philippe Vernoux 2, Alain Billard <sup>1</sup> and Pascal Briois 1,\***


Received: 29 March 2019; Accepted: 17 April 2019; Published: 23 April 2019

**Abstract:** Lanthanum perovskites are promising candidates to replace platinum group metal (PGM), especially regarding catalytic oxidation reactions. We have prepared thin catalytic coatings of Sr and Ag doped lanthanum perovskite by using the cathodic co-sputtering magnetron method in reactive condition. Such development of catalytic films may optimize the surface/bulk ratio to save raw materials, since a porous coating can combine a large exchange surface with the gas phase with an extremely low loading. The sputtering deposition process was optimized to generate crystallized and thin perovskites films on alumina substrates. We found that high Ag contents has a strong impact on the morphology of the coatings. High Ag loadings favor the growth of covering films with a porous wire-like morphology showing a good catalytic activity for CO oxidation. The most active composition displays similar catalytic performances than those of a Pt film. In addition, this porous coating is also efficient for CO and NO oxidation in a simulated Diesel exhaust gas mixture, demonstrating the promising catalytic properties of such nanostructured thin sputtered perovskite films.

**Keywords:** perovskite; catalytic coating; CO oxidation; cathodic sputtering method

#### **1. Introduction**

The family of perovskite oxides is known for its catalytic properties, hydrothermal stability, high recyclability and low cost compared with Platinum Group Metal (PGM) [1–3]. The general formula of perovskites is ABO3 where the larger size A-cation presents a 12-coordination number and the B-cation coordinates with 6 neighboring atoms [4]. Partial substitution of A and/or B atoms with other elements showing redox properties may enhance the catalytic activity due to the generation of structural defects such as anionic or cationic vacancies and/or modification of the oxidation state of B cations to maintain the electro-neutrality [5]. Lanthanum perovskites are promising candidates to replace noble metals (Pt, Pd, etc.) [6,7], especially regarding catalytic oxidation reactions. Lanthanum cobaltite (LaCoO3) is one of the most promising catalysts for the oxidation of gaseous pollutants such as carbon monoxide, unburnt hydrocarbons and nitrogen oxide [7,8]. Lanthanum cobaltites are used in many others fields due to their magnetic properties as well as their mixed ionic and electronic conductivity [9–12]. Most of these applications require the implementation of thin films that act as electrodes for fuel cells [13], thermoelectric processes [14], sensors [15], and magnetoresistance devices [16]. These electric, magnetic and electrocatalytic properties depend on the composition but also on the microstructure and morphology of coatings, which strongly depend on the preparation method [17,18]. Several techniques are reported for the synthesis of perovskite thin films such as the ground-frost [7], casting in band [18], chemical vapor deposition (CVD) [17], spray painting [19], sol-gel [7], atomic layer deposition (ALD) [20], laser pulsed deposition (PLD) [21], and physical vapor deposition (PVD) [22,23]. Regarding catalytic applications, the development of submicrometric catalytic films may optimize the surface/bulk ratio to save raw materials. Indeed, a thin catalytic coating can combine a large exchange surface with the gas phase with an extremely low loading. This is particularly suitable for air cleaning since gaseous pollutants only lick the surface of the catalyst. In addition, thin catalytic coatings could be deposited on hot substrates such as collector walls in thermal engine exhausts for removing pollutants or of radiators for improving indoor air quality. Few studies on the development of thin perovskite coatings for catalytic applications are reported in the literature [24–26]. The challenge lies in preparing adherent, thermally stable and pure crystallized perovskite films without any secondary parasite phase [18], showing appropriate compositions and nanostructures for catalysis [7,8,16,18]. The porosity and the specific surface areas have to be optimized to counter-balance the low quantity of materials involved in submicrometric catalytic coatings. In the present study, we used the cathodic co-sputtering magnetron method in reactive condition to prepare submicrometric nanostructured coatings of perovskites. This technique allows the deposition of pure and adherent coatings of complex oxides with a controlled and reproducible manner, while respecting the environment [27,28]. We have chosen to prepare Sr and Ag-doped lanthanum cobaltite as this perovskite composition is one of the most active for CO oxidation [8]. The partial substitution of La3<sup>+</sup> cations by Sr2<sup>+</sup> ones improves the thermal stability and then the specific surface area of pure LaCoO3 [29] and also enhances the surface concentration of oxygen vacancies involved in the oxidation catalytic mechanism [30]. The partial substitution of La3<sup>+</sup> by Ag<sup>+</sup> can also increase the catalytic properties of lanthanum cobaltites for oxidation reactions due to the formation of oxygen vacancies [31] and the stabilisation of Ag nanoparticles on the oxide surface [6,32]. The different parameters (intensities of current applied on the metallic targets, total pressure in the chamber, oxygen partial pressure, etc.) of the reactive magnetron sputtering preparation method were tuned to achieve pure and adherent coatings of LaCoO3 on alumina dense membranes. After a calcination at 500 ◦C, crystallized and dense cubic perovskite films of around 1.5 μm thick were achieved with the targeted La/Co stoichiometry. Different compositions of La1-*x*-*y*Sr*x*Ag*y*CoO3-<sup>α</sup> (*x* = 0.13–0.28, *y* = 0.14–0.48) doped perovskites were sputtered on alumina disks. We found that the incorporation of high contents of Ag can strongly modify the morphology of the coatings, increasing their porosity. The catalytic performance of the perovskite catalytic coatings for CO oxidation was found to be improved by the double substitution of Sr and Ag. The most active composition, La0.40Sr0.1Ag0.48Co0.93O3, displays similar catalytic performances than those of a Pt film, for CO oxidation. In addition, this porous coating is also active for CO and NO oxidation in a simulated Diesel exhaust gas mixture, demonstrating the promising catalytic properties of such nanostructured thin sputtered perovskite films.

#### **2. Results**

#### *2.1. Preparation and Characterization of LaCoO3 Catalytic Coatings*

Layers of LaCoO3 were synthetized on alumina substrates by using the reactive magnetron sputtering preparation method. Depositions have been performed in a reactive mode, mixing oxygen and argon in the chamber. To achieve the suitable La/Co ratio, the discharge current applied to the Co target was adjusted while maintaining a constant current of 1 A on the La target (Table 1). As expected, the La/Co ratio, estimated by Energy Dispersive Spectroscopy (EDS), decreases with the current dissipated on the Co target (Figure 1). We determined that a current intensity of 0.3 A is required to reach the targeted La/Co ratio. The diffractogram recorded on this as-deposited coating

(Figure 2) shows that the layer is amorphous as no peak is detected. The difference between the radii of La3<sup>+</sup> (136 pm) and Co3<sup>+</sup> cations (72 pm) suggests an amorphous as-deposited coating due to steric effects as predicted by the confusion principle [33–35]. Then, a calcination step was performed at different temperatures (from 100 ◦C to 500 ◦C) for 2 h in air. X-Ray Diffraction (XRD) patterns (Figure 2), obtained under a Bragg-Brentano configuration (θ/2θ), evidence that the oxide coating crystallizes from 500 ◦C as a cubic perovskite phase (JCPDS 01-075-0279). This temperature is approximately 100 ◦C lower than those reported by H. Seim et al. [20] and H.J. Hwang et al. [36] on LaCoO3 films prepared by PLD and sol-gel methods, respectively. This demonstrates that the reactive magnetron sputtering technique is efficient for preparing crystallized lanthanum cobaltite films at rather low temperatures.


**Table 1.** Sputtering parameters used for LaCoO3 coating.

\* sccm = Standard Cubic Centimeter per Minutes.

**Figure 1.** Evolution of the atomic La/Co ratio measured by Energy Dispersive Spectroscopy (EDS) as a function of the current dissipated on the Co target (ILa = 1 A). The total pressure is 1.5 Pa.

This calcination step at 500 ◦C was performed for all coatings (Figure 3). The thin film with the highest La/Co ratio is not crystallized and only presents small peaks of La2O3 (JCDPS 00-005-0602), probably coming from a high excess of La. On the opposite, for larger atomic ratios, coatings are crystallised with various space groups of the perovskite-type structure. Indeed, the film is cubic (Pm3m space group) for a ratio of 0.87, exhibits a rhomboedric symmetry (R-3c space group) for a ratio of 0.56 and again becomes cubic for a ratio of 0.3.

**Figure 2.** X-Ray diffractograms of LaCoO3 sputtered films prepared with a current dissipated on the Co target of 0.3 A (La/Co = 0.87) after different post-calcination treatments during 2 h in air from 100 ◦C to 500 ◦C.

**Figure 3.** XRD patterns as a function of the atomic composition ratio after calcination treatment at 500 ◦C for 2 h under air.

The morphology of the LaCoO3 coating with the suitable ratio (La/Co = 0.87) was observed by SEM. Figure 4 shows SEM images of the top view and the brittle cross section of the sample before and after the calcination step at 500 ◦C. The as-deposited coating (Figure 4a,c) covers the surface of the alumina substrate and follows its morphology. This film is quite dense, as shown in the cross-section image (Figure 4a) and adherent with a vitreous appearance characteristic of an amorphous material, in agreement with XRD (Figure 2) [37]. After calcination (Figure 4b), some cracks can be observed especially in the cross section (Figure 4d). The crystallized film is not sticking well with the alumina support and remains quite dense. The cracks were probably formed under stresses during the crystallization or the thermal treatment. In this latter case, the mismatch of thermal expansion coefficients between the perovskite film and the alumina substrate could be at the origin of the delamination (αLaCoO3 <sup>≈</sup> 20 <sup>×</sup> 10−<sup>6</sup> ◦C−<sup>1</sup> [38] and <sup>α</sup>Al2O3 <sup>≈</sup> 7 <sup>×</sup> 10−<sup>6</sup> ◦C−<sup>1</sup> [39]) of the coating. The thickness of the annealed coating is around 1.5 μm (Figure 4c,d), resulting in a 500 nm.h−<sup>1</sup> deposition rate. This result is in agreement with the thickness measured by tactile profilometry.

**Figure 4.** SEM images of the surface and the brittle cross section of the LaCoO3 sample synthesized with ICo = 0.3 A and ILa = 1 A: (**a**,**c**) as deposited coating and (**b**,**d**) after calcination treatment at 500 ◦C for 2 h under air.

#### *2.2. Preparation and Characterization of La1-x-ySrxAgyCoO3-*<sup>α</sup> *(LSACO) Catalytic Coatings*

The experimental parameters used for the synthesis of double substituted cobaltite (denoted as LSACO) coatings on alumina supports are reported in Table 2. The protocol is similar to that developed for the synthesis of pure lanthanum cobaltite (Ar flow rate = 50 sccm, O2 flow rate = 20 sccm, total pressure = 1.5 Pa, draw distance = 45 mm) except the sputtering time which was extended to 4 h to achieve thicker films. The control of the substitution rates of Sr and Ag was achieved by tuning the intensity of the currents dissipated on the metallic targets (Table 2). We assumed that Sr2<sup>+</sup> and also Ag<sup>+</sup> cations will partially substitute La3<sup>+</sup> cations in A sites of the perovskite. This hypothesis is based on the similarity of the ionic radius of La3<sup>+</sup> (136 pm) and Ag<sup>+</sup> (128 pm) [40]. The Ag content in the perovskites was gradually enhanced by increasing the current applied to the Ag target (LSACO-1 to 4). This was counter-balanced by a progressive decay of the current dissipated to the La target. The Co content has been slightly increased in LSACO-5 compared with LSACO-4, while maintaining constant the Ag, La, and Sr contents (Table 2)


**Table 2.** Sputtering parameters used for *LSACO* coatings.

The chemical composition of the as-prepared LSACO coatings was estimated by EDS analysis. Table 3 displays the variations of the atomic percent of each element, without considering the oxygen

level which is tricky to quantity by using EDS. The Sr and Ag contents inversely vary in LSACO-1, 2 and 3 while La and Co loadings are fairly stable. Let us note that the chemical composition of LSACO-2 and LSACO-3 is fairly similar with 11 at% of Ag and around 10 and 30 at% of Sr and La, respectively. The high current applied to the Ag target during the preparation of LSACO-4 and LSACO-5 leads to a significant increase in the Ag concentration and a concomitant drop of the La content. In LSACO-4 and LSACO-5, the Ag atomic percent reaches around 25% while the La level is approximately 20%. Surprisingly, the Co concentration decreases in LSACO-5 despite the largest applied current to the Co target. Except LSACO-5, the atomic ratio between cations located in A sites (theoretically La3<sup>+</sup>, Sr2<sup>+</sup> and Ag+) and Co3<sup>+</sup> ones located in B sites is close to the target of 1. The morphology of the as-deposited coatings (Figure 5) was found to drastically change with the chemical composition. The LSACO-1 coating only contains few perovskite clusters (Figure 5a, white spots at the top right) dispersed on the surface of the micrometric alumina grains (Figure 5, dark grey), despite of the 4 h deposition time. For LSACO-2 and LSACO-3, the number of perovskite clusters significantly increases on the surface of the alumina grains (white spots in Figure 5b,c). Interestingly, larger starfish shaped perovskites islands are growing on the substrate defects (cavity of alumina substrate or grain boundaries) showing arms like filaments. The morphology of the surface of LSACO-4 and LSACO-5 coatings (Figure 5d,e) are quite different. The alumina substrate is now fairly fully covered by a porous film with a wire-like morphology (Figure 5d,e). These wires have grown parallel to each other, leading void between each other, reaching a length of the order of 1 μm and a diameter of around 100 nm. Therefore, high Ag contents seem to strongly enhance the porosity and then the coverage of the films.

**Table 3.** Chemical composition determined by EDS of the as-prepared La1-x-ySrxAgyCoO3-<sup>α</sup> (LSCAO) perovskite coatings.


**Figure 5.** SEM images of the surface morphology of as-deposited LSACO coatings on alumina substrate: (**a**) LSACO-1, (**b**) LSACO-2, (**c**) LSACO-3, (**d**) LSACO-4, and (**e**) LSACO-5.

These series of LSACO coatings were annealed at 500 ◦C for 2 h in air. This calcination step was sufficient to crystallize all the films as a cubic perovskite phase (JCPDS 01-075-0279) whatever the composition. On the other hand, an additional XRD pattern at 44.3◦ on the diffractogramm of LSACO-4 and LSACO-5, corresponding to (2 0 0) planes of fcc metallic silver, proves the presence of metallic Ag, out of the perovskite structure. These results demonstrate the limited solubility of Ag in the perovskite in agreement with previous studies [32]. For high contents of Ag, reaching 25 at. %, i.e., around 10 at% in the overall oxide including the oxygen atoms, part of Ag is not incorporated into the perovskite structure. No XRD peaks corresponding to Ag◦ were observed for LSAC0-1, LSACO-2, and LSACO-3 (Figure 6), but the low content of Ag, below 5 at%, makes difficult their detection. Therefore, we cannot exclude the presence of metallic Ag on the surface of these samples.

**Figure 6.** XRD patterns after the calcination treatment at 500 ◦C for 2 h under air of the different LSACO coatings.

Figure 7 displays the surface morphology of the calcinated films. The alumina substrates are cracked probably due to the thermal treatment. The LSACO-1 coating only contains isolated perovskites clusters on the surface of alumina grains. As before calcination, the morphology of LSACO-2 and LSACO-3 coatings are similar with perovskite islands mainly located in the cavity and interstices of the alumina substrate. Filaments that were present around of the perovskite clusters have disappeared. Coatings containing high Ag loadings (LSACO-4 and LSACO-5) are the only ones able to fully cover the alumina substrate with a porous wire-like morphology. These nanowires are less ordered than before calcination and interlock, then decreasing the porosity of the film. Ag particles (Figure 7e) can be observed on the surface of LSACO-5 (white particles), confirming that a part of Ag was not incorporated into the perovkiste structure.

**Figure 7.** SEM images of the surface morphology of calcined LSACO coatings on alumina substrate: (**a**) LSACO-1, (**b**) LSACO-2, (**c**) LSACO-3, (**d**) LSACO-4, and (**e**) LSACO-5.

#### *2.3. Catalytic Performances of the Perovskite Coatings*

#### 2.3.1. Catalytic Performances for CO Oxidation

The catalytic activity of LSACO-2, LSACO-4, and LSACO-5 coatings, containing respectively 11, 25 and 28 at% of Ag, was measured for CO oxidation during heating ramps up to 400 ◦C. We have selected LSACO-4 and LSACO-5 as these films exhibit a porous and wire-like morphology. We have also tested the catalytic performances for CO oxidation of pure lanthanum cobaltite coatings. These samples have shown no activity below 500 ◦C in good agreement with their dense morphology. The catalyst LSACO-2 was also tested for comparison as a non-porous and non-covering representative layer. In addition, we have also tested, for comparison, a Pt coating (3 μm thick, 5 mg Pt/cm2) prepared by spray-painting and annealed at 500 ◦C. Figure 8 displays the Light-off (LO) curves of the different catalytic coatings for CO oxidation. The results show that the catalytic performances increase with the silver content in the film. Values of T20, temperature at 20% conversion, significantly decrease from 280 to 215 ◦C when increasing the Ag content. Furthermore, the catalytic performances of LSACO-5 are rather close to those of the Pt coating up to 220 ◦C. This underlines the promising catalytic activity of the sputtered Ag-doped perovskite coating.

Two successive LO up to 400 ◦C have been performed on LSACO-4 (Figure 9a). The value of T20 increases from 215 to 242 ◦C during the second LO. This indicates that a modification of the coating morphology during the first LO, most probably due to the Ag particles sintering. Nevertheless, the onset temperature slightly decreases from 175 to 150 ◦C (Figure 9a). Similar catalytic experiments have been carried out on LSACO-5 (Figure 9b) without any significant modification of the catalytic performances between the two successive LO. This indicates a better stability of the morphology of LSACO-5 in the presence of the reactive mixture up to 400◦C compared with LSACO-4. XPS measurements performed after catalytic tests evidence a La surface seggregation with a concommitant drop of the Co concentration. The two catalytic coatings show a similar Ag surface atomic concentration, i.e., around 10 at%, which is far lower from the loading estimated by EDS before the catalytic tests (Table 3). The Ag3d XPS peaks (Figure 10) show a binding energy of Ag3d5/2 at 367.7 eV, attributed to metallic silver. The catalytic properties during the second LO of the two catalytic perovskite coatings are fairly similar, in good agreement with an equivalent Ag surface concentration (Table 4). This confirms the direct link between the surface Ag concentration and the catalytic activity for CO oxidation.

**Figure 8.** Light-off curves for CO oxidation recorded on the sputtered perovskite coatings and on Pt film. Reactive mixture: CO/O2 = 3000 ppm/3%. Overall flow = 3.6 L/h.

**Figure 9.** Two successive LO recorded on (**a**) LSACO-4 and (**b**) LSACO-5. Reactive mixture: CO/O2 = 3000 ppm/3%. Overall flow = 3.6 L/h.

**Figure 10.** Ag3d XPS spectrum of LSAC-5.


**Table 4.** Surface composition determined by XPS of LSACO-4 and LSACO-5 after catalytic tests.

2.3.2. Catalytic Performances in a Model Lean Diesel Exhaust Gas

The catalytic performances of LSACO-4 were explored in a model lean diesel exhaust gas mixture containing 8% O2, 950 ppm CO, 270 ppm NO, 1000 ppm C3H8 and 10% H2O. The LO was recorded up to 375 ◦C to investigate the low temperature activity. Figure 11 shows the ability of the catalytic coating to oxidize CO and NO into CO2 and NO2, respectively. Below 350 ◦C, we did not observed any propane oxidation. However, the catalytic coating shows a remarkable activity for NO oxidation from around 225 ◦C. The NO conversion achieves 30% at 375 ◦C. CO conversion starts from 250 ◦C but seems to reach a maximum at only 18% from 350 ◦C. These results show that, despite a very low mass of catalyst and a complex mixture including the inhibiting effect of H2O, the perovkite catalytic coating can be active at low temperature for low temperature NO and CO oxidation.

**Figure 11.** Variation of CO and NO conversion as a function of temperature on LSACO-4. Reactive mixture: CO/O2/NO/C3H8/H2O: 950 ppm/8%/270 ppm/1000 ppm/10%. Overall flow: 3.6 L/h.

#### **3. Materials and Methods**

#### *3.1. Preparation of the Catalytic Coatings*

LaCoO3 perovskite coatings were synthesized by magnetron sputtering from two metallic targets of La (kurt J. Lekser, purity 99.9%, Hastings, England) and Co (kurt J. Lekser, purity 99.9%). The Alcatel SCM 604 reactor, described elsewhere [33], was a 90-l sputtering chamber where a pressure of 10−<sup>4</sup> Pa was maintained by a primary pump assisted by a molecular turbo pump. The La metallic target (thickness = 3 mm; diameter = 50 mm) and the Co one (thickness = 1 mm; diameter = 50 mm) were attached to two magnetrons distant from 120 mm and connected to a pulsed direct current generator (PINACLE+ purchased by advance energy). The distance between the two targets and the substrate holder (DT-S) can be independently modified. The draw distance between the targets and the alumina substrate was adjusted to 45 mm (Table 1). The targets are powered with pulsed currents to avoid any electric micro-arcs that can damage the quality of the films [41]. To avoid any accumulation of positive charge (Ar+) on the targets, the current is periodically stopped for very short dead time of 5 μS. Gases were introduced with mass-flow controllers (Brooks, 5850 SLA, Hatfield, PA, USA) and the total working pressure was measured with a MKS Baratron gauge. Films were deposited on dense alumina pellets (Keral 99, diameter = 16 mm, thickness = 0.60 mm purchased by Kerafol Gmbh, (Koppe Platz 1, Eschenbach i. d. Opf). The synthesis of doped perovskites with Sr and Ag was performed with the addition of two magnetrons into the sputtering chamber and metallic targets of Sr and Ag (thickness = 3 mm diameter = 50 mm, kurt J. Lekser, purity 99.9%). The Sr and Ag targets were powered by a two pulsed direct current generator (PINACLE+ and MDX500 both purchased by advance energy, Metzingen, Germany).

#### *3.2. Characterizations of the Catalytic Coatings*

The crystal structure of the coatings was determined thanks to a BRUKER D8 focus X-Ray diffractometer (Co Kα1+α2 radiations, in Bragg Brentano configuration and equipped with a LynxEye linear detector ( Bruker, Billerica, MA, USA). XRD patterns were collected under air during 10 min in the (20–80◦) scattering angle range by steps of 0.019◦. The surface and the morphology of films were observed with a scanning electron microscope (JEOL JSM 7800F, Akishima, Japan) equipped with an EDS detector allowing the estimation of the chemical composition of samples. The observation of brittle cross section by SEM led also to the determination of the thickness of films. Coating thickness was also determined by the step method with an Altysurf profilometer produced by Altimet society (Marin, France) equipped with tungsten micro force probe inductive allowing an accuracy of about 20 nm. Before each measurement, the calibration of the experimental device was realized with a reference sample number 787569 accredited by CETIM organization.

XPS spectra were recorded for each catalysts on an AXIS Ultra DLD from Kratos Analytical (Manchester, UK) using a monochromatized Al X-ray source (h*v* = 1486.6 eV) between 0 and 1200 eV with a pass energy of 40 eV. Sample were pretreated at 200 ◦C in He before measurements to clean the surface. Peaks were referenced using C1S peak of carbon (BE = 284.6 eV).

#### *3.3. Catalytic Activity Measurements*

A quartz reactor was operated under continuous flowing conditions at atmospheric pressure. The samples were placed on a fritted quartz, 18 mm in diameter, with the catalytic coating side facing the fritted quartz [42]. Gases were mixed by using mass-flow controllers (Brooks) to generate the different reactive mixtures. H2O vapor was introduced using an atmospheric pressure Pyrex saturator heated at 46 ◦C. Reactants and products were analyzed using a gas micro-chromatograph (SRA 3000 equipped with two TCD detectors, a molecular sieve and a Porapak Q column for O2, CO, C3H6, and CO2 analysis) and a CO2 Infra-Red analyzer (Horiba VA 3000, Horiba Europe Gmbh, Leichlingen, Germany). NO and NO2 concentrations were measured with IR and UV online analyzers (EMERSON NGA2000).

Reactants were Air Liquide certified standards of NO in He (8000 ppm), C3H8 in He (8005 ppm), CO in He (1%), O2 (99.999%), which could be further diluted in (99.999%). The carbon and nitrogen balance closure was found to be within 2%.

#### **4. Conclusions**

Different compositions of La1-*x*-*y*Sr*x*Ag*y*CoO3-<sup>α</sup> (*x* = 0.13–0.28, *y* = 0.14–0.48) doped perovskites were synthetized as thin coatings deposited on alumina disks with the cathodic co-sputtering magnetron method in reactive conditions. The control of different parameters during the sputtering process can tune the morphology of the catalytic films. In particular, we found that the incorporation of high Ag loadings can generate covering films with a porous wire-like morphology showing good catalytic activity for CO oxidation. The most active composition, La0.40Sr0.1Ag0.48Co0.93O3, displays similar catalytic performances than those of a Pt film. In addition, this porous coating is also efficient for CO and NO oxidation in a simulated Diesel exhaust gas mixture, demonstrating the promising catalytic properties of such nanostructured thin sputtered perovskite films.

**Author Contributions:** Investigation, M.A.P.Y. and L.L.; data curation, P.B. and P.V.; writing—original draft preparation, M.A.P.Y, P.B., and P.V.; writing—review and editing, P.B. and P.V.; formal analysis: A.B., P.V. supervision, A.B., P.B. and P.V.; project administration, P.B.; funding acquisition, P.B., P.V. and A.B.

**Funding:** The research was funded by "ADEME" in the frame of the ADIABACAT project and "Pays de Montbéliard Agglomération".

**Conflicts of Interest:** The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Effect of Preparation Method of Co-Ce Catalysts on CH4 Combustion**

#### **Sofia Darda 1,2, Eleni Pachatouridou 1, Angelos Lappas <sup>1</sup> and Eleni Iliopoulou 1,\***


Received: 21 January 2019; Accepted: 22 February 2019; Published: 27 February 2019

**Abstract:** Transition metal oxides have recently attracted considerable attention as candidate catalysts for the complete oxidation of methane, the main component of the natural gas, used in various industrial processes or as a fuel in turbines and vehicles. A series of novel Co-Ce mixed oxide catalysts were synthesized as an effort to enhance synergistic effects that could improve their redox behavior, oxygen storage ability and, thus, their activity in methane oxidation. The effect of synthesis method (hydrothermal or precipitation) and Co loading (0, 2, 5, and 15 wt.%) on the catalytic efficiency and stability of the derived materials was investigated. Use of hydrothermal synthesis results in the most efficient Co/CeO2 catalysts, a fact related with their improved physicochemical properties, as compared with the materials prepared via precipitation. In particular, a CeO2 support of smaller crystallite size and larger surface area seems to enhance the reducibility of the Co3O4/CeO2 materials, as evidenced by the blue shift of the corresponding reduction peaks (H2-TPR, H2-Temperature Programmed Reduction). The limited methane oxidation activity over pure CeO2 samples is significantly enhanced by Co incorporation and further improved by higher Co loadings. The optimum performance was observed over a 15 wt% Co/CeO2 catalyst, which also presented sufficient tolerance to water presence.

**Keywords:** Co3O4; CeO2; complete CH4 oxidation; hydrothermal synthesis; precipitation

#### **1. Introduction**

Today, CH4, N2O, and CO2 emissions represent approximately 98% of the total greenhouse gas (GHG) inventory worldwide, a percentage expected to increase even further in the 21st century. Methane is the main compound of natural gas (NG) and is today used in various industrial processes and also as energy source in gas turbines or natural gas fuelled vehicles. Heavy-duty (HD) natural gas–fuelled fleet represents today less than 2% of the total fleet, however this percentage is expected to grow to represent as much as 50% in the near future. Moreover, methane is considered a substantially more potent greenhouse gas (GHG) than CO2 and, hence, leakage of only a small percentage of methane from the supply chain could alter the net climate benefits of NG, which is lately suggested as an alternative fuel for heavy-duty (HD) transportation. Methane can be also a result of incomplete combustion processes, similarly to CO, while both of them are well-known environmentally detrimental pollutants and, thus, proper catalytic oxidation converters are needed for their abatement [1,2]. Noble metals and transition metals are the most commonly used components of such catalysts, investigated for methane and/or CO complete oxidation. Both these groups of catalysts have been extensively studied in order to develop efficient catalytic systems for combustion of methane, recognized as the lowest reactivity molecule among alkanes. All up to date research

efforts have disclosed the superiority of the noble metals (especially Pt, Pd, and their mixtures) on oxide supports [3–5]. However, there are many factors defining the effectiveness of these supported catalysts, including the nature and properties of the support, the metal loading, the size, shape, and electronic state of metal nanoparticles and their interaction/synergy with the support, their preparation method and, finally, any pretreatment conditions possibly applied for their improvement [6]. However, in spite of their excellent activity, their applications at a larger scale have been limited because of their prohibitively high cost, their shortage in worldwide reserves, and their prompt deactivation during methane oxidation, mainly due to sintering of supported noble metal nanoparticles [7]. On the other hand, transition metal oxides have recently attracted considerable attention as alternative catalytic materials, among which cobalt oxide is considered as the most promising catalyst for methane combustion. Most crucial property of the early transition metal oxides is the fact that they can generate oxygen vacancies, suggested as the necessary sites that can activate molecular oxygen for oxidation applications [5]. Co3O4 nanosheets are reported to show better catalytic activity than Co3O4 nanobelts and nanocubes for CH4 oxidation, in spite of their low specific surface area. Co3O4 catalysts with various structures (nanoparticles, two-dimensional and three-dimensional structures) were also investigated for methane catalytic combustion, exhibiting that enhanced catalytic performance of methane of the 2D-Co3O4 and 3D-Co3O4 catalysts is related with their noticeable reducibility and existence of abundant active Co3+ species [8,9]. Highly porous Co3O4 nanorods, with narrow pore-size distribution and a high surface area were synthesized by simple hydrothermal methods and similarly exhibited an enhanced catalytic performance for CH4 oxidation, especially at high GHSV (Gas Hourly Space Velocity) conditions [10]. Co3O4 nanomaterials were also synthesized by hydrothermal method and tested for low temperature CO oxidation and methane combustion in another report. In that case, addition of ethylene glycol and its varying concentration significantly influenced the size and shape of the Co3O4 oxides (resulting in nanosheets or nanospheres). During catalyst evaluation, it was evidenced that the Co3O4 nanosheets exhibited more active oxygen species than the Co3O4 nanospheres, a fact accounting for their higher activity in the oxidative reactions [11].

In addition to cobalt, several ceria-based, low-cost catalysts have recently attracted much attention as a reliever of costly, unstable noble metal catalysts, since they exhibit good performance in oxidation reactions of carbon monoxide and lower hydrocarbons. Once more, the synthesis method is very crucial; for example nanocrystalline CeO2, as small as 5 nm, were prepared by the precipitation method, using hydrogen peroxide as an oxidizer and is reported to present even 100 ◦C lower methane oxidation temperature, as compared with microcrystalline CeO2. Nabih et al. has also reported the synthesis of mesoporous ceria nanoparticles (NPs) by combining the sol-gel (SG) process with the inverse mini-emulsion technique. These materials were of high specific surface area and exhibited an even better catalytic performance than the reference samples, synthesized by precipitation [12,13]. The addition of MO*<sup>x</sup>* (M: di- or tri-valent transition metal ion) in CeO2 is interestingly reported to promote the catalytic activity for the oxidation reaction, a fact possibly related with improved redox properties and high oxygen storage capacity, induced by the synergistic effect between CeO2 and MO*<sup>x</sup>* [14–17]. A nanocomposite catalyst Co3O4/CeO2 is very recently reported to exhibit high activity in complete oxidation of CH4, suggesting a synergistic effect of CeO2 nanorods and the supported Co3O4 nanoparticles [7].

In addition to oxidation performance, catalyst tolerance to deactivation, caused by poisoning and thermal aging, is also a problem in all natural gas applications. Poisoning is either due to adsorption of impurities, e.g., S, P, Zn, Ca, and Mg, present in the exhaust gases, on the catalytic active sites or their reaction with the active sites causing the formation of non-active compounds. Sulfur in the NG vehicles exhaust gas actually originates from odorants and lubricating oils and less from gas itself. Sulfur is considered as the most noxious component readily deactivating Pd catalysts, recognized as the most efficient for methane oxidation. Only small amounts of SO2 can significantly inhibit catalyst activity by blocking the active noble metal sites by sulfur compounds [18–21]. Moreover, water is always present in the process of the catalytic combustion of methane, either as gas humidity

or as a combustion product. Water presence may affect the reaction kinetics, resulting in a serious inhibition of the oxidation reactions [14]. Thus, activity of Pd/Al2O3 is reported as severely inhibited by water at temperatures below 450 ◦C, a fact attributed to the formation of surface hydroxyl groups, which blocks active catalyst sites, while suppressed oxygen exchange between support and PdO via the formed hydroxyl groups is another recent explanation. Based on experiments and kinetic modelling two are the suggested routes of the water inhibition effect; rapid adsorption of water species on the active sites and slow build-up of hydroxyl species. Formation of these surface hydroxyls is enhanced by high concentration of oxygen and high concentration of water vapor [22]. In general, while various mechanisms have been proposed in the literature to explain the deactivation observed due to water presence, all studies agree on a strong effect of the support. Thus, tuning opportunely the metal-support interaction and the oxygen exchange capability of the support is a suggested way to deal with water inhibition problem over Pd/CeO2 catalysts, otherwise very highly active for methane oxidation in dry conditions [23]. In a recent study, Pt addition was suggested to improve the performance of Pd-modified Mn-hexaaluminate catalyst in the high-temperature oxidation of methane, especially in SO2 and water presence, correlating the improved water and sulfur resistance with the presence of particles of PtPd alloy [24]. In a similar study, presence of water vapor is reported to significantly inhibit the catalytic performance of CeO2-MO*<sup>x</sup>* (M = Cu, Mn, Fe, Co, and Ni) mixed oxide catalysts for CH4 combustion, while CeO2-NiO samples showed the best durability for CH4 wet combustion (T50: 528 ◦C, when 20% water vapor was added in the reaction stream) among all materials tested [14]. Interestingly, water presence in the reaction feed is even reported to promote methane conversion over a NiO catalyst, prepared from thermal decomposition of the corresponding metal nitrate. As activation of oxygen plays a crucial role in methane oxidation, in the surface kinetics controlled region, the observed high catalytic efficiency of NiO is attributed to its larger capability for oxygen adsorption, even at 500 ◦C. The promoting role of water could be due to the fact that H2O molecules could modify the NiO surface and, thus, promote further activation of O2 and/or CH4 on the surface [25].

In a series of recent studies, we have investigated the impact of synthesis parameters on the solid state properties of CeO2 materials. In detail, we followed four different, time- and cost-effective, preparation methods, i.e., thermal decomposition, precipitation, and hydrothermal method of low and high NaOH concentration, employing in all cases Ce(NO3)3·6H2O as the cerium precursor. A thorough characterization of all samples was carried out to gain insight into the impact of synthesis route on the textural, structural, morphological, and redox properties of the derived materials. The results revealed the superiority of the hydrothermal method towards the development of ceria nanoparticles of high specific surface area (>90 m2 g−1), well-defined geometry (nanorods) and improved redox properties. Employing CO oxidation as a probe reaction evidenced a direct quantitative correlation between the catalytic activity and the abundance of easily reduced, loosely bound oxygen species. More specifically, the rod-like morphology of the hydrothermally synthesized CeO2 nanoparticles, with well-defined (100) and (110) reactive planes, favored the enhanced reducibility and lattice oxygen mobility, rendering this material appropriate as catalyst or supporting carrier [26–28]. The current study aims in the preparation of novel Co-Ce catalytic materials targeting in the complete combustion of CH4. Co incorporation on an optimum selected CeO2 support aims to promote the catalytic activity for the oxidation reactions, via an attempted improvement of the redox property and high oxygen storage capacity, deliberately induced by the synergistic effect between CeO2 and CoO*x*. We purposely explored the effect of both the synthesis method of the CeO2 support (following either the hydrothermal or precipitation technique) and the metal loading (0, 2, 5, and 15 wt% Co) on the catalytic performance and durability for complete methane oxidation. Deactivation studies were also performed examining tolerance of the optimum selected catalysts to thermal aging and presence of water vapor in the reaction feed (wet CH4 combustion).

#### **2. Results and Discussion**

#### *2.1. Catalyst Characterization*

#### 2.1.1. Textural and Structural Characterization

The physicochemical properties of catalytic materials are presented in Table 1. As expected [26], the synthesis method of CeO2 greatly affects the specific surface area, the pore volume and the average pore diameter of the carrier. In addition, the mean particle size of CeO2 presents important differences between the two synthesis methods. Applying the hydrothermal method (CeO2-H), the formed ceria presents higher surface area (110.9 m2/g) and smaller CeO2 crystallites (11.8 nm) than the precipitated ceria (CeO2-P), where the growing up of the ceria crystallites (18.6 nm) resulted in decreased surface area (49.5 m2/g). The effect of synthesis method on the crystallite size is probably related with slower rates of crystal migration and crystallites grow up, when following the hydrothermal synthesis method. Incorporation of cobalt, on both CeO2 carriers, as well as the increase of its metal loading, affects the texture of the CeO2 support, decreasing the specific surface area of the derived, CeO2-supported, Co catalysts. On the contrary, in the case of pure Co3O4, the different synthesis method does not appear to significantly affect the specific surface area. However, applying the hydrothermal method (Co3O4-H), the formed Co oxide presents lower pore volume and average pore diameter and similarly smaller Co3O4 crystallites than Co3O4-P (33.3 nm as compared to 42.2 nm, respectively). Lower crystallite sizes (16.1–16.8 nm) of Co3O4 nanomaterials, synthesized by hydrothermal method are reported in the recent literature, probably related with lower calcination temperature, thus limiting aggregation of Co3O4 crystallites in the sample [11]. It is noteworthy that much smaller Co3O4 crystallites are also formed on the CeO2-H rather than the CeO2-P support (16.0 as compared to 76.6 nm, respectively), when incorporating 15 wt% Co on either support via wet impregnation, a fact indicating synergetic effects between the CeO2-H support and the incorporated Co oxide.


**Table 1.** Physicochemical properties of catalytic materials.

<sup>1</sup> Measured by XRD diffractograms using the Scherrer equation, at (2θ) 28.6◦, 47.5◦ for CeO2 and 31.3◦, 36.9◦ for Co3O4.

Representative XRD (X-Ray Diffraction) patterns of bare CeO2, Co3O4 and *x*Co/CeO2 prepared with either the precipitation or the hydrothermal method, are shown in Figure 1a,b respectively. As expected, both samples of CeO2 supports (CeO2-P and CeO2-H) exhibit characteristic peaks predominately located at 2*θ*: 28.6◦, 33.1◦, 47.5◦, 56.36◦, 59.1◦, 69.4◦, 76.9◦, and 79.2◦ and have face centered cubic structure with the lattice parameters *a* = *b* = *c* = 5.411 Å and *α* = *β* = *γ* = 90◦ [29], which are also preserved in the Co-supported catalysts.

Bare Co oxides (Co3O4-P and Co3O4-H) show diffraction peaks at 2*θ*: 19.0◦, 31.3◦, 36.9◦, 38.5◦, 44.8◦, 55.9◦, 59.4◦, and 65.2◦, which are attributed to the cobalt oxide Co3O4 and suggest that the cobalt precursor led to the formation of a face centered cubic unit cell of Co3O4 (space group *Fd*3*m*) with a

spinel type structure after calcination [9,11]. At low Co loadings (2 wt.%), no diffraction peak related to Co3O4 was observed, while increasing Co loading (to 5 wt.% and 15 wt.%) leads to the development of low intensity Co3O4 peaks (2*θ*: 31.3◦, 36.9◦, and 44.8◦) for Co/CeO2-P (Figure 1a) and Co/CeO2-H (Figure 1b) catalytic materials, respectively.

**Figure 1.** XRD patterns of CeO2, Co3O4 and *x*Co/CeO2 prepared following the (**a**) precipitation method; and (**b**) the hydrothermal method. (*x*: 2, 5 or 15 wt% Co loading).

#### 2.1.2. Morphological Characterization

The HR-TEM (High Resolution-Transmission Electron Microscopy) images further confirm the exposed planes of the two 15 wt.%. Co catalysts, supported on differently synthesized CeO2 carriers. In both cases, two different face-centered cubic phases are identified, on taken electron diffraction patterns: a face-centered cubic phase of CeO2 with a = 0.541 nm, and a second face-centered cubic phase of Co3O4, with a = 0.808 nm. However, the morphology of the Co3O4/CeO2 catalysts is further explored with HR-TEM. In detail, low magnification images in Figure 2 (Figure 2a,b correspond to CeO2-P and CeO2-H, respectively) show an overview of the two catalytic samples. A very well defined, crystalline hexagonal formation (nanocrystalline structures of CeO2 are forming plates) is observed in the first case, while in the second case the CeO2 nanocrystals show a nanorod structure, an expected morphology related with the hydrothermal synthesis applied during ceria preparation [26]. Moreover, in both HR-TEM micrographs showing a very well defined hexagonal formation of CeO2-P (Figure 2c), and a nanorod structure of CeO2-H (Figure 2d) respectively, the characteristic d-spacings of the (111) planes are evidenced for both CeO2 supports, in agreement with the XRD patterns (characteristic peak, predominately located at 2*θ*: 28.6◦) [7].

As shown in Figure 3, both CeO2 nanocrystals and Co3O4 nanoparticles are well crystallized, as evidenced by the clear lattice fringes in these HR-TEM micrographs. The coexistence of the Co3O4 and CeO2 cubic phases is evidenced by fast Fourier transformation analysis (FFT). However, Co3O4 nanoparticles are supported on bigger CeO2 nanocrystals with a size of about 25–30 nm (slightly larger than estimated via XRD patterns) over the Co/CeO2-P sample (Figure 3a). The experimental lattice spacings of 0.285 nm and 0.243 nm, as well as the lattice spacings of 0.312 nm, can be assigned to (220) planes and (311) planes of the Co3O4 cubic phase, and to the (111) planes of the CeO2 respectively. On the contrary, HR-TEM image of the Co/CeO2-H sample (Figure 3b) shows a much smaller CeO2 nanorod (<20 nm), a trend in agreement with the XRD characterization. A smaller Co3O4 nanoparticle with a size of about 15 nm, seems to be attached to these nanorod support. In this case, the experimental lattice spacings of 0.285 nm and 0.202 nm, as well as the lattice spacings of 0.312 nm, can be defined to (220) planes and (400) planes of the Co3O4 cubic phase, and to the (111) planes of the CeO2, respectively [7,9].

**Figure 2.** Low magnification images (overview) of the CeO2 specimens, synthesized using the precipitation, CeO2-P (**a**) or the hydrothermal CeO2-H (**b**) method. The corresponding HR-TEM micrographs of the CeO2-P sample (**c**), and the CeO2-H sample (**d**), respectively.

**Figure 3.** HR-TEM micrographs of the CeO2-P (**a**) and CeO2-H (**b**) specimens. As insets, FFT images taken at the white marked area are present for the two different cases, CeO2-P (**a**) and CeO2-H (**b**) samples, respectively.

#### 2.1.3. H2-TPR and O2-TPD Studies

The reducibility of the catalysts was also investigated with H2-TPR experiments and the corresponding reduction profiles are presented in Figure 4. Bare CeO2-P and CeO2-H (Figure 4a,b insets, respectively) exhibit two peaks; one at 500–515 ◦C and one at 780–790 ◦C. The former is attributed to the reduction of surface oxygen of ceria, while the second one to the reduction of the bulk oxygen of CeO2 [29,30]. Regarding pure Co oxides synthesized by different methods (Co3O4-P and Co3O4-H in Figure 4a,b insets, respectively), a broad double peak is observed in both cases, which is associated with the reduction of Co3O4 to CoO (T: 330–340 ◦C) and CoO to Co<sup>0</sup> (T: 400–460 ◦C) [30,31]. It is worth to mention that for both materials prepared with hydrothermal method (Co3O4-H and CeO2-H) a blue shift is observed; the reduction peaks are shifted to lower temperatures, which indicates enhanced reducibility of both oxide materials under study.

Concerning the H2-TPR profiles of *x*Co/CeO2 catalysts, all samples present three reduction peaks, which are also shifted to lower temperatures, in comparison with the corresponding bare CeO2 and Co3O4 oxides (Figure 4a,b), an observation assigned to the interaction of Co with ceria. The first two peaks are related to the reduction of Co3O4, while the third one is attributed to the reduction of bulk oxygen of ceria support. In addition, the corresponding H2 consumption tends to increase as expected, as Co loading increases from 2 to 5, and finally to 15 wt.%.

**Figure 4.** TPR profiles of bare CeO2, Co3O4 (insets) and *x*Co/CeO2 catalysts prepared with: (**a**) Precipitation and (**b**) Hydrothermal Method. (*x*: 2, 5m or 15 wt% Co loading).

As suggested in several studies, during methane combustion the lattice oxygen in the oxide surface contributes to the oxidation of CH*x*, thus leading to the formation of an oxygen vacancy, available for oxygen adsorption from the gas-phase. Both the adsorbed oxygen, as well as the surface lattice oxygen, are considered crucial for oxidation reactions, as they contribute to the adsorption/activation of oxygen and, thus, the desired oxidation of intermediates on the catalyst surface [11,32,33]. Oxygen-temperature programmed desorption (O2-TPD) experiments have evidenced that oxygen molecules were released from Co3O4 nanomaterials, suggesting that these oxygen molecules were adsorbed on the surface oxygen vacancies of the oxides [32]. These oxygen vacancies facilitate oxygen mobility of the oxide, forming the active oxygen species and, thus, enhancing the reaction activity with reductive molecules like H2 and CH4 [8,34]. Tuning the shape of the oxide catalyst, e.g. nanorods, nanocubes and nanosheets, is suggested to improve availability and abundance of these active oxygen species [26,35]. Especially, synthesis of Co3O4 catalysts of specific shape in order to provide more active oxygen species in the catalyst surface is reported in the recent literature [11].

Figure 5 shows the profiles of O2 temperature-programmed desorption (O2-TPD) of bare CeO2 (prepared either following the precipitation or the hydrothermal method) and the corresponding 15 wt.% Co/CeO2 supported catalysts. The bare ceria supports exhibit one peak at 418 ◦C, related with the adsorbed atomic oxygen evolved from the bulk of CeO2 support [36]. The incorporation of 15 wt.% Co to CeO2 carriers shifted the peak to lower temperatures, suggesting that O2 desorption is facilitated, especially for the 15Co/CeO2-H catalyst (observed shift from 418 to 376 ◦C) [30]. In addition, both Co-based catalysts exhibit two extra peaks at a higher temperature area (500–650 ◦C), which are related with the oxygen desorption from the surface of the Co oxides, while the peak observed at T > 700 ◦C corresponds to O2 desorption from the bulk of Co3O4 [11].

**Figure 5.** O2-TPD profiles of the two bare CeO2 supports (prepared with either the precipitation or the hydrothermal method) and the corresponding 15 wt.% Co/CeO2 supported catalysts.

#### *2.2. Catalytic Activity of Pure CeO2, Pure Co3O4 and Co3O4/CeO2 Oxides for Methane Oxidation*

The catalytic performances of all materials (pure CeO2, pure Co3O4 and Co3O4/CeO2 composite catalysts) in complete oxidation of methane were investigated using a fixed-bed reactor and are presented in Figure 6a. Below 500◦C, CH4 conversion was <10% for pure CeO2 supports, while the CeO2 nanorods (CeO2-H) exhibit a slightly improved performance, finally reaching 41% conversion at 600 ◦C (as compared to 31% conversion achieved with the CeO2-P sample, respectively). Co incorporation via deposition of Co3O4 nanoparticles on both CeO2 supports largely increased their catalytic activity for CH4 combustion, which is further enhanced as Co loading increases from 2 to 5 and, finally, 15 wt.%. However, an improved catalytic performance is always observed on the CeO2-H (ceria nanorods) supported materials. Co incorporation on CeO2 nanorods achieves 10% CH4 conversion

from 420 ◦C and finally reaches up to ~90% CH4 conversion at 600 ◦C (15Co/CeO2-H sample). It is worth to mention that on higher Co loadings (5 and 15 wt.%) a significant improvement (>15% higher conversion) is always observed, when using the CeO2 nanorods support. This is probably related with the optimized physicochemical properties of this CeO2-H support: smaller CeO2 cystallites, combined with higher surface area and possibly some synergetic effects between the deposited Co3O4 and the CeO2 carrier. These catalysts exhibit similar performance (T50% = 520 ◦C) with materials reported in the recent literature [7], evaluated however at milder conditions (lower GHSV:18,000 h−<sup>1</sup> and different feedstock ~6.67% CH4 and ~33.3% O2). Efficiency at high GHSV feeds is very important and still remains a significant challenge, as reported in the literature [10].

**Figure 6.** Catalytic performance in complete CH4 oxidation on pure CeO2, Co3O4 and nanocomposite Co/CeO2 catalysts. (**a**) CH4 conversion vs. temperature Flow rate: 900 cm3/min; GHSV~40,000 h−1; (**b**) Effect of WHSV on CH4 conversion for the optimum selected 15Co/CeO2-H catalyst. Feed: 0.5% vol. CH4 and 10% vol. O2, balanced with He.

In order to further exhibit superiority of our optimum material, 15Co/CeO2-H was further evaluated differentiating the reaction conditions used; varying GHSV between 10,000, 20,000, and 40,000 h−<sup>1</sup> (Figure 6b). As expected, 15Co/CeO2-H catalyst exhibits a significantly improved performance at lower GHSV reaching T50% = 440 ◦C, which is obviously much better when compared with similar reported catalysts (Co3O4/CeO2 exhibiting T50% = 475 ◦C at WHSV of 9000 mL g−<sup>1</sup> h<sup>−</sup>1) [7]. Moreover, the much lower temperature for 50% conversion of CH4 (T50%) on Co3O4/CeO2-H, as compared with the bare CeO2-H support (T50% = 440 ◦C as compared with T50% = 575 ◦C, respectively, at low GHSV: 10,000 h−<sup>1</sup> conditions), once more supports the significant role of the loaded Co3O4 nanoparticles for complete oxidation of methane. Finally, the superior performance of pure Co3O4 catalysts is very interesting, as these materials reached complete (100%) oxidation of CH4 from 520 ◦C independently of their synthesis method and thus deserve further investigation.

#### *2.3. Catalyst Stability Tests*

Thermal stability of a catalyst used in the CH4 catalytic combustion is very important because of its expected operation at high temperatures. In general, thermal aging may cause crystal growth and loss of catalytic surface. Thus, catalyst stability over time of the optimum selected material (15Co/CeO2-H) was tested under the flow of mixture of 0.5 vol% CH4 and 10 vol% O2, balanced with He. As shown in Figure 7, the conversion at 600 ◦C is rather stable (only a small decrease from 88 to 81% is observed) after 10 hours on stream, under a GHSV of 40,000 h<sup>−</sup>1. This is probably attributed to the thermal stability of the particular oxides (CeO2 and Co3O4), which are expected to deteriorate at higher operating windows (700–750 ◦C). Preservation of finely dispersed Co3O4/CeO2 species after aging at 750 ◦C is reported to retain high methane oxidation activity and tolerance to thermal ageing, suggesting once more the importance of the preparation method to achieve intimate contact between the cobalt–ceria species leading to good oxidation activity [17].

**Figure 7.** Response of CH4 combustion to long-term stability tests over the optimum selected 15Co/CeO2-H catalyst; Flow rate: 900 cm3/min; GHSV~40,000 h−1.; Reaction temperature 600 ◦C; Feed:. 0.5% vol. CH4 and 10% vol. O2, balanced with He.

The presence of ~10 vol.% of water in the NG exhausts still remains one of the main challenges for the required catalytic system to control the emissions of NG vehicles. Thus, in order to further test the water vapor effect on the catalytic performance of our optimum catalyst (15Co/CeO2-H) for methane oxidation, 5% and 10% water vapor was added in the process of CH4 combustion and the corresponding catalytic performance is shown in Figure 8 in comparison with the case using the dry reaction gas. The amount of water added significantly exceeds the stoichiometric amount produced during combustion, under the conditions used in this work (0.5% CH4, 10% O2). Increasing the water

vapor content in the reaction stream, slightly descends its catalytic performance, as T50% is increased from 520 to 530 ◦C, when water vapor is added, while increase of water content (from 5 to 10 wt.%) does not induce any further inhibition effect. Thus, water presence does not lead to a significant activity fall (observed increase in T50% is less than 20 ◦C), while water effect is even more limited at higher temperatures of methane combustion. On the contrary, effect of water presence is most evident on the initiation temperature of the CH4 combustion, increasing T10 value from 420 to 450 ◦C [14]. Removing water from the reaction feed completely restored the initial activity of the catalyst, without the need to subject the catalyst to any regeneration/thermal treatment. The restored catalyst efficiency of the used catalyst suggests that the limited water inhibition effect is probably related with adsorption of water species on the active sites at the low temperature range. There is no evidence of hydroxyl species' build-up or sintering of Co species, which could lead to a more severe/permanent deactivation, in spite the high concentration of oxygen (10 wt%) and high concentration of water vapor (10 wt%) used, expected to facilitate formation of surface hydroxyls [22]. The steam reforming reaction of methane might be involved, during the CH4 catalytic combustion in the presence of water vapor, in which CH4 is partially oxidized to CO and H2. However, analysis of the effluent gas hardly detected any CO, confirming that CH4 is totally oxidized to CO2 and H2O over 15Co/CeO2-H, even in the presence of water vapor (wet CH4 catalytic combustion).

**Figure 8.** CH4 catalytic combustion over 15Co/CeO2-H (fresh and used) with different contents of water vapor addition (0, 5 and 10 wt%). Flow rate: 900 cm3/min; GHSV~40,000 h−1.

#### **3. Materials and Methods**

The CeO2 carrier was prepared with two different methods; the precipitation method (CeO2-P) and the hydrothermal method (CeO2-H). Cerium nitrate hexahydrate (Ce(NO3)3·6H2O, 99% purity, supplied by Fluka, Steinheim, Germany) was used as a precursor salt for both synthesis methods. For the final production of 25 g of CeO2-P we initially prepared an aqueous solution (0.5 M) of cerium nitrate dissolving ~63 g of Ce(NO3)3·6H2O in 290 mL of double distilled water. Then a precipitating agent (25 vol% NH3 solution) was added at room temperature to the continuously stirred solution of the cerium nitrate, until the pH of the solution reached the value of 10. Then the pH remained stable at this value for additional 3 h. Following precipitation, the resulting precipitate was filtered, dried overnight at 110 ◦C and then calcined at 500 ◦C for 5 h under air flow. For the preparation of CeO2-H, 0.023 mol of cerium nitrate was dissolved in 46 mL of double distilled water. Then, 125 mL of a 1 M NaOH solution was added rapidly under vigorous stirring. The mixed solution was placed in a 1 L Teflon bottle, which was sealed and placed in an oven at 110 ◦C for 5 h. The as-obtained material was washed several times with double distilled water and ethanol, and then dried and calcined, following the same thermal treatment with the precipitated one.

The incorporation of cobalt over the CeO2 supports was conducted by applying the incipient wetness impregnation method. A series of Co catalysts supported on CeO2-P and CeO2-H, were prepared in three different metal loadings (2, 5 and 15 wt%). Incorporation of the Co active phase was accomplished by using Co(NO3)2·6H2O (99% purity, supplied by Merck, Darmstadt, Germany). Aqueous solution of the metal salt was impregnated on CeO2 supports, in successive stages, with intermediate drying in an oven at 100 ◦C for 30 min. The derived samples were dried at 110 ◦C overnight and finally calcined under air flow at 500 ◦C for 5 h. The as prepared composites are herein labeled as *x*Co/CeO2-P or H, where *x* is the cobalt loading.

For comparison reasons, bare Co3O4 was prepared by precipitation (Co3O4-P) and hydrothermal (Co3O4-H) method. Co(NO3)2·6H2O was used as a precursor salt and the same precipitation method was followed as for the synthesis of CeO2-P. For the preparation of Co3O4-H, 0.0515 mol of cobalt nitrate was dissolved in 51 mL of double distilled water and 125 mL of a 1 M NaOH solution was added rapidly under vigorous stirring. Then, the next steps (aging, washing, and thermal treatment) were the same with the hydrothermal synthesis of ceria.

The physicochemical characteristics of as-synthesized materials were evaluated by various complimentary techniques. The total Co loading (wt%) of the final catalysts was determined by the Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES -on a Perkin-Elmer Optima 4300DV apparatus, Waltham, MA, USA). The textural characteristics of the as prepared catalysts were determined by the N2 adsorption–desorption isotherms at –196 ◦C (Nova 2200e Quantachrome flow apparatus, Boynton Beach, FL, USA). Specific surface areas (m2g−1) were obtained according to the Brunauer-Emmett-Teller (BET) method at relative pressures in the 0.05–0.30 range. The specific pore volume (cm3g−1) was calculated based on the highest relative pressure, whereas the average pore size diameter (dp, nm) was determined by the Barrett-Joyner-Halenda (BJH) method. Prior to measurements the samples were degassed at 250 ◦C under vacuum. The crystalline structure of the catalysts was determined by powder X-ray diffraction (XRD) on a Siemens D 500 diffractometer (Bruker, Karlsruhe, Germany) operated at 40 kV and 30 mA with Cu Kα radiation (λ = 0.154 nm). Diffractograms were recorded in the 5–80◦ 2θ range and at a scanning rate of 0.02◦ s−1. The Scherrer equation was employed to determine the primary particle size of a given crystal phase based on the most intense diffraction peaks of CeO2 (28.6◦, 47.5◦) and Co3O4 (31.3◦, 36.9◦). The size, morphology and structure of the two 15 wt% Co-supported catalysts (either on CeO2-H or CeO2-P supports) were also investigated by transmission electron microscopy using a (JEOL JEM 2011 TEM, Zaventem, Belgium) operating at 200 kV with an atomic resolution of 0.194nm. The HR-TEM micrographs were analyzed using Digital Micrograph v2 software (Gatan, Inc., München, Germany).

Hydrogen temperature-programmed reduction (H2-TPR) experiments were performed in a different bench-scale unit, also available in our facilities for TPX studies, loading a fixed bed tubular reactor with 0.1 g of sample, while the reactor exit was coupled with a mass spectrometer (MS). A flow (50 cm3/min) of 5 vol% H2/He gas mixtures was used, while the temperature of the solid catalyst was increased to 800◦C at the rate of 10 ◦C/min. The effluent gas from the reactor was analysed using the MS detector (m/z = 2 was used for H2). Prior to reduction, the as prepared sample was loaded to the reactor and pre-treated at 300◦C for 1h under He flow. The catalyst sample was then cooled in He gas flow to room temperature, before recording the H2-TPR trace. Oxygen temperature-programmed desorption was performed (O2-TPD) in the same unit. The temperature was increased to 400 ◦C (heating rate 5 ◦C/min) under 20 vol% O2/N2 (50 cm3/min), for 30 min and were cooled to room temperature to adsorb oxygen for 30 min. After that, the samples were purged under He flow for 1 h, in order to purge physically adsorbed oxygen. The desorption step was conducted under He flow (50 cm3/min) up to 750 ◦C at a rate of 10 ◦C/min and the signal of O2 (m/z = 32) were detected by the mass spectrometer.

The catalytic combustion of CH4 was performed in a fixed-bed reactor loaded with 0.6 g of catalyst. The total gas flow rate was 900cm3/min, corresponding to a gas hourly space velocity (GHSV) of ~40,000 h−1. The composition of the feed was 0.5 vol% CH4 and 10 vol% O2, balanced with He. The effect of GHSV was also investigated at 40,000, 20,000, and 10,000 h−1. The CH4 conversion was monitored in the 300–600 ◦C range (in a decreasing temperature mode) and was held constant at each temperature for ~20 min prior taking measurements. The composition of the effluent gas from the reactor was analysed using gas chromatography (GC-FID, Agilent 7890, Walborn, Germany). Stability of optimum catalyst's performance (time-on-stream) was tested at 600 ◦C for 10 h (under 0.5 vol% CH4, 10 vol% O2 in He flow). In addition, the effect of H2O content (0, 5, and 10% vol. H2O in He) was studied on CH4 oxidation efficiency.

#### **4. Conclusions**

Various novel Co-Ce catalysts were prepared following different preparation techniques for the synthesis of ceria support (precipitation or hydrothermal method) and incorporating different loadings of Co on either support. The synthesis route results in ceria carriers of different physicochemical properties, further modified after Co incorporation. Hydrothermal synthesis leads to an improved CeO2 support, with smaller crystallite size, larger surface area and enhanced reducibility. The ceria support and, thus, the synthetic procedure is important in terms of dispersing the active Co component (smaller Co3O4 crystallites formed on the CeO2-H support) and exhibiting higher oxygen mobility (O2 desorption peak of larger area, additionally appearing at lower temperature), as deduced by comparing the 15Co/CeO2-H and 15Co/CeO2-P catalytic samples. In summary, a nanocomposite catalyst Co3O4/CeO2-H exhibits higher activity in complete oxidation of CH4, than Co3O4/CeO2-P and pure CeO2 materials. The higher dispersion of the deposited Co species and the enhanced reducibility of Co-Ce catalysts advocate synergistic effects of CeO2 nanorods and the supported Co3O4 nanoparticles. Moreover, the addition of water vapor in the reaction stream only slightly inhibited the catalytic activity of the optimum selected 15Co/CeO2-H catalyst, a fact mainly observed at lower temperature ranges. Thus, the corresponding material is very promising for methane oxidation purposes, as it additionally presented good water resistance properties in wet CH4 combustion.

**Author Contributions:** The experimental work was conceptualized and designed by E.P. and E.I.; S.D. and E.P. synthesized all catalytic materials, while S.D. helped in performing the evaluation tests and E.P. helped in characterization studies; A.L. provided all resources, while E.P. and E.I. analyzed all experimental results and wrote the original draft. The manuscript was amended and supplemented by all authors. All authors have given approval for the final version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors would like to thank Ioannis Tsiaousis for performing the HR-TEM characterization. **Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Electrochemical Promotion of Nanostructured Palladium Catalyst for Complete Methane Oxidation**

#### **Yasmine M. Hajar, Balaji Venkatesh and Elena A. Baranova \***

Department of Chemical and Biological Engineering, Centre for Catalysis Research and Innovation (CCRI), University of Ottawa, 161 Louis-Pasteur, Ottawa, ON K1N 6N5, Canada; yhaja059@uottawa.ca (Y.M.H.); bvenk044@uottawa.ca (B.V.)

**\*** Correspondence: elena.baranova@uottawa.ca; Tel.: +1-6135625800 (ext. 6302); Fax: +1-6135625172

Received: 4 December 2018; Accepted: 4 January 2019; Published: 6 January 2019

**Abstract:** Electrochemical promotion of catalysis (EPOC) was investigated for methane complete oxidation over palladium nano-structured catalysts deposited on yttria-stabilized zirconia (YSZ) solid electrolyte. The catalytic rate was evaluated at different temperatures (400, 425 and 450 ◦C), reactant ratios and polarization values. The electrophobic behavior of the catalyst, i.e., reaction rate increase upon anodic polarization was observed for all temperatures and gas compositions with an apparent Faradaic efficiency as high as 3000 (a current application as low as 1 μA) and maximum rate enhancement ratio up to 2.7. Temperature increase resulted in higher enhancement ratios under closed-circuit conditions. Electrochemical promotion experiments showed persistent behavior, where the catalyst remained in the promoted state upon current or potential interruption for a long period of time. An increase in the polarization time resulted in a longer-lasting persistent promotion (p-EPOC) and required more time for the reaction rate to reach its initial open-circuit value. This was attributed to continuous promotion by the stored oxygen in palladium oxide, which was formed during the anodic polarization in agreement with p-EPOC mechanism reported earlier.

**Keywords:** electrochemical promotion; NEMCA; palladium; ionic promoter; nanoparticles; yttria-stabilized zirconia

#### **1. Introduction**

Natural Gas Vehicles (NGVs) have gained considerable attention in the last decade due to much lower greenhouse gas emissions and lower price of methane compared to diesel or gasoline. Not only CH4 is abundant in natural gas form, but methane can also be produced using anaerobic digestion technologies of bio-derived sources [1–4]. Despite lower emissions of NGVs they often suffer from incomplete methane combustion. Because methane is also 23 times more potent in warming the atmosphere than carbon dioxide its complete conversion to CO2 is paramount. Therefore, the development of efficient low temperature catalysts for deep oxidation of methane (CH4) has recently attracted significant attention [5–10].

Palladium-based catalysts are considered the most efficient for methane activation in excess of oxygen and their activity depends on temperature, methane/oxygen ratio, and catalyst surface oxidation state and composition [5]. The nature of the active surface sites PdOx vs. Pd was a subject of several studies [6,7]. It was shown that chemisorbed oxygen on Pd metal is poorly active, whereas Pd oxidation with an optimum of 3–4 monolayers forms an active PdO catalyst. Chemisorption of a first layer of oxygen is fast; however, partial bulk oxidation is relatively slow [7].

The innovative field of catalysis that could boost complete methane oxidation reaction over Pd is electrochemical promotion of catalysis (EPOC) or non-Faradaic electrochemical modification of catalytic activity (NEMCA). This general, well-established phenomenon in catalysis aims at controlling in-situ both the activity and the selectivity of a catalyst through application of electric stimuli [8–11]. EPOC is observed with solid electrolyte materials that serve as catalyst support. Ions contained in these electrolytes (O2−, H+, Na+ OH−, etc.) are electrochemically pumped to the catalyst surface, where they act as promoting species leading to modification of catalyst electronic properties and as a result its catalytic activity and selectivity. More precisely, applying an anodic polarization results in the strengthening of electron-donor adsorbates, e.g., chemisorbed methane, and weakening of the binding strength of electron-accepting adsorbates, e.g., dissociatively chemisorbed oxygen [12]. The resulting electrochemical activation magnitude is much higher than that predicted by Faraday's law. [13,14]. The increase in the catalytic rate, Δ*r* (mol O/s) divided by the electrochemical rate, *I*/*nF* (*I* is current, F is Faraday's constant and *n* is number of electrons, 2 for O2−) is denoted as the apparent Faradaic efficiency, *Λ*, and the process is considered non-Faradaic when |*Λ*| is greater than 1. Another parameter used to quantify EPOC is the rate enhancement ratio, *ρ*, which is the ratio between the promoted closed-circuit catalytic rate, *r* and the unpromoted open-circuit catalytic rate, *r*o.

The electrochemical promotion of complete methane oxidation was investigated on palladium catalysts prepared using various methods as summarized in Table 1 [15–23]. Electrochemical promotion of Pd thick film catalyst electrode prepared using wet impregnation was investigated in the temperature range of 470–600 ◦C [18]. The rate enhancement of 40% and an apparent Faradaic efficiency of 1.85 were found in this work. The addition of CeO2 layer between the YSZ solid-electrolyte and Pd film catalyst increased the open-circuit catalytic rate but decreased the apparent Faradaic efficiency due to the higher electric resistance [18]. Another study on a Pd film catalyst prepared using commercial organometallic paste showed higher enhancement ratio (*ρ* = 5.6) and Faradaic efficiency (*Λ* = 579) at 560 ◦C; however, instability of the catalyst over time and rapid deactivation within 900 min of experiment was also noted [20]. Furthermore, the effect of CeO2 layer was studied in [24]. It was shown that presence of ceria increased catalytic activity of Pd due to the formation of an active PdO phase, which was stabilized by CeO2 acting as a continuous source of oxygen, similarly to the oxygen migration from the YSZ electrolyte under EPOC. In another study, an addition of porous YSZ layer between Pd film-catalyst and the dense YSZ solid-electrolyte resulted in a high open-circuit catalytic rate. The authors reported an enhancement ratio of 1.2 and apparent Faradaic efficiency, *Λ*, of 17 under fuel-rich conditions [23]. EPOC of sputtered Pd catalyst-electrode was compared to impregnated Pd for methane complete oxidation. The sputtered catalysts showed slightly higher *Λ* of 12 but similar enhancement ratio (*ρ* = 1.6) [21]. The effect of metal loading and catalyst thickness on EPOC of Pd was studies on Pd catalyst prepared by physical vapor deposition (PVD). It was found that metal loading and catalyst thickness have significant effect on the open-circuit rate and electrochemical promotion, where the thinner films resulted in the highest reaction rates per gram of catalyst at 500 ◦C [17]. A scaling-up of the system was attempted by electroless deposition of Pd in the channels of a YSZ monolith honeycomb; however, decrease in the conversion of methane occurred under positive and negative polarization [19].

Therefore, from previous EPOC studies on Pd for complete CH4 oxidation, it is clear that the catalyst preparation method has a strong influence on Pd morphology, structure, oxidation state and as a result, on its catalytic activity, degree of promotion and stability under open and closed circuit conditions. Furthermore, from the practical point of view it is essential to work with low loadings of noble-metal catalysts that exhibit high dispersion and large active surface area In the present work, we report electrochemical promotion of nano-structured, highly dispersed Pd catalyst prepared by polyol reduction method for CH4 complete oxidation in the temperature range of 400 and 450 ◦C and various gas compositions.



a

Continuous increase in closed-circuit

 rate post EPOC;

Electrolysis effect.

#### **2. Results and Discussion**

Transmission electron microscopy (TEM) was used to determine the palladium morphology and particle size (Figure 1a,b). The resulting Pd particles are spherical in shape with a diameter of approximately 5 nm, that are coalesced together in larger aggregates of roughly 50 nm in size. Figure 1c,d shows SEM images of as-prepared Pd/YSZ catalysts and the same catalyst after catalytic measurements under open circuit and EPOC conditions. It can be seen that as-prepared catalyst-electrode forms a highly dispersed, non-continuous layer on YSZ surface that consists of fine grains and pores. After the reaction, the "spent" catalyst shows much larger catalyst islands indicating a change in the morphology due to the catalyst agglomeration that takes place during the reaction. The resulting energy dispersive X-ray spectroscopy (EDS) spectrum (Figure 2a) of as-prepared Pd shows that the only element present is Pd. The x-ray diffraction pattern (XRD) contains several diffraction peaks of Pd (111), (200) and (220) corresponding to face-centered cubic (fcc) structure typical for bulk palladium metal. The crystallite size of Pd found from Pd (111) was 8 nm.

**Figure 1.** (**a**,**b**) TEM images of stand-alone Pd NPs at 5 nm and 50 nm scale; (**c**,**d**) SEM images of as-prepared (**c**) and post-experiment (**d**) Pd catalyst deposited on YSZ solid electrolyte.

**Figure 2.** (**a**) EDS spectrum and (**b**) XRD pattern of Pd catalyst.

The open-circuit catalytic rate of methane oxidation was tested at two different temperatures and at various gas compositions (Figure 3). It can be seen that for both temperatures the rate increased as a function of oxygen partial pressure. The increase was more significant at 4 kPa of methane where oxygen-to-methane partial pressures ratios were lower. The rate increase in fuel-rich condition is indicative of a Langmuir-Hinshelwood mechanism [15] where methane is able to competitively adsorb on palladium as seen at 450 ◦C under 4 kPa of CH4, while the quasi-stable value at higher ratio of oxygen (at 2 kPa of CH4) can be explained by an Eley-Rideal mechanism as CH4 reacts on the oxygen covered surface [25].

**Figure 3.** Effect of oxygen-to-methane ratio on catalytic rate at 350 and 450 ◦C.

Figure 4 shows the transient rate response to an application of positive current (20 μA) between the Pd working electrode and the counter electrode at 450 ◦C. Under open-circuit conditions (t < 0.5 h), the catalytic rate was at 8.35 × <sup>10</sup>−<sup>8</sup> mol/s. When a constant current was imposed, the reaction rate gradually increased due to pumping of O2<sup>−</sup> promoters to the catalyst surface. After 2 h of polarization, the closed-circuit rate increase was 180% higher than its corresponding open-circuit value. In addition, the non-Faradaic behavior resulted in an apparent Faradaic efficiency, *Λ*, of 610 denoting that the back-spillover of O2<sup>−</sup> at a I/2F rate gave a 610 times increase in the overall oxidation rate [26]. It should be noted that slight increase in the catalyst-working electrode potential (UWR) was observed upon positive polarization (from 0.39 to 0.42 V). Furthermore, the slow reaction rate increase did not reach a steady-state value even after 2 h of applied polarization and took over 1 h to reach the open circuit value observed before polarization. According to the mechanism of EPOC the reaction rate is due to the supply of oxygen ionic species from YSZ and the formation of an effective double

layer at the surface of the catalyst that changes the work function of Pd leading to weakening of the chemisorbed oxygen bond strength thus facilitating the –O2C desorption from the catalyst surface.

A similar behavior was observed under potentiostatic conditions (UWR = 0.25 V) and a stoichiometric flow of reactants at 425 ◦C (Figure 5). The catalytic rate slowly increased in value until it reached 7.0 × <sup>10</sup>−<sup>8</sup> mol s−<sup>1</sup> after 2 h. As in the galvanostatic conditions (Figure 4) the closed-circuit reaction rate was continuously increasing with time without reaching a steady-state. After 2 h, the rate enhancement ratio of 1.31 and, an apparent Faradaic efficiency of 1107 were obtained. The high *Λ* value is due to the low current that passed through the cell, which was sufficient to promote Pd catalyst.

**Figure 4.** Transient rate response of Pd nanoparticles to a current step change. o.c.: open-circuit. Conditions: T = 450 ◦C, 2 kPa of CH4 and 4 kPa of O2. Flow rate: 100 ccm.

The continuous increase of the catalytic rate in Figures 4 and 5 can be explained by the continuous oxidation of palladium catalyst to PdOx, which makes the catalyst more active for methane oxidation. In Figure 5, this continuous increase in catalytic rate occurred simultaneously with a decrease in current at a constant applied potential value indicating Pd oxide formation. Upon current or potential interruption the rate slowly returned to its initial state, because oxygen species stored in PdOx continued acting as sacrificial promoters for methane oxidation reaction.

**Figure 5.** Transient rate response of Pd nanoparticles to potential step changes. o.c.: open-circuit. Conditions: T = 425 ◦C, 2 kPa of CH4 and 4 kPa of O2. Flow rate: 100 ccm.

To confirm PdOx formation and oxygen storage effect on p-EPOC, the catalyst was polarized for a different duration 3, 6 and 10 h. As seen in Figure 6, the closed-circuit catalytic rate was continuously increasing with time under constant current (40 μA for 3 h) and potential (0.5 V for 6 and 10 h) application. It can be seen that even after ten hours of polarization, the closed-circuit rate kept on rising without reaching a steady-state. At the same time, the longer polarization time resulted in longer decrease of the open-circuit rate after polarization was stopped. A proportional relationship (as shown in the inset Figure 6) was found between the duration of EPOC and the time to reach the open-circuit rate ro.c. The slope of this relationship was 0.5. This persistent electrochemical promotion (p-EPOC) is due to the stored oxygen ions in PdOx that act as sacrificial promoters when the electrical circuit is open [27].

**Figure 6.** *Cont*.

**Figure 6.** (**a**) Transient rate response of Pd at different duration of EPOC and (**b**) the potential/current read at potentiostatic or galvanostatic application. Conditions: T = 425 ◦C, 2 kPa of CH4 and 4 kPa of O2. Flow rate: 100 ccm.

This indicates that during the positive polarization two parallel processes take place: i. Migration of Oδ<sup>−</sup> promoters to the gas exposed catalyst surface and ii PdOx formation at the three-phase boundary (tpb) according to the electrochemical reaction:

$$\text{Pd} + \text{xO}^{2-} \rightarrow \text{PdO}\_x + \text{xe}^- \tag{1}$$

Current decrease and potential increase upon positive potentiostatic (Figures 5 and 6) and glavanostatic polarization (Figure 4), respectively, confirms the formation of PdOx. Palladium oxide has lower conductivity than Pd metal, therefore current that flows through the solid-state cell or potential difference of the working catalyst-electrode (UWR) are the clear indication of an electrochemical oxide formation [28].

Figure 7 shows a transient rate response at a constant applied potential 0.5 V for 24 h. The reaction rate continuously increased for up to 20 h followed by 2 h of a steady-state rate and then a slight rate decrease. The continuous rate increase indicates constant catalyst activation due to the growth of PdO*x*, whereas somewhat rate decrease after 22 h of polarization at 425 ◦C may be linked with morphology change observed for the "spent" catalysts (Figure 1d). The open-circuit rate took more than 6 h to return to its initial value showing a persistent promotional effect.

**Figure 7.** Long period transient rate response of Pd to a potentiostatic step change. o.c.: open-circuit. Conditions: T = 425 ◦C, 2 kPa of CH4 and 4 kPa of O2. Flow rate: 100 ccm.

Figure 8 shows the catalytic rate and catalyst potential change upon application of constant current as low as 1 μA. This resulted in the continuous catalytic rate increase for 1, 2 and 4 kPa of O2, accompanied by catalyst potential (UWR) increase, confirming palladium oxide formation. The corresponding Λ values are summarized in Figure 9.

**Figure 8.** (**a**)Transient rate response of Pd catalyst at different O2 partial pressure under 1 μA galvanostatic application and (**b**) the corresponding catalyst potential response. Conditions: T = 450 ◦C, 2 kPa of CH4. Flow rate: 100 ccm.

In Figure 9, the effect of partial pressure on closed-circuit reaction rate was tested at different galvanostatic conditions. The highest increase in catalytic rate was found at slightly fuel-rich conditions resulting in a *ρ* value of 1.3; the higher rate can be explained by the advantaged adsorption of gaseous methane over oxygen. At this condition, gaseous methane can be expected to directly adsorb onto Pd, resulting in a competition between oxygen and methane adsorption following a Langmuir-Hinshelwood mechanism. In addition, the desorption of oxygen from the surface becomes facilitated at lower oxygen partial pressure in the atmosphere as the overall chemical potential of oxygen is reduced [29]. At a partial pressure ratio higher than the stoichiometric ratio, it is perceived that the catalytic rate increase is slightly lower. The slight decrease in the enhancement is due to the

competing adsorption of oxygen on the surface of Pd, putting slight mass-transfer limitations on the chemisorption of CH4.

**Figure 9.** Current effect on catalytic rate in function of methane/oxygen ratio. T = 450 ◦C. Flow rate: 100 ccm.

Similarly, the effect of temperature on closed-circuit reaction rate was tested at different galvanostatic conditions. Figure 10 shows that there is an increase in catalytic rate as a function of temperature at all applied positive current values for 400, 425 and 450 ◦C. It can be noticed that upon application of small current of 1 μA, a significant rate increase was detected, resulting in a logarithmic-shape relationship of rate increase versus applied current. In addition, a highest value of ~2400 was found for the apparent Faradaic efficiency, constructing that a very minimal current was able to result in a change of the Pd surface oxidation state and hence the adsorption strength of methane reactant [15,16,18,20].

**Figure 10.** Current effect on catalytic rate in function of temperature: 400, 425 and 450 ◦C. PCH4= 2 kPa and PO2 = 4 kPa. Flow rate: 100 ccm.

Table 1 compares EPOC of methane oxidation on Pd/YSZ found in this work to previous studies carried out on Pd catalyst-electrode deposited on YSZ solid-electrolyte. The table depicts where our results fall in comparison with previous experiments. It should be noted that the metal loading used in this work is the lowest in the temperature range of interest (T ≤ 450 ◦C), which is important for cold-start emission application. In agreement with previous work, Pd nanostructured catalyst synthesized by polyol method shows electrophobic type of EPOC, where only positive polarization promotes the reaction. Applied polarization led to the supply of Oδ<sup>−</sup> promoters from YSZ electrolyte to the catalyst surface, resulting in the formation of a more active phase of PdO, on the surface first and in the bulk gradually.

Ionic oxygen migration to the surface altered the adsorption properties of the catalyst surface, resulting in the weakening of gaseous oxygen adsorption and strengthening that of electron-donor methane. The alteration of the catalytic oxidation state was similar under both potentiostatic and galvanostatic application, which have resulted in a continuous increase in catalytic rate and a very slow open circuit rate decrease when the circuit was interrupted. This was explained by the formation of PdO during polarization and oxygen storage in its bulk, which was continuously providing the promoting oxygen species to the surface post-polarization according to p-EPOC mechanism [27].

#### **3. Experimental**

#### *3.1. Synthesis of Pd Nanoparticles*

Mono-metallic Pd NPs were synthesized using 0.133 g of palladium chloride (Fisher Scientific®, Canada) precursor salts dissolved in 25 mL of ethylene glycol and 0.8 M NaOH. The mixture was heated up to 160 ◦C and kept under stirring conditions for 3 h. The final colloidal solution was cooled down and washed repeatedly with ethanol.

#### *3.2. Catalyst Characterization*

X-ray diffraction (XRD) was performed on the fresh Pd sample using Rigaku Ultima IV multipurpose diffractometer (Rigaku, The Woodlands, TX, USA). The diffractometer was equipped with an X'Celerator detector with monochromatic CuKα radiation (λ = 1.5418 Å) at 40 kV and 44 mA with a divergence slit of 2/3 degree, a scan speed of 0.03 o/s and a scan step of 0.02 degrees between 30 and 80◦ 2θ.

The transmission electron microscopy (TEM) micrographs were obtained using JEOL JEM 2100F FETEM (JEOL, Peabody, MA, USA) operating with a field emission gun at an acceleration voltage of 200 kV. SEM micrographs were recorded using PhenomTM SEM (Nanoscience Instruments, Virginia, USA). Additional elemental analysis was performed using the energy dispersive X-ray spectroscopy (EDS) attachment.

#### *3.3. Electrochemical Cell and Reactor*

The solid electrolyte is a 19 mm diameter and 1 mm thickness disk of 8 mol % Y2O3-stabilized ZrO2 (YSZ) (TOSOH®, Grove city, OH, USA) fabricated following the procedure reported earlier [30]. Inert gold reference and counter electrodes were deposited on one side of the disk by applying thin gold paste coating (Gwent Group, Pontypool, UK) of 0.2 and 1 cm2 surface areas, respectively. This was followed by annealing in air at 500 ◦C. The catalyst-working electrode was deposited on the other side of the solid electrolyte disk (1 cm<sup>2</sup> surface area) opposing to the counter electrode. To this end, mono-metallic Pd were dispersed in isopropanol and 10 μL of a suspension were deposited at a time with intermediate drying at room temperature. The resulting total metal loading was 0.5 mg of Pd on YSZ. Catalytic measurements were carried out at atmospheric pressure in the single-chamber capsule reactor reported earlier [31,32]. The working electrode side of the electrolyte was pressed against a gold mesh (1 cm2) that served as a current collector, while the counter and reference electrodes were

pressed directly against gold wires [33]. Two type K thermocouples (Omega®, Quebec, Canada) were placed in vicinity of the electrochemical cell, one for temperature control and one for data acquisition.

#### *3.4. Catalytic and Electrochemical Measurements*

The reaction gases were CH4 (Linde, 99.99%), O2 (Linde, 99.99%), and He (Linde, 99.997%) as a carrier gas. The total flow rate was constant at 100 mL min<sup>−</sup>1. Gas composition was varied using MKS, 1259 C and 1261-C series flow meters and detected using non-dispersive infrared (NDIR) CO2 gas analyzer (Horiba, VA-3000, Burlington, Canada). Constant electric current or potential were applied using a potentiostat-galvanostat (Arbin Instruments®, MSTAT, College Station, TX, USA) connected to the electrodes of solid-electrolyte electrochemical cell.

#### **4. Conclusions**

Electrochemical promotion of Pd nanostructured catalyst was investigated for the methane oxidation reaction in the 400–450 ◦C temperature range. The promotion of the catalytic rate of Pd NPs was achieved under anodic polarization. Upon various potentiostatic and galvanostatic tests, non-Faradaic enhancement was achieved, most notably at 450 ◦C, under the application of 1 μA, where *Λ* was equal to ~3000, higher than any previous EPOC study on Pd. Continuous increase in the reaction rate was found under EPOC conditions, due to the Pd oxide formation in the vicinity of the tpb. A proportional relationship was found between the duration of polarization and the post-polarization time required to reach the initial open-circuit rate value. Post polarization, a persistent promotion (p-EPOC) was observed due to the promotion of the reaction by the stored oxygen, which was accumulated during positive polarization. Overall, our work revealed interesting behavior of Pd synthesized by polyol method, providing further insight into the application of electrochemical promotion for complete methane oxidation with highly dispersed Pd.

**Author Contributions:** Conceptualization, E.A.B.; Data curation, Y.M.H and B.V.; Formal analysis, B.V.; Funding acquisition, E.A.B.; Methodology, Y.M.H.; Project administration, E.A.B.; Supervision, E.A.B.; Validation, Y.M.H.; Writing—original draft, Y.M.H.; Writing—review & editing, E.A.B.

**Funding:** Financial support from Natural Science and Engineering Research Council (NSERC) Canada is greatly acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Synthesis of Sulfur-Resistant TiO2-CeO2 Composite and Its Catalytic Performance in the Oxidation of a Soluble Organic Fraction from Diesel Exhaust**

**Na Zhang 1, Zhengzheng Yang 1,\*, Zhi Chen 1, Yunxiang Li 1, Yunwen Liao 1,2, Youping Li 1, Maochu Gong <sup>3</sup> and Yaoqiang Chen 3,\***


Received: 13 May 2018; Accepted: 11 June 2018; Published: 14 June 2018

**Abstract:** Sulfur poisoning is one of the most important factors deteriorating the purification efficiency of diesel exhaust after-treatment system, thus improving the sulfur resistibility of catalysts is imperative. Herein, ceria oxygen storage material was introduced into a sulfur-resistant titania by a co-precipitation method, and the sulfur resistibility and catalytic activity of prepared TiO2-CeO2 composite in the oxidation of diesel soluble organic fraction (SOF) were studied. Catalytic performance testing results show that the CeO2 modification significantly improves the catalytic SOF purification efficiency of TiO2-CeO2 catalyst. SO2 uptake and energy-dispersive X-ray (EDX) results suggest that the ceria doping does not debase the excellent sulfur resistibility of bare TiO2, the prepared TiO2-CeO2 catalyst exhibits obviously better sulfur resistibility than the CeO2 and commercial CeO2-ZrO2-Al2O3. X-ray powder diffraction (XRD) and Raman spectra indicate that cerium ions can enter into the TiO2 lattice and not form complete CeO2 crystals. X-ray photoelectron spectroscopy (XPS), H2-temperature programmed reduction (H2-TPR) and oxygen storage capacity (OSC) testing results imply that the addition of CeO2 in TiO2-CeO2 catalyst can significantly enhance the surface oxygen concentration and oxygen storage capacity of TiO2-CeO2.

**Keywords:** cerium-doped titania; sulfur-tolerant materials; organic compounds purification; diesel oxidation catalyst; vehicle exhaust

#### **1. Introduction**

Diesel engines have been widely used in passenger cars and vans, due to excellent fuel efficiency and durability. However, diesel exhaust gases, such as carbon monoxide (CO), unburned hydrocarbons (HCx), nitrogen oxides (NOx), particulate matter (PM) and soluble organic fraction (SOF), are considered major sources of air pollutants [1–3]. Among these hazardous pollutants, SOF are the heavy liquid hydrocarbons (C > 16 [4,5], aromatics and oxygenated compounds [6]) adsorbed on soot, which mainly come from unburned fuel and lube oils [7,8]. The content of SOF is known to vary with engine operating conditions and can reach about 5–60% of the whole mass of the particulate matter [6,9–11]. Due to the fact that diesel SOF contains types of polycyclic aromatic hydrocarbons (PAHs) [12,13] which are recognized as strong carcinogens [14,15], purifying the SOF from diesel exhaust is an important and essential work.

Diesel oxidation catalyst (DOC) was employed to accelerate the oxidation and purification of diesel exhaust gases of CO, HCx and SOF. In recent decades, CeO2-ZrO2, Al2O3 and their mixed oxide-based catalysts were widely used as commercial DOC and displayed excellent catalytic CO and HCx oxidation activity. Focusing on the purification of SOF, CeO2-based oxygen storage materials (OSM) have been greatly impressed by considerable researchers due to the superior catalytic activity on hydrocarbons and SOF oxidation [16–18]. Meanwhile, controlled synthesized of nanostructured CeO2-based catalysts and their catalytic performance in diesel soot oxidation are lucubrated [19,20]. However, CeO2-based catalysts are easily poisoned by SO2 [21–24]. And SO2 is a subsistent in the diesel exhaust, since sulfur is present in almost all commercial diesel fuels [25–27], sulfur in fuels would be oxidized to SO2 and then emitted from diesel engines [28,29]. Furthermore, sulfur poisoning resulting from sulfur species accumulation is more destructive, since even using ultra-low sulfur diesel (ULSD), cumulative exposure of a catalyst over its lifetime in a heavy-duty diesel may amount to kilograms of sulfur [30]. A large amount of sulfur species accumulation inevitably results in the blocking of channels of monolithic catalyst, and hence the strike of diesel exhaust after-treatment system. Thus, it is important and realistic for a diesel oxidation catalyst to improve the sulfur resistibility.

TiO2 is known as an effective sulfur-resistant material [31–33], and our previous works [34–36] also prove that TiO2-based diesel oxidation catalysts display excellent sulfur resistibility. However, TiO2 is not active enough for catalytic diesel SOF combustion. Reports about the sulfur resistance catalyst for diesel SOF oxidation are still scarce. Considering all of this, in this work, CeO2 was introduced in TiO2 by a co-precipitation method, and its effects on sulfur resistibility and catalytic activity for diesel SOF combustion were investigated.

#### **2. Results and Discussion**

#### *2.1. Sulfur Resistibility*

The sulfur resistibility values of catalysts were measured by sulfur uptake testing. As shown in Figure 1, under SO2 exposure, the weight of all samples increased with time; final weight increments tended to be flat after 1–3 h SO2 exposure, except for CeO2. After 4 h, chosen as representative exposure time of simulative 160,000 km vehicle aged catalyst, the final weight increments of TiO2, TiO2-CeO2 and CeO2 catalysts are 1.63 wt. %, 2.01 wt. % and 4.72 wt. %, respectively. The normalized sulfur uptake results are calculated by supposing 1 g of sample as standard, and the results are listed in Table 1. The normalized sulfur uptake values of TiO2, TiO2-CeO2 and CeO2 catalysts are 166 μg/m2, 170 μg/m2, and 891 μg/m2, respectively. From the results it can be seen that the sulfur species accumulation is severe on CeO2 catalyst but is slight on both TiO2 and TiO2-CeO2 catalysts, which implies that the TiO2-based catalysts (TiO2 and TiO2-CeO2) present obviously better sulfur resistibility than the CeO2 catalyst. Since the non-sulfating material of TiO2 displays low SO2 adsorption and hence relieves sulfate generation and exhibits superior sulfur resistibility [32,37]. Furthermore, the introduction of moderate amounts of ceria in TiO2 has essentially no effect on the naturally excellent sulfur resistibility of TiO2.



<sup>a</sup> Surface area was calculated by BET method from the N2 adsorption-desorption results; <sup>b</sup> EDX results were obtained by detecting the simulative 160,000 km vehicle aged samples; <sup>c</sup> The normalized sulfur uptake = sulfur uptake/(100 × surface area) [37,38].

**Figure 1.** Sulfur uptake of the TiO2, TiO2-CeO2 and CeO2 catalysts.

Additionally, the accumulation amounts of sulfur species on simulative 160,000 km vehicle aged TiO2, TiO2-CeO2 and CeO2 catalysts tested by EDX (Table 1) are 1.02 wt. %, 1.04 wt. % and 4.26 wt. %, respectively, which shows the same trend with the sulfur uptake results. It is clear that the prepared TiO2 and TiO2-CeO2 present significantly less sulfur species accumulation and better sulfur resistibility than the CeO2 catalyst under the long-term exposure of diesel exhaust ambiences.

#### *2.2. Catalytic Performance*

Figure 2 shows the thermogravimetry-differential thermal analysis-differential thermogravimetry (TG-DTA-DTG) curves of the bulk lube without catalysts and the lube impregnated on catalysts; all of the DTA curves are the positive peaks indicating exothermic peaks, and all of the originally DTG negative peaks are inverted into positive peaks for a better readability of the graph. The combustion of lube without catalyst under air flow is shown in Figure 2a; the weight loss of bulk lube is tersely distinguished into two stages; about 90% of lube is deflagrated at 220–350 ◦C, and then the rest of 10% lubricating oil is consumed tardily after 350 ◦C till 500 ◦C, which implies that the commercial lube contains a fraction of hydrocarbons hard to pyrolyze (may come from the lubricant additive). The onset combustion temperature of *T*10% (the temperature at which 10% of the initial lube is converted) is about 264 ◦C, the lube combustion fastest temperature of *Tm* (the temperature of weightlessness fastest point in DTG curves) is about 324 ◦C, and the final reaction temperature of *Tf* (the temperature of lube is completely converted) is about 507 ◦C. As shown in DTA curves of Figure 2a, a sharp and large exothermic peak is seen at about 325 ◦C which result from the rapid pyrolysis of bulk lube. Due to the decomposition of lube being an exothermic reaction, once ignition occurs, the heat continually increases and accumulates, and hence, most of lube is removed rapidly. The multiple peaks at 400–500 ◦C imply that the commercial lube contains multifarious hydrocarbons (lubricant additives) that are hard to decompose. Figure 2b plots the lubricant oxidation on CeO2 catalyst. About 93% of the lube is rapidly oxidized between 140 ◦C and 280 ◦C, and the rest of 7% of lube is fully burnt at 280–340 ◦C. The onset combustion temperature of *T*10% is about 162 ◦C, the fastest weightlessness temperature of *Tm* is about 186 ◦C and the final reaction temperature of *Tf* is about 322 ◦C. Figure 2c shows the decomposition of lubricant with TiO2, the TG-DTG curves can be divided into four stages with different decomposition rates. About 16% of lube is decomposed at 205–266 ◦C, another 28% of the lube is burnt at 266–324 ◦C, about 49% of the lubricant is consumed at 324–396 ◦C and the rest of 7% of uninflammable lube is ignited between 396 ◦C and 420 ◦C. The onset combustion temperature of *T*10% is about 249 ◦C, the final reaction temperature of *Tf* is about 420 ◦C, and three obviously fast weightlessness peaks of lube combustion are observed at about 252, 289 and 363 ◦C. For the TiO2-CeO2 catalyzed lube

combustion (Figure 2d), about 97% of the lubricating oil is rapidly combusted between 180 ◦C and 330 ◦C, and the rest of about 3% of lube is burnt out at 334–362 ◦C. The *T*10%, *Tm* and *Tf* is about 212, 261 and 362 ◦C, respectively.

**Figure 2.** Simultaneous TG-DTA-DTG curves for simulating the catalytic performance for the combustion of SOF on (**a**) without catalyst; (**b**) CeO2; (**c**) TiO2 and (**d**) TiO2-CeO2 catalysts.

Due to the fact that the weight loss of lube can be ascribed either to evaporation or to combustion, the activity of prepared catalyst is also identified by the integrated area of the exotherm. The normalized DTA peak areas are described in units of μV· ◦C/(mg lube)/(mg sample), and then different catalysts can be directly compared on a common basis [39,40]. The larger the value of normalized DTA exotherm peak area, the greater the fraction of lubricant combusted verses evaporated, and the better the catalytic performance [39–41]. The oxidation activity data of catalysts for lube combustion are listed in Table 2. CeO2 catalyst exhibits an outstanding catalytic lube oxidation activity, which is consistent with our previous reports [40,41]. Although the introduction of TiO2 slightly lowers the combustion temperature of lube, bare TiO2 is not active enough for catalytic SOF oxidation. About 60% of lube is burnt at 350 ◦C in the lube/TiO2 sample, while the value of lube without catalyst sample is about 90%. This is because the lube oxidation is an exothermic reaction; once ignition occurs, the heat continually increases and accumulates. Thus, the burn of lube without catalyst (containing more lube oil) is more violently than the lube/TiO2. The TiO2-CeO2 catalyst obviously lowers the onset temperature of lube combustion and considerably promote the removing of lube resulting from combustion. Compared to TiO2, the prepared TiO2-CeO2 catalyst presents obviously lower SOF removal temperature and larger exothermal peak area, which indicates that the TiO2-CeO2 catalyst presents better catalytic lube combustion activity.


**Table 2.** Catalytic performances for the combustion of SOF over the prepared catalysts.

<sup>a</sup> Three obviously fast weightlessness peaks are observed over the lube/TiO2 sample.

#### *2.3. Catalyst Characterization*

#### 2.3.1. XRD and Raman Spectra

The XRD patterns of prepared catalysts are shown in Figure 3; both TiO2 and TiO2-CeO2 display only characteristic peaks which refer to the typical anatase structure of TiO2. The peaks of TiO2 are sharper than those of TiO2-CeO2, which indicates that the addition of CeO2 impedes the crystal growth and sintering and lower crystallinity of the TiO2-CeO2 composite materials. In the case of TiO2-CeO2, the typical reflections of CeO2 crystals at 28.7◦, 33.2◦, 47.7◦, 56.6◦ and 77.1◦ are not observed, and the positions of typical anatase structure of TiO2 shift obviously to smaller angles, which suggest that a complete CeO2 crystal is not formed and Ce ions (Ce4+ radius: 0.087 nm) possibly enter into the TiO2 (Ti4+ radius: 0.06 nm) lattice and resulting in an expansion of TiO2 unit cell, the unit cell volume of anatase tetragonal cell of TiO2 is 134.95 Å3, for TiO2-CeO2, the value enlarges to 135.15 Å3. Thus, it can be inferred that Ce ions entered into the TiO2 unit cell, and this is a possible reason why the addition of ceria into TiO2 has no effect on the naturally excellent sulfur resistibility of TiO2.

**Figure 3.** XRD patterns of the TiO2 and TiO2-CeO2 catalysts.

To further confirm the above conjecture, Raman spectra were employed. As can be seen in Figure 4, CeO2 presents a strong peak at about 464 cm−1, which can be associated with the cubic CeO2 [42]. TiO2 and TiO2-CeO2 catalysts show five visible peaks at 145, 196, 397, 517 and 639 cm−1, which are the A1g + 2B1g + 3Eg Raman-active modes of TiO2 anatase phase (the peak at 517 cm−<sup>1</sup> is complex of A1g and B1g) [43], for the TiO2-CeO2 catalyst, the characteristic Raman peak of CeO2 at 464 cm−<sup>1</sup> is not observed. This result further confirmed that a CeO2 crystal is not formed in the TiO2-CeO2 catalyst, which is consistent with the XRD result.

**Figure 4.** Raman spectra of the CeO2, TiO2 and TiO2-CeO2 catalysts.

#### 2.3.2. Nitrogen Adsorption-Desorption

The nitrogen adsorption–desorption isotherms and pore size dispersion of TiO2 and TiO2-CeO2 are shown in Figure 5; both TiO2 and TiO2-CeO2 are mesoporous materials and show distinct H3 and H4 complex hysteresis loops indicating slit pore features; compared to TiO2, the prepared TiO2-CeO2 displays obviously larger pore size; the textual features are listed in Table 3.

**Figure 5.** Nitrogen adsorption-desorption isotherms and pore size dispersion (inset) of the TiO2 and TiO2-CeO2 catalysts.

**Table 3.** The texture properties of TiO2 and TiO2-CeO2 catalysts.


The surface areas of TiO2 and TiO2-CeO2 catalysts are about 98 m2/g and 118 m2/g, respectively, which indicates that the introduction of CeO2 slightly increases the surface area of TiO2. The pore volume and average pore size of TiO2 are about 0.22 cm3/g and 7.2 nm, respectively; the pore volume and average pore diameter of TiO2-CeO2 catalyst are 0.26 cm3/g and 8.6 nm, respectively. It can be seen that the addition of CeO2 to TiO2 increases its surface area, pore volume and average pore size; this is possibly due to the addition of CeO2 which impedes crystal growth and sintering and lowers crystallinity of the TiO2-CeO2 composite materials, and this speculation is consistent with the XRD results (Figure 3). The larger surface area, pore volume and pore size can be advantages to the contacting, transmitting and diffusion of the lube molecules on the catalyst, and hence, be beneficial to the catalytic SOF oxidation activity.

#### 2.3.3. XPS

Figure 6 shows the XPS spectra of O 1s region of the CeO2, TiO2 and TiO2-CeO2 catalysts, all samples show two XPS peaks, the peak at about 530.1 eV can be assigned to lattice oxygen, and the peak with a binding energy of 532.1 eV is characteristic of surface adsorbed oxygen [37,44]. Due to the superior oxygen storage capacity, the peak of surface adsorbed oxygen of CeO2 is obviously stronger than the lattice oxygen; the surface adsorbed oxygen ratio is about 55%. For the TiO2 catalyst, the surface adsorbed oxygen ratio is about 38%, and the addition of CeO2 obviously increases the surface adsorbed oxygen ratio, where the value of TiO2-CeO2 is about 42%. Usually, surface adsorbed oxygen is more reactive than lattice oxygen due to its higher mobility [45,46], and our previous work also confirms that the adsorbed oxygen is more active than the lattice oxygen in the catalytic SOF oxidation reaction [40]. Thus, the addition of CeO2 in TiO2 enhances the amounts of surface adsorbed oxygen of TiO2-CeO2, which can be responsible for the improved catalytic activity of SOF oxidation.

**Figure 6.** XPS (O 1s) spectra of the CeO2, TiO2 and TiO2-CeO2 catalysts.

Additionally, the XPS (Ti2p) spectra of TiO2 and TiO2-CeO2 catalysts are both located at 458.5 eV (2p3/2) and 464.2 eV (2p1/2), which are characteristic of TiO2 species. Compared to bare CeO2, the cerium peaks of TiO2-CeO2 are very weak and indiscernible, which indicates that the surface concentration of Ce in TiO2-CeO2 catalyst is very low; this phenomenon further confirms that Ce dopants are not gathering on the surface and are possibly entering into TiO2 lattice, which is consistent with the XRD results (Figure 3).

#### 2.3.4. H2-TPR

The redox property of a catalyst is closely related to the catalytic performance. The redox behavior of the prepared catalysts is described by hydrogen temperature-programmed reduction (H2-TPR) and shown in Figure 7.

**Figure 7.** H2-TPR profiles of the CeO2, TiO2 and TiO2-CeO2 catalysts.

The TPR peak of individual TiO2 shows a weak peak at about 500–700 ◦C which may be ascribed to the reduction of TiO2, and this phenomenon has been observed by other researchers [47,48]. The CeO2 obviously shows a two-step reduction, the multiple low-temperature reduction peak at about 300–600 ◦C can be assigned to the reduction of surface oxygen, and the peak after 700 ◦C is ascribed to the reduction of CeO2 bulk oxygen [47]. For the TiO2-CeO2, the reduction peak at 300–700 ◦C can be ascribed to the reduction of TiO2 and surface oxygen of CeO2; interestingly, the reduction peak of bulk oxygen of CeO2 after 700 ◦C is poorly identified, which indicates that the TiO2-CeO2 catalyst exhibits lots of surface oxygen, and this result is consistent with XPS result. Consequently, the prepared TiO2-CeO2 catalyst with lots of surface oxygen species presents superior catalytic activity, due to the surface oxygen is highly reactive [45,46].

#### 2.3.5. OSC

The CeO2-based catalyst presents excellent oxygen storage capacity (OSC) and exhibits superior catalytic SOF oxidation activity [16,41]. The OSC of prepared catalysts were tested by an oxygen pulse injection technique and the results are listed in Table 4.


**Table 4.** The oxygen storage capacity (OSC) of TiO2, TiO2-CeO2 and CeO2-based catalysts.

<sup>a</sup> The prepared TiO2-CeO2 catalyst in this work is Ce0.1Ti0.9O2.

It can be seen that the OSC of TiO2 is about 2.9 μmol/g, which is very slight and may be within the measurement uncertainties. While the addition of CeO2 significantly increases the OSC of TiO2-CeO2 catalyst, the OSC is about 101 (μmol O2)/(g sample), and the normalized OSC of TiO2-CeO2 catalyst is about 524 (μmol O2)/(g CeO2). For the pure CeO2 sample, the OSC is about 73 (μmol O2)/(g sample). It should be mentioned that the static oxygen storage capacity testing for OSC of pure CeO2 is an undervalued result, due to the oxygen mobility of bare CeO2 being very low. Thus, a CeO2-based oxygen storage materials in our previous work [49] is used as reference, the normalized OSC of Ce0.35Zr0.60Nd0.05O2 is about 638 (μmol O2)/(g CeO2). It can be seen that the normalized OSC of prepared TiO2-CeO2 catalyst is almost as good as the CeO2-based oxygen storage materials. Based on the results, it can be suggested that although Ce ions enter into the TiO2 lattice, the OSC of CeO2 in TiO2 is not degraded, the TiO2-CeO2 still exhibits good OSC. The good OSC of TiO2-CeO2 is one of the reasons that TiO2-CeO2 catalyst presents excellent catalytic SOF oxidation activity, this is consistent with the catalytic performance results (Figure 2 and Table 2) and the reports [16,41].

#### **3. Materials and Methods**

#### *3.1. Catalyst Preparation*

The TiO2-CeO2 catalyst was prepared by a co-precipitation method. Desired TiOSO4·2H2O and Ce(NO3)3·9H2O mixture solutions with the molar ratio of Ti:Ce = 9:1, which was the optimal ratio to expose the single TiO2 crystal structure in our previous studies [35,36], were slowly added to NH3·H2O solutions under vigorous stirring. And then the precipitate was filtered and washed many times, after dried at 120 ◦C overnight and calcined for 3 h at 500 ◦C under airflow, the TiO2-CeO2 catalyst powder was obtained. The TiO2 and CeO2 catalysts were prepared by the same method.

The simulative 160,000 km vehicle aged sample was obtained by following reference [50] and our previous works [35,37]. Fresh catalyst was placed in a reactor and aged at 670 ◦C for 15 h and then at 250 ◦C for 15 h in the gases mixture at flow rate of 800 mL/min: 600 ppm C3H6, 1500 ppm CO, 200 ppm NO, 50 ppm SO2, 5% O2, 4% CO2, 8% vapor and N2 balance.

#### *3.2. Catalyst Evaluation*

The catalytic activity for SOF combustion of prepared catalysts were tested using TG-DTA method [16,39]. Due to the fact that the diesel SOF is comprised primarily of lube with a small amount of unburned fuel [51], commercial lubricating oil was often used to simulate the diesel SOF catalytic combustion [16,39].

For this test, the prepared powder catalyst was dried overnight at 120 ◦C to remove the effects of surface adsorbed water, and then impregnated with 5.0 wt. % commercial lube (Shell Helix HX7 5W-40, Shell Petrochemicals Company Limited, Jiaxing, China), the slurry of lube/catalyst mixture was stirred and mixed till a homogeneous state was obtained. About 10 mg of the lube/catalyst mixture powder was placed in the sample pan of TG-DTA unit (HCT-2, Beijing Henven Instruments, Beijing, China) and dried at 120 ◦C for 1 h to eliminate the effects of adsorbed substances (water, volatile matters etc.), and then heated to 550 ◦C with a temperature rate of 5 ◦C/min under airflow at 30 mL/min. The TG-DTA curves were recorded to determine the catalytic performance for SOF combustion of the prepared catalysts.

The test on lube without catalyst was carried out as a reference. About 10 mg of the lube was placed in the sample pan of TG-DTA unit and dried at 120 ◦C for 1 h, and then heated to 550 ◦C with a temperature rate of 5 ◦C/min under airflow at 30 mL/min.

#### *3.3. Catalyst Characterization*

Sulfur uptake was tested on a thermogravimetric analyzer (TGA) HCT-2 (Henven Instruments, Beijing, China). Consulting the references [38,52] and our previous works [37,53], about 15 mg of catalyst was placed in the sample crucible and pretreated under a 35 mL/min of N2 flow for 5 h at 300 ◦C, and then the gas mixture of 43 mL/min SO2(0.05 vol. %)-N2 and 31 mL/min O2(15.0 vol. %)-N2 was introduced at 300 ◦C for 4 h, the weight increase as a function of time was recorded by the TGA.

The sulfur cumulant of catalysts after mimicking 160,000 km vehicle aging were analyzed by an energy-dispersive X-ray (EDX) spectroscopy (IE-250, Oxford Instruments, Oxford, UK).

X-ray powder diffraction (XRD) patterns were collected on a powder X-ray diffractometer (DX-1000, Dandong Fangyuan Instrument Ltd., Dandong, China) employing Cu Kα radiation (λ = 0.1542 nm).

Raman spectra were recorded by a LabRAM HR Laser Raman spectrometer (HORIBA Jobin Yvon Inc., Paris, France) with an excitation wavelength of 532 nm.

N2 adsorption-description isotherms were measured on a QUADRASORB SI automated surface area and pore size analyzer (Quantachrome Instruments Ltd., Boynton Beach, FL, USA). The specific surface area and pore size were calculated by the Brunauer-Emmett-Teller (BET) method and Barret-Joyner-Halenda (BJH) method, respectively. Before adsorption measurements, the samples were degassed at 300 ◦C for 3 h under vacuum.

X-ray photoelectron spectroscopy (XPS) data were collected on a Kratos XSAM 800 spectrometer (Kratos Analytic Inc., Manchester, UK) with Al Kα radiation. The binding energy shifts of the samples were calibrated by fixing the C1s binding energy (BE 284.8 eV).

H2-temperature programmed reduction (H2-TPR) experiments were performed in a quartz tubular reactor. Samples were pretreated at 450 ◦C for 1 h under the N2 flow (35 mL/min) and then cooled to room temperature; after that, the samples were heated from room temperature to 800 ◦C with a heating rate of 10 ◦C/min under the flow of H2 (5.0 vol. %)-N2 mixture. The hydrogen consumption as a function of reduction temperature was recorded by a thermal conductivity detector (TCD) cell.

The oxygen storage capacity (OSC) of the samples was measured by a pulse injection technique [54]. The sample was firstly reduced in a H2 flow (30 mL/min) at 550 ◦C for 1 h; after cooling to 200 ◦C, an oxygen pulse was injected every 5 min to obtain a breakthrough curve, from which the OSC was calculated.

#### **4. Conclusions**

From the aforementioned results, it can be concluded that moderate amounts of ceria dopants in titania can obviously enhance the catalytic SOF oxidation activity of TiO2-CeO2 catalyst. Meanwhile, the prepared TiO2-CeO2 catalyst can maintain the naturally excellent sulfur resistibility of titania; the sulfur resistibility of TiO2-CeO2 is as well as the bare TiO2. The prepared TiO2-CeO2 catalyst significantly enhances the sulfur tolerance of conventional CeO2-based SOF oxidation catalysts and displays a good catalytic SOF oxidation activity. The TiO2-CeO2 catalyst exhibits a typical phase of anatase, and the cerium ions can enter into the TiO2 unit cell, impede the crystal growth and sintering and lower crystallinity of the TiO2-CeO2 composite materials and, hence, improve the surface area, pore volume and pore size of TiO2-CeO2 catalyst. Moreover, the addition of CeO2 in TiO2 can significantly enhance the surface oxygen concentration and oxygen storage capacity of TiO2-CeO2; the normalized oxygen storage capacity of TiO2-CeO2 is almost as good as the CeO2-based oxygen storage materials. The improvement of textual features and surface oxygen concentration of TiO2-CeO2 catalyst are the main reasons for the enhancement of catalytic SOF oxidation activity.

**Author Contributions:** Z.Y. and Y.C. conceived the project; N.Z., Z.Y. and Z.C. performed the experiments; Y.L. (Yunxiang Li), Y.L. (Yunwen Liao), Y.L. (Youping Li) and M.G. carried out the data analysis; N.Z. and Z.Y. wrote the paper.

**Acknowledgments:** This work was supported by the National Natural Science Foundation of China (21703174), Meritocracy Research Funds of China West Normal University (17YC146) and Doctor Startup Foundation of China West Normal University (15E012). Zhi Chen is particularly grateful to the Sichuan Provincial Students' Innovation and Entrepreneurship Training Program (201610638094).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
