TiO2 for selective catalytic oxidation of ammonia to nitrogen

Journal Pre-proofs Full Length Article Reaction properties of ruthenium over Ru/TiO2 for selective catalytic oxidation of ammonia to nitrogen Jung Hun Shin, Geo Jong Kim, Sung Chang Hong PII: DOI: Reference:

S0169-4332(19)33723-7 https://doi.org/10.1016/j.apsusc.2019.144906 APSUSC 144906

To appear in:

Applied Surface Science

Received Date: Revised Date: Accepted Date:

22 July 2019 21 October 2019 29 November 2019

Please cite this article as: J. Hun Shin, G. Jong Kim, S. Chang Hong, Reaction properties of ruthenium over Ru/TiO2 for selective catalytic oxidation of ammonia to nitrogen, Applied Surface Science (2019), doi: https:// doi.org/10.1016/j.apsusc.2019.144906

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

© 2019 Published by Elsevier B.V.

Reaction properties of ruthenium over Ru/TiO2 for selective catalytic oxidation of ammonia to nitrogen

Jung Hun Shina, Geo Jong Kimb, Sung Chang Honga†

a

Department of Environmental Energy Engineering, Graduate School of Kyonggi University,

94-6 San, Iui-dong, Youngtong-ku, Suwon-si, Gyeonggi-do 443-760, Republic of Korea b

University of Kentucky Paducah Extended Campus Program, 209 Crounse Hall, 4810 Alben

Barkley Drive Paducah, Kentucky 42002, United states of America E-mail address; [email protected]

†

Telephone; +82-31-249-9733, Fax: +82-31-254-4905

Abstract The objective of this study was to investigate reaction characteristics of selective catalytic oxidation of ammonia to nitrogen (NH3-SCO) using ruthenium-based catalysts. Ruthenium was compared with platinum, a metal generally used as an NH3 oxidation catalyst. Experiment was carried out by depositing active metal onto TiO2. 1Ru/TiO2 was superior to 1Pt/TiO2 in NH3 conversion and N2 yield. Therefore, various analyses were conducted to determine the influence of catalytic properties of 1Ru/TiO2 on reaction activity. X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and Field EmissionTransmission Electron Microscope (FE-TEM) analyses confirmed that ruthenium was present in RuO2. Such a structure was confirmed to be excellent for adsorption ability of NH3 and oxygen through analysis of NH3-TPD, NH3-TPO, H2-TPR, and O2-chemisorption. In

1

addition, in situ diffuse reflectance infrared Fourier transform spectroscopy (DIRFTS) analysis suggested that Ru/TiO2 had excellent ability to convert NH3 to NO. The converted NO then reacted with adsorbed NH3 to generate N2. Based on results of NH3 injection at various concentrations, it was found that the rate at which NH3 was converted to NO from the Ru/TiO2 surface affected N2 selectivity. Keywords: Ruthenium; NH3-SCO; Selective catalytic oxidation; I-SCR; Mechanism

2

1. Introduction Gaseous ammonia is generated from various fixed/mobile pollution sources such as gas slip from selective catalytic reduction (SCR) of NOx using NH3 or urea in deNOx process, industrial wastewater, chemical process, and semiconductor manufacture process [1]. Ammonia can have serious effects on human health and the environment [2]. For example, ammonia has a serious odor at 50 ppm or less. High concentrations of ammonia can cause serious injuries and burns to the respiratory tract [3]. It has been recently reported that the production of PM2.5 is strongly influenced by the emission of ammonia [4]. Therefore, developing a technology that can control ammonia is becoming very important. There are various techniques for removing ammonia, such as adsorption, absorption, biofiltration, and catalytic oxidation. In the case of adsorption and absorption, operating costs associated with additional processing of adsorbent and absorbent can be significantly high. In the case of biofiltration, it is operated under specific conditions. Various factors such as water content, pH, light, oxygen availability, nutrition, and temperature can have significant impact on its performance. Thus, selective catalytic oxidation of ammonia to nitrogen (NH3-SCO) method that can convert ammonia selectively to nitrogen and water without additional steps under various conditions is a limelight technology [5]. Various catalysts for NH 3-SCO have been studied. They can be divided into noble metal and transition-metal oxide depending on the active metal. Many transition-metal oxide catalysts such as Fe2O3 [6], CuO [7], MoO3 [8], CoO3 [9], MnO2 [10], and CeO2 [11] have been found to have superior N2 selectivity. However, high temperatures of 300 °C or higher are required. Therefore, researchers have focused on noble metals such as Pt [12, 13], Pd [14], Rh [15], Ru [16], and Ag [17] that have excellent activities over a wide temperature range. Among these catalysts, research on Pt-

3

based catalysts having the highest reaction activity has been actively conducted. However, Pt-based catalysts are mainly converted to N2O which has a problem of low N2 selectivity [18-20]. Therefore, it is important to develop catalysts with excellent N2 selectivity at temperature below 300°C to replace Pt-based catalysts. Ruthenium is a heterogeneous catalyst with excellent efficiency for CO [21-22], HCl [2324], and NH3 [25] oxidation. According to Wang et al. [26], a coordinately unsaturated (cus) Ru atom exposed on the surface of Ruthenium oxide (RuO2) is excellent for ammonia adsorption and dissociative adsorption of molecular oxygen. Many researchers believe that adsorbed ammonia is converted from the surface of RuO2 to NO molecules and adsorbed [27, 28]. By using RuO2-CuO/Al-ZrO2 composites, Cui et al. [29] have recently shown an excellent N2 yield rate of 100% up to about 250°C. These effects were due to conversion of NH3 to NO molecules by RuO2 in the catalyst. Adsorbed NO and NH3 were then converted by CuO to N2 through SCR reaction. The reaction formula of SCR is as follows:

4NH3 + 4NO + O2 → 4N2 + 6H2O

(1)

The reaction in which oxidation/reduction of NH3 proceeds with one catalyst is called an internal-selective catalytic reduction (I-SCR) mechanism [5,13,30,31]. It has excellent N2 yield. However, this technology is economically difficult to become commercially available because very high contents of ruthenium and copper are required. Therefore, there is a need to develop a catalyst with excellent NH3 conversion rate and high N2 yield utilizing low content of ruthenium. For this reason, studies on the role and mechanism of ruthenium based catalysts for NH3-SCO reaction are essential.

4

Thus, the objective of this study was to investigate reaction characteristics of selective catalytic oxidation of ammonia to nitrogen (NH3-SCO) using ruthenium-based catalysts. To determine reaction characteristics of ruthenium-based catalyst by NH3-SCO reaction, the catalyst was prepared by supporting ruthenium on reducible support TiO2. Platinum was produced with the same method for comparing catalytic activity and properties. The relationship between catalyst performance and physicochemical properties was determined through X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), Field EmissionTransmission Electron Microscope (FE-TEM), NH3-Temperature Program Desorption (NH3-TPD), NH3-Temperature Program Oxidation (NH3-TPO), H2-Temperature Program Reduction (H2-TPR), and O2-Chemisorption analyses. In addition, the mechanism of ruthenium based catalyst was proposed through in situ diffuse reflectance infrared Fourier transform spectroscopy (in situ DRIFTS) analysis.

2. Experiments 2.1 Catalyst preparation Ru/TiO2 and Pt/TiO2 were manufactured by the wet impregnation method. Commercialized TiO2 G5 (Millennium/Crystal, 100% anatase) was used as a support. Ruthenium nitrosylnitrate (RuNO(NO3)3/Ru 31.3%; Alfa Aesar) and platinum hydroxide (Pt(OH)2; SNS Co.) were used as active metals. Calculated amounts of ruthenium nitrosylnitrate (1wt.% Ru) and platinum hydroxide (1wt.% Pt) were dissolved in 60 ml distilled water at room temperature and stirred for 1 hr. The mixture was then added to a TiO2 support and stirred for 1 hr to form a slurry. Using a rotary vacuum evaporator at 65 °C, water was removed at a pressure of 60 mmHg. It was dried at 110 °C for 24 hr in a dry oven to obtain a powder

5

which was then heated at a rate of 10 °C/min in a furnace at 400 °C for 4 hr. The catalyst was prepared at a constant size through a 40-50 mesh filtering.

2.2 Catalytic testing The NH3-SCO experimental device consisted of a gas injection, a reactor, and an analysis of the reaction gas. The experimental device was a continuous flow fixed reactor. The gas supplied to the reactor was adjusted for flow rate using a MFC (Mass Flow Controller; MKS Co.). The composition of the gas included 30~2,000 ppm NH3 and 10% O2 with 6% moisture in N2. The water injected into the reactor was designed so that the N2 gas could pass through the bubbler and contain water. The temperature was maintained at 45 °C using a circulator on the outside of a bubbler so that the amount of water supplied was constant. The gas supply pipe was entirely manufactured as a stainless pipe and maintained at 180 °C using a heating band so that the water in the reaction gas was not condensed. Total flow rate through the reactor was 500 ml/min and the space velocity was 60,000 hr-1. Approximately 0.3 g of catalyst was used for each test. The reactor used was Quartz tube and the catalyst layer was fixed using quartz wool. The reaction temperature of the catalyst was adjusted using a PID(Proportional-Integral-Differential) controller and K-type thermocouple fixed on the catalyst layer. Prior to experimenting with the catalyst, samples were pretreated at 400 C for 0.5 hr under 10% O2/N2 condition. Concentrations of reactants or products were measured as follows. NO concentration of inlet or outlet was measured using an NO analyzer (ZRE gas analyzer; Fuji Electric Co.). N2O concentration was measured by using an N2O analyzer (gas analyzer; Fuji Electric Co.). Concentrations of NO2 and NH3 were measured using a detector tube of 9L, 3L, 3La, or 3M (Gastec Co.). When SCO reaction reached steady state, NH3

6

conversion and N2 yield (%) were calculated from gas concentration according to equations (2) and (3): NH3 conversion(%) = ( N2 yield(%) =(

[𝑁𝐻3 ]𝑖𝑛 −[𝑁𝐻3 ]𝑜𝑢𝑡 [𝑁𝐻3 ]𝑖𝑛

) x 100

[𝑁𝐻3 ]𝑖𝑛 −[𝑁𝐻3 ]𝑜𝑢𝑡 −[𝑁𝑂]𝑜𝑢𝑡 −[𝑁𝑂2 ]𝑜𝑢𝑡 −2[𝑁2 𝑂]𝑜𝑢𝑡 [𝑁𝐻3 ]𝑖𝑛

(2) ) x 100

(3)

2.3 Characterization of Catalysts Specific surface area of NH3 oxidation catalyst was measured using an ASAP (Accelerated Surface Area and Porosimetry) 2010 instrument (Micromeritics) and calculated using Brunauer-Emmett-Teller (BET) model. Each sample was analyzed after removing gas under vacuum for 2 hr at a temperature of 300°C. XRD was performed by using an X’Pert PRO MRD multipurpose x-ray diffractometer (Panalytical Co.). The radiation source was CuKα (λ = 1.5056 Å) and the generator was 30 kW. 2θ was measured in the range of 10~90° at a scanning rate of 6°/min. X-ray fluorescence (XRF) was performed by using a ZSX PrimusⅡ spectrometer (Rigaku) to measure bulk quantities of elements in catalysts. FETEM analysis was performed using JEM-2100F transmission electron microscope (JEOL Co.). The voltage of electron gun was 200 kV. Sample contained a small amount of powdered catalyst in distilled water. It was then evenly dispersed using an ultrasonic vibrator. The sample was then dropped onto a Cu grid and dried at 103℃ oven for 30 min before analysis. Particle size was estimated as a result of measuring approximately 50 particles. XPS was performed by using an ESCALAB 210 spectrometer (VG Scientific) and Al Kα wavelength (1486.6 eV) was used as the excitation source. After removal of moisture

7

completely from the catalysts by drying at 100℃ during 24h, they were analyzed without surface sputtering or etching in order to keeping a vacuum pressure at 10-6 Pa. Spectra were analyzed by using the XPS PEAK software (ver. 4.1). NH3-TPD was measured using quadrupole mass spectrometer (PFEIFFER VACUUM Co.). The reactor was charged with 0.3 g of catalyst ground to less than 100 μm. Total flow supplied to the catalyst was maintained at 50 ml/min. The catalysts was pretreated by injecting 10% O2/He for 30 min to remove H2O and impurities from the sample surface. And 5% NH3/Ar was injected under 70 ℃ for 1 hr to ammonia adsorption. For removing physical adsorption species, Ar was injected before starting NH3-TPD. Data corresponding to NH3 (m/e 15) were recorded utilizing quadrupole mass spectrometer during TPD analysis and the temperature was ramped up to 700 °C at a rate of 10 °C/min. NH3-TPO had the same pretreatment process as NH3-TPD. However, before starting NH3TPO, physically adsorbed species were removed for 1 hr in a 10% O2/He atmosphere. Data corresponding to NH3 (m/e 15), N2 (m/e 28), NO (m/e 30), and N2O (m/e 44) were recorded using the quadrupole mass spectrometer during TPO analysis. The temperature was raised to 700 ℃ at a rate of 10 ℃/min. H2-TPR was performed using Autochem 2920 Chemisorption Analyzer (Micromeritics Co.). TCD (Thermal conductivity detector) was used for concentration measurement. The reactor was charged with 0.3 g of ground catalyst (less than 100 μm). In order to remove impurities and water on the catalyst surface, 50 ml/min of 10% O2/He gas was supplied to the catalyst at a temperature of 400 °C. The temperature was lowered to 60 °C. After stabilizing the catalyst surface by supplying 10% H2/Ar gas, the temperature was raised to 700 °C at a rate of 10 °C/min.

8

O2-Chemisorption was also measured using the Autochem 2920 Chemisorption Analyzer (Micromeritics CO.). TCD (Thermal conductivity detector) was used to evaluate oxygen mobility capacity of the catalyst. The catalyst was pretreated by injecting 10% H2/Ar for 0.5 hr at 200 ℃ to remove active metal oxygen. O2 pulses of 200 ppm were then introduced at room temperature until there was no variation between two consecutive peaks. Cumulative oxygen consumption was calculated by summing the total of each pulse. In situ DRIFTS analysis was performed using iS10 FTIR Spectrometer (Thermo Fisher Co.) fitted with a CaF2 window. Spectra were collected by accumulating 30 scans at 4 cm-1 resolution using a mercury cadmium telluride (MCT) detector.

3. Results and Discussion 3.1. Catalytic Activity for Ammonia SCO. Figure 1 shows NH3-SCO catalytic activity of 1Ru/TiO2 and 1Pt/TiO2 loaded with active metals (Ru, Pt) by using an wet impregnation method with TiO2 having anatase phase. Figure 1(a) shows NH3 conversion and N2 yield according to the function of temperature of catalysts when 200 ppm of ammonia is injected. In the case of 1Pt/TiO2, activity did not appear up to 225°C with T90 (temperature of 90% conversion) showing at 270°C. On the other hand, 1Ru/TiO2 showed NH3-SCO activity at a lower temperature than 1Pt/TiO2 and T90 was shown at 258°C. The difference in activities of these catalysts was also confirmed by N2 yield. The N2 yield of 1Ru/TiO2 was better in the whole temperature range compared to that of 1Pt/TiO2. The maximum N2 yield of 1Pt/TiO2 was 39.9% at 250°C while that of 1Ru/TiO2 was 65.3% at the same temperature. The exhaust gas that determines N2 yield is shown in Figure 1(b). NOx produced by the side reaction during NH3 oxidation could be classified into NO, NO2, and N2O. The reason

9

why 1Ru/TiO2 N2 yield is decreased is because it has a large effect on NO and NO2. On the other hand, it was found that the exhaust gas of 1Pt/TiO2 was affected by N2O more than by NO and NO2. N2O is a greenhouse gas that contributes to global warming 250 times more than that of CO2 [32,33]. In addition, 2 moles of NH3 can react in the NH3-SCO reaction to produce 1 mole of N2O which can greatly affect the reduction of N2 yield. In addition, N2O is more difficult to remove than NO or NO2 [34,35].

3.2. Characterization of catalysts 3.2.1. Structure analysis Structures of 1Ru/TiO2 and 1Pt/TiO2 were determined by XRD and TEM analyses. XRD analysis results of TiO2, 1Ru/TiO2 and 1Pt/TiO2 are shown in Figure 2. TiO2 characteristic diffraction peaks revealed the anatase phase. However, different from 1Pt/TiO2, new peaks were observed at 27.96o and 35.10o for 1Ru/TiO2. According to previous studies [25,36], new peak is known to be the rutile form of RuO2 (110 and 101). TEM analysis results clearly revealed its structural characteristics as shown in Figure 3. The particle diameter observed from TEM image of 1Pt/TiO2 (Figure 3(b)) was confirmed to be about 3-7 nm. However, the particle diameter of 1Ru/TiO2 (Figure 3(a)) was observed to be about 29 nm. Table 1 summarizes active metal contents and specific surface areas of catalysts using XRF and BET analyses. It was confirmed that the particle diameter of 1Ru/TiO2 catalyst was larger despite the fact that both catalysts had the same active metal content and specific surface area. These characteristics of ruthenium have been reported in many previous studies. According to Taiwo et al. [37], RuO2 has a rutile phase. Since RuO2 has a similar structure when it is deposited to TiO2 of rutile structure, bonding occurs more easily, leading to excellent

10

dispersion. It has been shown that anatase TiO2 can lead to agglomeration of Ru and low dispersion [38-41]. Thus, it can be assumed that the use of support (anatase TiO2) as described above can increase the particle size due to agglomeration of ruthenium. Results of XRD and TEM analyses confirmed that the active metal of 1Ru/TiO2 was aggregated and presented in the form of rutile RuO2. According to previous studies [16,26,42,43], the RuO2 structure has a dangling bond in which some bonds are broken due to atomic coordination unsaturation of the crystal surface, thus facilitating chemical bond with ammonia molecules or oxygen easily.

3.2.2. XPS surface chemical analysis X-ray photoelectrion spectroscopy (XPS) analysis were carried out to elucidate the chemical state of 1Pt/TiO2 and 1Ru/TiO2 catalysts. The XPS spectra results are shown in Figure 4. As shown in Figure 4(a), two main spectral signal with binding energy located at 458.8 and 464.4 eV belonged to Ti 2p3/2 and Ti 2p1/2 of Ti4+[44]. The peak of Ti 2p3/2 and Ti 2p1/2 on the both catalysts appeared a difference about 5.7 eV, which was consistent with state of Ti4+. Also, The binding energy of the Ti 2p peaks slightly could shift than that of pure Ti 2p, because of the metal support interaction by the addition active metal[44]. In Figure 4(a), binding energies of Ti 2p of pure TiO2 were 258.77 and 264.47 eV, respectively. 1Ru/TiO2 and 1Pt/TiO2 were shifted to lower binding energies(458.56 and 464.26 eV). It is means that there is a weak metal support interaction. But, as the Ti 2p peak of both catalysts existed in the same binding energy, it is suggested carefully that interaction strength with TiO2 doesn’t differ. Figure 4(b) showed the Pt 4f spectrum results of a 1Pt/TiO2 catalyst. Kim et al. [61], noted that the valence state of active Pt metal exists as Pt4+ (74.24 eV) and Pt2+ (72.64 eV),

11

while Pt4+ and Pt2+ contribute to PtOx species respectively. These result can confirm the active Pt metal on 1Pt/TiO2 surface was mainly existed as a PtOx species in Figure 4(b). However, the relationship between valence state (Pt2+, Pt4+) and N2 yield in NH3-SCO has not yet been clearly identified. Therefore, the impact on oxidation was not addressed in this study. The Ru 3d spectra of 1Ru/TiO2 catalyst was plotted in Figure 4(c). Unfortunately, the binding energy of Ru 3d3/2 was overlapped with of C 1s. According to Cui et al. [16], C1s and C=O carbon species on the sample surface were located at the binding energy of 284.6 and 288.2 eV. Also, they have mentioned that the components of the binding energy position of 281 and 284.2 eV can be attributed to Ru 3d5/2 and Ru 3d3/2 in RuO2, respectively, and the peak strength at the binding energy of 281 eV increases with the amount of RuO2 load. In figure 4(c), We have confirmed the Ru 3d5/2 and Ru 3d3/2 corresponding to RuO2 species. These results can confirm, as with the XRD results(Figure 2), that 1Ru/TiO2 catalyst actually exist as RuO2 and TiO2. The O 1s spectra of 1Ru/TiO2 and 1Pt/TiO2 was displayed in Figure 4(d). In the O 1s results, two overlapping oxygen species were observed. The first peak observed in the binding energy of 529-530 eV contributes to the lattice oxygen (Oα), while the second peak overlapped in the binding energy of 532 eV contributes to the surface adsorbed oxygen (Oβ)[29]. The surface molar fractions of O 1s in the both catalysts were summarized in Table 2. The Oβ/(Oα+Oβ) ratio of the Ru/TiO2 and Pt/TiO2 were 56.64% and 39.38% respectively. Also, The ratio of Oβ/(Oα+Oβ) on Ru/TiO2 catalyst was larger than that of Pt/TiO2. As mentioned earlier (3.2.1), it was known that the RuO2 on the surface of 1Ru/TiO2 is easily coupled with the surface oxygen species(Oβ). It was noted commonly for surface adsorbed oxygen species(Oβ) to play an important role on catalytic activity in NH3-

12

SCO[29,62]. The role of the surface adsorbed oxygen species are discussed in more detail later in this study.

3.3. Reaction properties 3.3.1. NH3 adsorption NH3-TPD analysis was performed to compare ammonia adsorption properties of catalysts. Results are shown in Figure 5. These catalysts contained broad NH3 desorption peaks at 80200 °C and 230-350 °C corresponding to Bronsted and Lewis acid sites, respectively [44, 45]. NH3 adsorption intensity of 1Ru/TiO2 was larger than that of 1Pt/TiO2. When integrating and comparing overall NH3 desorption area of the two catalysts, it was found that 1Ru/TiO2 was 1.75 times wider than 1Pt/TiO2. These results were also confirmed by NH3-TPO analysis as shown in Figure 6(a). It was confirmed that these adsorption characteristics were consistent with characteristics of dangling bonds of RuO2 described above. Also, the availability of a large amount of ammonia could be considered as the reason why NH3 conversion of 1Ru/TiO2 was excellent compared to 1Pt/TiO2 as shown in Figure 1. Figure 6(b) shows outlet gas in NH3-TPO analysis. As a result, it was confirmed that peaks of N2, NO, and N2O gas were generated as the reaction progressed at about 230~300 ℃ for both catalysts. 1Ru/TiO2 was converted to N2 more than 1Pt/TiO2. On the other hand, N2O area of 1Pt/TiO2 was generated about 2.974 times wider than 1Ru/TiO2 and NO was generated about 1.197 times wider. Accordingly, the formation of NOx by NH3 oxidation of 1Ru/TiO2 was suppressed. In addition, the L acid area observed in NH3-TPD was found to disappear as oxygen was injected (NH3-TPO). The area was similar to the temperature at which the

13

reaction activity of both catalysts began, suggesting that the formation of N2 and NOx in the NH3-SCO reaction might have consumed ammonia of L acid.

3.3.2. Mechanism Studies NH3-SCO has various mechanisms, including I-SCR [13, 30, 31], hydrazine [46-48], and HNO [14, 49] mechanisms. Therefore, in situ DRIFTS analysis was performed to compare the mechanism of the two catalysts. Before carrying out the analysis, samples were pretreated with 10 vol.% O2/N2 for 30 min at a temperature of 400 °C. Figure 7 shows results of adsorption of 200 ppm ammonia without oxygen at a temperature of 300 °C for 30 minutes. Figure 7(a) shows results of 1Ru/TiO2. NH stretching vibration band of ammonia was found at 3372, 3270, 3245, and 3159 cm-1 [14,45,49]. In addition, the Lewis acid site was observed at 1620 cm-1 [14,45,50]. Amide with various oscillations by dissociating the adsorbed ammonia was observed at 1347, 1326 cm-1 (NH2-wagging), and 1220-1150 cm-1(N-N stretching and NH2 rocking) [45]. Also, linearly-NO adsorbed to ruthenium oxide was observed at 1844 cm-1. Bidentate and bridged nitrate was revealed at 1603 cm-1 and 1241 cm1

[51]. Despite the condition that oxygen was not injected, peak growth of NH stretching

vibration band (3372, 3270, 3245, 3159 cm-1), linearly-NO (1844 cm-1), bidentate (1603 cm1

) and bridged (1241 cm-1) nitrate increased from about 3 minutes and became stable from

10 minutes. This makes it possible to propose that the catalyst can generate linear-NO and bidentate nitrate using internal oxygen for 10 minutes. In addition, NO and nitrate generated can react with NH3 through SCO mechanism. The mechanism by which oxidation/reduction of NH3 takes place on one catalyst surface is called I-SCR. Some previous studies have stated that I-SCR mechanism can improve N2 selectivity [52,53]. Based on 1Pt/TiO2 results shown

14

in Figure 7(b), the NH stretching vibration band of ammonia appearing at 3384, 3316, 3275, 3207 cm-1, Lewis acid site appearing at 1620 cm-1 [14,45,50], and NH2-wagging appearing at 1325 cm-1 [44,45] were found. In addition, linearly-NO adsorbed on Pt at 1760 cm-1 [54, 55] and bidentate nitrate at 1540 cm-1 [56] were observed. Different from 1Ru/TiO2, 1Pt/TiO2 produced various peaks at the same time as ammonia injection. It formed a small amount of nitrate. Furthermore, the peak regarded as N2O was observed at 2300-2150 cm-1 [14]. Results confirmed that the selectivity for N2O was large via NH3 oxidation as a peak was not observed in this 1Ru/TiO2. Thus, 1Pt/TiO2 rarely progresses to I-SCR. As shown in Figure 8, NH3 injection was stopped for 200 ppm pre-adsorbed catalyst and O2 was injected by about 10%. Figure 8(a) shows result of 1Ru/TiO2. All peaks except those for bridge nitrate (1603, 1580 cm-1) decreased simultaneously with oxygen injection. On the other hand, the peak corresponding to bridge nitrate increased rapidly. This confirms that 1Ru/TiO2 can convert adsorbed NH3 into nitrate species very quickly. Moreover, the converted nitrate can be detached from NO and NO2 by decomposition reaction. As shown in Figure 1(b), it is possible to confirm that the exhaust gas of 1Ru/TiO2 at 300°C is almost desorbed by NO or NO2. This might be due to insufficient NH3 required for the SCR reaction. Although oxygen is injected, from results of 1Pt/TiO2 shown in Figure 8(b), 3384, 3316, 3275, 3207cm-1, and NH2-wagging (1325 cm-1) and 2300-2150 cm-1 peak regarded as N2O corresponding to the NH stretching vibration band decreased slowly. On the other hand, linearly-NO (1760 cm-1) and bidentate nitrate (1540 cm-1) decreased and disappeared [5456]. However, considering N2 selectivity, 1Pt/TiO2 proceeded NOx formation by NH3 direct oxidation than SCR reaction [18].

15

Figure 9 shows result of injecting 10% O2 onto sample adsorbed on NH3 at 180 ℃. As shown in Figure 9(a) for the result of 1Ru/TiO2, the formation of linear-NO (1852 cm−1), bidentate nitrate (1558 cm-1, 1241 cm−1) was observed, although the reaction did not progress [51,57]. However, this was not observed in the result of 1Pt/TiO2 shown in Figure 9(b). Thus, 1Ru/TiO2 has a better ability to generate NO or nitrate needed for I-SCR reaction than 1Pt/TiO2. NO generation is an important factor in the progress of I-SCR reaction. Although 1Ru/TiO2 can generate NO necessary for I-SCR reaction, as no reaction activity is processed, the oxygen species in the catalysts required to convert to NO and the oxygen species required for I-SCR reaction might be different from each other. The present study also confirmed that oxygen in the catalyst could affect reaction activity based on DRIFTs results. Therefore, H2-TPR analyses were performed to determine reducibilities of catalysts. Results are displayed in Figure 10. 1Ru/TiO2 had peaks at 95.5 °C, 120 °C, 151.59 °C, and 314.34 °C while 1Pt/TiO2 had those at 96.38 °C and 286.39 °C. According to previous studies using ruthenium catalyst [58, 59], results of H2-TPR showed that peak was generated when ruthenium oxide species was reduced to metallic at a temperature of 206 °C or less. Also, peak was generated when TiO2 was reduced at ruthenium due to hydrogen species spillover at temperature 300~800 oC. Kim et al. [60] have mentioned that H2 consumption peaks observed at 127 °C and 210~530 °C are attributed to the reduction of PtOx species and the reduction of surface oxygen on TiO2 support, respectively. Thus, H2 consumption peak at temperatures below 200 °C of both catalysts was judged to be oxygen species of active metal (Ru, Pt) and named active metal oxygen, whereas the peak produced between 300 °C and 500 °C was surface oxygen species of TiO2 by spillover and named TiO2 surface oxygen. Based on results of Figure 9, active metal oxygen can be used to investigate

16

oxygen species required for NH3 to be converted to NO. Because active metal oxygen could be used at temperatures below 200 °C where NO3 peak was observed, SCR reaction did not progress. The conversion to NO was better because 1Ru/TiO2 had much more active metal oxygen than 1Pt/TiO2. Thus, 1Ru/TiO2 could be used for SCR to generate N2 by reaction of NH3 and NO. The peak of TiO2 surface oxygen began to increase at about 225oC when 1Ru/TiO2 activity appeared. TiO2 surface oxygen by spillover might have been used for the SCR reaction. The amount of active metal oxygen in catalyst was compared through O2-chemisorption analysis. Results are shown in Figure 11. Before conducting this analysis, the adsorption of active metal oxygen was considered by reducing the sample at 200 °C under 30% H2/N2 atmosphere. As a result, the final cumulative oxygen content was 1.011×10-6 mmol/g in 1Ru/TiO2 and 2.190×10-7 mmol/g in 1Pt/TiO2. These results suggest that 1Ru/TiO2 has more active metal oxygen that can switch NH3 to a larger amount of nitrate than 1Pt/TiO2.

3.4. N2 selectivity by NH3 injection at different concentrations According to our study results, 1Ru/TiO2 had better N2 yield than 1Pt/TiO2 by I-SCR reaction. However, in Figure 1(a), 1Ru/TiO2 did not have complete N2 selectivity. As shown in Figure 8, NH3 was rapidly switched to nitrate in the catalyst. Thus, NH3 required for SCR reaction was lacking. This confirms the effect of concentration of injected NH3 on N2 yields of both catalysts. Experiments were conducted at 300 °C so that NH3 conversion rate was 100%. Results are shown in Figure 12. When 30 ppm of NH3 was injected, N2 yield of 1Ru/TiO2 was significantly decreased from 54.6% to 0%. As mentioned above, NOx desorption took place because NH3 required for the SCR reaction was lacking as adsorbed ammonia was

17

rapidly converted to nitrate. On the other hand, N2 selectivity of 1Pt/TiO2 was slightly increased from 8.8% to 12%, although the increase was not statistically significant. N2 yield of 1Pt/TiO2 showed a tendency to decrease gradually when ammonia injection concentration was increased. However, that of 1Ru/TiO2 showed a tendency to increase when the concentration of ammonia injected was increased. Its N2 yield increased up to 95.5%. These results confirmed that N2 yield of 1Ru/TiO2 could show a large difference depending on the concentration of NH3 injected. Figure 13 shows reaction characteristics of the surface with different NH3 concentrations through DRIFT analysis under actual reaction conditions where NH3 and O2 are simultaneously injected. Results were obtained after 30 min. When 200 ppm NH3 was injected, a large intensity of bridge nitrate (1603, 1580 cm-1) was generated on the surface of 1Ru/TiO2, indicating that most of its surface was converted to nitrate. Therefore, since the intensity of L acid required for the SCR reaction was very low, the amount of NH3 used as a reducing agent might be small and the N2 yield was low as shown in Figure 12. The peak of L acid point was confirmed at 3372, 3263, 3241, 3153, 1628 cm-1 when 2,000 ppm of NH3 was injected and the peak of bidentate nitrate could be confirmed at 1594 and 1235 cm-1. These results suggest that NO can be converted to N2 by reacting with NH3. Therefore, the rate of converting 1Ru/TiO2 into nitrate in the catalyst at the time of NH3 adsorption has a great influence on N2 yield.

3.5. NH3 oxidation mechanism on Ru/TiO2 NH3 oxidation mechanism of 1Ru/TiO2 is proposed as shown in Figure 14. Ammonia was mainly adsorbed to L acid site and B acid site in 1Ru/TiO2, although some of them were

18

converted to intermediate species NH2 by dissociation. In addition, some of the adsorbed NH3 could react with active metal oxygen to form nitrate at a temperature under 200 °C. NH3(ad) → NH2(ad) + H+

(4)

3Ru-O + NH3(ad) → Ru-NO3-(ad) + H+

(5)

At a temperature when reaction activity of the catalyst takes place, nitrate can be decomposed and desorbed with NO or NO2 in the catalyst. However, the SCR reaction in which NH3 and NO react to form N2 is in progress if there is extra NH3 adsorbed to the catalyst. Therefore, the rate of converting NH3 to nitrate in the NH3-SCO reaction of 1Ru/TiO2 is judged to be a very important factor in N2 selectivity. The SCR reaction can be confirmed by using TiO2 surface oxygen. When there is no more reaction due to deficiency of oxygen in the catalyst, active metal oxygen and TiO2 surface oxygen used for the reaction will be restored by meteorological oxygen and the reaction is maintained. Ru-NO3-(ad) → Ru-NO(ad) or Ru-NO2(ad)

(6)

Ru-NO(ad) + NH3(ad) + Ti-O → N2(g) + H2O(g)

(7)

Ru-□ + O2(g) → Ru-O

(8)

Ti-□ + O2(g) → Ti-O

(9)

4. Conclusions In this study, NH3-SCO reaction characteristics and mechanism of 1Ru/TiO2 were compared with those of 1Pt/TiO2 at temperature below 300 °C. RuO2 on the surface of 1Ru/TiO2 has a very excellent ability to adsorb NH3 and O2 molecules in NH3-SCO reaction. In addition, RuO2 has excellent ability to convert NH3 to NO3- species. These characteristics are influenced by active metal oxygen. Moreover, NO3- was decomposed in to NO or NO2 and

19

reacted with adsorbed NH3 and O2 to produce N2 and H2O(I-SCR). But, 1Pt/TiO2 was generated N2O by relying on NH3 direct oxidation. As a result, 1Ru/TiO2 is superior to 1Pt/TiO2 in NH3 conversion rate and N2 yield. However, N2 yield of 1Ru/TiO2 can be greatly affected by the concentration of NH3 injected. These results are related to the rate at which adsorbed NH3 is converted to NO. To increase N2 yield, studies that control the rate of conversion to NO depending on NH3 concentration injected and temperature range are needed.

5. Acknowledgements This research was supported by the Center for Environmentally Friendly Vehicle (CEFV) as a Global-Top Project of the Ministry of Environment (MOE), Republic of Korea.

20

References [1] Y. Pan, S. Tian, D. Liu, Y. Fang, X. Zhu, Q. Zhang, B. Zheng, G. Michalski, Y. Wang, Environ. Sci. Technol. 50 (2016) 8049-8056 [2] Q. Zhang, H. Wang, P. Ning, Z. Song, X. Liu, Y. Duan, Appl. Sur. Sci., 419 (2017) 733743 [3] B. Wang, Y. Wu, X. Wang, Z. Chen, C. He, Sens. Actuator. B-Chem. 190 (2014) 157164 [4] Y. Wu, B. Gu, J.W. Erisman, S. Reis, Y. Fang, X. Lu, X. Zhang, Environ. Pollut. 218 (2016) 86-94 [5] L. Chmielarz, M. Jabtonska, RSC Adv. 5 (2015) 43408-43431 [6] K. Gora-Marek, K. Brylewska, K. A. Tarach, M. Rutkowska, M. Jablonska, M. Choi, L. Chmielarz, Appl. Catal. B: Enviorn. 179 (2015) 589-598 [7] C. Chen, Y. Cao, S. Liu, J. Chen, W. Jia, Appl. Sur. Sci, 480 (2019) 537-547 [8] L. Lietti, G. Ramis, G. Busca, F. Bregani, P. Forzatti, Catal. Today 61 (2000) 187-195 [9] K. Shojaee, B. S. Haynes, A. Montoya, Proc. Combust. Inst., 36 (2017) 4365-4373 [10] Z. Qu, R. Fan, Z. Wang, H. Wang, L. Miao, Appl. Sur. Sci., 351 (2015) 573-579 [11] S.M. Lee, H.H. Lee, S.C. Hong, Appl. Catal. A-Gen. 470 (2014) 189-198 [12] G. Olofsson, A. Hinz, A. Andersson, Chem. Eng. Sci. 59 (2004) 4113-4123 [13] H. Ma, W. F. Schneider, ACS catal., 9.3 (2019) 2407-2414

21

[14] L. Zhang, H. He, J. Catal. 268 (2009) 18-25 [15] L. Gang, B.G. Anderson, J. van Grondelle, R.A. van Santen, J. Catal. 199 (2001) 107114 [16] H. Wang, P. Ning, Q. Zhang, X. Liu, T. Zhang, J. Fan, J. Wang, K. Long, Appl. Catal. A: Gen., 561 (2018) 158-167 [17] L. Zhang, C. Zhang, H. He, J. Catal. 261 (2009) 101-109 [18] S. Shrestha, M. P. Harold, K. Kamasamudram, A. Kumar, L. Olsson, K. Leistner, Catal. Today, 267 (2016) 130-144 [19] S. Shrestha, M. P. Harold, K. Kamasamudram, A. Yezerets, Catal. Today 231 (2014) 105-115 [20] M. Sun, S. Wang, Y. Li, H. Xu, Y. Chen, Appl. Surf. Sci. 402 (2017) 323-329 [21] D.W. Goodman, C.H.F Peden, M.S. Chen, Surf. Sci. 19 (2007) 124-126 [22] S. Scire, R. Fiorenza, A. Gulino, A. Cristaldi, P. M. Riccobene, Appl. Catal. A: Gen., 520 (2016) 82-91 [23] N. Lopez, J.G. Sequra, R.P. Marin, J.P. Ramirez, J. Catal. 255 (2008) 29-39 [24] E.V. Kondratenko, A.P. Amrute, M.M. Pohl, N. Steinfeldt, C. Mondelli, J.P. Ramirez, Catal. Sci. Technol. 3 (2013) 2555-2558 [25] J. P. Ramirez, N. Lopez, E.V. Kondratenko, J. Phys. Chem. C 114 (2010) 16660-16668 [26] Y. Wang, K. Jacobi, W.D. Schone, G. Ertl., J. Phys. Chem. B 109 (2005) 7883-7893 [27] C.C. Wang, Y.J. Yang, J.C. Jiang, J. Phys. Chem. C 113 (2009) 2816-2821

22

[28] C.C. Wang, Y.J. Yang, J.C. Jiang, D.S. Tsai, H.M. Hsieh, J. Phys. Chem. C 113 (2009 ) 17411-17417 [29] X. Cui, L. Chen, Y. Wang, H. Chen, W. Zhao, Y. Li, J. Shi, ACS Catal. 4 (2014) 21952206 [30] M. Jablonska, M. Nocun, K. Golabek, R. Palkovits, Appl. Sur. Sci., 423 (2017) 498-508 [31] M. jabionska, R. Palkovits, Appl. Catal. B: Environ., 181 (2016) 332-351 [32] H. Zhu, Y. Li, X. Zheng, Appl. Catal. A: Gen., 571 (2019) 89-95 [33] G. Grzybek, P. Stelmachowski, S. Gudyka, P. Indyka, Z. Sojka, N. Guillen-Hurtado, V. Rico-Perez, A. Bueno-Lopez, A. Kotarba, Appl. Catal. B: Environ., 180 (2016) 622-629 [34] S. Parres-Esclapez, M.J. Illan-Gomez, C. Salinas-Martinez de Lecea, A. Bueno-Lopez, Int. J. Greenh. Gas Control., 11 (2012) 251-261 [35] M. Konsolakis, F. Aligizou, G. Goula, I.V. Yentekakis, Chem. Eng. J. 230 (2013) 286295 [36] M.T. Uddin, Y. Nicolas, C. Olivier, T. Toupance, M. M. Muller, H.J. Kleebe, K. Rachut, J. Ziegler, A. Klein, W. Jaegermann, J. Phys. Chem. C, 117 (2013) 22098-22110 [37] T. Omotoso, S. Boonyasuwat, S.P. Crossley, Green Chem. 16 (2014) 645-652 [38] J.M.G. Carballo, E. Finocchio, S. Garcia, S. Rojas, M. Ojeda, G. Busca, J.L.G. Fierro, Catal. Sci. Technol 1 (2011) 1013-1023 [39] K. Seki, Catal, Surv. Asia, 14 (2010) 168-175

23

[40] N. Aranda-Perez, M.P. Ruiz, J. Echanve, J. Faria, Appl. Catal. A-Gen. 531 (2017) 106118 [41] G. Zhou, R. Dou, H. Bi, S. Xie, Y. Pei, K. Fan, M. Qiao, B.Sun, B. Zong, J. Catal. 332 (2015) 119-126 [42] S.I. Shah, S. Hong, T.S. Rahman, J. Phys. Chem. C 118 (2014) 5226-5238 [43] S. Hong, A. Karim, T.S. Rahman, K. Jacobi, G. Ertl, J. Catal. 276 (2010) 371-381 [44] S.M. Lee, S.C. Hong, Appl. Catal. B-Environ. 163 (2015) 30-39 [45] F. Wang, J. Ma, G. He, M. Chen, C. Zhang, H. He, ACS Catal., 8 (2018) 2670-2682 [46] W.K. Offermans, A.P.J. Jansen, R.A. van Santen, Surf. Sci. 600 (2006) 1714-1734 [47] M. Lin, B. An, N. Niimi, Y. Jikihara, T. Nakayama, T. Honma, T. Takei, T. Shishido, T. Ishida, M. Haruta, T. Murayama, ACS Catal., 9.3, (2019) 1753-1756 [48] G. Novell-Leruth, J.M. Ricart, J. Perez-Ramirez, J. Phys. Chem. C 112 (2008) 1355413562 [49] W. Chen, Z. Qu, W. Huang, X. Hu, N. Yan, Fuel, 166 (2016) 179-187 [50] C. Wang, J. Wang, J. Wang, T. Yu, M. Shen, W. Wang, W. Li, Appl. Catal. B: Environ., 204 (2017) 239-249 [51] Z. Liu, S. Zhang, J. Li, J. Zhu, L. Ma, Appl. Catal. B: Environ., 158-159 (2014) 11-19 [52] M. Jablonska, A. Krol, E. Kukulska-Zajac, K. Tarach, L. Chmielarz, K. Gora-Marek, J, catal. 316 (2014) 36-46 [53] G. Qi, J.E. Gatt, R.T. Yang, J. Catal. 226 (2004) 120-128

24

[54] P.S. Dimick, J.L. Kross, E.G. Roberts, R.G. Herman, H.G. Stenger, C.E. Lyman, Appl. Catal. B-Environ. 89 (2009) 1-11 [55] A. Bourance, O. Dulaurent, S. Salasc, C. Sarda, C. Bouly, D. Bianchi, J. Catal. 204 (2001) 77-88 [56] A. Filtschew, D. Stranz, C. Hess, Phys. Chem. Chem. Phys. 15 (2013) 9066-9069 [57] Z. Zhang, L. Chen, Z. Li, P. Li, F. Yuan, X. Niu, Y. Zhu, Catal. Sci. Technol., 6 (2016) 7151-7162 [58] J. Xu, X. Su, H. Duan, Q. Lin, X. Liu, X. Pan, G. Pei, H. Geng, Y. Huang, T. Zhang, J. Catal. 333 (2016) 227-237 [59] D. Li, R. Li, M. Lu, X. Lin, Y. Zhan, L. Jiang, Appl. Catal. B-Environ. 200 (2017) 566577 [60] S.S. Kim, H.H. Lee, S.C. Hong, Appl. Catal. B-Environ. 119(2012) 100-108 [61] G. J. Kim, D. W. Kwon, S.C. Hong, Jour. Phys. Chem. C, 120 (2016) 17996-18004 [62] H. Wang, Q. Zhang, T. Zhang, J. Wang, G. Wei, M. Liu, P. Ning, Appl. Sur. Sci., 485 (2019) 81-91

25

Figure Captions Figure 1. Effect of active metal on NH3-SCO oxidation over 1Ru/TiO2 and 1Pt/TiO2 catalysts. (a) NH3 conversion / N2 yield, (b) Outlet NOx concentration. Reaction condition: NH3: 200 ppm; O2: 10 vol.%; H2O, 6 vol.%; S.V.: 60,000 hr-1. Figure 2. XRD patterns of 1Ru/TiO2 and 1Pt/TiO2 catalysts. Figure 3. FETEM and STEM images of 1Ru/TiO2 and 1Pt/TiO2 catalysts with energy dispersive X-Ray (EDX) of highlighted regions showing peaks of Ti, Ru, and Pt. (a),(c) Corresponded images of 1Ru/TiO2. (b),(d) Corresponded images of 1Pt/TiO2. Figure 4. XPS spectra of Ti 2p (a), Pt 4f (b), Ru 3d (c), O 1s (d) of 1Ru/TiO2 and 1Pt/TiO2 catalyst. Figure 5. NH3-TPD profiles of 1Ru/TiO2 and 1Pt/TiO2 catalysts. Figure 6. NH3-TPO profiles of 1Ru/TiO2 and 1Pt/TiO2 catalysts. (a) NH3, (b) NOx (NO, NO2, N2O). Figure 7. In situ DRIFT spectra of (a) 1Ru/TiO2 and (b) 1Pt/TiO2 at 300℃. Surfaces are purged under 10 vol.% O2/N2 mixture at 400℃ for 30 min. Surfaces are saturated with NH3 (200 ppm) for 30 min. Figure 8. In situ DRIFT spectra of (a) 1Ru/TiO2 and (b) 1Pt/TiO2 at 300℃. Surfaces are saturated under NH3 (200 ppm) at 300℃ for 30 min. Surfaces are then purged under 10 vol.% O2/N2 mixture for 30 min. Figure 9. In situ DRIFT spectra of (a) 1Ru/TiO2 and (b) 1Pt/TiO2 at 180℃. Surfaces are saturated under NH3 (200 ppm) at 180℃ for 30 min. They are then purged under 10 vol.% O2/N2 mixture for 30 min. Figure 10. H2-TPR profiles of 1Ru/TiO2 and 1Pt/TiO2 catalysts.

26

Figure 11. Amounts of surface adsorbed oxygen (Oβ) of 1Ru/TiO2 and 1Pt/TiO2 catalysts. Figure 12. Effect of NH3 concentration on SCO reaction over 1Ru/TiO2 and 1Pt/TiO2 catalysts. Reaction conditions: NH3, 30~2,000ppm; O2, 10 vol.%; H2O, 6 vol.%; S.V.: 60,000hr-1. Figure 13. In situ DRIFT spectra of 1Ru/TiO2 and 1Pt/TiO2 at 300℃. Surfaces are purged under 10 vol.% O2/N2 mixture at 400℃ for 30 min. Surfaces are then saturated with NH3 (200 or 2,000ppm) + O2 (10 vol.%) for 30 min. Figure 14. Schematic representation of selective catalytic oxidation of NH3 to nitrogen over Ru/TiO2 catalyst (NH3-SCO)

Table Captions Table 1. Physical properties of TiO2, 1Ru/TiO2 and 1Pt/TiO2 catalysts. Table 2. The information of binding energy and molar fraction of Oα and Oβ over 1Ru/TiO2 and 1Pt/TiO2 catalysts.

27

Graphical abstract

28

Highlights

 RuO2 has excellent adsorption capacity of ammonia molecules and oxygen.  The major mechanism of NH3-SCO reaction over Ru/TiO2 was internal-selective catalytic reduction(I-SCR)  It is important to control that NH3 adsorbed convert properly to NO over Ru/TiO2

29

Table 1. The physical properties of TiO2, 1Ru/TiO2 and 1Pt/TiO2 catalysts.

Catalyst

Active metal contenta (%)

SBETb

Total pore volumeb

Mean pore sizeb

(m2/g)

(cm3/g)

(nm)

TiO2

-

280.67

0.4491

7.4074

1Ru/TiO2

1.03

120.91

0.4275

14.989

1Pt/TiO2

1.04

114.08

0.4255

14.076

(a) XRF analysis (b) BET analysis

30

Table 2. The information of binding energy and molar fraction of Oα and Oβ over 1Ru/TiO2 and 1Pt/TiO2 catalysts.

Binding energy (eV)

Molar percent of O 1s (%)

Catalyst Oα

Oβ

Oα/OTa

Oβ/OTa

1Ru/TiO2

529.5

532

43.36

56.64

1Pt/TiO2

529.5

532

60.62

39.38

(a) Ototal = (Oα + Oβ)

31

Fig. 1.

32

Fig. 2.

33

Fig. 3.

34

35

Fig. 4.

36

Fig. 5.

37

Fig. 6.

38

Fig. 7.

39

Fig. 8.

40

Fig. 9.

41

Fig. 10.

42

Fig. 11.

43

Fig. 12.

44

Fig. 13.

45

Fig. 14.

46

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:

47