Thermocatalytic Decomposition of Dimethyl Methylphosphonate Based on CeO₂ Catalysts with Different Morphologies

Abstract

The catalytic performances of the catalysts and decomposition mechanisms of dimethyl methylphosphonate (DMMP), a commonly used nerve agent simulant, are well understood based on previous studies. However, the effects of the morphology of the catalyst on DMMP decomposition performance and mechanisms remain unexplored. Thus, in this work, experimental studies were conducted on the thermocatalytic decomposition of DMMP on CeO₂ nanomaterials with different morphologies, e.g., irregular nanoparticles, nanorods, and nanocubes. From the performance evaluation, CeO₂ nanorods exhibited higher DMMP thermocatalytic decomposition performance as compared to irregular nanoparticles and nanocubes. The primary reaction pathways were the same on all three morphologies of materials, according to in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) study, whereas side reaction paths showed variable behaviors. According to the catalytic reaction mechanism study, the surface lattice oxygen played a vital role in the thermocatalytic decomposition of DMMP and the accumulation of phosphates, carbonates, and formates were the main factors for deactivation of the catalyst. The behavior of CeO₂ catalyst with different morphologies in the thermocatalytic decomposition of DMMP was revealed in this work, and this will be useful for the future design of high-performing catalysts for the efficient degradation of chemical toxicant.

1. Introduction

Sarin is a nerve agent that caused massive casualties during World War I and II [1]. Since the Chemical Weapon Convention entered into force, the safe disposal of chemical warfare agents (CWAs) stockpiles has become a major challenge. In an attempt to dispose of the CWAs, various methods, e.g., incineration [2], hydrolysis [3,4,5], chemical reaction [6,7,8,9], photolysis [10,11,12], etc., have been used. Among the commonly used disposal methods, thermocatalytic decomposition is one of the more efficient and low-cost approaches. During the thermocatalytic decomposition of CWAs, a catalyst must be used to enhance the removal efficiency. Thus, from the perspective of CWAs disposal, the lifetime of a catalyst is one of the most important parameters used in the evaluation of a catalyst’s performance. It is customary to refer to the protection time as the catalyst’s lifetime when 100% conversion is accomplished [13,14].

It is worth noting that in laboratory research, dimethyl methylphosphonate (DMMP) is often used as a simulation of sarin, it exhibits low toxicity while possessing a similar structure with sarin. Among the various catalysts, metal oxides are commonly applied as catalysts for the thermocatalytic decomposition of DMMP, whereby they cleave the P−OCH₃ and P−CH₃ bonds [15,16,17]. Generally, P−OCH₃ could be easily broken by single valent metal oxides, such as Al₂O₃ [18], MgO [19], ZnO [20], Y₂O₃ [21], etc. For instance, the research on the thermocatalytic decomposition of DMMP on Al₂O₃ [18] has shown that one of the methoxy bonds in DMMP was removed at approximately 200 °C, while the other methoxy bond was removed at 300 °C. However, the P−CH₃ bond remained intact on the surface of Al₂O₃ even at temperatures above 400 °C. Such an observation was expected since the P−CH₃ bond is typically resistant to cleavage during thermocatalytic decomposition, which also happened for other single valent metal oxide catalysts, such as magnesium oxide and lanthanum oxide. As such, based on the literature, utilizing single valent metal oxide catalysts in the thermocatalytic decomposition process is inadequate to achieve full decomposition of DMMP. In contrast, it was recently shown that multivalent metal oxide catalysts are more favorable in the decomposition of DMMP, especially in the cleavage of P−CH₃ bond. Iron oxides have the unique ability to activate the P−CH₃ bond at room temperature. This phenomenon was attributed to the existence of a low energy pathway for the oxidative cleavage of P−CH₃ bond due to Fe(III)/Fe(II) redox couple [16,22]. Similarly, multivalent manganese oxide [14] and copper oxide catalysts [15,23] show scission of P−OCH₃ and P−CH₃ bonds.

Being an important rare-earth metal oxide material, multivalent CeO₂ possesses an outstanding oxygen storage/release capacity with reversible redox variations between Ce³⁺/Ce⁴⁺, as well as the formation and removal of oxygen vacancies [24], which makes it appealing in various applications such as catalysis, sensing, electrochemistry, optics, and biomedical applications [25,26,27,28,29]. Mitchell et al. studied the room temperature decomposition of DMMP based on a series of alumina-supported cerium and iron oxide catalysts, and the optimal impregnation ratio of these two metals was obtained at the maximum decomposition degree of DMMP [30]. Based on further research, the incomplete decomposition mechanism of DMMP at room temperature based on alumina-supported cerium catalyst was revealed [31]. Chen et al. reported DMMP decomposition on fully oxidized and partially reduced ceria films and reaction mechanism for DMMP decomposition on CeO₂ [32]. Maija et al. calculated the DMMP adsorption and decomposition on (110) and (111) CeO₂ surfaces with density functional theory modeling. Two possible reaction pathways were proposed [33]. The current studies on the catalytic decomposition of DMMP by CeO₂ are focused on elucidating the catalytic performance and mechanism. Despite the significant advancements in the development of CeO₂ in the thermocatalytic decomposition of DMMP, it is worth noting that the morphology, size, and surface properties of CeO₂ nanomaterials have an important influence on their catalytic performances [34]. For instance, Laura et al. investigated the complete oxidation of toluene based on CeO₂ nanomaterials with different morphologies [35]. According to their results, CeO₂ nanorods exhibited a higher concentration of defects associated with their exposed crystalline planes, which led to their high oxidative catalytic activity. Moreover, CeO₂ nanoparticles have shown maximum catalytic decomposition activity for naphthalene [36], which is related to the high Ce³⁺/Ce⁴⁺ ratio and the oxygen vacancy on the surface of CeO₂. When the particle diameter is less than the Bohr radius, the quantum confinement leads to an increase in the effective band gap of the material by decreasing the particle size [37,38]. Currently, there is still a lack of research on the thermal catalytic decomposition of sarin simulant, i.e., DMMP, by CeO₂ with different morphologies and sizes. As such, it is crucial to have a systematic and deep exploration into the effects of physicochemical and surface properties on the catalytic performance of CeO₂ with different morphologies so as to design CeO₂ catalysts with high thermocatalytic DMMP decomposition performance.

Herein, CeO₂ with different morphologies and sizes, e.g., irregular nanoparticles, nanorods, and nanocubes, were prepared via a hydrothermal synthesis method by adjusting the hydrothermal reaction temperature and the concentration of NaOH. The effects of the different physicochemical and surface properties of CeO₂ nanomaterials with different morphologies and sizes on their thermocatalytic decomposition of DMMP were investigated in this work. Finally, the DMMP decomposition mechanism based on CeO₂ nanomaterials was studied and proposed.

2. Materials and Methods

2.1. Synthesis of CeO₂ Nanomaterials

Ce(NO₃)₃·6H₂O (99.9%) and NaOH (99.9%) were purchased from Macklin, Shanghai, China. All the chemicals and reagents were used without further purification. In a typical procedure, solution A was prepared by dissolving 2.60 g Ce(NO₃)₃·6H₂O in 10 mL deionized water. Then, 24 g NaOH was dissolved in 50 mL deionized water to prepare solution B (6 mol/L NaOH solution). After which, solution A was dripped into solution B with continuous stirring at room temperature for 30 min. The mixture was placed in a 100 mL Teflon-lined autoclave and it was hydrothermally treated for 24 h at 100 °C (which resulted in the sample with rod-shaped morphology) and another batch at 180 °C (which resulted in cube-shaped morphology). After cooling down to room temperature naturally, the precipitates were centrifuged and repeatedly rinsed in ethanol and deionized water. The precipitates were then dried at 100 °C for eight hours and calcined for four hours at 300 °C. After the calcination process, CeO₂ nanorods (CeO₂nr) and CeO₂ nanocubes (CeO₂nc) were obtained. CeO₂ nanomaterials with different sizes were prepared by changing the concentration of NaOH. In this work, two variations of CeO₂ nanorods were prepared with 6 mol/L and 12 mol/L NaOH, and they were denoted as 6MCeO₂nr and 12MCeO₂nr, respectively. Two variations of CeO₂ nanocubes were prepared with 6 mol/L and 12 mol/L NaOH, and they were denoted as 6MCeO₂nc and 12MCeO₂nc, respectively. One variation of irregular nanoparticle was prepared with solution B (2 mol/L NaOH solution). The mixture was placed in a 100 mL Teflon-lined autoclave and it was hydrothermally treated for 24 h at 100 °C. After cooling down to room temperature naturally, the precipitates were centrifuged and repeatedly rinsed in ethanol and deionized water. The precipitates were then dried at 100 °C for eight hours and calcined for four hours at 300 °C. It was denoted as 2MCeO₂np.

2.2. Characterization of CeO₂ Nanomaterials

In order to record the X-ray diffraction (XRD) patterns, a SmartLab (Rigaku, Saitama, Japan) with Cu Kα radiation (λ = 1.5418) was used. Utilizing a Nova 4200e (Quantachrome Instruments, Boynton Beach, FL, USA) and the Brunauer–Emmett–Teller (BET) model, the sample’s specific surface area was determined at 77 K in nitrogen. The sizes and morphologies of all samples were examined by a JEM 2100 high-resolution transmission electron microscope (JEOL, Tokyo, Japan). H₂−temperature-programmed reduction (H₂−TPR) was carried out using an AutoChem II 2920 (Micromeritics, Norcross, GA, USA). The samples (100 mg) were heated 3 h at 300 °C with He flowing at 50 mL/min, and cooled to 50 °C. In order to execute H₂−TPR, the sample was heated at a rate of 10 °C/min from room temperature to 900 °C in the presence of 10% H₂/Ar mixture (40 mL/min). X-ray photoelectron spectroscopy (XPS) studies were carried out utilizing an Al-K radiation-powered Thermo Fisher ESCALAB 250Xi (Thermo Fisher Scientific, Waltham, MA, USA) equipment. Based on the carbon deposit C (1s) binding energy (BE) of 284.8 eV, the measurement was calibrated. The Avantage program version 5.976 (Thermo Fisher Scientific, USA) was used to evaluate XPS data. All recorded peaks of the corrected spectra were all fitted with a Gaussian–Lorentzian shape function. Raman spectroscopy was conducted using a HORIBA LabRam HR Evolution (Longjumeau, France). A Thermo Scientific ICS-5000 was used for ion chromatography (Thermo Fisher Scientific, USA). Aqua regia and hydrofluoric acid were used to dissolve the used catalyst. The colorless gelatinous materials on the wall at the end of the reaction tube were collected using 0.2 M NaOH solution. The eluent used was 15 mM KOH buffer. The in situ DRIFTS was conducted using a Nicolet iS 50 (Thermo Fisher Scientific, Waltham, MA, USA) during the reaction process to monitor the thermocatalytic decomposition of DMMP. During the reaction, in situ DRIFTS was conducted with 64 scans at a resolution of 8 cm⁻¹ (to minimize the collection time).

2.3. Evaluation of Thermocatalytic Decomposition Performance

The catalyst’s capability for thermocatalytic degradation was assessed using a fixed-bed reactor with an inner diameter of 4 mm. The DMMP steam was produced by passing compressed air through a bubbler containing DMMP at a flow rate of 50 mL/min. The temperature of the bubbler was 30 °C. The gas line was heated at 80 °C to prevent the condensation of DMMP vapor. The initial DMMP content into the fixed-bed reactor was 5.32 g/m³. The DMMP content in the reaction tail gas were detected by Gas chromatograph (Agilent 6890N, Agilent Technologies, Inc, chromatographic column DB−1701, Santa Clara, CA, USA) outfitted with FID detector. To verify the thermal stability of DMMP at 300 °C, the decomposition experiment of blank DMMP was carried out in an empty reactor. For this, 0.66 g CeO₂ nanocubes, 0.42 g CeO₂ irregular nanoparticles, and 0.42 g CeO₂ nanorods (20−40 mesh) were loaded onto the fixed bed which was heated at 300 °C and gave a gas hourly space velocity (GHSV) of 10,000 h⁻¹. Protection time was used to evaluate of the catalytic decomposition performance (duration needed to achieve DMMP conversion 100%). DMMP conversion rate can be defined based on Equation (1).

$$\mathrm{DMMP}\mathrm{conversion}=\left(1-\frac{C_{out}}{C_{in}}\right)\times 100\%$$

(1)

(C_in is the initial DMMP content into the fixed-bed reactor, C_out is the DMMP content in the reaction tail gas).

2.4. Thermocatalytic Decomposition Reaction Tail Gases Analysis

A CATLAB microreactor (HIDEN ANALYTICAL, Warrington, UK) was used to carry out a thermocatalytic decomposition micro-reaction. The reaction tube was filled with 0.05 g CeO₂ nanomaterials, then heated at 300 °C for 2 h under a helium atmosphere (with a flow rate of 100 mL/min). The sample was naturally cooled to room temperature under a helium flow and subsequently heated from room temperature to 300 °C at 10 °C/min. DMMP vapor is brought into the reaction tube with a mixture of 80% argon and 20% oxygen as the carrier gas. The carrier gas (50 mL/min) bubbled in a flask containing DMMP at 10 °C. Mass spectrometry (HIDEN ANALYTICAL, Warrington, UK) was used to monitor the reaction tail gas.

3. Results and Discussion

3.1. Characteristics of CeO₂ Nanomaterials

The crystallographic information of the as-prepared samples was determined using XRD, and the result is depicted in Figure 1. Based on the recorded XRD spectra for each sample, the detected peaks corresponded well with the characteristic peaks in the XRD spectrum of fluorite structure CeO₂ (JCPDS 34−0394, Space Group Fm3m), which indicates that CeO₂ with different morphologies were successfully prepared. By employing Scherrer’s equation, the crystallite sizes of 2MCeO₂np, 6MCeO₂nr, 12MCeO₂nr, 6MCeO₂nc, and 12MCeO₂nc could be calculated to be 6.86 nm, 7.50 nm, 9.61 nm, 39.56 nm, and 49.46 nm, respectively (

Table 1. Synthesis conditions and physicochemical characteristics of the as-prepared CeO₂ nanomaterials with different morphologies.

Catalyst	Temperature (°C)	CNaOH (mol/L)	Morphology	Mean Length (nm)	Crystallinity Size (nm)	SBET (m2/g)
2MCeO2np	100	2	Particles	5.02	6.86	128.89
6MCeO2nr	100	6	Rods	38.72	7.50	126.1
12MCeO2nr	100	12	Rods	133.83	9.61	102.19
6MCeO2nc	180	6	Cubes	18.78	39.56	31.13
12MCeO2nc	180	12	Cubes	57.95	49.46	14.609

). The specific surface areas (S_BET) of the as-prepared CeO₂ nanomaterials were obtained with N₂ adsorption-desorption measurements. In Table 1, irregular CeO₂ nanoparticles and nanorods possessed significantly larger S_BET as compared to the CeO₂ nanocubes. Furthermore, from the BET results, the S_BET of CeO₂ nanomaterials with the same morphology increased with the reduction in size.

Figure 2 shows the TEM images of the five as-prepared CeO₂ nanomaterials prepared via the hydrothermal method. Based on the results, it can be observed that the hydrothermal temperature and NaOH concentration used in the preparation process could result in significant differences in the morphologies of the samples. The formation mechanism of the CeO₂ nanomaterials with different morphologies can be attributed to the well-known dissolution/recrystallization process [39]. Figure 2e shows the TEM image of the irregular CeO₂ nanoparticles prepared with 2 M NaOH at 100 °C. It can be observed that as the concentration of NaOH increased beyond 2 M NaOH (Figure 2a,b), the as-obtained CeO₂ gradually exhibited a change in its morphology into nanorods due to the anisotropic growth of Ce(OH)₃ nuclei along the (110) direction. However, as the hydrothermal reaction temperature increased to 180 °C, Ce(OH)₃ would become unstable and it could oxidize to form CeO₂ nanocubes (Figure 2c,d). The CeO₂ nanocubes grew along each edge length direction with increasing NaOH concentrations. According to the TEM results, the average lengths of 6MCeO₂nr, 12MCeO₂nr, 6MCeO₂nc, 12MCeO₂nc, and 2MCeO₂np were 38.72 nm, 133.83 nm, 18.78 nm, 57.95 nm, and 5.02 nm, respectively, and this result is summarized in Figure 2f and Table 1.

H₂−TPR was used to examine the reducibility of the surface oxygen and bulk phase oxygen in the CeO₂ nanomaterial (Figure 3). The low temperature reduction peaks located between 300 °C and 650 °C correspond to the reduction of surface oxygen species, while the high temperature reduction peak located above 650 °C relates to the reduction of bulk phase oxygen species [40]. The surface oxygen reduction temperatures of the irregular nanoparticles and nanorods were significantly lower than that of the nanocubes. Such a result indicates that the irregular nanoparticles and nanorods possessed better activity at low temperatures as compared to the nanocubes. Furthermore, a very small amount of H₂ was consumed during the surface oxygen reduction for the nanocubes, which may be due to the extremely small surface areas exhibited by the nanocubes that resulted in the smaller amount of surface adsorbed oxygen. Interestingly, 12MCeO₂nr and 12MCeO₂nc possessed more active bulk oxygen due to their lower bulk oxygen reduction temperature than the same morphologies of 6MCeO₂nr and 6MCeO₂nc samples, respectively, which makes them more promising as DMMP thermocatalytic catalysts. This is confirmed later in the mechanism discussion section that DMMP is decomposed on CeO₂ mainly by the recycling between consuming surface oxygen and replenishment of bulk oxygen and the external oxygen. Therefore, the more active the bulk oxygen is, the more beneficial it is to the transfer of oxygen on CeO₂.

Oxidation state of Ce and the nature of the surface oxygen species were characterized by X-ray photoelectron spectroscopy (Figure 4). The Ce 3d XPS spectra of the as-prepared CeO₂ nanomaterials consisted of two main peaks (V₀, V) and three satellite peaks (V`, V``, V```) that can be ascribed to Ce3d<sub>5/2</sub>, and the remaining two main peaks (U<sub>0</sub>, U) and three satellite peaks (U`, U``, U```) that can be ascribed to Ce3d<sub>3/2</sub>. The binding energy peaks of V (882.14 eV), V`` (888.51 eV), V``` (898.05 eV), U (900.63 eV), U`` (907.32 eV), and U``` (916.45 eV) are attributed to Ce⁴⁺, whereas Ce³⁺ is linked to the photoelectron excitations of V₀ (880.2 eV), V` (884.7 eV), U<sub>0</sub> (898.65 eV), and U` (903.09 eV) [35]. The presence of Ce³⁺ indicates the presence of non-stoichiometric CeO_2−X on the surface of CeO₂ nanomaterials. Oxygen vacancies could be formed during the transformation of Ce⁴⁺ into Ce³⁺ by the following defect reaction: O_O^X↔V_O`` + 2e + $\frac{1}{2}$O(g), where O<sub>O</sub><sup>X</sup>, V<sub>O</sub>``, and e are oxide ions in the lattice, doubly charged oxygen vacancies, and electrons in the conduction band made up of Ce 4f energy states, respectively [41]. It indicates the presence of oxygen vacancies on the surface of CeO₂ nanomaterials. Oxygen vacancies can play a critical role in the redox mechanism and oxygen migration toward the surface of CeO₂ nanomaterials. According to the previously reported methods [35,42], Ce 3d spectra of the samples were fitted to calculate the content of Ce³⁺ (

Table 2. XPS, Raman characterizations, and catalytic performance of various CeO₂ nanomaterials.

Catalyst	XPS			Raman		ICP	Protection Time (min)	MSTC (gDMMP/gcat)	SSTC (gDMMP/m2)
	Ce3+/(Ce3+ + Ce4+)	Oα/Oall	Oβ/Oall	FWHM460	I600/I460	Ce(wt%)
2MCeO2np	19.64	58.94	41.06	27.27	0.0370	77.9	350	0.222	0.00072
6MCeO2nr	25.32	51.08	48.92	33.40	0.0388	73.3	420	0.266	0.00089
12MCeO2nr	25.15	62.74	37.26	41.26	0.0461	77.3	490	0.310	0.00128
6MCeO2nc	19.04	70.10	29.90	12.41	0.0118	72.2	210	0.085	0.00179
12MCeO2nc	20.82	60.70	39.30	16.55	0.0132	72.3	378	0.152	0.00688

). Based on the fitted result, the content of Ce³⁺ in nanorods was higher than those in the irregular nanoparticles and nanocubes. Such a result may be due to the different oxygen vacancy formation energy at different exposed crystal planes for CeO₂ with different morphologies [43]. The O1s spectra of the as-prepared CeO₂ nanomaterials could be deconvoluted into two different peaks [44] as shown in Figure 5. The binding energy peak located at 528.5 eV–529.5 eV (labeled as Oα) was attributed to superficial lattice oxygen. The peak located at 530 eV–531.5 eV (labeled as Oβ) is assigned to oxygen vacancies, surface adsorbed O²⁻/O⁻, surface adsorbed O₂, hydroxyl groups, and carbonates. The ratios of Oα and Oβ were calculated based on the area of the deconvoluted peaks. As shown in Table 2, the Oα/O_all and Oβ/O_all of all five samples were different, which may be due to the various surface structures of CeO₂ materials.

To investigate the presence of oxygen vacancy, Raman spectroscopy was performed for all samples. As shown in Figure 6, a strong band located around 460 cm⁻¹ could be detected, which can be attributed to the symmetric stretch F_2g mode of CeO₂. Furthermore, the broad peak located at about 600 cm⁻¹ (inset of Figure 6) was attributed to the defect-induced (D) mode due to the oxygen vacancy [35,45]. The intensity ratio of the two peaks located at 600 cm⁻¹ and 460 cm⁻¹ (I₆₀₀/I₄₆₀) can be used to quantify the relative concentration of oxygen vacancy (Table 2). In Table 2, CeO₂ nanorods exhibited the highest oxygen vacancy concentration among all three morphologies. The content of Ce in different CeO₂ nanomaterials was obtained by ICP-OES as shown in Table 1. This result indicates that the content of Ce in each CeO₂ nanomaterial is less than the theoretical calculation of Ce. The variation trend of the calculated value obtained after multiplying the ratio of Ce³⁺ by the content of Ce was consistent with that of I₆₀₀/I₄₆₀, although the distribution of Ce elements on the surface may be different from the ICP data.

3.2. Catalytic Performance Testing

In order to evaluate the performance of a catalyst, the protection time is a crucial fac-tor [13]. Figure 7 shows the protection times of various CeO₂ catalysts, and no distinct regularity in the protection time for samples with different morphologies was observed. This result may be due to the difference in the filling weight for CeO₂ nanomaterials with different morphologies to achieve the same bed volume. Therefore, in order to avoid the influence of catalyst loading quantity on the protection time, the mass specific treatment capacity (MSTC) of the catalyst towards DMMP was calculated and shown in Table 2. Based on the result, the MSTCs of catalysts can be arranged in the following order; 12MCeO₂nr (0.310 g_DMMP/g_cat) > 6MCeO₂nr (0.266 g_DMMP/g_cat) > 2MCeO₂np (0.222 g_DMMP/g_cat) > 12MCeO₂nc (0.152 g_DMMP/g_cat) > 6MCeO₂nc (0.085 g_DMMP/g_cat). As such, according to the trend, CeO₂ nanorods exhibited superior MSTC as compared to the irregular nanoparticles and nanocubes. Furthermore, the surface area specific treatment capacity (SSTC) of the catalyst towards DMMP was also calculated and the corresponding results are summarized in Table 2. Interestingly, the SSTC of CeO₂ nanocubes was higher than those of CeO₂ nanorods and irregular nanoparticles. This result is due to the extremely low specific surface area of CeO₂ nanocubes as compared to the other two morphologies. When comparing CeO₂ nanomaterials with the same morphology, 12MCeO₂nr and 12MCeO₂nc with larger crystal sizes possessed larger MSTC and SSTC. Thus, based on these results, the morphology and crystal size of CeO₂ can significantly affect its catalytic performance.

Table 3. Protection time on various catalysts.

Catalyst	Reaction Condition	DMMP Concentration	Protection Time	Reference
0.5% Pt-Al2O3	299 °C, flow rate 8.85 L/min	3.5 g/m3	8 h	Graven et al. [46]
Cu2-HA	400 °C, flow rate 100 mL/min	3.58 g/m3	7.5 h	Lee et al. [47]
1.6% Pt-TiO2	300 °C, flow rate 100 mL/min		8 h
10% V/Al2O3	400 °C, flow rate 50 mL/min	1300 ppm	12.5 h	Cao et al. [13]
1% Pt/Al2O3			8.5 h
10% Cu/Al2O3			7.5 h
Al2O3			4.0 h
10% Fe/Al2O3			3.5 h
10% Ni/Al2O3			1.5 h
10% V/SiO2			25 h
CuO/γ-Al2O3	350 °C, flow rate 100 mL/min	4.0 g/m3	1.8 h	Dong et al. [48]
CuO-1% CeO2/γ-Al2O3			2.1 h
CuO-5% CeO2/γ-Al2O3			3.9 h
CuO-10% CeO2/γ-Al2O3			1.8 h
CeO2	400 °C, flow rate 100 mL/min	8.46 g/m3	2.33 h	Dong et al. [49]
10% Cu/Ce			4.2 h
20% Cu/Ce			4.43 h
50% Cu/Ce			5.36 h
80% Cu/Ce			2.33 h
CuO			0.93 h
2MCeO2np	300 °C, flow rate 50 mL/min	5.32 g/m3	5.8 h	This work
6MCeO2nr			7.0 h
12MCeO2nr			8.1 h
6MCeO2nc			3.5 h
12MCeO2nc			6.3 h

provided a brief summary of the protection times for various catalysts reported in the literature.

3.3. Reaction Mechanism for Thermocatalytic Decomposition of DMMP

12MCeO₂nr and 12MCeO₂nc with superior catalytic performances and 2MCeO₂np were selected for the study on the mechanism of thermocatalytic decomposition of DMMP. After the protection tests, the severely deactivated catalyst collected at the front end of the bed was analyzed using XPS in Figure 8. Based on Figure 8a–c, distinct P2p peaks were observed for all samples, which suggests that the surface of CeO₂ is covered by phosphorus species. The O1s spectra presented in Figure 8d–f could be deconvoluted into four peaks, i.e., surface lattice oxygen (O_Lat), phosphorus oxygen, adsorbed oxygen (O_Ads), and adsorbed water/hydroxyl oxygen [50].

The presence of phosphorus oxygen species further indicates the accumulation of P-containing byproducts on the surface of CeO₂, which essentially suppress the decomposition of DMMP. Meanwhile, the Ce³⁺ content in the sample after the reaction was also calculated from the fitted XPS spectra (Figure 8g–i). Variation of Ce³⁺ content in clean and deactivated catalysts are summarized in

Table 4. Variation of Ce³⁺ content in clean and deactivated catalysts.

Catalyst	Before Decomposition Reaction	After Decomposition Reaction	ΔCe3+	ΔCe3+/Time
	Ce3+/(Ce3++ Ce4+)	Ce3+/(Ce3++ Ce4+)
2MCeO2np	19.64	34.99	15.35	0.044
12MCeO2nr	25.15	35.01	9.86	0.020
12MCeO2nc	20.82	30.57	9.75	0.026

. Ce³⁺ content in the spent sample as compared to that in the unused sample (ΔCe³⁺) increased obviously (Table 4). Combined with the mechanism analysis of CeO₂ in the reaction pathway inference (Scheme 1), the surface oxygen of CeO₂ participates in the catalytic DMMP decomposition reaction, and the lattice oxygen migrates onto the surface, causing the formation of oxygen vacancies, and extra electrons are transferred to Ce⁴⁺, eventually, Ce⁴⁺ cations are reduced to Ce³⁺. However, the accumulation of the byproducts on the surface of the catalyst can inhibit the replenishment of external oxygen into the CeO₂ lattice, and Ce³⁺ could not be oxidized to Ce⁴⁺, this will increase the Ce³⁺ content. Therefore, it is speculated that the change in Ce³⁺ content may be related with the life of the catalysts (protection time). As such, the variation of Ce³⁺ content before and after catalytic decomposition reaction per unit protection time (ΔCe³⁺/Time) of three catalyst materials was calculated (Table 4). Based on the result, the ΔCe³⁺/Time value was inversely proportional to the protection time, which further confirms our previous speculation. Among all samples, 12MCeO₂nr possessed the smallest ΔCe³⁺/Time and the longest protection time, which indicates its superior catalytic performance.

Ion chromatography is used to qualitative analysis of P-containing byproducts on the surface of the spent CeO₂ catalysts and the colorless gelatinous species at the end of the catalytic reaction tube. The results of ion chromatograph indicate the presence of PO₄³⁻ at both sites as shown in

Table 5. Content of PO₄³⁻ on various CeO₂ materials and at the end of the reaction tube after catalytic decomposition of DMMP.

Samples	PO43−
2MCeO2np	3.508 (g/kg)
12MCeO2nr	1.450 (g/kg)
12MCeO2nc	1.720 (g/kg)
2MCeO2np reaction tube tail residues	0.2943 (mg/L)
12MCeO2nr reaction tube tail residues	0.0722 (mg/L)
12MCeO2nc reaction tube tail residues	0.1654 (mg/L)

. Therefore, the byproduct on the surface of CeO₂ is CePO_4. The colorless gelatinous PO₄³⁻-containing species is soluble in water, so it is not CePO₄. We speculate the colorless gelatinous species is H₃PO₄. For the reason that the boiling point of H₃PO₄ is lower than the catalytic reaction temperature, the weak adsorption of H₃PO₄ on the catalyst results in the desorption of H₃PO₄ from the catalyst bed by air flow. H₃PO₄ should form steam getting into the air; on the contrary, it was found on the wall at the end of the reaction tube. This phenomenon occurs because the catalytic reaction tube has a constant temperature zone, where the catalyst bed is in, and the temperature at the end of the reaction tube is lower than that in the constant temperature zone, so H₃PO₄ steam was condensed on the tube wall.

The tail gas produced during the catalytic decomposition of DMMP on 2MCeO₂np, 12MCeO₂nr, and 12MCeO₂nc were qualitatively analyzed by mass spectrometry to determine its components. The mass spectra of the tail gas produced by 2MCeO₂np, 12MCeO₂nr, and 12MCeO₂nc are displayed in Figure 9. According to the results, signals belonging to methanol, CO₂, H₂O, and H₂ were observed. Moreover, the mass-to-charge ratio of CO was also 28 in addition to the fragment signals of methanol and CO₂. As such, gas chromatography was employed to confirm the presence of CO in the produced tail gas.

The variations in the surface species were identified by employing time-resolved in situ DRIFTS over 2MCeO₂np, 12MCeO₂nr, and 12MCeO₂nc at 300 °C (Figure 10). The corresponding functional groups in accordance with the literature [51,52] are shown in

Table 6. Assignments of IR frequencies in DMMP.

Vibrational Mode	IR Frequencies (cm−1)
	DMMP [51]	DMMP Gas Phase [21,52]	DMMP Liquid Phase This Work
νa(P−CH3)	2992	3014	2996
νa(O−CH3)	2957	2962	2958
νs(P−CH3)	2926	2924	2927
νs(O−CH3)	2852	2860	2854
δa(O−CH3)	1465	1467	1465
δs(O−CH3)	1450	/	/
δs(P−CH3)	1317	1314	1315
ν(P=O)	1242	1276	1257
ρ‖(O−CH3)	1186	1188/1190	1187
νa(C−O)	1058	1070	1061
νs(C−O)	1034	1049	1034
ρ‖(P−CH3)	916	914	914
ν(P−O)	822	816	821
ν(P−O)	789	/	788
ν(P−C)	714	/	713

ν = stretching; δ = deformations; ρ = rocking; s = symmetric; a = asymmetric.

and

Table 7. Assignments of IR frequencies in surface species.

Vibrational Mode	IR Frequencies (cm−1)
	2MCeO2np	12MCeO2nr	12MCeO2nc	Reference [53]	Reference [54]	Reference [56]
ν(OH)	3701	/	3709	3710	3724	3720
ν(OH)	3659	3655	/	3660	3656/3651	3652
ν(OH)	/	3554	3594	3600	3616	/
νs(P−CH3)	2930	2928	2914	/	/	/
νs(O−CH3)	2854	2874	2876	/	/	/
ν(Ce−O−CH3) [55]	2813	2813	2813	/	/	/
2δ(C−H)	2716	2716	/	/	2721	/
νa(OCO) 1	1615	1620	/	1599	1609	/
νa(OCO) 2	1535	/	/	1553	1549	1552/1550
ν(CO3) 3 [57]	1396	/	/	1354	/	/
δ(C−H)	1365	1363	/	1371	1367	1375/1372
δs(P−CH3)	/	1318	1318	/	/	/
ν(CO3) 5 [57]	/	1318	1318	/	/	/
ν(CO3) 5 [57]	/	/	1236	/	/	/
ν(CO3) 4 [57]	1294	/	/	1289/	/	/
ν(P=O) [32,55]	1262	1279	1272	/	/	/
νa(P−O) [32,55]	1213	1208/	1130	/	/	/
νs(P−O) [32,55]	1056	1169	1050	/	/	/

¹. monodentate formate species; ². bidentate (bdt) formate bound to two Ce⁴⁺ atoms; ³. monodentate carbonate; ⁴. bidentate carbonate; ⁵. bridged carbonate.

. According to Figure 10, the peaks from 3500 cm⁻¹ to 3800 cm⁻¹, could be associated with the vibration mode of the surface hydroxyl on CeO₂ [53,54]. With longer reaction durations, the vibration intensity of partial surface hydroxyl increased negatively, which indicates the participation of surface hydroxyl during the decomposition of DMMP. The disappearance of ν(P−O) vibrational modes, i.e., 821 cm⁻¹ and 788 cm⁻¹, and ν(P−C) vibrational mode, i.e., 713 cm⁻¹, implies the removal of −OCH₃ and −CH₃ groups from the DMMP molecule. The ν_a(O−CH₃), δ_a(O−CH₃), ρ‖(O−CH₃), ν_a(C−O), ν_s(C−O), ν_a(P−CH₃), and ρ‖(P−CH₃) vibration modes that are ascribed to −OCH₃ and P−CH₃ groups disappeared. However, with prolonged testing duration, the peaks belonging to ν_s(P−CH₃), ν_s(O−CH₃), and δ_s(P−CH₃) still exist and are inferred to be the peaks of intermediate products. The peak at 2813 cm⁻¹ could be assigned to ν(Ce−O−CH₃). As Morris et al. reported, due to the P−OCH₃ cleavage, −OCH₃ combined with Ti ion to form surface methoxy. Compared to DMMP, the vibrational peak of surface methoxy will shift to a lower frequency [55]. It is worth noting that the intense peak belonging to ν(P=O) at 1257 cm⁻¹ vibrational mode shifted to 1262 cm⁻¹, 1279 cm⁻¹, and 1272 cm⁻¹ for CeO₂np, CeO₂nr, and CeO₂nc, respectively. This observation indicates that the DMMP chemisorbed to the Lewis acid sites of CeO₂, and/or bonded with the surface hydroxyl group of CeO₂ via H bond [32,52]. New distinct peaks located at 1208–1050 could be detected, which corresponds to P−O stretching in PO_x [32,55]; 2716 cm⁻¹ and 1365 cm⁻¹ belong to 2δ(C−H) and δ(C−H) of bidentate formate [54]; 1615 cm⁻¹ and 1620 cm⁻¹ belong to ν_a(OCO) in monodentate formate bonded with Ce³⁺; 1535 cm⁻¹ belongs to ν_a(OCO) of bidentate formate bonded with Ce⁴⁺ [56] (Figure 10a,b). Moreover, 1396 cm⁻¹ and 1294 cm⁻¹ that belong to ν(CO₃) in monodentate and bidentate carbonate could be detected, respectively (Figure 10a) [57]. In Figure 10b,c, δ_s(P−CH₃) vibration mode of products coincided with that of ν(CO₃) in bridged carbonate at 1318 cm⁻¹ [57], indicating that bridged carbonate may exist on the surface of CeO₂nr and CeO₂nc. In addition, 1236 cm⁻¹ that belongs to ν(CO₃) of bridged carbonate could be detected (Figure 10c) [57]. According to the above in situ DRIFTS analysis, the difference of surface products between the three morphologies lies in the carbon containing byproducts. Monodentate formate, bidentate formate, monodentate carbonate, and bidentate carbonate were observed on 2MCeO₂np. Monodentate formate and possibly bridged carbonate are considered to be the main carbon-containing byproducts of 12MCeO₂nr. The main carbon-containing byproduct on 12MCeO₂nc was bridged carbonate. The accumulation of phosphate, formate, and carbonate species causes the masking of the active site, which ultimately deactivates the catalyst.

Based on the mass spectroscopy, ion chromatography, and in situ DRIFTS studies, the catalytic decomposition of DMMP pathways based on 2MCeO₂np, 12MCeO₂nr, and 12MCeO₂nc are proposed. The 3D model schematics of CeO₂np, CeO₂nr, and CeO₂np were depicted based on different morphologically exposed crystal plane [58]. From the in situ DRIFTS analysis, the main reaction pathways based on the CeO₂ samples with three different morphologies are the same (Scheme 1Ⅰ). Firstly, the electron-rich P=O in DMMP adsorbs onto the Lewis acid active site of CeO₂, and/or bonded with the surface hydroxyl group of CeO₂ via H bond [33,59]. Then, the surface hydroxyl group provides the nucleophilic attack to the phosphorus, and the methoxy group combines with the surface hydrogen to produce gaseous methanol. Cerium methyl phosphate is then formed by further losing the methoxy group (Scheme 1a) [15,17,22,32]. The scission of the P−CH₃ bond through the surface lattice oxygen of the CeO₂ nucleophilic attack leads to the formation of CePO₄ and a transient methoxy intermediate that contains lattice oxygen (Scheme 1b). After sacrificing the C−H bond of the transient methoxy intermediate, CO and H₂ are generated on the oxygen-deficient surface of CeO₂ [60]. However, based on the tail gas analysis, the formation of CO, H₂, CO₂, and H₂O were confirmed (Figure 9), which implies that a part of CO and H₂ were oxidized to CO₂ and H₂O under an oxygenated environment. The methanol generated via the main reaction can continue to react with the surface lattice oxygen in CeO₂, which produces a variety of products (Scheme 1Ⅱ–Ⅳ). Such a process is complex, and it is similar to the reports by Overbury and Foster group [54,56]. Combined with the different types of carbon containing byproducts via further methanol reaction, the side reaction pathways of 2MCeO₂np, 12MCeO₂nr, and 12MCeO₂nc are different. According to Scheme 1Ⅱ, The lattice oxygen and methanol initially interact to produce an adsorbed methoxy and a hydroxyl group adsorbed on the surface (Scheme 1d) [54]. Then, the methoxyl group reacts with hydroxyl or surface lattice oxygen, whereby CO, H₂, CO₂, and H₂O are formed during a complete dehydrogenation (Scheme 1e), while monodentate formates are formed during partial dehydrogenation (Scheme 1f) [54]. In addition, the methoxyl group reacts with hydroxyl and it dehydrates to form a bidentate coordinated methoxyl group (Scheme 1g) [56]. This is then further dehydrogenated by reacting with two adjacent hydroxyl groups to form a bidentate coordinated formate and hydroxyl group (Scheme 1h) [56]. Moreover, CO₂ adsorption on surface oxygen could result in monodentate carbonate and bidentate carbonate (Scheme 1i) [57,61]. When compared to the side reaction pathway of 2MCeO₂np (Scheme 1Ⅱ), the reactions of Scheme 1d–g on 12MCeO₂nr (Scheme 1Ⅲ) are the same as that on 2MCeO₂np. In addition, based on in situ DRIFTS analysis, the bridging carbonate may exist on 12MCeO₂nr, which is formed via the complete dehydrogenation of the surface methyl group (Scheme 1j) [60]. Based on in situ DRIFTS analysis, no formate is formed on the surface of 12MCeO₂nc, therefore, no reaction pathway of Scheme 1f occurred on 12MCeO₂nc (Scheme 1Ⅳ); the other reaction pathways are the same as 12MCeO₂nr. According to the proposed reaction path, surface oxygen (surface lattice oxygen or adsorbed oxygen) participates in the oxidative decomposition of DMMP. It is well known that CeO₂ possesses an excellent oxygen storage and release capacity. In higher temperature reactions, the mechanism of CeO₂ is primarily the Mars–van Krevelen mechanism [62]. Thus, when the surface oxygen is consumed, the bulk oxygen and the external oxygen (such as oxygen, water molecules) will replenish the oxygen vacancy for the next catalytic DMMP decomposition cycle (Scheme 1k) [53]. The catalyst will eventually become completely deactivated when the byproducts fully cover the active sites on its surface.

4. Conclusions

In this work, various CeO₂ catalysts with three different morphologies, i.e., irregular nanoparticles, nanorods, and nanocubes, were synthesized via a modified preparation method, and they were investigated for their thermocatalytic performance in DMMP decomposition. Based on the results, the as-prepared CeO₂ nanorods exhibited superior MSTC as compared to the irregular CeO₂ nanoparticles and nanocubes. For the CeO₂ nanomaterials with the same morphology, i.e., 12MCeO₂nr and 6MCeO₂nr, it is noted that 12MCeO₂nr with larger crystal sizes exhibited higher protection time, MSTC, and SSTC than 6MCeO₂nr with smaller crystal sizes. This same conclusion exists for CeO₂ nanocubes. These results indicated that the morphologies and crystal sizes of the CeO₂ nanomaterial could significantly affect the catalytic performance. The various CeO₂ nanomaterials with different morphologies and particle sizes possessed different redox activity (H₂−TPR), surface oxygen species (XPS), relative oxygen vacancy concentration (Raman), and the ability of the catalyst to resist poisoning. However, these properties did not show direct relationships with the catalytic performance of the material, which indicates that the catalytic performance of CeO₂ nanomaterials is not only affected by a single factor, instead it may be a result of a set of factors, and it will be an interesting work in our future study. It is speculated that the change in the Ce³⁺ content before and after the protection test could be due to the poisoning of the catalyst, therefore affecting the protection time. Based on the storage and release characteristics of CeO₂, the catalytic decomposition of DMMP was performed through the cyclic process that involved the consumption of surface lattice oxygen and adsorbed oxygen, and the timely replenishment of bulk and external oxygen. The main reaction path of the catalytic DMMP decomposition was similar for all three morphologies samples, the gas products were methanol, CO, H₂, CO₂, and H₂O, and the surface byproduct was phosphates. However, during the side reaction that involved methanol, the reaction paths of three morphologies samples are different. On the surface of CeO₂ irregular nanoparticles, carbonate and formate were produced, while formate and possibly carbonate were produced on the surface of CeO₂ nanorods, only carbonate was formed on the surface CeO₂ nanocubes. The CeO₂ catalyst was inactivated due to the accumulation of phosphorus oxygen species and carbon oxygen species on the surface of CeO₂. Thus, based on the collective results, a basic understanding of the catalytic DMMP decomposition based on CeO₂ with different morphologies was attained, and this understanding will contribute to the future design and preparation of high-performing catalysts for the degradation of chemical toxicants.