Nitrogen oxides (NO_x), which arise from automobile exhaust gas and the combustion of fossil fuels, are a major source of air pollution. They are responsible for photochemical smog, acid rain, ozone depletion, and greenhouse effects ^{[1, 2]}. Selective catalytic reduction of NO_x by NH₃ (NH₃-SCR) is currently regarded as the most promising technology for removing NO_x from the exhaust of vehicles and stationary combustion processes. A typical commercial catalyst for NH₃-SCR is V₂O₅-WO₃(MoO₃)/TiO₂ ^{[3, 4]}. However, inevitable problems related to the toxicity of vanadium species, high conversion of SO₂ to SO₃, and the relatively narrow window temperature for this catalyst have led to increased interest in developing novel, highly active, stable, environmentally friendly, and vanadium- free NH₃-SCR catalysts for controlling NO_x emissions.

CeO₂ is a key component in automotive three-way catalysts for the treatment of exhaust gases owing to its large oxygen storage capacity and excellent redox properties ^{[5, 6]}. However, it is well known that both redox and acidity is necessary in NH₃-SCR catalysts ^{[7, 8]}. The redox properties of the catalyst appear to be a key factor in controlling the reactivity of de-NO_x catalysts at low temperatures, while surface acidity plays an important role in the SCR reaction at high temperatures ^{[9, 10, 11]}. Because ceria is a basic solid ^[12], it is inevitable that its redox properties and acidity must be adjusted to obtain a wide operating temperature window for CeO₂-based NH₃-SCR catalysts for reducing NO_x emissions. For example, Shan et al. ^[13] reported a novel Ce-W mixed oxide catalyst that exhibited nearly 100% NO_x conversion in a wide temperature range from 250 to 425 °C, in which the W species increased the amount of Brönsted and Lewis acid sites on the catalyst surface. The modification was not only beneficial in improving the low-temperature activity of the catalyst by facilitating “fast SCR” reaction, but also enhanced its activity at high temperature by inhibiting the unselective oxidation of NH₃ ^{[14, 15]}. Furthermore, modification of CeO₂ with niobate has also been revealed to be an effective method of improving NH₃-SCR activity, ascribed to the resulting significant increase in surface acidity ^[16]. Si et al. ^{[17, 18]} showed that sulfate and nickel modifications effectively enhanced the activity and selectivity of ceria-zirconia catalysts for NH₃-SCR. The nickel modification improved the Lewis acidity of the CeO₂-ZrO₂ catalysts, while Brönsted acid sites formed by the sulfate modification facilitated ammonia adsorption instead of ammonia oxidation at high temperatures. Furthermore, Si et al. ^{[19, 20]} reported that a zirconium phosphate@Ce_0.75Zr_0.25O₂ catalyst exhibited high NH₃-SCR activity within a wide temperature range of 250-450 °C. Loading zirconium phosphate onto Ce_0.75Zr_0.25O₂ was found to reduce the strong interaction between phosphate and cerium as well as introduce acid sites onto the surface of the Ce_0.75Zr_0.25O₂. Very recently, Li et al. ^{[21, 22]} reported that a Ce-P-O catalyst prepared by a hydrothermal method showed excellent NH₃-SCR activity and K₂O and SO₂/H₂O resistance within a broad temperature range, and thus displayed promise for the de-NO_x process of stationary sources at medium or high temperatures. However, the structure of these Ce-P-O catalysts was not confirmed, and thus the structure-activity relationship remained unclear.

In this work, a series of H₃PO₄-modified CeO₂ samples were prepared by impregnation of ceria with phosphoric acid solution and evaluated for NH₃-SCR. We found that H₃PO₄ had a significant promotional effect on the CeO₂ catalyst toward the NH₃-SCR reaction. The bulk and surface structures of this catalyst were characterized using different methods. Furthermore, the structure-activity relationship of the H₃PO₄- modified catalysts was thoroughly investigated.

CeO₂ was prepared using the citric acid method previously described in the literature ^[17]. A certain amount of Ce(NO₃)₃·6H₂O was dissolved in distilled water and then mixed with aqueous citric acid to produce a solution with a citric acid to total metal ions ratio of 1.1:1. The solution was stirred vigorously at room temperature for at least 1 h, and then the water was removed by evaporation. The resulting solid was decomposed to produce the oxide by annealing in air at 200 °C. The obtained oxide was then ground and further calcined at 500 °C for 3 h in a muffle oven.

H₃PO₄-modified CeO₂ samples were prepared by wet incipient impregnation of the prepared ceria with H₃PO₄ solution. The obtained mixture was first dried at 110 °C overnight and then calcined at 500 °C for 3 h in static air. The resulting catalysts were denoted as xH₃PO₄-yCeO₂, where “x/y” denotes the P/Ce molar ratio.

Powder X-ray diffraction (XRD) patterns were obtained using a D/Max-IIB diffractometer and Cu K_α radiation (λ = 1.5406 Å). The working voltage and current of the X-ray tube were 40 kV and 30 mA, respectively. The samples were scanned within the range 2θ = 10°-80° at a scanning rate of 5°/min.

Infrared spectra were recorded with a Fourier-transform infrared (FT-IR) spectrometer (Bruker Vertex 70 FTIR) using the KBr wafer technique for sample preparation.

Raman spectra were recorded on a Raman 960 FT-Raman spectrometer using a laser with an excitation wavelength of 1064 nm and a resolution of 8 cm^-1. All Raman spectra presented in this paper are original and unsmoothed.

X-ray photoelectron spectroscopy (XPS) was conducted on a scanning X-ray microprobe (Thermo ESCALAB 250) using Al K_a radiation (1486.7 eV). The binding energies of Ce 3d, P 3p, and O 1s were calibrated using the C 1s peak (BE = 284.8 eV) as standard.

The specific surface areas of the samples were obtained from N₂ adsorption-desorption analysis at -196 °C using a Micromeritics ASAP 2010. The catalyst samples were degassed at 250 °C overnight prior to the N₂ physisorption.

Pyridine-IR spectra were recorded on a Nicolet 410 FT-IR spectrometer. The samples were heated at 200 °C for 30 min and then cooled to room temperature in vacuum of 10^-4 mmHg. The pre-treated samples were exposed to pyridine as the probe molecule at room temperature, after which the pyridine was evacuated at 100 °C. All the IR measurements were performed at room temperature.

Temperature-programmed desorption of NH₃ (NH₃-TPD) was performed using a quadrupole mass spectrometer to record the signals of NH₃ (m/z = 15 for NH). Prior to the TPD experiments, the samples (50 mg) were pretreated at 500 °C in a flow of pure Ar for 1 h and allowed to cool to room temperature. The samples were then exposed to a flow of 1000 ppm NH₃/Ar at room temperature for 2 h, followed by Ar purge for another 1 h. Finally, the temperature was raised to 800 °C in flowing Ar at a rate of 10 °C /min.

For the H₂ temperature-programmed reduction (H₂-TPR) experiments, the samples (50 mg) were pretreated at 500 °C in a flow of air for 1 h and allowed to cool to room temperature. The temperature was then raised to 900 °C at a rate of 10 °C/min in a flow of 5 vol% H₂/Ar. A thermal conductivity cell was used to detect the consumption of H₂.

SCR activity measurements of the catalysts were performed in a fixed-bed quartz flow reactor. The reaction conditions were as follows: 500 ppm NO, 500 ppm NH₃, 5% O₂, Ar balance, GHSV = 20000 mL/(g·h). The gas concentrations in the outlet stream were continuously monitored with an online quadrupole mass spectrometer (QIC-20, Hiden, UK). The following ion peaks (m/z) were monitored to determine the concentration of products and reactants: 15 (NH₃), 30 (NO), 44 (N₂O), and 46 (NO₂). NO_x conversion was calculated as X_NOx= (1 - [NO_x]_out/[NO_x]_in) × 100%, and N₂ selectivity was calculated as S_N2 = 1 - 2[N₂O]_out/([NO_x]_in - [NO_x]_out + [NH₃]_in - [NH₃]_out) × 100%, where [NO_x] = [NO] + [NO₂].

The catalytic performance of the pure and H₃PO₄-modified CeO₂ catalysts for NH₃-SCR of NO in the presence of oxygen are shown in Fig. 1. As expected, the pure CeO₂ showed very poor NO_x conversion throughout the whole temperature range and poor N₂ selectivity at temperatures higher than 400 °C. The poor high-temperature SCR activity of CeO₂ was attributed to the strong redox properties of ceria, which led to the over-oxidation of NH₃ by gaseous oxygen ^{[23, 24, 25]}. However, the catalytic efficiency was significantly improved after H₃PO₄ modification, and the loading of H₃PO₄ played a pivotal role in broadening the operating temperature window. The window was greatly broadened as the P/Ce molar ratio was increased from 1:10 to 1:1, and more than 80% NO conversion was achieved in the temperature range of 250-550 °C over the H₃PO₄-CeO₂ sample. However, upon higher H₃PO₄ loading, such as for the 1.5H₃PO₄-CeO₂ sample, the low-temperature SCR activity was decreased slightly in comparison with that of H₃PO₄-CeO₂.

Fig. 1. NOx conversion and N2 selectivity over the pure CeO2 and H3PO4-modified CeO2 samples. Reaction conditions: 500 ppm NO, 500 ppm NH3, 5% O2, Ar balance, GHSV = 20000 mL/(g·h).

The XRD patterns of H₃PO₄/CeO₂ samples with various H₃PO₄ loadings are shown in Fig. 2. For all of the samples, the typical diffraction lines of the cubic CeO₂ structure (JCPDS No. 34-0394) were clearly visible. At low P/Ce ratios, no phases except CeO₂ were observed, suggesting that the phosphorus species may have been intercalated into the CeO₂ lattice structure or existed as amorphous phosphate species. However, with increasing H₃PO₄ loading, CeP₂O₇ (JCPDS No. 16-0584) phase appeared in the H₃PO₄-CeO₂ sample, while in the 1.5H₃PO₄-CeO₂ sample, CeP₂O₇ became the main phase, and weak peaks of monazite CePO₄ (JCPDS No. 32-0199), Ce(PO₃)₃, and Ce(PO₃)₄ could also be observed ^{[26, 27, 28, 29]}. The diffraction peaks of CeO₂ in the H₃PO₄-modifed ceria samples were shifted towards lower 2θ values as the H₃PO₄ content increased. This may be ascribed to an increase in lattice parameters caused by part of the phosphorus species entering the fluorite structure of ceria ^{[30, 31]}.

Fig. 2. XRD patterns of the pure and H3PO4-modified CeO2 samples. (1) CeO2; (2) H3PO4-10CeO2; (3) H3PO4-2CeO2; (4) H3PO4-CeO2; (5) 1.5H3PO4-CeO2.

The BET surface areas of the samples are presented in Table 1. The surface area of the catalysts first decreased with increasing P content and then increased at higher P/Ce molar ratio. The lowest surface area appeared at the molar ratio of 1:2. These results indicated that there was no clearly defined relationship between catalytic performance and BET surface area in our study. However, in Li’s work ^{[21, 22]}, the authors showed that Ce-P-O(h), prepared by a hydrothermal method, was a novel and efficient catalyst for NH₃-SCR with NO conversions above 90% within the range of 200-550 °C, a better catalytic performance than our H₃PO₄-CeO₂ catalyst. Although the two catalysts had the same composition (Ce, P, and O), our H₃PO₄-CeO₂ catalyst had a much smaller BET surface area (17.6 m²/g) than that reported for the Ce-P-O(h) catalyst (82.4 m²/g). Thus, the larger specific surface area of Ce-P-O(h) catalyst is the likely reason for its excellent performance.

Table 1 Physical properties and XPS data of the catalysts Sample Binding energy (eV) Ce3+/(Ce4++Ce3+) (%) SBET a Ce 3d5/2 P 2p O 1s (m2/g) CeO2 882.4 — 529.3 16.4 35.6 H3PO4-10CeO2 882.6 132.7 529.7 25.6 32.8 H3PO4-2CeO2 882.5 133.2 529.9 25.3 10.7 H3PO4-CeO2 882.8 133.6 530.4 24.1 17.6 1.5H3PO4-CeO2 883.0 133.4 530.6 23.0 36.1 a Specific surface area measured by the Brunauer-Emmett-Teller (BET) method.

Table 1 Physical properties and XPS data of the catalysts

The Raman spectra of the H₃PO₄-modified CeO₂ samples are displayed in Fig. 3. The band at 465 cm^-1 in the spectra of all samples was attributed to the symmetric breathing mode of O atoms around each Ce⁴⁺ cation in the cubic CeO₂ phase ^{[32, 33]}. As the P/Ce ratio was increased from 1/10 to 1.5/1, the intensity of this band at 465 cm^-1 decreased monotonously, which meant that the amount of cubic CeO₂ phase continuously decreased. Meanwhile, a new weak band appeared at 1034 cm^-1, which could be assigned to the symmetric stretching of PO₃ terminal groups from cerium pyrophosphate ^{[34, 35]}. Further increasing the P/Ce ratio resulted in an obvious increase in the proportion of the cerium pyrophosphate phase, and pyrophosphate became the main phase in the 1.5H₃PO₄- CeO₂ sample, consistent with the XRD result.

Fig. 3. Raman spectra of the H3PO4-modified CeO2 samples measured at 100-2000 cm-1.

Fig. 4 shows the FT-IR spectra of the pure and H₃PO₄-modified CeO₂ catalysts. Several new bands at 470, 532, 748, 954, and 1223 cm^-1 appeared in the H₃PO₄-CeO₂ samples as the H₃PO₄ loading was increased. These bands could be attributed to O-P bonds, while the bands at 1223 and 748 cm^-1 could be ascribed to P₂O₇^4- and P₄O₁₂^4-, respectively ^{[36, 37, 38, 39, 40]}. Additionally, as P/Ce ratio was increased from 1/50 to 1.5/1, the intensity of the bands at 1223 and 748 cm^-1 increased monotonically, consistent with the Raman results.

Fig. 4. IR spectra of the H3PO4-modified CeO2 samples.

To obtain a better understanding of their surface characteristics and concentration of active species, the CeO₂-based catalysts were investigated using XPS. The electron-binding energies (BE) of the photoelectron peaks pertaining to Ce 3d, P 2p, and O 1s are shown in Table 1. The obtained values agreed well with those previously reported in the literature ^{[14, 15, 41]}. The representative photoelectron peaks of Ce 3d and O 1s for both the pure CeO₂ and H₃PO₄-modified CeO₂ samples are depicted in Figs. 5 and 6, respectively. As shown in Table 1, the binding energy of the Ce 3d and O 1s core-levels of the H₃PO₄-modified CeO₂ samples was shifted to higher values with respect to pure CeO₂, indicating the interaction between Ce and P species. As shown in Fig. 5, the Ce 3d spectra of the CeO₂-based samples with different H₃PO₄ loadings were very complicated. By performing peak-fitting deconvolution, the Ce 3d spectra could be separated into eight bands. The sub-bands labeled u' and v' represent the 3d¹⁰4f¹ initial electronic state corresponding to Ce³⁺, while the sub-bands labeled u, u'', u''', v, v'', and v''' represent the 3d¹⁰4f ⁰ state of Ce^{4+ [14, 42]}. The Ce³⁺ ratios of the samples, calculated as Ce³⁺/(Ce³⁺+Ce⁴⁺), are shown in Table 1. The Ce³⁺ ratio in H₃PO₄-modified CeO₂ was significantly higher than that in CeO₂. Meanwhile, it can be seen that the Ce³⁺ ratio increased remarkably with H₃PO₄ loading until it reached a maximum for the H₃PO₄-10CeO₂ sample, after which further increases led to a slightly decreased Ce³⁺ ratio. As indicated by the XRD, IR, and Raman results, besides Ce(IV) pyrophosphate, other phosphates such as the monazite CePO₄ phase appeared in the H₃PO₄-modified CeO₂ samples as the H₃PO₄ loading was increased, which led to an increased Ce³⁺ ratio. However, the highly stable CePO₄ has been claimed to be responsible for the locking of a considerable fraction of the Ce³⁺/Ce⁴⁺ redox couple, which may reduce the redox ability of these catalysts ^[20].

Fig. 5. Ce 3d XPS spectra of the pure and H3PO4-modified CeO2 samples. (1) CeO2; (2) H3PO4-10CeO2; (3) H3PO4-2CeO2; (4) H3PO4-CeO2; (5) 1.5H3PO4-CeO2.

Fig. 6. O 1s XPS spectra of the CeO2-based samples with different H3PO4 loadings. (1) CeO2; (2) H3PO4-10CeO2; (3) H3PO4-2CeO2; (4) H3PO4-CeO2; (5) 1.5H3PO4-CeO2.

Fig. 6 shows the O 1s XPS spectra of the samples in the range of 524-542 eV. There were three kinds of oxygen species present in the samples. The peak centered at 531.5 eV was assigned to non-bridging oxygen (P-O), the peak centered at 533.0 eV was caused by bridging oxygen (P-O-P), and the peak centered at 529.5 eV is ascribed to lattice Ce-O ^{[43, 44, 45]}. As shown in Fig. 6, the relative intensities of both the P-O-P and P-O contributions increased and the lattice Ce-O peak intensity significantly decreased with increasing H₃PO₄ content. These results confirm that phosphorus species appeared on the surface of the catalysts after H₃PO₄ modification as expected.

NH₃-TPD is a frequently used method for determining the number of acid sites on solid acid catalysts as well as acid strength distribution. Fig. 7 shows NH₃-TPD profiles of the pure CeO₂ and H₃PO₄-modified CeO₂ catalysts. Pure CeO₂ showed two broad NH₃ desorption peaks, one at about 120 °C and the other at around 250 °C, which indicate weak acid strength. However, it is noteworthy that the acid site strength changed significantly after H₃PO₄ modification. With increasing H₃PO₄ loading, the weak acid site peak decreased gradually until it completely disappeared, while a peak at around 600 °C, assignable to strong acidity, dramatically increased in intensity. As discussed above, the surface area of the catalysts initially decreased with increasing P content and then increased at higher P/Ce molar ratio (Table 1). The lowest surface area was observed for the sample with a P/Ce molar ratio of 1:2. Moreover, the pore paths accessible by NH₃ were partly blocked by the phosphate formed as a result of the P enrichment of the surface of the catalyst. As a result of these two factors, the amount of desorbed NH₃ from 1.5H₃PO₄-CeO₂ and H₃PO₄-2CeO₂ was much lower than that from H₃PO₄- 10CeO₂. However, the H₃PO₄ modification increased the acidic strength of the catalysts, which facilitated the adsorption of NH₃ for the catalysts and simultaneously favored the SCR reaction.

Fig. 7. NH3-TPD profiles of the pure CeO2 and H3PO4-modified CeO2 catalysts. (1) CeO2; (2) H3PO4-10CeO2; (3) H3PO4-2CeO2; (4) H3PO4-CeO2; (5) 1.5H3PO4-CeO2.

The adsorbed pyridine IR technique was used to determine the acidic properties of the CeO₂-based catalysts. As shown in Fig. 8, the IR spectra of pyridine adsorbed on CeO₂ exhibited bands at 1456, 1485, 1595, 1545, and 1642 cm^-1. The bands at 1456 and 1595 cm^-1 are assignable to Lewis acid sites, while the band at 1490 cm^-1 is mainly attributed to Lewis and Brönsted acid sites, and the bands at 1545 and 1642 cm^-1 are caused by Brönsted acid sites ^{[32, 46]}. Thus, both Lewis and the Brönsted acid centers were found on the surfaces of the CeO₂ catalysts. However, the pyridine-IR profiles exhibited an obvious change after H₃PO₄ modification. The bands attributed to Brönsted acidic sites showed an obvious increase in intensity with the amount of H₃PO₄ modification, while the amount of Lewis acidic sites decreased considerably after H₃PO₄ modification. The enhanced Brönsted acidity is probably connected with POH surface groups on the H₃PO₄-modified CeO₂ catalysts. Although the nature of the NH₃ adsorption sites was still under debate, Li et al. ^{[47, 48, 49]} showed that Brönsted acid sites efficiently catalyzed the decomposition of ammonium nitrite (an important intermediate in the SCR cycle), which promoted the NH₃-SCR reaction. Furthermore, more Brönsted acidic sites might provide more moderate acid sites exhibiting less oxidation activity in comparison with Lewis acid sites, which would be beneficial for avoiding NH₃ consumption at high temperatures ^{[17, 25]}. Thus, H₃PO₄ modification enhances the high temperature NH₃-SCR activity of CeO₂ owing to the higher ratio of Brönsted to Lewis acid centers.

Fig. 8. IR spectra of pyridine adsorbed on the pure CeO2 and H3PO4-modified CeO2 catalysts. (1) CeO2; (2) H3PO4-2CeO2; (3) H3PO4-CeO2; (4) 1.5H3PO4-CeO2.

The redox properties of the CeO₂-based samples were investigated by H₂-TPR. The H₂-consumption profiles obtained from the TPR experiments are plotted in Fig. 9. As expected, the reduction profile of pure CeO₂ exhibited two peaks ^{[17, 50]}. The first peak, centered at around 500 °C, was attributed to the reduction of the uppermost layers of Ce⁴⁺, while the second peak, at above 800 °C, originated from bulk reduction. However, the H₂-TPR profile was significantly affected after H₃PO₄ modification. The position of the CeO₂ reduction peak shifted to higher temperature with increasing H₃PO₄ content, which indicated that the oxidation ability of the catalysts was decreased. This difference in the redox behavior between the H₃PO₄-CeO₂ catalyst and pure CeO₂ may have arisen from interaction between Ce and P species.

Fig. 9. H2-TPR profiles of the CeO2 and H3PO4-modified CeO2 catalysts.

Phosphorus species, produced by the decomposition of zinc dialkyldithiophosphate present in engine oil, can lead to the poisoning of automotive catalysts because the formed monazite (CePO₄) accumulates on the surface and in the subsurface regions of the three-way catalysts. This deteriorates the oxygen storage and release properties of CeO₂ ^{[32, 33, 34]}. However, in this work, we found that the addition of H₃PO₄ to CeO₂ catalysts promoted the catalytic efficiency of NH₃-SCR, and the operating temperature window was greatly broadened. Based on XRD, IR, Raman spectroscopy, and XPS analyses, it was concluded that in the samples with low P/Ce ratio (<1:2), the phosphorous species were highly dispersed on the surface of the CeO₂, while part of the species diffused into the subsurface region and even became incorporated into the CeO₂ lattice. The incorporation of P into the CeO₂ lattice may occur through substitution of anions with PO₄^3- or P₂O₇^4-, which would lead to the formation of Ce(III) and excess oxygen in the lattice ^{[32, 33, 51]}, consistent with the XPS results shown in Table 1. However, as the P/Ce ratio increased, amorphous pyrophosphates, orthophosphates, and metaphosphate appeared in the samples. With further increases in the H₃PO₄ loading, crystallization of the amorphous CeP₂O₇ occurred on the H₃PO₄-CeO₂ sample and a mixture of crystalline CeP₂O₇, CePO₄, Ce(PO₃)₃, and Ce(PO₃)₄ was formed on the 1.5H₃PO₄-CeO₂ catalyst, in which CeP₂O₇ as the main phase. Our previous work has demonstrated that rare earth pyrophosphates used as solid acid catalysts show excellent catalytic performance for the Beckmann rearrangement reaction and the vapor-phase dehydration of glycerol ^{[52, 53]}. Furthermore, we found that polyphosphate possessed much stronger surface acidity than CePO₄ ^[53]. Therefore, the increase in acidity on the samples with higher H₃PO₄ loading may have been caused by the formed crystalline polyphosphate.

As is well known, both redox and acidic properties are necessary for the SCR reaction ^{[7, 8]}. Previous studies have showed that the redox properties of a catalyst governs its low-temperature activity, while its surface acidity determined its SCR activity at high temperatures ^{[9, 10, 11]}.

Besides acid properties, redox properties have been claimed to be another important controlling factor for the De-NO_x reaction, especially at high temperatures ^{[1, 3]}. A high reaction temperature was unfavorable for NH₃ adsorption but promoted H-abstraction of the adsorbed NH₃. Strong oxidation ability would result in higher catalytic activity for NH₃ oxidation to N₂ or even NO_x, causing decreases in the NH₃-SCR activity and N₂ selectivity of the catalyst at high temperatures. The present H₃PO₄ modification decrease the oxidation ability of the CeO₂-based catalysts and inhibited NH₃ oxidation on the catalysts. Therefore, the high-temperature NH₃-SCR activity and selectivity of the catalysts were improved after H₃PO₄ modification. In summary, the addition of H₃PO₄ to CeO₂ in the present study increased its acidic strength and modified its redox ability, which led to a very broad operation temperature window.

The addition of H₃PO₄ to CeO₂ enhanced its activity for NH₃-SCR remarkably, especially at high temperatures. The modification resulted in an increased acidic strength and reduced redox ability, which were responsible for the very broad operation temperature window of the H₃PO₄-modified CeO₂ catalysts.