Ordered Nanostructure Catalysts Efficient for NOx Storage/Reduction (NSR) Processes

Abstract

NOx emissions in the atmosphere can cause various environmental problems, which should be strictly controlled and regulated. Furthermore, because of the limited amount of crude oil resources in the world and severe global warming, the development of fuel-efficient vehicles has long been desired. Accordingly, efficient NOx storage and reduction catalysts have been developed over the decades, called NSR (NOx storage/reduction) catalysts. In the present article, recent advances in NSR catalysts which possess ordered nanostructures will be summarized, including our noble Pt/KNO₃/K-titanate nanobelt (KTN), Pt-KNO₃/CeO₂ and Pt-KNO₃/ZrO₂ catalysts, as well as nanoporous Ni-phosphate (VSB-5) and Co-substituted VSB-5 catalysts.

1. Introduction

1.1. Overview of the Development on NSR Catalysts

Emissions of NO_x in the atmosphere can cause various environmental problems. Therefore, it is an urgent requirement that NO_x emissions should be strictly controlled and regulated [1]. On the other hand, due to the limited quantity of raw oil resources in the world and seriously increasing global warming, fuel-efficient vehicles become desirable. Most current gasoline-burning engines are operating with a very narrow air-to-fuel ratio (A/F) range near the stoichiometric value (A/F = 14.7, rich-burn) with little or no O₂ in the exhaust, while lean-burn engines (A/F = 20–25) can burn fuel more efficiently [2,3]. However, the conventional three-way catalyst is no longer effective for NO_x removal in an oxidative atmosphere. Therefore, an effective NO_x removal method for lean-burn exhaust becomes desirable from the environmental and catalytic point of view. NO_x removal technologies under the lean-burn condition include direct catalytic decomposition of NO_x, catalytic NO_x storage/reduction (NSR) [1,4,5,6,7], and selective catalytic reduction of NO_x with urea/NH₃ [8,9] or hydrocarbons [10,11,12].

The NSR catalysts operate under cyclic fuel lean–rich conditions. Under the lean-burn condition, NO_x is absorbed onto the catalyst, and under the rich-burn condition, stored NO_x is reduced by H₂, CO, and hydrocarbons. In real systems, a storage period of 1–2 min, followed by 3–5 sec of rich conditions, is typically adopted. To achieve this cycle, NSR catalysts are composed of three essential elements: (1) a high-surface-area metal oxide, such as Al₂O₃, as a support; (2) basic elements, such as Ba or K, for the NO_x storage sites; and (3) precious-metal redox sites, such as Pt, Pd, and Rh, for the redox reaction for oxidizing NO to NO₂ and reducing stored NO_x to N₂ by various reductants. Typical model of an NSR catalyst is Pt-Ba/Al₂O₃, which was developed by Toyota’s research group [6,7,8,9]. Although considerable amounts of experimental and theoretical investigations of Pt-Ba/Al₂O₃ have been conducted, serious problems still remain with respect to sulfur poisoning [6,13,14,15,16]. Meanwhile, high-NO_x storage NSR catalysts are desirable because increasing NO_x storage capacity can decrease the necessary catalyst and, in turn, decrease the amount of precious metals required and the cost.

1.2. Application of Nanotechnology for NSR Processes

In order to obtain better NSR catalysts, the application of nanotechnology seems to be promising. As for support materials other than Al₂O₃, titanium dioxide has been proposed as an efficient support since it has a relatively higher resistance to sulfur poisoning [6]. Previous reports have demonstrated that K-based catalysts showed good performance in NO_x storage in lean-burn atmospheres [15,16,17,18,19,20]. Most researchers believe that the reaction of K and TiO₂ will cause the loss of NO_x storage capacity. However, recent studies on K₂Ti₂O₅ prepared by a solid-state reaction showed high NO_x storage capacity (1.2 mmol/g) at 550 °C [21,22]. The structural transformation between K₂Ti₂O₅ and K₂Ti₆O₁₃ was considered as the mechanism for NO_x storage in Pt/K₂Ti₂O₅ [22]. The lower NO_x storage capacity at relatively lower temperature (200–400 °C) and the slower NO_x storage rate even at higher temperatures (550 °C) have resulted in the limited application of K₂Ti₂O₅-based catalysts.

Alkali-titanate nanomaterials have been widely investigated [23,24,25,26,27]; however, the usual applications have directed the formation of functional materials by neutralizing them with acid [28,29,30]. Recently, we reported on the preparation of potassium-titanate nanobelts (KTN) with the hydrothermal method and employed them as supports for Pt-KNO₃-impregnated NSR catalysts [31]. The obtained catalysts exhibited a high NO_x storage capacity (1.27–2.27 mmol/g) at 350 °C, the maximal value of which (2.27 mmol) was the highest NO_x storage capacity ever reported in the relevant literature. From the results of isothermal NO_x storage and lean-rich cycling experiments, as well as various characterizations of catalysts before and after the reaction, we were able to demonstrate that the reduction of stored KNO₃ resulted from the formation of K-rich surface layers of KTN [32]. The migration of K⁺ from and back to the K-rich layer might be the mechanism for the storage and reduction of NO_x in those catalysts, which will be discussed deeply in this paper. We also compared NSR behaviors of various alkali-metal-nitrate-impregnated KTN catalysts and found that the KNO₃-impregnated catalyst showed the best performance. The XPS and XRD analyses of NaNO₃-impregnated Pt/KTN catalysts revealed that the migration of K⁺ from KTN bulk to the surface nitrate salts proceeds more predominantly than the migration of Na⁺, which could be the reason why KNO₃-impregnated KTN exhibited the best catalytic performance. We have extended our study to CeO₂ and ZrO₂ supports since they have unique catalytic behavior for oxygen transfer and storage, and we investigate NSR behavior of Pt-K/CeO₂ and Pt-K/ZrO₂ catalysts in this article.

Since NO is one of the main atmospheric pollutants that can cause acid rain, photochemical smog, depletion of stratospheric ozone, and greenhouse effects, the removal of NO has been one of the hottest topics around the world. Great efforts have been focused on the exploitation of novel kinds of catalysts to realize practical applications or to illuminate the reaction mechanism for NSR processes [33,34]. Among various proposed catalysts, ion-exchanged zeolites have shown particularly high catalytic activities for NSR processes, which presenting high catalytic performance at low temperatures [35,36,37]. Additionally, notable attention has been concentrated on the fundamental studies of the mechanism of the process by using in situ FT-IR, XPS, TPD, and so on. A significant amount of fundamental research using supported noble or transition-metal catalysts has been carried out to study the nature of catalytic removal of NO by H₂ or CO [38,39,40]. Transition-metal-containing zeolites are the most promising SCR catalysts due to their high performance in selective catalytic reduction of nitrogen oxides by hydrocarbons on Ni-ZSM-5 and Co-ZSM-5, etc. [41,42,43]. In these reports, NO and CO have been employed as the probe molecule to monitor the surface of transition-metal ions [41]. The adsorbed species with high stability have been identified as intermediates contributing to the catalytic efficiency. IR spectroscopy, especially in situ IR, provides significant information about the nature of the bonds formed between the surface of catalysts and adsorbates [43,44,45,46].

Several years ago, we used nanoporous nickel phosphate VSB-5 for the first time as the active component for NO and CO adsorption [47]. To the best of our knowledge, it was the first time that nickel was used as a catalyst with all nickel ions located in the framework for the reaction, which is different from Ni species in ion-exchanged zeolites. Nickel phosphate VSB-5 (Versailles-Santa Barbara) is a nanoporous material with a 24-ringopen framework and high thermal stability [15]. In contrast to traditional aluminosilicate zeolites, VSB-5 has 1-D channels with a larger pore size of 1.1 nm, and can realize the construction of nanostructure or can be a pathway for large molecules. Additionally, the different valences and various coordination numbers of the nickel ions make VSB-5 one of the most promising materials for a wide range of applications. For example, the unsaturated Ni²⁺ coordinate sites in the framework of VSB-5 are accessible to H₂ molecules in the pores, making VSB-5 one of the best candidates for hydrogen storage material in comparison with ZSM-5 and active carbon. However, although more and more attention has been focused on VSB-5, the understanding of active, unsaturated Ni²⁺ sites is not clear yet, which may hinder the further application of VSB-5 materials. Additionally, to the best of our knowledge, the use of NO or CO as probe molecules for investigating active sites in the pores of VSB-5 has never been reported, which may also benefit the application of VSB-5 in the de-NOx process.

2. Results and Discussion

2.1. NSR Behavior over Various Nano Structured Catalysts

2.1.1. Potassium Cation Transfer Mechanism in NO_x Storage and Reduction Processes over Pt/K-Titania Nanobelt Catalysts

Since the pioneering work of NO_x storage-reduction (NSR) catalysts by Toyota’s research group, many studies have been reported on the mechanism of NSR reaction over Pt-BaO/Al₂O₃ catalysts. It is generally accepted that oxidation of NO to NO₂ over Pt is a necessary initial step before NO_x storage can take place over BaO, forming Ba(NO₃)₂ surface layers on the bulk BaO particles. Recently, we have found that a Pt/KNO₃/K-titanate nanobelt (KTN) catalyst has the highest NO_x storage capacity (2.3 mmol/g-cat) among those reported in the literature, with excellent NSR performance. We have studied the mechanism of this NSR process by kinetic measurements, in situ XRD, TEM, XPS, and FT-IR measurements during the reaction and found that the main NO_x storage phase was a KNO₃-like composition, which transformed into K-rich surface layers of KTN in the reduction process. A novel “K cation transfer mechanism” was proposed from and back to K-rich titanate surface layers during storage-reduction cycles.

Figure 1 is the TEM image of the Pt/20 wt%KTN catalyst after H₂ reduction, which was a needlelike nanobelt structure several μm in length and a few nm in width. The Pt metal particles were highly dispersed on all the catalysts, with the particle sizes around 1–2 nm. The nanobelt structure was maintained completely up to 26–33 wt% KNO₃ loading after the reaction. However, most of the nanobelts were destroyed on spent Pt/41 wt%KNO₃/KTN. Combining the results of isothermal NO_x storage experiment and TEM observation, it is concluded that preservation of the nanobelt structure, as well as the addition of suitable amounts of KNO₃, were the key factors to obtain high NO_x storage capacity.

Time courses of isothermal NO_x storage are shown in Figure 2 over varied catalysts at 350 °C. The storage process includes two stages: at the first stage, the NO_x was completely trapped, no outlet NO being detected; at the second stage, NO_x was partially trapped, the outlet NO being increased concomitantly with time. The time for the complete NOx absorption stage was elongated with the increase in added KNO₃ until 33 wt% and then shortened. No other nitrogen species, such as NO₂ and N₂O, were detected during these processes. Therefore, the total amount of NOx was estimated from this area, which is called QMS estimation.

Figure 3 shows the dependence of the NO_x storage capacity upon the added amount of KNO₃ on KTN. The highest storage capacity of 2.27 mmol/g was achieved when 33 wt% KNO₃ was added to Pt/KTN. TEM images showed that the nanobelt structure was completely maintained even after the redox cycle, when the amount of modified KNO₃ was lower than 26 wt%. However, a small amount of nanobelt was collapsed to amorphous potassium-titanate particles after reaction over Pt/33%KNO₃/KTN, and most of the nanobelt was transformed into an amorphous form in the case of Pt/41%KNO_3/KTN. Accordingly, optimization of the amount of added KNO₃ is the key factor to keep the nanobelt structure, as well as to obtain higher NO_x storage capacity.

To more deeply evaluate the NO_x storage and reduction rates, as well as the stability of the catalysts, lean–rich cycling tests were conducted, as shown in Figure 4, under 6 min storage and 4 min reduction cycles. During the NO_x storage process, no NO was detected within 6 min, while when the gas flow was switched to H₂, only N₂ was detected as N species during reduction cycle, indicating the excellent NOx trapping capacity with high stability.

As shown in Figure 5, when a mixed gas of NO and O₂ was introduced to the reduced sample at 573 K, a strong IR band of adsorbed NO₃⁻ species was observed at 1369 cm⁻¹, which disappeared quickly just after replacing the gas phase with H₂, showing the fast response of NSR processes.

In situ XPS measurements were carried out to identify NO_x-storage and H₂-reduced states more clearly. After NO_x storage, the existence of surface KNO₃ species was confirmed by strong K 2p (292.9 and 295.5 eV), N 1s (407.1 eV), and O 1s (532.7 eV) transition peaks. However, the XPS peaks of Ti 2p and Pt 4f transitions were scarcely observed, indicating that the KTN support surface was mostly covered with KNO₃ overlayers. After in situ reduction by H₂, the N 1s peak disappeared; instead, the intensities of Ti 2p (457.8 and 463.5 eV) and Pt 4 f (69.5 and 73.0 eV) peaks were increased significantly, indicating the complete removal of KNO₃ from the surface of KTN support. At the same time, the binding energy of the K 2p peak shifted to the same position as KTN support (292.0 and 294.8 eV), although the intensity ratio of (K 2p) / (Ti 2p) was greater than that of KTN itself. These results clearly indicate that in the reduction process, potassium cations may be transferred into KTN bulk, forming K-rich surface layers of KTN.

Figure 6 summarizes a mechanism of the preparation of active phase and NO_x storage/reduction over Pt/KNO_3/KTN catalysts. During the storage process, NO is oxidized on Pt, forming NO₂, which reacts with K cations of the K-rich surface layer to form KNO₃. Formed NO_x storage species may easily move over the surface and finally cover the entire surface of KTN, achieving high NO_x storage capacity. During the NO_x reduction process, the formed KNO₃ is reduced to N₂ and H₂O, and K cations move back to KTN, forming K-rich surface layers of KTN again.

2.1.2. Activity and Selectivity-Controlling Factors for NO_x Storage-Reduction Process by Pt-Alkali Metal Nanocomposites Supported on TiO₂ with Different Crystal Structures

As described in the previous section, we have found that K-titanate nanobelt- (KTN: K₂Ti₈O₁₇) supported Pt/KNO₃ NSR catalysts exhibited a high NO_x storage capacity and excellent redox behavior. In this section, we employed different crystal-structured TiO₂ with various loading amounts of Pt (0.5–5 wt%) and MeNO₃ (Me = Na, K, Rb; 1.0–40 wt%) to elucidate the activity and selectivity-controlling factors for NSR processes.

The K-titanate nanobelt (K₂Ti₈O₁₇: KTN) was prepared by hydrothermal treatment using TiO₂ (P-25) and KOH. Various amounts of Pt and KNO₃ were loaded on KTN (350 m³), ST-01(Anatase; 270 m²)), MT150 A (Rutile; 100 m²), and P-25(A/R = 0.8; 50 m²) supports by a conventional impregnation method. After reduction of the catalyst (0.1 g) at 623 K, a NO_x storage process (lean; 80 mL/min of 930 ppm NO/7% O_2/He at 623 K) was performed on a gas-flow system, and the output gas was analyzed by mass spectroscopy, followed by a rich process of 80 mL/min of 4%H₂/He flow. The lean-rich cycle experiments were carried out using an interval of 6 min of NO_x (930 ppm) and 4 min of H₂ (4%) with 80 mL/min flow.

Table 1. Full trapping period, NO stored amount and N₂ desorbed.

Catalysts	Surf. Area. (Pore vol.)	MeNO3 Loaded (wt%)	Pt Size TEM (nm)	Full Trapping Period (min)	NO Stored (mmol/g)	N2 Desorb. (mmol/g)
TiO2(A) [ST-01]	275 m2/g (0.63 cm3/g)	K (20)	2–3	2	0.92	0.45
		K (33)		18	2.45	1.11
TiO2(R) [MT150 A]	101 m2/g (0.36 cm3/g)	K (20)	3–4	5	0.77	0.40
		K (33)		1	1.71	0.66
TiO2(A/R) [P-25]	54 m2/g (0.10 cm3/g)	K (20)	3–4	13	1.53	0.56
		K (33)		8	1.28	0.67
K2 Ti8 O17 [KTN]	320 m2/g (0.01 cm3/g)	K (20)	1–2	25	1.57	0.67
		K (33)		34	2.53	1.10
		Na (28)	2–3	23	1.50	0.71
		Rb (27)	3–4	16	1.28	0.59

summarizes NO_x storage and reduction properties of 1.5 wt% Pt-Alkali metal (Pt:Me = 1:31 & 1:61) catalysts supported on various carriers. The results indicated that full NO trapping periods and NO stored amounts were strongly dependent on surface area, pore volume, and Pt particle size. The KTN- and ST-01- (anatase) supported catalysts showed more efficient catalytic behavior than the others.

Figure 7 shows the lean-rich cycle experiments over 1.5 wt%Pt-33 wt% KNO₃/KTN, and capacity was completely recovered by a 4 min H₂ reduction, indicating an excellent NO_x trapping capacity with high stability. However, in the latter case, a considerable amount of NO was detected at the later stage of the lean cycle, indicating an important role for the ordered-layer structure of K cations.

Table 2. Binding-energy (eV) changes during NO_x stored and reduction processes of 1.5 wt% Pt-33 wt% KNO₃/KTN and 1.5 wt%Pt-33 wt%KNO₃/TiO₂(ST-01) by in situ XPS.

Catalysts	Condition	Pt 4f7/2	Ti 2p3/2	K 2p3/2	N 1s	O 1s
Pt-KNO3/KTN	NOx stored	— (0)	457.7 (0.01)	292.7 (2.3)	407.1	532.8 (1)
	H2 reduced	69.2 (1)	457.7 (1)	292.2 (1)	—	529.4 (1)
Pt-KNO3/TiO2 (ST-01)	NOx stored	73.7 (0.8) 70.5 (0.2)	457.8 (1.1)	292.4 (0.9)	406.9	529.2 (0.8) 532.1 (0.2)
	H2 reduced	70.5 (1)	458.0 (1)	292.7 (1)	—	530.0 (1)

summarizes the in situ XPS results of the NO_x-storage and H₂-reduced states. In the case of KTN-supported catalyst, only N 1s, O 1s, and K 2p transitions assignable to KNO₃ were observed in the NSR state, without any peaks assignable to KTN. Accordingly, NO is oxidized into NO₂ and combined with the K-rich layer to form KNO₃, which covers the entire surface of KTN support.

In the reduced state, all the transitions assigned to KNO₃ disappeared completely, indicating a reduction of KNO₃, and K⁺ moved back to KTN, forming a K-rich layer. On the other hand, in the case of TiO₂ (ST-01) supported catalysts, the Ti 2p and K 2p peaks did not change much during NO_x redox cycles, indicating poor dispersion of K cations on the surface, which would be the reason for their lower NO_x storage capacity and slower response. Table 2 summarizes the in situ XPS results of the NO_x-storage and H₂-reduced states. In the case of KTN-supported catalysts, only N 1s, O 1s, and 2p transitions assignable to KNO₃ were observed in the NSR state, without any peaks assignable to KTN. Accordingly, NO was oxidized into NO₂ and combined with the K-rich layer to form KNO₃, which covered the entire surface of the KTN support. In the reduced state, all the transitions assigned to KNO₃ disappeared completely, indicating the reduction of KNO₃, and K⁺ moved back to KTN, forming a K-rich layer. On the other hand, in the case of TiO₂- (ST-01) supported catalysts, the Ti 2p and K 2p peaks did not change much during NO_x redox cycles, indicating poor dispersion of K cations on the surface, which would be the reason for their lower NO_x storage capacity and slower response. The following three factors may be important for efficient NSR catalysts in Pt-alkali metal/TiO₂ systems: (1) smaller Pt particles for NO oxidation and KNO₃ reduction, (2) facile migration of alkali cations from and back to subsurface layers during redox cycles, and (3) the existence of a certain surface structure to maintain high dispersion of alkali cations.

2.1.3. Mechanistic Study of NO_x Storage and Reduction Behavior by In Situ XPS and FTIR over Pt/K/CeO₂ and Pt/K/ZrO₂ NSR Catalysts

Since the pioneering work of NO_x storage-reduction (NSR) catalysts by Toyota’s research group, many studies have been reported on the mechanism of NSR reaction over Pt-BaO/Al₂O₃ catalysts. The generally accepted mechanism for this reaction is “NO₂ transfer from Pt to BaO” forming Ba(NO₃)₂ surface layers. As previously mentioned, we have proposed a novel “K⁺ transfer mechanism from and back to K-rich titanate surface layers” for Pt/KNO₃/K-titanate NSR catalysts [33]. In the present study, we investigated the mechanism of NSR processes over Pt/KNO_3/CeO₂ and Pt/KNO₃/ZrO₂ by means of in situ XPS and FT-IR during the reaction. We found that under lean conditions, KNO₃-like species are stored below the PtCeO₃(PtO·CeO₂) top layers, and under rich conditions, K⁺ may be stabilized in the K₂ Ce₂O₄ (K₂O·Ce₂O₃) top layers in a way similar to the “K⁺ transfer mechanism”.

Figure 8 shows the time courses of outlet NO during the NOx storage process at 350 °C. Over ceria-supported catalysts, the full trapping period and the amounts of stored NOx increased with the increase in added KNO₃. Figure 9 shows the lean-rich cycle experiments over Pt/2.5 K/CeO₂ and Pt/3.5 K/CeO₂ catalysts. In the case of 3.5 mmol of KNO₃-added catalyst, almost all of the NO_x was stored for 6 min, and the capacity was completely recovered by a 4 min H₂ reduction, indicating an excellent NO_x trapping capacity with high stability. However, over 2.5 mmol of KNO₃-added catalyst, a small amount of NO was detected at the later stage of the lean cycle.

Table 3. Full trapping period (min), Amount of stored NO and Desorbed N₂ (mmol/g-cat).

Catalysts	Trapping Period	NOx Stored	N2 Desorbed
Pt/2.5 K/ZrO2	20.7	1.2	0.7
Pt/3.5 K/ZrO2	18.3	1.3	0.8
Pt/4.9 K/ZrO2	15.8	1.5	0.9
Pt/2.5 K/CeO2	18.6	1.1	0.7
Pt/3.5 K/CeO2	24.1	1.6	0.9
Pt/4.9 K/CeO2	28.6	2.4	1.5

summarizes the full NO trapping periods during a lean process, total amounts of stored NO_x, and desorbed N₂ over various catalysts. Both ZrO₂ and CeO₂-supported catalysts exhibited excellent NRS behavior at 350 °C. In the case of CeO₂-supported catalysts, the full trapping period and the amount of stored NO_x increased with the increase in added KNO₃, and the highest storage capacity of 2.4 mmol/g with the longest trapping period, of 29 min, was achieved in the case of 4.9 mmol of KNO₃-added catalyst. The particle size of Pt was very small and could not be detected by TEM after H₂ reduction. On the contrary, over ZrO₂-supported catalysts, a decrease in the trapping periods and nearly constant NO_x storage amounts were observed with the increase in added KNO₃.

In situ XPS measurements were carried out to identify NO_x-stored and H₂-reduced states (Figure 10 and

Table 4. B.E.(eV) changes during NO_x NSR processes of Pt-3.5 K/CeO₂ by in situ XPS. (Numbers in parentheses show relative intensity ratio of oxidized and reduced catalysts.).

	Pt4 f7/2	K2 p3/2	Ce3 d5/2	N 1s	O 1s
Reference	71.0	292.7	883.1	398.1	533.2
NOx stored	72.3 (5)	292.7 (0.3)	889.0 (1.2)	407.1	532.6, 529.0
H2 reduced	70.6 (1)	293.6 (1)	885.6 (1)	-	531.6, 529.0

). After NO_x storage, the existence of KNO₃ species was confirmed by K 2p(292.7 eV), N 1s(407.1 eV) and O 1s (532.6 eV) transition peaks, although their intensities were not as strong as those of Pt 4f (72.3 eV;Pt²⁺) and Ce 3d (889.0 eV;Ce⁴⁺) peaks. These results suggest that KNO₃-like species are stored below the PtCeO₃(PtO·CeO₂) top layers. After in situ reduction by H₂, the N 1s peak disappeared with the large increase in K2 p (293.6 eV) and the shift of Ce 3d(88.5 eV;Ce³⁺) peaks. At the same time, the Pt4 f (70.6 eV;Pt⁰) peak decreased significantly, indicating the complete removal of KNO₃, together with the reduction of Pt. Accordingly, under rich conditions, K⁺ may be stabilized as K₂Ce₂O₄ (K₂O·Ce₂O₃) on the top layers of ceria support and phase transfer between PtCeO₃(lean), and K₂Ce₂O₄(rich) surface nanocomposites is responsible for high capacity and durability, which is realized by facile combination of Ce³⁺⇔ Ce⁴⁺ and Pt⁰ ⇔ Pt²⁺ redox cycles. On the contrary, over Pt-K/ZrO₂, because of the absence of a Zr³⁺⇔ Zr⁴⁺ redox cycle, sintering of PtO₂ (lean) and K₂O(rich) easily takes place during lean-rich cycles, which causes the deactivation.

2.1.4. Comparison of NO_x Storage and Reduction Behaviors among Ceria-Supported Pt-KNO₃, Cu-KNO₃, and Co-KNO₃ NSR Catalysts

As mentioned in the previous section, we have reported that Pt and a KNO₃-loaded K-titanate nanobelt (K₂Ti₈O₁₇) exhibited the highest NO_x storage capacity (2.3 mmol/g), at 350 °C, among the relevant literature and proposed a novel K cation transfer mechanism from and back to K-rich titanate surface layers during storage-reduction cycles [33]. We extended our study to a KNO₃-added Pt/CeO₂ system, which also exhibited high NO_x storage capacity and good redox cycle behavior as NSR catalysts. Moreover, we substituted Pt with Cu or Co and compared their catalytic behaviors by applying a kinetic investigation, as well as in situ XPS and FT-IR measurements during the reaction. It is revealed that under lean conditions, KNO₃-like species are stored underneath the MeCeO₃ (MeO·CeO₂; Me = Pt, Cu or Co) top layers, and under rich conditions, K⁺ may emerge to the surface and stabilize in the K₂Ce₂O₄ (K₂O·Ce₂O₃) top layers in a way similar to the “K⁺ transfer mechanism of Pt/KNO₃/K-titanate nanobelt”.

Various catalysts were prepared by a conventional impregnation method. After reduction of catalysts (0.2 g) at 350 °C, a NO_x storage process (lean; 80 mL/min of 930 ppm NO/7%O₂/He) was performed on a gas-flow system, and the output gas was analyzed by mass spectroscopy, followed by a rich process of 80 mL/min of 4%H₂/He flow. The lean-rich cycle experiments were carried out.

Table 5. Full trapping periods, amounts of stored NO, and desorbed N2 (mmol/g-cat.) at 350 °C.

Catalysts	Trapping Period	NOx Stored	N2 Desorbed
Pt-2.5 K/CeO2	19 min	1.1	0.6
Pt-3.3 K/CeO2	24 min	1.6	0.9
Pt-4.9 K/CeO2	29 min	2.4	1.1
Cu/1.3 K/CeO2	16 min	1.9	0.8
Co/1.3 K/CeO2	15 min	1.1	0.5

summarizes the full NO trapping periods during a lean process, total amounts of stored NO_x, and desorbed N₂ over various catalysts at 350 °C. In the case of Pt(0.08 mmol/g-cat.)- xKNO_3/CeO₂ catalysts, the full trapping period and the amount of stored NO_x increased with the increase in added KNO₃(x mmol/g-cat.), and the highest storage capacity of 2.4 mmol/g with the longest trapping period of 29 min was achieved in the case of 4.9 mmol of KNO₃-added catalyst. It is worth noticing that Cu- and Co-loaded catalysts showed better NSR behaviors than the Pt-loaded catalyst, probably because of the participation of Cu or Co cations as NO_x storage species. The Pt particles were very small and could not be detected by XRD or TEM, whereas weak XRD peaks of Cu or Co metals were detected after H₂ reduction.

Figure 11 shows the lean-rich cycle experiments at 350 °C over Pt(0.08 mmol)-KNO₃ (3.5 mmol)/CeO₂ and Cu or Co(0.88 mmol)-KNO₃(1.3 mmol)/CeO₂ catalysts. In the cases of Pt and Cu catalysts, complete NO_x removal and recovery by H₂ reduction were observed, whereas a small amount of NO was detected at the later stage of the lean cycle in the case of Co catalysts. In situ XPS measurements were carried out to identify NO_x-stored and H₂-reduced states (

Table 6. Binding-energy (eV) changes during NO_x stored and reduction processes of Pt-2.5 K/CeO₂ and Cu-1.3 K/CeO₂ by in situ XPS.

Catalysts	Condition	Pt 4f7/2	Cu 2p3/2	K 2p3/2	Ce 3d5/2	N 1s	O 1s
	reference	71.0	932.7	292.7	883.1	398.1	533.2
Pt/KNO3 /CeO2	NOx stored	72.3 (5)		292.7 (0.3)	889.0 (1.2)	407.1	532.6, 529.0
	H2 reduced	70.6 (1)		293.6 (1)	885.6 (1)	-	531.6, 529.0
Cu/KNO3 /CeO2	NOxstored		933.7 (10)	292.7 (0.1)	889.0 (2.5)	406.9	532.5, 529.0
	H2 reduced		932.6 (1)	293.6 (1)	885.6 (1)		531.7, 529.0

). After NO_x storage, the existence of KNO₃ species was confirmed by K 2p(292.7 eV), N 1s(407.1 eV), and O 1s(532.6 eV) transition peaks, although their intensities were not as strong as those of Pt 4f(72.3 eV;Pt²⁺), Cu 2p (933.7 eV; Cu²⁺) and Ce 3d(889.0 eV; Ce⁴⁺) peaks. These results suggest that KNO₃-like species are stored in the buried form below the top layers. After in situ reduction by H₂, the N1 s peak disappeared with the large increase in K2 p(293.6 eV) and the shift of Ce 3d (88.5 eV;Ce³⁺) peaks. Accordingly, under rich conditions, K⁺ may be stabilized as K₂ Ce₂O₄ (K₂O·Ce₂O₃) on the top layers of ceria support.

Cu/KNO₃/CeO₂ and Co/KNO₃/CeO₂ catalysts exhibit excellent NSR behavior comparable with Pt/KNO₃/CeO₂, which is desirable for practical use by substituting a precious metal with base metals. Facile K⁺ transfer through the subsurface layers of CeO₂ may be responsible for these superior catalytic performances.

2.1.5. Preparation of Nanoporous Nickel(II) Phosphate VSB-5 and Their Adsorption and Catalytic Behavior in NO-CO Reactions

VSB-5 can be synthesized between pH7.3 and 11.0 and over a wide range of Ni/P ratios and concentrations, employing various organic amines or inorganic basic salts, or even with aq. NH₃ as basic reagents. They exhibit quite similar XRD patterns, with the molecular formula of Ni₂₀ [(OH)₁₂(H₂O)₆] [(HPO₄)₈(PO₄)₄]·12 H₂ O, although their crystal morphologies and inner surface structure of micropores seem to be rather different, with various preparation conditions. In the present study, we prepared two kinds of catalysts, VSB-5(TEA) and VSB-5(NH₃), by employing two extremely different basic reagents, triethylamine (TEA) and NH₃. Changes to their morphology and pore structures were studied by N₂ and CO physisorption at 77 K, NO and CO adsorption at higher temperatures, and the catalytic reduction of NO by CO at 323–573 K. The effect of NaOH treatment and the effects of substituting Ni with Co and Fe ions were investigated on NO adsorption and catalytic behavior of NO-CO reaction. The role of pores and the mechanism of NO-CO reaction were investigated by employing FT-IR spectroscopy.

The results of various catalyst characterizations are summarized in

Table 7. Surface Area, adsorption amount, P/Ni ratios and initial rates of the reaction at 523 K.

Bases (Treat.)	S.A.	Ads.amt. (77 K)		NO(a).amt. (cm3 g−1)		SEM/ EDX	NO-CO Initial Rate (10−7 mol/min−1)
	m2 g−1	N2	CO	293 K	353 K	P/Ni	CO2	N2 O	N2
VSB-5(TEA)	15.0	2.5	3.8	4.7	4.4	0.595	1.4	2.5	1.0
VSB-5(NH3)	318.0	93.5	106.0	36.1	35.3	0.625	3.0	3.0	2.5
TEA(NaOH treat.)	28.4	5.8	13.5	16.0	15.0	0.718	14.0	13.0	2.5
NH3(NaOH treat.)	108.0	32.1	36.0	12.6	12.3	0.477	20.0	14.0	33.0

. The surface area of VSB-5 (TEA) (15.0 m² g⁻¹) estimated from N₂ physisorption was abnormally small compared with that of VSB-5(NH₃) (318.0 m² g⁻¹), suggesting that the end of needle-like crystals of VSB-5(TEA) must be blocked by large amounts of phosphate anions, along with the fringe of the end. After NaOH treatment, its surface area increased two-fold due to the removal of end-capped phosphate anions from the fringe. In the case of VSB-5(NH₃), the surface area decreased by 1/3 by NaOH treatment.

Figure 12 shows SEM photos of various VSB-5 before and after NaOH treatment. The crystal size of VSB-5 (TEA) was one order of magnitude larger than that of VSB -5(NH₃), the structure of which did not change much by NaOH treatment. However, in the case of VSB-5(NH₃), an apparent collapse of the morphology began to take place, which could be the reason for the decrease in surface area accompanied by the considerable decrease in P/Ni ratio in SEM/EDX measurements.

Figure 13 shows the adsorption isotherms of NO at 353 K over VSB-5(TEA) and VSB-5(NH₃) before and after NaOH treatment. The rate of NO adsorption (solid line) was rather slow over all the catalysts investigated. Especially in the case of VSB-5(TEA), it required a few days to reach equilibrium of adsorption. This may also correspond to the assumption that the end of the crystal must be blocked. The rate of NO desorption was even slower (broken line), and a certain hysteresis phenomenon was observed between adsorption and desorption curves. As summarized in Table 7, the extrapolated amount of adsorbed NO was correlated well to the surface area of VSB-5, although almost independent of the adsorption temperatures, indicating a strong and activated NO adsorption.

It is interesting to note that the effect of NaOH treatment for NO adsorption was completely opposite between VSB-5(TEA) and VSB-5(NH₃). In the former case, the amount of NO adsorption increased by three times, whereas it decreased by half in the latter case, which corresponds well to the changes of p/Ni ratios by NaOH treatment as follows: from 0.595 to 0.718 in the former case and from 0.625 to 0.477 in the latter. Based on these experimental results, it is reasonable to suppose that there is a strong electronic interaction between adsorbed NO and phosphate anions inside the micropores. In the last columns of Table 7, initial rates of product formation were summarized in NO-CO reactions over various VSB-5 catalysts at 523 K. At the initial stage of the reaction, the main products were CO₂ and N₂ O, with a lower amount of N₂, indicating the successive reduction mechanism of NO into N₂ through an N₂ O intermediate. After NaOH treatment, the initial rates increased by nearly one order of magnitude, indicating the increase in active sites by NaOH treatment.

Figure 14 shows the time courses of NO-CO reaction over VSB-5(TEA) together with 20 mol% Co- and Fe-substituted catalysts at 523 K. Over Co-substituted VSB-5, the reaction rate was enhanced several times, and N₂ O was formed quickly at the initial stage, showing maximum at 30 min. After that, the amount of N₂ was increased with the decrease in N₂ O, indicating the successive reduction mechanism. The profile of time course was completely different in the case Fe-substituted VSB-5, and N₂ was the main product from the beginning, with a small amount of N₂ O, indicating the operation of a direct reduction pathway of NO into N₂.

Two kinds of VSB-5 were prepared by employing TEA and NH₃ as bases. The crystal size of VSB-5 (TEA) was one order of magnitude larger than that of VSB-5(NH₃), with a surface area one order of magnitude smaller. The end of the former crystal must be blocked by large amounts of phosphate anions, which were partially removed by NaOH treatment. Adsorption and desorption isotherms of NO at 293–353 K showed the hysteresis curves, indicating that the inner space of micropores must be crowded by phosphate anions. However, once NO molecules can get into the pores, they may be stabilized by a certain electronic interaction with phosphate anions. These catalysts were very active for NO-CO reaction at 323–573 K with the successive reduction mechanism. Substitution of Ni²⁺ ions with Co²⁺ and Fe²⁺ largely enhanced their catalytic activity, indicating that coordinately unsaturated metal²⁺ ions in the framework of VSB-5 are the active sites for this catalytic reaction.

2.1.6. Unusual Storage of NO and Its Efficient Removal by Various Reducing Agents inside One-Dimensional Micropores of Nickel (II) Phosphate VSB-5 Catalysts

Nickel (II) phosphate VSB-5 is a novel microporous material with a 24-ring open-framework of NiO₆ octahedra (1-D channels), which has outstanding properties and high thermal stability. In the present study, we prepared two kinds of catalysts, VSB-5(TEA), VSB-5(NH₃), by employing extremely different basic reagents, trimethylamine (TEA) and ammonia (NH₃). Substitution of Ni²⁺ with other metal²⁺ ions (V, Fe, Co) was carried out to prepare efficient catalysts for NO storage and removal by various reducing agents, such as H₂, CO, NH₃, and CH₄.

As summarized in

Table 8. Surface area, adsorption amounts, and initial rates of various reactions at 523 K.

Catalysts	S.A.	NO(a).amt. (cm3 g−1)		Initial Rates of Product Formation (10−7 mol/min−1)
	m2 g−1	293 K	353 K	NO-H2 Reaction		NO-CO Reaction		NO-CH4 Reaction
				N2	N2O	N2	N2O	N2(N2O)	CH3CN
VSB-5(TEA)	15.0	4.7	4.4	14.3	2.3	3.5	0.8	0.6	0.2
VSB-5(NH3)	318.0	36.1	35.3	18.6	3.5	5.8	1.4	1.3	0.5

, the surface area of VSB-5 (TEA) estimated from N₂ physisorption was very small compared with that of VSB-5(NH₃), suggesting that the end of needle-like crystals of VSB-5(TEA) must be blocked by large amounts of phosphate anions, along with the fringe of the end. The rates of both NO adsorption and desorption were very slow over both catalysts, and a certain hysteresis phenomenon was observed between adsorption and desorption curves, suggesting that the inner space of micropores must be crowded by phosphate anions. However, once NO molecules can get into the pores, they may be stabilized by a certain electronic interaction with phosphate anions and activated for catalytic reactions.

In the last three columns of the table, initial rates of product formation are summarized during NO-H₂, NO-CO, and NO-CH₄ reactions over both catalysts at 523 K. In the case of NO-H₂ and NO-CO reactions, the main products were N₂, CO₂ and H₂O, with a lower amount of N₂ O. In the NO-CH₄ reaction, a considerable amount of CH₃CN was formed additionally. Substitution of Ni²⁺ ions with V²⁺, Fe²⁺, or Co²⁺ enhanced their catalytic activities by more than one order of magnitude, indicating that well dispersed, coordinately unsaturated metal²⁺ ions in the framework of VSB-5 are active for these catalytic reactions.

2.1.7. Effect of Co(II), Fe(II), and V(II) Substitution upon Unusual NOx Storage and Efficient Removal by Various Reducing Agents inside 1-D Channels of Microporous Nickel (II) Phosphate Catalysts

As already shown in Table 8, the surface area of VSB-5 (TEA) (15.0 m² g⁻¹) estimated from N₂ physisorption was abnormally small compared with that of VSB-5(NH₃) (318.0 m² g⁻¹), suggesting that the end of needle-like crystals of VSB-5(TEA) must be blocked by large amounts of phosphate anions, along with the fringe of the end. In the case of VSB-5(NH₃), the surface area decreased by 1/3 by NaOH treatment. The SEM photos of various VSB-5 showed that the crystal size of VSB-5(TEA) was one order of magnitude larger than that of VSB-5(NH₃). On the other hand, XRD patterns of all the samples did not change much, showing characteristic diffraction patterns assignable to typical VSB-5. The rate of NO adsorption was rather slow over all the catalysts investigated. Especially in the case of VSB-5(TEA), it required a few days to reach equilibrium of adsorption. The rate of NO desorption was even slower, and a certain hysteresis phenomenon was observed between adsorption and desorption curves. As can be seen in the table, the extrapolated amount of adsorbed NO was correlated well to the surface area of VSB-5, although almost independent of the adsorption temperatures, indicating a strong and activated NO adsorption in these temperature ranges. From these results, it is reasonable to suppose that there is a strong electronic interaction between adsorbed NO and phosphate anions inside the micropores. In the last columns of the table, initial rates of product formation are summarized in NO-CO reactions over various VSB-5 catalysts at 523 K. At the initial stage of the reaction, the main products were CO₂ and N₂ O, with lower amounts of N₂, indicating the successive reduction mechanism of NO into N₂ through N₂ O intermediate. Figure 15 shows the time courses of NO-CO reactions over VSB-5(TEA), together with 20 mol% Co- and Fe-substituted catalysts at 523 K. Over Co-substituted VSB-5, the reaction rate was enhanced several times, and N₂ O was formed quickly at the initial stage, showing its maximum at 30 min. After that, the amount of N₂ was increased with the decrease in N₂ O, indicating the successive reduction mechanism. The profile of time course was completely different in the case Fe-substituted VSB-5, and N₂ was the main product from the beginning, with a small amount of N₂ O, indicating the operation of a direct reduction pathway of NO into N₂. Two kinds of VSB-5 were prepared by employing TEA and NH₃ as bases. The crystal size of VSB-5(TEA) was one order of magnitude larger than that of VSB-5(NH₃), with a surface area one order of magnitude smaller. Adsorption and desorption isotherms of NO at 293–353 K showed hysteresis curves, indicating that the inner space of micropores must be crowded by phosphate anions. However, once NO molecules can get into the pores, they may be stabilized by a certain electronic interaction with phosphate anions. These catalysts were very active for NO-CO reactions at 323–573 K with the successive reduction mechanism. Substitution of Ni²⁺ ions with Co²⁺ and Fe²⁺ largely enhanced their catalytic activity, indicating that coordinately unsaturated metal²⁺ ions in the framework of VSB-5 are the active sites for this catalytic reaction.

2.1.8. Catalytic Reduction of NO by H₂ over Co-Substituted Nanoporous Ni Phosphates, VSB-5

The removal of NO remains an urgent environmental problem. Great effort has been devoted to the development of new efficient catalysts to realize practical applications and to understand an intrinsic reaction mechanism. Recently, nickel phosphate VSB-5 and its cobalt substituent have been synthesized, possessing one-dimensional open frameworks (D = 1.1 nm) and exhibiting some interesting catalytic behaviors. In the present study, we first studied the catalytic reduction of NO by H₂ over these materials, applying kinetic investigation as well as FT-IR observation during the reaction.

Nickel phosphate, VSB-5, and cobalt- (10–20% molar ratio) substituted Ni-phosphate (Co-VSB-5) were synthesized in hydrothermal conditions according to the literature. The catalytic NO-H₂ reaction (10 Torr (0.18 mmol) each, 200 mg catalyst) was carried out in a closed gas-circulation system with an online analysis by GC. FT-IR spectroscopy was employed to observe adsorbed species during the reaction. The catalysts were characterized by XRD, XPS, TEM, TG, TPD, and N₂ adsorption techniques.

Co-VSB-5 has a large amount of water molecules in its pores, and to obtain the catalytic activity, it is necessary to remove a certain amount of water molecules by evacuation at higher temperatures (Figure 16).

Accordingly, the catalyst was first evacuated at 300 °C for 20 hrs before reaction. As shown in Figure 17, the catalytic activity of VSB-5 for NO-H₂ reaction was low. However, substitution of Ni with Co enhanced the reaction rates by more than one order of magnitude. The amount of adsorbed NO was also increased by higher-temperature evacuation up to 350 °C (Figure 18), which corresponded well to the catalytic activities.

It is worth noting that 20 mol% Co-VSB-5 catalysts were very active for the hydrogen dissociation process, and H₂–D₂ exchange reaction took place even at −80 °C, as shown in Figure 19. An HZ (Co-VSB-5)-D₂ hydrogen exchange reaction also took place around 150 °C, suggesting the incorporation of surface OH or HPO₃ groups into the reaction with NO. As shown in Figure 19C, hydrogen dissociation was significantly inhibited by the presence of NO, suggesting competitive adsorption of NO and H₂ at the active sites.

3. Mechanistic Investigation and Design of NSR Catalysts

As is well known, conventional three-way catalysts (TWC) are highly efficient in reducing NO_x, CO, and unburned hydrocarbon emissions, though only within a narrow window region around stoichiometric air/fuel ratio air/fuel (A/F = 14.7). However, under lean-burn conditions, which are favorable to reduce CO₂ emissions and to realize high fuel-efficiency, NO_x emissions fall to very low levels in the case of TWA catalysts. To improve this situation, NO_x storage-reduction catalysts (SR) are promising. Effective NO_x removal techniques for lean-burn exhaust are eagerly desired, and NO_x storage–reduction (NRS) catalysts have been recognized as promising. In the case of the original Pt-Ba/Al₂ O₃ catalysts first reported by Toyota’s research group, an NOx storage-reduction mechanism can be formulated as follows:

Lean-burn condition; 2 NO + O₂ → 2 NO₂ (on Pt), 2 NO₂ + BaO → Ba(NO₃)₂

Rich-burn condition; Ba(NO₃)₂ +4 H₂ → N₂+ 4 H₂ O + BaO.

NSR catalysts are composed three essential elements: (1) a high-surface-area metal oxide, such as Al₂ O₃, as a support, (2) basic elements, such as Ba or K, for NO_x storage sites, and (3) precious-metal redox sites, such as Pt, Pd, and Rh, for the redox reaction for oxidizing NO to NO₂ and reducing stored NO_x to N₂ by various reductants. Correspondingly, the following three factors may be important as efficient NSR catalysts in Pt-alkali metal/metal oxide support systems: (a) smaller Pt particles for NO oxidation and (alkali metal) -NO₃ reduction, (b) facile migration of alkali cations from and back to subsurface layers during redox cycles, and (c) the existence of a certain surface structure to maintain high dispersion of alkali cations. In this section, detailed NSR mechanisms over various nanostructured catalysts will be discussed by changing these essential factors.

3.1. Transition-Metal Redox Sites

Pt plays several key roles in the NOx storage/reduction reaction steps, such as NO oxidation under lean-burn conditions, NOx storage under lean-burn conditions, and NOx reduction under rich-burn conditions. As mentioned in the previous section, we extended our KTN catalysts to a KNO₃-added Pt/CeO₂ system, which also exhibited high NO_x storage capacity and good redox cycle behavior as NSR catalysts. Moreover, we substituted Pt with Cu or Co and compared their catalytic behaviors by applying a kinetic investigation, as well as in situ XPS and FT-IR measurements during the reaction.

Cu/KNO₃/CeO₂ and Co/KNO₃/CeO₂ catalysts exhibit excellent NSR behavior comparable with Pt/KNO₃/CeO₂, which is desirable for practical use by substituting a precious metal with base metals. Facile K⁺ transfer through the subsurface layers of CeO₂ may be responsible for these superior catalytic performances. Accordingly, under rich conditions, K⁺ may be stabilized as K₂Ce₂O₄ (K₂O·Ce₂O₃) on the top layers of ceria support, and phase transfer between PtCeO₃(lean) and K₂Ce₂O₄(rich) surface nanocomposites is responsible for high capacity and durability, which is realized by the facile combination of Ce³⁺⇔ Ce⁴⁺ and Pt⁰ ⇔ Pt²⁺ redox cycles. On the contrary, over Pt-K/ZrO₂, because of the absence of a Zr³⁺⇔ Zr⁴⁺ redox cycle, sintering of PtO₂ (lean) and K₂O(rich) easily takes place during lean-rich cycles, which causes the deactivation.

3.2. NOx Trapping Sites

BaO is the most commonly studied NOx trapping component among metal oxides, with a major contribution to the overall NOx trapping capacity. Besides the NOx storage step, BaO also plays a role in the NO oxidation, NOx release, and NOx reduction steps. Although different metal-oxide supports and precious-metal components affect the functions of BaO in NSR reactions, in this section, we still focus on the general functions of BaO and of the Pt/BaO/Al₂ O₃ catalyst in NSR reactions.

(a)

NO oxidation step; as reviewed in the previous section, BaO presents some negative effects in NO oxidation as a result of its blockage of the Pt surface and its stabilization effect on Pt oxides (inactive for NO oxidation).

(b)

NOx storage step; because of the complexity of this step, it is difficult to identify and differentiate mechanisms that occurred during the NOx trapping process. Both nitrites and nitrates are detected on the NSR catalyst after NOx trapping. Therefore, it is well accepted that there must be a nitrite route and a nitrate route for the NOx trapping on BaO. Within the nitrite route, it is proposed that NO is oxidized on Pt sites and directly trapped by nearby BaO sites to form Ba nitrites. The Ba nitrites are finally oxidized to Ba nitrates. Within the nitrate route, it is proposed that NO is oxidized to NO₂ on Pt sites. NO₂ spills over to the BaO site to form Ba nitrate with the evolution of NO. When Ba loading is high or low, there are more or fewer Ba sites that are close to the Pt sites, which leads to the predominance of the nitrite route or nitrate route. At high temperatures, no matter whether NO or NO₂ is used as the NOx precursor, similar NOx trapping capacity is observed on the Pt/BaO/Al₂ O₃ catalyst. This is because NO can be easily oxidized to NO₂ at high temperatures. However, at low temperatures, the NOx-trapping capacity of BaO is low when using NO as a NOx precursor due to limited NO oxidation into NO₂. With NOx trapping BaO, the H₂O and CO₂ in the exhaust could react to produce barium hydroxide and carbonate. The thermodynamic stability of BaCO₃ is higher than that of Ba (OH)₂. In contrast to BaO, in order to trap NOx, Ba(OH)₂ and BaCO₃ must decompose first and then react with NOx. Therefore, the presence of H₂O and CO₂ in the exhaust gas phase will affect the rate of NOx trapping. In addition, CO₂ decreases more NOx-trapping capacity than H₂O.

(c)

NOx release step; the release of NOx is very important for efficient regeneration of NOx-trapping sites and the following NOx reduction process. NOx release is induced by the decomposition of Ba nitrites and Ba nitrates upon high temperature and rich excursion (reductant introduction). The stability of Ba nitrites and Ba nitrates decreases with increasing temperature and decreasing O₂ partial pressure, causing their decomposition so as to release NOx. Apart from the reductant gases and O₂, the presence of other gases, such as H₂O and CO₂, also affects NOx release during the rich-burn cycle. The presence of H₂O reduces NOx release, while CO₂ increases NOx release.

3.3. Surface Structure for Storage

In the cases of VSB-5-type NSR catalysts, once NO molecules can get into the pores, they may be stabilized by a certain electronic interaction with phosphate anions. These catalysts were very active for NO-CO reaction at 323–573 K with the successive reduction mechanism. Substitution of Ni²⁺ ions with Co²⁺ and Fe²⁺ largely enhanced their catalytic activity, indicating that coordinately unsaturated metal²⁺ ions in the framework of VSB-5 are the active sites for this catalytic reaction.

3.3.1. Adsorption-Desorption Behavior of NO and CO on Nanoporous VSB-5

A large amount of NO and/or CO can be adsorbed into the pores of VSB-5, and strong and sharp bands were observed by in situ FT-IR investigation. They exhibit unusual behavior in the adsorption and desorption process. The amount of adsorption could be promoted by increasing the adsorption temperature, which could produce more active sites. The adsorption of CO and NO is competitive, and the adsorbed CO may be replaced by NO due to the stronger bonding strength of the latter. In this section, in situ FT-IR and temperature-programmed desorption (TPD) techniques were employed to investigate the adsorption/desorption behaviors of NO and CO. Following the pretreatment of the sample by 250 °C evacuation, 30 torr of NO was introduced, and the spectrum was taken at room temperature, as shown in Figure 20 (RT). Two peaks (1873 and 1858 cm⁻¹) and a shoulder (1886 cm⁻¹) were observed in the in situ FT-IR spectra. These three bands may be assigned to the mononitrosyl species adsorbed on three kinds of Ni²⁺ at specific positions corresponding to three kinds of Ni in the framework of VSB-5. The main peak (1873 cm⁻¹) may be assigned to NO adsorbed on Ni²⁺, while the weak peaks at 1858 and 1886 cm⁻¹ may be attributed to the NO adsorbed on Ni^2−δ and Ni^2+δ, respectively. The Ni^2−δ may result from the Ni²⁺ coordinated with more electrically negative species, such as oxo species, while the Ni^2+δ may come from the Ni²⁺ linked with less electrically negative species.

The sample was heated to different temperatures under an NO atmosphere and kept for 30 min at each temperature. Then, the temperature was lowered to room temperature in the presence of NO, and the FT-IR spectra were recorded, as shown in Figure 20 (150, 200, 250 °C). The integrated intensity of absorbed species increased about 14% when the temperature was raised to 150 °C, suggesting that more NO was adsorbed on some sites at high temperatures, which were inaccessible at lower temperatures. It is known that five kinds of OH groups exist in the channels of VSB-1 and interact with each other during the heating process. Considering the similarity of the structure, it is reasonable to suppose that the heating process may cause the interactions between P-OH groups. These interactions had an influence on the coordination of nickel cations and produced some unsaturated sites, providing more active locations for the adsorption of NO within the channels. However, the integral intensities decreased when the temperature was higher than 200 °C. After heating to 250 °C, the intensity was lower than that at room temperature. This may be caused by the destruction of some NO adsorption sites during the further heating process. Additionally, decomposition of NO presented, and the formation of N₂ and N₂ O was also observed during the NO adsorption at higher temperatures in the online analysis of GC.

Different amounts of NO with varied pressures were introduced after pretreatment and adsorbed for 30 min at room temperature. Then, the FT-IR spectra were taken, as shown in Figure 21. Two peaks and a shoulder were observed at 1875, 1858, and 1886 cm⁻¹, respectively, under the pressure of 7 × 10⁻² torr NO. The intensities of the absorption peaks increased a little with increasing pressure, and the integrated intensity for 30 torr was 1.54 times more than that of 7 × 10⁻² torr, although there is not much difference between the adsorption at 10 torr and 30 torr. The position of the main band shifted from 1875 to 1873 cm⁻¹ when the pressure was increased to 1 torr and remained unchanged with increasing pressure.

This shift suggests that the adsorbed NO species may interact with the surrounding species from the framework of VSB-5 and form different kinds of mononitrosyl species. It was reported that NO preferred to form dinitrosyl species on the Ni²⁺ surface, while no dinitrosyl species were observed in the above experiments. One possible reason may be that the Ni²⁺ ions were highly occupied, and no more vacancies were available for dinitrosyls. The other possible reason may be the decomposition of dinitrosyl complexes to linear mononitrosyl species at ambient temperature. At the same time, the band at 1858 cm⁻¹ became more obvious and changed into a clear band when the pressure increased to 1 torr. These bands had very narrow full width at half maximum (FWHM), suggesting that the Ni sites were homogeneously distributed, as proposed by the reported literature.

Figure 22 shows the desorption spectra of adsorbed NO after the above heating-up/cooling-down adsorption process. These adsorbed species could be desorbed by evacuation, and more species were removed with the rise of evacuating temperature due to thermal behaviors. The weak band at 1886 cm⁻¹ was less stable and converted to a weaker shoulder after evacuation at 100 °C. A total of 75% of the integral intensity was lost by evacuating for 30 min at 150 °C. As for the species adsorbed at room temperature, similar behaviors were observed during the desorption process (spectra not shown). However, the adsorbed species were less stable compared to those adsorbed at high temperatures, and 94% of integral intensity was lost after evacuating for 30 min at 150 °C. Complete removal of the species was achieved by evacuation at 200 °C. These results indicate that the species adsorbed at raised temperatures have higher stability, and their stability may come from the components adsorbed at the sites inaccessible at low temperatures.

3.3.2. Isotopic ¹⁵NO Adsorption

To investigate the adsorption/desorption behaviors in more detail, labeled ¹⁵NO was employed following the pretreatment. The introduction of 10 torr ¹⁵NO at room temperature afforded the spectrum shown in Figure 23 (curve a). Similar adsorption characteristics as those of ¹⁴NO presented, and two main bands and a shoulder with narrow FWHM were observed at 1842, 1826, and 1853 cm⁻¹. The positions of these bands showed about 30 cm⁻¹ red shift compared with those of ¹⁴NO, which confirmed that these adsorbed bands were from the NO species. Three bands could be assigned to the mononitrosyl species, and no dinitrosyl species were observed, which is in correspondence with the results of ¹⁴NO adsorption. An increase in the pressure or temperature under the NO atmosphere caused behaviors analogous to those of ¹⁴NO adsorption/desorption (spectra not shown).

Following the adsorption of ¹⁵NO at ambient temperature and the following removal of gaseous ¹⁵NO by evacuation, 10 torr ¹⁴NO was introduced, and the adsorption was operated at a different temperature. The spectra were recorded at room temperature, as shown in curves b-e in Figure 23. Intense adsorption bands attributed to the characteristics of ¹⁴NO were observed with the simultaneous decrease in isotopic ¹⁵NO bands. The decrease in intensity of isotopic ¹⁵ NO after the introduction of ¹⁴NO indicates that adsorbed isotopic ¹⁵NO species were replaced by the gas phase of ¹⁴NO. The loss of intensity grew with the increase in temperature, and almost all the absorbed ¹⁵NO species were replaced after the temperature rose to 200 °C. At the same time, the intensities of adsorbed ¹⁴NO species increased with temperature due to more sites being available at high temperature or being released by the dissociation of labeled ¹⁵NO.

However, the absorbed ¹⁵NO species became stronger once the sample was heated to 150 °C under 10 torr of ¹⁵NO atmosphere. The spectrum is shown in Figure 24 (curve a). ¹⁴NO was then introduced after the gaseous isotopic ¹⁵NO was removed by evacuation, and the spectra are shown in Figure 24 (curve b-f). Adsorption peaks from the ¹⁴ NO species emerged, and the intensity of ¹⁵NO species simultaneously decreased a little. However, the intensity of ¹⁵NO species was still stronger than that of ¹⁴NO species adsorbed at room temperature, which is different from the result shown in Figure 24 (curve b). The intensities of ¹⁴NO increased with the rise in temperature, and the maximum arrived by heating to 200 °C, while at the same time, the intensities of ¹⁵NO species decreased. However, obvious bands of ¹⁵NO were still observed, even when the temperature was raised to 250 °C. This further confirms that the species absorbed at raised temperatures was much more stable than those adsorbed at room temperature.

3.3.3. CO Adsorption

Figure 25 shows the in situ FT-IR spectra of adsorbed CO after procedure to those those of NO adsorption. One main peak at 2194 cm⁻¹ was observed after the introduction of 10 torr CO at room temperature (curve at RT). Following the adsorption at room temperature for 60 min, the sample was heated for 30 min at corresponding temperatures and then cooled down to room temperature. The deconvolution indicates that this main peak was composed of two peaks, at 2185 and 2195 cm⁻¹, respectively, as shown in the inset. These peaks may be assigned to the Ni²⁺-CO species adsorbed at different Ni sites. No additional carbonyl peak was observed due to its weaker bond strength compared to that of the nitrosyl species.

The intensities of main peaks increased with prolonged adsorption time, indicating that more CO was adsorbed during this period (spectrum was not shown). The integral intensity of the peaks increased with the rise of temperature and reached the maximum at 200 °C. This indicates that CO might arrive at inaccessible sites at lower temperatures, which might be achievable during the heating process. Increasing the pressure or raising the temperature could promote the integral intensities and stability, which is in accordance with the behaviors of NO adsorption. This confirms that the heating process assists CO molecules to reach inaccessible sites at low temperatures. The maximum amount of CO adsorption was realized when the temperature was raised to 200 °C, which is also similar to the case of NO. A sharp decrease in the intensity for species adsorbed at ambient temperature was observed after evacuation at room temperature, and complete removal of the adsorbed species was completed after evacuation at 150 °C, which demonstrates that the CO species adsorbed at room temperature were unstable and also weaker than the NO species. These results, together with the fact that the stretching frequency was higher than that of gaseous CO (2143 cm⁻¹), indicate that the VSB-5 matrix had high electrophilicity and σ-carbonyls may be formed.

3.3.4. Co-Adsorption of NO and CO

Figure 26 shows the spectra of co-adsorption experiments of NO and CO. A total of 5 torr of CO and NO, each, was added simultaneously to the catalyst disk following pretreatment. A strong peak at 1876 cm⁻¹ and two shoulders attributed to the mononitrosyl species were observed, which is the same as those observed for NO adsorption in the absence of CO. Only a very weak band at 2196 cm⁻¹, attributed to carbonyls, appeared, indicating the adsorption of CO was blocked by NO. The intensity of the Ni²⁺-NO species increased when the temperature was raised to 150 °C, while the intensity of Ni²⁺-CO species remained as before. This confirms that the adsorption of NO and CO was competitive, and the formed nitrosyl species were stronger than the carbonyls.

To obtain more information, NO preadsorption was conducted firstly by introducing NO (10 Torr) and kept at ambient temperature for 30 min. After removing gaseous NO by evacuation, CO (10 Torr) was introduced and liquid nitrogen was used to trap all the time. The results are shown in Figure 27.

Carbonyl species were formed, and intense band at 2196 cm⁻¹ was observed at room temperature. The intensities of carbonyls increased with rising temperatures, while the intensities of nitrosyls decreased at the mean time. The reappearance of the carbonyls demonstrated that CO adsorption could take place under low-equilibrium NO pressure conditions as liquid nitrogen acted as a cold trap (P_NO = 7 × 10⁻² torr). This further verifies that the adsorption of NO and CO was competitive and may take place at the same site. NO can be catalytically reduced by CO at high temperatures, based on the results of the online GC analysis in a closed gas-circulation system. A certain amount of N₂, N₂ O, and CO₂ was produced at 250 °C. The intensities of adsorbed NO did not decrease much, even when the temperature was higher than 200 °C, suggesting that the adsorbed species might not be the direct reaction intermediate; however, the adsorbed compounds should act as reservoirs of the NO-CO reaction.

In this study, VSB-5 was first used as an active component for the adsorption of NO and/or CO and demonstrated catalytic activity for reduction of NO by CO. A large amount of NO and/or CO can be adsorbed in the micropores of VSB-, and novel adsorption behaviors were first observed. Strong and sharp adsorption bands of mononitrosyl species were observed at ambient temperature. An increase in temperature might produce more uncoordinated sites that were inaccessible at low temperatures for adsorption; this resulted in the increase in the quantity and intensity of adsorbed species. Species adsorbed at high temperatures became more stable than those adsorbed at ambient temperature, and they are difficult to be replaced by the labeled gas. The adsorption of CO and NO is competitive, and the adsorption of CO may be blocked by NO since the latter has stronger bond strength.

4. Experimental

4.1. Catalyst Preparation

The K-titanate nanobelt (K₂ Ti₈ O₁₇: KTN) was prepared using hydrothermal treatment, with TiO₂ and KOH as starting materials. In XRD measurements of the obtained KTN, all reflection peaks were assigned to a monoclinic phase of K₂ Ti₈ O₁₇ (JCPDS No. 84–2057). Nanobelts several micrometers in length and tens of nanometers in width with bending and twisting features were observed using TEM. Specifically, 3 g of TiO₂ was added to 100 mL of 10 mol/L KOH aqueous solution and stirred for 30 min. The suspension was transferred to the autoclave, sealed, and heated to 130 °C for 4 days. The white solid product was collected by filtration, washed with deionized water several times, and finally dried at 100 °C overnight.

Alkali metal nitrates of various kinds (designated as MNO₃: M = Li, Na, K, Rb, and Cs) and 1.5 wt% Pt were dispersed onto the KTN support using a conventional impregnation method. The impregnation amounts of MNO₃ were, respectively, 13 (M = Li), 17 (Na), 20 (K), 27 (Rb), and 33 wt% (Cs), which corresponded to 2 mmol/g on catalysts. Mixtures of the KTN and aqueous solutions of H₂ PtCl₆ and MNO₃ were put into an oven at 70 °C until complete evaporation of water was achieved, with subsequent calcination at 350 °C for 2 h. The catalysts were designated as Pt-xxMNO₃/KTN, where xx signifies the weight percent of MNO₃.

Nickel phosphate VSB-5 (Versailles-Santa Barbara) catalysts were prepared by the following preparation conditions: in the cases of VSB-5(TEA), Co-VSB-5(TEA), and Fe-VSB-5(TEA), the molar ratios of Ni: M: P: TEA: H₂ O = (1-x): x: 0.92: 3: 200 with pH = 10 (x = 0.2 for Co and Fe), and in the case of VSB-5(NH₃), the molar ratio of Ni: P: H₂ O= 1: 0.63: 100 with NH₃ aq adjusting pH = 7.7. After stirring for 1 hr at room temperature, the solution was put into an autoclave and kept at 442 K for 4 days. NaOH treatment was carried out following calcination of dried samples at 623 K by stirring in 1 M aq. NaOH solution (pH = 13.2) for 6 h.

4.2. Catalyst Characterization

Adsorption isotherms were taken by a Belsorp MAX apparatus at 77 K (N₂, CO), 293 K (NO), and 423 K (NO). The samples were thoroughly evacuated at 627 K for 24–36 h before the measurement. NO-CO reaction (10 Torr each) was carried out with a glass-made closed gas-circulation system with online analysis by G.C.

The XRD patterns were recorded using a diffractometer (MultiFlex; Rigaku Corp.) with a Cu K1 X-ray source (50 kV, 30 mA). Then, X-ray photoelectron spectroscopy (XPS) was conducted using a spectrometer (JPM-9010 MC, JEOL). The XPS apparatus had a preparation chamber in which a sample could be heated in situ and reduced by hydrogen (in situ hydrogen reduction). The C1 s (284.3 eV) peak was used as the standard reference for binding energies.

The IR spectra of adsorbed species were measured as follows: self-standing pellets (50 mg) were pressed and treated in situ in the IR cell. The FT-IR spectra were recorded at room temperature on a JASCO FT/IR 660 instrument equipped with an MCT detector at a resolution of 4 cm⁻¹ with 512 scans. Before adsorption, NO was further purified by passing it through a liquid-nitrogen cold trap and additional fraction distillation. The wafer was heated at different temperatures under vacuum for 1 h as pretreatment. The intensity of the adsorbed species at room temperature increased with the pretreatment temperature, and the intensity pretreated at 250 °C was about 5 times that of the intensity evacuated at 150 °C. The pretreatment temperature was set as 250 °C, considering the thermal stability and the removal of the species blocked in the micropores. The adsorption/desorption time was set at 30 min for all the FT-IR experiments

For TPD measurement, 300 mg VSB-5 powder was put into a quartz tube. As a pretreatment, the sample was heated to 250 °C with a ramp of 5 °C min⁻¹ and kept for 1 h under He flow (50 mL min⁻¹). The adsorption of NO (0.1 vol% NO balanced with He) was operated at room temperature or at 150 °C for 30 min. The amount of adsorbed NO was calculated from the D-value between the input and output. The following TPD was operated as below: 50 mL min⁻¹ He flowed for 15 min to remove the freely gaseous NO; then, the temperature was raised (5 °C min⁻¹) under He flow (50 mL min⁻¹); the out-coming species were recorded by quadruple mass spectrometer (QME200, PFEIFFER).

4.3. Procedure for NOx Storage/Reduction

The catalytic tests were performed on a fixed-bed gas-flow system equipped with a pulse gas-feed system, with an online quadruple mass spectrometer (QMS) as detector. Before every catalytic test, the sample was reduced by 80 mL/min 8%H₂/He flow in the quartz tube reactor from room temperature to 350 °C, with a ramp rate of 10 °C/min, and maintained at 350 °C for 30 min. During this reduction procedure, outlet gas was analyzed by a QMS and used for temperature program reduction (TPR) profiles. The signal of He was used as reference to calculate the relative signal intensity of other gases. For example, the relative signal intensity of H₂ was calculated as I_m/e = 2/I_m/e = 4. For the isothermal NO_x storage test, 80 mL/min 930 ppm NO/7%O₂/He was flowed through 100 mg catalyst at 350 °C, and the outlet gas was analyzed by QMS. After the adsorption, the gas flow was switched to He for 15 min flushing. Then, the NO_x storage catalyst was reduced by 4% H₂/He at the same temperature, and the outlet gas was recorded by QMS. After flushing with He for 15 min, the processes of NOx storage and reduction were repeated, and the data obtained from the second time was used for analysis.

For the lean-rich cycling test, 50 mg of catalyst was used. The lean-rich cycles (6 min of 930 ppm NO/7% O₂/He and 4 min of 4% H₂/He with 160 mL/min flow rate) were adopted to check the kinetic property and stability of the catalysts. To simulate the gas composition of mobile exhaust during rich condition, 930 ppm NO/4% H₂/He was also used as reductant for lean-rich cycling test. In this case, 160 mL/min 930 ppm NO/7% O₂/He flow was flowed through 50 mg catalyst for 3 min and then switched to 160 mL/min 930 ppm NO/4% H₂/He flow for 30 sec.

The hydrogenation of CO was conducted in a closed gas-circulation system with a liquid nitrogen cold trap, along with the circulation loop, which simplified the product analysis and which prevented secondary reactions, such as olefin hydrogenation and water–gas shift reaction. The reaction products were analyzed using online gas chromatography (TCD and FID). The catalyst was reduced by 200 torr of H₂ at 573 K for 5 h before reaction. The CO–H₂ reaction was conducted at 473 K with 1:2 ratios of CO and H₂ (total pressure = 150 torr).

5. Conclusions and Perspective

Since the introduction of the NOx storage-reduction (NSR) concept in 1995 by Toyota, many studies have been carried out to obtain a better understanding of the NSR mechanism and to obtain more efficient materials and improved engine control for the NOx abatement. In the present article, we first addressed the overview of the development of the NSR catalysts, followed by the application of nanostructured materials for NSR processes. In the case of Pt-KNO₃/KTN catalysts, we proposed a noble K+ migration mechanism from the K-rich surface layer to the bulk KTN, which achieves a fast response of storage and reduction processes with high capacity. We also applied microporous VSB-5 as the NO storage nanospace, which showed activated adsorption and desorption behavior and advantages for NSR processes. We also tried to substitute the Ni²⁺ ion of the VSB-5 framework with CO²⁺ and Cu²⁺ ions, which caused a drastic increase in the NSR ability, with one order of magnitude higher NSR capacity. However, it is desirable to discover more stable and affordable materials for effective NSR catalysts in the near future.

In conclusion, we summarize the important future studies for NSR catalysis as follows:

(1)

Although there is a wide general agreement concerning the mechanism of NSR processes, diversified and sometimes contradictory results have also been reported, mainly due to different experimental conditions used by the various researchers, such as different lean and regeneration periods, different reducing gases, different catalyst preparation and pretreatment, and different mass- and heat-transfer conditions. NOx storage is a sequential process (initially, nitrite and eventually, nitrate formation), which still does not clearly state the nature of the Ba species, nor does it account for the NOx slip during the storage process. The NOx reduction mechanism and the product selectivity under rich conditions are dependent on the reductants. NH₃, which has recently been considered to be an intermediate reductant, has not been universally observed and accepted by researchers.

(2)

The noble metal Pt is widely used by researchers for the oxidation of NO and the subsequent reduction of stored NOx. However, the nature of the Pt particles, such as the loading, dispersion, size and the morphology, and metal-support interaction influencing the catalytic activity are still the subject of debate. There is still no conclusive understanding of the optimum loading, shape, and dispersion of Pt particles for the best catalytic activity. Investigations have been carried out with other noble metals, such as Pd and Rh. Furthermore, the potential of metal alloys needs to be explored.

(3)

The storage component best investigated and commonly used is Ba; however, the nature of the storage component is still not clear. According to some researchers, BaCO₃ is the dominant storage component under real conditions, whereas others claim that BaO is a better potential storage component than BaCO₃. Researchers have revealed that the proximity of Ba and the noble metal is important in the storage process; however, systematic investigations are still lacking regarding this aspect. Overall, it is accepted that BaAl₂ O₄ is formed at higher temperatures during the NSR operation, although the exact role of BaAl₂ O₄ is unknown. Storage materials, other than Ba, such as K, Mg, and Ca showed promising performance in the storage step, as well as less sulfur poisoning, which needs to be explored more profoundly.

(4)

Besides stabilizing the dispersed noble metal particles, the support material can also affect the oxidation state of the noble metal and the spillover process, which has an effect on the oxidation, storage, and regeneration (reduction) steps, and thus, the overall performance of the NSR catalyst. For Ba-based lean traps, NOx storage capacity seems to follow the basicity of the oxide support, while the rate of NOx reduction tends to decrease with stronger basicity of the support. An open question is the role of the support in the spillover processes occurring in the NSR system. Furthermore, for a deeper understanding of the surface processes during NSR, the structure and chemistry of the interfaces of the support with catalytic (noble metals) and storage components (alkali and alkaline earth metals) need to be investigated more thoroughly, particularly under dynamic conditions.