Challenges and breakthroughs in post-combustion catalysis: how to match future stringent regulations

P. Granger
Unité de Catalyse et de Chimie du Solide, Université de Lille Sciences et Technologies, 59650 – Villeneuve d'Ascq, France. E-mail:

Received 16th May 2017 , Accepted 21st June 2017

First published on 22nd June 2017

This short overview briefly summarizes the prominent evolutions and scientific breakthroughs in the development of end-of-pipe technologies with respect to the standard regulations of atmospheric pollutant emissions from automotive exhaust. Up to now, several strategies have been implemented to fulfill more and more stringent emission limits and to overcome the extensive use of critical materials. This led to the elaboration of cheaper and more compact systems with close-coupled technologies suitable for light duty vehicles and also to preserve the competitiveness of car manufacturers. But all these improvements could not be sufficient to face short-term, more severe limitations as well as more realistic test driving cycles since the actual ones usually underestimate the emissions of current atmospheric pollutants especially from diesel powered engines. Presently, several scenarios have been envisioned which account for the partial replacement of liquid fossil fuels by alternative fuels for urban travel and the emergence of hybrid engines powered by natural gas or bio-fuels accompanied by more simple and efficient end-of-pipe systems. In this short overview, particular attention was paid to the past and present academic contributions which led and/or inspired important technical and commercial applications.

image file: c7cy00983f-p1.tif

Pascal Granger

Pascal Granger (born in 1964) is Distinguished Professor at the University of Lille and a member of the French Chemical Society. His research is focused on reaction mechanisms and kinetics of catalytic reactions involved in post-combustion catalysis and catalytic DeNOx and DeN2O processes for stationary sources. His research interest is also focused on the development of perovskites as potential substitutes of PGM. He is a co-editor of two Wiley Books dedicated to perovskite materials (2016) and one book series entitled ‘Past and Present in DeNOx catalysis’ (2007) published by Elsevier.

1. Past, present, and future needs in emission control of automotive exhaust

There is still a growing demand for more efficient technologies to clean up air partly due to the introduction of tougher standards which could attain technical limits for catalytic end-of-pipe technologies. Indeed, a high efficiency, considering trace amounts of atmospheric pollutants, unsteady-state operating conditions, substitution of critical materials such as platinum group metals (PGM) and rare earth elements, and a longer lifetime, could be an unattainable goal. Academia faced significant scientific obstacles in the past two decades to improve both the efficiency and the durability of catalytic systems especially those dedicated to lean-burn applications. Presently, new constraints have arisen for improving the life cycle and recyclability of end-of-pipe catalysts with more simple and green methods for their preparation as starting points.1,2 In addition, all types of motorization systems will be affected by the implementation of new regulations on atmospheric pollutant emissions i.e. gasoline and diesel engines. All these requirements led to sophisticated technologies to clean up exhaust gas jointly with the use of reformulated fuels generating less harmful atmospheric pollutants and careful monitoring of combustion. Regarding this last point, significant improvements in engine design and fuel injection modes were obtained. New scenarios as well as the introduction of new motorization systems could comply with the implementation of stricter regulations. However, the operating window is usually narrow and could be finally operational for a short period of time.

1.1 Future need to meet Gasoline Direct Injection (GDI) emissions regulations

Diesel engines have proven their efficiency compared to gasoline engines in terms of yield with lower CO2 emissions. Simultaneously, serious obstacles associated with this technology were essentially due to soot and NOx emissions, with an important technological threshold with regard to the selective reduction of NOx to nitrogen in the presence of a large excess of oxygen. This issue was tackled in the Euro 6 regulations and represented a significant gap compared to Euro 5, as exemplified in Table 1.
Table 1 NOx limits of diesel cars and related emission control costs11
European regulations NOx limits (g km−1) Emission control cost ($ per vehicle)
Euro 3 0.5 −392
Euro 4 0.25 −513
Euro 5 0.18 −822
Euro 6 0.08 −1239

The recent expansion of Gasoline Direct Injection (GDI) technology is essentially related to significant advantages compared to conventional port fuel injection engines (PFI). This technology is under development but several advantages have already been identified such as fuel economy, gain in performance, better drivability and related lower CO2 emissions provided by the engine combustion mode.3,4 All these advantages have been summarized in a recent review article which essentially focused on the state of the art of exhaust particulate filter technology5 and could partly explain that the market share of vehicles equipped with GDI engines is estimated to grow by 40–60% in 2017, taking all new registered gasoline vehicles in Europe into account.6 The societal costs and benefits associated with the implementation of Gasoline Particulate Filters in GDI vehicles have already been discussed.7 Indeed, some significant drawbacks are also identified to be related to higher particulate number (PN) and particulate mass compared to PFI engines (see Fig. 1). The predominance of sub-20 nm primary particles renders soot formed from GDI engines potentially hazardous to human health.8 More stringent European standard regulations will set an upper limit of 6.0 × 1011 particulate number (PN) km−1 after September 2017 (ref. 9), which suggests strong technical efforts to comply with these new regulations. Practically, the basic Diesel Particle Filter (DPF) concept can be a good starting point, being a cost-effective solution.7 However, GDI engines produce a much lower particulate number compared to conventional diesel. As a consequence, a lower accumulation of soot induces detrimental effects on the particle filtration efficiency.10 Although some debates have arisen with regard to the health benefits related to more careful control of particulate number instead of mass emissions, the installation of efficient Gasoline Particulate Filters (GPFs) to control PN emission will be probably mandatory and could be a significant limitation on the development of such type of motorization.

image file: c7cy00983f-f1.tif
Fig. 1 Comparison of particulate number and particulate mass emissions from gasoline and diesel engines (reproduced with permission from ref. 5).

1.2. Alternative to current gasoline and diesel engines

Generally speaking, an important question arises regarding the evolution of the market share of alternative fuel powered engines. Indeed, the implementation of more stringent regulations with respect to atmospheric pollutants, i.e., NOx and particulate matter from gasoline and diesel engines, will likely induce additional cost investments to develop more efficient technologies. As a consequence, this could strongly affect the competitiveness of car manufacturers.

As illustrated in Table 1, the automotive industry already granted significant investments especially to match the emission limits for atmospheric pollutant emissions from diesel cars. Particular attention was paid to diesel engines especially within the Euro 6 standard regulations. As a consequence, sophisticated catalytic exhaust gas treatments for mobile sources have been elaborated, involving extensive utilization of critical materials, i.e., PGM such as platinum and rhodium and rare earth elements. Hence, securing reliable, sustainable and undistorted access to certain raw materials such as PGM is an important issue in European countries because the automotive industry is consuming 65–80% of the demand, which is predicted to grow due to the implementation of more stringent regulations within Euro 7.

In addition, the emission ceilings under simulated driving test conditions using the NEDC driving cycle underestimate those encountered under real vehicle running conditions. This is true for the amount of CO2 emitted (by 20–30%) and even more for NOx emissions for diesel vehicles.11 This led the European Commission (EC) to the implementation of more restrictive standard rules (Euro 6 up to Euro 6d standard) and to the definition of new testing protocols for light-duty vehicles through the Worldwide Harmonized Light-duty Vehicle Testing Procedure (WLTP). WLTP consists of chassis dynamometer tests for the determination of emissions and fuel consumption from light-duty vehicles especially those powered by diesel engines. This procedure will be complemented by Real Driving Emission (RDE) tests in agreement with the Euro 6d standard regulations with the aim of measuring the pollutant emissions, such as NOx, emitted by cars while driven on the road under “real life” conditions. Europe will be the first to introduce this RDE procedure which should ensure that cars deliver low emissions under on-road conditions. This new certification test should strictly comply with the 80 mg km−1 limit set for NOx emissions within the Euro 6 standard.

As a consequence, the adoption of more stringent regulations (Euro 7) together with more realistic certification tests could increase significantly the demand for platinum group metals to comply with future regulations and could strongly affect the use of diesel vehicles with the perspective of losing market share and favoring the emergence of alternative motorization systems coupled with simpler end-of-pipe technologies. In practice, the replacement of PGM is not trivial. Presently, the best practical issue which has arisen is the substitution of expensive PGM (Pt or Rh) by the cheaper palladium.

Future scenarios in France take into account a sharp decrease in the number of light-duty vehicles from 40 million in 2010 to 25 million in 2050 (ref. 12), thanks to the development of railroad and river transports. In parallel, new concepts of mobility and transportation modes are arising. This is likely because nearly 90% of the European population is exposed to health risks due to abnormally high exposure levels of particulate pollution of approximately one third above the permitted limits. To further reduce atmospheric pollutants, the regulation limit for CO2 emission of 130 g km−1 set by the European Commission (EC) in 2012 will be changed to 95 g km−1 in 2021. Hence, a decrease in liquid fossil fuel consumption by approximately 50% is expected, which implies the development of alternative fuels for urban travel and the emergence of hybrid engines powered by natural gas or bio-fuels. The development of natural gas engines offers an alternative to diesel vehicles in terms of lower particulate emissions due to the homogeneous combustion process and also to the emergence of new sectors related to the commercialization of gasoline direct injection engines that are more efficient in terms of CO2 emissions but unfortunately producing a larger number of soot particles than conventional spark ignition engines (see Fig. 1).13

Natural gas is abundant and low-cost with a limited environmental impact associated with its multiple use, securing the demand compared to the growing dependency on imported fuels.14,15 Consequently, the introduction of alternatives to liquid fuel in the US or heavy-duty vehicles is a fast growing market. Old NGV engines running under lean-burn conditions with oxidation catalysts to control CO and oxygenate emissions are progressively replaced by stoichiometric spark-ignited engines employing gas recirculation and three-way catalysts to comply with more stringent NOx emission standards.16 There is an urgent need to reorient car European manufacturers towards the development of alternative fuels and hybrid motorization systems generating lower CO2 atmospheric pollutants and particulate matter. In practice, the implementation of combined approaches including fuel optimization to after-treatment systems is mandatory. In this context, a high cost-efficiency is reachable by developing engines powered by natural gas. Natural gas is flexible and can adapt itself to every type of engine from light- to heavy-duty vehicles and is more competitive with a lower cost, i.e., lower than 50% with respect to gasoline and 30% for diesel.

2. Lean-NOx trap versus selective catalytic reduction: towards coupling technologies

2.1. Advantages and drawbacks of single technologies

2.1.1. The selective catalytic reduction of NOx by unburnt hydrocarbons. The selective catalytic reduction of NOx by unburnt hydrocarbons (HC-SCR) is likely the most feasible technology for the abatement of NOx emissions from diesel powered vehicles with two distinct methods. Unburnt hydrocarbons can be used for the reduction of NOx according to a passive control. Alternately, an active control can be implemented with the injection of added diesel fuel upstream of the catalytic converter. This latter option is currently implemented when operating with a modern turbo charged common rail injected diesel engine.17

Since the first discovery associated with the development of Cu–ZSM5 HC-SCR catalysts using alkene and alkanes as reducing agents, a wide number of catalysts have been developed, especially supported metallic catalysts. Previous investigations dealt with Cu2+ ion-exchanged ZSM-5 (Cu-ZSM-5) zeolites in order to elucidate both their NO decomposition and SCR activities.18 A critical point was their low thermal stability – the zeolite structure collapsing at high temperature. A recent investigation provided more insight towards a better understanding of deactivation phenomena under hydrothermal conditions19 associated with (i) reversible migration of copper species, (ii) partial dealumination and (ii) irreversible formation of diffuse CuAl2O4. The authors concluded that the formation of a stable inactive aluminate phase is initiated by Al extraction from the zeolite framework.

However, most lean-burn catalytic after-treatment systems exhibited major drawbacks in terms of efficiency because of too low concentrations of residual HC to reduce NOx in a large excess of oxygen. The most recent promising finding was related to silver-based catalysts prepared according to simple classical wet impregnation of alumina materials, drastically broadening the operating window to selectively convert NO into nitrogen.20–23 Unfortunately, their low sulfur tolerance illustrated in Fig. 2 led to unsuccessful practical developments with severe deactivation ascribed to strong adsorption of SO3 coming from SO2 oxidation.

image file: c7cy00983f-f2.tif
Fig. 2 NOx conversion on Ag/γ-Al2O3 as a function of reaction temperature before and after exposure to a SO2 or SO2/SO3 containing mix for 21 h (reproduced with permission from ref. 22).

Regarding the use of Platinum Group Metals (PGM), high efficiencies are usually reported depending on the composition of the support. Indeed, complete hydrocarbon combustion depends on the acidic properties of the supports24 and determining the temperature at which maximum NO reduction to nitrogen occurs. Among the existing varieties of catalysts, one of the key issues is related to better knowledge of elementary surface processes, especially those involving the metal–support interface. Obviously, such knowledge is a prerequisite for further improvements in terms of catalytic activity and selectivity. Burch et al. first explored the kinetics of the NO–C8H18–O2 reaction on Pt/Al2O3 (ref. 25) highlighting the importance of the Pt–Al2O3 interface. Crucial mechanistic information was reported with Pt sites mostly covered by dissociated oxygen species under lean conditions preventing NO dissociation (see Fig. 3). Hence, in a large excess of oxygen, once adsorbed, NO is readily oxidized to chemisorbed NO2 species further reacting with CxHy at the metal–support interface. Further investigations revealed more complex surface processes with preferential interactions between ad-NOx species and more reactive oxygenates coming from the partial oxidation of CxHy.

image file: c7cy00983f-f3.tif
Fig. 3 Mechanism scheme for the NO–C8H18–O2 reactions at 200 °C over Pt/Al2O3 (reproduced with permission from ref. 25).

This original kinetic study brought important information highlighting the positive reaction order of oxygen on the overall catalytic process over PGM catalysts, whereas the opposite trend is usually encountered on three-way catalysts (TWC). As a matter of fact, the importance of NO2 as a key intermediate was pointed out, with its formation being kinetically limited at low temperature. This academic finding likely inspired further practical achievements especially the implementation of diesel oxidation catalysts upstream of catalytic NOx abatement systems.

Several attempts illustrated the use of bulk perovskites as HC-SCR catalysts. However, the optimization of such materials remained on the lab scale, emphasizing the important challenge of obtaining good balance in redox properties especially in a large excess of oxygen. Basically, the ABO3 structure of perovskites can accommodate a wide number of components in A- and B-sites and can stabilize various distorted structures. This is probably the starting point regarding the use of perovskites for NOx reduction as well as for the oxidation of unburnt hydrocarbons. Theoretically, it is possible to tune the properties of these materials by controlling the degree of substitution in the A- and B-sites, the structural environment and the valence of the A and B components. In practice, it is hard to find a good balance although previous investigations reported to some extent some successful achievements on the lab scale. The relevant comparisons between LaCo1−xCuxO3 with previous investigations on active Cu-zeolites26 and Cu/MCM-41 (ref. 27) revealed high catalytic HC-SCR properties if low coordination isolated copper ions are stabilized. These authors found that the coordination and surface chemical environment of isolated copper species in zeolite substrates are of great importance. By analogy, the introduction of copper cations in the B site of LaCoO3 in a specific distorted geometrical environment led to significant enhancement in the catalytic properties. Additional examples come from the optimization of La1−xKxCo1−yPdyO3−δ composition, with x = 0, 0.1 and y = 0, 0.05, showing the stabilization of more active ionic Pd species for NO oxidation to NO2. In that case, it was proposed that unusually high oxidation states for Pd3+ or Pd4+, coordinated in a distorted octahedral environment, can be more active than isolated Pd2+ ions stabilized in a square-planar coordination symmetry.

In general, the development of perovskite-based materials for pollutant emissions control in post-combustion catalysis28,29 is often related to the capacity to enhance the formation of structural defects and stabilize cations in the B-site in unusual oxidation states. Further imbalances induced by partial substitution of La3+ by an alkali metal ion having a lower oxidation state will be compensated by an increase of the valence of B cations as well as the creation of oxygen vacancies. Incorporation of copper into potassium-containing perovskites also led to the improvement of thermal stability, especially concerning the potassium stability during the reaction at high temperature.30 Despite significant breakthroughs in controlling the properties of perovskite systems through appropriate substitutions, some key issues still remain for properly monitoring the formation of defective sites and preventing phase segregations at the surface. The nature and density of different oxygen species i.e. lattice O2− species, O or O2 from gaseous O2 adsorption on oxygen vacancies are key parameters influencing the distribution of reactivity at the surface. The evolution of surface composition can strongly influence the catalytic properties.

2.1.2. Towards more efficient catalytic technologies for the abatement of NOx.
The key role played by diesel oxidation catalysts (DOC). Diesel oxidation catalysts are essential in current automotive pollution control systems. Placed upstream of the current catalytic technologies for the abatement of NOx and particulate matter, DOC ensures that CO and unburnt hydrocarbons are efficiently converted through oxidation reaction. In addition, the formation of NO2 is promoted to improve the performances of other devices (NOx and soot removal). Accordingly, significant strategies were developed both for improving the DOC efficiency and developing on-board diagnostic systems capable of detecting deteriorations due to deactivation phenomena. Post-injection strategy approaches (active diagnostic) were found to be an efficient way to detect whether there is oxidation or not on DOC.31,32 Diesel particulate filters took advantage of this strategy with residual temperatures rising beyond 600 °C, which is suitable for their regeneration.

Generally, DOCs are composed of PGM and Pt/Al2O3 can be recognized as a benchmark system being more active than other PGM-based catalysts but sensitive to particle sintering under lean conditions and to poisoning effects.33–35 Indeed, it was found that NO oxidation to NO2 is altered by particle sintering whereas propene conversion would be more affected by sulfur poisoning.36 The presence of propene can also inhibit the rate of NO oxidation on Pt based catalysts. Subsequent Pd incorporation into Pt attenuated this effect by promoting the combustion of propene at lower temperature.37 For monolithic catalysts, improved performance can be attained on zone-coated Pt-based catalysts with increased flow rates and higher platinum concentration in the upstream portion compared to Pt homogeneously distributed along the reactor channels. The heat released from the exothermic combustion in the front part of the zone-coated catalyst can attenuate the propene inhibiting effect releasing Pt sites for NO oxidation to NO2.38 It was also predicted from kinetic modeling studies that CO also has a detrimental effect during engine cold start on Pt-based DOC. This effect is attenuated after Pd incorporation39 and is explained by changes in the pre-exponential factor of the reaction rate constant instead of changes in the activation energy for CO oxidation which remained constant irrespective of the Pt[thin space (1/6-em)]:[thin space (1/6-em)]Pd ratio. In fact, an increase in Pd sites enhances the rate of CO oxidation which occurs more readily on Pd than on Pt. The partial substitution of platinum by palladium also presents additional advantages. As previously explained, the replacement of costly Pt by cheaper palladium is presently the best economical option. In addition, Pd addition is assumed to improve the thermal resistance of Pt to particle sintering by forming a Pt–Pd alloy.40,41 Recently, another concept was investigated to slow down the rate of platinum sintering in air at high temperature (T = 800 °C). In fact, these operating conditions can favor the formation of volatile oxidic Pt species which accelerates the particle sintering according to the Ostwald ripening mechanism earlier described.42 The formation of volatile PGM species was previously pointed out on Pt-based catalysts, which can further diffuse and accumulate on SCR catalysts situated downstream of the DOC and damage their catalytic performances.43 Xiong et al. showed that those volatile Pt species can be trapped by PdO, leading to smaller Pd–Pt particles (∼20 nm instead of 500 nm for single Pt particles).41 Hence, Pd coexisting with Pt in alloyed particles would slow down Pt volatilization.44

Let us note that the thermal stability and related investigations to understand deactivation phenomena are often related to the methodology implemented, which to some extent can lead to distorted observations if aging under lab conditions is performed using simulated conditions instead of “real life” conditions for identifying the prevalence of sintering or poisoning effects. Careful control of the structural properties and surface composition of bimetallic Pt–Pd particles according to appropriate synthesis methods can lead to significant changes in their catalytic properties at low temperature. On the other hand, these can be insignificant at high temperature because the thermodynamics prevails, leading to an equilibration associated with surface Pd enrichment and then levelling their catalytic performances.

Greater stabilization of small Pt particles can be envisioned by using zirconia as a support alone and/or incorporated into silica. In addition, zirconium incorporation strengthens surface acidity and then improves the sulfur tolerance.45 However, the most important issue still remains to be the hypothetical complete substitution of PGM. Previous attempts with CuO/Ce0.8Zr0.2O2 led to promising observations especially for samples with low Cu loading, improving significantly the catalytic activity for NO oxidation to NO2. However, the use of perovskites led to more prominent catalytic performances in NO oxidation to NO2 (ref. 46) (see Fig. 4), which suggests that those solids could potentially rival supported PGM catalysts especially Sr-doped LaCoO3 systems exhibiting higher conversion than benchmark Pt-based catalysts.

image file: c7cy00983f-f4.tif
Fig. 4 NO-to-NO2 oxidation capacity of (a) La1−xSrxCoO3 and (b) La1−xSrxMnO3 perovskites with x ranging from 0 to 0.5, together with a model Pt based catalyst (reproduced with permission from ref. 46).

Urea-SCR as the most promising technology?. The development of urea-SCR catalysts is likely the most promising technology to selectively reduce NOx to nitrogen globally in a large excess of oxygen. Indeed, no fuel penalty inherent to this technology could be taken into account. The NOx emission control does not need the implementation of cycling regimes compared to lean-NOx trap technologies, and a good efficiency can be obtained at low temperature. Despite the feasibility and the high performances achieved from this technology, the automotive industry faces serious technical obstacles. Basically, fast reduction of NOx to nitrogen can take place when the following operating conditions are fulfilled: NOx/NH3 = 1 and NO/NOx = 0.5 (with NOx = NO + NO2) according to eqn (1).
2NH3 + NO + NO2 → 2N2 + 3H2O(1)

In practice, the optimization of these fast-SCR conditions is a complex task. Indeed, starting from a solution of urea, the dosage strategy of urea must be carefully monitored to get an optimal distribution of ammonia released from the thermal decomposition of urea (eqn (2)) and the sequential slow hydrolysis of isocyanic acid (HNCO) according to eqn (3).

(H2N)2CO → HNCO + NH3(2)
HNCO + H2O → NH3 + CO2(3)

Indeed, undesired side reactions can occur i.e., polymerization of HNCO according to eqn (4) and (5), damage the SCR catalysts, and lead to lower ammonia concentration. This is an important issue because the operating window to get an optimal NOx/NH3 is narrow. From a practical viewpoint, the injection of a slight excess of urea could not be applicable as it causes hypothetical ammonia slip which must be avoided because of the intrinsic toxicity of ammonia.

image file: c7cy00983f-u1.tif(4)
image file: c7cy00983f-u2.tif(5)

There are also kinetic restrictions which should be taken into account to ensure a good SCR catalytic efficiency. The introduction of modeling tools provided an important input especially for designing the ammonia/urea dosage strategies in order to avoid the ammonia inhibiting effect at low temperature and an optimal surface coverage of ammonia.47 Finally an important issue is also related to the oxidation reaction of NO to NO2 to get an NO2 to NOx ratio close to 0.5, corresponding to the fast-SCR conditions. This reaction is kinetically limited at low temperature. The existence of unusual bi-modal conversion profiles on Cu-zeolites has been reported elsewhere.48 Joshi et al. combined experimental and kinetic modeling to reconstruct and explain the existence of two distinct conversion ranges, around 250 °C and 450 °C, which reflect the superimposition of four different reactions such as the standard-SCR, the parasitic NH3 oxidation, the NO oxidation and direct NO oxidation by oxygen. The mathematical model implemented by those authors takes mass transfer phenomena into account including bulk and washcoat diffusion. At low temperature, the standard-SCR proceeds rapidly and the conversion is attenuated when the parasitic ammonia oxidation becomes predominant above 300 °C. At high temperature, NO oxidation to NO2 governs the overall process, accelerating by one order of magnitude the NOx reduction to nitrogen.

From benchmark SCR-vanadium-based catalysts to more thermally stable and selective systems. The usual supported vanadia-based catalysts as benchmarks for ammonia-SCR have been extensively investigated to clarify the reaction mechanism and the surface properties for establishing relevant structure–activity relationships.49–53

There is a relative consensus on the vanadium coverage dependency of the selectivity with preferential formation on isolated surface oxidic vanadium species at a low coverage which can aggregate to three-dimensional oligomeric and polymeric metavanadate species on highly loaded vanadium-based catalysts. Ultimately, the detrimental formation of V2O5 nanoparticles must be avoided because of sublimation above 650 °C, emphasizing the low thermal stability of supported vanadia based catalysts.54 Moreover, significant production of N2O can considerably limit their application for the treatment of diesel exhaust gas.55

Obviously, doped-zeolite materials seem to be the most promising SCR system. Significant improvements in their structural properties have been recently obtained which considerably enlarge their applicability in post-combustion catalysis. However, despite significant practical advances, some controversies still exist regarding the nature of active species previously assigned to the segregation of monomeric, dimeric or very small ionic clusters,56 the related reaction mechanisms and also the occurrence of various parallel and successive reactions, i.e., ammonia-SCR, ammonia-oxidation and NO oxidation to NO2 which could notably enhance the rate of NOx reduction when the fast-SCR conditions are attained. As a matter of fact, kinetic features and related surface catalytic functionalities are things difficult to rationalize because of the interplay between copper redox sites and Brønsted acidity. Interestingly, Gao et al.57 pointed out that the huge mechanistic information already obtained on vanadia-based catalysts could not be strictly applicable to new varieties of catalysts especially transition metal doped-zeolites. By way of illustration, the ammonia stabilized as NH4+ on Brønsted acid sites on V-based systems as reactive intermediates58 could not be consistent with the rate enhancement observed at a low Si/Al ratio on Fe–ZSM5 corresponding to a higher density of Brønsted acid sites.59 Indeed, experimental evidence revealed that ammonia adsorbed on Lewis sites reacts much more readily with NO than NH4+.60 As a matter of fact, Gao et al. concluded that Brønsted acid sites in a zeolite substrate would mainly contribute as a reservoir instead of a provider of reactive NH4+.57

The formation of multiple sites especially on Cu exchange zeolites having a chabazite-type structure (Cu-SSZ-13) depends on various parameters such as copper loading and copper ion mobility. In addition, the Si/Al ratio makes any attempt of rationalization more complicated, altering the copper location sites, and the Brønsted acid sites useful for ammonia storage.61 All these observations emphasize the fact that relevant relationships between the surface and the catalytic properties need careful in situ physicochemical characterization.

The NOx storage reduction systems. The NOx Storage-Reduction method (NSR), which is composed of Lean-NOx Trap (LNT) catalysts, was discovered and industrially developed by Toyota in the early 1990s.62 Today this technology fulfills the Euro 6 regulation. At first glance, these systems can be considered as three-way catalysts (TWCs) further modified by adding barium oxide able to trap ad-NOx species. Hence, the efficiency of such systems resides in the management of successive lean/rich cycles to store ad-NOx under lean conditions as nitrites and nitrates and then ensure their reduction to nitrogen under fuel rich conditions on PGM particles.63 The most common LNT catalysts contains precious metals, typically Pt or Rh, and basic storage materials, commonly Ba2+ or K+, to stabilize those ad-NOx species. A beneficial effect on the storage capacity of Pt/K/Mn/Al2O3–CeO2 was demonstrated and explained by the formation of a specific phase characterized by operando FTIR spectroscopy and TEM measurements.64 γ-Al2O3 is currently used and ensures a high specific surface area and can play an important role; even though NOx is loosely bound on alumina, this support can improve the efficiency at low storage temperature. Let us note that in practice, the presence of a significant amount of PGM leads to a very effective diesel oxidation catalyst (DOC) for enhancing the production of NO2 from NO oxidation. Earlier investigations led to some amendments to the suggested reaction mechanisms including a reversible surface spillover step of NO2 between Pt sites and BaO sites.65

Despite important kinetic and thermodynamic limitations and the overconsumption of fuel due to direct injection to ensure a rich atmosphere during short periods of regeneration (1–2 second), such technology is still attractive, being cost-efficient compared to the urea-SCR technology. In practice, the storage capacity of LNT can involve different parallel and sequential reaction pathways (see Fig. 5) elucidated from temperature-programmed adsorption/desorption experiments and in situ infrared spectroscopic measurements.66

image file: c7cy00983f-f5.tif
Fig. 5 Different reaction pathways for the storage of nitrates based on ref. 50.

This process can be limited at low temperature due to the slow NO oxidation to NO2 being usually the rate-determining step: the optimal temperature was found to be around 300 °C.67 It was also shown that the NOx storage materials can deteriorate once exposed to H2O and CO2. The overall efficiency also depends on the nature of reducing agents decreasing on Pt–LNT according to the following sequence: H2 > CO > C3H6. This latter ranking largely opens the potentiality of hydrogen. Indeed a recent investigation68 found a rate enhancement in NOx reduction over a model Pt/Rh/BaO/CeO2/Al2O3 LNT when fast purges (less than 10 second) are performed in the presence of hydrogen. Under these unsteady-state conditions, lower amounts of ammonia are released in a wide temperature range. In fact, the lower ammonia production compared to usual conditions was explained by a combination of enhancement of ammonia oxidation by oxygen species from ceria and ammonia adsorption on the support. The production of N2O as an undesired side product is also an important issue. N2O primarily forms on commercial LNT catalysts after switching from lean to rich conditions when PGM are still partially oxidized after a long exposure to lean conditions. A secondary production is detected during the rich-to-lean switch ascribed to surface reactions between NH3, CO and/or isocyanate with residual stored NOx.69 Regarding the production of N2O during the lean-rich switch, a high O-coverage prevents the dissociation of NO during the early stage. In fact, the production of N2O is no longer observed when the reductant scavenges O-adatoms and the metallic character of the PGM is restored. This is a likely explanation why ammonia extensively formed at an increasing purge period. This explanation seems consistent with in situ FTIR observations on Pt/Ba/Al2O3 during cycling under lean conditions (750 ppm NO2 in N2/air) at 200 °C and then cooled down at RT and regenerated under a flow of propene/air/N2 at increasing temperature up to 600 °C. The CO produced during this regeneration step can act as a probe molecule adsorbed on Pt, showing at the early stage infrared bands ascribed to CO adsorbed on electrophilic Pt sites progressively shifting to lower wavenumbers characteristic of CO adsorbed on Pt0 sites as the dominant species at the highest temperature.70

Particular attention was paid in the past decade to the sulfur tolerance of LNT systems. Indeed, the presence of SOx can lead to stronger stabilization of sulfate species more thermally stable than nitrates and requiring regeneration strategies.71,72 Olsson et al.71 found that a complete regeneration can be achieved at 700 °C with a high concentration of hydrogen (5000 ppm) occurring more readily in the presence of CO2, suggesting the involvement of barium carbonates in the faster regeneration process.71 In fact these authors found that sulfate species can strongly be adsorbed on alumina, leading to high coverages of sulfates also on alumina. Hence, the capacity of alumina to store significant amounts of sulfates would partly protect or delay the sulfation of barium oxide. Some findings on the lab scale could be of practical interest for managing the regeneration. By way of illustration, Kima et al.72 found that a single desulfation after heavy sulfation is more efficient than sequential processes including two and four desulfation steps, leading to the restoration of higher catalytic performances in the former case.

As a matter of fact, the degradation in LNT performance is usually the result of a combination of several factors. As previously mentioned, residual sulfur accumulated on the storage barium oxide substrate in the aged LNT generates a significant loss of NOx storage efficiency. However, thermal sintering of the PGM plays a crucial role in decreasing the contact between Pt and Ba and then altering spillover processes, as discussed earlier. As a consequence, the regeneration becomes harder with an irreversible deactivation, inducing changes in the selectivity behavior with a more extensive production of ammonia in the temperature range 150–450 °C.73 The development of LNT catalysts more resistant to sulfur poisoning and more thermally stable have been explored with different viewpoints consisting in modifying the LNT composition and/or by adopting novel preparation methods which can be easily scalable. Regarding the first option, CeO2 incorporation into Pt/Ba/Al2O3 led to significant improvement because first, CeO2 exhibits a higher sulfur tolerance without a longer formation time of BaS and secondly, it inhibits particle sintering taking place during the regeneration step at 600 °C. Subsequent comparison between Pt/Ba/Al2O3 and Pt/Ba/CeO2/Al2O3 has shown that the residual sulfur content on regenerated Pt/Ba/Al2O3 LNT is five times higher than on Pt/Ba/CeO2/Al2O3.74 The second option is more focused on the development of alternative preparation methods especially obtained from Flame Spray Pyrolysis (FSP) suited for preparing solids on a larger scale. Basically, an organic solution containing the metal precursors is sprayed in a flow of oxygen as carrier gas. Afterwards, the solvent acting as a fuel is readily ignited by flamelets. By carefully monitoring the different steps of the solvent evaporation, decomposition, nucleation, coalescence and condensation processes, it is possible to obtain nano-sized grains without inhomogeneity in the composition. Such a method has been profitably used for the synthesis of bimetallic Pt and Pd LNT. It was found that selective deposition of PGM was possible with preferential deposition on an alumina support or on storage materials (K2CO3).75 As illustrated in Fig. 6, it was found that Rh incorporation led to higher catalytic efficiency as well as better resistance to SO2 poisoning. However, the most prominent observation is provided by the comparison with LNT prepared by conventional wet impregnation. The superiority of solids prepared by FSP was ascribed to the selective deposition of PGM compared to wet impregnated samples because of their short residence times at high temperature, whereas for impregnated catalysts the time-scale drastically differs accounting for an impregnation-drying-calcination sequence occurring over a period of several hours.

image file: c7cy00983f-f6.tif
Fig. 6 NOx conversion of selected flame-made catalysts, Al–Ba 1 Pt and 0.5 Pt Al–Ba 0.5 Rh, and an impregnated Pt/Ba/Al2O3 reference catalyst during cycling at different space velocities: 38[thin space (1/6-em)]000 h−1 (open symbols) and 300[thin space (1/6-em)]000 h−1 (full symbols). In (a), the flame-made catalysts Al–Ba 1Pt (circles) and 0.5 Pt Al–Ba 0.5 Rh (triangles) are compared. In (b), the catalytic performances of the wet-impregnated Pt/Ba/Al2O3 reference catalyst (squares) under the same testing conditions are shown at 300 °C and at 400 °C (reproduced with permission from ref. 75).

3. New insights into coupled end-of-pipe technologies

Prior to envisaging more compact and close coupled technologies, it seems important to identify the different technical features with advantages counterbalanced by drawbacks. A priori it is not trivial for coupling SCR, LNT and DPF to be running under different operating conditions. Regarding SCR and LNT, both technologies run in unsteady state regimes which currently take place according to the NO/NO2 ratio (reaching 1 for the fast-SCR conditions), whereas it is related to lean/rich cycles for LNT systems. As already pointed out, some analogies exist for the abatement of NOx on NOx Storage and SCR catalytic systems, with common reactive ad-NOx species i.e. nitrites and nitrates as true intermediates that can be further reduced to nitrogen. As previously explained in section 2.1.2, this implies the settlement of a diesel oxidation catalyst upstream of the LNT and SCR system to convert CO and unburnt hydrocarbons but also to produce NO2 from NO oxidation to suit the NO/NO2 ratio that corresponds to optimal operating conditions and to combust soot. Generally DOC and LNT contain high PGM loadings whereas SCR catalysts are composed of cheaper transition metals which to some extent can be less thermally stable and intrinsically much less active. This is an important aspect especially for coupling SCR and DPF. Increasing their surface density at a higher metal loading often means significant loss of selectivity. The complexity of coupling technologies is also inherent in various interactions between atmospheric pollutants and the different catalytic functionalities. For instance, Artioli et al.76 found that the presence of soot has a detrimental effect on the LNT efficiency, reducing the NOx storage capacity in the temperature range 200–350 °C. However, soot can directly interact with ad-NOx species according to eqn (6) leading to low temperature NOx emissions, as exemplified in Fig. 7. Hence, soot can favor the decomposition/reduction of ad-NOx involving surface mobility of adsorbed nitrates. However, the subsequent production of CO2 can slightly prevent the on-going NOx storage since the formation of barium carbonate has a negative effect.77
image file: c7cy00983f-t1.tif(6)

image file: c7cy00983f-f7.tif
Fig. 7 TPD run after NOx adsorption at 350 °C (1000 ppm NO + O2 (3% v/v) in He + H2O (1% v/v) + CO2 (0.1% v/v)) over (A) PtBa/Al2O3 catalyst; (B) PtBa/Al2O3/soot mixture (reproduced with permission from ref. 76).

Hence, coupling effects due to soot oxidation perturbing the kinetics of NOx removal will not be reviewed in this present contribution although this issue is essential and will probably grow in the near future with the introduction of gasoline direct injection. Indeed, upcoming legislation will impose the introduction of particulate filters for gasoline engines with a technical solution combining the three way catalytic functionality and the filter in one device: the so-called catalysed gasoline particulate filter. Recently, kinetic modeling attempts showed that the optimization of such systems is influenced by heat transfer and internal and external mass transfer.78

3.1. Coupled LNT–SCR technologies

First attempts with double bed catalytic systems have shown that it was possible to form ammonia on Pt during the regeneration stage of LNT and then promote the ammonia-SCR on acidic solids under lean conditions.79

Subsequent investigations80–84 confirm the relevance of this combined approach with a model Pt/Ba/Al LNT upstream and a Fe-beta SCR catalyst downstream, minimizing ammonia slip and reaching a superior NOx conversion to nitrogen of ∼99.5% at 300 °C.80 Indeed, stored ammonia further reacts under lean conditions with NOx. In addition, the LNT ensures through NO oxidation that a suitable concentration of NO2 needed to activate the fast-SCR process is available. The overall efficiency of the system also depends on the hydrogen amount i.e. concentration and time exposure81,84 which promotes the ammonia formation at low temperature and then enhances the N2 yields (see Fig. 8). Such a trend likely stimulated diverse strategies consisting in regenerating first the LNT by using reformate and then producing a high concentration of ammonia further stored and converted on an SCR catalyst under lean conditions.81 However, an optimum concentration of hydrogen was observed. Indeed, high H2 concentrations further induce significant ammonia inhibiting effects. The solution consisting in using reformate (CO + H2 mixture) for regenerating the LNT also exhibits some limitations especially on Pt-based LNT due to a strong CO-inhibiting effect. It is worthwhile to note that different behaviors have been found when LNT and SCR catalysts are physically mixed.83 In this specific case, Corbos et al.83 did not observe a significant effect of the nature of the reductant (CO or H2) and the reduction time.

image file: c7cy00983f-f8.tif
Fig. 8 Impact of H2 concentration and regeneration duration on the formation of ammonia during the storage phase under 500 ppm NO + 5 vol% O2 (120 s) regeneration under hydrogen (reproduced with permission from ref. 84).

The development of spectroscopic approaches on the lab scale led to important mechanistic information through the characterization of stored nitrites and nitrates on a model Pt–Ba/Al2O3 combined with Fe–ZSM5 at low and high temperature, respectively.82 These ad-NOx species can further react to produce ammonia. In fact, these observations can be explained by a two-step reaction involving first the reaction of NOx with H2 to produce ammonia and then the reduction of NOx by ammonia, which is considered as the slow step.

In practice, the development of commercial end-of-pipe systems is usually a complex task particularly when different technologies are combined involving engine management and different catalyst formulations. Some attempts consisting in coupling LNT and SCR technologies by using dual-layer monolithic catalysts led to prominent results.85 Fe or Co-doped ZSM5 zeolites as SCR catalysts were deposited on top of Pt/Rh/BaO/CeO2 LNT. During the regeneration period, undesired ammonia formed on the LNT can be captured and stored on the SCR system and then consumed after switching to lean conditions, further improving the overall efficiency of NOx reduction to nitrogen. Some improvements can be obtained after ceria incorporation which enhances the formation of ammonia at low temperature and improves the thermal stability compared to barium. Of course, the introduction of a second layer induces some limitations due to internal diffusion limitations which alter the efficiency. It also causes an increase of the number of design parameters, i.e., catalyst loading and layer thickness.86 Numerical simulations can be useful for designing and optimizing such systems and related operating strategies. Different models developed for this purpose usually take chemical and physical processes into account under wide operating conditions which can lead to excessively long computations. Increasing the number of parameters is numerically demanding and sometimes restricted to the steady state to preserve an acceptable error. Recently, new approaches mimicking realistic driving cycle conditions led to acceptable accuracy of computing the concentration profiles in the washcoat.86 For the sake of simplicity, global kinetic models were preferably implemented for modelling the performance of LNT + SCR systems, indicating more advantages to inserting the SCR downstream at a cooler place to trap ammonia more efficiently when the LNT is regenerated and especially on less selective aged systems which produce more ammonia.87

Predictive mathematical tools are helpful for the optimization of more complex architectures with close-coupled technologies: DOC, DPF, NSR and SCR catalysts. Zukerman et al.88 predicted the dynamic responses of close coupled NSR and SCR after-treatment systems. Kinetic models related to ammonia formation and oxidation during the regeneration under rich conditions were developed in order to obtain an optimal balance between stored and reactive NOx and ammonia species determining the efficiency of the global system (see Fig. 9).

image file: c7cy00983f-f9.tif
Fig. 9 (A) θNO and θNH3 (the ratios of NO and NH3 to catalyst capacity, respectively) in a NOx storage unit under EC1 conditions as a function of time (at experimental time and at optimal time). (B) NO conversion and the ratio between NH3 and adsorbed (stored) NH3 for two volume configurations under EC1 conditions (reproduced with permission from ref. 88).

Several attempts demonstrated the existence of a non-ammonia reaction pathway for NO reduction to nitrogen in coupled LNT–SCR technologies.89 Indeed during the regeneration step by injecting a mixture composed of CO, hydrogen and propene, NO and propene slipping from the LNT can react over the SCR catalyst. Cycling experiments also demonstrated that propene can be stored and continues to react when the coupled system shifts to lean conditions, offering an additional reaction pathway for the mitigation of NOx under lean and rich conditions, as exemplified in Fig. 10.

image file: c7cy00983f-f10.tif
Fig. 10 Comparison of cycle-averaged NOx conversion and selectivity to NH3 for LNT-only and LNT–SCR systems using 1% CO + 0.33% H2 + 3333 ppm C3H6 as a reductant (reproduced from permission from ref. 89).

Alternative technologies have been explored, for instance, instead of coupling LNT and SCR, SCR can be associated with three-way-catalysts well-known to produce ammonia during the fuel rich period. Such a strategy could also be attractive for gasoline engines avoiding ammonia slip. However, in a passive-SCR configuration, the objective is deliberately the enhancement of ammonia production to feed downstream SCR. In a recent investigation, DiGiulio et al.90 compared the performance of passive-ammonia SCR systems under steady state and lean/rich cycling conditions. They showed an air-to-fuel ratio dependency of the ammonia production. Indeed, a lower air-to-fuel ratio is needed to produce the same amount of ammonia under cycling conditions compared to steady-state conditions. These authors observed a decrease in ammonia formation under sufficiently rich conditions according to the following sequence: TWC with s high PGM loading ≥ a combination of TWC with high and low PGM loadings ≥ TWC with a low PGM loading ∼ lean NOx trap. A joint beneficial effect was also related to the inclusion of a NOx storage functionality and then lowering the NOx slip during the lean period which led the authors to an important questioning point related to the feasibility of improving the NOx storage capacity of conventional TWC.

Actually, the environmental benefit due to the emergence of efficient close coupled LNT and SCR technologies is closely related to the cost investments provided by such installations. As a matter of fact, previous attempts on the lab scale have shown a constant NOx conversion, and a reduction of PGM loading can be envisioned on coupled systems compared with a single LNT system.91 Such an opportunity would be possible according to a combined technology earlier developed by Ford's catalyst suppliers including: – desulfation of LNT at lower temperature – advanced SCR catalysts close to those encountered in commercial zeolite-based SCR applications – synergetic coupling of the LNT and SCR catalyst performance preserving high efficiency even on aged systems.

3.2. SCR systems coated on DPF

Numerous investigations discussed different architectures combining DOC + DPF + SCR or DOC + SCR + DPF.92 One of the major advantages provided by this latter configuration is the fast light-off of the SCR catalysts. In addition, they are not altered by the high temperature reached during the regeneration of DPF, allowing the use of benchmark vanadium based SCR catalysts. On the other hand, low NOx concentrations are available, which lowers the efficiency of passive DPF regenerations. Guan et al.92 also inventoried in their review article the advantages provided by a DOC + DPF + SCR configuration preferred to comply with the Euro 6 and US 2010 regulations. In this specific case, the removal of soot is promoted but the SCR is not sufficiently efficient particularly during the engine cold start which also suffers from much lower NO2 concentrations to fulfill the fast-SCR conditions.93 This configuration rules out the use of V-based SCR catalysts and currently Fe- and Cu-doped zeolites are preferred since they exhibit higher hydrothermal stability. Both cases can satisfactorily meet the US and Euro standard regulations.

In order to reduce the total volume, several attempts reported the performances of the SCR functionality coated on the DPF to form an SCRF. Hence, a single catalytic converter is used for controlling NOx and particulate emissions.94,95 As previously discussed, the use of modeling tools can help in optimizing the wash-coat loading and distribution. The validation of such mathematical kinetic models was reported for heavy and light duty vehicles,96 being capable of predicting NOx conversion and ammonia slip and taking into account possible interactions between SCR and DPF functionalities. As a result, the authors did not find significant impact of soot on NOx conversion. On the other hand, the presence of NOx induces a significant delay in soot removal.

As a matter of fact, the coating of the SCR functionality on the particulate filter must take several parameters into account to avoid significant pressure drops. As has been recently found,97 the washcoat amount and its location inside the pores of the filter materials can impact the soot efficiency and the related pressure drop, as illustrated in Fig. 11. In practice, a current two-step filtration process usually occurs with depth filtration at the early stage when soot enters the porous structure of the substrate and induces a sharp increase of the pressure drop. In the second stage, the soot accumulates on the surface of the wall filter, forming a soot cake which will act as a filter and corresponding to a lower increase of the pressure drop. As shown in Fig. 11, an increase of the wash-coat amount from 90 g L−1 to 150 g L−1 induces a strong detrimental effect on the pressure drop, reaching more readily critical values. Hence, a more open porous structure is mandatory for SCRF compared to conventional DPF. The SCR wash-coat should coat the pore structure rather than the wall filter which means that improvements in wash-coating technologies are also an important issue.98 In passive regeneration, soot and NOx removal is driven by the formation/consumption of NO2;97 (i) the passive soot oxidation is slowed down for NO2/NOx ≤ 0.5 because of the occurrence of the fast-SCR and joint diffusion phenomena, depleting NO2. (ii) Such trends are much less accentuated for NO2/NOx ≥ 0.5 since the NO2-SCR occurs much less readily, preserving NO2 available for soot oxidation. In this configuration, a synergy effect on the rate of NOx reduction is observed (see Fig. 12).

image file: c7cy00983f-f11.tif
Fig. 11 Soot-loading characteristics of 90 (left) and 150 g L−1 (right) SCR/DPF samples configured such that the catalyst was present predominantly on the upstream portion of the filter microstructure and on the inlet channel wall (for the 150 g L−1 sample) (reproduced with permission from ref. 97).

image file: c7cy00983f-f12.tif
Fig. 12 SCR/DPF NOx reduction efficiency at 35[thin space (1/6-em)]000 GHSV; NH3/NOx = 1; NO2 = 250 ppm; and NO2/NOx = 0.55, 0.6, and 0.65; effect of 4 g L−1 initial soot loading on NOx reduction efficiency (reproduced with permission from ref. 97).

Recent investigations reported greater stabilization when the SCR material is co-extruded with the ceramic substrate suppressing the inherent complexity of the coating process. However, the authors also mentioned that higher amounts of SCR catalysts are needed to manufacture the hybrid materials. The low specific surface area must be enhanced to increase conversion levels.99

Typically, SCR catalyst materials must exhibit high thermal stability. Recently, Cu2+ ion exchanged in a Cu-SSZ-13 chabazite (CHA) structure was found to be more efficient in terms of activity and selectivity for ammonia-SCR than benchmark Cu-ZSM-5 and Cu-beta catalysts.100,101 Interestingly, Cu-SSZ-13 was found to be more resistant to thermal degradation and less sensitive to deactivation through hydrocarbon inhibition compared to other Cu-doped zeolite-based systems, which opens new opportunities both for the replacement of conventional V2O5/WOx(MoOx)/TiO2 catalysts and for coupling the SCR function to the diesel particulate filter. A faster clustering effect of copper moieties occurs in the supercages of the zeolite Y, ruling out any opportunity for subsequent deposition on the DPF substrate. The superiority of Cu-SSZ-13 recently attracted the scientific community to obtain more insight into the structural and catalytic features.102 Two types of Cu ions were characterized inside the CHA network at different ion exchange positions with the distribution changing with the copper loading: at low loadings, Cu2+ ions occupy six-membered rings, while they are preferentially localized in the large cages of the CHA structure at high Cu loadings.102 Cu2+ in planar 6-membered rings was found to be more active to the SCR. On the other hand, Cu-oxo species selectively catalyze the decomposition of NO.103 Gao et al.104 also investigated the structure relationship in ammonia-SCR as well as the slowest competitive ammonia oxidation on Cu-SSZ-13. They suggested that NH3-SCR is mass transfer limited due to the small pore opening, emphasizing the fact that the thickness of the washcoat and pore diffusion limitation must be taken into account for further optimization. Further textural changes in meso-microporous Cu-SAPO-34 led to greater accessibility owing to the mesopores thus remarkably enhancing the diffusion rate.105

4. Natural Gas Vehicle (NGV) three-way-catalyst: prospects and challenges

As previously mentioned, a major advantage of natural gas powered engines is related to the very low amount of particulates, but this could be partly counterbalanced by the residual emissions of methane in the exhaust which is well-recognized as a potent greenhouse gas. Hence, the emissions of NOx and unburnt methane from stoichiometric (λ = 1) and lean-burn (λ = 1.3) NGV engines can be regulated through the implementation of palladium–rhodium three-way catalysts or platinum–palladium oxidation catalysts, respectively. It is worth noting that both regimes lead to significantly lower CO2 emissions.106 However, in general stoichiometric conditions are advantageous because the simultaneous abatement of NOx and methane is easier than under lean conditions, and greater resistance to poisoning effects in the presence of water (>10 vol%) and sulfur (<0.5 ppm) is observed.

Up to now, the majority of academic investigations have been performed under lean-burn conditions (large excess of oxygen), being more favorable for combusting methane at lower temperature. Important technical issues are addressed by academia in order to stabilize the presence of oxidic palladium species responsible for the catalytic performances. A critical aspect in the development of stoichiometric NGV catalysts is related to the low stability of PdO under these operating conditions, which decomposes to much less reactive Pd0 species at high temperature. Hence, some thermodynamic constraints must be considered due to the equilibrium between PdOx and Pd0. Palladium in association with platinum was found to improve the thermal durability and the impact of sulfur poisoning.107

Catalytic technologies for methane abatement have already been investigated. Palladium is the metal of choice for generally highly loaded benchmark catalysts106 (>200 g ft−3, more than twice the amount used in conventional catalysts) to compensate the deactivation process due to thermal sintering. Indeed, a high temperature in the range 400–500°C108 is needed to activate the cleavage of the C–H bond (439.3 kJ mol−1).109 Important issues in the development of such technology are generally associated with: (i) thermal deactivation at high temperature, (ii) design of catalytic systems more resistant to poisoning effects, especially sulfur compounds in the exhaust gas, and (iii) high activity at the lowest temperatures typically during the engine cold start. All these items have been reviewed elsewhere.110

Plasma-assisted catalytic systems could fit the last requirement with an in-plasma configuration (Fig. 13(a)) leading to synergy on the rate of methane conversion by combining non-thermal plasma producing highly active radicals towards methane unselectively converted into CO to a large extent even at room temperature and a highly selective 2 wt% Pd/Al2O3 catalyst which selectively converts CO into CO2.111 On the other hand, the post-plasma configuration (Fig. 13(b)) does not provide an extra conversion compared to the previous option. This approach was also verified on conventional Pt and Pd supported on γ-Al2O3 with a much lower PGM content in the range 0.36–1.66 wt%, suggesting the opportunity to significantly reduce the use of PGM using this approach.112 However, in practice the environmental approach provided by plasma-assisted catalysis needs more insight to prove its economic feasibility.

image file: c7cy00983f-f13.tif
Fig. 13 Configuration of an in-plasma catalysis reactor (a) and a post-plasma catalysis reactor (b) (reproduced with permission from ref. 111).

Different approaches explored the possibility to promote the catalyst efficiency electrochemically by applying a potential between the catalyst supported on a solid electrolyte as a working electrode and a counter electrode deposited on the same electrolyte substrate. This concept, labelled non-faradaic electrochemical promotion of catalysis (NEMCA), was initially discovered by Stoukides and Vayenas.113 By this way, it is possible to monitor the metal–support interaction and to induce reverse spillover of promoters from the solid electrolyte to the catalyst surface.114 Such a behavior has been profitably developed for NOx storage/reduction systems by using Pt/K–βAl2O3/Pt operating under fixed lean conditions, the regeneration occurring under the same working lean-burn conditions as those of storage. Relevant practical achievements were also obtained for the selective reduction of NOx by unburnt hydrocarbons, enhancing the activity and selectivity to nitrogen production below 300 °C by using low overpotentials.115 Hence, through this approach it is possible to overcome fuel penalties compared to a classical LNT technology.116 This concept has been successfully implemented by Matei et al.117 for methane combustion on Pd deposited on Y2O3-stabilized ZrO2 (YSZ) with a gain of one order of magnitude obtained on Pd/YSZ compared to the undoped YSZ. The electrochemically induced effect of the catalytic activity was preserved under fuel rich conditions, which is related to the stabilization of the core–shell structure. As explained by the authors, the metallic Pd core would provoke a higher electron density in the conduction band of the particles which would facilitate electron extraction from surface oxidic palladium species compared to fully-oxidized palladium particles.117 Some questioning points arise regarding the slow commercial development despite the significant boost in catalytic activity due to electrochemical assistance. Indeed, a lower use of critical materials, especially expensive PGMs, could be envisioned.118 This has been discussed in a short review article emphasizing the fact that total oxidation reactions at reduced oxygen partial pressure could be industrially relevant.119 A second argument in favor of the development of such a technology for the purification of automotive exhaust is also related to the scale of the potential of application and the investment required.

Previous investigations on Pd-doped perovskites showed that the combination of perovskites and palladium can be suited to methane combustion because of greater stabilization of PdO. Eyssler et al.120 found that Pd2+ in strong interaction with LaFeO3 coexists with Pdn+ (with n > 2) diluted inside the perovskite lattice. The distribution of both oxidic Pd species depends on the calcination temperature, as well as the reaction temperature, the optimal activity being obtained on 2 wt% Pd/LaFeO3 calcined at 500 °C which can outperform conventional Pd/Al2O3 in terms of greater stabilization of PdO. LaFeO3 was found to protect PdO crystallites from thermal sintering, further minimizing their subsequent decomposition to Pd0. On the other hand, partial dissolution of palladium species inside the perovskite lattice together with a loss of specific surface area when calcination is performed at 1000 °C has significant detrimental effects on the catalytic activity. Similar trends were also observed by Yoon et al.121 on Pd supported on LaAlO3, who obtained higher TOF values than those calculated on Pd-substituted LaAlO3. This rate enhancement was explained by an electron back-donation from more the electronegative La to Pd thus enhancing the thermal stability of PdO.

Perovskite-based materials have been extensively investigated in the past decades due to their intrinsic oxygen mobility. As explained, these solids exhibit remarkable catalytic properties for total oxidation reactions especially methane combustion. However, their practical applications on an industrial scale are relatively scarce, and probably, among the different examples the most demonstrated in post-combustion catalysis are related to their development at the beginning of the 1970s as two-way catalysts122,123 and more recently in three-way catalysis based on the fact that under cycling conditions the self-regenerative behavior of perovskite-based materials can protect PGM particles from irreversible thermal sintering.124,125 As mentioned, their rather low specific surface area generally restricts their application to high temperature. This could be an important issue for future applications for developing efficient systems working at low temperature. In practice, significant improvements have been recently obtained through simple and non-aqueous routes.126 New developments in colloidal crystal templating methods using polymer spheres as templates can lead to three-dimensionally ordered macroporous perovskite systems thus developing significantly their porous structure.127,128 The ease of tuning the morphology with preferentially exposed surface planes having high intrinsic activity via the selection of an appropriate surfactant can be considered as a significant breakthrough. In addition, the development of meso-macropore networks minimizes diffusion limitations, as shown on La0.6Sr0.4MnO3 for methane combustion.

Also, CexZr1−xO2 mixed oxides have been extensively investigated for Pd-only three-way catalysts.129 Zhao et al. found that the textural properties, i.e., pore volume and pore size distribution, influence the oxygen storage capacity and related catalytic performances. However, the elemental composition and the structural properties also determine the OSC properties, as illustrated in Fig. 14 and 15 showing the more favorable pyrochlore-type structure (Fig. 14(c)).130 Different stable and metastable phases have already been characterized, showing that among the two metastable tetragonal phases (t and t′′), the t′′ phase developed better redox properties.131

image file: c7cy00983f-f14.tif
Fig. 14 Schematic figure of the atomic configuration of three types of CeO2–ZrO2 mixed oxides (CZ). (a) M-CZ was a mixture of CeO2, ZrO2 and CeO2–ZrO2 solid solution; (b) S-CZ was a solid solution of CeO2 and ZrO2; (c) R-CZ was a solid solution of CeO2 and ZrO2 with a pyrochlore-type structure, that is, Ce and Zr atoms were arranged regularly (reproduced with permission from ref. 130).

image file: c7cy00983f-f15.tif
Fig. 15 Specific OSC (the amount of OSC per mole of Ce) of CeO2–ZrO2 mixed oxide (CZ) as a function of ZrO2 content. (○) M-CZ; (□) S-CZ; (■) R-CZ. The amount of OSC was the mole of oxygen released when these materials were treated at 1173 K in air for 15 min followed by reduction under a stream of 20% hydrogen in nitrogen at 773 K (reproduced with permission from ref. 130).

The presence of defective structures was found to improve the oxygen mobility and OSC properties. Such behavior can be enhanced via the incorporation of yttrium and praseodymium.132 The importance of the preparation method has also been pointed out because the crystallite size of CeO2–ZrO2 influences the OSC properties and the porous structure. All these structural and textural features can govern subsequent deposition of palladium and its relative stability under the usual temperature for the combustion of methane. Indeed, one of the major problems in methane conversion especially near stoichiometric and/or fuel rich conditions is related to deactivation by coke deposition. In this sense, the use of OSC materials can be useful to circumvent this problem and needs an appropriate balance between oxygen mobility of the support materials and the metal dispersion.133 As a matter of fact, these authors suggested that the continuous removal of carbon deposits from methane decomposition takes place at the metal–support interface through a spillover process. Hence, the extent of the interfacial parameter plays a crucial role and can be significantly altered with an increase of the Pt platinum size, reducing the catalyst efficiency. Such an explanation is in rather good agreement with the TAP study of methane adsorption showing irreversible adsorption of methane on the reduced surface of a model three-way Pd/γ-Al2O3 with subsequent activation, producing CO and H2 – thanks to the spillover process of OH groups. These surface processes are quasi-completely suppressed after aging at 980 °C due to the growth of Pd particles. On the other hand, the adsorption was partly preserved on Pd–Rh/γ-Al2O3 showing that Rh in close interaction with Pd can inhibit particle sintering.134 A similar behavior was also characterized on the same catalysts for NOx adsorption and activation, emphasizing the role of the metal–support interface.135

From a fundamental viewpoint, better clarification on the nature of oxygen could be a prerequisite for further improvements. It is generally accepted that the catalytic oxidation of refractory hydrocarbons depends on the nature of reactive surface oxygen species in mixed oxide materials. It was suggested that the ground state of free oxygen molecules can be activated by a sequential electron transfer leading to the formation of reactive superoxide (O2), peroxide (O22−) and oxide species (O2−). The formation of reactive oxygen species could also occur without electron transfer by means of chemically active singlet oxygen (1Δg) formation.136 Such a hypothesis has been recently highlighted on a LaMnO3–Pd/YSZ dual bed reactor, allowing significant lowering of palladium loading (≤0.2 wt%). Indeed, a drastic rate enhancement was observed on the oxygen exchange which positively influences the rate of methane oxidation in connection with the creation of more reactive oxygen species. Hence, its validation could provide interesting prospects regarding the feasibility of developing dual bed catalytic systems for the abatement of methane at low temperature.

5. Conclusions

Presently, significant breakthroughs have been obtained from a practical viewpoint particularly for treating lean-burn exhaust gas, whereas TWC is a mature technology for gasoline engines. Significant investments led to smart end-of-pipe systems by coupling after-treatment technologies with fine tuning of engines. The automotive industry will probably face additional critical issues especially the abatement of low emissions of finer particles for gasoline engines. One question arising from this premise is certainly related to the impact of additional efforts which will have to be made by car manufacturers to comply with more US, Japanese and European standard regulations and preserve their competitiveness. This will concern the use of critical materials especially PGM and rare earth elements.

Let us note that more and more stringent regulations also stimulated academia, and notable advances were obtained in the development of more efficient durable catalytic systems. It is not trivial to develop nanostructured catalysts with well-defined hierarchical porous structures and fine control of the dispersion of the active phases which can be conserved under severe operating conditions, i.e., at high temperature, wet and corrosive atmospheres favorable for thermal sintering and bulk and surface reconstruction processes leading to more thermodynamically stable but much less active systems. In this sense, the stabilization of PGM dispersion as well as doped-zeolite structures is remarkable.

Finally, stricter regulations will appear in the near future which will require more engineering efforts and likely collaborative approaches regarding the reduction of fuel consumption to comply with CO2 regulation emissions and achieve improved combustion efficiency at low temperature, which means that coupled end-of-pipe technologies must run efficiently in the same operating low temperature regime. In this context, close-coupled systems could be of practical interest to avoid significant heat loss and comply with an important issue related to the development of smaller and cheaper end-of-pipe technologies.

Notes and references

  1. W. Yang, R. Zhang, B. Chen, N. Bion, D. Duprez, L. Hou, H. Zhang and S. Royer, Chem. Commun., 2013, 49, 4923–4925 RSC .
  2. S. Kaliaguine, A. Van Neste, V. Szabo, J. E. Gallot, M. Bassir and R. Muzychuk, Appl. Catal., A, 2001, 209, 345–358 CrossRef CAS .
  3. W. Piock, G. Hoffman, A. Berndorfer, P. Salemi and B. Fusshoeller, SAE Int. J. Engines, 2011, 4, 1455–1468 CrossRef .
  4. S. Philipp, R. Hoyer, F. Adam, S. Eckhoff, R. Wunsch, C. Schoen and G. Vent, SAE Technical Papers, 2013, vol. 2, code 97364 Search PubMed.
  5. B. Guan, R. Zhan, H. Lin and Z. Huang, J. Environ. Manage., 2015, 154, 225–258 CrossRef CAS PubMed .
  6. A. Mamakos, G. Martini, P. Dilara and Y. Drossinos, JRC Scientific and Policy Report EUR 25297, 2012 Search PubMed .
  7. A. Mamakos, N. Steininger, G. Martini, P. Dilara and Y. Drossinos, Atmos. Environ., 2013, 77, 16–23 CrossRef CAS .
  8. A. Liati, D. Schreiber, P. D. Eggenschwiler, Y. A. Rojas Dasilva and A. C. Spiteri, Combust. Flame, 2016, 166, 307–315 CrossRef CAS .
  9. B. Gieachskiel, B. Alföldy and Y. Drossinos, J. Aerosol Sci., 2009, 40, 639–651 CrossRef .
  10. C. Saito, T. Nakatani, Y. Miyairi, K. Yuuki, M. Makino, H. Kurachi, W. Heuss, T. Kuki, Y. Faruta, P. Kattouah and C. Vogt, SAE Technical Papers, 2011, vol. 2, code 91197 Search PubMed.
  11. Future_of_Diesel_Engines_updated.pdf  .
  12.  .
  13. P. Whitaker, P. Kapus, P. Ogris and P. Hollerer, SAE Int. J. Engines, 2011, 4, 1498–1512 CrossRef .
  14. S. J. Curran, R. M. Wagner, R. L. Graves, M. Keller and J. B. Green Jr, Energy, 2014, 75, 194–203 CrossRef .
  15. A. Raj, Johnson Matthey Technol. Rev., 2016, 60(4), 228–235 CrossRef .
  16. G. Karavalakis, M. Hajbabaei, Y. Jiang, J. Yang, K. C. Johnson, D. R. Cocker and T. D. Durbin, Fuel, 2016, 175, 146–156 CrossRef CAS .
  17. K. Arve, H. Backman, F. Klingstedt, K. Eranen and D. Y. Murzin, Appl. Catal., B, 2007, 70, 65–72 CrossRef CAS .
  18. M. Iwamoto, H. Furukawa, Y. Mine, F. Uemura, S. Mikuriya and S. Kagawa, J. Chem. Soc., Chem. Commun., 1986, 1272–1273 RSC .
  19. P. N. R. Vennestrøm, T. V. W. Janssens, A. Kustov, M. Grill, A. Puig-Molina, L. F. Lundegaard, R. R. Tiruvalam, P. Concepción and A. Corma, J. Catal., 2014, 309, 477–490 CrossRef .
  20. P. Sazama, L. Čapek, H. Drobná, Z. Sobalík, J. Dědeček, K. Arve and B. Wichterlová, J. Catal., 2005, 232, 302–317 CrossRef CAS .
  21. P. Sazama and B. Wichterlova, Chem. Commun., 2005, 4810–4811 RSC .
  22. J. P. Breen, R. Burch, C. Hardacre, C. J. Hill, B. Krutzsch, B. Bandl-Konrad, E. Jobson, L. Cider, P. G. Blakeman, L. J. Peace, M. V. Twigg, M. Preis and M. Gottschlin, Appl. Catal., B, 2007, 70, 36–44 CrossRef CAS .
  23. M. K. Kim, P. S. Kim, J. H. Baik, I.-S. Nam, B. K. Cho and S. H. Oh, Appl. Catal., B, 2011, 105, 1–14 CrossRef CAS .
  24. J. M. Garcıa-Cortés, J. Pérez-Ramırez, M. J. Illán-Gómez, F. Kapteijn, J. A. Moulijn and C. Salinas-Martınez de Lecea, Appl. Catal., B, 2001, 30, 399–408 CrossRef .
  25. R. Burch and T. C. Wtaling, J. Catal., 1997, 169, 45–54 CrossRef CAS .
  26. Y. Wan, J. X. Ma, Z. Wang, W. Zhou and S. Kaliaguine, J. Catal., 2004, 227, 242–252 CrossRef CAS .
  27. Y. Teraoka, K. Kanada and S. Kagawa, Appl. Catal., B, 2001, 34, 73–78 CrossRef CAS .
  28. Y. Teraoka, H. Nii, S. Kagawa, K. Jansson and M. Nygren, Appl. Catal., B, 2000, 194–195, 35–41 CrossRef CAS ; D. Fino, G. Saracco and V. Speccia, Appl. Catal., B, 2003, 43, 243–259 CrossRef .
  29. F. E. López-Suárez, A. Bueno-López, M. J. Illán-Gómez and J. Trawczynski, Appl. Catal., A, 2014, 485, 214–221 CrossRef .
  30. J.-P. Schön, C. Dacquin, C. Dujardin and P. Granger, Top. Catal., 2017, 60(3–5), 300–306 CrossRef .
  31. C. Guardiola, V. Dolz, B. Pla and J. Mora, Control Eng. Pract., 2016, 56, 148–156 CrossRef .
  32. C. Guardiola, B. Pla, P. Piqueras, J. Mora and D. Lefebvre, Appl. Therm. Eng., 2017, 110, 962–971 CrossRef CAS .
  33. M. M. Azis, X. Auvray, L. Olsson and D. Creaser, Appl. Catal., B, 2015, 179, 542–550 CrossRef CAS .
  34. A. Arvajová and P. Koči, Chem. Eng. J., 2017, 158, 181–187 CrossRef .
  35. A. Morlang, U. Neuhausen, K. V. Klementiev, F. W. Schutze, G. Miehe, H. Fuess and E. S. Lox, Appl. Catal., B, 2005, 60, 191–199 CrossRef CAS .
  36. M. H. Wiebenga, C. H. Kim, S. J. Schmieg, S. H. Oh, D. B. Brown, D. H. Kim, J.-H. Lee and C. H. F. Peden, Catal. Today, 2012, 184, 197–204 CrossRef CAS .
  37. X. Auvray and L. Olsson, Appl. Catal., B, 2015, 168–169, 342–352 CrossRef CAS .
  38. A. Abedi, J.-Y. Luo and W. S. Epling, Catal. Today, 2013, 207, 220–226 CrossRef CAS .
  39. K. Daneshvar, R. K. Dadi, D. Luss, V. Balakotaiah, S. B. Kang, C. M. Kalamaras and W. S. Epling, Chem. Eng. J., 2017, 323, 347–360 CrossRef CAS .
  40. R. Chen, Z. Chen, B. Ma, X. Hao, N. Kapur, J. Hyun, K. Cho and B. Shan, Comput. Theor. Chem., 2012, 987, 77–83 CrossRef CAS .
  41. H. Xiong, E. Peterson, G. Qi and A. K. Dathye, Catal. Today, 2016, 272, 80–86 CrossRef CAS .
  42. T. W. Hansen, A. T. DeLaRiva, S. R. Challa and A. K. Datye, Acc. Chem. Res., 2013, 46, 1720–1730 CrossRef CAS PubMed .
  43. G. Cavataio, H. W. Jen, J. W. Girard, D. Bobson, J. R. Warner and C. K. Lambert, SAE Int. J. Fuels Lubr., 2009, 2, 204–216 CrossRef CAS .
  44. T. R. Johns, R. S. Goeke, V. Ashbacher, P. C. Thüne, J. W. Niemantsverdriet, B. Kiefer, C. H. Kim, M. P. Balogh and A. K. Datye, J. Catal., 2015, 328, 151–164 CrossRef CAS .
  45. M.-Y. Kim, E. A. Kyriakidou, J.-S. Choi, T. J. Toops, A. J. Binder, C. Thomas, J. E. Park II, V. Schwartz, J. Chen and D. K. Hensley, Appl. Catal., B, 2016, 187, 181–194 CrossRef CAS .
  46. J. A. Onrubia, B. Pereda-Ayo, U. De-La-Torre and J. R. González-Velasco, Appl. Catal., B, 2017, 213, 198–210 CrossRef CAS .
  47. E. Tronconi, I. Nova, C. Ciardelli, D. Chatterjee, G. Bandl-Konrad and T. Burkhardt, Catal. Today, 2005, 105, 529–536 CrossRef CAS .
  48. S. Y. Joshi, A. Kumar, J. Luo, K. Kamasamudram, N. W. Currier and A. Yezerets, Appl. Catal., B, 2015, 165, 27–35 CrossRef CAS .
  49. M. Inomata, A. Miyamoto and Y. Murakami, J. Catal., 1980, 62, 140–148 CrossRef CAS .
  50. J. A. Dumesic, N.-Y. Topsøe, H. Topsøe, Y. Chen and T. Slabiak, J. Catal., 1996, 163, 409–417 CrossRef CAS .
  51. N.-Y. Topsøe, Science, 1994, 265(5176), 1217–1219 Search PubMed .
  52. H. Kamata, K. Takahashi and C. U. I. Odenbrand, J. Catal., 1999, 185, 106–113 CrossRef CAS .
  53. R. Pérez Vélez, I. Ellmers, H. Huang, U. Bentrup, V. Schünemann, W. Grünert and A. Brückner, J. Catal., 2014, 316, 103–111 CrossRef .
  54. G. Madia, M. Elsener, M. Koebel, F. Raimondi and A. Wokaun, Appl. Catal., B, 2002, 39, 181–190 CrossRef CAS .
  55. M. Yates, J. A. Martín, M. A. Martín-Luengo, S. Suárez and J. Blanco, Catal. Today, 2005, 107–108, 120–125 CrossRef CAS .
  56. S. Brandenberger, O. Krocher, A. Tissler and R. Althoff, Catal. Rev.: Sci. Eng., 2008, 50, 492–531 CAS .
  57. F. Gao, N. M. Washton, Y. Wang, M. Kollár, J. Szanyi and C. H. F. Peden, J. Catal., 2015, 331, 25–38 CrossRef CAS .
  58. G. Busca, L. Lietti, G. Ramis and F. Berti, Appl. Catal., B, 1998, 18, 1–36 CrossRef CAS ; M. Kantcheva, V. Bushev and D. Klissurski, J. Catal., 1994, 145, 96–106 CrossRef .
  59. R. Q. Long and R. T. Yang, J. Catal., 1999, 188, 332–339 CrossRef CAS .
  60. T. Yu, J. Wang, M. Q. Shen and W. Li, Catal. Sci. Technol., 2013, 3, 3234–3241 CAS .
  61. F. Gao, E. D. Walter, M. Kollar, Y. L. Wang, J. Szanyi and C. H. F. Peden, J. Catal., 2014, 319, 1–14 CrossRef CAS .
  62. N. Takahashi, H. Shinjoh, T. Iijima, T. Suzuki, K. Yamazaki, K. Yokota, H. Suzuki, N. Miyoshi, S.-I. Matsumoto, T. Tanizawa, T. Tanaka, S.-S. Tateishi and K. Kasahara, Catal. Today, 1996, 27, 63–69 CrossRef CAS .
  63. S. Roy and A. Baiker, Chem. Rev., 2009, 109, 33–39 CrossRef PubMed .
  64. T. Lesage, J. Saussey, S. Malo, M. Hervieu, C. Hedoin, G. Blanchard and M. Daturi, Appl. Catal., B, 2007, 72, 166–177 CrossRef CAS .
  65. L. Olsson, H. Persson, E. Fridell, M. Skoglundh and B. Andersson, J. Phys. Chem. B, 2001, 105, 6895–6906 CrossRef CAS .
  66. I. Nova, L. Castoldi, L. Lietti, E. Tronconi, P. Forzatti, F. Prinetto and G. Ghiotti, J. Catal., 2004, 222, 377–388 CrossRef CAS .
  67. L. Liu, Z. Li, S. Liu and B. Shen, Mech. Syst. Signal Process., 2017, 87, 195–213 CrossRef .
  68. A. Wei-Lun Ting, M. Li, M. P. Harold and V. Balakotaiah, Chem. Eng. J., 2017, 326, 419–435 CrossRef .
  69. S. Bártova, P. Kočí, D. Mráček, J. A. Pihl, J.-S. Choi, T. J. Toops and W. P. Partridge, Catal. Today, 2014, 231, 145–154 CrossRef .
  70. Z. Liu and J. A. Anderson, J. Catal., 2004, 224, 18–27 CrossRef CAS .
  71. L. Olsson, M. Fredriksson and R. J. Blint, Appl. Catal., B, 2010, 100, 31–41 CrossRef CAS .
  72. D. H. Kima, A. Yezerets, J. Li, N. Currier, H.-Y. Chen, H. Hess, M. H. Engelhard, G. G. Muntean and C. H. F. Peden, Catal. Today, 2012, 197, 3–8 CrossRef .
  73. J. Wang, Y. Ji, G. Jacobs, S. Jones, D. J. Kim and M. Crocker, Appl. Catal., B, 2014, 148–149, 51–61 CrossRef CAS .
  74. J. H. Kwak, D. H. Kim, J. Szanyi and C. H. F. Peden, Appl. Catal., B, 2008, 84, 545–551 CrossRef CAS .
  75. R. Büchel, S. E. Pratsinis and A. Baiker, Appl. Catal., B, 2012, 113–114, 160–171 CrossRef PubMed .
  76. N. Artioli, R. Matarrese, L. Castoldi, L. Lietti and P. Forzatti, Catal. Today, 2011, 169, 36–44 CrossRef CAS ; L. Castoldi, N. Artioli, R. Matarrese, L. Lietti and P. Forzatti, Catal. Today, 2010, 157, 384–389 CrossRef .
  77. R. Matarrese, L. Castoldi, L. Lietti and P. Forzatti, Top. Catal., 2009, 52, 2041–2046 CrossRef CAS .
  78. B. Opitza, A. Drochnerb, H. Vogelb and M. Votsmeier, Appl. Catal., B, 2014, 144, 203–215 CrossRef CAS ; B. Opitz and M. Votsmeier, Chem. Eng. Sci., 2016, 149, 117–128 CrossRef .
  79. T. Nakatsuji, M. Matsubara, J. Rouistenmäki, N. Sato and H. Ohno, Appl. Catal., B, 2007, 77, 190–201 CrossRef CAS .
  80. A. Lindholm, H. Sjövall and L. Olsson, Appl. Catal., B, 2010, 98, 112–121 CrossRef CAS .
  81. A. Kouakou, F. Dhainaut, P. Granger, F. Fresnet and I. Louis-Rose, Top. Catal., 2009, 52, 1734–1739 CrossRef CAS .
  82. P. Forzatti and L. Lietti, Catal. Today, 2010, 155, 131–139 CrossRef CAS .
  83. E. C. Corbos, M. Haneda, X. Courtois, P. Marecot, D. Duprez and H. Hamada, Appl. Catal., A, 2009, 365, 187–193 CrossRef CAS .
  84. C. Dujardin, A. Kouakou, F. Fresnet and P. Granger, Catal. Today, 2013, 205, 10–15 CrossRef CAS .
  85. Y. Liu, M. P. Harold and D. Luss, Appl. Catal., B, 2012, 121–122, 239–251 CrossRef CAS .
  86. J. Rink, B. Mozaffari, S. Tischer, O. Deutschmann and M. Votsmeier, Top. Catal., 2017, 60, 225–229 CrossRef CAS .
  87. D. Chatterjee, P. Koči, V. Schmeißer, M. Marek, M. Weibel and B. Krutzsch, Catal. Today, 2010, 151, 395–409 CrossRef CAS .
  88. R. Zukerman, L. Vradman, M. Herskowitz, E. Liverts, M. Liverts, A. Massner, M. Weibel, J. F. Brilhac, P. G. Blakeman and L. J. Peace, Chem. Eng. J., 2009, 155, 419–426 CrossRef CAS .
  89. J. Wang, Y. Ji, Z. He, M. Crocker, M. Dearth and R. W. McCabe, Appl. Catal., B, 2012, 111–112, 562–570 CrossRef CAS .
  90. C. D. DiGiulio, J. A. Pihl, J. E. Parks II and M. D. Amiridis, Catal. Today, 2014, 231, 33–45 CrossRef CAS .
  91. L. Xu and R. W. McCabe, Catal. Today, 2012, 184, 83–94 CrossRef CAS .
  92. B. Guan, R. Zhan, H. Lin and Z. Huang, Appl. Therm. Eng., 2014, 66, 395–414 CrossRef CAS .
  93. G. Koltsakis, O. Haralampous, C. Depcik and J. C. Ragone, Rev. Chem. Eng., 2013, 29(1), 1–61 CrossRef CAS .
  94. T. Ballinger, J. Cox, M. Konduru, D. De, W. Manning and P. Andersen, SAE Int. J. Fuels Lubr., 2009, 3, 369–374 CrossRef .
  95. M. Naseri, S. Chatterjee, M. Castagnola, H.-Y. Chen, J. Fedeyko, H. Hess and J. Li, SAE Int. J. Engines, 2011, 4, 1798–1809 CrossRef .
  96. T. C. Watling, M. R. Ravenscroft and G. Avery, Catal. Today, 2012, 188, 32–41 CrossRef CAS .
  97. K. G. Rappé, Ind. Eng. Chem. Res., 2014, 53, 17547–17557 CrossRef .
  98. M. Naseri, S. Chatterjee, M. Castagnola, H. Y. Chen, J. Fedeyko, H. Hess and J. Li, SAE Int. J. Engines, 2011, 4, 1798–1809 CrossRef .
  99. M. Schütt, M. Gallinger and R. Moos, Top. Catal., 2017, 60, 204–208 CrossRef .
  100. D. W. Fickel, E. D'Addio, J. A. Lauterbach and R. F. Lobo, Appl. Catal., B, 2011, 102, 441–448 CrossRef CAS .
  101. D. W. Fickel and R. F. Lobo, J. Phys. Chem. C, 2010, 114, 1633–1640 CAS .
  102. J. H. Kwak, H. Zhu, J. H. Lee, C. H. F. Peden and J. Szanyi, Chem. Commun., 2012, 48, 4758–4760 RSC .
  103. U. Deka, I. Lezcano-Gonzales, B. M. Weckhuysen and A. M. Beale, ACS Catal., 2013, 3, 413–427 CrossRef CAS .
  104. F. Gao, E. D. Walter, E. M. Karp, J. Luo, R. G. Tonkyn, J. H. Kwak, J. Szanyi and C. H. F. Peden, J. Catal., 2013, 300, 20–29 CrossRef CAS .
  105. J. Liu, F. Yu, J. Liu, L. Cui, Z. Zhao, Y. Wei and Q. Sun, J. Environ. Sci., 2016, 48, 45–48 CrossRef PubMed .
  106. A. Raj, Johnson Matthey Technol. Rev., 2016, 60(4), 228–235 CrossRef .
  107. G. Corro, C. Cano and J. L. G. Fierro, J. Mol. Catal. A: Chem., 2010, 315, 35–42 CrossRef CAS .
  108. Y. Wang, H. Shang, H. Xu, M. Gong and Y. Chen, Chin. J. Catal., 2014, 35, 1157–1165 CrossRef CAS .
  109. B. C. Enger, R. Lødeng and A. Holmen, Appl. Catal., A, 2008, 346, 1–27 CrossRef CAS .
  110. P. Gelin and M. Primet, Appl. Catal., B, 2002, 39, 1–37 CrossRef CAS .
  111. H. Lee, D.-H. Lee, Y.-H. Song, W. C. Choi, Y.-K. Park and D. H. Kim, Chem. Eng. J., 2015, 259, 761–770 CrossRef CAS .
  112. P. Da Costa, R. Marques and S. Da Costa, Appl. Catal., B, 2008, 84, 214–222 CrossRef CAS .
  113. M. Stoukides and C. G. Vayenas, J. Catal., 1981, 70, 137–146 CrossRef CAS .
  114. J. Nicole, D. Tsiplakides, C. Pliangos, X. E. Verykios, C. Comninellis and C. G. Vayenas, J. Catal., 2001, 204, 23–34 CrossRef CAS .
  115. P. Vernoux, F. Gaillard, C. Lopez and E. Siebert, J. Catal., 2003, 217, 203–208 CAS .
  116. A. de Lucas-Consuegra, A. Caravaca, M. J. Martín de Vidales, F. Dorado, S. Balomenou, D. Tsiplakides, P. Vernoux and J. L. Valverde, Catal. Commun., 2009, 11, 247–251 CrossRef CAS .
  117. F. Matei, C. J. Jiménez-Borja, J. Canales-Vázquez, S. Brosda, F. Dorado, J. L. Valverde and D. Ciuparu, Appl. Catal., B, 2013, 132–133, 80–89 CrossRef CAS .
  118. A. de Lucas-Consuegra, A. Caravaca, J. González-Cobos, J. L. Valverde and F. Dorado, Catal. Commun., 2011, 15, 6–9 CrossRef CAS .
  119. N. A. Anastasijevic, Catal. Today, 2009, 146, 308–311 CrossRef CAS .
  120. A. Eyssler, A. Winkler, P. Mandaliev, P. Hug, A. Weidenkaff and D. Ferri, Appl. Catal., B, 2011, 106, 494–502 CrossRef CAS .
  121. D. Y. Yoon, Y. J. Kim, J. H. Lim, B. K. Cho, S. B. Hong, I.-S. Nam and J. W. Choung, J. Catal., 2015, 330, 71–83 CrossRef CAS .
  122. W. F. Libby, Science, 1971, 171, 499–500 CAS .
  123. R. J. H. Voorhoeve, J. P. Remeika Jr., P. E. Freeland and B. T. Mathias, Science, 1972, 177, 353–354 CAS .
  124. Y. Nishihata, J. Mizuki, T. Akao, H. Tanaka, M. Uenishi, M. Kimura, T. Okamoto and J. Mizuki, Nature, 2002, 418, 164–167 CrossRef CAS PubMed .
  125. H. Tanaka, H. Fujikawa and I. Takahashi, SAE Technical Papers, 1993, code 90902 Search PubMed.
  126. W. Yang, R. Zhang, B. Chen, N. Bion, D. Duprez, L. Hou, H. Zhang and S. Royer, Chem. Commun., 2013, 49, 4923–4925 RSC .
  127. M. Sadakane, T. Horiuchi, N. Kato, C. Takahashi and W. Ueda, Chem. Mater., 2007, 19, 5779–5785 CrossRef CAS .
  128. A. Hamidreza, H. Dai, J. Deng, Y. Liu, B. Bai, Y. Wang, X. Li, S. Xie and J. Li, J. Catal., 2013, 307, 327–339 CrossRef .
  129. B. Zhao, Q. Wang, G. Li and R. Zhou, J. Alloys Compd., 2010, 508, 500–506 CrossRef CAS .
  130. S. Matsumoto, Catal. Today, 2004, 90, 183–190 CrossRef CAS .
  131. M. Yashima, H. Arashi, M. Kakihana and M. Yoshimura, J. Am. Ceram. Soc., 1994, 77, 1067–1071 CrossRef CAS .
  132. H. He, H. X. Dai, K. W. Wong and C. T. Au, Appl. Catal., A, 2003, 251, 61–74 CrossRef CAS ; H. He, H. X. Dai and C. T. Au, Catal. Today, 2004, 90, 245–254 CrossRef .
  133. F. B. Passos, E. R. de Oliveira, L. V. Mattos and F. B. Noronha, Catal. Today, 2005, 101, 23–30 CrossRef CAS .
  134. F. Dhainaut, S. Pietrzyk, M. Chaar, A. C. van Veen and P. Granger, Appl. Catal., B, 2012, 126, 239–248 CrossRef .
  135. Y. Renème, F. Dhainaut, Y. Schuurman, C. Mirodatos and P. Granger, Appl. Catal., B, 2014, 160–161, 390–399 CrossRef .
  136. M. Richard, F. Can, D. Duprez, S. Gil, A. Giroir-Findler and N. Bion, Angew. Chem., Int. Ed., 2014, 53, 11342–11345 CrossRef CAS PubMed .

This journal is © The Royal Society of Chemistry 2017