Progress and future challenges in controlling

Applied Catalysis B: Environmental 70 (2007) 2–15 www.elsevier.com/locate/apcatb

Progress and future challenges in controlling automotive exhaust gas emissions Martyn V. Twigg * Johnson Matthey Catalysts, Royston, Herts SG8 5HE, United Kingdom Available online 30 June 2006

Abstract By the early 1970s increased use of cars in some major cities had resulted in serious concerns about urban air quality caused by engine exhaust gas emissions themselves, and by the more harmful species derived from them via photochemical reactions. The three main exhaust gas pollutants are hydrocarbons (including partially oxidised organic compounds), carbon monoxide and nitrogen oxides. Engine modifications alone were not sufficient to control them, and catalytic systems were introduced to do this. This catalytic chemistry involves activation of small pollutant molecules that is achieved particularly effectively over platinum group metal catalysts. Catalytic emissions control was introduced first in the form of platinum-based oxidation catalysts that lowered hydrocarbon and carbon monoxide emissions. Reduction of nitrogen oxides to nitrogen was initially done over a platinum/rhodium catalyst prior to oxidation, and subsequently simultaneous conversion of all three pollutants over a single three-way catalyst to harmless products became possible when the composition of the exhaust gas could be maintained close to the stoichiometric point. Today modern cars with three-way catalysts can achieve almost complete removal of all three exhaust pollutants over the life of the vehicle. There is now a high level of interest, especially in Europe, in improved fuel-efficient vehicles with reduced carbon dioxide emissions, and ‘‘leanburn’’ engines, particularly diesels that can provide better fuel economy. Here oxidation of hydrocarbons and carbon monoxide is fairly straightforward, but direct reduction of NOx under lean conditions is practically impossible. Two very different approaches are being developed for lean-NOx control; these are NOx-trapping with periodic reductive regeneration, and selective catalytic reduction (SCR) with ammonia or hydrocarbon. Good progress has been made in developing these technologies and they are gradually being introduced into production. Because of the nature of the diesel engine combustion process they produce more particulate matter (PM) or soot than gasoline engines, and this gives rise to health concerns. The exhaust temperature of heavy-duty diesels is high enough (250–400 8C) for nitric oxide to be converted to nitrogen dioxide over an upstream platinum catalyst, and this smoothly oxidises retained soot in the filter. The exhaust temperature of passenger car diesels is too low for this to take place all the time, so trapped soot is periodically burnt in oxygen above 550 8C. Here a platinum catalyst is used to oxidise higher than normal amounts of hydrocarbon and carbon monoxide upstream of the filter to give sufficient temperature for soot combustion to take place with oxygen. Diesel PM control is discussed in terms of a range of vehicle applications, including very recent results from actual on-road measurements involving a mobile laboratory, and the technical challenges associated with developing ultra-clean diesel-powered cars are discussed. # 2006 Elsevier B.V. All rights reserved. Keywords: Vehicle emissions; Autocatalysts; Three-way catalysts (TWCs); NOx-traps; Selective catalytic reduction (SCR); Diesel particulate filters

1. Introduction Even by the 1940 and 1950s air quality problems were experienced in some urban cities because of the increasing number of cars [1–6]. This was especially noticeable in locations such as the Los Angeles’ basin where temperature inversions trap and recycle polluted air [7]. By the 1960s cars had been in large-

* Tel.: +44 1763 253 141; fax: +44 1763 253 815. E-mail address: [email protected]. 0926-3373/$ – see front matter # 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2006.02.029

scale mass production for many years, and they gave personal mobility to an increasing range of people. But, oxidation of gasoline in the engine to CO2 and H2O was far from completely efficient, Scheme 1, so the exhaust gas contained significant amounts of unburned hydrocarbons and lower levels of partially combusted products like aldehydes, ketones, and carboxylic acids, together with large amounts of CO. Unburned fuel and other hydrocarbons formed by pyrolysis, and various oxygenated species are referred to as ‘‘hydrocarbons’’ and designated HC. At the high temperature during the explosive combustion in the cylinder N2 and O2 react to establish the endothermic

M.V. Twigg / Applied Catalysis B: Environmental 70 (2007) 2–15

3

Scheme 2. Scheme 1.

equilibrium with nitric oxide (NO). This equilibrium is then frozen as the hot product gases are rapidly cooled and ejected into the exhaust manifold. The combination of NO and any of its oxidised form nitrogen dioxide (NO2) is referred to as NOx and more than a thousand ppm can be present in exhaust of a gasoline engine. The three major primary pollutants in the exhaust gases from cars are therefore NOx, HC and CO. In some American cities irritating photochemical smogs became so frequent that air quality was a major health concern. The origin of these photochemical smogs was two of the primary pollutants from cars. They were of concern in their own right, but they underwent photochemical reactions to generate ozone, a strong irritant, as well as low levels of other even more noxious compounds [8]. Fig. 1 shows the increase in atmospheric oxidant levels during a day in summer in Los Angeles during the 1970s; peak levels were reached during the early afternoon. This trend followed the sunlight intensity, and it was established ozone was the main ‘‘oxidant’’ that was produced via the photochemical

Scheme 3.

dissociation of NO2 followed by the reaction of the atomic oxygen formed with O2, as shown in Scheme 2 in which ‘‘M’’ is a ‘‘third body’’ that removes energy that would otherwise cause the dissociation of O3. However, it is mainly NO that is formed in engines, and not NO2, and the rate of NO oxidation to NO2 in air is extremely slow. The oxidation of NO to NO2 is a third-order reaction the rate of which depends on the square of the very low NO concentration [9], as illustrated in Scheme 3, so the formation of NO2 could not have resulted from the direct oxidation of NO. The actual formation of NO2 in air, the ozone precursor, actually involves free radical oxidation of HC (or more slowly with CO), and one of the more important series of free radical reactions leading to NO2 is shown in Scheme 4. Overall the process corresponds to the oxidation of hydrocarbon in the presence of NO to give NO2, an aldehyde, and water. The reactive aldehyde can undergo further reactions with NO2 to give, for example, peroxyacetylnitrate (PAN) according to Scheme 5. PAN is a very strong lachrymator [10,11], and even minute traces of it cause serious eye irritation and painful breathing. Levels of tailpipe pollutants from American cars in the mid1960s were typically HC 15 g/mile; CO 90 g/mile; NOx 6 g/ mile [12]. Engine modifications could not alone meet the

Scheme 4.

Fig. 1. Variation of ambient atmospheric ‘‘oxidant’’ levels in a California City during a Summer day in the 1970s. The ‘‘oxidant’’ is mainly ozone, and peaked in early afternoon.

Scheme 5.

4

M.V. Twigg / Applied Catalysis B: Environmental 70 (2007) 2–15

demands of the 1970 Clean Air Act [13] that was designed to make significant improvements in air quality over a reasonable period of time. As a result, catalytic systems were introduced to control exhaust emissions. 2. Early autocatalysts

Scheme 6.

2.1. Choice of catalyst types The exhaust of an internal combustion engine is a demanding environment, and very unlike the steady-state operation of most chemical plant catalytic processes [14]. The catalyst must function at low temperature, resist effects of thermal excursions up to about 1000 8C, tolerate the presence of poisons (especially sulphur species) and not be affected by gas flow pulsations and severe mechanical vibrations. At first it was necessary to oxidise HC and CO, and base metal catalysts, for example those containing copper and nickel were tested. It was established they were sensitive to poisoning (initially by lead and halide from antiknock additives in the fuel, and sulphur dioxide derived from sulphur compounds originally present in the fuel and lubrication oils), and furthermore they did not have good thermal durability [15,16]. Platinum group metal catalysts were found to be very active, and a huge amount of work was done with ruthenium, but its oxides are so volatile it was not possible to prepare a catalyst that did not lose ruthenium during use [17]. Even iridium oxides are too volatile at high temperatures, so this metal could not be used in practical catalysts [18]. However, especially platinum, as well as palladium and rhodium met the requirements of having the nobility to remain metallic under most operating conditions, and not have volatile oxides that led to metal loss, and these metals have been used in autocatalysts since their introduction [19]. Of these platinum is the most noble, but catalysts containing it when very hot and exposed to oxygen for long periods can sinter through a process involving migration of oxide species. Palladium forms a more stable oxide, and this is catalytically active in oxidation reactions. Rhodium oxide (Rh2O3) is quite readily formed from the metal under hot oxidising conditions, and this can readily undergo reactions with catalyst support compounds such as alumina as shown in Scheme 6. The main role of rhodium is in NOx reduction, and

since it is metallic rhodium that is active in this reaction, it is important this can be made available rapidly when any oxidising conditions return to being slightly reducing. However, this can be a relatively slow process, because for example, there are no active centres for the dissociation of reductant species, especially hydrogen. This re-reduction process can itself be catalysed by the presence of a small amount of other metallic species that provide, for example, a mechanism for hydrogen dissociation and accelerated reduction of rhodium oxide. Table 1 lists some of the physical properties of platinum group metals together with those for base metals that are used in industrial catalysts. Frequently two or more metals are used in combination in autocatalysts. Platinum/palladium was used in some of the early oxidation catalysts, as was platinum/rhodium that was also used under rich conditions for NOx reduction. Today three-way catalysts (see below) most commonly contain palladium/rhodium although platinum/rhodium is still used on some cars. 2.2. Early oxidation catalysts The first cars with oxidation catalysts injected air into the rich (excess fuel and therefore reducing) exhaust gas to provide oxygen for oxidation of HCs and CO. Some traditional pelleted platinum catalysts, from the chemical process industry were used, in a flat radial flow-like reactor. This configuration was not ideal because of gas-by-pass, but at that time the conversions required were not as high as today so sufficient conversions could be achieved. However, attrition of the pellets caused by their movement against each other under the influence of the pulsating gas flow and vibration of the vehicle was a major concern. An alternative catalyst structure made use of a ceramic monolithic honeycomb. For strength reasons they had relatively low porosity that made them unsuitable as a

Table 1 Physical properties of some selected metals and their oxides relevant to their behaviour in autocatalysts Metal

Atomic number

Atomic weight

Density (g cm3)

mp (K)

Reduction potential (V) Mn+ ! M0 (n)

Oxide stability under operating conditions

Platinum Iridium Palladium Rhodium Osmium Ruthenium Copper Cobalt Nickel Iron

77 46 45 76 44 29 27 28 26 78

195.08 192.22 106.42 102.91 190.2 101.07 63.33 58.93 58.69 55.85

21.45 22.56 12.02 12.41 22.59 12.37 8.96 8.90 8.90 7.87

2045 2683 1825 2239 3327 2583 1357 1768 1726 1808

1.19 (2) 1.16 (3) 0.92 (2) 0.76 (3) N/A (2) N/A (2) 0.34 (2) 0.28 (2) 0.30 (2) 0.44 (2)

Unstable oxides Moderately stable oxides Stable oxides Stable oxides Very volatile oxides Very volatile oxides Stable oxides Stable oxides Stable oxides Stable oxides

Numerical data from: J. Emsley, The Elements, 2nd ed., Clarendon Press, Oxford, 1993. mp = melting point.

M.V. Twigg / Applied Catalysis B: Environmental 70 (2007) 2–15

Fig. 2. An example of how a cordierite monolith is retained in a stainless steel mantle with an intumescent mat. Vermiculite in the mat exfoliates when heated and permanently retains the monolith in place (Courtesy Corning Inc.).

catalyst support [20]. This was overcome by applying a thin layer of high surface area catalytically active material to the channel walls [21]. This layer, typically 20–150 mm thick depending on the nature of the honeycomb and the application concerned, is referred to as a washcoat. The process of applying it is called washcoating and the washcoat surface area is typically about 100 m2/g. The monoliths are made from cordierite that has an exceptionally low coefficient of thermal expansion that is needed to prevent them from cracking when thermally stressed during use. Monoliths are manufactured by extruding [22] a suitable mixture of clay, talc, alumina and water with various organic additives, that is dried and fired at high temperature. During this process cordierite, 2MgO2Al2O35SiO2, is formed [23]. Fig. 2 shows one way a ceramic monolith can be retained in a stainless steel mantle that is welded into the exhaust system. It is wrapped in an intumescent mat typically containing inorganic fibres (such as rock wool), vermiculite and an organic binder. When first used the converter experiences temperatures that decompose the organic binder and cause the vermiculite to exfoliate. The force of this expansion is considerable, and it exerts a pressure on the monolith that keeps it firmly in place for the life of the vehicle. The long-term retaining force provided by the exfoliation of vermiculite is amazing, as was the impact of fitting oxidation catalysts in the exhaust systems of cars. There was a very considerable reduction in HC and CO emissions, but there was little or no effect on the NOx emissions. 2.3. Early Nox emissions control Nitric oxide is a thermodynamically unstable compound (enthalphy of formation DHf = +90 kJ/mol), it is a free radical yet under practical conditions in the presence of oxygen dissociation does not take place [24], and it can only be converted to nitrogen via a reductive process. The first approach for controlling NOx from car engines was to reduce it over a

5

Fig. 3. Schematic arrangement of oxidation catalyst and air injection point used initially to lower HC and CO emissions. The lower later modification used air injection after a platinum/rhodium catalyst operating under rich conditions to reduce NOx, then HC and CO were oxidised in a second stage after air injection. In this way all three pollutants were controlled in a two stage process.

platinum/rhodium catalyst in rich exhaust gas before air was added to permit oxidation of HC and CO over an oxidation catalyst [25]. This arrangement, and the earlier oxidation catalyst only system are illustrated schematically in Fig. 3. The selectivity of the catalyst used and the conditions employed for the NOx reduction had to ensure a high degree of selectivity so as not to reduce NOx to NH3 or SO2 to H2S. These reactions shown in Scheme 7 are undesirable because of the smell and toxicity of the products. Good selectivity was achieved and this system enabled markedly lower emissions of HC, CO and NOx to be achieved in a reliable way. 3. Three-way gasoline catalysts (TWCs) The engines with the earliest catalytic emissions control systems were fuelled via carburettors that could not precisely control the amount of fuel that was mixed with the intake air. Often the air/fuel ratio moved randomly either side of the stoichiometric point, and it was observed a platinum/rhodium catalyst could, under appropriate conditions, simultaneously convert CO and HC (oxidations) and reduce NOx with high efficiency [26]. This catalyst concept became known as a threeway catalyst (TWC), because all three pollutants are removed from the exhaust gas simultaneously. Application of the TWC required three elements: 1. Electronic fuel injection (EFI) so precise amounts of fuel could be metered to provide a stoichiometric air/fuel mixture.

Scheme 7.

6

M.V. Twigg / Applied Catalysis B: Environmental 70 (2007) 2–15

2. An oxygen sensor in the exhaust to provide an electrical signal indicating if the engine is running rich or lean. 3. A microprocessor to control a feedback-loop using oxygen sensor signals to determine the amount of fuel to be injected under specific conditions to maintain the exhaust gas close to the stoichiometric point. Scheme 9.

By the early 1980s all of the elements necessary for the operation of TWCs were available. This became a more efficient means of controlling HC, CO and NOx emissions than the earlier two catalyst systems, and it was also more cost effective. Soon TWCs were universally adopted. 3.1. Oxygen storage components During the development of TWC formulations cerium compounds were incorporated that are redox active: under lean conditions (oxidising) they absorb oxygen, and under rich (reducing) conditions oxygen is released from them [27]. In this way the composition of the exhaust is buffered around the stoichiometric point, and this enhances conversion of all three pollutants, and especially NOx. The reactions involved in oxygen storage are illustrated in Scheme 8, and make use of the two easily accessible oxidation states of cerium at exhaust gas temperatures. The total oxygen storage capacity (OSC) is directly related to the amount of cerium oxide present, although not all of this is available during short engine transients for kinetic reasons. Since the introduction of oxygen storage components into TWCs there was a move to using increasingly more thermally stable components, and minimising negative interactions between them. For example, it is possible to optimise the environment around platinum, and if this is different from that which is optimal for rhodium it is advantageous to physically divide the catalyst into two (or more) layers containing wellseparated different active metal dispersions with their specific promoter packages [28]. Usually platinum and palladium function best in oxidation roˆles, and they are often located in the bottom part of a two-layer TWC. Rhodium in the top layer is then exposed to all of the reductant species that reduce NOx before the exhaust gases diffuse to the lower layer where they are oxidised. Physical separation into layers enhances overall catalyst performance and life by preventing alloy formation, separating otherwise incompatible promoters, and encouraging desired reactivity by matching catalytic functionality by imposing appropriate diffusing conditions on reactants. By the correct use of promoters, particularly alkaline earth and lanthanide oxides, it was possible to modify the catalytic properties of palladium so it can function as a TWC and catalyse reduction of NOx as well as oxidation of CO and HC [29]. This entails interplay between catalysis by palladium

Scheme 8.

metal and its oxide, the presence of which can be controlled by close contact with cations that stabilise surface oxygen. Again separating the catalyst coating into two layers can minimise cross-contamination, and help obtain long lasting high activity. Perhaps the alkaline promoted NOx reduction reaction with palladium-only TWCs involves the water gas shift reaction that produces hydrogen which very efficiently reduces NOx as shown in Scheme 9. Here it is postulated surface formate intermediates may be involved in converting CO to H2 as in some copper catalysed synthesis gas reactions [30], although other mechanisms involving reduced cerium species are also possible. 3.2. TWC substrate types Extruded ceramic monoliths are widely used for TWC production, but in some situations monoliths made by rolling metal foils are used. For example, the use of thin foil can provide, when appropriately coated, a catalyst with low backpressure characteristics that can be advantageous on high performance cars. These metal-based catalysts can be welded directly into the exhaust system [31]. More recently there have been advances in extruding thin wall ceramic monoliths, and these have been widely used. They have relatively low thermal mass and high geometric surface area that facilitate fast catalyst light-off after the engine has started. The decision about which type of substrate is used, metallic or ceramic, depends on a balance between these properties and the overall system cost. 3.3. Control of H2S emissions TWCs can be exposed to periods of lean driving, for example during decelerations when there may be fuel cut-off to help improve fuel economy. Then sulphur dioxide present in the fuel as organosulphur compounds can be oxidised to sulphur trioxide which in turn can be captured by basic catalyst components that are converted to sulphates such as aluminium and cerium sulphates. Later under slightly rich conditions, for example during hard accelerations stored sulphate can be reduced to hydrogen sulphide, that has a particularly objectionable odour. Careful control of the engine operating parameters (calibration), and modified catalyst formulations have enabled good control of these undesirable emissions [32], but of course the presence of high levels of sulphur compounds in gasoline is itself undesireable because SO2 is an unwanted pollutant, and today sulphur fuel levels are very much less than they were. To moderate H2S emissions catalyst formulations often contained components such as iron, nickel, and

M.V. Twigg / Applied Catalysis B: Environmental 70 (2007) 2–15

7

manganese compounds that acted as getters to capture reduced sulphur species and prevent them from being released into the exhaust gas. Later when conditions were slightly oxidising the trapped sulphur compounds were released as much less odourous SO2.

characterising them in detail [34] it is possible to use these data to simulate the overall behaviour in a computer model. Such models are very useful in designing systems for new engine/ vehicle combinations by predicting tailpipe emissions for different possible layouts. A recently reported [35] model incorporated reactions (1)–(9):

3.4. On-board diagnostics (OBD)

H2 þ 0:5O2 ! H2 O

(1)

Legislation now demands the functioning of TWCs is periodically interrogated during use, and if performance is lower than a predetermined level the incident is reported and stored in the on-board computer [33]. If poor performance persists a malfunction indicator lamp (MIL) is turned on, so the driver can have the fault corrected. The OBD system makes use of two oxygen sensors, one upstream and one downstream of the catalyst. By running slightly lean for a short period the oxygen storage component in the catalyst is converted into its fully oxidised form, at which point the engine is run slightly rich and the time taken for the gas exiting the catalyst to become slightly rich, as detected by the second oxygen sensor, is a direct measure of the oxygen storage capacity. This measurement is related to catalytic performance, and so it can be used as a criterion for the OBD requirement. In practice this approach, or a modified alternative form, works very well, and Fig. 4 illustrates the fundamentals of monitoring OSC using two oxygen sensors.

CO þ 0:5O2 ! CO2

(2)

C3 H6 þ 4:5O2 ! 3CO2 þ 3H2 O

(3)

C3 H8 þ 5O2 ! 3CO2 þ 4H2 O

(4)

H2 þ NO ! H2 O þ 0:5N2

(5)

CO þ NO ! CO2 þ 0:5N2

(6)

C3 H6 þ 9NO ! 3CO2 þ 3H2 O þ 4:5N2

(7)

C3 H8 þ 10NO ! 3CO2 þ 4H2 O þ 5N2

(8)

Ce2 O3 þ 0:5O2 Ð 2CeO2

(9)

3.5. Computer modelling TWCs Much is now known about the individual reactions and processes involved in the operation of TWCs, and by carefully

The oxidation reactions (1)–(4), the reduction reactions (5)– (8), and the catalyst’s reversible uptake and release of oxygen reaction (9) describe the overall chemistry taking place. The model developed had reactions involved in oxygen storage and release combined in a single rate expression that gave an acceptable mathematical description when tested against emissions from a SULEV car. The validating car had a rich start-up strategy with air injected into the exhaust gas to ensure very rapid heating of the underfloor catalyst following a cold

Fig. 4. Arrangement of two oxygen sensors upstream and downstream of a three-way catalyst for monitoring catalyst characteristics during driving. When the catalyst is active the oxygen level oscillations are damped by the oxygen storage components in the catalyst, should deactivation take place the oscillations break through the catalyst as illustrated by the dashed traces.

8

M.V. Twigg / Applied Catalysis B: Environmental 70 (2007) 2–15

Fig. 5. Decrease in emissions from American cars over the period 1970–1990. The introduction of oxidation catalysts markedly lowered emissions of HC (left graph, upper curve) and CO (right graph) before NOx control was introduced (left graph, lower curve). Since 1990 emissions continued to decrease as increasingly stringent legislation was introduced.

start. There was good agreement between measured and simulated temperatures in the catalyst, and small differences were attributed to difficulties of temperature measurement. There was also good agreement between simulated and measured cold start emissions. Special attention was given to transient behaviour, and in general good agreement with the simulation was obtained. The model helped in understanding the observed overall behaviour of the TWC on the vehicle. Table 2 California (CARB) Emissions standards post-1994 Year

Emissions (g/mile, FTP test) Category

PM

CO

NOx

a

Tier 1 Tier 1

0.25 0.25b 0.25c

3.40 3.40 3.40

0.40 0.40 0.40

2004

TLEV1 d LEV2e,f

0.125 0.075

3.40 3.40

0.40 0.05

0.08 0.01

2005

LEV1 d ULEV2e,f

0.075 0.040

3.40 1.70

0.40 0.05

0.08 0.01

2006

ULEV1 d SULEV2e,f,g

0.040 0.010

1.70 1.0

0.20 0.02

0.04 0.01

2007

ZEV1 ZEV2

0 0

0 0

0 0

0 0

1993 1994 2003

HC

Note: LEV = low emission vehicles, SULEV = super low emission vehicles, ZEV = zero emission vehicles. a NMHC = non-methane hydrocarbons, i.e., all hydrocarbons excluding methane. b NMOG = non-methane organic gases, i.e., all hydrocarbons and reactive oxygenated hydrocarbon species such as aldehydes, but excluding methane. Formaldehyde limits (not shown) are legislated separately. c FAN MOG = fleet average NMOG reduced progressively from 1994 to 2003. d LEV1 type emissions categories phasing out 2004–2007. e LEV2 type emissions limits phasing in 2004 onwards. f LEV2 standards have same emission limits for passenger cars and trucks