WO2004053314A1

WO2004053314A1 - Desulphation of nox traps

Info

Publication number: WO2004053314A1
Application number: PCT/GB2003/004582
Authority: WO
Inventors: Anthony Collier; Mark Brogan; David Skipp; Philip Retman
Original assignee: Ford Global Technologies, Llc
Priority date: 2002-12-12
Filing date: 2003-10-24
Publication date: 2004-06-24
Also published as: GB0228925D0; GB2396122A

Abstract

The invention teaches a method of desulphating a NOx trap, which comprises passing a reducing gas through the NOx trap while it is operating at a temperature in excess of a predetermined minimum, wherein the conditions necessary for desulphation are maintained for a length of time that is predetermined in dependence solely upon the temperature of the NOx trap during the desulphation process.

Description

DESULPHATION OF NOx TRAPS

The present invention relates to optimisation of the operating conditions of an engine to assure desulphation of a NOx trap arranged in the engine exhaust system.

Increasing pressure to improve fuel economy has led to the development of lean-burn operating gasoline vehicles, such as direct injection spark ignition (DISI) . Under lean operating conditions, the conventional three-way catalyst (TWC) is inefficient at reducing NOx to N₂. It is widely accepted, therefore, that alternative approaches such as a close-coupled TWC in conjunction with an under floor lean NOx trap (LNT) will be utilized to achieve increasingly stringent emissions. Such a configuration is shown schematically in Figure 1.

In the LNT, NO is oxidized to N0₂ over Pt and stored on alkaline earth metals (e.g. Ba) in the form of nitrate during lean engine operation. Due to the finite capacity of NOx adsorption sites on the LNT, the LNT must undergo a periodic rich regeneration. Under such conditions, the nitrate decomposes to NO and 0₂ and the NO is subsequently reduced to N₂ over the Rh component of the LNT, in the same manner as would occur in a typical 3-way catalysis.

Sulphur poisoning of LNT's in which the NOx storage capacity can be severely depleted by the formation of barium sulphate is a well known problem. The rate and degree of poisoning is dependent on the fuel sulphur concentrations and mileage accumulated. To maintain the NOx trapping efficiency periodic desulphation (DeSOx) of the LNT is required. Decomposition of barium sulphate requires high temperatures, typically greater than 600°C, under a controlled rich (reducing) air: fuel (AFR) mixture for efficient release of sulphur.

A key step in developing effective DeSOx strategies is the ability to repeatably sulphate the LNT to enable DeSOx to be attempted. For the sulphation technique to be useful it must be capable of consistent replication of real world sulphation.

One option is to use commercially available gasoline to generate the sulphur dioxide (S0₂) to load the LNT either by real world vehicle running or under controlled chassis/engine testing. However, the sulphur content of such fuels can lead to unacceptable long times to sulphate the LNT for effective DeSOx development. The fuel sulphation ageing (FSA) rate can be increased by doping the fuel with specific sulphur dopants. However, fuel doping adds complexity and can affect other fuel parameters which may impact on engine performance. The FSA, whether with or without sulphur dopants, also does not allow the sulphur to be placed directly in front of the LNT, but rather sulphation proceeds via transport across the starter TWC employed for stoichiometric operation. This results in an additional problem, in that the amount of sulphur adsorbed on the TWC will affect the amount of sulphur accumulated on the LNT, making investigations difficult.

As will be discussed in greater detail below, it has been demonstrated on a gasoline direct injection spark ignition (DISI) engine with lean NOx trap (LNT) after- treatment, tested on a dynamic dynamometer facility, that NOx traps that are rapidly sulphated by injecting sulphur dioxide as a gas immediately upstream of the emission control system closely replicate slow sulphur accumulation.

Using the rapid sulphation by the injection of S0₂ upstream of an LNT, the inventors have been able to conduct a series of sulphur loading and desulphation (DeSOx) experiments. The experiments investigated the effect of sulphur loading, LNT temperature, and air: fuel ratio (AFR) on the DeSOx efficiency. The study showed, surprisingly, that LNT DeSOx rate is fundamentally linked to the LNT temperature and that other parameters, such as sulphur load, are of far lesser importance to the desorption rate.

Thus, in accordance with the invention, there is provided a method of desulphating a NOx trap, which comprises passing a reducing gas through the NOx trap while it is operating at a temperature in excess of a predetermined minimum, wherein the conditions necessary for desulphation are maintained for a length of time that is predetermined in dependence solely upon the temperature of the NOx trap during the desulphation process.

The invention will now be described further, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic diagram of an engine having an exhaust system containing a TWC followed by a LNT; Figure 2 is a similar diagram showing a conventional engine rig for sulphating a LNT by the FSA method;

Figure 3 is a similar diagram showing an engine rig for suplhating a LNT by the RSA method of the present invention; Figure 4 is a graph showing a comparison between the release of H₂S and S0₂ during DeSOx of LNT's that have been sulphated by the FSA and RSA methods; and

Figure 5 is a graph of the time taken to halve the sulphur of an LNT during desulphation for different values of sulphur loading and different AFR' s during the desulphation process.

The conventional method of ageing a NOx trap, herein referred to as fuel sulphation ageing (FSA) is based on loading the lean NOx trap (LNT) 5 with S0₂ formed from combustion of fuel. Early work showed that the three way catalyst (TWC) 3 prior to the LNT 5 also adsorbed sulphur. This complicated the DeSOx investigations of a sulphur loaded LNT using FSA. Consequently, during FSA the TWC 3 was removed as shown in Figure 2. During the FSA the LNT 5 was held at a constant temperature -of 350°C with periodic purging activated. The calculation of the sulphur loaded using FSA was based on fuel flow and fuel sulphur concentration (-100 ppm) .

The method of ageing an emission control system, such as a NOx trap, in accordance with the present invention, herein referred to as rapid sulphation ageing (RSA) , is carried out with the rig shown in Figure 3 by injecting S0₂ at 7 into the exhaust stream 9. This method involves the use of a mass flow controller system which is designed to enable supply of S0₂ (2% in the cylinder) gas to specific parts of the exhaust system at flows up to 10 litres/minute. The injection time was also used to adjust the injection of sulphur.

The same sulphation cycle as for the FSA is used with the S0₂ injected upstream of the LNT 5. The calculation of the sulphur loaded using RSA is based on concentration and density of sulphur in the cylinder, and injection flow rate.

Experiments as now to be described were carried to demonstrate the viability of the RSA method of the invention.

Before assessing the new RSA approach against the traditional FSA approach, it is important to consider the experimental boundaries involved with FSA. It was assumed that the accumulation of S0₂ through chemisorption on adsorbent materials in the LNT during the FSA cycle is representative of that which will occur in the field. Clearly, under real world driving conditions, the range of temperatures and AFR' s to which the LNT will be exposed could result in a number of different sulphur species, chemisorbed on different adsorbents on the LNT and even to the formation of bulk sulphates and sulphides.

It is well known that certain surface and bulk sulphates in the LNT form the most stable sulphur phases and therefore predominate the sulphur adsorption from the gas phase (see Asik, J.R., Meyer, G.M., Meyer, G.M. (2000) Lean NOx trap desulphation through rapid air fuel modulation. SAE 2000-01-1200 and Golovin, A.V. & Asik, J. Modelling and experiments for lean NOx trap desulphation by high frequency A/F modulation. SAE 2000-01-1201) . It is also well established that these sulphates are stable over a very wide range of temperatures and AFR' s . Therefore it is a reasonable assumption that the FSA method is representative of sulphur accumulation on the LNT in the field under most driving conditions.

Clearly it needs to be shown that the RSA method of the invention correlated with the FSA method. To do this, it is not only important to show that the uptake of NOx on the LNT is equivalent both after sulphation and desulphation but also, and perhaps more importantly, that the desorption rate of sulphur and the sulphur speciation during desorption is equivalent in both methods .

It is well established, and indeed intuitive, that the rate of desorption of any desorbing gas is a strong function of the adsorption strength of the prior bound species (see Bond, G.C. (1962), Catalysis by Metals, Academic Press London and New York) . Therefore if similar desorption rates of sulphur species from the LNT are observed after independent sulphur adsorptions, it is a strong indication that the adsorbed species in both cases are equivalent.

To confirm that the injection calculations were valid, and that the LNT was being exposed to the desired sulphur loads, a mass spectrometer (MS) was placed pre-LNT during an injection sequence. The calculations showed that the sulphur going into the LNT (as measured by the mass spectrometer) agreed within 5% of the intended injection quantity.

Following confirmation that the correct mass of S0₂ was being injected it was necessary to check that all of the sulphur injected was adsorbed on the LNT and no sulphur slip occurred. Any slippage observed during the RSA technique would have serious implications on any conclusions drawn from DeSOx tests. By running the MS downstream of the LNT over the entire injection period it was confirmed that little or no sulphur slippage occurred.

Having established that rapid S0₂ injections could be used to load the LNT to controllable sulphur levels, the next step was to compare the RSA to the traditional FSA method.

In the first series of tests, the LNT was loaded with sulphur followed by DeSOx at 750 °C (Table 1) . In the second series of tests, the sulphur load was followed by DeSOx at 650°C (Table 2). The mass spectrometer was used to monitor the release rates for S0₂ and H₂S associated with the different sulphation methods. As shown in Figure 4, similar chemistries and hence adsorption behaviour were seen for both sulphation methods over the desulphation temperature investigated.

It is clear from Figure 4 that the shape of the HS and S0₂ curves are similar for both the FSA and RSA methods. In both cases a sharp and rapid release of S0₂ was observed on transition to net rich AFR at 750°C. This is a result of decomposition of the adsorbent sulphate formed in the LNT. Later an increasing level of release of H₂S was noted. This is because it takes some time to completely remove the oxygen from the LNT due to the oxygen storage capacity (OSC) of the system. Once the OSC is consumed the surface of the LNT and the precious metals contained therein will be in a reduced state. Under these conditions the precious metal will reduce desorbing S0₂ from the LNT to H₂S. The mass spectrometer was also used to calculate the recovery and repeatability of the sulphation methods. As can be seen from Tables 1 & 2, there was generally good recovery agreement between the two methods; with the RSA approach giving better repeatability.

TABLE 1 - Comparison of RSA to FSA at 750 °C

It is reasonable to postulate that a small concentration of S0₂ in the gas phase over a long period of time will, having more time to penetrate into the bulk of the adsorbent particle, be bound more strongly than a large concentration of S02 in a short period of time. It has been determined that this was not the case for the S0₂ adsorbed on the LNT under these conditions. The last row in Table 2 shows the mass of sulphur released after injecting the same sulphur mass at a greatly reducing flow rate. Hence, this compared 'rapid' (order of minutes) to 'slow' (order of hours - similar to times used for FSA) for the injection technique. It was observed that the mass of S0₂ released and the percentage of sulphur recovered was very similar to the three repeat RSA tests. As shown by the similar recoveries of sulphur for the 'slow' versus 'rapid' injections of S02, the rate of sulphur accumulation on the LNT does not have a strong influence on the desorption rate and therefore, not on the absorption strength of the adsorbed species.

TABLE 2 - Comparison of RSA to FSA at 650 °C

The final checks on the injection technique were to verify that the same level of poisoning was achieved for both RSA and FSA methods. This was evaluated by measuring the NOx capacity of the LNT after identical sulphur loadings using FSA and RSA. Identical poisoning responses were found. Also after desulphation, the NOx adsorption capacity of the LNT was found to be equivalent in both cases. These results again confirm that RSA was a suitable technique for assessing sulphation and DeSOx fundamentals.

After having proved that the RSA sulphation method of the invention gives a reasonable approximation to sulphur poisoning under real world driving conditions this technique was used to explore the parameters which are most important to the rate of sulphur removal and sulphur speciation during the desulphation.

The investigations explored the effect on DeSOx of following parameters:

* LNT air to fuel ratio (AFR)

* mass of sulphur loaded on the LNT * LNT temperature

A study of the effect of AFR on the amount of sulphur removed during DeSOx showed that though in the range from 0.9 λ to 0.98 λ, the ratio of H₂S to S0₂ depended on the AFR, the mass of sulphur removed was not AFR dependent. It was also found that the desulphation time was independent of the mass of sulphur adsorbed on the LNT.

Figure 5 shows the influence of LNT temperature on desulphation rate. The plot includes all the DeSOx data collected at different sulphur loadings and AFR settings. It is clear that regardless of AFR ratio (rich of stoichiometric) and sulphur content in the LNT the predominant and perhaps the only important parameter in the rate of removal of sulphur is the LNT temperature.

The high fit for the trend line through all the points shows that the desulphation rate can be represented by an Arrhenius equation :

_{k = A} * _{θ e}(-EJ RT)

where k is the desulphation reaction rate, R is the ideal gas constant, T is the LNT temperature, E_a is the activation energy in joules per mol, A is the frequency factor (related to the fraction of collisions between the adsorbing species, in this case S0₂, and the adsorbent) and θ is the fractional surface coverage of the adsorbed species on the adsorbent.

The extremely close fit of the desulphation data to the Arrhenius plot illustrates very well that the desulphation process on the LNT can be explained by conventional theories in catalysis relating to desorption kinetics from a metallic or metal oxide surface. It also demonstrates that the desorption of sulphur is a highly activated process and therefore is strongly dependent on the adsorbents used in the LNT.

The results obtained from the research outlined above can be put to use in optimising the desulphation time of the NOx trap in a vehicle. The aim of a lean burn engine is to reduce fuel consumption by operating with a lean mixture whenever possible. However, DeSOx of the LNT, which is essential if the vehicle is to comply with regulations relating to NOx emissions, requires the engine to be operated with rich mixture which is contrary to the aim of reducing fuel consumption. The invention teaches that only moderate enrichment of the AFR is necessary to desulphate the NOx trap. The invention also enables an optimum time for the desulphation to be calculated or derived from a look-up table as a function of LNT temperature only, so that the engine need not be run rich for any longer than the time needed to desulphate the LNT.

The RSA sulphation loading technique and the strategy for optimisation of desulphation are described herein primarily with reference to LNTs used with lean-burn operating gasoline engines. However, it will be clear to those skilled in the art of engine emissions control that the RSA technique is equally applicable to the sulphation of other engine emission control systems which require desulphation strategy development, including those used in conjunction with diesel internal combustion engines. Likewise, the strategy for optimisation of desulphation according to the invention will be equally applicable to NOx traps used in conjunction with diesel internal combustion engines, with appropriate adjustment of the desulphation temperatures .

Claims

1. A method of desulphating a NOx trap, which comprises passing a reducing gas through the NOx trap while it is operating at a temperature in excess of a predetermined minimum, wherein the conditions necessary for desulphation are maintained for a length of time that is predetermined in dependence solely upon the temperature of the NOx trap during the desulphation process.

2. A method as claimed in claim 2, wherein the desulphation time is derived from a look-up table.

3. A method as claimed in claim 2, wherein the desulphation time is computed from the Arrhenius equation :

k = A * θ e^^E"^IRT)

where k is the desulphation reaction rate, R is the ideal gas constant, T is the LNT temperature, E_a is the activation energy in joules per mol, A is a frequency factor (related to the fraction of collisions between the adsorbing species and the adsorbent) and θ is the fractional surface coverage of the adsorbed species on the adsorbent.