GASOLINE ENGINE WITH AN EXHAUST SYSTEM FOR COMBUSTING PARTICULATE MATTER

Abstract

A gasoline engine having an exhaust system comprises means for trapping particulate matter (PM) from the exhaust gas and a catalyst for catalysing the oxidation of the PM by carbon dioxide and/or water in the exhaust gas, which catalyst comprising a supported alkali metal. The invention further includes a method of combusting PM from a gasoline engine in CO.sub.2 and/or H.sub.2O from the exhaust gas at temperatures in excess of 500 C., which method comprising trapping the PM and contacting it with a catalyst comprising a supported alkali metal.

Claims

1. A system comprising a three way catalyst (TWC) adapted for simultaneously oxidizing CO and HC and reducing NOx of an exhaust gas from a gasoline engine, means for trapping particulate matter (PM) of <100 nm from the exhaust gas and a catalyst for catalysing the oxidation of the PM by at least one of carbon dioxide (CO.sub.2) and water (H.sub.2O) in the exhaust gas, which catalyst consists of a supported alkali metal, wherein the supported alkali metal is disposed downstream of the TWC and is disposed relative to the means for trapping in order to effect contact between the supported alkali metal and the particulate matter.

2. A system according to claim 1, wherein the alkali metal is selected from the group consisting of lithium, sodium, potassium, rubidium, caesium and a mixture of any two or more thereof

3. A system according to claim 2, wherein the alkali metal is the mixture and the mixture is a eutectic mixture.

4. A system according to claim 1, wherein the alkali metal is present in the catalyst at from 1% to 20% by weight of the catalyst.

5. A system according to claim 1, wherein the support is selected from the group consisting of alumina, ceria, zirconia, titania, a silica-alumina, a zeolite, a mixture thereof and a mixed oxide of any two or more thereof.

6. A system according to claim 1, wherein the support alumina selected from the group consisting of alpha-alumina, theta-alumina, gamma-alumina and a mixture of any two more thereof.

7. A system according to claim 1, wherein the alkali metal is potassium and the support is selected from the group consisting of alpha-alumina, zirconia, ceria and a mixture of any two or more thereof.

8. A system according to claim 5, wherein the support comprises the mixed oxide and each component of the mixed oxide is present in an amount of from 10 wt % to 90 wt % by total catalyst weight.

9. A system according to claim 8, wherein the mixed oxide is a binary mixed oxide and the ratio of the cationic components present is in a ratio of from 20:80 to 50:50.

10. A system according to claim 1, wherein the trapping means comprises a filter.

11. A system according to claim 10, wherein the filter comprises a wall-flow filter, a foam or a wire mesh.

12. A system according to claim 11, wherein the fitter is the foam and the foam is a metal oxide.

13. A system according to claim 1, wherein the trapping means is coated, at least in part, with the catalyst.

14. A method of combusting PM of <100 nm comprising contacting an exhaust gas from a gasoline engine with a three-way catalyst adapted for simultaneously oxidizing CO and HC and reducing NOx of the exhaust gas, trapping the PM and contacting the trapped PM with a catalyst consisting of a supported alkali metal, said catalyst adapted to oxidize the PM by at least one of carbon dioxide (CO.sub.2) and water (H.sub.2O) at temperatures in excess of 500 C.

15. A method according to claim 14, wherein the trapping step by electrophoresis or filtration.

16. A system according to claim 1, wherein the support is a mixed oxide of any two or more selected from the group consisting of alpha-alumina, zirconia and ceria.

17. A system according to 16, wherein each component of the mixed oxide is present in an amount of from 10 wt % to 90 wt % by total catalyst weight.

18. A system according to claim 12, wherein the metal oxide is selected from the group consisting of alpha Al.sub.2O.sub.3, ZrO.sub.2/MgO, ZrO.sub.2/CaO/MgO and ZrO.sub.2/Al.sub.2O.sub.3.

19. A system according to claim 1, wherein the elemental alkali metal is present in the catalyst at from 5% to 15% by weight of the catalyst.

20. A method according to claim 14, wherein said catalyst is adapted to oxidize the PM at temperatures in excess of 550 C.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order that the invention may be more fully understood, the following figures are provided by way of illustration only wherein reference is made to the accompanying drawings in which:

[0016] FIG. 1 is a bar chart of temperature ( C.) against oxygen source comparing the activity of a 10 wt % K/Ceria catalyst with a control;

[0017] FIG. 2 is a schematic diagram of a laboratory simulation test rig for trap evaluation;

[0018] FIG. 3 is a trace showing time-resolved particle number density from the four lowest-size stages (channels) of an electrical low pressure impactor (ELPI), during five minute portions of a nominal steady state test at the equivalent road speed of 80 km/h performed on a 1.8 litre GDI engine fitted with a wall flow particulate filter;

[0019] FIG. 4 shows the wall-flow filter performance: time variation of particle number density, of the FIG. 3 system over the final ten minutes of the European ECE+EUDC drive cycle for two sizes (in nm) denoted by number beside the curves;

[0020] FIG. 5 is a bar chart showing the performance of a cylinder-wire electrostatic precipitator to trap PM of four sizes (in nm) measured at cylinder outlet over a range of potential differences applied between the grounded cylinder and the wire relative to a control; and

[0021] FIG. 6 shows the performance, as represented by % collection efficiency by number against mobility diameter D.sub.p (nm), of a wire-cylinder trap wherein the inner surface of the cylinder is coated with a catalyst according to the invention compared to an identical uncoated device.

DETAILED DESCRIPTION OF THE INVENTION

[0022] In use, we believe that the at least one alkali metal can be present as elemental metal, its oxide, its hydroxide or its carbonate, although possibly also as its nitrate or its sulfate. Whilst the catalysts of the invention are especially suited to catalysing the oxidation of PM with CO.sub.2 and/or H.sub.2O, it will be appreciated that such catalysts are also capable of catalysing the oxidation of PM with any O.sub.2 present in the exhaust gas, for example as seen in GB-A-2232613. The combustion of PM in O.sub.2 can occur at lower temperatures than in CO.sub.2 and H.sub.2O.

[0023] Preferably, the at least one alkali metal is lithium, sodium, potassium, rubidium or caesium or a mixture of any two or more thereof, preferably potassium. Where the at least one alkali metal is a mixture, we prefer that the mixture is a eutectic mixture.

[0024] The elemental alkali metal can be present in the catalyst at from 1% to 20% by weight of the total catalyst, preferably from 5-15% by weight and most preferably 10% by weight.

[0025] The support for the at least one alkali metal can be alumina, ceria, zirconia, titania, a silica-alumina, a zeolite or a mixture or a mixed oxide of any two or more thereof. Advantageously, we prefer that the surface area of the support is as low as possible. This is because we believe that the alkali metal catalyst can be mobile, e.g. molten, at its working temperatures. If the support is porous, the mobile alkali metal can migrate into the pores away from the surface of the support, thus effectively reducing the surface area of the alkali metal and with it the activity of the catalyst. Thus, where the support is alumina, we prefer to use alpha-alumina (surface area 1-5 m.sup.2 g.sup.1) or theta-alumina (90 m.sup.2 g.sup.1), although gamma-alumina (140 m.sup.2 g.sup.1) can also be used, or a mixture of any two or more thereof.

[0026] Most preferably, the at least one alkali metal is potassium and the support is alpha-alumina, zirconia or ceria or a mixture or mixed oxide of any two or more thereof.

[0027] Where the support is a mixed oxide, each component oxide of the mixed oxide can be present in an amount of from 10% to 90% by weight of total catalyst weight. For example, in a binary mixed oxide, the ratio of the cationic components can be from between 20:80 and 50:50 by weight of the total catalyst.

[0028] The substrate for the supported at least one alkali metal can be the internal surface of a conduit carrying the exhaust gas which conduit can be of metal or ceramic, or a metal or ceramic flow-through or wall-flow filter monolith, foam, e.g. metal oxide foam, or wire mesh.

[0029] The means for trapping can comprise any apparatus suitable for trapping the PM of the appropriate particle size, e.g. between about 10-100 nm. Such means can include a wall-flow filter made from a ceramic material of appropriate pore size such as cordierite. Where a filter is used, the filter can be coated, at least in part, with the catalyst in order to effect contact between the catalyst and the PM.

[0030] Alternatively, the trapping means can comprise a foam e.g. a metal oxide foam, wherein the foam can act as a filter. The foam per se can comprise the support, i.e. the foam is both support and substrate, or the foam can be the substrate and a support e.g. comprising a high surface area metal oxide washcoat, can be coated thereon. Any suitable metal oxide foam can be used, but we prefer materials having the sufficient robustness for use in exhaust systems. Non-limiting examples of foams include alpha-Al.sub.2O.sub.3, ZrO.sub.2/MgO, ZrO.sub.2/CaO/MgO and ZrO.sub.2/Al.sub.2O.sub.3.

[0031] Another substrate is a wire mesh and this can comprise the support, i.e. the wire mesh can be both support and substrate, or it can be a substrate to which a catalyst support is coated. Where the wire mesh is both support and substrate, suitable pre-treatment of the wire mesh can be used to key the coating to the wire mesh. Such techniques are known in the art and include a sol base layer or a NaOH(aq) surface pre-treatment. These techniques can also be used to key the catalyst to metal surfaces such as the flow-through monolith and conduit surfaces described above.

[0032] Alternatively, more conventional filter technology such as wall-flow filters used for diesel applications and well known to persons skilled in the art can be used as a catalyst support. A further, less preferred, trapping means utilises thermophoresis and is described in our GB-A-2350804.

[0033] Alternatively, or in addition, the trapping means can comprise a discharging electrode and a collecting electrode for electrostatic deposition of the PM and power means for applying a potential difference between the discharging electrode and the collecting electrode, i.e. the deposition method involves electrophoresis.

[0034] In embodiments of the present invention, geometries of the discharging and collecting electrodes can include: (i) a first cylinder and a wire disposed coaxially therein; (ii) in a variation of the embodiment at (i), a plurality of further wires extending longitudinally relative to the cylinder, which wires are arranged equidistantly and radially about the coaxially disposed wire; (iii) in a variation on the embodiments at (i) and (ii), the first cylinder comprises inner and outer surfaces and an insulating layer is disposed therebetween, which first cylinder is coaxially disposed within a second cylinder, the arrangement being such that the power means is capable of applying a potential difference between the coaxial wire and the inner surface of the first cylinder and between the outer surface of the first cylinder and the second cylinder; (iv) a plurality of first or second cylinders of embodiment (i), (ii) or (iii) above arranged in parallel; and (v) a pair of plates arranged in parallel and a plurality of wires arranged equidistantly in parallel to each other and disposed midway between the plates.

[0035] In one embodiment according to the invention, the collecting electrode in the electrophoretic trapping means can be earthed. In embodiment (i), (ii) and (iv) above, the collecting electrode can be the or each first cylinder; in embodiment (iii), the collecting electrode can be the inner surface of the first cylinder and the second cylinder; and in embodiment (v), the collecting electrodes can be the plates.

[0036] In the electrophoretic trap embodiments of the present invention, the catalyst can be coated, at least in part, on the collecting electrode.

[0037] In a particularly preferred embodiment, the first or second cylinder is a section of the exhaust pipe carrying the exhaust gas. This arrangement has the advantage that it is relatively non-intrusive to the typical configuration of a vehicle exhaust system, i.e. the presence of the coaxial wire(s) and the means for holding the coaxial wire in its coaxial arrangement and for holding any additional wires parallel thereto, e.g. ceramic insulators or ceramic tubes, do not alter the pressure drop in the exhaust pipe. Accordingly, there is little or no impact on the efficiency of the engine.

[0038] An advantage of the present invention incorporating an electrostatic trap, such as the cylinder-wire embodiments, is that they have a relatively low power consumption e.g. about 10 W, therefore there is minimal impact on the overall efficiency of the vehicle.

[0039] According to a further aspect, the invention provides a method of combusting PM from a gasoline engine in CO.sub.2 and/or H.sub.2O from the exhaust gas at temperatures in excess of 500 C., which method comprising trapping the PM and contacting the trapped PM with a catalyst comprising a supported alkali metal. The temperatures in excess of 500 C. can be e.g. 550 C., 600 C., 650 C. or 700 C.

EXAMPLES

[0040] In order that the invention may be more fully understood, the following Examples are provided by way of illustration only:

Example 1

Catalyst Preparation

[0041] A 10 wt % K/Al.sub.2O.sub.3, 10 wt % K/CeO.sub.2 and 10 wt % K/ZrO.sub.2 (as the elemental alkali metal based on the total weight of the catalyst) was prepared by wet impregnation. In each case the impregnation medium was an aqueous solution of KNO.sub.3. A mixture of the correct amounts of the support and impregnation solution was heated to evaporate the water and the material was calcined at 500 C. for 2 hours. Three alumina supports were used: alpha-, theta- and gamma-. We understand that the alkali metal is present as K.sub.2O in each catalyst, although some residual KNO.sub.3 may be present post-calcination.

Example 2

Catalyst Ageing

[0042] Pelletised catalysts of Example 1 were aged in a simulated stoichiometric gasoline exhaust gas mixture of nitrogen, water, carbon monoxide, hydrogen, oxygen, sulphur dioxide at 850 C. for 2 hours and 16 hours.

Example 3

Preparation of Catalyst/Particulate Matter Samples

[0043] Fresh catalyst of Example 1 and aged catalyst from Example 2 were each mixed with about 30-40% w/w carbonaceous PM (BP2000high surface area graphite, a simulant for gasoline exhaust PM) lightly in a mortar and pestle to ensure thorough mixing. Light mixing was employed in order to simulate real conditions in an exhaust system in which the gasoline PM would loosely contact a trap device including a catalyst. Thus heavy grinding, which would promote too tight a contact between the PM and the trap, was avoided.

Example 4

Sample Testing

[0044] The samples of Example 3 were tested in synthetic gas streams including oxygen, water or CO.sub.2 to simulate the temperature and concentration of oxygen, water and CO.sub.2 components found in a gasoline exhaust gas. This approach has the advantage that oxidation in oxygen and the steam gasification and reverse-Boudouard reactions can be studied in isolation, whereas in practice the exhaust gas will comprise a mixture of oxygen, water vapour and CO.sub.2 among others and the three processes can occur simultaneously.

[0045] The samples of Example 3 were tested using Thermogravimetric Analysis (TGA) to determine the temperature where significant weight loss of the samples occurred, indicating the point where combustion or any phase changes within each sample took place. The technique operates using a microbalance, which measures weight change as a function of temperature. When the combustion temperature is reached, the carbon is removed as either CO.sub.2 or CO and therefore the weight of the sample decreases.

[0046] Temperature Programmed Oxidation (TPO) was used to measure the effect of water as a source of oxygen for combustion. TPO involves passing helium through a water saturator at room temperature and then over the catalyst. The production of CO.sub.2 and CO is then monitored as a function of temperature.

[0047] The conditions used for each of these techniques are described in Table 1.

TABLE-US-00001 TABLE 1 Conditions used for the Combustion Reactions Flow-rates Final Temp. Ramp-rates Technique Gases (ml min.sup.1) ( C.) ( C. min.sup.1) TGA Air 100 700 15 10% CO.sub.2/Ar 100 1200 15 Ar 100 1200 15 TPO H.sub.2O/He 75-80 900 10

[0048] The results for the samples are given in the attached Tables 2, 3 and 4. FIG. 1 summarises the results for 10 wt % K/Ceria, showing that the temperatures for combustion of PM in oxygen, water and CO.sub.2 are within the normal operating temperature of a gasoline exhaust system for a European passenger vehicle.

[0049] Reactions involving gas phase reactants can use conventional heterogeneous catalysts. However, catalysts for reacting solid carbon with carbon dioxide, steam or oxygen, preferably have different properties from conventional heterogeneous catalysts. In order to catalyse the reaction with solid carbon it is understood that the catalyst should come into contact with the carbon material to create a boundary across which oxygen can be transferred. To achieve this, it is preferred that the active component of the catalyst is mobile at reaction temperature. This is illustrated by comparison of the potassium-based catalysts with a Pt/Zirconia catalyst. Combustion in oxygen using 10 wt % K/Zirconia is initiated at 391 C., (an improvement on the non-catalytic reaction that takes place at approximately 550 C.). In comparison, when 1 wt % Pt/Zirconia is used, temperatures as high as 523 C. are required. The reason behind this observation is probably that at 350 C. the active potassium oxide begins to melt and so becomes mobile although still retained within the pore structure of the zirconia. In contrast, platinum oxide decomposes at approximately 550 C., and the metal particles formed are immobile until higher temperatures are reached, consequently limiting the contact between catalyst and soot at lower temperatures.

[0050] As can be seen from the results shown in Tables 2, 3 and 4, the catalysts of the present invention reduce the combustion temperature of simulated gasoline PM to gasoline exhaust gas temperatures. Furthermore, it can be seen that lower surface area supports retain catalyst activity at lower temperatures following high temperature ageing. This is believed to be because the metal compounds are mobile at active temperatures and they can migrate deep into the smallest pores of higher surface area supports, thereby reducing the effective surface area of the alkali metal and making it less available to catalyse the relevant reaction.

[0051] Furthermore, the results in the Tables 2, 3 and 4 show that the support material can have a role to play in the chemistry of the process. The initial combustion temperatures of the reactions of oxygen, water and carbon dioxide with carbon can be reduced further when using 10 wt % K/Ceria and 10 wt % K/Zirconia than for -alumina supported catalyst. The benefit is most noticeable in the case of 10 wt % K/Ceria for the reaction of carbon dioxide with PM: combustion starts to take place at 712 C. for the fresh catalyst, increasing slightly to 746 C. and 748 C. for the 2 hours and 16 hours aged samples. In the case of 10 wt % K/Zirconia, the most significant improvement occurs in the steam gasification reaction. Initial combustion is detected at 477 C. for the fresh catalyst again increasing slightly to 477 C. and 532 C. for the 2 hours and 16 hours aged samples.

[0052] It should be noted that all of the above catalysed reactions, with the exception of 1 wt % Pt/Zirconia, occur at significantly lower temperature than the non-catalysed reactions. This improvement could be due to the redox properties of these support materials i.e. both ceria and zirconia can undergo oxidation or reduction readily depending on the conditions (CeO.sub.2Ce.sub.2O.sub.3, ZrO.sub.2Zr.sub.2O.sub.3).

[0053] Another property of the catalysts of the present invention is that it is understood that they can activate carbon dioxide and water in order to transfer the oxygen to trapped PM. At present the reason for this improvement is not fully understood, but we believe it is the ability of the alkali metals to form reactive intermediates in the presence of carbon dioxide and water that make them the most suitable catalysts for this purpose. Again, comparison of the potassium based catalysts with lwt%Pt/Zirconia, relative to the non-catalysed reactions, show that the alkali metals can reduce the initial combustion temperatures by up to 300 C. unlike the platinum catalyst, which only reduces the temperature by approximately 30-50 C.

Example 5

Particulate Filters

[0054] 25 mm-diameter152 mm length cores were extracted from Corning EX-80 100/17 cordierite wall-flow filter.

Tests in Laboratory Flow Rig

[0055] Carbon particles were generated by spark discharge between graphite electrodes in a Palas GFG1000 carbon aerosol generator (CAG). 10-200 nm agglomerates from the

[0056] CAG, in an inert carrier/dilution gas (argon/nitrogen), were then dispersed in a synthetic gas mixture which can include water and hydrocarbon vapours. While the elemental carbon fraction of particulates from PFI engines is normally not more than 40%, considerable variability from similar light-duty gasoline vehicles has been found, with significant elemental carbon emissions in some cases. GDI engines can produce particulates more closely resembling those from diesel engines, up to 72% carbon being reported, so this simulation approach is particularly relevant to GDI engines.

[0057] The particle-laden gas flowed in a straight, 50 mm diameter, insulated, stainless steel pipe, 3 m in length, to which after-treatment devices could be attached in-line. An in-line process heater and heating tape were used to achieve the desired uniform gas temperature. Particles were sampled using 6.5 mm stainless steel probes and measurements of number concentration and size distribution were made using a TSI Scanning Mobility Particle Sizer (SMPS) comprising a model 3081L electrostatic classifier (EC) and a model 3022 condensation particle counter (CPC). To keep sample temperature below the 36 C. limit, samples were diluted with particle-free nitrogen in a three-stage ejector system.

[0058] The laboratory simulation test rig for trap evaluation is shown schematically in FIG. 2.

[0059] For the tests reported here, the synthetic gas was composed only of particle-free air together with the CAG carrier/dilution gas. Tests were carried out at a fixed gas mass flow rate and temperatures T between ambient and 400 C. and for filter exposure times up to 20 hours. [Temperatures were limited to a maximum of 400 C. by the heating sources used in the laboratory test rig. However, the trends shown in the results obtained allow us to predict system performance at higher temperatures, e.g. at least 500 C.] The chosen gas flow rate and particle production rate provided particle mass concentrations of 1.06 mg/m.sup.3 and 0.46 mg/m.sup.3 at T=20 C. and 400 C. respectively, within the range 0.1-10 mg/m.sup.3 typical of gasoline engine exhausts. The filter space velocity (gas actual volume flow rate divided by filter volume) was approximately 44000 h.sup.1 at 20 C. and 101000 h.sup.1 at 400 C. Measurements were taken at locations 100 mm upstream and downstream of the filter. The SMPS particle size window was set to 9-422 nm and scan times for up and down scans were chosen as 120 seconds and 60 seconds, respectively; the time interval between measurements upstream and downstream was 5 min. A sample dilution ratio of 60 was used.

[0060] The results obtained are shown in Table 5.

TABLE-US-00002 TABLE 5 Total Number particle number particle concentration Temper- Air flow rate concentration average peak ature (L/min at average reduction reduction ( C.) T = 22 C.) (%) (%) 22 35 54 50 200 35 57 54 400 35 54 46

[0061] At 400 C., the measured capture efficiency as a function of particle size was between 65% and 90% throughout the ultrafine size range, with increased scatter below 20 nm where measurement uncertainty was expected to be highest and above 100 nm where the smallest numbers of particles were detected. Little change in the pattern was seen after an exposure time of 19 hours, when the pressure drop had increased by a factor of more than four. (With effective continuous or intermittent oxidation of trapped PM, pressure drop should be of less concern.) The time-averaged size distributions show no shift in mode size across the filter

Tests on GDI Engine

[0062] A filter of identical type, but with 118 mm diameter, was fitted to the exhaust system of a Mitsubishi Carisma car with 1.8 litre GDI engine, for comparison with the standard exhaust system. The filter location was approximately one third of the way between engine exhaust manifold and tailpipe exit. With the car on a chassis dynamometer, raw exhaust samples were extracted at the tailpipe though a stainless steel sampling probe and line. Samples were first diluted in a Dekati two-stage Diluter, with a dilution ratio of 74, then analyzed using a Dekati Electrical Low Pressure Impactor (ELPI), a 12-channel real-time particle size spectrometer covering the size range 0.03-10 m.

[0063] Measurements with and without the filter were made at two conditions: (a) nominally steady state at an equivalent road speed of 80 km/h, where fuel injection was in the stratified overall-lean mode with a measured equivalence ratio of around 1.8; (b) transient, over the European ECE+EUDC drive cycle. FIG. 3 shows the time-resolved particle number density from the four lowest-size stages (channels) of the ELPI, during five-minute portions of the steady-state tests. The baseline measurements (dashed curves, left-hand scale) were taken immediately after the car had been taken through a drive cycle, and the plot includes periods in both 4th and 5th gears. With the filter fitted and pre-loaded during one hour of operation (solid curves, right-hand scale), the number density had fallen by an order of magnitude, for all sizes in the range 30-170 nm. Beyond 170 nm, particles were not detected in significant numbers, with or without the filter. Thus the filter appeared to perform as well as in the laboratory simulation, an unsurprising result.

[0064] Over the ECE and lower-speed portion of the EUDC cycle, the filter reduced tailpipe number concentrations in the ultrafine range to values very significantly below the baseline levels, as illustrated in FIG. 4. No significant change in filter performance over the cycle was seen between filters pre-loaded for 1 hour and for 4 hours.

Example 6

Electrostatic Separators

[0065] A wire-cylinder electrostatic trap consisting of a grounded stainless-steel pipe of 50 mm diameter and 1 m length with an axial discharge electrode of 0.1 mm tungsten wire was constructed. The dimensions were selected on the basis of estimates of the axial distance required for 100% particle capture at typical gas velocities in a laboratory simulation, supported by analytical modelling. The wire was energized using a Start Spellman SL300 high voltage power supply and kept both taut and centrally located by an insulated tensioning device which allows the trap to be installed in-line with a simulated exhaust pipe, in place of the filter shown in FIG. 2. Particle charging was expected to occur through ion bombardment in the corona discharge, prior to transport to the pipe wall by electrophoretic drift.

[0066] A simulated gasoline engine exhaust (generated as described in Example 5) dispersed in particle-free nitrogen, was fed to the trap at ambient temperature. The residence time within the trap was <1 s, ensuring turbulent flow. With the field, at V=7 kV and with a corona current around 1 mA, the particle numbers at outlet were many orders of magnitude smaller than at inlet, over most of the size range and for all the radial sampling positions (measured within a 15 min period). In the 10-20 nm range, the reduction in particle number was still two orders of magnitude. Evidence that particles had been deposited, rather than remaining gas-borne with a radial shift, was provided by observation of a thin, uniform, black deposit on the trap wall after some 3 hours of running.

[0067] The experiment was repeated over a range of voltages and the results are shown in FIG. 5. As can be seen, the particulate separation in this arrangement reaches a plateau above about 12 kV.

Example 7

Catalysed ESP

[0068] A simulated gasoline engine exhaust was heated to 400 C. and fed to the trap (described in Example 6) at that temperature. The residence time within the trap was 0.6 s, ensuring a realistic turbulent flow. With the electric field maintained at 8 kV, the particle trapping efficiency was measured as a function of particle diameter. After completion of this test, the simulated exhaust pipe (see FIG. 2) was replaced by an identical pipe which had been coated in 10 wt % K/gamma-alumina catalyst. The results, shown in FIG. 6, appear to indicate that the catalyst does not materially affect the performance of the separator, and if anything, provides a slight improvement in collection efficiency for larger particle sizes.

TABLE-US-00003 TABLE 2 Combustion Characteristics for Fresh and Aged Catalysts in O.sub.2 (using TGA) 2-Hour Aged 16-Hour Aged Fresh Initial Initial Initial Combustion Combustion Combustion 2-Hour Aged T.sub.50 16-Hour Aged Catalyst Temp. ( C.) Temp. ( C.) Temp. ( C.) Fresh T.sub.50 ( C.) ( C.) T.sub.50 ( C.) None 550 625 10% K/-Alumina 410 510 527 448 560 579 10% K/-Alumina 414 502 499 444 551 548 10% K/-Alumina 410 461 468 439 564 542 10% K/Ceria 407 499 479 444 574 576 10% K/Zirconia 391 479 505 432 564 593 Key: T.sub.50 is the temperature at which 50% of the weight of soot is removed.

TABLE-US-00004 TABLE 3 Combustion Results for Fresh and Aged catalyst in CO.sub.2 (using TGA). 2 hour-aged 16 hour-aged Fresh Initial Initial Initial Combustion Combustion Combustion 2 hour-aged T.sub.50 16-hour aged T.sub.50 Catalyst Temp. ( C.) Temp ( C.) Temp ( C.) Fresh T.sub.50 ( C.) ( C.) ( C.) None 979 1130 10% K/-Al.sub.2O.sub.3 743 808 836 808 891 917 10% K/-Al.sub.2O.sub.3 743 773 813 804 888 923 10% K/-Al.sub.2O.sub.3 731 739 757 808 819 819 10% K/CeO.sub.2 712 746 748 786 817 832 10% K/ZrO.sub.2 730 748 763 804 815 836 Key: T.sub.50 is the temperature at which 50% of the weight of soot is removed.

TABLE-US-00005 TABLE 4 Combustion Results for Fresh and Aged catalyst in H.sub.2O (using TPO). 2 hour-aged 16 hour-aged Fresh Initial Initial Initial Combustion Combustion Combustion 2 hour-aged T.sub.50 16 hour-aged T.sub.50 Catalyst Temp. ( C.) Temp ( C.) Temp ( C.) Fresh T.sub.50 ( C.) ( C.) ( C.) None 850-900 >900 10% K/-Al.sub.2O.sub.3 505 694 673 751 >900 >900 10% K/-Al.sub.2O.sub.3 525 699 690 709 >900 >900 10% K/-Al.sub.2O.sub.3 505 585 669 736 840 >900 10% K/CeO.sub.2 550 597 599 753 773 815 10% K/ZrO.sub.2 477 477 532 755 759 823 Key: T.sub.50 is the temperature of half-peak maximum temperature.