The Reduction of Nitric Oxide in Automobile Emissions
The Reduction of Nitric Oxide in Automobile Emissions
Stabilisation of Catalysts Containing Ruthenium
Catalysts containing ruthenium are used to reduce nitric oxide in automobile exhaust emissions but volatile oxides formed under transient oxidising conditions during operation can severely affect them. However, basic oxides such as alkaline earths or rare earths can stabilise the ruthenium by forming ruthenates, which have little tendency to volatilise. The ruthenates are synthesised either before the catalyst is made or in position on the catalyst support, which may be monolithic or pelleted.
In the development of catalysts for the reduction of nitric oxide in the purification of automotive exhaust emissions one of the desired attributes of a catalyst is its ability to promote selectively the reduction of nitric oxide NO to molecular nitrogen (and not to ammonia as happens over a large majority of catalysts), in the presence of hydrogen-containing compounds in the gas stream (1, 2, 3). Ruthenium-containing catalysts have been found to have a pronounced selectivity for reduction to molecular nitrogen (4, 5, 6), and several laboratories have pursued recently the development of practical ruthenium-containing catalysts.
The NOx-reduction catalysts are intended to operate in an overall reducing atmosphere in an automobile exhaust which is, on the average, a product of the carburetion of a combustible mixture richer than the stoichiometric mixture. However, relatively frequent excursions occur in the operation of the vehicle into conditions where the exhaust contains, for certain time intervals, a net excess of oxygen. Further, in some embodiments of the so-called dual catalyst system it is desirable to use the NOx-reduction catalyst as a carbon monoxide and hydrocarbon oxidation catalyst during the warm-up period, deliberately admitting an excess of oxygen during that time.
With the use of ruthenium-containing catalysts it soon became apparent that their stability under oxidising conditions is poor. Analysis of spent catalysts revealed severe losses of the active component, readily explained by the formation and removal of the volatile tetroxide (7).
Various ways were explored to minimise the tendency of the ruthenium to volatilise under oxidising conditions while still preserving the high activity and selectivity of the catalysts.
We have attempted the stabilisation of the ruthenium (or its oxide) on the catalyst by the incorporation of the basic oxide of barium. The reasoning leading to this suggestion was couched in the following chemical terms: since the highly oxidised states of ruthenium (Ru3+ and higher) give rise to acidic oxides, their interaction with basic oxides such as barium oxide will lead to the formation of double mixed oxides which should be more stable with respect to volatilisation in the presence of oxygen. Indeed, the higher oxidation states of Ru do form an extensive series of ruthenates, many of the latter well characterised in the literature (8). In parti cular, the ruthenates of the alkaline-earth metals have been investigated by Ward and co-workers (9). They tend to crystallise mainly with the perovskite structure.
Barium ruthenate BaRuO3, in particular, has been well characterised (9a). It has a close-packed stacking of close-packed BaO3 layers with the Ru ions in octahedral sites. There is metal-metal bonding between the ruthenium ions with an interatomic Ru-Ru distance of 2.55 Å. This is a closer approach than in ruthenium metal (2.65 Å). This structural property may have a bearing on the remarkable catalytic activity of the barium ruthenate which will be discussed later.
Another class of basic metal oxides useful for the stabilisation of ruthenium in an oxidising atmosphere is that of the rare-earth metals. With all the lanthanides, except lanthanum itself, one obtains compounds of the type (RE)2Ru2O7 which crystallise as cubic pyrochlores (10). The interaction between La2O3 and RuO2 leads to a perovskite phase (11) which has recently been well characterised by Bouchard and Weiher (12). It is worth noting that La2O3 is the most basic among the rare-earth oxides and therefore is most suitable for stabilisation. The perovskite LaRuO3 is considered to be unusual, as it is the only example of an oxide lattice where all the ruthenium is in the formally trivalent state. It is well known that the valence state of the B metal ion in the perovskite formula ABO3 can be varied in wide limits by substituting mono-, di- or trivalent atoms in the A position. On the other hand, different atoms can be also substituted into the B position. This gives a very wide range of possible ruthenium-containing ‘stabilised’ catalysts.
The work referred to in the present article is limited specifically to the stabilisation by BaO or La2O3. The subject-matter is the basis for several U.S. patent applications.
Preparation of Catalysts
We have followed essentially two routes in the preparation of the stabilised catalysts. In one technique BaRuO3 or LaRuO3 is pre-synthesised. The presynthesised ruthenates were finely ground and incorporated on to monolithic supports either from American Lava Corp. or from Corning Glass Works using very dilute (2 wt. per cent) suspensions of colloidal ‘Dispal’ alumina as the binding agent. There is an obvious disadvantage to the technique, and this is the limit on the particle size achievable by the grinding. At best, the particle diameter is of the order of a few microns (or a few 104 Å), which leaves the major part of the valuable ruthenium atoms in the bulk and inaccessible to the reacting molecules on the surface. On the other hand, there is the obvious advantage that every ruthenium atom has been stabilised as the ruthenate.
The second technique consists of the impregnation of either a pelleted or monolithic support, first by a solution of the nitrate of the stabilising metal barium or lanthanum, followed by calcination to convert the nitrate to oxide, secondly by the solution of ruthenium trichloride, RuCl3, which is then dried at 110°C and reduced in hydrogen at ∼450°C. The very small particles of the ruthenium metal (< 10 Å diameter) are then ‘fixed’ by rapid heating in air at 900°C. The fast fixation process assures very good dispersion of the resulting ruthenate. There is no direct evidence, however, that every ruthenium atom has been deposited in the vicinity of the stabilising oxide so as to assure the formation of the ruthenate.
Catalysts containing from 200 to 2000 p.p.m. ruthenium by weight were prepared by both methods.
Stability of Ruthenates in Oxidising Atmospheres
To check the stability of the BaRuO3 and LaRuO3 and compare it to that of RuO2, thermogravimetric experiments were performed using the DuPont automatic balance. Figure 1 indicates the vast difference in the weight loss between the ruthenates and RuO2 at 1180°C in flowing air.
Table I shows the percentage loss of ruthenium in a fast-flowing blend of nitrogenoxygen from supported pelleted catalysts containing ruthenium alone or with BaO or La2O3. It should be pointed out that, this being one of the earliest preparations, the ruthenium was not prefixed by prereduction and subsequent rapid oxidation and therefore was relatively poorly dispersed (4.6 per cent of the ruthenium atoms are surface atoms). None the less, the vast improvement in stability due to the presence of BaO and La2O3 is quite obvious.
The cycling conditions and the possibility of switching from oxidising to reducing exhaust in the operation of an engine requires the knowledge of the changes that the stabilised catalyst undergoes under such treatment. For this purpose presynthesised ruthenates were subjected in the TGA balance to alternate oxidising and reducing conditions. The reducing agents were hydrogen or carbon monoxide (in nitrogen). Since the use of carbon monoxide leads under certain conditions to the formation of barium or lanthanum carbonates, we give only the data for reduction by hydrogen, and only the results for BaRuO3 are presented. The results for LaRuO3 are analogous. Figure 2 indicates that the BaRuO3 is reduced to BaO + Ru and that the weight is restored upon exposure to air. X-ray diffraction patterns after the four cycles indicate the restoration of the ruthenate, BaRuO3. It is possible that frequently repeated oxidation-reduction cycling engenders the migration of the ruthenium to form gradually a separate oxide phase. To prevent this, the stabilisation of very small particles as achieved by the fixation is very important. Another possible stratagem is to dilute the perovskite ABO3 in the B position by nonvolatile ions such as Mn4+(9a) or others. In lanthanum ruthenate, where ruthenium is trivalent, dilution by trivalent ions such as Ni3+ and others should be feasible.
Activity and Selectivity of Ruthenates and Stabilised Ruthenium Catalysts
Under reducing conditions at relatively elevated temperatures the barium (or lanthanum) ruthenate reverts, as shown in the previous section, to the ruthenium metal and the stabilising oxide. Hence, its activity and selectivity under these conditions might be expected to be the same as in unstabilised Ru catalysts.
At lower temperatures and under a redox potential not quite sufficient to reduce the ruthenates, the catalyst might be required to catalytically reduce nitric oxide while the ruthenium is still coordinated in the ruthenate structure.
The available literature indicated that from simple considerations the ruthenates could be expected to be active catalysts. Thus, first, the perovskite-type ruthenates have been shown to have metallic conductivity, for which convincing explanations have been proffered (12, 13). Secondly, the Ru-Ru distances in some of the structures were shown to be actually shorter than in the metal (9a).
The metallic character should satisfy one of the important conditions required for the redox catalytic activity, the ability to chemisorb both electron-donor and electron-acceptor adsorbents. The close proximity of the ruthenium atoms should provide the dual sites required for the pairing of nitrogen atoms in the reduction of nitric oxide to nitrogen.
Tables II and III give the comparison between a pelleted catalyst containing 2000 p.p.m. of ruthenium alone and the catalysts containing the same amount of ruthenium but stabilised by 3 per cent barium as BaO or 3 per cent lanthanum as La2O3. As seen from the tables, the stabilisation does not impair the activity and affects the selectivity only very slightly, if at all. At the conditions of these runs the catalysts were not reduced to the metallic state of ruthenium, because the temperature was too low (see Fig. 2).
Table IV gives the activity and selectivity in laboratory conditions of a presynthesised BaRuO3 catalyst deposited on to a monolithic support. Similar behaviour is exhibited by the presynthesised LaRuO3.
The monolithic catalysts made in the laboratory were subjected to prolonged durability tests on engines mounted on dynamometers. Many formulations were tested, but we limit ourselves here to the presentation of a few examples only. The compositions of the catalysts of which the dynamometer tests results are discussed here are given in Table V. The last column in the table gives the performance of a fresh catalyst in a high-space-velocity screening test.
|Catalyst Identification||Stabilising Element||Ru Loading p.p.m.||NO Reduction Activity in Screening Dynamometer Test at 230,000/h, A/F ratio 13.8, 1000°F|
The operating parameters of the engine dynamometer are given in Table VI. The engine was a six-cylinder 200 CID Ford engine, and the exhaust of each cylinder was led over a three inch diameter, three inch long monolithic catalyst.
Lead-sterile fuel and ashless oil were used for the durability tests.
Figures 3, 4 and 5 give, respectively, the durability testing results in the form of percentage of nitric oxide removed and ammonia formed as a function of air/fuel ratio at increasing times of running for catalysts M–176A, M–177AP2 and M–168. The decrease in activity for the first two catalysts is very minor, notwithstanding the low loading of ruthenium. On Fig. 5 it can be seen that catalyst M–168 has suffered a sharp drop in activity in the period between 79 and 138 hours of operation. This was traced to a sharp overheating associated with a malfunction of the system. The malfunction resulted in the physical damage shown in Fig. 6. Nevertheless, a substantial activity was still exhibited by the catalyst as shown by the plots on Fig. 5 at times beyond 138 hours of operation.
The durability results accumulated in actual vehicle operation are included here to demonstrate that, notwithstanding the encouraging laboratory and dynamometer activity and stability, the catalysts deteriorated more rapidly in vehicle tests than in laboratory or dynamometer evaluation. The deterioration was faster than required for the satisfaction of the U.S. Federal 1976 requirements. As a substantial number of these catalysts have been analysed and some have shown no appreciable loss of the active component ruthenium, it is obvious that other causes of deterioration, such as poisoning or overheating (or both) operate also in the catalysts for the reduction of NOx, which is hardly surprising.
The data in Table VII were obtained on cars equipped with exhaust gas recirculation devices, with the exception of vehicle 22–C–58. The EGR was calibrated to substantially reduce inlet levels of NOx, as seen in the table. This may have resulted in increased HC and CO emissions, which does, however, not bear directly on the subject-matter of the paper. This explains the low level of NOx emissions without the use of the catalysts. The data in the table clearly indicate that in vehicle operation the ruthenium catalysts deteriorate. The stabilisation procedure is designed to minimise the deterioration associated with the volatilisation of the Ru. The extent of success in preventing volatilisation can be judged from the analytical data of the next paragraph.
|Vehicle||Initial CVS (C/H) Results, NOx||Durability CVS (C/H) Results, NOx|
|Wt. % Catalyst||With Catalyst||Efficiency (%)||Mileage||Wt. % Catalyst||With Catalyst||Efficiency (%)|
|21A91 (M−177B) Galaxie, 429 CID||1.40||0.45||68||12,000a||1.05||0.77||26.6|
|31A73 (M−190) Galaxie, 400 CID||1.27||0.48||62||16,000b||1.25||0.77||38.4|
|22−C−58 (M−176B and M−177) Capri, 2.6 I, d||2.5||0.47||82||8,000b||2.0||0.92||45.5|
|110−T−713 (M−150C2)c Galaxie, 351 CID||1.19||0.39||67||10,000a||1.49||0.87||41.6|
|110−T−714 (M−190) Galaxie, 351 CID||1.69||0.44||74||12,500b||1.20||0.64||46.6|
|110−T−718 (M−150G)c T-Bird, 429 CID||0.83||0.26||76.6||5,000a||1.14||0.93||18.4|
Analysis of Used Catalysts
To assess the extent of loss of ruthenium from used catalysts the chemical analysis of this element was carried out by two different analytical techniques.
The X-ray fluorescence analysis using the Siemens SRS−1 instrument was carried out for a series of elements present on the catalyst. These included both the active elements added in the preparation of the catalysts and some of the contaminants accumulated during testing. Since this article deals specifically with stabilisation, the analysis results presented here are limited to ruthenium only. As mentioned before, the presence of other contaminants may have an effect on the catalyst deterioration noticed in Table VII and this matter is now under intensive investigation.
A word of caution is in order with respect to the X-ray fluorescence analysis. The presence of several contaminants on the used catalysts requires the preparation of a large matrix of standard samples. The matrix was prepared using a support from only one of the suppliers, while the actual catalysts were made with different support materials. Since all of them (with the exception of catalyst M–150G) were aluminosilicates, the effect on the analysis should not be large. Secondly, not all the interelement intensity corrections, associated with minor contaminants, were accounted for in the X-ray fluorescence analysis.
The activation analysis for ruthenium was carried out by irradiation for three minutes in the core of the Ford Nuclear Reactor at the University of Michigan. After a decay period of ∼14 days, the samples were counted on a high-resolution gamma ray spectroscopy system. The 0.497 MeV radiation from 103Ru (half life 39.5d) is analysed by computer for the quantitative determination. A ground specimen of the sample is compared with another specimen to which a known amount of ruthenium has been added. Both samples are irradiated simultaneously, and the flux variations are not more than 1 per cent.
Examination of Table VIII shows considerable improvement in the prevention of ruthenium volatilisation under the conditions of dynamometer testing. This particular cycle consisted of 48 minutes of testing at 870°C, 140,000/h, under reducing conditions and 6 minutes of testing at 650°C, 50,000/h under oxidising conditions. The stabilisation without reduction followed by fixation brings about improvement, but not complete prevention. The incorporation of the fixation step virtually eliminates ruthenium loss under the dynamometer test conditions.
|a: Fresh Catalysts|
|Catalyst Designation||Nominal Ruthenium Content p.p.m.||Average Ruthenium Content X-ray Fluorescence p.p.m.||Average Ruthenium Content Activation Analysis p.p.m.|
|b: Dynamometer Tested Catalysts|
|Catalyst Designation||Hours of Service||Nominal Ruthenium Content p.p.m.||Average Ruthenium Content X-ray Fluorescence p.p.m.|
|c: Vehicle Tested Catalysts|
|Catalyst Designation||Mileage||Nominal Ruthenium Content p.p.m.||Average Ruthenium Content X-ray Fluorescence p.p.m.||Average Ruthenium Content Activation Analysis p.p.m.|
|M−150G (right) a||5000||2000||1053||2144|
|M−150G (left) a||5000||2000||851||not measured|
|M−150C−2A a||10,000||2000||540||not measured|
|M−150C−2B a||10,000||2000||600||not measured|
Under vehicle conditions the loss from the first generation of stabilised catalysts (the M–150) series is still higher than acceptable. The data from the second generation of catalysts are not all available as yet, as these are still being tested on vehicles and the mileage accumulation is continuing (see Table VII).
In the single case where the analysis is available (catalyst M–177B, after 12,000 miles, Table VIII), there is still an apparent loss of ruthenium. This qualification is associated with the fact that no analysis is run on a fresh catalyst and the nominal intended loading can be different from the actual loading. It is, however, realised that the achieved degree of stabilisation can still be substantially improved, while mastering the technique of catalyst preparation. Further improvements can be expected by minimising the opportunities for phase separation during repeated oxy-reduction of the catalyst. This will mainly be achieved by ruthenium dilution and smaller particle-size distribution. On the other band, better control of the air/fuel ratio in the operation of the vehicle system should also minimise the loss of ruthenium.
Another area currently under intensive investigation is the poisoning of the ruthenium-containing catalysts in the environment exhaust. A detailed effort using the techniques used previously in the poisoning studies of oxidation catalysts (14) is in progress.
We owe gratitude to many more individuals at the Ford Motor Company than we can list by name. Most importantly, we acknowledge the help of Joe Kummer, Vern Bergman, Eric Daby, Carol Smith, Gene Hancock, Henry Stepien and numerous members of the Technical Services Department.
G. H. Meguerian and C. R. Lang, S.A.E. Preprint 710,291, 1970
M. Shelef and H. S. Gandhi, Ind. Engng. Chem., Prod. Res. Dev., 1972, 11, 2
R. L. Klimisch and G. J. Barnes, Environmental Sci. Tech., 1972, 6, 543
M. Shelef and H. S. Gandhi, Ind. Engng. Chem., Prod. Res. Dev., 1972, 11, 393
R. L. Klimisch and K. C. Taylor, Environmental Sci. Tech., 1973, 7, 127
K. Otto and M. Shelef, Z. Phys. Chem. ( Frankfurt ), 1973, in press
For a concise summary on the vapour pressure of ruthenium oxides see: Gmelin’s Handbuch der Anorganische Chemie, 8 Auflage, System Number 63, Erg.–Band, 1970, 232 et seq.
Ibid., 252 et seq.
(a) P. C. Donohue,, L. Katz and R. Ward, Inorg. Chem., 1965, 4, 306 ; (b) Ibid., 1966, 5, 335 and 339 . (See also references cited therein)
M. C. Montmory and F. Bertant, Compt. Rend., 1961, 252, 1471 (cited in ref. 7, p. 258 )
F. Bertant,, F. Forrat and M. C. Montmory, Compt. Rend., 1959, 249, 829
R. J. Bouchard and J. F. Weiher, J. Solid State Chem., 1972, 4, 80
J. B. Goodenough, Bull. Soc. Chim. Fr., 1965, 1200
M. Shelef et al ., ‘Poisoning of Monolithic Noble Metal Oxidation Catalysts in Automobile Exhaust Environment’, A.I.Ch.E. 74th Nat. Mtg., New Orleans, Mar. 11 – 15, 1973