108
Ind. Eng. Chem. Res. 1999, 38, 108-117
Behavior of Non-Promoted and Ceria-Promoted Pt/Rh and Pd/Rh Three-Way Catalysts under Steady State and Dynamic Operation of Hybrid Vehicles Sergio Tagliaferri, Rene´ A. Ko1 ppel, and Alfons Baiker* Laboratory of Technical Chemistry, Swiss Federal Institute of Technology, ETH-Zentrum, CH-8092 Zu¨ rich, Switzerland
The suitability of newly developed non-promoted and ceria-promoted palladium-rhodium automotive catalysts for the exhaust gas control of a hybrid drive system has been tested by light-off experiments with steady and cycling feed stoichiometry, and pulsed-flow operation. The dynamic behavior of the honeycomb-type catalysts has been compared to the performance of standard honeycomb platinum-rhodium catalysts. Light-off tests carried out in the range 150500 °C indicated significant differences in the conversion of NOx, CO, and hydrocarbons during warm-up, depending on the catalyst composition and whether λ-cycling was applied or not. Appropriate λ-cycling substantially improved the behavior of all catalysts in the lower light-off region. At higher temperatures cycling afforded lower conversions of all target components. Under pulsed-flow operation with an air pulse preceding the exhaust pulse (filling of the cylinder with air), asymmetric λ-cycling with longer rich half-cycles resulted in CO and HC conversions as well as N2 yields similar to or higher than those without an air pulse for the ceria-promoted catalysts. The catalytic tests suggest that ceria-promoted palladium is competitive to Pt-RhCe in a hybrid vehicle application. Introduction In view of tightened regulations for low-emission vehicles, hybrid vehicle technology is progressing worldwide in an attempt to optimize overall vehicle performance, fuel economy, and emissions.1 Hybrid drive systems generally make use of the synergetic combination of a combustion engine with an electric motor. As part of an interdisciplinary project, we are working on emission control catalysis for an advanced hybrid concept, consisting of an internal combustion engine, an electric motor, a continuous variable transmission, and a flywheel as a short time energy storage medium.2,3 This configuration allows operation of the combustion engine close to its economic optimum in a so-called intermittent mode. Driving energy is taken from the flywheel, which is recharged by operating the combustion engine for 3 s in intervals of about 17 s, thus resulting in a pulsed-flow operation of the catalytic converter. Each exhaust pulse is preceded by an air pulse, which results from the filling of the cylinders with air at start-up and shut-off.3 Various studies on the transient behavior of threeway catalysts (TWC) have been performed. Because of the steplike response of the oxygen sensors, catalytic converters in automobiles are periodically forced about the stoichiometric air-fuel ratio at a frequency of about 1 Hz and a small amplitude.4-7 In a recent review, Silveston8 concluded that this periodic forcing suppresses rather than enhances conversions under normal operating conditions in the 400-600 °C temperature range. Other authors reported that under cycling conditions the catalytic activities of three-way automotive * To whom correspondence should be addressed. Tel.: (+41-1) 632 31 53. Fax: (+41-1) 632 11 63. E-mail: baiker@ tech.chem.ethz.ch.
catalysts can be superior compared to those under static conditions, depending on the temperature, cycling period, and feed stream conditions.9-11 Cycling at temperatures below the light-off temperature was found to increase conversions of NOx, CO, and hydrocarbons, whereas the effect of λ-cycling was negative at higher temperatures.12 It was shown that the addition of base metal oxides such as ceria to the catalyst formulations can buffer excursions into the lean or rich region.13 Palladium has been used in exhaust catalysts since the very beginning in the early 70s, when in the U.S. the first automotive exhaust gas cleaning systems were commercialized.5-7 Two major factors have lead to a renewed interest in palladium as the main noble-metal component for three-way catalysts. The first factor is purely economical, since palladium is presently the cheapest precious metal applicable in exhaust catalysis. Second, Pd-based techniques seem suitable in meeting the stringent future emission regulations, which will require better cold-start performance.14-17 Several research groups have presented a new generation of palladium catalysts with and without the addition of Rh.10,15-20 Fleet tests with new technology Pd-Rh catalysts have shown that Pd-based catalysts may be suitable for practical use in automotive exhaust catalysis.21 To enhance the performance of Pd-based catalysts, different approaches have been tested, such as the use of Pd-Pt mixtures,22 modification of the noble-metals impregnation method,23 changes of the wash-coat crystalline structure,24 and alteration of the wash-coat composition. The classical promotion by ceria has been complemented by adding lanthana and other rare earth or alkaline metal oxides.10,20,25,26 Better NOx reduction performance19,20 with and without the addition of base metal oxides, as well as improved high-temperature stability14,20 and better HC conversion behavior,27 have
10.1021/ie980249d CCC: $18.00 © 1999 American Chemical Society Published on Web 11/25/1998
Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 109 Table 1. Composition and Denotation of Tested Catalysts catalyst denotation Pd-Rh Pd-Rh-Ce Pt-Rh Pt-Rh-Ce
Pd
wash-coat composition (wt %) Pt Rh Al2O3 CeO2
1 1 1 1
0.2 0.2 0.2 0.2
98.8 86.8 98.8 86.8
12 12
been reported. The addition of ceria as one of the most important promoters was observed as stabilizing the alumina wash coat,28,29 increasing and stabilizing the noble-metal dispersion,30 enhancing the oxygen storage capacity of the catalysts,29 and enhancing the activity for the water-gas shift reaction (WGSR).29,30 In this study, a set of commercially available new technology non-promoted and ceria-promoted Pd-Rh catalysts have been compared with analogous standard Pt-Rh three-way catalysts. λ-cycling and pulsed-flow experiments have been carried out to study the dynamic behavior of the catalysts. The experiments were specifically designed to target the application in hybrid drive systems with intermittent operation of a combustion engine. Time-resolved analysis of the different components of the synthetic exhaust gas by combined FTIR spectroscopy and mass spectrometry allowed a detailed view of the dynamic behavior. Experimental Section Catalysts. The four catalysts (Degussa AG) tested consisted of a ceramic honeycomb carrier with 400 cells/ in.2. The wash-coat loading was 110 g L-1 with the composition (wt %) as denoted in Table 1. Palladiumbased catalysts were produced using a new technology developed by the company, whereas the Pt-based catalysts corresponded to state-of-the-art catalytic systems. The catalysts had a length of 15 cm and a diameter of 2.5 cm. To reduce the volume of the catalyst to 12.73 cm3, the outermost channels were sealed with a catalytically inert ceramic paste. Before the catalytic tests were performed, the samples were conditioned for 5 h at 600 °C in a simulated exhaust with λ ) 1 (see the experimental procedures). Apparatus. Catalytic experiments were carried out using a fully computer-controlled apparatus, which has been described in detail elsewhere.31 The synthetic exhaust gas mixture used for laboratory tests contained CO and H2 at a ratio of 3:1, C3H6 (500 ppm), C3H8 (500 ppm), O2, NO (2000 ppm), CO2 (12%), H2O (10%), and N2 (balance). The flows of the reactant gases were regulated by mass-flow controllers (Brooks 5850) and mixed in special gas-mixing chambers. Water was fed by a diaphragm pump (Lewa Lab M3) and evaporated into the preheated feed stream. The λ-value of the simulated exhaust, which represents the ratio between the available oxygen and the oxygen needed for full conversion to CO2, H2O, and N2, is defined by eq 1:
λ)
2cO2 + 2cCO2 + cH2O + cCO + cNO 2cCO + cH2 + 10cC3H8 + 9cC3H6 + 2cCO2 + cH2O
(1)
where ci designates the corresponding gas concentration of component i at the converter inlet. The numerator corresponds to air-derived species and the denominator to fuel-derived species. The λ-value could be adjusted by altering the flows of O2 and the CO/H2 mixture. For
a stoichiometric feed (λ ) 1), the concentrations of these three components were CO (0.51%), H2 (0.17%), and O2 (0.73%). Via two simultaneously operated solenoid valves additional O2 and CO/H2 streams could be added to the synthetic exhaust gas to facilitate rapid λ-cycling. While changing the λ-value, the total flow rate was held constant by simultaneously adjusting the flow rate of N2. Analysis. The product stream leaving the reactor was cooled to 150 °C and subsequently directed to the analysis system which consisted of an FT-IR spectrometer (Bruker IFS-66) with a heatable gas cell (100 cm3 volume) and a quadrupole mass spectrometer (Balzers GAM 400). NO, NO2, N2O, NH3, CH4, C3H6, C3H8, CO, CO2, and H2O were analyzed by FT-IR spectroscopy and O2 and H2 by mass spectrometry. The analytical system permitted the quantitative analysis with a resolution of up to 15 measurements/s. The mass spectrometer was not used for the time-resolved experiments. With exception of the time-resolved experiments, where only one scan at a 4 cm-1 resolution was recorded per spectrum, in all other experiments 120 scans with a 2 cm-1 resolution were averaged for each analysis point. The spectra were quantified on the basis of characteristic absorption frequencies of the different components and by using the software OPUS 2.0 by Bruker, which is based on principle component regression. The accuracy in the concentration measurements was (5% of the calibration range maximum for all components, as found by measurements with calibration gas mixtures. Experimental Procedures. Experiments were carried out at atmospheric pressure using a gas flow rate of 10.625 L(NTP) min-1, corresponding to a gas hourly space velocity (GHSV) of 50 000 h-1 with regard to the total catalyst volume. Estimating a void fraction of about 0.5 results in GHSV ranging from 145 000 (150 °C) to 265 000 h-1 (500 °C) in the investigated temperature range. Considering the particular operating characteristics of a hybrid engine, special focus was given to the light-off behavior. Thus, the following experimental tests were performed. Light-off experiments were carried out either with steady inlet gas composition or with forced λ-cycling by rising the gas inlet temperature from 100 to 500 °C at a rate of 3 °C min-1. Measurements were taken from 150 to 500 °C. Forced λ-cycling with different amplitudes and frequencies was achieved by periodically changing the stoichiometry of the feed composition according to the following: mode 1, no cycling, constant λ ) 1; mode 2, cycling at a frequency of ν ) 0.67 Hz with an amplitude of 0.02 (i.e., λ ) 1 ( 0.02; mode 3, ν ) 0.67 Hz, λ ) 1 ( 0.05; mode 4, ν ) 1.0 Hz, λ ) 1 ( 0.05; mode 5, where an asymmetric amplitude was used with a lower λ of 0.954 and an upper one of 1.05 and the frequency was ν ) 0.67 Hz. The amplitude was of nearly rectangular shape.31 Time-resolved cycling experiments with an amplitude of λ ) 1 ( 0.05 and different frequencies were carried out at a temperature of 310 °C, which lies in the middle of the warm-up region. The experiments aiming at determining the frequency dependence of the catalytic behavior were conducted at 310 °C with a constant amplitude of λ ) 1 ( 0.05. Sweep tests were performed at 500 °C with steady and cycled feed streams by increasing the time-averaged λ-value from 0.95 to 1.05. The amplitude in the cycled sweep curves was (0.02 (i.e., λ ) x ( 0.02. Note that x is not the same as the time-averaged λ-value.
110 Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999
Figure 1. Effect of the temperature and cycling mode on the lightoff performance of the reduction of NO to N2. The yield of N2 is defined as: YN2 ) XNO - YN2O - YNO2 - YNH3, where the conversion of NO is XNO ) (cNO,in - cNO,out)/cNO,in × 100, the yields of NO2 and NH3 are Yi ) ci/cNO,in × 100 and, for N2O, YN2O ) 2cN2O/cNO,in × 100. (M1) Operating mode 1: λ ) 1. (M2) Cycling mode 2: λ ) 1 ( 0.02; cycling frequency ν ) 0.67 Hz. (M3) Cycling mode 3: λ ) 1 ( 0.05; ν ) 0.67 Hz. Heating rate ) 3 °C min-1. Symbols: (thin line) Pd-Rh/Al2O3 (Pd-Rh); (thick line) Pd-Rh/Al2O3-CeO2 (PdRh-Ce); (- - -) Pt-Rh/Al2O3 (Pt-Rh); (- - -) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce). Denotation of catalysts used in text are given in parentheses.
Pulsed-flow experiments were carried out at 400 °C and 1.7 bar. The exhaust gas was pulsed with a flow rate of 10.625 L(NTP) min-1 through the reactor for 3 s, followed by a period of 17 s, with no gas flowing through the converter. The λ-value was either kept constant or cycled symmetrically or asymmetrically during the 3 s of the pulse. The performance of the catalysts was examined for the different intermittent operating modes A-D. For modes A and C a stoichiometric exhaust with the constant λ ) 1 was used without (A), and with (C) a preceding air plug. For operating modes B and D the λ-value was cycled during the pulse with a frequency of 1 Hz and an amplitude of λ ) 1 ( 0.05, without (B) and with (D) a preceding air plug. Cycling started with a rich exhaust gas and was asymmetric (0.6 s rich/0.4 s lean) in the case of mode D. In experiments with an air plug, an air pulse with a duration of 0.2 s and a flow rate of 3.187 L(NTP) min-1 preceded the exhaust pulse, simulating air which is transferred into the cylinders. Results Light-Off Experiments. A. NOx Conversion. The light-off performance for NOx conversion to N2 of the different catalysts is presented in Figure 1 for steady and cycled operation modes, employing a temperature ramp of 3 °C min-1. For a steady stoichiometric feed (M1), the temperature to achieve 50% N2 yield, T50, was substantially lower for the platinum-based catalysts PtRh and Pt-Rh-Ce. The ceria-promoted catalyst PdRh-Ce possessed the highest T50 of about 350 °C. All catalysts showed a similar efficiency for N2 formation at higher temperatures. Cycling differently affected the efficiency of the catalysts for N2 formation. A summary
of the changes of the T50 values for the cycling modes 2-5 compared to those of the non-cycled feed (M1) is given in Table 2. In general, the light-off temperatures for the Pd-based catalysts were substantially lowered, whereas T50 of Pt-Rh and Pt-Rh-Ce showed only a comparably small decrease upon cycling. For higher amplitudes a marked effect was observed for Pd-RhCe, where T50 dropped by about 110 °C. Cycling with mode 3 resulted in comparable NOx conversion to N2 for all catalysts (Figure 1). Interestingly, the addition of ceria positively influenced the activity of the Pd-based catalysts, whereas T50 increased for the corresponding Pt-based catalysts. Regarding the influence of higher cycling frequencies, 50% N2 yield was not reached in the measured temperature range for Pd-Rh (M4, Table 2). Note that with an increasing temperature the catalysts show a complex behavior with regard to N2 formation which is a consequence of the formation of undesired byproducts as described next. B. Formation of Byproducts. The temperaturedependent production of N2O and NH3 for the four catalysts under operating modes 1-3 is depicted in Figures 2 and 3. In the lower temperature range (T < 350 °C) N2O is preferably formed, and in the upper part of the warm-up phase NH3 appeared as the predominant byproduct. Under steady feed conditions, N2O yields up to 60% were observed for catalysts containing Pt, whereas less than 40% N2O was found for the Pdbased samples. When cycling was applied, N2O production decreased for the platinum catalysts. Moreover, the temperature range for N2O formation became narrower. Addition of ceria to the catalyst formulation significantly reduced the formation of nitrous oxide for all operation modes. Compared to N2O, only small amounts of ammonia were produced with the non-cycled feed (M1), as illustrated by Figure 3. Upon cycling, the onset of ammonia formation was shifted to lower temperatures and substantially higher yields were obtained, with the effect being more pronounced for higher cycling amplitudes. Applying mode 3 the fully formulated exhaust catalyst Pd-Rh-Ce produced NH3. Note that ammonia formation (Figure 3) supplemented nitrogen formation (Figure 1) to give approximately 100% yield in nitrogencontaining species for higher temperatures, thus explaining at least partly the complex behavior observed with mode 3. C. Efficiency for N2 Formation. Regardless of whether NOx content in the exhaust is emitted as NO, NO2, N2O, or NH3, all gases have a detrimental impact on the environment. Note that legislation currently only sets limitations for NO and NO2. NOx and NH3 are not only toxic but are also involved in the formation of ground ozone, which is a serious contamination problem during the summer. Nitrous oxide contributes to the greenhouse effect and is known as an “ozone killer”. Consequently, the goal has to be to achieve maximum overall N2 yields in the warm-up phase. In view of this point, the N2 yield has been integrated in the temperature range 150-500 °C. Division of these values by 350 °C gives the average N2 yield attained during warmup. As-derived average N2 yields and corresponding CO and C3H8 conversions are summarized in Table 3 for all catalysts and operating modes. Under steady feed conditions, the Pt-based catalysts showed the best performance during warm-up with average N2 yields in the range of 50-57%. The addition of ceria had a beneficial effect on Pt-Rh, but a negative impact on
Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 111 Table 2. Effect of Cycling on Temperatures To Achieve 50% Conversion/Yield, ∆T50 (°C), for Cycling Modes M2-M5 Compared to Mode M1 (No Cycling) (Negative Values Indicate That T50 Increased upon Cycling) ∆T50, CO conversion
∆T50, C3H8 conversion
∆T50, N2 yield
catalyst
M2a
M3
M4
M5
M2
M3
M4
M5
M2
M3
M4
M5
Pd-Rh Pd-Rh-Ce Pt-Rh Pt-Rh-Ce
25 25 95 60
25 -5 75 -55
0 65 95 85
10 10 90 75
10 25 10 10
15 75 20 35
-80 30 -65 -5
40 65 25 -30
45 35 30 30
85 110 40 10
b 115 20 5
80 100 50 30
a
Operating modes M1-M5 are described in the Experimental Section. b N2 yield never reaches 50% between 150 and 500 °C.
Figure 2. Influence of the temperature and cycling mode on the yield of N2O. (M1) Operating mode 1: λ ) 1. (M2) Cycling mode 2: λ ) 1 ( 0.02; cycling frequency ν ) 0.67 Hz. (M3) Cycling mode 3: λ ) 1 ( 0.05; ν ) 0.67 Hz. Heating rate ) 3 °C min-1. Symbols: (thin line) Pd-Rh/Al2O3 (Pd-Rh); (thick line) Pd-Rh/ Al2O3-CeO2 (Pd-Rh-Ce); (- - -) Pt-Rh/Al2O3 (Pt-Rh); (- - -) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce).
Figure 3. Influence of the temperature and cycling mode on the yield of NH3. (M1) Operating mode 1: λ ) 1. (M2) Cycling mode 2: λ ) 1 ( 0.02; cycling frequency ν ) 0.67 Hz. (M3) Cycling mode 3: λ ) 1 ( 0.05; ν ) 0.67 Hz. Heating rate ) 3 °C min-1. Symbols: (thin line) Pd-Rh/Al2O3 (Pd-Rh); (thick line) Pd-Rh/ Al2O3-CeO2 (Pd-Rh-Ce); (- - -) Pt-Rh/Al2O3 (Pt-Rh); (- - -) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce).
Pd-Rh. In contrast, ceria addition to Pd-Rh had a positive influence on N2 yield for cycled operating modes, resulting in the highest average N2 yield for the catalyst Pd-Rh-Ce and mode 3. For the standard catalyst Pt-Rh-Ce higher cycling amplitudes reduced the performance. The last column of Table 3, denoted with “m”, has been calculated by using for each temperature the best mode with respect to the N2 yield. In practice, this would correspond to changing the cycling mode during the warm-up phase to achieve the highest possible N2 yield. This strategy allowed us to reach a strongly enhanced performance for all catalysts as compared to that of mode 1. The addition of ceria had a positive effect on N2 yield for both Pt-Rh and Pd-Rh. D. CO Conversion. The light-off performance for CO conversion is shown in Figure 4 for the different catalysts and cycling modes. For all catalysts 50% CO conversion was obtained between 185 and 235 °C and >90% conversion in the range 250-280 °C with mode 1. At higher temperatures, CO conversion dropped similarly as compared to that observed for the N2 yield. The ranking of the T50 values is Pt-Rh-Ce < Pt-Rh < Pd-Rh-Ce < Pd-Rh, emphasizing that the presence of ceria lowers T50. As shown in Table 3, similar average CO conversions during warm-up were achieved with catalysts Pd-Rh-Ce, Pt-Rh, and Pt-Rh-Ce and mode 1. Table 2 lists the change of the T50 values for CO conversion when cycling was applied. Generally, cycling
had a positive influence on CO conversion with the effect being more distinct for the platinum-based catalysts. Calculated average CO conversions for the temperature range 150-500 °C (Table 3) reveal that catalyst PdRh-Ce was quite insensitive to cycling. For the platinum catalysts cycling improved CO conversion performance, except for the catalyst Pt-Rh-Ce and mode 3. Under cycling conditions, the Pd-based catalyst showed a significantly poorer performance than the platinum catalysts. The drop in CO conversion for temperatures exceeding 300 °C was reinforced by cycling for the catalysts without ceria. The effect was more pronounced for higher amplitudes and frequencies (modes 3 and 4). E. Hydrocarbon Conversion. Propane has been chosen as the low-reactivity HC component to compare the light-off performance of the catalysts. Similar to that observed for N2 formation, the T50 value for C3H8 conversion of the Pt-based catalysts was about 40-60 °C lower than that for the Pd-based samples under steady feed conditions (Figure 5). Ceria addition positively influenced the performance of Pt-Rh but had a negative impact on the performance of Pd-Rh. This behavior is also reflected by the calculated average propane conversion shown in Table 3, which upon addition of ceria was lower for PdRh and higher for PtRh. Depending on the mode applied, cycling had a slightly positive or even negative effect on propane conversion for the catalysts Pd-Rh, Pt-Rh, and Pt-
112 Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 Table 3. Light-Off Performance in the Temperature Range 150-500 °C for the Different Operating Modes, Expressed as Average Conversions and Yields, Respectively CO conversion (%)
C3H8 conversion (%)
N2 yield (%)
catalyst
M1a
M2
M3
M4
M5
M1
M2
M3
M4
M5
M1
M2
M3
M4
M5
mb
Pd-Rh Pd-Rh-Ce Pt-Rh Pt-Rh-Ce
75 84 85 85
83 81 96 92
75 80 93 79
69 84 91 99
63 79 93 94
54 51 63 68
55 56 64 67
61 66 65 70
38 56 48 63
55 65 63 58
40 33 50 57
41 47 58 63
47 58 54 42
24 42 47 45
26 43 62 49
57 64 66 69
a Operating modes M1-M5 are described in the Experimental Section. b Column m denotes the “mixed mode” light-off procedure (see text).
Figure 4. Dependence of CO conversion on the temperature and cycling mode. (M1) Operating mode 1: λ ) 1. (M2) Cycling mode 2: λ ) 1 ( 0.02; cycling frequency ν ) 0.67 Hz. (M3) Cycling mode 3: λ ) 1 ( 0.05; ν ) 0.67 Hz. Heating rate ) 3 °C min-1. Symbols: (thin line) Pd-Rh/Al2O3 (Pd-Rh); (thick line) Pd-Rh/ Al2O3-CeO2 (Pd-Rh-Ce); (- - -) Pt-Rh/Al2O3 (Pt-Rh); (- - -) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce).
Figure 5. Dependence of propane conversion on the temperature and cycling mode. (M1) Operating mode 1: λ ) 1. (M2) Cycling mode 2: λ ) 1 ( 0.02; cycling frequency ν ) 0.67 Hz. (M3) Cycling mode 3: λ ) 1 ( 0.05; ν ) 0.67 Hz. Heating rate ) 3 °C min-1. Symbols: (thin line) Pd-Rh/Al2O3 (Pd-Rh); (thick line) Pd-Rh/ Al2O3-CeO2 (Pd-Rh-Ce); (- - -) Pt-Rh/Al2O3 (Pt-Rh); (- - -) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce).
Rh-Ce. In contrast, cycling showed a clearly beneficial effect for Pd-Rh-Ce, the T50 of which decreased by up to 75 °C (Table 2) and which achieves a similar performance in propane oxidation as Pt-Rh-Ce (Table 3). Influence of the λ-Cycling Frequency. Figure 6 depicts the changes in concentrations of the exhaust gas components with time upon λ-cycling with an amplitude of λ ) 1 ( 0.05 for Pd-Rh-Ce and Pt-Rh-Ce. The effect of forced λ-cycling with different frequencies on the concentrations of C3H8, CO, NO, and NH3 is shown for a temperature of 310 °C. The black arrow indicates 1 lean half-cycle. At a frequency of 0.3 Hz, qualitatively similar cyclic concentration-time profiles were observed for all catalysts. The concentration maxima of both CO and propane appeared at the end of the rich half-cycle. For C3H8 a second maximum, coinciding with the NO peak at the end of the lean half-cycle, was observed. Substantial amounts of ammonia, with the maximum appearing in the middle of the lean half-cycle, were produced with both catalysts. Increasing the cycling frequency resulted in a decrease of the concentration maxima for all exhaust gas components. For 1.0 Hz, concentrations were very low and no cyclic effects were discernible for Pt-Rh-Ce, whereas oscillations with intervals of 1 s were still observable for the corresponding palladium-based catalyst.
In Figure 7 the influence of different cycling frequencies on the conversions of CO and HC (C3H8 and C3H6) and on the N2 yields is shown for a constant cycling amplitude of λ ) 1 ( 0.05 and a temperature of 310 °C. To magnify the clarity of the effects, presentation with the period length in seconds (1/cycling frequency) as the x-axis has been chosen. On the left side of the main chart, conversions measured for the steady operation mode (λ ) 1) are shown for comparison. For a steady feed (mode 1), the platinum-based catalysts showed slightly higher HC conversions and markedly higher N2 yields than the palladium-containing catalysts. In general, the addition of ceria had a slightly positive influence. Upon being cycled with increasing period lengths (lower frequency), the performance of the catalysts is characterized by maxima in HC and CO conversion and N2 yield at a cycling frequency of about 1 Hz. For the Pd-based catalysts N2 yields increased markedly if λ-cycling was applied, whereas HC conversion was only slightly higher and CO conversion was lowered. In the case of the platinum-containing catalysts cycling had a detrimental effect on HC and CO conversion but resulted in slightly increased N2 yields for period lengths between 0.67 and 2 s. In the case of palladium catalysts, ceria promotion clearly enhanced CO and HC conversion and, for period lengths up to 1 s, also N2 yield. Except for the lowest period length, the addition of ceria to Pt-
Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 113
Figure 6. Effect of the cycling frequency on the concentrations of the most significant exhaust gas components with time at 310 °C and λ ) 1 ( 0.05 for Pd-Rh/Al2O3-CeO2 (Pd-Rh-Ce, left) and Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce, right). The arrows indicate one lean half-cycle. Symbols: (- - -) CO; (thick line) C3H8; (thin line) NO, (- - -) NH3.
Figure 7. Performance of the different catalysts under operation with cycled feed stoichiometry. Conversions of CO and C3H8 and yield of N2 as a function of the different cycling frequencies at 310 °C for a cycling amplitude of λ ) 1 ( 0.05. Small charts on the left: Conversions and yields for a stoichiometric feed (λ ) 1). Symbols: (b) Pd-Rh/Al2O3 (Pd-Rh); (O) Pd-Rh/Al2O3-CeO2 (Pd-Rh-Ce); (9) Pt-Rh/Al2O3 (Pt-Rh); (0) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce).
Rh only slightly influenced HC and CO conversion but decreased N2 yield. At typical cycling frequencies of 1 Hz, as induced by the closed-loop control of an automobile, and at the temperature of 310 °C, which lies in the middle of the warm-up region, Pd-Rh-Ce showed catalytic performance similar to that of the standard TWC Pt-Rh-Ce. Sweep Curves. Sweep tests with and without λ-cycling were performed to study the sensitivity of the
Figure 8. Sweep curve for Pd-Rh/Al2O3-CeO2 (Pd-Rh-Ce, left) and Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce, right) at 500 °C and constant feed (no cycling). Symbols: Yields of (9) N2 and (0) NH3; (4) concentration of H2, conversions of (b) CO, (O) HC (total of C3H6 and C3H8) and (3) NOx (total of NO, NO2 and N2O).
catalysts to excursions into the lean or rich region. Figure 8 compares the sweep curves for catalysts PdRh-Ce and Pt-Rh-Ce measured at 500 °C with constant feed (no cycling). NOx conversion was almost complete on the rich side but decreased rapidly to zero for λ > 1 (lean side). With regard to the reaction products emerging from NOx conversion, N2 appeared as the main product around the stoichiometric point, whereas for decreasing λ-values, both catalysts produced increasing amounts of ammonia, reaching yields around 80% for λ ) 0.95. Note that the decline of the N2 yield and production of ammonia on the rich side was less pronounced for Pd-Rh-Ce compared to that for PtRh-Ce. With both catalysts CO and HC conversions higher than 90% were measured in the lean region, whereas for rich exhausts (λ < 1) CO conversion was significantly lower. HC conversion remained high for Pd-Rh-Ce and slightly decreased for Pt-Rh-Ce. To quantify the sensibility of the catalytic behavior to deviations from the stoichiometric feed composition, the different conversion and yield values have been integrated in the range λ ) 0.975-1.025. Division by 0.05 gave the average conversions and yields listed in Table 4 for steady and cycled feed conditions. Conversion of HC was complete for the measured λ-range, independently of whether cycling was applied or not. With regard to CO conversion, the presence of ceria had a positive impact on activity, with the effect being more pronounced for Pd catalysts with cycling and for Pt catalysts without cycling. Concerning the N2 yield, which considerably defines the width of the λ-window, ceria had only a beneficial effect for the Pd-containing catalyst. However, maximum N2 yields were higher than 90% for all catalysts, as shown in the last column of Table 4, which denotes the maximum yield which is achieved at λ ≈ 1. Experiments with Pulsed Flow. In view of the application of the catalysts in the hybrid vehicle with intermittent operation of the combustion engine, simulation of the pulsed-flow operation was performed at 400 °C, which is exceptionally low for TWC applications. The performance of the catalysts was compared for the different operating modes A-D described in the Experi-
114 Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 Table 4. Average Conversions and Yields Obtained in Sweep Experiments at 500 °C in the λ-Range of 0.975-1.025 CO conversion (%) catalyst
steady feed
Pd-Rh Pd-Rh-Ce Pt-Rh Pt-Rh-Ce
82 84 79 94
a
with
cyclinga 79 92 88 90
HC conversion (%) steady feed 99 99 96 99
with
cyclinga 99 99 99 97
N2 yield (%) steady feed
with cyclinga
max.
47 54 49 47
55 57 60 50
98 99 98 98
ν ) 0.67 Hz; λ ) 1 ( 0.02.
by appropriate operating modes being chosen, high C3H8 conversions and N2 yields as well as low CO concentrations can be achieved with the ceria-containing catalysts. In contrast, the air plug had a detrimental effect on the catalytic behavior of the non-promoted catalysts. Discussion
Figure 9. Influence of air plug and its compensation on average C3H8 conversion, N2 yield, and CO concentration during an exhaust pulse of 3 s at 400 °C for intermittent operating modes. (A) Exhaust pulse with λ ) 1. (B) Symmetric λ-cycling with λ ) 1 ( 0.05, 1 Hz. (C) Air plug followed by exhaust pulse with λ ) 1. (D) Air plug followed by asymmetric λ-cycling with λ ) 1 ( 0.05, 1 Hz, periods 0.6 s rich/0.4 s lean.
mental Section. To compare the effect of the intermittent operating modes on the catalysts performance, the concentrations of the exhaust components were integrated over the period of the pulse and divided by 3 s. As the inlet concentration of CO was not constant for the experiments with cycled feeds, concentrations instead of conversions are given. Figure 9 depicts the average C3H8 conversions, N2 yields, and CO concentrations during an exhaust pulse of 3 s at 400 °C. Independent of the operating mode applied, the presence of ceria had a strongly positive effect on C3H8 conversion and CO concentration, with the effect being more pronounced for the Pd catalyst. Moreover, N2 yields were higher for Pd-Rh-Ce than for Pd-Rh when cycling was applied. Comparing mode A with mode B indicates that cycling positively influenced the behavior of catalyst Pd-Rh-Ce, whereas the effect of cycling was detrimental for the other catalysts. Using a preceding air plug followed by an exhaust pulse with λ ) 1 (mode C) resulted in significantly lower N2 yields for all catalysts and in slightly increased HC conversions for the Pt-based samples. The combination of an air plug with a cycled exhaust pulse (mode D) resulted in the best overall performance of the ceria-promoted catalysts Pd-Rh-Ce and Pt-Rh-Ce. The result indicates that
Light-Off Behavior. An objective of the present study was to examine how the newly developed Pd/Rh TWCs compare to conventional Pt/Rh catalysts with regard to the conversion of the main target species CO, HC, and NOx during warm-up with constant feed (λ ) 1) and with forced λ-cycling. Under steady feed conditions the platinum-based catalysts show better lightoff performance for C3H8 than for Pd, whereas with λ-cycling at higher amplitudes comparable behavior is observed for the fully promoted Pd and Pt catalysts. In the higher temperature region and using cycling modes with higher amplitudes the Pd catalysts show better performance than the Pt catalysts. When oxygen is present in the feed stream, oxidation is the most important HC-removing process, whereas steam reforming becomes significant after oxygen has been consumed.32 Ceria shows beneficial effects in modes with λ-cycling, especially at higher amplitudes and frequencies. The platinum-based catalysts also show higher activity for CO oxidation, except for cycling mode 3, where Pd-Rh-Ce reaches the same average CO conversion during warm-up as the standard three-way catalyst Pt-Rh-Ce. The drop in CO conversion above 300 °C can be ascribed to higher steam-reforming activity of the catalysts, thus producing additional CO. Ceria promotion generally enhanced the efficiency of the palladium catalyst for CO oxidation (Table 3), whereas no clear effect is observable for the Pt catalysts. An improvement of CO oxidation activity by ceria was occasionally reported for platinum metal-based catalysts.28,31,33,34 Kim29 attributed the improved CO removal upon ceria promotion of TWCs to the enhanced activity for the water-gas shift reaction and, in part, to the additional oxygen storage it provides to the TWC. With regard to NOx conversion to N2, the palladiumbased catalysts are substantially less active without cycling than the Pt catalysts and ceria promotion furthermore decreases the effectiveness of the Pd catalyst. Cycling shifts the light-off curves for NOx conversion to lower temperatures, resulting in similar catalytic performances, expressed as the average N2 yield attained during warm-up (Table 3), of Pd-Rh-Ce compared to the platinum-based catalysts for cycling modes 3 and m. For temperatures exceeding 300 °C, all catalysts show a more or less pronounced drop of the N2 yield and a concomitant increase in NH3 formation (Figure 3), especially for higher cycling amplitudes. This observation coincides with the achievement of full HC conversion and a drop in CO conversion, which is more
Ind. Eng. Chem. Res., Vol. 38, No. 1, 1999 115
Figure 10. Concentration of H2 in the outlet exhaust gas during light-off. Modes are specified in the caption of Figure 1. (M1) Operating mode 1: λ ) 1. (M2) Cycling mode 2: λ ) 1 ( 0.02; cycling frequency ν ) 0.67 Hz. (M3) Cycling mode 3: λ ) 1 ( 0.05 ν ) 0.67 Hz. Heating rate ) 3 °C min-1. Symbols: (thin line) PdRh/Al2O3 (Pd-Rh); (thick line) Pd-Rh/Al2O3-CeO2 (Pd-Rh-Ce); (- - -) Pt-Rh/Al2O3 (Pt-Rh); (- - -) Pt-Rh/Al2O3-CeO2 (Pt-Rh-Ce).
pronounced for the Pd catalysts. Moreover, a significant increase in the H2 concentration emerges (Figure 10), which shows a similar trend for all catalysts at higher temperatures as observed for the NH3 yield. In auto exhaust, NH3 can be formed by the direct reaction of NO with H2, via the hydrolysis of an isocyanate intermediate, or by the direct reaction of NO with hydrogen produced via a water-gas shift reaction (WGSR).35 In a study of the reduction of NO with a mixture of CO and H2 in the presence of H2O and O2 over supported Rh and Pt catalysts, Voorhoeve et al.36 reported NH3 formation via an isocyanate intermediate. By investigating the role of steam reforming and the water-gas shift reaction in the behavior of three-way catalysts, Whittington et al.32 found that both reactions become important only at temperatures higher than those needed to facilitate oxidation (i.e., only when the monolithic catalyst bed is depleted of oxygen). According to these findings, we can assume that only small quantities of CO and H2 are available in the lower temperature region, as these components are oxidized first, whereas with an increasing temperature steam reforming becomes more important and hydrocarbons are converted to CO and H2. CO further reacts to CO2 and H2 via the WGSR. Consequently, NH3 formation becomes evident simultaneously with the appearance of a full HC conversion and the decrease of CO conversion and N2 yield, respectively. As has been described in the literature,37 the addition of metal oxides can have an influence on the formation of undesired byproducts. The formation of N2O in the lower part of the warm-up phase (