alumina catalysts in automobile engine

Performance of platinum/alumina catalysts in automobile engine exhaust with ... Catalytic Converter for Removing CO and HC Emissions from a Two-Stroke...
0 downloads 0 Views 977KB Size
I n d . Eng. Chem. Res. 1988,27, 30-36

30

Horio, M.; Hayashi, H.; Morishita, K. Proceedings of the Eighth International Conference on Fluidized Bed Combustion, Houston, 1985, p 655. Kunii, D.; Levenspiel, 0. Fluidization Engineering; Wiley: New Yotk, 1969. Roberts, A. G.; Stanton, J. E.; Wilkins, D. M.; Beacham, B.; Hoy, H. R. Inst. Fuel Symp. Ser. 1975, 1. Toomey, R. D.; Johnstone, H. F. Chem. Eng. Prog. Symp. Ser. 1953,

49, 51. Van Heerden, G.; Nobel, P.; Van Krevelen, D. W. I n d . Eng. Chem. 1953, 45, 1237. Wen, C. Y., Leva, M. AIChE J . 1956, 2, 482.

Rzceiued f o r review January 5, 1987 Revised manuscript received September 9, 1987 Accepted September 25, 1987

Performance of Pt/A1203Catalysts in Automobile Engine Exhaust with Oscillatory Air/Fuel Ratio Byong K. Cho Physical Chemistry Department, General Motors Research Laboratories, Warrera, Michigan 48090-9055

Performance of Pt/A120, catalysts for the removal of CO, NO, and hydrocarbons in engine exhaust was investigated using a fixed-bed reactor connected to a dynamometer-mounted engine. The effect of air/fuel ratio oscillation on the catalyst performance was examined over a wide temperature range. Results showed good activity of fresh Pt/A1203catalysts for the oxidation of CO and hydrocarbons as well as for the reduction of NO, when the feed composition oscillated symmetrically around a time-average stoichiometric point. Below the reaction lightoff temperatures, a feed composition oscillation at 0.1 Hz around a time-average stoichiometric point yielded higher conversions for all three pollutants than steady feed operation, whereas this trend reverses above the reaction lightoff temperatures. The presence of excess oxygen above the stoichiometric amount in engine exhaust lowered considerably the reaction lightoff temperatures for the oxidation of CO and hydrocarbons but severely suppressed the NO reduction activity of Pt/A120, catalysts. The performance and reaction selectivity of the catalysts under engine exhaust conditions were further discussed in comparison with those under laboratory conditions in light of the chromatographic and antichromatographic effect occurring in the catalyst pellet. The essential requirement for an effective three-way catalyst in automobile catalytic converters is high conversion of CO, NO, and hydrocarbons. In order to satisfy this requirement, the three-way catalysts contain noble metals such as Pt, Pd, and Rh as active components; Pt and Pd are used for the oxidation of carbon monoxide and hydrocarbons, while Rh is used for its good activity to reduce nitric oxide as well as to oxidize carbon monoxide. Noble metals contribute significantly to the cost of three-way catalysts. Rh is currently being used at loadings above the mine ratio in most commercial three-way catalysts (Taylor, 1984). This has prompted extensive efforts to develop Rh-free three-way catalysts with adequate NO reduction activity (Adams and Gandhi, 1983; Yokota et al., 1985; Muraki et al., 1986). A recent laboratory study by Cho and Stock (1986) has shown that the initial NO decomposition activity of Pt/ A1203 catalysts is comparable to that of Rh/A1203catalysts, although both Pt/A1203 and Rh/A120, catalysts were found to be deactivated due to the gradual accumulation of surface oxygen produced from NO decomposition. These interesting similarities between Pt/A1,03 and Rh/A1,03 observed in laboratory feedstreams prompted us to examine the performance of Pt/A1203 catalysts in an exhaust environment. A detailed understanding of the dynamic behavior of noble metal catalysts under transient feedstream conditions is of practical importance in automotive emission control because three-way catalytic converters operate under the conditions of periodically fluctuating feed composition. There are numerous reports in the reaction engineering literature that a catalyst’s activity or selectivity can be enhanced under certain conditions by forced composition cycling (e.g., Billimoria and Bailey, 1978; Cutlip, 1979;Abdul-Kareem et al., 1980; Fiolitakis et al., 1983; Cho and West, 1986; Silveston et al., 1986). During warmed-up 0888-5885/S8~2621-0030$01.50f 0

operation of three-way catalytic converters, steady operation at the stoichiometrically balanced feed composition has been preferred to cyclic operation, and thus a high frequency and/or small amplitude of composition cycling is desirable (Schlatter et al., 1983; Taylor, 1984). Similar observations were reported by Cho and West (1986) for warmed-up CO oxidation activity over Pt/A1203. However, below the reaction lightoff temperature of CO oxidation (Le., during warm-up), Cho and West (1986) observed a significant enhancement of catalyst activity during cyclic operation compared with steady operation. This enhancement of catalyst activity during cyclic operation of CO oxidation over Pt/Al,03 can be explained in terms of the chromatographic and antichromatographic concept introduced by Cho (1983). For commercial three-way catalysts containing Pt, Pd, and Rh, engine dynamometer experiments (Herz, 1981; Herz and Kiela, 1983) and laboratory reactor experiments (Taylor and Sinkevitch, 1983) have been reported under cyclic operating conditions. Under steady operating conditions, Jones et al. (1971) have shown that a Pt catalyst can be effective for the simultaneous removal of CO, NO, and hydrocarbons in the engine exhaust when the A/F ratio is kept close to stoichiometry. Summers and Monroe (1981) observed a considerable deterioration of three-way performance of the Pt catalysts upon aging in the engine exhaust. In this work we undertake engine-dynamometer experiments to investigate the catalytic activity of Pt/A1203 catalysts for the removal of CO, NO, and hydrocarbons in engine exhaust. Specifically, the following three questions are addressed in this study regarding the practical applicability of the above-mentioned laboratory findings reported previously (Cho and West, 1986; Cho and Stock, 1986). The first is whether improved catalyst performance for CO oxidation could be achieved under engine exhaust 0 1988 American Chemical Society

Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 31 Table I. Catalysts"

Pt loading, wt '70 impregnation depth, fim dispersion, 70 method of impregnation

immeenation surface uniform 0.1 0.1 40 1600 50 100 nonaqueous aqueous

"Support: bead radius = 1600 fim; bulk density = 0.4731 g/cm3;

BET surface area = 113 m2/g.

Figure 1. Schematic diagram of the experimental system (1,engine; 2, dynamometer; 3, packed-bed catalytic converter; 4, packed-bed tubular reactor; 5, blank alumina bed; 6, furnace; 7 , reservoir; 8, fuel injectors; 9, flow restrictor; 10, thermocouple (inlet temperature); 11, thermocouple (reactor temperature); 12, additional air injector; 13, analyzer train; 14, exhaust line).

conditions during cyclic operation compared with steady operation. The second is whether improved catalyst performance can similarly be achieved for NO reduction and hydrocarbon oxidation. The third is whether fresh Pt/A1,03 catalysts are active for NO reduction under engine exhaust conditions.

Description of Experiments Catalysts. Two Pt/Al2O3catalysts with different impregnation depths were used in this study: a surface-impregnated catalyst and a uniformly impregnated catalyst. These are the same catalysts as those used in the previous laboratory experiments for CO oxidation (Cho and West, 1986), since one of the purposes of this work is to compare the major findings of the earlier laboratory study with results obtained using engine exhaust. Some of the important characteristics of the catalysts are listed in Table I; a detailed description of the catalyst preparation procedure and postimpregnation heat treatment conditions can be found in the above reference. Experimental Procedure. A schematic diagram of the experimental system is shown in Figure 1. The gasoline engine (5.7-L V8) was operated on Indolene Clear fuel at 1800 rpm with an intake manifold pressure of 50 kPa (abs). Most of the engine exhaust flow was directed to the inlet of a Type-160 GM catalytic converter, while a small portion of the exhaust was diverted, by adjusting a flow restrictor, to the fixed-bed reactor which was used in this study. The reactor was made of 25-mm-0.d. stainless steel tube packed with 30 cm3 of Pt/Al2O3 catalyst and was preceded by a 25-mm-0.d. stainless steel tube packed with blank spherical y-alumina pellets of 3.2-mm diameter. This blank alumina bed was placed in an electrical furnace and served as either a precooler or a preheater depending on the desired inlet temperature to the reactor. The temperature of the engine exhaust was approximately

565-575 "C measured just before the converter, and the space velocity of the exhaust through the reactor was about 25000 h-l (STP). The analyses of CO, NO, 02,and hydrocarbons at both the reactor inlet and outlet were done on a dry basis (i.e., after the removal of HzO from the exhaust) by a dispersive IR detector, chemiluminescence, an electrochemical sensor, and flame ionization, respectively. These analyzers were able to measure only the time-average values of the concentrations under stabilized engine operating conditions due to their slow response times. The average air/fuel ratio (A/F ratio) unc'er constant operating conditions was computed from the measured time-average concentrations of CO, NO, 02,and hydrocarbons (expressed in volume percent) using A/F = 14.63/[1 + 0.02545(Cco + C H + ~ 9 C ~ -c 2C0, - CNO)] (1) assuming that the fuel has a H/C ratio of 1.91 and the hydrocarbons in the exhaust are present as C3H6(Monroe, 1986). At the stoichiometric point, the A/F ratio is 14.63 with this fuel. The concentration of hydrogen in the exhaust was assumed to be at one-third that of CO (Taylor, 1984). It is worth noting that the engine exhaust contains approximately 20 ppm SO2 and about 10% each of H,O and COz (Taylor, 1984). Two fuel injectors were used to cycle the composition of the engine exhaust; one maintained a steady base-line composition, while the other was turned on and off by an electronic timing circuit to generate a richer composition pulsation to A/F than the base-line composition. For stoichiometric cycling, which is defined as composition cycling about a time-average stoichiometric A/F, the time-average A/F was kept at 14.63 with a cycling amplitude of 0.5 A/F. For cyclic operation about a timeaverage lean A/F, the time-average A/F was increased from 14.63 to 15.13 by introducing additional air to the engine exhaust just before the blank alumina bed while keeping the A/F cycling amplitude at 0.5 A/F. Two different cycling frequencies (0.1 and 1 Hz) were used for all catalyst,s studied; these two frequencies were chosen to represent slow and fast feedstream transients. Without real-time measurements of exhaust composition at the reactor inlet, we do not know exactly the extent of the gas-phase mixing as a function of cycling frequency. But from the previous studies (Sell et al., 1980; Schlatter et al., 1983), we know that at 1 Hz the amplitude of the concentration oscillation may not quite reach either end point due to attenuation by mixing. A t 0.1 Hz, however, the amplitude should reach both end points.

Catalyst Performance Effect of Feed Composition Cycling. The activity of the Pt/A1203catalyst with a narrow surface Pt band (40 pm) was measured under steady and cyclic feed conditions for the removal of CO, NO, and hydrocarbons from the engine exhaust. The cycling scheme involved a symmetric cycling around the stoichiometric point with an amplitude of 0.5 A/F. Parts a, b, and c of Figure 2 compare the conversion performance of this catalyst under steady feed conditions with that under cyclic feed conditions for CO, hydrocarbons, and NO, respectively. In comparing the catalyst performance, we used the conversion data plotted as a function of the reactor inlet temperature rather than the reactor bed temperature, because the temperature within the bed is dependent upon the activity of the catalyst sample being tested. For convenience in our data analysis, we define reaction lightoff temperature as the

32 Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 a. CO

c. NO

b. Hydrocarbons

100 r

7

.*

000

E

20

,

0 360

400

440

480500

Inlet Temperature (OC)

360

400

440

460500

Inlet Temperature (W)

360

400

440

460500

Inlet Temperature (OC)

Figure 2. Comparison of cycled performance with steady performance of the surface-impregnated Pt/Al2O3 catalyst for CO, hydrocarbons, and NO conversions: ( 0 )steady operation, (A) cyclic opcyclic operation a t 0.1 Hz. eration a t 1 Hz, (0)

temperature where the catalyst activity shows a maximum gradient during steady operation (i.e., the inflection point on the conversion vs temperature curve during steady operation). It is noted in Figure 2 that there is little difference in catalyst performance between the steady operation and the fast (1Hz) cyclic operation. However, a large difference in catalyst performance is observed between the steady operation and the slow (0.1-Hz) cyclic operation; below the reaction lightoff temperature, the cyclic operation gives better catalyst performance than the steady operation for all three pollutants, while the reverse is true above the reaction lightoff temperature. The relatively flat conversion profiles in the high temperature regime indicate that the catalyst activity is limited by mass transfer to the catalytic surface. It is interesting to note that the CO conversion performance during the slow cyclic operation (0.1 Hz) reaches this mass-transfer-controlled regime a t a much lower inlet temperature than that during either the fast cyclic operation (1.0 Hz)or the steady operation. On the other hand, the fast cyclic operation for the conversion of NO and hydrocarbons does not reach the mass-transfer-controlled regime until the inlet temperature becomes as high as 500 “C. This interesting aspect will be discussed in detail later in connection with the warm-up behavior of the catalysts. Effect of Impregnation Depth. The activity of the uniformly impregnated catalysts was measured under the same operating conditions that were used for the surface-impregnated catalysts, so that the effect of impregnation depths on the steady and cyclic performance of the catalysts can be isolated. For a direct comparison of the experimental results obtained using different batches of catalysts, it was necessary to correct for small variations in the catalyst amount and feed flow rate from one run to another (Cho and West, 1986). In this work the catalyst activity was normalized by matching the steady feed CO conversion performance of all catalysts at 500 “C, beyond which the catalyst activity was mainly controlled by the external mass-transfer resistance for both catalysts. In fact, matching of the CO conversion at 500 “C was obtained by reducing the CO conversion level for the surface-impregnated catalysts by approximately 5 %. Accordingly, the conversion levels of CO, NO, and hydrocarbons for the surface-impregnated catalysts over the entire temperature range studied were reduced by 5%. Direct comparisons of the results for both the surface-impregnated and uniformly impregnated catalysts are made in parts a, b, and c of Figure 3 for the conversion of CO, hydrocarbons, and NO, respectively. The filled symbols stand for the uniformly impregnated catalysts, while the open symbols are for the surface-impregnated catalysts. It is seen that the qualitative behavior of these two catalysts is nearly the same for all three pollutants even

360 400 440 460 320 360 400 440 460 Inlet Temperature (OC) Inlet Temperature (012)

320

320

360 400 440 460500 Inlet Temperature (OC)

Figure 3. Effect of the bandwidth of noble metal impregnation: (0,0) steady operation, ( 0 ,cyclic ~ ) operation at 0.1 Hz,(open symbols) surface-impregnation, (filled symbols) uniform impregnation.

0

200

300

400

500

600

Reactor Bed Temperature (OC) Figure 4. Comparison of the reaction lightoff temperature for steady feed CO oxidation over Pt/A120Bcatalysts under laboratory conditions with that under engine exhaust conditions: (0)laboratory feed stream, ( 0 )engine exhaust.

though there are slight differences in the level of conversion performances. Also the reaction lightoff temperatures measured a t the 50% conversion level of each reactant under steady operating conditions are essentially the same. From these observations, we might speculate that a wide-band impregnation may be preferable to the narrow-band impregnation in view of poison accumulation on the catalyst in actual emission control applications (Cho and Oh, 1985). The performance improvement by cyclic operation over steady operation is significant for both catalysts below the reaction lightoff temperatures. Warmup Performance of Catalysts. 1. Reaction Lightoff Temperature for CO Oxidation: Comparison with Laboratory Experiments. For a logical (if not direct) comparison of the reaction lightoff temperature in the laboratory CO oxidation experiments with that in the engine dynamometer experiments, we define CO equiualent in the engine exhaust as CO equivalent CCo + CH2 + 9 c H C - CNo (2) Physically this CO equivalent is the total concentration (in volume percent) of CO, H,,and hydrocarbons which can be oxidized by oxygen and is weighted by the oxygen-atom requirement. Figure 4 compares the results of temperature run-up experiments for steady-state CO oxidation over the surface-impregnated Pt/A1,0, catalyst in engine exhaust (closed circles) and in the CO-02 laboratory feedstream (open circles). The laboratory data, taken from Cho and West (1986), were obtained in a feed composition of 0.6% CO and 0.3% O2with Nzas the background gas. The data from engine-dynamometer experiments were obtained in an exhaust composition of 0.7% CO equivalent and 0.35% 0, with Nz, H,O, and CO, as background gases. It should be noted that the feedstream conditions of the engine-

Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 33 a. Steady Operation 100

z e

"1

,w::

.

m'

.m.

60

1

a. Surface Impregnation

b. Cyclic Operation at 0.1 Hz

b. Uniform Impregnation

A A ~ A A A A A

A A A A A

40

,tr)t ,

, ,

,

, ,

400 440 480 500 Inlet Temperature (OC)

360

-003

t

!'

2 0 t

1,

360

300

,

, ,

I

,

,

,

400 440 480 500 Inlet Temperature (OC)

Figure 5. Change of reaction selectivity due to cyclic operation: ( 0 ) CO, (A)hydrocarbons, (m) NO.

dynamometer experiments were approximately matched to those for the laboratory study in terms of CO equivalent and O2concentration. Interestingly, the catalyst lightoff temperature for CO oxidation in the engine exhaust is approximately 150 "C higher than that in the laboratory feedstream, even though the space velocity used in the engine-dynamometer experiments was somewhat lower than that for the laboratory experiments (25000 vs 30 000 h-l). This difference in the reaction lightoff temperature suggests that the CO oxidation activity of Pt/A1,03 catalysts is inhibited by the presence of additional components such as water vapor, sulfur, phosphorus, nitric oxide, hydrocarbons, etc., in the engine exhaust. Note that sulfur is detrimental to CO oxidation over Pt/A120, catalysts, and nitric oxide and hydrocarbons can compete with CO and O2 for active catalytic sites. Phosphorus accumulates in the catalyst very slowly compared to the time scale of this experiment (Cho and Oh, 1985), so that the catalysts were probably virtually free of phosphorus poisoning effects. 2. Selectivity among the Reactions. We have observed in Figures 2 and 3 that below the reaction lightoff temperature the Pt/A1203catalysts perform significantly better under 0.1 Hz cyclic operating conditions than they do under steady operating conditions. This can be attributed to the antichromatographic effect during the cyclic operation at low temperatures (Cho, 1983; Cho and West, 1986). In a multiple reaction system (as encountered in automobile exhaust catalysis), this antichromatographic effect can also play a profound influence on the selectivity of reactions during cyclic operation. Figure 5, which was constructed from the data presented in Figure 2, shows the performance of the surface-impregnated catalyst for the removal of CO, NO, and hydrocarbons under steady and cycled conditions. Under the steady operating condition (Figure 5a), lightoff occurs at the reactor inlet temperature of approximately 410 "C for all reactions, and the catalyst performances below the reaction lightoff temperature are nearly the same. However, under the slow cycling condition (0.1 Hz; Figure 5b), the catalyst becomes most active for the oxidation of hydrocarbons and only moderately active for CO oxidation followed by the reduction of NO. In other words, the relative rates of the multiple reactions are altered markedly by the slow feed composition cycling. We speculate that this change in reaction selectivity is related to the competitive reaction-adsorption dynamics resulting in the chromatographic and antichromatographic interactions. 3. Effect of Cyclic Operation on the Reactor Temperature. The change of reaction selectivity due to cyclic operation could play a very important role in the warm-up behavior of catalytic converters, because the selectivity change will inevitably lead to changes in the thermal behavior of the reactor.

400 500 600 Inlet Temperature (OC)

300

400 500 600 Inlet Temperature (OC)

Figure 6. Reactor bed temperature as a function of the inlet temsteady operation, (A) cyclic operation at 1 Hz, (0) perature: (0) cyclic operation at 0.1 Hz.

The effect of cyclic operation on the reactor bed temperature was examined by measuring separately both the reactor inlet temperature and the reactor bed temperature, as shown in parts a and b of Figure 6 for surface-impregnated and uniformly impregnated catalysts, respectively. It is noted that the general shape of the curve is nearly the same for both surface impregnation and uniform impregnation. There is, however, a smaller difference in the reactor temperature between 0.1-Hz cyclic operation and steady operation for the uniformly impregnated catalysts than that for the surface-impregnated catalysts. This can be attributed to the increased damping effect on the gas-phase composition fluctuation for the uniformly impregnated catalyst compared with the surface-impregnated catalyst (Cho and West, 1986). That is, when the active metal is spread over a wide band in the pellet, the increased diffusional distance tends to dampen the fluctuation of the reactant composition exerted on the external surface of the catalyst pellets. The reactor inlet temperature of 410 "C at which the curves for both the cyclic operation and steady operation cross each other corresponds to the reaction lightoff temperature observed for the steady feed experiments. In the regime below 410 "C, the reactor bed temperature at a given inlet temperature is higher for the cyclic operation than for the steady operation. This trend reverses above 410 "C, in agreement with the earlier observations of the conversion performance of the catalysts. This result indicates that the heat generation during the slow cyclic operation below 410 "C is substantially larger than that during the steady operat,ion or the fast cyclic operation. Also interesting to note is that the thermal response of the reactor during the cyclic operation is smoother than that during the steady operation which exhibits a sharp temperature rise a t the reaction lightoff temperature. In considering the practicality of cyclic operation during converter warm-up, it is worth noting that the converter warm-up from a cold start during the FTP (Federal Test Procedure) (Simanaitis, 1977) is achieved in a time period of approximately 90-200 s for a typical commercial three-way catalyst system (Oh and Cavendish, 1985; Monroe, 1986), depending on the converter size and the condition of the catalyst. Thus, a cycling period of 10 s, which yielded signficant improvement in the performance of our Pt/A1203catalyst during warm-up, is within the time scale of the warm-up period of commercial three-way catalyst systems. However, the time frame available for effecting improved performance by cyclic operation is not the total time to catalyst lightoff but the difference in time between the time to initial significant activity during cyclic operation and the time to reaction lightoff during steady operation. This difference should be much larger than the cycling period for cyclic operation. Unfortunately, we cannot determine from this work whether the cycling period of 10 s satisfies this condition or not.

34

Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 600

t E,

400

U

al 300 250 250 300

400

500

600

Inlet Temperature (OC)

Figure 9. Reactor bed temperature as a function of the inlet temperature for the time-average A/F ratio of 15.13: ( 0 )steady opera).( cyclic operation a t 0.1 Hz. tion, (A)cyclic operation a t 1 Hz,

CO HC NO 02 Chemical Species

J

Figure 7. Effect of additional air on the feed composition to the reactor: (blank column) standard engine exhaust, (filled column) air added to the standard engine exhaust. a

CO

c NO

b. Hydrocarbons

t L 250 300

400

500

Inlet Temperature (OC)

Inlet Temperature (OC)

300 400 500 Inlet Temperature (OC)

Figure 8. Effects of excess oxygen on the performance of Pt/Al2O3 catalysts: (0)steady operation, (A) cyclic operation at 1 Hz, (0) cyclic operation a t 0.1 Hz.

4. Effect of Excess Oxygen. In view of the oxygeninduced deactivation of the NO decomposition activity of Pt/A1,0, catalysts reported recently by Cho and Stock (19861, the effect of excess oxygen on the cyclic performance of Pt/A1,03 was examined by deliberately increasing the oxygen concentration in the engine exhaust. Additional air was injected into the engine exhaust gas upstream of the blank alumina bed (see Figure 1) in order to increase the time-average A/F ratio from 14.63 to 15.13. Figure 7 compares the reactor inlet concentrations for the stoichiometric A/F ratio of 14.63 with those for the A/F ratio of 15.13. When the A / F ratio was increased by adding air, the concentrations of CO, NO, and hydrocarbons were reduced as expected due to the dilution effect of the additional air. Figure 7 shows that the dilution effect on the concentrations of CO, NO, and hydrocarbons is much smaller compared with the magnitude of the increase in oxygen concentration. Thus, it is reasonable to assume that the difference in the catalyst performance between the A / F ratios 14.63 and 15.13 is primarily due to the effect of excess oxygen in the exhaust gas. Figure 8 shows the conversion performance of the uniformly impregnated Pt/A1203 catalyst at a time-average A/F ratio of 15.13. The difference in catalyst performance between steady operation and fast cyclic operation (1Hz) is indistinguishable, whereas the effect of slow cyclic operation (0.1 Hz) is rather significant. As expected, the lightoff temperatures for the oxidation of CO and hydro-

carbons are significantly reduced upon injecting additional air, while the reduction of NO is severely suppressed. The former is consistent with the previous findings on CO oxidation (Cho and West, 1986), and the latter with the oxygen-poisoning effect on the NO decomposition activity of Pt/A1203catalysts (Cho and Stock, 1986). At the inlet temperatures below 325 O C , CO and hydrocarbons can be removed from the engine exhaust more efficiently during the slow cyclic operation than during the steady or fast cyclic operation, whereas this trend reverses at the inlet temperatures above 325 "C. This is consistent again with the previous findings on CO oxidation with excess O2 in the laboratory reactor (Cho and West, 1986) and in the engine exhaust using Pt/Pd/A1203 catalysts (Hegedus et al., 1980). The slightly improved performance of the Pt/A120, catalyst for NO reduction during A/F cycling compared with the steady performance under the net oxidizing condition (i.e., A/F = 15.13) over the entire temperature range investigated can be attributed to the more severe deactivation of the Pt/A1,03 catalyst due to the oxygen accumulation under the steady feed conditions. That is, the Pt/A1203catalyst is under a deactivated state during the steady operation in the net oxidizing environment, whereas it may have a chance to be regenerated during the cyclic operation when it is in the stoichiometric (A/F = 14.63) half-cycle. These observations can be summarized as follows. 1. For the stoichiometric time-average feed composition, the catalyst performs better under cyclic operation a t 0.1 Hz than under steady operation for all three pollutants below the reaction lightoff temperature, whereas this trend reverses above the reaction lightoff temperature. 2. For the net-oxidizing time-average feed composition, the catalyst performs better under cyclic operation at 0.1 Hz than under steady operation for the oxidation of CO and hydrocarbons below the reaction lightoff temperature, whereas this trend reverses above the reaction lightoff temperature. 3. For the net-oxidizing time-average feed composition, the catalyst performs better under cyclic operation at 0.1 Hz than under steady operation for NO reduction over the entire temperature range of practical interest. Figure 9 shows the reactor bed temperature as a function of the inlet temperature for the uniformly impregnated catalyst for the time-average A/F ratio of 15.13. These results differ in Figure 9 from the stoichiometric cyclic operation shown in Figure 6; the evolution patterns of the reactor bed temperature are essentially the same for steady operation, fast cyclic operation at 1 Hz, and slow cyclic operation at 0.1 Hz. That is, when there is abundant

Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 35 oxygen available, the catalyst performance exhibits a lesser dependence on the operational mode of the reactor.

Summary and Discussion Our previous studies (Cho, 1983; Cho and West, 1986) have culminated in this work where the theory and laboratory experimental results were tested under more realistic engine exhaust conditions. Previously we reported an improvement in the catalyst performance under cyclic operation compared with steady operation for CO oxidation over Pt/A120, below the reaction lightoff temperature (Cho and West, 1986). In this work, we observed similar performance characteristics for the Pt/A1,0, catalysts for all three emission components (CO, NO, and hydrocarbons) under engine exhaust conditions with a composition cycling frequency of 0.1 Hz. This result demonstrates the validity of our previous proposal for a hybrid scheme for converter operation: cyclic operation below the converter lightoff temperature and steady (or close to steady) operation above the converter lightoff temperature. It should be noted though that this improvement in the catalyst performance by cyclic operation becomes noticeable only when the cycling period is sufficiently long, as we have seen in the conversion performance as well as in the reactor temperature. This does not necessarily mean that the cycling period should be as long as 10 s. It rather suggests that a cycling frequency between 0.1 and 1Hz has a strong possibility of improving catalyst performance significantly when the catalyst temperature is below the reaction lightoff temperature. The reason why the fast cycling at 1 Hz performs nearly the same as the steady operation is obvious as stated in the Experimental Section: at this fast cycling frequency, the fluctuation in the engine exhaust composition is smeared out due to dispersion to the level close to the steady operation by the time the gas stream reaches the catalytic surfaces. The enhanced catalyst performance below the reaction lightoff temperature can be explained by the antichromatographic effect by which the reactant population dynamics on the catalytic surface are directed toward a more even distribution under the cyclic operation than under the steady operation. That is, the antichromatographic effect helps initiate the surface bimolecular reaction at low temperature by promoting the mixing (or collision) of reactants on the surface, which results in the release of reaction enthalpy, thereby promoting the reaction further. The beneficial effect of cyclic operation below the reaction lightoff temperature is most pronounced for the oxidation of hydrocarbons; the hydrocarbons are preferentially oxidized a t a lower temperature than the CO is oxidized or NO is reduced. Nevertheless, the conversion of hydrocarbons still increases with temperatures above 400 OC where the CO conversion is already at its plateau, indicative of a mass-transfer-controlled regime. These observations suggest that the hydrocarbons in the engine exhaust can be classified into two groups depending on their oxidation rate over Pt/A1203: one group which oxidizes more easily than CO, and another group which is more difficult to oxidize than CO. This classification is consistent with the literature data (Kuo et al., 1971) that differentiated between fast-oxidizing hydrocarbons such as propylene and slow-oxidizing hydrocar.bons such as methane. The heat of reaction released from the fastoxidizing hydrocarbons helps initiate the oxidation of CO and the slow-oxidizing hydrocarbons, and later the reduction of NO. This rather sequential pattern of reaction, which we may call a domino effect, is a direct result of changing selectivity among the multiple reactions under cyclic operating conditions. The changing selectivity in

turn is a result of an adsorption-reaction interaction in which the chromatographic and antichromatographic effects play important roles in determining the performance of catalysts. We note, however, that during steady operation the multiple reactions do not seem to occur sequentially but occur rather simultaneously at a higher temperature than during cyclic operation. The fairly good activity of the Pt/A1203catalyst for NO reduction is in agreement with our recent laboratory study on NO decomposition (Cho and Stock, 1986). The reaction lightoff temperature for CO oxidation under engine exhaust conditions is about 150 " C higher than that under laboratory conditions observed by Cho and West (1986). Neglecting the effect of a small difference in space velocity used in these two experiments, we attribute the higher reaction lightoff temperature under engine exhaust conditions to the additional chemical species such as sulfur, water vapor, nitrogen oxides, the hydrocarbons present in engine exhaust. Note particularly that sulfur is known for its poisoning effect on Pt/A1203 catalysts (Taylor, 1984). These findings suggest that future work be directed toward the improvement of the lightoff performance of Pt/A1,0, catalysts under engine exhaust conditions. We have noted that the reaction lightoff temperatures measured by 50% conversion of reactants are nearly the same for both surface impregnation and uniform impregnation, and the advantage of the surface impregnation starts to show up only above the reaction lightoff temperature. Even this advantage is far from significant. On the basis of the results of this study, we are led to believe that cyclic operation generally performs better than steady operation below the reaction lightoff temperature. It is noted, however, that the cycling frequency should be slower than 1 H z to affect a noticeable improvement in catalyst activity. In order to take advantage of this rule in the automobile catalytic converter, it is necessary to develop an A/F control mechanism which is operable at ambient temperature. To use a feedback control scheme for this purpose, like the one currently in use, it is necessary to develop an oxygen sensor which is operable at low temperature (preferably at the ambient temperature). Otherwise an open-loop control mechanism can be considered as an alternative to the feedback control mechanism below the reaction lightoff temperature.

Acknowledgment The author gratefully acknowledges D. R. Monroe for his helpful consultations on the experimental system and M. J. D'Aniello, K. D. T. Dang, and M. G. Zammit for preparation of catalyst samples. The engine-dynamometer experiments were conducted by D. J. Upton. Registry No. Pt, 7440-06-4; CO, 630-08-0; NO, 10102-43-9; 02,

7782-44-7.

Literature Cited Abdul-Kareem, H. K.; Silveston, P. L.; Hudgins, R. R. Chem. Eng. Sci. 1980, 35, 2077. Adams, K. M.; Gandhi, H. S. Ind. Eng. Chem. Prod. Res. Deu. 1983, 22, 207. Billimoria, M. R.; Bailey, J. E. A C S Symp. Ser. 1978, 65, 526. Cho, B. K. Ind. Eng. Chem. Fundam. 1983,22, 410. Cho, B. K.; Oh, S.H. Ind.,Eng. Chem. Process Des. Deu. 1985, 24, 897. Cho, B. K.; Stock, C. J. Presented at the 1986 Annual Meeting of American Institute of Chemical Engineers, Miami Beach, FL, Nov 1986. Cho, B. K.; West, L. A. Ind. Eng. Chem. Fundam. 1986, 25, 158. Cutlip, M. B. AIChE J . 1979, 25, 502.

36

I n d . Eng. C h e m . Res. 1988, 27, 36-41

Fiolitakis, E.; Schmid, M.; Hofmann, H.; Silveston, P. L. Can. J . Chem. Eng. 1983,61, 703. Hegedus, L. L.; Oh, S. H.; Baron, K. U.S. Patent 4222236, 1980. Herz, R. K. Ind. Eng. Chem. Prod. Res. Deu. 1981, 20, 451. Herz, R. K.; Kiela, J. B.; Sell, J. A. Ind. Eng. Chem. Prod. Res. Deu. 1983, 22, 387. Jones, H. J.; Kummer, J. T.; Otto, K.; Shelef, M.; Weaver, E. E. Environ. Sci. Technol. 1971, 5, 790. Kuo, J. C. M.; Morgan, C. R.; Lassen, H. G. SAE Paper 710289, 1971. Monroe, D. R., General Motors Research Laboratories, Warren, MI, private communication, 1986. Muraki, H.; Shinjoh, H.; Fujitani, Y. Appl. Catal. 1986, 22, 325. Oh, S. H.; Cavendish, J. C. AIChE J . 1985, 31, 943. Schlatter, J. C.; Sinkevitch, R. M.; Mitchell, P. J. Ind. Eng. Chem. Prod. Res. Deu. 1983, 22, 51.

Sell, J. A.; Herz, R. K.; Monroe, D. R. SAE Paper 800463, 1980. Silveston, P. L.; Hudgins, R. R.; Adesina, A. A.; Ross, G. S.; Feimer, J. L. Chem. Eng. Sci. 1986, 41, 923. Simanaitis, D. J. Auto. Eng. 1977, 85(8), 34. Summers, J. C.; Monroe, D. R. Ind. Eng. Chem. Prod. Res. Dev. 1981, 20, 23. Taylor, K. C.; Sinkevitch, R. M. Ind. Eng. Chem. Prod. Res. Deu. 1983, 22, 45. Taylor, K. C. In Catalysis-Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer-Verlag: West Berlin, 1984; Vol. 5, p 119. Yokota, K.; Muraki, H.; Fujitani, Y. SAE Paper 850129, 1985.

Received for review February 4, 1987 Accepted September 24, 1987

Kinetics of Absorption of Oxygen in Aqueous Alkaline Solutions of Polyhydroxybenzenes Anand V. Patwardhan and Man Mohan Sharma* Department of Chemical Technology, University of Bombay, Matunga, Bombay 400 029, India

Kinetics of oxygen absorption in aqueous alkaline solutions of polyhydroxybenzenes (PHBs) such as pyrogallol (PG), p-tert-butylcatechol (PTBC), tert-butylhydroquinone (TBHQ), 2,3,5-trimethylhydroquinone (TMHQ), and gallic acid (GA) was studied in a jet apparatus, a stirred cell, and a model stirred contactor, a t 29 f 1 “C. In the cases of PTBC and TBHQ, the reaction was found to be first order in oxygen as well as PTBC or TBHQ. The intrinsic second-order rate constants were in the range of 1 X 103-1 X lo5 m3/(kmol.s). In the case of PG, GA, and TMHQ, the system conformed to the instantaneous reaction regime. The theory of gas absorption with an instantaneous reaction was used to calculate the diffusivity of dissolved P G and GA. At very low partial pressures of oxygen, the oxygen absorption in alkaline P G can become predominantly gas film controlled. Polyhydroxybenzenes (PHBs) are phenols with two or more hydroxyl groups. These are also called benzenepolyols. The absorption of oxygen in the aqueous alkaline solutions of PHB and substituted PHB, such as 1,2,3benzenetriol or pyrogallol (PG), 3,4,5-trihydroxybenzoic acid or gallic acid (GA), p-tert-butylcatechol (PTBC), tert-butylhydroquinone (TBHQ), and 2,3,5-trimethylhydroquinone (TMHQ), was studied. The absorption of oxygen in aqueous alkaline solutions of PHBs and their substituted derivatives is relevant in several contexts with respect to the understanding of functioning of antioxidants, new methods of removing oxygen from certain streams, etc. The classic example is the use of alkaline PG in the Orsat apparatus for analysis of gaseous mixtures containing oxygen (Vogel, 1975). PG appears to be the strongest reducing agent among the benzenepolyols, as is evident from the fact that its aqueous alkaline solutions absorb oxygen from air and darken rapidly. PG, as an antioxidant, is useful in protecting decomposition of alkali cellulose (Langmaack, 1971). A mixture of defatted rice bran and alkalized PG is useful in protecting foodstuffs from oxygen. GA and its propyl ester are useful as antioxidants for various duties. The dihydroxybenzenes, namely PTBC and TBHQ, are used as antioxidants, antiozonants, monomer inhibitors, etc. There is scanty information in the literature on the kinetics of absorption of oxygen in aqueous alkaline solutions of PHBs and their derivatives. It is, however, apparent that these reactions are extremely fast and mass transfer is accompanied by chemical reaction in the diffusion film. It was, therefore, thought desirable to study the kinetics of absorption of oxygen in aqueous alkaline solutions of PG, GA, PTBC, TBHQ, and TMHQ.

Previous Studies Recently, Rothe (1985) has reported the production of nitrogen-rich inert gas from air by the absorption of oxygen in alkaline P G solution in a packed column. The spent PG solution has been claimed to be regenerated by heating for the desorption of oxygen. Takahashi et al. (1980) have reported the kinetics of absorption of oxygen in alkaline solution containing hydroquinone. The reaction was found to be first order in oxygen and first order in hydroquinone. Takeuchi et al. (1980) have reported the kinetics of absorption of oxygen in aqueous alkaline solutions containing the sodium salt of 1,4-naphthohydroquinone-2-sulfonic acid (NHQS). The reaction was found to be first order in oxygen and first order in NHQS. Catechol and hydroquinone are converted to the corresponding 0-and p-benzoquinones by most oxidizing agents. The chemistry of oxidation of PG and substituted PG has been studied by various workers. Nierenstein (1915) observed that the oxidation of PG in potassium hydroxide solution gives, among other products, 2,3,2’,3‘,2”,3”hexahydroxyphenoquinone. Campbell (1951) has reported OH OH OH O H OH OH

o 0> 2,3, 2 ’ , 3 ,

f,3’-h e x a h y d r o x y p h e n o q u i n o n e

that the oxidation of 4,6-di-tert-butylpyrogallol in alkaline solution with air gives an orthoquinone, which then rearranges to give further products. Purpurogallin, a redbrown to black mordant dye, is obtained from electrolytic

0888-5885/8~/262~-0036$01.50/0 0 1988 American Chemical Society