2219
Znd. Eng. C h e m . Res. 1991,30,2219-2225
V = Nabla operator Literature Cited
Krantz, W. B.; Goren, S. L. Finite-Amplitude, Long Waves on Liquid Films Flowing Down a Plane. Znd. Eng. Chem.F u d a m . 1970,9,
Bierman, V. M.Das Fliej3en ebener Schichten inkompreaeibler viswelastheher Flhigkeiten mit einer in Ablaufrichtung infinitesimal geetBrten freier Oberfliiche. Rheol. Acta 1968, 7, 138. Camerlengo, A. L.; OBrien, J. J. Open Boundary Conditions in RQ tating Fluids. J. Comput. Phys. 1980, 35, 12. Degani, D.; Gutfinger, C. Levelling of a Newtonian Fluid on a Horizontal Surface. Zsr. J. Technol. 1974, 12, 191. Degani, D.; Gutfinger, C. A Numerical Solution to the Leveling Problem. Comput. Fluids 1976,4, 149. Grimshaw, R.Theory of Solitary Waves in Shallow Fluids. In Encyclopedia of Fluid Mechanics; Cheremisinoff, N. P., Ed.; Gulf Publishing: Matawan, NJ, 1986; 1986; Vol. 2. Hirt, C. W. A Technique for Including Surface Tension Effects in Hydrodynamic Calculations. J. Comput. Phys. 1969,4, 97. Keunings, R.;Bousfield, D. W. Analysis of Surface Tension Driven Leveling in Viscoelastic Films. J. Non-Newtonian Fluid Mech.
Krantz, W. B.; Goren, S. L. Stability of Thin Liquid Filma Flowing Down a Plane. Znd. Eng. Chem. Fundam. 1971,10,91. Malamataris, N. G. Computer-aided analysis of flow on moving and unbounded domains: Phase-change fronts and liquid leveling. Ph.D. Dissertation, The University of Michigan, 1991. Orchard, S. E. On Surface Levelling in Viscous Liquids and Gels. Appl. Sei. Res. 1962, 11,451. Orlanski, I. A Simple Boundary Condition for Unbounded Hyperbolic Flows. J. Comput. Phys. 1976, 21, 251. Quach, A. Polymer Coatings. Physics and Mechanics of Leveling. Znd. Eng. Chem. Prod. Res. Dev. 1973,12,110. Wasden, F. K. Studies of mass and momentum transfer in free falling wavy films. Ph.D. Dissertation, University of Houston,
107.
1987, 22, 219.
Kheahgi, H. S. The Motion of Viscous Liquid Films. Ph.D. Dissertation, University of Minnesota, 1984. Kheshgi, H. S.; Scriven, L. E. Measurement of Liquid Film Profiles by Moire Topography. Chem. Eng. Sci. 1983,38,525. Kheshgi, H. S.; Scriven, L. E. Penalty finite element analysis of unsteady free surface flows. Finite Elem. Fluids 1984,5, 393.
1989.
Wasden, F. K.; Dukler, A. E. Insights into the Hydrodynamics of Free Falling Wavy Films. AZChE J. 1989,35, 187. Zorll, W. Untersuchungen fiber Verlauf und Laufneigung bei flhigen Lackflmen mit kleinen OberfXchenunebenheiten. Mitt. Dtsch. Blechverarbeitung Oberflaechenbehandl. 1970, 11, 217.
Received for review December 11, 1990 Revised manuscript received April 29, 1991 Accepted May 24,1991
Room Temperature Purification of Humid Air Containing Low Levels of Carbon Monoxide Ravindra Yaparpalvi and Karl T. Chuang* Department of Chemical Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2C6
The use of clean air is mandatory in numerous applications in order to meet the increasingly stringent indoor air pollution standards. A catalytic process for the purification of humid air containing low levels of carbon monoxide (CO) was developed. During preliminary screening many supported metal catalysts containing various active ingredients were examined for room temperature oxidation of CO. The results indicate that platinum supported on hydrophobic styrene-divinylbenzene copolymer meets the catalyst requirements. It was observed that the activity of this catalyst was enhanced by water vapor. This is a marked contrast to platinum on hydrophilic supports which deactivates in the presence of water vapor. In all experiments, steady-state multiplicity and oscillations were observed. The phenomena were similar to those reported in the literature for high CO concentrations and high temperatures.
Introduction Carbon monoxide (CO), a priority air pollutant in a variety of industrial and domestic environments, has a lengthy history of serious health implications (Stewart, 1975). A combination of the widespread Occurrence of low (ppm) level concentrations of CO in the ambient air, mostly generated by vehicular emissions, and reports of ita adverse effects upon long-term exposure to even ppm level concentrations (Greek and Dorweiler, 1990) has created great concem, prompting development of methods that control exposure of CO to humans in both industrial and nonindustrial environments. Control of such emissions via catalytic oxidation offers potential advantages over other methods, e.g., adsorption, in terms of equipment size and operating simplicity. However, apart from being low in CO concentration, the polluted air is often saturated with water and available only at room temperturea. Hence the catalysts must be active at room temperatures and must not be poisoned by the presence of water vapor. Catalytic oxidation of CO has received extensive attention in the literature. The majority of these studies has, however, been carried out mainly in relation to automobile 0888-5885/91/ 2630-2219$02.50/0
emission control, and therefore, reaction conditions involve temperatures in excess of 300 O C . There are innumerable situations in industrial and domestic sectors wherein the use of clean air for various applications is imperative. Typical examples are breathing air, workshop air, and air for agitation in fermenters. While the current technology appears to be quite satisfactory in removing various airborne contaminants like dirt particles and oil, the same can not be said for the removal of low-level CO at ambient temperatures. This has largely been due to the failure of the commercial hopcalite catalyst to effectively remove CO from air in the presence of water vapor. Furthermore, in purifying large volumes of air, ambient temperature operation that requires no feed preheating is preferred. The main objective of the present work is to develop a catalytic process suitable for the removal of CO from polluted air streams. According to Hiross Canada Inc. (2550 Dunwin Dr., Mississauga, Ontario, Canada L5L 155) the reactor should operate at room temperature with a space velocity of greater than 400OOO h-* to be economically competitive. T w o classes of catalyst are normally used for the oxidation of CO by 02. They are supported noble metale and 1991 American
Chemical Society
2220 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991
the oxides of transition metals. All noble metal based catalysts investigated so far are mostly supported on inert, highly porous materials such as Si02and A1209.However, these catalysts must be heated to prevent a gradual loss in activity due to the accumulation of capillary condensed water in the catalyst support. Similarly the metal oxide catalysts show no activity when exposed to air containing water vapor. The mechanism of the catalytic oxidation of CO on noble metal catalysts is discussed in review articles (Engel and Ertl, 1979; Golodets, 1983). There is general agreement that dissociative chemisorption of O2on the catalyst metal surface is a necessary prerequisite for the reaction to take place with chemisorbed CO and that the reaction is adequately explained by the Langmuir-Hinshelwood mechanism. Preadsorbed CO has been reported (Conrad et al., 1978) to act as an inhibitor for O2 adsorption and results in low rates of oxidation. Adsorption characteristics of CO and O2on noble metal surfaces have been published (Engel and Ertl, 1982; Ford, 1970). The established configurations of CO adsorbed on a metal surface are the linear (M-CO) and the bridge bonded (M4O-M), and the linear species is reported to be catalytically more active. Tucker (19641, on the basis of LEED study, has concluded that there may be a limited mobility of CO on Pt(100) at room temperature. Similarly, an initial sticking coefficient of 0.92 for CO and 0.1 for O2on platinum surfaces at room temperature has been reported (Campbell et al., 1981; Norton et al., 1984). Therefore CO is believed to adsorb on platinum much more strongly than 02. Cant et al. (1978) found that the CO oxidation activity of noble metal catalysts, supported on y-alumina, silica-alumina, etc., at relatively low temperatures (330-480 K) was high in the beginning and then reduced to a constant value. They found that the CO/O2 ratio and temperature were the critical factors. Multiplicity of steady states and oscillations have been reported for CO oxidation at high temperatures and high concentrations (Razon and Schmitz, 1986). The brief literature review provides the background for the interpretation of our experimental results.
Experimental Section Selection of Support and Catalyst Material. It has been reported (Katz, 1953) that water vapor does not get adsorbed on Pt-group metals to an appreciable amount. If the metals are supported on a hydrophobic material, the presence of water vapor in the feed air will not cause capillary condensation, and therefore, the reactant gases can always reach the catalyst active sites. The use of polymeric hydrophobic catalyst supports has been recognized for the oxidation of hydrogen (Chuanget al., 1987). In the present work hydrophobic styrene-divinylbenzene copolymer (SDB) was found to satisfy our requirements. During the development stage a number of SDB-supported catalysts containing active ingredients such as Pt, Ir, Pd, Rh, and oxides of copper, cobalt, silver, tin, etc., in various combinations and proportions were prepared and tested as candidates for room temperature oxidation of CO. However, experimentation beyond the screening stage was carried out only with Pt-Ir/SDB Catalysts which showed the highest activity compared to other catalysts as well as to SDB-supported catalysts containing equivalent loadings of only Pt. Preparation of Support and Catalyete. The SDB support was obtained by polymerizing divinylbenzene in ethylvinylbenzene. The reaction was carried out in 2methyl-1-pentanol using 2,2-azobis(2-methylpropionitrile) as initiator. The SDB so obtained was crushed and
Table I. Characteristics of Pt-Ir/SDB Catalysts metal area, % % cat. % metal m2/a no. loading Pt 11 3.05 1.8 0.2 1 2.0 5.4 0.6 4.18 2 6.0 9.0 1.0 10.0 3 10.0
size, mesh 10-14 10-14 10-14
screened to different sizes. It had a BET area of 465 m2/g. The Pt-Ir/SDB catalysts were prepared by the impregnation of SDB with an ethanol solution containing a given amount of H2PtCg and Ir(NH3)&l9. The mixture was installed in a rotary evaporator that operated at 95 "C and under a slight vacuum. The dried product was then reduced in hydrogen at 200 "C until the pH of the furnace outlet became neutral. The platinum was found to distribute uniformly, and the percent loading of the catalyst was based on the initial weight of the metal complexes. The metal area was determined by H2/02titration and the characteristics of these catalysts are listed in Table I. Other supported catalysts tested during the development stage were prepared by impregnation of the support by appropriate strength solutions of the corresponding metal salts, namely, H2PtC&, Pd(NH3)&l2.H20, Pd(NO3I2, RhC13, and nitrates of copper, cobalt, silver, etc. Some hydrophilic supports were also used to compare results. These were Y zeolite (type AW500, Union Carbide), Silicalite (SR-115, UOP), y-alumina (Kaiser, BET area 341 m2/g),and semihydrophobic alumina (Met Pro). Following drying, each preparation was calcined at 573-673 K and subsequently reduced at the same temperature with a hydrogen/nitrogen gas stream overnight. Equipment. Experimental runs were carried out with the use of a flow reactor system consisting essentially of a feed purification train, a water vapor saturator, and a fixed-bed integral reactor. Two reactors of different dimensions were employed. A primary reactor of 38.1 mm i.d. X 345 mm long was used to demonstrate catalyst performance under operating conditions encountered in practice, while a secondary reactor of 25.4 mm i.d. X 190 mm long was used to test for catalyst activities during development stage. Both the reactors are made of stainleas steel. The inlet and outlet lines to the primary reactor system consisted of 12.7 mm i.d. copper pipe, whereas a 6.35 mm i.d. high-strength Teflon tubing served as inlet and outlet lines for the secondary reactor. The feed gas to the reactor system was air. This air was drawn from a compressed air supply line (approximately 70 psig) and was bone dry (pressure dew point -40 "C). A stabilizer was installed in the line to subdue any fluctuations in supply air pressure. The air was further purified by passing it through a purification train in four stages of filtration as follows: (stage 1)Domnick Hunter grade A 0 1-pm prefilter for particulate removal; (stage 2) Domnick Hunter grade SL soda line filter for C02removal; (stage 3) Domnick Hunter grade AA O.Ol-pm high-efficiency coalescing filter; (stage 4) Domnick Hunter grade AC activated carbon filter for organic vapor removal. A part of or all the purified air was then passed through a water vapor saturator to obtain desired humidity. The saturator was a 152.4 mm i.d. X 914.4 mm high stainless steel column and was filled with water to the required level. A sparger was installed to ensure complete saturation. The humidity and temperature of air just before it entered the reactor were continuously monitored by a Vaisala Series HMP 124B (Cole-Parmer) humidity and temperature transmitter. This sensor has a temperature measurement range of -20 to +80 OC and relative humidity (RH) measuring range of 0 to 100% RH. The impurity level of CO in air was achieved by continuously injecting the ap-
Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991 2221
Results and Discussion
To Von
To Vont
Figure 1. Schematic diagram of the flow reactor system. PS, pressure stabilizer; HS, humidity and temperature sensor; F12, filtration stages 1 and 2; F34, filtration stages 3 and 4; MFC,mass flow controller; DPC,differential pressure cell; WVS,water vapor saturator; FM, flowmeter; PG, pressure gauge; PR, primary reactor; SR, secondary reactor; SV, solenoid valve; SC, source cylinder.
propriate amount of CO. Essentially, CO flow from source cylinder (technical grade, Linde-Union Carbide) was controlled by means of a mass flow controller (Unit Instruments) in order to obtain the desired CO concentrations in inlet air. The catalyst in the reactor was supported on a fine stainless steel wire mesh along with glass wool/glass beads packing in order to arrest any catalyst movement. A schematic diagram of the complete flow reactor system is shown in Figure 1. The CO concentrations in air at the reactor inlet and outlet were continuously monitored and recorded by use of a 2000 series ECOLYZER CO analyzer. This electrochemical cell analyzer had two scale ranges of 0-60 and 0-600 ppm. The rise time (90%) of the instrument was 25 s. The precision and minimum detectable sensitivity of the analyzer were il% (full scale) and 0.5% (fullscale), respectively, and the performance of the instrument was not affected by the presence of gases like COz,NHs, NzO, and SOz in the environment. The analyzer was calibrated on a regular basis to ensure accuracy and reproducibility in analysis using standardized CO in air mixtures (6 ppm, 197 ppm) supplied by Matheson. Procedure. The experimental runs were mainly directed toward investigating the effect of CO concentration in the inlet air on the catalyst activity measured in terms of steady-state conversions (CO conversion in this study is defined as ppm of CO oxidized divided by CO ppm level in the inlet air and has also been referred to as oxidation efficiency). Similarly runs were conducted to study the effect of various operating variables such as catalyst composition, water vapor content in inlet air, total flow rate, and total pressure on the oxidation efficiency of the catalyst. Therefore, starting with a regenerated catalyst, the experiment would begin by setting air flow rate, etc., to a desired value and adjusting the inlet CO concentration to the lowest value of interest. When steady state was reached, as indicated by a constant outlet concentration of CO on the analyzer, the CO ppm level in inlet air was given a step raise and again the steady-state outlet concentration was noted. This incremental increase in inlet CO concentration was continued until the catalyst oxidation efficiency dropped to the low conversion regime. In the reverse direction experiment, the run ended when the CO inlet concentration was decreased from its highest value in steps to its lowest value. Multiplicity of steady states was observed and will be discussed later. Regeneration of the catalyst was carried out by shutting off CO feed and flushing with purified air for a period of approximately 1h. This procedure was found to recover the catalyst activity previously poisoned by CO.
The experimental conditions, such as the inlet CO ppm level, in this investigation were deliberately adjusted to cause poisoning of the catalyst and do not reflect a failure of the catalyst performance under actual operating conditions. The objective was to mark the tolerance limits and study the behavior of the catalyst under various operating conditions that may be encountered in industrial applications. Extraneous Catalytic Effects. Before presenting the experimental results, it is worthwhile to address the issue of transport disguises. It is known that heat- or masstransfer effects, caused by intrareactor, interphase, or intraparticle gradients, can disguise the results and lead to misinterpretations. A blank experimental run, carried out with only the catalyst support in the reactor, exhibited no CO conversion indicating the reactor and support materials to be noncatalytic. In the flow system, the flow rate was varied while the space velocity, defined in this study as the volumetric flow rate of the combined gas stream (i.e., air + CO + water vapor) entering the catalyst bed divided by the volume of the catalyst bed, was kept constant. Accordingly, runs were carried out using different catalyst bed depths and, in the meantime, maintaining the same space velocity. It was observed that similar values of CO conversions were obtained at a fixed space velocity, thus indicating the absence of mass-transfer effects. Since the CO feed concentrations used in this study were less than 300 ppm and in most runs below 100 ppm, the temperature increases as a result of heat released from the CO oxidation was less than 3.0 OC. This was confirmed by similar readings of the thermocouples at the inlet and outlet of the reactor. Hence it can be assumed that the reactor was operated isothermally for all the experiments. Effect of CO Concentration. Three catalysts listed in Table I were tested to determine the effect of feed concentration on catalyst activity. Figure 2 shows the steady-state CO conversion versus feed concentration, at two different flow rates for each catalyst. As expected, when the total flow rate and hence the space velocity (SV) decreased, the overall conversion increased. Therefore, the curve for the higher flow rate appears on the left side of the lower flow rate curve. As CO feed ppm level was increased from zero to higher values, the reactor outlet conversion was close to about 100%, defined as a high conversion regime. After the inlet CO concentration was raised to a certain point, the conversion dropped to a low value, defined as a low conversion regime. The exact point of drop depended mainly on the CO inlet concentration, air flow rate, loading of active material, and catalyst particle size. The features observed during this transition from a high conversion regime to a low conversion regime have been further discussed in the following paragraphs. The steady-stateconversionsof CO as a function of inlet concentration in consecutive cycles using 9% Pt-1% Ir/ SDB catalyst are presented in Figure 3. Similar to many reports on CO oxidation (Razon and Schmitz, 1986), hysteresis phenomena were observed. The term hysteresis here refers to a multiplicity of steady states, and the actual steady state was observed to depend on the past history of operation of the reactor. In particular, different CO conversions were obtained for the same inlet CO concentration depending on the direction in which the CO inlet concentration was varied. Beginning with a regenerated catalyst, desired inlet conditions were maintained at a low CO inlet concentration until steady state was reached, giving one data point on the top part of the curve. Then
2222 Ind. Eng. Chem. Res., Vol. 30, No.9, 1991 R
d 0
100
.d
b
$
E: 0 V 0
-\.. -
40 20 60
V
I
I
I
L\
I
I
(4
AA\
8. "88
I
0
I
AAM-A-AA.
I
A>
1
A.&A
R
d ."VI0
.
9.8-.-.
Ll
gC
60 W_
0
u
M
0 V
R
.-$ VI
s
100
-
a0
40
-
20
-
60
0
u 0
u
0
I
I
I
I
I
I
0-A-A-A-A-A-A-A-A-A-A-A
I 8 8 .
(c)
\8
-
\W ' 8 1
1
I
I
I
Feed Concentration (ppm CO) Figure 2. Effect of feed concentration on CO conversion at 23 O C . Catalyst volume, 30 mL. (a) 1.8% Pt-0.2% Ir/SDB. SV 7200 h-l (A); 29800 h-l (m). (b) 5.4% Pt-0.6% Ir/SDB. SV: 29800 h-I (m); 68400h-' (A). (c) 9.0%R-1.0% Ir/SDB. SV 29800 h-' (A);68400 h-' (m).
loo
t \
\
0
50
I
\
100
150
200
250
300
Feed Concentration (ppm C O ) Figure 8. CO conversion hysteresis. SV, 68400 h-l; T, 24 OC; RH, O%, catalyst, 9% Pt-1% Ir/SDB.
the inlet CO concentration was raised, and the experiment again proceeded to a steady state, giving another data point. This procedure was repeated in obtaining the curve denoted in Figure 3 as 1A. The CO concentration was then decreased, step by step, until a new steady state was reached each time. This gave rise to the curve denoted as 1B in Figure 3. Thus for a given CO inlet concentration between the high conversion and low conversion regimes, there was a transition regime, which shows different steady-state conversions. During the transition regime,
oscillations in conversion were always observed. These oscillations started when the inlet CO concentration reached a certain point and ended when the CO concentration fell below this value during an experimental run in the reverse direction. These observations, along with different explanationssuggeated by various authors CRazon and Schmitz, 1986) for observed oscillations at high temperatures and concentrations, lead us to believe that the oscillations in this study were caused by varying surface concentrations of the reacting species. Furthermore, because of the low CO concentrations involved, long transients for a complete transformation from a high conversion regime to low conversion regime were observed during tests. The time required to reach a complete saturation of CO on the surface was of the order of hours. Figure 3 also shows that the catalyst activity was reduced in going from the first to the second cycle. The second cycle run, represented by curves 2A and 2B, was carried out immediately following the cycle 1 run without regenerating the catalyst. This suggests that in the absence of the regeneration step, the adsorbed CO from the previous experiment was not completely removed and therefore the transition from the high to low conversion regime occured at a lower ppm. Following cycle 2, the catalyst was flushed with purified air for a period of approximately 1h and the activity was again tested in cycle 3. The catalyst exhibited the same activity as that obtained in cycle 1,thus indicating that the activity can be recovered by flushing with air at ambient temperature. Cant et al. (1978) and Tamhankar and LaCava (1989) have reported that the initial interaction of the platinum surface with CO permanently blocked some of the active sites and required high temperature for regeneration. In our study, the Pt-Ir catalyst supported on the hydrophobic SDB seems readily regenerable at low temperatures, indicating that the active ingredient and/or support surface may play an important role in the regeneration. The occurrence of two conversion regimes and an unstable region wherein multiplicity and oscillations may be explained by the ability of platinum to adsorb CO and O2at various operating conditions. It is generally agreed that the rate of CO oxidation is controlled by the reaction between the adsorbed CO and 02. At room temperatures CO is adsorbed more strongly than O2 (Engel and Ertl, 1982) on the platinum surface. Thus adsorbed CO inhibits the adsorption of oxygen. The large CO coverage at high ppm values results in a low conversion regime due to the lack of adsorbed oxygen. On the other hand, the adsorption of CO is not affected by the presence of adsorbed oxygen at the beginning of the experimental run. This results in a high conversion regime where the CO molecules impinging upon the catalyst surface react with high probabilities. As the CO concentration in the feed is continually increased, the number of adsorbing CO molecules exceeds the number of CO molecules removed by the oxidation and a transition to CO-covered surface occurs. At this point the catalyst activity is dependent on the rate of oxygen adsorption and its subsequent dissociation on the surface. These observations can also be explained by the following rate expression which has been found by many studies (McCarthy et al., 1975) to satisfactorily reflect the reaction mechanism of CO oxidation in air: r=
k[COI (1 + K[C0])2
(1)
At low CO ppm levels above equation reduces to r = k[CO], which implies that the reaction rate on surface
Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991 2223 100
w
80
e
.-
13
60
> c 0
u 0
60
t1
40
40
0
u
V
20
20
0
10
20
30
40
50
60
70
0
Space Velocity, L/h-mg metal Figure 4. Effect of metal loading on catalyst activity. Feed concentration, 50 ppm; T,23 O C ; RH, 0%.).( 1.8% Pt-0.2% Ir/SDB; (A) 5.4% Pt-0.6Z Ir/SDB; (A)9.0% Pt-1.0% Ir/SDB.
Feed Concentration (ppm CO) Figure 5. Eftect of water vapor on catalyst activity. SV, 420000 h-l; T,23 O C . Hydrophilic catalyet: ).( dry; (A) wet. Hydrophobic catalyst: ( 0 )dry; (A) wet.
covered predominantly by oxygen is proportional to the impingement rate of CO molecules. At higher CO ppm levels, the rate becomgs inversely proportional to CO concentration, Le., r = k / [ C O ] ,which describes that the CO oxidation reaction is inhibited by the presence of CO in the feed air. Effect of Metal Loading. To determine the effect of metal loading on the catalyst activity, the amount of noble metals (Pt-Ir) present in the catalyst sample is varied from 124 to 760 mg. The flow rate per milligram of metal present in the catalyst sample is used to compare the catalyst activity. A typical comparison at a fixed CO feed concentration is presented in Figure 4. It can be seen that the catalyst containing the lowest loading of metal (1.8% Pt-0.2% Ir) yields the best conversion of CO per unit weight of noble metal. This is reasonable because the noble metal is better dispersed at a lower loading. On the other hand, the higher the loading, the higher is the total metal area obtained. This can be seen from data given in Table I. The reaction rate was found to increase with increasing total metal area, and hence the catalyst containing the highest loading (9% Pt-1% Ir) exhibited the best catalyst activity. It should be noted that the catalyst was inactive if the total metal loading was below 1% . This may be attributed to the fact that the reaction moved into the low conversion regime before the samples were taken for analyses. Effect of Water Vapor. All the experimental results presented so far pertain to catalyst activities obtained under dry feed conditions. To determine the influence of water vapor on the catalyst activity, additional experiments were carried out with humid air. Two noble metal supported catalysts were used for the purpose: one supportsd on SDB (9% Pt-1% Ir/SDB) and the other supported on Y zeolite with similar platinum content, each representing hydrophobic- and hydrophilic-type catalysts. Figure 5 compares the performance of the two catalysts under dry and humid feed conditions. It can be seen that, in the case of a hydrophilic support, the catalyst is active under dry conditions but almost inactive in the presence of water vapor. This is due to the accumulation of capillary condensed water in catalyst pores similar to that observed by Sherwood (1980).The condensation results in increased diffusional resistance for the reactants across the liquid film covering the catalyst active sites. On the other hand,
the hydrophobic catalyst is not very active under dry conditions, but its activity increases when water vapor is introduced into the feed gas. The reason for the improvement in catalyst activity may be attributed to the change in adsorption characteristics of CO and/or O2by water adsorbed on the surface. In the caae of hydrophobic catalysts, the surface is free of adsorbed water when the dry air is used. Since adsorbed water can only be obtained through the moisture mixed in the feed, it is reasonable to assume that adsorbed water has beneficial effects for the oxidation of CO. With hydrophilic catalysts, there always exists adsorbed water and surface hydroxyl group at room temperatures and thus the catalysts showed good activity under dry feed conditions. Again, hydrophilic catalysts are not suitable for humid feeds because of capillary condensation. CO Oxidation by N20. The oxidation of CO by nitrous oxide (N20)was carried out to compare the results obtained from using O2 (air) as the oxidation agent. The reaction mechanism in both cases is similar and differs only in the way the oxidant is formed on the surface; i.e., dissociative adsorption of 0,needs two adjacent vacant sites whereas N20 needs only one site. Tests were conducted with dry and humid feed using 9% Pt-1% Ir/SDB catalyst. Before each run the catalyst was flushed with N20 (purified grade, Linde-Union Carbide) for a period of approximately 6 h. A typtical set of results is presented in Figure 6. The data show observations similar to those made during the runs with dry and humid air. In the beginning of the run, the catalyst surface was essentially covered with adsorbed N20. Addition of low ppm CO into the feed stream resulted in rapid formation of COz, confirmed by the complete conversion of CO. When the available oxidant on the surface was continually removed by the reaction, a transition from N20- to CO-covered surface occured. As a result, the catalyst exhibits low conversion. Similar to oxidation using air,oscillationswere also observed during the transition regime. Complete conversion of CO was obtained when water was introduced into the N20 stream. This is also shown in Figure 6. The similarity of the results indicate that either two adjacent vacant sites are not essential for the dissociation of adsorbed O2into atomic form or the atomic oxygen is mobile so that the availability of any two vacant sites is the only requirement for dissociative adsorption. It may be con-
2224 Ind. Eng. Chem. Res., Vol. 30,No. 9,1991
6\"
80
-
60
-
40
-
LlJ
d .4
$
z V
0
V
20 0
0
20
150
200
250
Feed Concentration (ppm CO)
-
I
01 0
100
50
I
I
I
I
I
I
5
10
15
20
25
30
I
I
35
40
eeed Concentration (ppm CO) Figure 6. CO oxidation by N20. Catalyst (30 mL), 9% Pt-1% Ir/SDB; SV, 23500 h-*; T,23 O C ; (m) dry; (A)50% RH.
cluded that the number of vacant sites available for the adsorption of oxidant is the rate-determining step at high CO concentrations. Effect of Catalyst Particle Size. Since platinum was found by X-ray analysis to deposit on the SDB support uniformly across the catalyst particle, the influence of intraparticle diffusion on catalyst activity was studied by varying the particle size. For the purpose, the catalpt (9% Pt-1% Ir/SDB) in three different sizes, 5/9,14/20,and 20/35 mesh, were tested. Figure 7a shows the conversion of CO as a function of inlet concentration under dry air conditions. The results indicate that the catalyst activity varies with the particle size. It is clear from the figure that the catalyst with smallest particle size (20/35mesh) drops to the low conversion regime faster than the other two sizes. One possible explanation is that CO tends to adsorb on the outer layer of the particle and a large portion of the platinum deposited on small particles is covered by adsorbed CO. The high CO coverage results in low catalyst activity. On the other hand, the platinum located in the interior of a large particle may be exposed to a lower CO concentration due to diffusional effects. Therefore, the rate of CO oxidation may be expressed by rate = kl[CO] + k 2 / [ C O ] (2) The first and second terms in the right-hand side of the equation represent, respectively, the internal and external contribution to the reaction. It can be concluded that small particles are better for applications involving low CO concentrations. The particle size should be increased as the feed concentration increases. The effect of feed humidity on catalyst activity for the three particle sizes was also investigated. Figure 7b shows the performance of the catalysts at a fixed inlet CO concentration by comparing the observed CO conversion as a function of amount of water vapor in the inlet air. The results obtained are in marked contrast to those obtained in the absence of water vapor. The catalyst with smallest particle size (20/35mesh) was poisoned by CO at an inlet concentration of 185 ppm. However, in the presence of water vapor, complete conversion was obtained with 25% RH in the feed. To ascertain the effect of water vapor on the catalyst activity, a test run was carried out for a period of 10 h (instead of the normal 3 h), keeping the CO concentration at 185 ppm and maintaining 25% RH in the feed. No change in catalyst activity was observed at the
1
1
40v
20
0
0
25
50
75
100
Percentage Relative Humidity, %RH Figure 7. Effect of catalyst particle size on activity. Catalyst (35 mL), 9% Pt-1% Ir/SDB; SV,92600 h-l; T,22 OC. (a) Dry conditions: (m) 5/9 mesh; ( 0 )14/20 mesh; (V) 20/35 mesh. (b) Humid conditions, 185 ppm CO in feed (m) 5/9 mesh ( 0 )14/20 mesh; (V) 20135 mesh.
end of the experiment, thus affirming the role of water vapor in improving the catalyst activity. By increasing water vapor content from 25% to 50% and higher relative humidity, a slight 'dip" in conversion was observed with particle sizes 5/9 mesh and 14/20 mesh, while no appreciable "dip" was noticed with 20/35 mesh size. Catalyst Regeneration/Reaction Studies. The catalyst, when used in practice, would encounter various environmental conditions: humidity, temperature, etc. In order to assess the behavior of the catalyst to these changes, experimental runs were carried out under the following conditions: (1) dry air reaction/dry air regeneration; (2)dry air reaction/wet air regeneration; (3)wet air reaction/dry air regeneration; (4) wet air reaction/wet air regeneration. Figure 8a shows the results obtained during dry air reaction and dry air regeneration (case 1). It can be seen that, in the absence of water vapor and at the high space velocity employed, the catalyst activity drops steadily in the f i t hour of operation. After the reaction reached the low conversion regime, the catalyst regeneration was carried out by shutting off CO feed and flushing with air for a period of 45 min. The run was repeated with the regenerated catalyst, and it was observed that the catalyst retraced the same activity profile. This suggests that the catalyst activity can be regained without using a high temperature regeneration procedure. In dry reactionlwet regeneration (case 2) experiments, the reaction was carried out under dry air conditions at a CO concentration of 50 ppm. The regeneration step involved a contact of catalyst with CO-free air at 5% and 92% relative humidity. Upon regeneration, the reaction was once again carried out under dry conditions. It was observed that the higher the relative humidity used in the
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2225 100
g
'
i
during CO oxidation was studied. Contrary to conventional noble metal supported catalysts, water vapor was found to enhance catalyst activity. Similar to CO oxidation at high CO concentrations and high temperatures, multiplicity and kinetic oscillations were also observed.
Acknowledgment 0 ' 0
I
I
1
I
I
1
I
10
20
30
40
50
60
70
I
We thank E. A. Symons, K. Marcinkowska, and W. Graham of Atomic Energy of Canada Limited for providing useful comments. This study was supported by a grant from National Sciences and Engineering Research Council of Canada.
Literature Cited 2o 0
1 20
30
50
60
0
10
0 ' 0
I
I
I
I
I
I
10
20
30
40
50
60
40
70
70
Elapsed Time, minutes
Figure 8. Catalyst regeneration/reaction results. Catalyst (30mL), 20/35 mesh 9% Pt-1% Ir/SDB; SV,420,000h-l; T, 23 OC; pressure, 515 kPa; feed concentration, 50 ppm CO. (a) Dry reaction/dry regeneration. (b) Dry reaction/regeneration with 5% RH).( and with 92% RH (A). (c) Wet reaction/dry regeneration and wet reaction/wet regeneration.
regeneration the greater was the initial activity obtained. However, after 45 min of operation the conversion dropped to a value similar to that in the dry regeneration. This is presented in Figure 8b. It seems that water adsorbed during wet regeneration momentarily improved the activity and once the adsorbed water was removed from the catalyst surface, the catalyst repeated its performance under dry air conditions. The conversion data for cases 3 and 4 are summarized in Figure 8c. Clearly, both dry and wet air regeneration were effective so long as the reaction was carried out under wet conditions. The results further emphasize the importance of water vapor in order to maintain catalyst activity at the high flow rates employed in the present investigation.
Conclusions A hydrophobic catalyst process for the effective removal of low levels of CO from air was developed. Several catalysts containing various active ingredients on different supports were examined. Pt-Ir metal supported on styrene-divinylbenzene copolymer exhibited the highest activity and is suitable for commercial applications. In the course of process development, the behavior of the catalyst
Campbell, C. T.; Ertl, G.; Kuipers, H.; Segner, J. A Molecular Beam Investigation of the Interaction of CO with a Pt (111)Surface. Surf. Sci. 1981,107, 207-219. Cant, N.W.; Hicks, P. C.; Lennon, B. S. Steady-State Oxidation of Carbon Monoxide Over Supported Nobel Metal with Particular Reference to Platinum. J. Catal. 1978,54,372-383. Chuang, K. T.; Quaiattini, R. J.; Thatcher, D. R. P.; Puissant, L. J. Development of a Waterproofed Catalyst Recombiner for Removal of Airborne Tritium. Appl. Catal. 1987,30,215-224. Conrad, H.; Ertl, G.; Kueppers, J. Interaction Between Oxygen and Carbon Monoxide on a Pt (111)Surface. J. Surf. Sci. 1978,76, 323-342. Enael, T.: Ertl, G. The Oxidation of CO on Platinum Metals. Adu. Eatal. .1979;28, 1-78. Engel, T.; Ertl, G. Oxidation of Carbon Monoxide. In The Chemical Physics of Solid Surfaces and Heterogeneous Catalvsis: King. D. A.,-Woodruff, D. P , Eds.; Elsevier: -Amsterdam, "1982; 4, Chapter 3,pp 73-93. Ford. R. R. Carbon Monoxide Adsomtion on the Transition Metals. Adu. Catal. 1970,21,51-145. Golodets, G. I. The oxidation of Carbon Monoxide. In Heterogeneous Catalytic Reactions Involving Molecular Oxygen; Roes, J. R. H., Ed.; Elsevier: Amsterdam, 1983;Chapter 10. Greek, W. P.; Dorweiler, V. P. Regulation of Carbon Monoxide: Are Current Standards Safe? Enuiron. Sci. Technol. 1990,24,32-33. Katz, M. The Heterogeneous Catalytic Oxidation of Carbon Monoxide. Adu. Catal. 1953,5,177-216. McCarthy, E.; Zahradnik, J.; Kuczynski, G. C.; Carberry, J. J. Some Unique Aspect of CO Oxidation on Supported Pt. J. Catal. 1975, 39,29-35. Norton, P. R.; Griffiths, K.; Bindner, P. E. Interaction of O2with Pt (100): 11. Kinetics and Energetics. Surf. Sci. 1984,138,125-147. Razon, L. F.; Schmitz, R. A. Intrinsically Unstable Behavior During the Oxidation of CO on Platinum. Catal. Reu.-Sci. Eng. 1986, 28 (l),89-164. Sherwood, A. E. Kinetics of Catalysed Oxidation in Air at Ambient Temperature. F'roceedings: Tritium Technology in Fission, Fusion, and Isotopic Applications, American Nuclear Society, National Topical Meeting; American Nuclear Society: Miamisburg, OH, 1980;pp 213-218. Stewart, R. The Effect of Carbon Monoxide on Humans; Annu. Reo. Pharmacol. 15, 1975,409-423. Tamhankar, S.S.;LaCava, A. I. Catalytic Oxidation of Traces of CO and H2 in the Ultra-Purification of Inert Gases. Gas Sep. Purif. 1989,3,128-132. Tucker Jr., C. W.Low Energy Diffraction Study of CO Adsorption on the (100) Face of Platinum. Surf. Sci. 1964,2,516-521.
VZ.
Received for reuiew December 18,1990 Accepted June 3,1991
