- Email: [email protected]
Catalytic ignition and extinction of hydrogen: comparison of simulations and experiments
surface science ELSEVIER
Surface Science 364 (1996) L625-L630
Surface Science L e t t e r s
Catalytic ignition and extinction of hydrogen: comparison of simulations and experiments D . G . V l a c h o s *, P.-A. Bui Department of Chemical Engineering, University of Massachusetts, Amherst, MA 01003-3110, USA Received 15 January 1996; accepted for publication 6 May 1996
Abstract
Surface ignition and extinction of hydrogen/air mixtures on platinum surfaces are modeled using a detailed surface kinetic mechanism and transport phenomena. It is shown that the platinum surface can be poisoned by different adsorbates, and the dynamic process of catalytic ignition and extinction is associated with a phase transition from one poisoning species to another. For certain temperatures, multiple poisoned states of the surface coexist. Comparison of simulations with experiments is conducted, and it is shown that the self-inhibition of hydrogen catalytic ignition is caused by poisoning of platinum by atomic hydrogen.
Keywords: Catalysis; Computer simulations; Hydrogen; Models of nonlinear phenomena; Platinum; Surface chemical reaction
Ignitions, extinctions and multiplicities in surface reactions are important in many applications, including partial oxidation reactors for chemical synthesis and catalytic removal of pollutants. These instabilities delimit the regime of operation of many industrial catalytic reactors and play a major role in reactor safety. Many years ago, some pathological cases of ignition and extinction were observed. In particular, the ignition temperature of CO in air was found to rise with composition [ 1], and the extinction temperature to drop with composition. This behavior cannot be explained using a one-step surface reaction with either a positive or a negative order in fuel kinetics. The increase of ignition temperature with inlet composition (self-inhibition) was later observed for hydrogen and olefins but * Corresponding author. Fax: + 1 413 5451647; e-mail: [email protected]
not for paraffins [2,3]. A clear understanding of the mechanism causing this self-inhibition for some fuels has not been yet achieved. Previous studies on prototype reaction systems with simple surface chemistry have shown that competition between adsorbates for catalyst sites can lead to rate multiplicities (see, for example Refs. [4-7]). These multiplicities can occur under isothermal conditions where the exothermicity of surface reactions is not a prerequisite as a feedback mechanism. Despite the fact that there are numerous studies on surface multiplicities and phase transitions [4-8], there are no studies on surface multiplicities of oxidation reactors using detailed surface kinetics. Thus, how previous results can be generalized for detailed surface reaction mechanisms is not clear. On the other hand, the advance in surface chemistry for simple adsorbates on some catalysts allows detailed modeling of these systems [9,10].
0039-6028/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved PII S0039-6028 (96)00798-4
D. G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
model a stagnation point-flow geometry wnere a premixed hydrogen/air mixture impinges on a fiat surface, as shown schematically in the inset of Fig. 1. The fluid mechanics, mass and energy conservation equations are all solved simultaneously, as described elsewhere [ 11 ]. The surface chemistry on a Pt surface is modeled through appropriate mass-transfer boundary conditions [9,12]. The inlet composition is fixed. The composition above the surface, which determines the state of the surface, is an unknown of the problem in the continuum flow regime (relatively low mean free path as compared to the characteristic dimensions of the reactor). The unknown surface composition is determined from the surface boundary condition of species. We perform simulations at atmospheric pressure and an ambient temperature of 25°C. A secondorder finite difference scheme is used to discretize the differential equations. Netwon's method and arc-length continuation techniques are employed to solve the coupled algebraic equations. A typical simulation involves ~ 103 equations and ~ 103
1200 1000 ~Freezing ! Heating T~ [K] 800 the, , ~ a l ~ 600 auto
y
Onset of
, chemistry gas-phase
unknowns and requires a few minutes on a DEC AXP 3000/600 computer for a one-parameter continuation run. Table 1 shows the steps included in the surface reaction mechanism 1-13]. Reactions follow the Langmuir-Hinshelwood mechanism. The stable product in the H2 + 0 2 reaction is H20. H2 and 02 dissociate upon adsorption to H* and O*, respectively (* denotes adsorbed species). H* and O* desorb by bimolecular association to form H2 and 02, respectively (here, we use second-order kinetics in desorption). H* reacts with O* to form OH*. Since the desorption of OH* is highly activated, OH* is preferentially converted to H20*. H20* desorbs readily at relatively low temperatures to form H20 in the gas phase. The gasphase mechanism consists of 40 irreversible reactions among nine gas-phase species (H2, 02, OH, H, O, HO2, H202, H20 and N2) [ 14]. The solid lines in Fig. 1 show the surface temperature T~ and the mole fraction Y~of H 2 at the first discretization node (i.e. at the gas-surface interface), versus the heat flux Pw provided to the surface (by resistive heating of the Pt foil) for 1% H2 in air including both gas-phase and surface mechanisms. Neither radiation nor heat losses at the back of the Pt foil are considered here because the ignition temperatures are low. At zero power, the surface temperature is equal to room temper-
~ ,
400
Table 1 H2/air surface reaction mechanism; the reaction rate constant is k=ko e -~/RTs, where R us the ideal gas constant; the units of k0 are in moles, centimeters, seconds or sticking coefficient
200 10-2 Ys H 2 10-3
10-4
i Catalytic-,~,~o \
,,,
i ....
~ ,oq\, ,
-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 Pw [cal/cm2 s] Fig. 1. Surface temperature and surface mole fraction of H z versus the heat flux for a mixture of 1% H2 in air with a strain rate of 50 s-1. The solid line indicates the coupled homogeneous-heterogeneous (HH) process with gas-phase and surface chemistry, and the circles correspond to the catalytic process alone without gas-phase chemistry. The inset is a schematic illustration of the stagnation flow reactor.
Reaction
k0
E (cal mol- 1)
H*O* ~ O H * + * OH* + *--*H* + O* H*+OH*~*+H20* • + H20* ~ H * + OH* 2OH*~O* +H20* O* + H 2 0 * ~ 2 O H * H2+2*o2H* 2H*~H2+2* 02+2*420* 20*~O2+2* H20 + * ~ H 2 0 * HzO*~H20+ * OH*--*OH + *
0.166E + 07 0.166E + 00 0.149E+09 0.299E + 05 0.166E+07 1.0 0.166E+05 0.04 0.166E+05 0.1 0.166E +05 0.249E + 05
2500 5000 15000 37000 12300 31000 0 18000 0 52000 0 10800 48000
D.G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
ature. As the power increases, the surface temperature increases almost linearly, and the reactivity of H2 is negligible. At a certain critical value of power, a turning point is found (indicated with a vertical arrow) where the system ignites. The corresponding temperature is called an ignition temperature Tiv Upon ignition, the system jumps to the ignited branch. An increase in surface temperature and a decrease in the H2 mole fraction will then be observed. In simulations, an intermediate unstable branch connects the ignited and extinguished branches. For reasons of clarity and due to the multiple curves shown in Fig. 2, all branches are drawn using the same marking. As the power increases on the ignited branch, at ,-~1000 K a change in the slope of the T~-Pw solid curve is observed. At this point, the surface mole fraction of H2 decreases sharply with the power input to the surface, indicating the onset of gas-phase chemistry (see below). As the power decreases from high values along the ignited branch, a critical point is reached where surface reactions cannot be further sustained. This
10-1
C • ~gn.
0
Onset of gas-phase ~_~mistry
10-5
".......... H;o.
,[/ 10-9
10-13 10-1 0
r
~
.~xf ~''
;:....
10-5 '
-
....
t
vacancies
. . . O. .H * "
10-9 O~"~O@ign. 10-13
, ~ , ,/ , ~ , , ,, , , ~ ,
-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 Pw[cal/cm2s]
Fig. 2. Surface coverage of species versus the heat flux for the c o u p l e d h o m o g e n e o u s - h e t e r o g e n e o u s process. The c o n d i t i o n s are the s a m e as in Fig. 1. U p o n ignition, a t r a n s i t i o n from a n H * - c o v e r e d surface to a n O * - c o v e r e d surface occurs. U p o n extinction, a t r a n s i t i o n from a n H z O * - c o v e r e d surface to an H * - c o v e r e d surface occurs.
point corresponds to an extinction indi vertical arrow. The extinction temperature l~x 1s below room temperature for these conditions, i.e. freezing of the surface is needed to extinguish surface reactions. The temperature on the ignited branch where surface reactions can be self-sustained, i.e. at zero power, is called an autothermal point. This point divides the space into the heating (Pw>0) and freezing (Pw<0) subspaces shown in Fig. 1. To examine the role of gas-phase chemistry in oxidation, simulations were performed while suppressing gas-phase reactions. The results are shown by circles in Fig. 1. Gas-phase chemistry has essentially no influence on catalytic ignition and extinction temperatures. However, deviations between the two simulations, with and without the gasphase chemistry, are observed above ~1000 K. When the mole fraction of H2 is plotted versus temperature instead of power, the change in slopes shown in Fig. 1 corresponds to an actual turning point (ignition) [ 12], similar to homogeneous ignitions we have found in earlier work [ 11,14]. Thus, this change in slope at ~ 1000 K indicates the onset of homogeneous reactions. Fig. 2 shows the coverages 0 of surface species versus the power input to the surface for the same conditions. For temperatures below T~g,the surface is covered with atomic hydrogen H*. Thermodynamically, atomic oxygen is preferred at low temperatures on Pt compared to atomic hydrogen due to the high heat of adsorption of O 2. However, the entire system is not in thermodynamic equilibrium. The high rate of adsorption of H2, due to its low molecular mass and its higher sticking coefficient compared to 02, results in preferential dissociation of H2. Since the activation energy for desorption of H* is relatively high and the temperatures are low, the fraction of free Pt sites (vacancies) is very low, and 02 cannot dissociate (an "H*-poisoned" surface). As a result, the surface reaction rates and the conversion of the fuel are low. A similar behavior has been observed in CO oxidation [8]. Upon ignition, the surface becomes covered with atomic oxygen O*, i.e. during the transient process of catalytic ignition, a phase transition from an H*-covered surface to an O*-covered surface
D.G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
transition occurs because during ignition the small fraction of H2 (1%) becomes almost completely oxidized. 02 is then in such an excess above the surface that its rate of adsorption is higher than that of H2. Consequently, 02 dissociates to O*. This transition from an H*-covered surface to an O*-covered surface is due a change in partial pressures of reactants above the surface, as determined by the surface boundary condition. Since the desorption of O* is highly activated (an activation energy of ~52 kcal mol-1), at temperatures below the homogeneous ignition temperature, the surface is partially poisoned by O*. A small fraction of vacancies (~ 2%) at intermediate temperatures allows for a small reactivity on this branch. The poisoning of the extinguished and ignited branches explains the relatively fiat curves in the Ysn2- Pw plot. On the ignited branch, as the power decreases, OH* and H20* increase at the expense of O*. Near extinction (low temperatures), H20* cannot desorb and blocks most catalyst sites (a product poisoning). During extinction, the surface converts from ( H 2 0 * + OH*)-blocked to H*-blocked. Our simulations indicate that the multiple poisoned phases can coexist at the same temperature. Outside the multiplicity regime, the surface is covered by one of the adsorbents (H* or O*), and within the multiplicity regime, the surface can be covered by one of the adsorbents, the product (H20*), and an intermediate (OH*). The above simulations are for 1% H2 in air. Next we examine the effect of composition on Tig and compare with experiments. The catalytic ignition temperature is higher than experimental data at similar compositions. Sensitivity analysis shows that the ignition temperature is affected primarily by the activation energy for desorption of H*, and the sticking coefficients of H 2 and 02 (data not shown). As an example of illustrating the role of desorption of H*, simulations were performed for three activation energies indicated in Figs. 3 and 4. Fig. 3 shows the effect of flow velocity on the catalytic ignition temperature. The solid lines are obtained from simulations using detailed chemistry by varying the strain rate (a velocity gradient outside the boundary layer [ 11 ]). For a nozzle at a fixed distance from the surface, the flow velocity
120 Ed.n,= 18 Kcal/mol 100 simulations 80
¢,
zx
6O experiments 40 I
2Oo
i
I
1
I 3
i
2
i
I
i
4
Flow velocity, U/Uo Fig. 3. Effect of flow velocity on the catalytic ignition temperature for 1% H2 in air from stagnation flow simulations (solid lines) for three values of activation energy of desorption of H* and experiments on Pt wires (points). The scaling strain rate in the simulations is 50 s- 1, and the scaling velocity in the experiments is 5 cm s- 1.
160 140 - detailed chemistry Ea.rt"= 18 Kcal/mol 120 100
1
~
17__
~ 80 60 4o
20 O~
one-st I
I
1
i
I
2
,
I
3
i
I
4
,
I
5
i
I
,
6
[%v] I-I2 Fig. 4. Effect of composition on the catalytic ignition temperature of H2. The points are experimental data on Pt wires, and the solid lines are simulations using detailed and one-step surface kinetics with a strain rate of 50 s-a.
in simulations is proportional to the strain rate. The points are experimental results obtained on Pt wires at a low Reynolds number [2]. Simulations show a slight increase in ignition temperature with increasing flow velocity. However, both experiments and simulations indicate that the effect of transport phenomena on ignition is minor for these conditions, i.e. kinetics controls catalytic ignition. A different activation energy for desorption shifts the ignition curve but does not change the qualitative response of the system.
D.G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
Fig. 4 shows the effect of composition on ignition. Experimentally, a self-inhibition of H 2 on its ignition is observed I-2,10,15]. Detailed chemistry predicts correctly this trend independently of the details of the reaction rate constants (e.g. Ed,n.) as far as the surface is blocked by H* before ignition occurs. Simulations have also been performed with a one-step surface chemistry H2+ ½ 0 2 ~ H 2 0 . The rate of the global surface reaction was taken to be first order in H2 and zero order in 02, as suggested experimentally [16]. The activation energy and pre-exponential were adjusted so that ignition happens at a positive power, and the ignition temperature is close to experiments for 1% H2 in air. The rate expression is rs=8.3× 104 e-4040/TscH2 in mols cm-2.s, where Cn: is the concentration of H 2 just above the surface in mol cm -3, and T~ is the surface temperature. The one-step surface chemistry predicts the opposite trend from that found experimentally, as shown in Fig. 4. This promotion of ignition with composition is due to the fact that ignition, in the case of a global reaction, is caused by the exothermicity of the surface reaction. As the composition increases, more heat is liberated, and thus ignition occurs at lower temperatures. Qualitatively, the promotion of ignition with composition is insensitive to the details of the parameters of the onestep kinetics. Fig. 2 indicates that catalytic ignition is controlled by the slow desorption of H* and the availability of free Pt sites. Thus, the self-inhibition of H 2 on its ignition is caused by an increase in the partial pressure of H 2 above the surface with increasing composition, which results in a more effective blocking of Pt. A higher surface temperature is then needed for desorption of H* to occur. Given the fact that the surface chemistry has been derived mainly at low pressures and is used here at atmospheric pressure, the qualitative agreement between experiments and simulations is very good. Here, we do not adjust kinetics parameters to fit the experimental data; rather, we illustrate that self-inhibition is caused by blocking of the catalyst
by the fuel or one of its fragments for m, hydrocarbons. Improvements between experiments and simulations by including adsorbate-adsorbate interactions may also be possible. In conclusion, the catalytic ignition of H2/air mixtures near Pt surfaces was studied using a stagnation point-flow model with detailed surface kinetics. We have shown that there is a small influence of transport on catalytic ignition for the regime of flow velocities investigated. The selfinhibition found experimentally is also reproduced in simulations, and has been attributed to partial poisoning of the catalyst by atomic hydrogen at low temperatures. Catalytic ignition is controlled by desorption of atomic hydrogen. Upon ignition, the surface becomes poisoned by atomic oxygen until at higher temperatures homogeneous ignition takes place. Near extinction, the surface is poisoned by the product. These surface phase-transitions occur as the temperature and/or the partial pressures of reactants (composition) change due to a competition of adsorbates for catalyst sites. Coexistence of various poisoned states by intermediates and the product have been observed for the first time using a mean-field model, and Monte Carlo studies on this subject are in progress.
Acknowledgements Acknowledgement for partial support of this research is made to the Donors of The Petroleum Research Fund, administered by the American Chemical Society, and to the Engineering Computing Services of the University of Massachusetts.
References [1] M.A.A.Cardoso and D. Luss, Chem. Eng. Sci. 24 (1969) 1699. [2] P. Cho and C.K. Law, Combust. Flame 66 (1986) 159. I-3] X. Song, W.R. Williams, L.D. Schmidt and R. Aris, in: TwentyThird Symposium(International)on Combustion (The CombustionInstitute, Pittsburgh, 1990) p. 1129. I4] C.G. Takoudis, L.D. Schmidt and R. Aris, Surf. Sci. 105 (1981) 325.
D. G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
zhos, L.D. Schmidt and R. Aris, Surf. Sci. 249 [6] L.F. Razon and R.A. Schmitz, Catal. Rev. Sci. Eng. 28 (1986) 89. [7] P. Gray and S.K. Scott, Chemical Oscillations and Instabilities, Nonlinear Chemical Kinetics (Clarendon, Oxford, 1990). [8] R.M. Ziff, E. Gulari and Y. Barshad, Phys. Rev. Lett. 56 (1986) 2553. [9] J. Warnatz, M.D. Allendorf, R.J. Kee and M.E. Coltrin, Combust. Flame 96 (1994) 393. [10] M. Fassihi, V.P. Zhdanov, M. Rinnemo, K.-E. Keck and B. Kasemo, J. Catal. 141 (1993) 438.
[11] D.G. Vlachos, L.D. Schmidt and R. Aris, Combust. Flame 95 (1993) 313. [12] P.A. Bui, D.G. Vlachos and P.R. Westmoreland, in: Twenty Sixth Symposium (International) on Combustion (The Combustion Institute, Pittsburgh, 1996) (in press). [13] W.R. Williams, C.M. Marks and L.D. Schmidt, J. Phys. Chem. 96 (1992) 5922. [14] D.G. Vlachos, Combust. Flame 103 (1995) 59. [15] H. Ikeda, J. Sato and F.A. Williams, Surf. Sci. 326 (1995) 11. [16] H. Ikeda, P.A. Libby, F.A. Williams and J. Sato, Combust. Flame 93 (1993) 138.
Surface Science 364 (1996) L625-L630
Surface Science L e t t e r s
Catalytic ignition and extinction of hydrogen: comparison of simulations and experiments D . G . V l a c h o s *, P.-A. Bui Department of Chemical Engineering, University of Massachusetts, Amherst, MA 01003-3110, USA Received 15 January 1996; accepted for publication 6 May 1996
Abstract
Surface ignition and extinction of hydrogen/air mixtures on platinum surfaces are modeled using a detailed surface kinetic mechanism and transport phenomena. It is shown that the platinum surface can be poisoned by different adsorbates, and the dynamic process of catalytic ignition and extinction is associated with a phase transition from one poisoning species to another. For certain temperatures, multiple poisoned states of the surface coexist. Comparison of simulations with experiments is conducted, and it is shown that the self-inhibition of hydrogen catalytic ignition is caused by poisoning of platinum by atomic hydrogen.
Keywords: Catalysis; Computer simulations; Hydrogen; Models of nonlinear phenomena; Platinum; Surface chemical reaction
Ignitions, extinctions and multiplicities in surface reactions are important in many applications, including partial oxidation reactors for chemical synthesis and catalytic removal of pollutants. These instabilities delimit the regime of operation of many industrial catalytic reactors and play a major role in reactor safety. Many years ago, some pathological cases of ignition and extinction were observed. In particular, the ignition temperature of CO in air was found to rise with composition [ 1], and the extinction temperature to drop with composition. This behavior cannot be explained using a one-step surface reaction with either a positive or a negative order in fuel kinetics. The increase of ignition temperature with inlet composition (self-inhibition) was later observed for hydrogen and olefins but * Corresponding author. Fax: + 1 413 5451647; e-mail: [email protected]
not for paraffins [2,3]. A clear understanding of the mechanism causing this self-inhibition for some fuels has not been yet achieved. Previous studies on prototype reaction systems with simple surface chemistry have shown that competition between adsorbates for catalyst sites can lead to rate multiplicities (see, for example Refs. [4-7]). These multiplicities can occur under isothermal conditions where the exothermicity of surface reactions is not a prerequisite as a feedback mechanism. Despite the fact that there are numerous studies on surface multiplicities and phase transitions [4-8], there are no studies on surface multiplicities of oxidation reactors using detailed surface kinetics. Thus, how previous results can be generalized for detailed surface reaction mechanisms is not clear. On the other hand, the advance in surface chemistry for simple adsorbates on some catalysts allows detailed modeling of these systems [9,10].
0039-6028/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved PII S0039-6028 (96)00798-4
D. G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
model a stagnation point-flow geometry wnere a premixed hydrogen/air mixture impinges on a fiat surface, as shown schematically in the inset of Fig. 1. The fluid mechanics, mass and energy conservation equations are all solved simultaneously, as described elsewhere [ 11 ]. The surface chemistry on a Pt surface is modeled through appropriate mass-transfer boundary conditions [9,12]. The inlet composition is fixed. The composition above the surface, which determines the state of the surface, is an unknown of the problem in the continuum flow regime (relatively low mean free path as compared to the characteristic dimensions of the reactor). The unknown surface composition is determined from the surface boundary condition of species. We perform simulations at atmospheric pressure and an ambient temperature of 25°C. A secondorder finite difference scheme is used to discretize the differential equations. Netwon's method and arc-length continuation techniques are employed to solve the coupled algebraic equations. A typical simulation involves ~ 103 equations and ~ 103
1200 1000 ~Freezing ! Heating T~ [K] 800 the, , ~ a l ~ 600 auto
y
Onset of
, chemistry gas-phase
unknowns and requires a few minutes on a DEC AXP 3000/600 computer for a one-parameter continuation run. Table 1 shows the steps included in the surface reaction mechanism 1-13]. Reactions follow the Langmuir-Hinshelwood mechanism. The stable product in the H2 + 0 2 reaction is H20. H2 and 02 dissociate upon adsorption to H* and O*, respectively (* denotes adsorbed species). H* and O* desorb by bimolecular association to form H2 and 02, respectively (here, we use second-order kinetics in desorption). H* reacts with O* to form OH*. Since the desorption of OH* is highly activated, OH* is preferentially converted to H20*. H20* desorbs readily at relatively low temperatures to form H20 in the gas phase. The gasphase mechanism consists of 40 irreversible reactions among nine gas-phase species (H2, 02, OH, H, O, HO2, H202, H20 and N2) [ 14]. The solid lines in Fig. 1 show the surface temperature T~ and the mole fraction Y~of H 2 at the first discretization node (i.e. at the gas-surface interface), versus the heat flux Pw provided to the surface (by resistive heating of the Pt foil) for 1% H2 in air including both gas-phase and surface mechanisms. Neither radiation nor heat losses at the back of the Pt foil are considered here because the ignition temperatures are low. At zero power, the surface temperature is equal to room temper-
~ ,
400
Table 1 H2/air surface reaction mechanism; the reaction rate constant is k=ko e -~/RTs, where R us the ideal gas constant; the units of k0 are in moles, centimeters, seconds or sticking coefficient
200 10-2 Ys H 2 10-3
10-4
i Catalytic-,~,~o \
,,,
i ....
~ ,oq\, ,
-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 Pw [cal/cm2 s] Fig. 1. Surface temperature and surface mole fraction of H z versus the heat flux for a mixture of 1% H2 in air with a strain rate of 50 s-1. The solid line indicates the coupled homogeneous-heterogeneous (HH) process with gas-phase and surface chemistry, and the circles correspond to the catalytic process alone without gas-phase chemistry. The inset is a schematic illustration of the stagnation flow reactor.
Reaction
k0
E (cal mol- 1)
H*O* ~ O H * + * OH* + *--*H* + O* H*+OH*~*+H20* • + H20* ~ H * + OH* 2OH*~O* +H20* O* + H 2 0 * ~ 2 O H * H2+2*o2H* 2H*~H2+2* 02+2*420* 20*~O2+2* H20 + * ~ H 2 0 * HzO*~H20+ * OH*--*OH + *
0.166E + 07 0.166E + 00 0.149E+09 0.299E + 05 0.166E+07 1.0 0.166E+05 0.04 0.166E+05 0.1 0.166E +05 0.249E + 05
2500 5000 15000 37000 12300 31000 0 18000 0 52000 0 10800 48000
D.G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
ature. As the power increases, the surface temperature increases almost linearly, and the reactivity of H2 is negligible. At a certain critical value of power, a turning point is found (indicated with a vertical arrow) where the system ignites. The corresponding temperature is called an ignition temperature Tiv Upon ignition, the system jumps to the ignited branch. An increase in surface temperature and a decrease in the H2 mole fraction will then be observed. In simulations, an intermediate unstable branch connects the ignited and extinguished branches. For reasons of clarity and due to the multiple curves shown in Fig. 2, all branches are drawn using the same marking. As the power increases on the ignited branch, at ,-~1000 K a change in the slope of the T~-Pw solid curve is observed. At this point, the surface mole fraction of H2 decreases sharply with the power input to the surface, indicating the onset of gas-phase chemistry (see below). As the power decreases from high values along the ignited branch, a critical point is reached where surface reactions cannot be further sustained. This
10-1
C • ~gn.
0
Onset of gas-phase ~_~mistry
10-5
".......... H;o.
,[/ 10-9
10-13 10-1 0
r
~
.~xf ~''
;:....
10-5 '
-
....
t
vacancies
. . . O. .H * "
10-9 O~"~O@ign. 10-13
, ~ , ,/ , ~ , , ,, , , ~ ,
-0.2-0.1 0.0 0.1 0.2 0.3 0.4 0.5 Pw[cal/cm2s]
Fig. 2. Surface coverage of species versus the heat flux for the c o u p l e d h o m o g e n e o u s - h e t e r o g e n e o u s process. The c o n d i t i o n s are the s a m e as in Fig. 1. U p o n ignition, a t r a n s i t i o n from a n H * - c o v e r e d surface to a n O * - c o v e r e d surface occurs. U p o n extinction, a t r a n s i t i o n from a n H z O * - c o v e r e d surface to an H * - c o v e r e d surface occurs.
point corresponds to an extinction indi vertical arrow. The extinction temperature l~x 1s below room temperature for these conditions, i.e. freezing of the surface is needed to extinguish surface reactions. The temperature on the ignited branch where surface reactions can be self-sustained, i.e. at zero power, is called an autothermal point. This point divides the space into the heating (Pw>0) and freezing (Pw<0) subspaces shown in Fig. 1. To examine the role of gas-phase chemistry in oxidation, simulations were performed while suppressing gas-phase reactions. The results are shown by circles in Fig. 1. Gas-phase chemistry has essentially no influence on catalytic ignition and extinction temperatures. However, deviations between the two simulations, with and without the gasphase chemistry, are observed above ~1000 K. When the mole fraction of H2 is plotted versus temperature instead of power, the change in slopes shown in Fig. 1 corresponds to an actual turning point (ignition) [ 12], similar to homogeneous ignitions we have found in earlier work [ 11,14]. Thus, this change in slope at ~ 1000 K indicates the onset of homogeneous reactions. Fig. 2 shows the coverages 0 of surface species versus the power input to the surface for the same conditions. For temperatures below T~g,the surface is covered with atomic hydrogen H*. Thermodynamically, atomic oxygen is preferred at low temperatures on Pt compared to atomic hydrogen due to the high heat of adsorption of O 2. However, the entire system is not in thermodynamic equilibrium. The high rate of adsorption of H2, due to its low molecular mass and its higher sticking coefficient compared to 02, results in preferential dissociation of H2. Since the activation energy for desorption of H* is relatively high and the temperatures are low, the fraction of free Pt sites (vacancies) is very low, and 02 cannot dissociate (an "H*-poisoned" surface). As a result, the surface reaction rates and the conversion of the fuel are low. A similar behavior has been observed in CO oxidation [8]. Upon ignition, the surface becomes covered with atomic oxygen O*, i.e. during the transient process of catalytic ignition, a phase transition from an H*-covered surface to an O*-covered surface
D.G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
transition occurs because during ignition the small fraction of H2 (1%) becomes almost completely oxidized. 02 is then in such an excess above the surface that its rate of adsorption is higher than that of H2. Consequently, 02 dissociates to O*. This transition from an H*-covered surface to an O*-covered surface is due a change in partial pressures of reactants above the surface, as determined by the surface boundary condition. Since the desorption of O* is highly activated (an activation energy of ~52 kcal mol-1), at temperatures below the homogeneous ignition temperature, the surface is partially poisoned by O*. A small fraction of vacancies (~ 2%) at intermediate temperatures allows for a small reactivity on this branch. The poisoning of the extinguished and ignited branches explains the relatively fiat curves in the Ysn2- Pw plot. On the ignited branch, as the power decreases, OH* and H20* increase at the expense of O*. Near extinction (low temperatures), H20* cannot desorb and blocks most catalyst sites (a product poisoning). During extinction, the surface converts from ( H 2 0 * + OH*)-blocked to H*-blocked. Our simulations indicate that the multiple poisoned phases can coexist at the same temperature. Outside the multiplicity regime, the surface is covered by one of the adsorbents (H* or O*), and within the multiplicity regime, the surface can be covered by one of the adsorbents, the product (H20*), and an intermediate (OH*). The above simulations are for 1% H2 in air. Next we examine the effect of composition on Tig and compare with experiments. The catalytic ignition temperature is higher than experimental data at similar compositions. Sensitivity analysis shows that the ignition temperature is affected primarily by the activation energy for desorption of H*, and the sticking coefficients of H 2 and 02 (data not shown). As an example of illustrating the role of desorption of H*, simulations were performed for three activation energies indicated in Figs. 3 and 4. Fig. 3 shows the effect of flow velocity on the catalytic ignition temperature. The solid lines are obtained from simulations using detailed chemistry by varying the strain rate (a velocity gradient outside the boundary layer [ 11 ]). For a nozzle at a fixed distance from the surface, the flow velocity
120 Ed.n,= 18 Kcal/mol 100 simulations 80
¢,
zx
6O experiments 40 I
2Oo
i
I
1
I 3
i
2
i
I
i
4
Flow velocity, U/Uo Fig. 3. Effect of flow velocity on the catalytic ignition temperature for 1% H2 in air from stagnation flow simulations (solid lines) for three values of activation energy of desorption of H* and experiments on Pt wires (points). The scaling strain rate in the simulations is 50 s- 1, and the scaling velocity in the experiments is 5 cm s- 1.
160 140 - detailed chemistry Ea.rt"= 18 Kcal/mol 120 100
1
~
17__
~ 80 60 4o
20 O~
one-st I
I
1
i
I
2
,
I
3
i
I
4
,
I
5
i
I
,
6
[%v] I-I2 Fig. 4. Effect of composition on the catalytic ignition temperature of H2. The points are experimental data on Pt wires, and the solid lines are simulations using detailed and one-step surface kinetics with a strain rate of 50 s-a.
in simulations is proportional to the strain rate. The points are experimental results obtained on Pt wires at a low Reynolds number [2]. Simulations show a slight increase in ignition temperature with increasing flow velocity. However, both experiments and simulations indicate that the effect of transport phenomena on ignition is minor for these conditions, i.e. kinetics controls catalytic ignition. A different activation energy for desorption shifts the ignition curve but does not change the qualitative response of the system.
D.G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
Fig. 4 shows the effect of composition on ignition. Experimentally, a self-inhibition of H 2 on its ignition is observed I-2,10,15]. Detailed chemistry predicts correctly this trend independently of the details of the reaction rate constants (e.g. Ed,n.) as far as the surface is blocked by H* before ignition occurs. Simulations have also been performed with a one-step surface chemistry H2+ ½ 0 2 ~ H 2 0 . The rate of the global surface reaction was taken to be first order in H2 and zero order in 02, as suggested experimentally [16]. The activation energy and pre-exponential were adjusted so that ignition happens at a positive power, and the ignition temperature is close to experiments for 1% H2 in air. The rate expression is rs=8.3× 104 e-4040/TscH2 in mols cm-2.s, where Cn: is the concentration of H 2 just above the surface in mol cm -3, and T~ is the surface temperature. The one-step surface chemistry predicts the opposite trend from that found experimentally, as shown in Fig. 4. This promotion of ignition with composition is due to the fact that ignition, in the case of a global reaction, is caused by the exothermicity of the surface reaction. As the composition increases, more heat is liberated, and thus ignition occurs at lower temperatures. Qualitatively, the promotion of ignition with composition is insensitive to the details of the parameters of the onestep kinetics. Fig. 2 indicates that catalytic ignition is controlled by the slow desorption of H* and the availability of free Pt sites. Thus, the self-inhibition of H 2 on its ignition is caused by an increase in the partial pressure of H 2 above the surface with increasing composition, which results in a more effective blocking of Pt. A higher surface temperature is then needed for desorption of H* to occur. Given the fact that the surface chemistry has been derived mainly at low pressures and is used here at atmospheric pressure, the qualitative agreement between experiments and simulations is very good. Here, we do not adjust kinetics parameters to fit the experimental data; rather, we illustrate that self-inhibition is caused by blocking of the catalyst
by the fuel or one of its fragments for m, hydrocarbons. Improvements between experiments and simulations by including adsorbate-adsorbate interactions may also be possible. In conclusion, the catalytic ignition of H2/air mixtures near Pt surfaces was studied using a stagnation point-flow model with detailed surface kinetics. We have shown that there is a small influence of transport on catalytic ignition for the regime of flow velocities investigated. The selfinhibition found experimentally is also reproduced in simulations, and has been attributed to partial poisoning of the catalyst by atomic hydrogen at low temperatures. Catalytic ignition is controlled by desorption of atomic hydrogen. Upon ignition, the surface becomes poisoned by atomic oxygen until at higher temperatures homogeneous ignition takes place. Near extinction, the surface is poisoned by the product. These surface phase-transitions occur as the temperature and/or the partial pressures of reactants (composition) change due to a competition of adsorbates for catalyst sites. Coexistence of various poisoned states by intermediates and the product have been observed for the first time using a mean-field model, and Monte Carlo studies on this subject are in progress.
Acknowledgements Acknowledgement for partial support of this research is made to the Donors of The Petroleum Research Fund, administered by the American Chemical Society, and to the Engineering Computing Services of the University of Massachusetts.
References [1] M.A.A.Cardoso and D. Luss, Chem. Eng. Sci. 24 (1969) 1699. [2] P. Cho and C.K. Law, Combust. Flame 66 (1986) 159. I-3] X. Song, W.R. Williams, L.D. Schmidt and R. Aris, in: TwentyThird Symposium(International)on Combustion (The CombustionInstitute, Pittsburgh, 1990) p. 1129. I4] C.G. Takoudis, L.D. Schmidt and R. Aris, Surf. Sci. 105 (1981) 325.
D. G. Vlachos, P.-A. Bui/Surface Science 364 (1996) L625-L630
zhos, L.D. Schmidt and R. Aris, Surf. Sci. 249 [6] L.F. Razon and R.A. Schmitz, Catal. Rev. Sci. Eng. 28 (1986) 89. [7] P. Gray and S.K. Scott, Chemical Oscillations and Instabilities, Nonlinear Chemical Kinetics (Clarendon, Oxford, 1990). [8] R.M. Ziff, E. Gulari and Y. Barshad, Phys. Rev. Lett. 56 (1986) 2553. [9] J. Warnatz, M.D. Allendorf, R.J. Kee and M.E. Coltrin, Combust. Flame 96 (1994) 393. [10] M. Fassihi, V.P. Zhdanov, M. Rinnemo, K.-E. Keck and B. Kasemo, J. Catal. 141 (1993) 438.
[11] D.G. Vlachos, L.D. Schmidt and R. Aris, Combust. Flame 95 (1993) 313. [12] P.A. Bui, D.G. Vlachos and P.R. Westmoreland, in: Twenty Sixth Symposium (International) on Combustion (The Combustion Institute, Pittsburgh, 1996) (in press). [13] W.R. Williams, C.M. Marks and L.D. Schmidt, J. Phys. Chem. 96 (1992) 5922. [14] D.G. Vlachos, Combust. Flame 103 (1995) 59. [15] H. Ikeda, J. Sato and F.A. Williams, Surf. Sci. 326 (1995) 11. [16] H. Ikeda, P.A. Libby, F.A. Williams and J. Sato, Combust. Flame 93 (1993) 138.