Langmuir 1985,1, 488-495
488
Unimolecular Reactions of NO, N20, NO2, and NH3 on Rh and Ptt G. A. Papapolymerou and L. D. Schmidt* Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 55455 Received December 28, 1984. In Final Form: March 25, 1985 The decomposition kinetics of NO, NzO, NOz, and NH, on polycrystallineRh and Pt and their inhibition by Ozand H2 are examined and compared in a differential flow reactor at pressures between 1 X and 3.0 torr and temperatures between 600 and 1800 K. It is shown that the rates of all four reactions on both metals and at all pressures and temperatures can be fit quantitativelywith simple Langmuir-Hinshelwood (LH) unimolecular rate expressions with an accuracy of *15% under all conditions. N20and NH3decompose exclusively to Nz,while NO2decomposes only to NO. NO decomposes to N2,while NzO is also produced with a rate from 2% at 1800 K and up to equal that of Nz production at 900 K (on Rh). At temperatures above 1300 K the NO decomposition rate is faster on Rh than on Pt by up to 2 orders of magnitude. Above 800 K the rate of NOz decomposition is faster than any of the other rates measured here, becoming flux limited above 1100 K with a reaction probability of 0.06. Inhibition studies showed that small amounts of Ozcoadsorbed on Rh and Pt with either NO or NzO severely inhibited NO decomposition. Oxygen inhibition is stronger on Rh, while hydrogen inhibition is stronger on Pt. It was also found that the surface of Rh undergoes faceting and pitting under reaction conditionswhich changes the surface area approximately by a factor of 2, while Pt remains smooth.
Introduction While rates of most homogeneous unimolecular reactions can generally be described quantitatively through the Lindemann mechanism (RRKM theory), no analogous descriptions of heterogeneous reactions are universally accepted. It is obvious that surface reactions should be more complex because they do not occur by isolated collisions of molecules and because they involve averaging over adsorbed complexes, binding states, and crystal planes. There is one simple model of surface reactions: the Langmuir-Hinshelwood (LH) mechanism. This assumes a single binding state with coverage-independent parameters (a Langmuir isotherm) and that the adsorbed reactant reacts as a unimolecular event through the steps A, S G+A, (1)
+
B,
-
B,
+S
(3)
to yield (4) where k R is the reaction rate coefficient kR
= k R o exp(-ER/RT)
and KA the adsorption equilibrium constant KA = k , / k d = KAOexp(EA/RT)
(5) (6)
In these expressions eA is the coverage (fraction of saturation density) of reactant A, E , and EA are reaction activation energy and heat of adsorption, respectively, and KRO and KAo are preexponential factors for reaction and adsorption respectively. Equation 4 assumes that step 2 is slow compared to steps 1and 3 and that B is not strongly or competitively adsorbed. A number of studies of unimolecular reactions on nobel metals have shown that many rates can be fit quite well +This work partially supported by NSF under Grant DMR82126729.
by the LH expression. Examples include NH3 on polycrystalline Pt, Fe, Ni, and W,1-3 NH3 on several crystal planes of Pt,4 and NO on Pt.596 It should be noted that only by obtaining steady-state rates over wide ranges of temperature and pressure can unique fits of rate expressions and the temperature dependences of rate parameters be determined. This requires measurements a t sufficiently high pressures that steady states are obtained with reactant coverages from zero to saturation and also low reactant conversions (below 10%) so that differential rates are obtained without product inhibition. These requirements usually prevent kinetics on supported catalysts from being analyzed in this detail. Situations where formation of compounds such as oxides and nitrides occur should not be applicable to the LH mechanism, because the LH mechanism assumes one monolayer coverage. Therefore nobel metals should be the best surfaces to test the LH mechanism. To test the LH model, g a m must be sufficiently pure and catalyst surfaces must be clean and remain free of contaminants throughout experiments. (1) (a) Loffler, D. G.; Schmidt, L. D. J. Catal. 1976,41, 440. (b) Loffler, D. G.; Schmidt, L. D. J. Catal. 1976,44,244. (2)McCabe, R. W. J. Catal. 1983,79, 445. (3)Grosman, M.; Loffler, D. G. J. Catal. 1983,80,188. (4)Loffler, D. G.; Schmidt, L. D. Surf. Sci. 1976,59, 195. (5)Amirnazmi, A.; Boudart, M. J. Catal. 1976,39,383. (6)Mummey, M. J.; Schmidt, L. D. Surf. Sei. 1981,109,29. (7)Takoudis, C. G.; Schmidt, L. D. J. Catal. 1983,80,274. (8)Avery, N. R. Surf. Sci. 1983,131,501. (9)Cambell, C. T.;White, J. M. Appl. Surf. Sci. 1978,1, 347. (10)Root, T.W.; Schmidt, L. D.; Fisher, G. B. Surf. Sci. 1983,134,30. (11)Gorte, R. J.; Schmidt, L. D.; Gland, J. L. Surf. Sci. 1981,109,367. (12)Ho,P.;White, J. M. Surf. Sci. 1984,137,103. (13)Gland, J. L.; Sexton, B. A. Surf. Sci. 1980,94,355. (14)Ertl, G.; et al. Surf. Sci. 1981, 107, 220. (15)Nieuwenhuys, B. E. Surf. Sci. 1983,126,307. (16)Fisher, G. B.; Schmieg, S. J. J. Vac. Sci. Technol., A 1983,l(2), 1064. (17)Norton, P. R.; Richards, P. J. Surf. Sci. 1974,41,293. (18)Procop, M.; Volter, J. Surf. Sci. 1972,33,69. (19)McCabe, R. W.; Schmidt, L. D. h o c . Znt. Vac. Congr., 7th, 1977. (20)Gorodetakii, V. V.; et al. Surf. Sci. 1981,108,225. (21)Gland, J. L.; Kollin, E. B. J. Vac. Sci. Technol. 1981,18(2),604. (22)Yates, J. T.;Thiel, P. A; Weinberg, W. H. Surg. Sci. 1979,84,427. (23)Eisenberg, R.; Hendriksen, D. E. Adu. Catal. 1979,28,79. (24)D.+lgren, D.; Hemminger, J. C. Surf. Sci. 1982,123,L739. (25)Pirug, G.;Bonzel, H. P. J. Catal. 1977,50,64.
0743-7463/85/2401-0488$01.50/0 0 1985 American Chemical Society
Langmuir, Vol. 1, No. 4, 1985 489
Unimolecular Reactions
The major objective of this study is to compare reactions of several different molecules containing nitrogen and to determine differences between Pt and Rh. In this paper we examine in detail the unimolecular reactions of NO, N20,and NH, on Rh and of NO2on Pt and Rh. Rates of other reactions have been reported previously, and we have repeated experiments on these reactions in our system to compare their rates. Inhibitions by O2 and H2 are also examined on both metals. In a future paper we will examine several bimolecular reactions involving these species. The possible surface reactions of these speties are
-
NO2
NO2 -+ N20 N20
NH3 -+ '/& + 3/2H2 AH0298 = +10.9 kcal/mol
+ '/202 m 0 2 9 8 = +13.5kcal/mol '/2N2 + O2 AH0298 = -8.09kcal/mol N2 + 1/202 AH0298 = -l9.5kcal/mol NO + 1/2N2 AH0298 = +2.1 kcal/mol NO
NO -+ '/2N2O + ' / 4 0 2 AH0298 = -2.1 kcal/mol
(7) (8)
(8a) (9) (9a)
(loa)
Below 750 and 450 K the decompositions of NO2 and NH3 to form NO and N2, respectively, have positive AGO values (at 1atm), while those of NO and N 2 0 decompositions have negative AGOvalues a t all temperatures. The equilibrium constant K is smaller than 1for the NO2and NH3 decompositions baow 750 and 450 K, but otherwise greater than 1. In experiments reported here (0.010-10 OR and less than 10% conversions) equilibrium conditions should not be approached for any reactants except for NO2 a t low temperatures. Experimental Section Reactions were carried out in a 400-cm3, six-way crow stainless steel reactor. Upstream and downstream valves controlled the flow rate and pumping speed, respectively. Gases were pumped out of the reactor by a pechanical pump with a cold trap (dry iceacetpne or ice-water for the decomposition of NOz) to reduce hydrocarbon back streaming. The base pressure in the reactor was abdut torr. Total reactor pressures between and 10 torr were measured by a capacitance manometer. Partial pressures of gases were measured by leaking gases into a quadrupole mass spectrometer system to a pressure of lo4 torr from a base presswe of lo+' torr. Reactor partial pressures were calibrated against mass spectrometer readings by passing known mixtures of gases through the reactor. Mass spectrometer signals were found to be proportional to partial pressures at all pressures. Cracking of NO, N 2 0 and NH3 on the hot tungsten filament of the mass spectrometer produced about 10% backgrounds of their reaction products. NOz cracked up to 25% on the mass spectrometer f i i e n t , and rates of NOz consumption were calibrated by comparing with those of NO and O2 production. Reactant flow rates were determined by measuring the rate of pressure decrease from flasks of known volume. Reactor residence times were adjusted from 1 to 10 s. The major advantages of this reactor geometry are a n accurately known catalyst area, ease of variation of reactor pressure, flow rate and catalyst temperature, and, most important, the applicability of the mixed reactor mass balance equation for determining absolute reaction rates: (11)
In eq 11, ri is the rate per unit area of consumption or formation of species i (in molecules/(cm2s), F the volumetric flow rate (L/s),
AP,the change in partial pressure of species i between reactor and feed conditions (torr), No Avogadro's number, A , the wire area (cm2),T the gas temperature (300 K), and R the gas constant (torr L/(mof K)). Reactant species concentrations in the reactor are uniform because of large gas diffusivities at these low pressures, and conversions were maintained below 10% to obtain differential rates. No transients or steady-state multiplicity were observed in any experiments, and steady states were established within 1-3 s after the temperature of the wire had attained a desired value. A series of molecular sieves and activated charcoal traps immersed in dry ice-acetone baths was used to further purify the NzO and NO (99.9% and 99.5% minimum purity, respectively). NH3 (99.99%) was passed through a trap a t -35 "C and N O p (99.5%) was used without further purification. We found that NOp reacted very readily with activated charcoal to produce a colorless gas ( N p ,CO, and Cop).At atmospheric pressure all of the NOz decomposed in the charcoal trap. The reactions of N O p with carbon are very exothermic and heated the activated charcoal trap to above 200 "C, a potentially dangerous situation. The catalysts were resistively heated 99.999% purity Pt and Rh wires of 0.025- or 0.013-cm diameter and lengths from 3 to 15 cm, which corresponds to areas between 0.12 and 1.2 cm2. Several wires of different area were used to obtain the data for the decomposition of each gas, and 4-5 different wires were used for each reaction to check reproducibility of data. The wires were spot welded to heavy Ni leads and the temperature was measured by Pt-13% Rh thermocouples with an estimated precision of *2%. All data were reproducible on each surface and on different samples and over long periods of time to within & E % . Studies in reactors containing to determine surface cleanliness have shown that, after heating either Pt or Rh for 2 h at 1700 K in 1 X torr of 02,the surfaces are free of contaminants, with rates identical with those measured here. In these experiments the wires were heated for 2-3 h at 1800 K and in 0.100 torr of 02.Since O2 is one of the products in the decomposition reactions of the oxidizing species N20, NO, and NOz, surfaces exposed to these gases should be continuously cleaned of contaminants (mainly carbon and sulfur). Wires were also heated in O2 each day before beginning rate measurements. Results Figures 1-5 show rates vs. surface temperature of N2 production from the NH,, NO, and N20 and of NO production from the NO2 decomposition reactions. Solid curves are fits to the data using eq 4. NH,, NO, and N20 decompositions on Pt have been reported previ~usly.'~~~' We have duplicated those results and will compare rates in later sections. Figure 6 shows the rates of all four reactions on Rh vs. the partial pressures of the corresponding gases at various temperatures. At high temperatures all rates are first order in reactant pressure and all became pressure independent a t sufficiently low temperatures, as can be seen in Figures 1-6. Solid curves in all figures are from LH rate expressions indicated in the text. NHP The rate of NH, decomposition on Rh increases rapidly until 800-900 K and then more slowly up to 1800 K (Figure 1). The reaction probability (rate/reactant flux) at 1800 K is 0.005 and reaction is not flux limited. Figure 6 shows that at 800 K the rate is still first order in pressure below 0.500 torr, but at 600 K it becomes zeroth order in pressure above 0.100 torr. From the fit of the data we obtained the following rate expression: 8.9 x 10'' eXp(-4230/RT)P~~$ rN2
=
1
+ 3.1 X
exp(16790/R7')PNH,
(12)
Rates calculated from this expression are indicated by the solid curves in Figures 1, 6, and 7. It is evident that all (26)Hasenberg, D.M.;Schmidt, L. D. Surf. Sci., in press. (27) Fisher, G . B. Chem. Phys. Lett. 1981, 79 (3), 452.
Papapolymerou and Schmidt
490 Langmuir, Vol. 1, No, 4, 1985 l
0
l
Pressure 9 In7Torr
10’~
A
r
10l8
I
IO‘*
‘R
1
10”
‘R
molecules 1o’6
t 1015
u 600
1000
1014
1400
Figure 1. Plot of reaction rate rR for N2 formation vs. surface temperature for NH, decomposition on polycrystalline Rh. The solid curves are rates calculated using the Langmuir-Hinshelwood model, eq 12. 1ole
Pressure in Torr = 3 0
1100
1500
T( K )
T(K)
Figure 3. Plot of rR for N2 formation vs. surface temperature for N 2 0 decompositionon polycrystalline Rh. Solid curves are calculated from eq 14.
,
1017
‘R
‘R
molecules
molecules
10’6
(Xd
1015
10’~
T(K)
Figure 2. Plot of rR for Nz formation vs. surface temperature for NO decomposition on polycrystalline Rh. Solid curves are calculated from eq 13. The data indicated by triangles are rates of NzO formation at 0.200 torr by bimolecular processes.
data points agree with this expression to within *15%. NO. Decomposition of NO to Nzand O2 occurs readily on Rh, and at high temperatures this rate is faster by up to 2 orders of magnitude than on Pt. The reaction probability approaches 0.002 at 1700 K. The rate expression for the NO decomposition on Rh is 1.2 X lozoexp(-17 580/RT)PNo (13) rNz = 1 + 2.7 X exp(18180/RT)PNo The triangles in Figure 2 are measured rates of N 2 0 production from the NO decomposition at 0.200 torr. Rates of N20 formation as low as 1 x 1014molecules/(cm2 s) could be measured in this system. On Rh the rate of N 2 0 formation is higher than that of N2 formation below 1000 K, but it is less than 10% of the rate of N2 a t tem-
T(K)
Figure 4. Plot of rR for NO formatisn vs. surface temperature for NOz decomposition on polycrystalline Rh. Solid curves are calculated from eq 17.
peratures above 1200 K. N20 is also produced in the bimolecular reactions of NO + H2 and of NO NH, on both Pt and Rh, having roughly the same dependence on temperature but with a rate of formation from 1to 2 orders of magnitude higher than that from NO decomposition. In a later paper we shall describe N20 production from bimolecular reactions in greater detail. Here we note that the rate of N20 formation by bimolecular processes from NO decomposition is small on Rh and Pt. Since rates of N 2 0 production are small and could be associated with bimolecular reactions with H2 and CO contaminants, we do not include them in the analysis of NO unimolecular decomposition. N20. N20 decomposes on Pt and Rh to produce N2and O2 exclusively. The reaction probability on Rh at 1800 K is 0.006 and is not flux limited as indicated by the rapidly rising rate with temperature shown in Figure 3. No NO
+
Unimolecular Reactions
Langmuir, Vol. 1, No. 4, 1985 491
1O2O lola
I P N ~ )= 0 20 Torr
‘R
molecules
‘R
(-)
1
lo”
f
10’6
1015
10’5
000
T(K)
1200
1600
T(K)
Figure 5. Plot of rR for NO formation vs. surface temperature for NOz decomposition on polycrystalline Pt. Solid curves are calculated from eq 17 with Pt parameters.
Figure 7. Hydrogen inhibition of rRvs. surface temperature for NH3decomposition on polycrystalline Rh at partial pressures of NHBand Hz as indicated. Solid curves are calculated from eq 19.
Schmidt’ for the N20 decomposition on Pt. We obtained a slightly moaified rate expression for this reaction on P t 2.7
1019
rNz
‘R
)
10”
10‘6
N20 1000 K
1014
10-2
16’
100
1
X
10l8 e~p(-i530O/RT)P~,~
+ 1.2 x
eXp(30780/RT)P~,o
(15)
This expression gives rates approximately a factor 3-7 lower than those reported previously, and we believe that this expressibn is an accurate representation of the N20 decomposition on Pt rather than the expression reported previ~usly.~ NOP. The r.ate of NO2 decomposition is very fast on both mtals above 1000 K and becomes flux limited above 1100-1300 K, with a reaction probability of 0.06 as shown in Figures 4 and 5. The decomposition of NO2 is almost identical on Pt and Rh. Below 900 K the rate becomes zero order in pressure. If a unimolecular reaction is flux limited, then the forward rate in eq 1 is comparable to the reaction rate eq 2. However, the form of eq 4 is still preserved with KA now being a steady-state parameter rather than the adsorption equilibrium constant
1018
molecules (cm2-sec
=
10’
P (Torr)
Figure 6. Plot of rRfor Nz and NO formation vs. reactant partial pressures at 1700, 1000, and 800 K on polycrystalline Rh. Solid lines are calculated from the LH expressions eq 12, 13, 14, and 17. Reactions and temperatures are as indicated.
was detected during N20 decomposition a t any temperatures on either Pt or Rh, with a minimum limit of detectability of about l X l O I 4 molecules/(cm2 s). Figure 6 shows that on Rh ht high temperatures the rate of N20 decomposition is also first order in pressure, but it becomes independent of pressure a t relatively high temperatures. The following rate expression was obtained from the fit of the data: 1.5 X rNz
=
1
+ 1.1 x
e~p(-20860/RT)P~,~ eXp(31730/RT)P~,o
(14)
We have repeated the experiments of Tokoudis and
In the high-temperature limit the rate becomes rR = k,PA if kR >> kd. Since ka = So/(2~MRT,)1/2, rR should be nearly independent of surface temperature as long as So does not depend strongly on temperature. The following rate expressions were obtained for the NO2 decomposition on Rh: 1.5 x 1 0 1 9 ~ ~ ~ ~ rNO
=
1 -k 2.2 x
lo4
eXp(30280/RT)P~o,
(17)
On Pt the rate equation is basically the same except that slightly different fits were obtained with KAO and EA of 8.5 X lo* and 27 700 cal/mol, respectively. NO2 decomposes almost exclusively to NO. Some N2 is produced, but above 1100 K its production rate is no more than 0.5% that of NO and can be accounted for entirely as arising from decomposition of the product NO.
Papapolymerou and Schmidt
492 Langmuir, Vol. I , No. 4, 1985 1018
c
I
I
t 10'' 'R
(*
'R
t
tL
10'6
c
t
L
"
1000
1400
1100
T(K)
Figure 8. Oxygen inhibition of rR vs. surface temperature for N20 decomposition on polycrystalline Pt at partial pressures of N20 and O2as indicated. Solid curves are calculated from eq 15, 18, and 21.
A t lower temperatures (C1100 K) no Nz is detected because both the rate of N2 production (from the decomposition of NO) and the partial pressure of the product NO are low. On Pt no N2 was detected at comparable NO partial pressures because the decomposition rate of NO is lower than that on Rh by a factor of 10-100 at temperatures above 1100 K. Other side reactions of NOz and NO that may be occurring involve homogeneous or heterogeneous formation of NO3, N203,N204,and N20k We observed none of these species or their fragments in our mass spectrometer. Dimerization to N204was also shown to be negligible as expected from thermodynamic equilibrium calculations (PNo2/PN204 = 100 and a t T = 25 "C). Cracking of NO2 on the stainless steel walls of the reactor did not occur to a measurable extent. This was ascertained by admitting NO2into the reactor and shutting both inlet and outlet valves. In a typical experiment the NOz was allowed to remain in the reactor for about 5 min at a pressure of 0.500 torr. No change in pressure could be detected (pressure change less than 0.2%),and when the wire was heated iii this batch system the pressure increased to 0.750 torr within 15-20 s as predicted by eq 8 and then it remained constant. O2 and HzInhibition. For all rates measured here we used conversions less than 10% and for NO decomposition less than 3% in order to keep product concentrations low and prevent product inhibition. Competitive adsorption of products will of course affect rates. In the LH model B should be given by the the rate for the process A expression
-
1500
T(K)
Figure 9. Oxygen inhibition of r R vs. surface temperature for N20 decomposition on polycrystalline Rh at partial pressures of N20 and O2as indicated. Solid curves are calculated from eq 14, 18, and 22. adding H2 to NH3 and O2 to NO and N20 to determine how they inhibit these unimolecular reactions. Comparable studies of H2 and N2 inhibition of NH3decomposition on Pt and O2inhibition of NO decomposition on Pt have been reported previously.'V6 NH3 + H2 on Rh. Figure 7 shows the effect of adding H2 to NH3 on Rh for NH3pressures of 0.200 and 0.050 torr. Without H2 the rate is given by eq 12, but Hzdecreases the rate significantly. However, inhibition becomes less pronounced ljelow 800 K and the rates a t various partial pressures of H2 approach the rate when PH2 = 0.00 torr with decreasing temperdture. This indicates that on Rh, H2 is less strongly bound than NH3. On Pt, H2 inhibited the decomposition rate more strongly at low temperatures.' We were able to fit the rates quantitgtively in the presence of H2using the LH form of eq 18 to yield rNH, = 8.9 x 10" exp(4230/RT)PNH8/[1 + 3.1 x exp(l6 ~ ~ O / R T ) P N H3.5~ X lo-' ~ X P ( ~ ~ ~ O / R T ) P H , ~ / ~ ] (19) Note that reaction and KNHBterms remain identical in the presence of H2 A similar fit was obtained by Ldffler and Schmidt1 for NH3 on Pt. They obtained an inhibition of KHFH,3l2= 9.8 x
lo4
eXp(27 ~ ~ O / R T ) P H ?(20) /~
for that system. We also found that the best fit was obtained for inhibition proportional to PW3l2 (m = 3/2) rather
than m = ' I 2 or 1, which are expected for simple dissociative or nondissociative adsorption. Possible mechanisms leading to m = f were discussed by Loffler and Schmidt.' N 2 0 O2on Pt. Figure 8 shows the effect of adding O2 to NzO on Pt for N20 partial pressures of 0.200 and 0.050 torr. Below 1100 K the decomposition of N20 becomes leas inhibited by 02,indicating that N20adsorption is stronger than O2 adsorption on this surface. The best fit we obtained for the O2 inhibition term was KoZpo2= 9.5 x exp(18700/RT)Po, (21)
+
where KB is the adsorption equilibrium constant for species B, analogous to K Agiven by eq 6. We insert an exponent m in the equation to account for possible dissociative adsorption ( m = 1/2) or other forms of inhibition, although in a simple nondissociative situation we expect m = 1. Molecular nitrogen does not chemisorb on Pt or Rh, but H2,02,and NO do. NO cannot be added independently because it decomposes, but we have studied the effects of
with other terms identical with those without O2 (eq 15). Atomic oxygen might be expected to inhibit as Po2'I2.
Langmuir, Vol. 1, No. 4,1985 493
Unimolecular Reactions Table I. Comparison of Rate Parameters Rh reaction NHS -+ Nz + H2 NO -+ NZ + 02 NzO -+ Nz + 0 2 NO2 NO + O2
kRO, molecules/
-
(cm2s) 2.9 x 1023 4.4 x 1023 1.4 X 10% 6.8 X lou
Pt
ER,
kcal/mol 21.0 35.8 52.6 30.3
KAO,
torr-’
3.1 X lo” 2.7 X LO“‘ 1.1 X
2.2
X
EA, kcal/mol 16.8 18.2 31.7
IO4
kRO, molecules/
(cm2s) LO x 1023
8.0 x 1019 2.2 x 1024 1.9 x 1024
ER,
kcal/mol 20.9 13.5 36.1 27.7
KAO,
torr-’
4.4 X
7.0 X 1.2 X IO4 8.5 X IO4
EA,
kcal/mol 16.7 8.2 30.8
Inhibition 3.5 x 10-1 1.6 X lo-’ 5.1 X 10”
‘R molecules
10l6
10’5
1500
1100
T(K )
Figure 10. Oxygen inhibition of rR vs. surface temperature for NO decomposition on polycrystalline Rh a t partial pressures of NO and O2as indicated. Solid curves are calculated from eq 13, 18, and 23.
However, the fit with m = i / 2 was found to be inferior to that for m = 1. N20 O2on Rh. Figure 9 shows the effect of adding O2 to N 2 0 on Rh for N20 partial pressures of 0.200 and 0.050 torr. Oxygen strongly inhibits the N20 decomposition, particularly a t low temperatures. Below 1500 K the rates of N 2 0decomposition at the various partial pressures of O2 become parallel to that corresponding to pure NzO, indicating that competition between coadsorbed O2 and N20 is similar. The following best fit for the O2inhibition was obtained:
+
Ko,Po2 = 5.1 X
exp(26800/RT)Po,
(22)
As on Pt, assumption of m = 1yielded better agreement with data than m = ’(> NO O2on Rh. Figure 10 shows the effect of adding O2to NO on Rh for NO pressures of 0.200 and 0.050 torr. We note that very small amounts of O2strongly inhibit the NO decomposition. Addition of only 3% of O2 to NO decreases the rate by 25% a t 1700 K and by 70% at 1200 K. The fit of eq 18 gave for the O2 inhibition term
+
KoY0, = 1.6 X
lo-’ exp(20700/RT)Po,
(23)
Mummey and Schmidt6 obtained a similar fit for NO on Pt. They obtained an inhibition term of Kelp,, = 1.56 exp(9550/RT)Po, (24)
9.8 X IO* 1.6 9.5 X
5.1 20.7 26.8
18.5 9.5 18.7
In both cases m = 1agreed well with inhibition data, while m = ‘I2 did not. Morphology of the Rh Surface. The surface morphology of the Rh wires was found to change during reaction. After heating in O2 and in oxygen-containing gases for 3-4 h and above 1300 K, surfaces were observed to develop a dull appearance. Reaction rates were also observed to increase by about a factor of 2 as Rh wires aged. No such changes in appearance or reaction rates occurred on Pt. On Rh pits and grooves form a t grain boundaries and pyramidal protrusions begin to grow and attain a size of about 2 pm. Our observations suggest that different gases seem to produce these protrusions a t different rates, the highest being for O2 followed by NO2, N20,NO, and NH3 In the bimolecular reaction of NO with H2, the protrusions grew very rapidly, within 1/2 h a t a pressure of 0.200 torr. The protrusions on each grain are single crystals of a different orientation than the grain, but this change of morphology of the Rh wires appears to affect reaction kinetics only as an area change, since neither the activation energies nor the heats of adsorption appeared to be altered. The only affect of this surface activation was to increase all rates by about a factor of 2. This was ascertained by making complete runs of the rate vs. temperature a t low pressures, then activating the wires, repeating the runs, and comparing rates. The geometrical area change (surface area of a pyramid/base area) was estimated from SEM pictures to be from 1.8 to 2.2, consistent with the rate changes. All rates reported here were taken on activated Rh wires, but rates are reported on the basis of the initial wire area.
Discussion Reaction Parameters. The quantitative agreement of all rates with LH unimolecular expressions (solid curves in Figures 1-6) shows that this is an excellent model to fit unimolecular reaction rates, and it suggests that the mechanism implied (adsorption-desorption equilibrium and coverage-independent parameters) is a reasonable description of these processes. Therefore the quantities of kR should be related to barrier crossing events, and those in KA should be related to adsorption and desorption of species A. The reaction rate coefficient kR in eq 5 should be kR = k m eXp(-&/RT) = vonoeXp(-ER/Rn (25) where uo is a frequency of reaction barrier impingement, ER is the barrier height, and no is the saturation reactant density. Assuming vo = 10l2and no = 1015molecules/cm2, we expect kRO 10’‘ to be the “normal” reaction preexponential. From Table 1we see that kROvaries from 1019 to loz6molecules/(cm2 8). All values are thus below those predicted by a vibrational frequency but consistent with
-
Papapolymerou and Schmidt
494 Langmuir, Vol. 1, No. 4 , 1985
c
IOi4
r
600
1000
1400
T(K)
Figure 11. Comparison plot of rR vs. surface temperature for all four decomposition reactions on both Rh and Pt at reactant partial pressure of 1.00 torr. Solid lines are calculated from LH rate expression (4) with parameters from Table I. reductions in k, due to steric factors, different mobilities in intermediate states, and other statistical mechanical theories of unimolecular reaction rates. Note that the size of kROhas no correlation with the magnitude of rR. Also there is no “compensation effect” (linear relationship between kR and E R ) in kR. Measured reaction activation energies shown in Table I vary from 13.5 to 53 kcal/mol and all are larger than the heats of adsorption determined by reaction from KA. (If E R were less than EA,the rate would go through a maximum and then decrease with increasing temperature.) The adsorption equilibrium constant should be
(26) from which the preexponential K A should ~ be approximately IO4 torr-’ if So = 1 and vdo = 1012s-l. Values in torr-l, again in Table I range from 1 X lo* to 7 X reasonable agreement with simple theories and consistent with possible variations in soand vd@ I t is interesting to compare heats of adsorption E A obtained from LH fit of reaction data to heats of chemisorption (desorption activation energies) of these gases obtained under UHV conditions by the use of TPD techniques. Heats of adsorption obtained in this study from LH fib of reaction rate data range from 8 to 32 kcal/mol. For NH, the heat of adsorption has been reported to be 8.6 kcal/mo121v28for the weakly bound state and between 16 and 23 kcal/moP1 for the more strongly bound states, and the measured value of 17 kcal/mol in these reactions is essentially identical with the heat of chemisorption of NH3. Vavere and H a n ~ e obtained n~~ a reaction activation (28) Sexton, B. A.; Mitchell, G. E. Surf. Sci. 1980, 99, 523. (29) -. , Vnvere. - .. , A.:, Hansen. . -. R. S. J. Catal. 1981. 69. 158. (30)Daniel, W. M.; et ai. Surf. Sci. 1981, I l i , 189. (31) Gland, J. L. Surf. Sci. 1978, 71, 327. ~
energy of 19 kcal/mol for the decomposition of NH, on Rh(lOO), -(llO), and -(111). The value of 21 kcal/mol obtained in this study is therefore in good agreement. Adsorption of N20 on Pt(ll1) has been studied by Avery et al. using EELS.* They estimated a heat of adsorption of approximately 6 kcal/mol, and no decomposition of NzO was detected. The values were obtained from reaction, 31 and 32 kcal/mol, are thus much higher, and we find that N 2 0 decomposes readily on both metals. The heats of adsorption of NO on clean Rh and Pt are 26 and between 25 and 33 kcal/mol, respectively.”12*wThe reaction value of 18 kcal/mol on Rh is consistent with this, although somewhat lower, but on Pt the reaction value of 8 kcal/mol is much lower. We repeated the results of Mummey and Schmidt and obtained good agreement with those parameters. The reaction preexponential factor for NO on Pt is also low, and parameters for this system are qualitatively different from the other reactions. Comparisons of Reactants and Metals. Figure 11 shows a comparison of all eight rates vs. temperature on Rh and Pt for reactant pressures of 1.0 torr. Rates go from 1014molecules/ (cm2s), the minimum detectable rate, to almost 1020 molecules/(cm2 s), the mass-transfer limit. This corresponds to reaction probabilities from below to nearly 0.10, and each reaction rate varies over many orders of magnitude as the temperature increases. NO2 decomposition is identical on both metals and, as noted previously, it becomes nearly independent of temperature in the first-order regime because the reaction is flux limited. Even in the low-temperature zero-order regime the rate is essentially identical on Rh and Pt. However, if we account for the increase in area of Rh (rates are calculated from the initial area of the wires), the rate should be approximately twice as high on Pt. NH, decomposition is also nearly identical on Pt and Rh, being a factor of 2 higher on Rh than on Pt at all temperatures. By accounting for the area increase on Rh, NH, decomposition is virtually identical on Pt and Rh. Also note that the rate of NH3 decomposition is higher than any of the others between 500 and 850 K. N 2 0decomposition differs by up to a factor of 6 between Rh and Pt, with curves crossing at about 1250 K at 1torr. The temperature dependence of NzO decomposition on Rh is the largest of any of these reactions. NO decomposition on Rh also has a much larger temperature dependence than on pt, and the rate a t high temperatures is up to 100 times faster on Rh. However, NO decomposition on Pt is faster than on Rh below 950 K. It is also evident that low-temperature behaviors of NO and NzO are quite different from those at high temperatures. N 2 0 on Rh has the lowest rate below 1000 K, followed by NO decomposition on Rh and N 2 0 decomposition on Pt. The cause of this is the high activation energies measured for these three reactions as Table I shows. On the other hand NO decomposition on Pt is highest below 450 K and the measured activation energy is 13.5 kcal/mol. Product Inhibition. Table I shows values of KB and EBfor product inhibition (species B) of NH, by H2 and Experimental values of KBo deviate of NO and N20 by 02. considerably from the expected lo4 torr-l, and for 02,EB values deviate from experimental heat of adsorption of 42-51 kcal/mol on On Rh(ll1) and at high coverages the heat of adsorption of O2 is 24.5 kcal/mol,1° while Fisher16 found that on Rh(100) and at coverages above half a monolayer two binding states for O2exist with heats of adsorption of 50 and 62 kcal/mol. While these could be representative of more weakly bound states, we suggest that these values have less fundamental signifiPt.113’4315
Unimolecular Reactions cance than do reactant heats of adsorption. This uncertainty arises mainly from inaccuracies in fitting these parameters and questions of the order of the inhibitor pressure m in eq 18. However, the solid curves in Figures 7-10 show that these expressions still give a quantitative representation of the inhibitions. While NH3 decomposition is nearly identical on Rh and Pt, H2 inhibition is much stronger on Pt. Hydrogen inhibits the NH3 decomposition differently on Pt than on Rh. On Pt H2 inhibits the rate very strongly below 1300 K, but not a t all above 1500 K.' On Pt the rate curves approach that corresponding to PH2= 0.0 torr above 1300 K, while on Rh this happens below 800 K (Figure 7). This is attributable simply to the different strengths of H2 adsorption (Table I) on Pt and Rh. Heats of adsorption of H2 are reported as 16-17 kcal/ mol on Pt foils and wires.17J8 McCabe and Schmidtlg obtained 17 and 13 kcal/mol for the heats of adsorption of H2 on Pt(ll1) and -(110) and from 12 to 27 kcal/mol for the four different binding states of H,on Pt(100). Two weakly bound states of 9 and 10 kcal/mol were also found on P t ( l l 0 ) and -(loo). Nieuwenhuys et report that the heat of adsorption of H2on a Rh filament and a t low coverages is about 18 kcal/mol and decreases to 11 kcal/mol a t saturation, while Yates et a1.22reported an initial EA of 19 kcal/mol for Hz on Rh(ll1). The value therefore of 18 kcal/mol (Table I) on polycrystalline Pt wire is in good agreement with above data, while the value of 5 kcal/mol for Rh is much lower. This low value of EA for H2 desorption from Rh may be indicative of a weakly bound state with coadsorbed NH3 affecting H2desorption. NO decomposition is much more strongly inhibited by O2 on Rh than on Pt. A t 1300 K with 0.200 torr of NO and 0.020 torr of O2the rate is inhibited by a factor of 11 on Rh, while on Pt it is only inhibited by a factor of 2.3. N20 is more strongly inhibited by O2on Rh, particularly below 1600 K, but the difference between metals is less than for NO. Also note that on both metals NO is inhibited by O2much more strongly than is N20, and that O2 inhibits the decomposition of both NO and N 2 0 more strongly on Rh than on Pt. Bonding. Relative reactivities have no simple relationship with heats of adsorption of the reactants. NO has the highest measured heat of adsorption but is quite unreactive, while N 2 0 appears to adsorb only physically on Pt(111)8s30 but reacts readily at high temperatures in these experiments. It is interesting to speculate on bonding properties and intermediates that can lead to decomposition. NH3 adsorbs through the lone pair of electrons on the N atom (18 kcal/mol). Once NH, dissociates, continued dissociation and atom recombination to N2 and H2are the only energetically favored processes. In low-pressure, low-temperature experiments NH3 on Pt is fairly unreactive with all of the NH3 desorbing without dissociation, although some dissociation occurs above 600 K.21927p28 The free radical NO adsorbs on transition metals through a 40 bond on the N atom and a 2a* back-bond, analogous to bonding of CO. However, NO favors bridge bonding and bent configurations more than does CO, and it is probable that bent and lying-down configurations are necessary for unimolecular d i s s o c i a t i ~ n In . ~ ~low-pressure experiments NO desorbs from Pt(ll1)ll with no dissocia-
Langmuir, Vol. 1, No. 4, 1985 495 tion and only -50% reacts on Pt(100). On Rh(ll1) NO dissociates completely a t low coverages and -55% a t saturation.1° On Rh(100) and at saturation -54% of NO dissociates.12 N20 is a very stable gas molecule and apparently does not chemisorb or decompose on transition metals in lowpressure experiments.8 It has been shown by EELS8 that N 2 0 bonds through the terminal N atom and inclined at an angle to the surface estimated at 35'. Adsorption occurs by donor bond formation without significant back-donation into the unoccupied 3a antibonding orbital. However, decomposition probably requires a lying-down configuration, and the N-N and N-O bonds may favor side-bonded adsorption of N20 far more than for NO. A molecular precursor state may be kinetically important for NzO decomposition under UHV c o n d i t i o n ~ . ~ ~ NO2is quite reactive, although it is doubtful that, being a nonlinear molecule (bond angle of 135')) nondissociative forms would be stable. The unpaired electron may be shared in a resonance structure between the N and 0 atoms of the NO2molecule. Clearly the N-0 bonds in NOz are more readily attacked at transition-metal surfaces with NO and 0 being the primary adsorbed species.24 In contrast to CO,, NO2appears to adsorb and dissociate readily.
Summary We believe that these rate expressions are accurate representations of these reactions on polycrystalline Pt and Rh. All rates have been reproduced on many wires and foils, and for NO on Pt and NH3 on Rh, surfaces have been shown to be clean by AES. The rate expressions presented here should be accurate to within f15% for pressures and 10 torr and temperatures between between 5 X 500 and 1800 K. These expressions should therefore be useful in predicting behavior in complex gas mixtures as long as inhibitors or species capable of undergoing bimolecular reactions are not present. We note that only in the pressure range used here is accurate measurement possible; higher pressures introduce mass-transfer limitations, and at lower pressure the rates become too small to be detectable. A remarkable feature of unimolecular surface reactions on transition metals is that they can be fit accurately by Langmuir-Hinshelwood rate expressions. Thus, in spite of crystallographic differences, multiple binding states, and multiple-step reaction processes, averaging of these effects produces rates that do not deviate significantly from the LH model a t any temperature or pressure. Most or all of these reactions must proceed by a sequence of steps, and many species should be coadsorbed (NH, and NO,). However, in each there appears to be a single rate limiting step so that in the absence of 0 and H inhibitors the rates are given quantitatively by the LH model. It is also evident from these experiments that in many cases the rate parameters kRO,ER,KAo,and EA are in reasonable agreement with expected values, although in some cases these parameters clearly differ from these values. Registry No. Rh, 7440-16-6; Pt, 7440-06-4; NO, 10102-43-9; NZO, 10024-97-2;NOz, 10102-44-0; NH,, 7664-41-7;0 2 , 7782-44-7; Hz,1333-74-0.
