Langmuir 1991, 7, 2999-3005
2999
Recombination of 0, N, and H Atoms on Silica: Kinetics and Mechanism Young C. Kim and Michel Boudart' Department of Chemical Engineering, Stanford University, Stanford, California 94305 Received March 20,1991. In Final Form: May 20, 1991
The probability of recombination, y, was measured for recombination of 0, N, and H atoms on pure silica between 700 and 1250 K (HT),300 and 700 K (MT),and at 194 K (LT). The rate of recombination is first order with respect to the atom concentration from LT to HT. The Arrhenius plots, y vs 1/T, are very complex. All observations are explained by assuming a surface with a small fraction of active sites that irreversibly bind chemisorbed atoms. Everything happens as if the active sites are surrounded by collection zones within which all atoms striking the surface are adsorbed reuersibly with an assumed sticking probability of unity. These atoms then diffuse on the surface. Some of them reach the active sites where they can recombine with the chemisorbed atoms. At LT, all atoms striking the surface reach the active sites. As a result of desorption at MT, the collection zones shrink with increasing temperature. At HT, only atoms striking active sites directly from the gas phase lead to recombination. All data were fitted to an analytical solution of the diffusion-reaction equation obtained for a model where the active sites are distributed uniformly. A new result of this work is the identification of the active sites for recombination of H on silica at HT as surface hydroxyl groups.
Introduction The rate of recombination of oxygen,l-15 and hydrogen2932 atoms on various surfaces has been studied by many workers using a variety of experimental techniques for measuring atom concentration in the gas phase. Recently, the surface recombination of oxygen and nitrogen atoms has attracted new attention as it affects
* Author to whom correspondence should be sent.
(1) Voevodski, V. V.; Lavrovskaya, G. K. Dokl. Akad. Nauk, USSR 1948,63,151. (2)Linnett, J. W.; Marsden, D. G. H. Proc. R. SOC. London, A 1956, A234,489. I31 Kaufman. F. - J. Chem. Phvs. 1958.28.352. (4)Herron, J. T.; Schiff, H. 1: Can. J.' &em. 1958,36, 1159. (5)Elias, L.;Ogryzlo, E. A.; Schiff, H. I. Can. J. Chem. 1959,37,1680. (6)Greaves, J.%;;Linnett, J. W. Trans. Faraday SOC. 1959,55,1346. (7)Greaves, J. C.; Linnett, J. W. Trans. Faraday SOC. 1959,55,1355. (8)Krongelb, S.;Strandberg, M. W. P. J.Chem. Phys. 1959,31,1196. (9)Elias, L.; Morgan, J. E.; Schiff, H. I. J. Chem. Phys. 1960,33,930. (IO) Mavroyannis, C.; Winkler, C. A. Can. J. Chem. 1961,39, 1601. (11) Marshall, T. C. Phys. Fluids 1962,5,743. (12)Berkowitz-Mattuck, J. B. In The Structure and Chemistry of Solid Surface; Somorjai,G. A., Ed.; John Wiley & Sons: New York, 1969; pp 60-81. (13)Hacker, D. S.;Marshall, S. A.; Steinberg, M. J.Chem.Phys. 1961, 35, 1788. (14)Melin, G.A.;Madix, R. J. Trans. Faraday SOC. 1971,67,198. (15)Greaves, J. C.;Linnett, J. W. Trans. Faraday SOC.1958,54,1323. (16)Wentink, T.; Sullivan, J. 0.; Wray, K. L. J.Chem. Phys. 1958,29, 231. (17) Herron, J. T.; Franklin, J. L.; Bradt, P.; Dibeler, V. H. J. Chem. Phys. 1959,30,879. (18)Young,R. A. J. Chem. Phys. 1961,34,1292. (19)Marshall, T. C. J. Chem. Phys. 1962,37, 2501. (20)Sancier, K. M.; Fredericks, W. J.; Hatchett, J. L.; Wise, H. J. Chem. Phys. 1962,37,860. (21)Kretschmer, C. B.; Petersen, H. L. J. Chem. Phys. 1963,39,1772. (22)Sancier, K. M. J. Chem. Phys. 1965,42,1240. (23)Evenson, K. M.; Burch, D. S. J. Chem. Phys. 1966,45,2450. (24)Wise, H.;Wood, B. J. In Adoanced in Atomic and Molecular Physics; Bates, D. R., Estermann, I., Eds.; Academic Press: New York, 1967;Vol. 3. (25)Kaufman, F. Progress in Reaction Kinetics; Pergamon Press: New York, 1969; Vol. 1,p 1. (26)Rahman, M. L.; Linnett, J. W. Trans. Faraday SOC. 1971,67,170. (27)Rosner, D.E. NASA Contract. Rep. 1973,CR-134124. (28)Smith, W. V. J. Chem. Phys. 1941,11, 110. (29)Wood, B.J.; Wise, H. J. Phys. Chem. 1958,29,1416. (30)Green, M.; Jenning, K. R.; Linnett, J. W.; Schofield, D. Trans. Faraday SOC. 1959,55,2152. (31)Tsu, K.; Boudart, M. In Actes du Congres International de Catalyse, 2nd; Technip: Paris, 1961;Vol. 1, p 539. (32)Melin, G. A.;Madix, R. J. Trans. Faraday SOC. 1971,67,2711. \-,
.
the performance of heat shields on reusable space vehicle^.^^-^* In the past, space vehicles have reentered into the atmosphere in free fall. The Space Shuttle Orbiter (SSO), however,was designed to travel repeatedly between the surface of the earth and outer space. Upon high-speed reentry through the atmosphere, a shock wave in front of the vehicle dissociates 02 and N2 molecules into 0 and N atoms. A fraction of these atoms recombine on the thermal insulation surfaces of the SSO,generating heat. This heat of recombination is responsible for up to half of the increase in SSO surface temperature during reentry. High temperature requires thicker insulation tiles, thus reducing the payload capacity of the vehicle. A coating material less active toward atom recombination allows the use of thinner tiles. The coating material on the insulation tiles currently used on the SSO is a reaction-cured glass (RCG) made of silica (94%), B203 (4%), and SiB4 (2%).33 In this paper we report measurements of the recombination probability, y (the probability that an atom impinging to the surface will recombine), of oxygen, nitrogen, and hydrogen atoms on a silica surface. The emphasis of the present work was to obtain reliable values of y over a large temperature range so as to elucidate the elementary steps involved in the surface recombination. With added understanding, such a study might assist the design of new coating materials for SSO vehicles.
Experimental Section The diffusion tube technique first described by Smith28has been used successfully30~31 for measuring the apparent recombination probability, YA (=-yF, where the subscript A denotes an apparent value and F is the roughnessfactor of the surface of the tube). In our apparatus, shown in Figures 1 and 2, atoms generated in a microwave discharge diffused down a cylindrical tube closed at one end. These atoms recombined on the tube (33)Goldstein, H. E.;Leiser, D. B.; Katvala, V. US.Patent40 093 771 1978. (34)Stewart, D. A.;Rakich, J. V.; Lanfranco, M. J. AIAA Pap. 1981, 81-1143. (35)Rakich, J. V.; Stewart, D. A.; Lanfranco, M. J. AIAA Pap. 1982, 82-0944. (36)Zoby, E. V.; Gupta, R. N.; Simmonds, A. L. AIAA Pap. 1984, 84-0224. (37)Howe, J. T.J. Spacecr. Rockets 1985,22, 19. (38)Scott, C.D. J. Spacecr. Rockets 1985,22,489.
0 1991 American Chemical Society
Kim and Boudart
3000 Langmuir, Vol. 7, No. 12, 1991 Storage Vessels
"2
U
Mn0'si02 Molecular Sieve Trap Metering Valve
m To precision
Double metering Valve
To Reactor
Discharge Vessel
I
m
I -
f
To Vacuum
HB U
Figure 1. Gas purification and storage system. Ple**"ra
Double Metering
atom recombination on the thermocouple probe raised the probe temperature. It has been shownzsthat if the reaction is first order with respect to the gas-phaseatom concentration, the temperature rise of the probe, AT, drops exponentiallywith distance X from the atom generator along the tube AT = constant X (exp(-uX/Ro))
(1)
u = [UR0y*/D]"2
(2)
where
Thsrmocouole
vacuum
GPlgB
Figure 2. Gas delivery, pressure control, and diffusion reactor. wall with the apparent recombination probability YA. The depletion of atoms due to wall recombination down the tube was measured with a Pt/Pt-10% Rh thermocouple probe traveling along the axis of the tube. When the discharge was turned on,
and u is the gas kinetic velocity of the atoms, Ro the radius of the diffusion tube, and D the diffusion coefficientof atoms in the gas phase. Values of D in eq 2 were calculated as described by Hirschfelder et al.39 The collision cross sections of atoms and molecules were provided by Margenau40and Yun and Mason.41 The apparent recombination probability of atoms impingingto the surface was obtained fromthe slope of straight line in (In AT) vs (X/Ro) plot. Details of the experimental system shown in Figures 1and 2 can be found elsewhere.42The diffusion reactor was made of a fused quartz tube (General Electric, type 204, 99.98% silica), referred to as silicain what follows. Atubular furnace with three independent heating zones was used to maintain constant temperature. The systemwas evacuated to 10" mbar by oil diffusion and mechanical pumps. Traps cooled at liquid nitrogen temperature prevented backstreaming of pump fluids. A copperwire trap placed before the pumps prevented atoms from accumulating in the high-pressure chamber of the mechanical pump, where they could react violently with pump oil. The silica tube (1.8 cm i.d. and 2.1 cm 0.d.) was introduced in the diffusion tube (2.2 cm i.d.) and was evacuated to 1o-B mbar for a day. All gases (Liquid Carbonic, NZand Hz 99.999%, 02 (39) Hirschfelder, J. 0.; Bird, R. B.; Curtiss, C. F. Molecular Theory
of Gases and Liquids; Wiley: New York, 1954; p 539.
(40) Margenau, H. Phys. Reo. 1944,66, 303. (41)Yun, K. S.; Mason, E. A. Phys. Fluids 1960,5, 380. (42) Kim, Y.C. Ph.D. Thesis, Stanford University, 1991.
Langmuir, Vol. 7, No. 12, 1991 3001
Recombination of 0, N, and H Atoms on Silica T I K 2000 1000
1°.li., 10-2
500
300
200
2
5
'
,no\ 0 8
9 17
15 $2-1
12
0
8 \ 0
16
Figure 3. Plot of recombination probability, y, versus 1/T for hydrogen atom recombination: 0 ,this work on silica (filled data at HT were used for fitting purpose); 0,Tsu and Boudart on Pyrex (ref 31). 99.995% purity) were further purified by passage through molecular sieve traps at DryIce-acetone temperature (194 K). The reaction pressure was maintained at 0.27 mbar. A microwave discharge powered by a microwave unit (Raytheon,PGM 10x1) surrounded a section of the diffusion tube. The atoms produced by the discharge in that section of the reactor diffused down the tube and recombinedon the wall. Beforethe dischargewas turned on, the temperature of the probe, T, was constant along the experimentalsection of the diffusion tube within 2 K. The value of AT defined in eq 1 is the temperature of the probe T' when the discharge is turned on, minus T. After the atom generator was allowed to run for 0.5 h at room temperature (RT), AT was recorded to obtain YA at RT. Then the reactor was heated to the highest experimental temperature (1250 K) and YA was obtained. As reaction temperature was decreasedto RT, values of y~ at other temperatureswere collected. The value of - y measured ~ at RT was the samewithin experimental error as that before the high-temperatureexperiments. A typical sequence of chronologically numbered experiments is shown in Figure 3. Experiments at temperatures below RT were carried out in a diffusion tube in a constanttemperature bath containing ice or DryIceacetone. The value of YA at RT was the same before and after low-temperature experiments. To estimatethe surface roughness factor, F, of the silica tube, YA was measured on a series of modified silica tubes containing various amounts of pure silica powder (Cab-0-Sil,Grade PTG, 99.8% silica)on the inner wall of the tube. About 1g of the silica powder, which has a specific surface area, 200 m2 gl, was stirred in 100 cm3 of deionized water for 1h. About 10 cm3 of mixture was introduced inside the silica tube, and the sample tube was dried at RT in air. Tubes with different amount of silica powder were prepared in this way. Measurement of YA was carried out on each sample tube after evacuating at lo+ mbar for a day and after running the atom generator for 0.5 h.
Results For all runs, the plots of In A T vs X / R o gave straight lines42a t all temperatures investigated. This justifies the first order assumed in deriving eq 1. Figure 4 shows that the value of YA for nitrogen recombination increases with the amount of silica powder added onto the tube surface. The increase of the roughness factor due to the addition of silica powder, Fp,was calculated as the ratio of the surface area of silica powder introduced to the geometric
50
100
150
200
Figure.4. Linear increaseof yA with surface roughness: nitrogen recombination at RT; F roughness factor due to silica powder, (~SBET)/~?TR&O; Ro,ra&s of tube; LO,length of tube; m,weight of silica powder introduced, g; SBET BET surface area of silica powder, m2g-l. surface area of the tube, 2rRoL0,where Ro and LOare radius and length of the silica tube, respectively. The measured value, YA, increases linearly with increasing Fp
(3) = rF + rFp up to Fp 70 (Figure 4). For Fpvalues higher than -70, diffusion of atoms to the external surface of the tube starts to limit the rate of recombination. According to eq 3, the slope and the intercept of the straight line in Figure 4 are y and yF, respectively. The roughness factor of the tube and y for nitrogen recombination are 2.4 and 2.5 X 10-4, respectively. The above estimatian is based on the assumption that the true recombination probability is identical for both the silica tube and the powder. Bikermana reported a roughness factor of glass varying between 1.6 and 5.4. de Boer also estimated about 3 for glass,44 and Tsu and Boudart3I estimated about 3.8 for Pyrex. The value of 2.4 obtained in this work appears reasonable and has been used to calculate y from YA. The values of y for recombination of H, N, and 0 atoms are shown in Figures 3, 5, and 6, respectively. The results will now bediscussed first for N and 0 atoms, then for H atoms.
-
YA
Discussion The observed first-order kinetics can be explained if recombination occurs between an atom on the surface and an atom in the gas phase (Eley-Rideal mechanism). But to explain the complex temperature dependence of 7 , it is assumed as in previous work31that the surface is covered with a small fraction of active sites, denoted by *, which by definition hold chemisorbed atoms irreversibly. The chemisorbed atoms A* recombine with atoms A arriving at the active sites. In addition to the direct arrival of atoms from the gas phase, atoms reversibly adsorbed on the rest of the surface diffuse along the surface to the active sites. Here is a proposed sequence of elementary (43) Bikerman, J. J. Surface Chemistry, 2nd ed.; Academic Press: New York, 1958; p 195. (44) de Boer, J. H. The Dynamic Character of Adsorption; Oxford University Press: London, 1953.
Kim and Boudart
3002 Langmuir, Vol. 7, No. 12, 1991 Energy
T / K
I
4
10’2
1
Site
10.6
0
3
2
1
4
5
6
T-’/ K-’ Figure 5. Plot of recombination probability, y, versus 1/T for nitrogen atom recombination on silica: 0 , this work on silica (filleddata at HT were used for fitting purpose); 0,Marshall (ref 19); +, Rosner (ref 27). T / K j200;
lppp,
I
5po
,
2po
3po
,
Ir.’j 10’2
0
0
0
10’
0
1
2
3
4
5
6
T”/ K-’ Figure 6. Plot of recombination probability, y, versus 1 / T for oxygen atom recombination: 0 , this work on silica (filled data at HT were used for fitting purpose); 0,Berkowitz-Mattuck on Vycor (ref 12); - - -,flight experiment on reaction cured glass (ref 34);-, flight experiment on reaction cured glass (ref 36).
steps for adsorption, desorption, surface diffusion, and recombination at the active sites: A + * A* irreversible adsorption at active site * A A, reversible adsorption anywhere else A, A, surface diffusion A, A desorption A, + A* A, + * Langmuir-Hinshelwood (LH) recombination A + A* A, + * Eley-Rideal (ER) recombination Active sites on the surface are replenished with irreversibly
-----
(45) Kruyer, S. Koninkl. Ned. Akad. Wetenschap. Proc. 1953, B56, 274. (46) Gomer, R. In Surface Mobilities on Solid Materials; Binh, V. T., Eds.; Plenum Press: New York, 1983; p 7. (47) Lewis, B.;Anderson, J. C. Nucleation and Growth of Thin Films; Academic Press: New York, 1978. (48) Barrer, R. M. Br. J . Appl. Phys. 1954, Suppl. 3, 41 and 49.
Langmuir, Vol. 7, No. 12, 1991 3003
Recombination of 0, N , and H Atoms on Silica Active Site
Table 11. Activation Energy and Preexponential Factor at HT for Atom Recombination on Silica atoms
E/kJ mol-'
N
14 2.8 17 3.8 41 f 6.0
0 H
* *
d 1.9 x 10-3 2.0 x 10-3 1.9 x 10-1
Error bars for rp are shown in Figures 3, 5,and 6. Table 111. Estimation of Ed by Curve Fitting
oxygen/ nitrogen
E'/kJ mol-'
0.2 41
Ed/ kJ mol-'
51 f 1
37*2
B cp
E/kJ mol-' Figure 8. Sketch of the surface model with scattered active sites. Table I. High Temperature Data for Atom Recombination on Silica nitrogen
oxygen
hydrogen
T/K
y x 104
T/K
y x 104
T/K
1260 1250 1140
3.8 3.7 3.5 2.9 2.0
1260 930 800 690 590
4.4 3.0 2.5 2.0 1.2
1240 1240 1140 980 920 920
1030 920
y x 104
~ _ _ _ _
36 33 15 11 11 8.6
hydrogen
0.5 10-2 2 x 10-3 16 does not apply does not apply
P
d
reaction depends on the relative magnitude of the diffusion distance, X D ,and the half distance between active sites, b. At LT, X Dis larger than b, so that the collection zones overlap and all atoms impinging on the surface reach the active sites. At MT, the collection zones do not overlap and shrink with increasing temperature. At HT, XD is smaller than a, so surface diffusion does not contribute to recombination that proceeds only by direct impingement of a gas-phase atom on a chemisorbed atom at an active site (step ER). Thus, eq 8 becomes
transition state, the standard entropy increase for desorption49gives a value of Vd equal to 1015s-' in order of Y = (o exp(-E/RT) (9) magnitude. If there is no change in standard entropy for surface diffusion, V D is 1013 s-', as reported by others.50 The values of the preexponential factor, (o, and activation in order of magnitude. Therefore, 0 was assumed to be energy, E , were obtained by a least-squares fit of eq 9 from Oxygen and Nitrogen Recombination. The concept the data a t H T shown in Table I and in Figures 5 and 6. of X D was used to account for the observed variation of As shown in Table 11,almost identical values of (o for oxygen y from LT to HT. Figure 8 shows active sites scattered and nitrogen recombination (ca. 2 X indicate that uniformly on the surface. Everything happens as if each about 0.2 % of the surface is covered by active sites ( b active site is surrounded by a collection zone, the area 20a). Also, the activation energy is almost the same for within a distance X Dfrom the active site. On the average, nitrogen and oxygen recombination (ca. 16 kJ mol-'). atoms striking the surface within the collection zones reach With the values of P, p, (o, and E shown in Table 11, active sites before desorption and may recombine with results for oxygen and nitrogen from LT to HT were the chemisorbed atom. The diffusion-reaction equation analyzed by a least-squares fit of eq 8. This yielded the on the surface, where the active sites are spaced at a value of the only adjustable parameter, Ed, namely, 51 kJ distance 2b apart, was solved by several a ~ t h o r s . ~ ~ p ~mol-l ~ - ~(Table ~ 111). The calculated curves shown in Figures The probability, P, that atoms striking the surface reach 5 and 6 fit the data from LT to H T reasonably well. The to the active sites is value of Ed for oxygen and nitrogen is compared below with that for hydrogen. 2 a X ~I i ( b / x D ) K l ( a / x D )- I i ( a / x ~ ) K l ( b / x ~ ) Activation Energy for Recombination. Let us p=+(o examine the value of E. According to Hir~chfelder?~ the b2 Io(a/XD)K,(b/XD)+ Ii(b/XD)Ko(a/XD) activation energy of an exothermic step, AB + C A (7) BC, is given to a rough approximation by 0.055 D(A-B), where Ii and Ki are the modified Bessel functions of ith where D(A-B) is the dissociation energy of A-B. This order. The first term is the probability of reaching the estimation can be applied to the elementary steps LH or active site by surface diffusion. The second term, (o, in ER the above expression is the probability of direct impingement on the active sites and is equal to the fraction of the -A A * + A, surface covered by active sites, ( a / b)2. Recombination where A represents N or 0 atoms. The value of the probability is the probability P times the reaction probactivation energy (16 f 2 kJ mol-') indicates the dissoability of arriving atoms at the active sites, exp(-E/RT), ciation energy, D(*-A), is about 290 f 30 kJ mol-'. This as explained above value satisfies the assumption of exothermicity of the y = P exp(-E/RT) (8) elementary step because the binding energies of dinitrogen and dioxygen are higher than D(*-A). The chemiIn this type of diffusion-limited reaction, the net rate of sorbed atom at the active site thermally desorbs at a certain temperature. (49) Boudart, M.; Loffler, D. G. Catal. Lett. 1990,6, 317. (50) George, S. M.; DeSantolo, A. M.; Hall, R. B.Surf. Sei. 1985,159, According to the rule of thumb used in the study of L425. temperature-programmed d e s ~ r p t i o nany , ~ ~chemisorbed (51) Kuan, D. Y.; Davis, H. T.; Aris, R. Chem. Eng. Sci. 1983,38 (5),
-
-
+
+
719. ~.
(52) Rumpf, F.; Poppa, H.! Boudart, M. Langmuir 1988, 4 , 722. (53) Henry, C. R. Surf. Scz. 1989,223, 519.
(54) Hirschfelder, J. 0. J. Chem. Phys. 1941, 9, 645. (55) Smith, A. W.; Quets, J. M. J. Catal. 1965, 4, 163.
+
Kim and Boudart
3004 Langmuir, Vol. 7, No. 12, 1991 Table IV. For All Atoms (0,N, and H),y = exp(-E/RT) at LT in Order of Magnitude y a t 194 K
atoms
N
0
H
EJ k J mol-I 14 17 16a
calcd
exptl
2 x 10-4 3 x 10-5 5 x 10-5
3 x 10-5 8 X lo* 5 x 10*
For H recombination, an average of E for 0 and N recombination was taken. a
species on the surface readily desorb at a thermal desorption peak temperature in kelvin: Td = D(*-A)/0.06/ 4.2, where 4.2 is the conversion factor between kilocalorie and kilojoule. Chemisorbed atoms with a binding energy 290 f 30 kJ mol-1 should desorb around 1150 f 140 K. Above this temperature, it is expected that fewer and fewer active sites are occupied by the chemisorbed atom with increasing temperature. A falloff of y at 1250 K was reported by RosnerZ7(Figure 5) in a study of nitrogen recombination on silica. This supports the role of the chemisorbed atom as the active site. Thus, the value of E obtained in the curve fitting is consistent with the estimated activation energy required to break the bond *-A. At LT, when XD is larger than b, atoms adsorbed anywhere on the surface diffuse to active sites, P = 1. If all the atoms striking the surface are adsorbed, as assumed above, the recombination probability (eq 8) becomes y = exp(-E/RT) (10) With the values of E obtained for 0 and N recombination at HT, y at 194 K was calculated from eq 10 (Table IV), and the measured values of y agree within an order of magnitude. Note that the measured y for H recombination at 194 K is also not much different from that for 0 recombination (Table IV). When the average value of E (16 kJ mol-l) in 0 and N recombination is used for H recombination, the calculated y is also consistent with the measured y for H within an order of magnitude. For all three atoms, y = exp(-E/RT) at LT. High values of q‘ and E’ for H recombination (Table 11)indicate that for H, besides the active sites responsible for recombination of 0,N, and H, there are different active sites for H recombination at HT. Hydrogen Recombination. Looking at H T values of (aand E (Table 11),we see values for H atom recombination higher than for 0 and N recombination, namely q‘ and E’. We suggest that these (a’ and E’ values be interpreted in terms of H atom recombination with hydrogen of the OH groups on the silica surface. This is based on the following evidence. Zhuravlev and Kiselev56measured the number density of surface OH groups for more than 40 different samples of crystalline and amorphous silica (10-950 m2 g-l) by isotope exchange between DzO and the surface OH groups. They found 5 nm-2 for all their samples regardless of specific surface area. Various authors5‘ have reported similar values with thermogravimetric or ion exchange methods. This surface concentration of OH groups on silica indicates that every surface silicon binds to one OH g r o ~ p and ~ ~ that p ~ a~ silica surface at RT is fully covered (56) Zhuravlev,L. T.; Kiselev,A. V. InProceedings of the International Symposium on Surface Area Determination; Everett, D. H., Ottewill, R. H., Eds.; Butterworths: London, 1970; p 55. (57)Boehm, H.P. In Advances in Catalysis and Related Subjects; Eley, D. D., Pines, H., Weisz, P. B., Eds.; Academic Press: New York, 1960; Vol. 16,p 225. (58)de Boer, J. H.; Vleeskens, J. M. Koninkl. Ned. Akad. Westenschap Roc. 1958, B61,2.
by surface OH groups. After the surface was heated to 1270 K under vacuum,6O the surface number density of OH groups decreasesto 1nm-,, as the surface is dehydrated by reaction between two surface OH groups forming a siloxane group on the surface and a desorbed water molecule. After our experiment at 1250 K, the surface of silica contained the same surface number density of OH groups,correspondingto a fractional surface coverageequal to 0.2. This indicates that a H atom does not break the siloxane group to generate a surface OH group. The residual OH groups appear to provide the additional active sites for hydrogen recombination at HT Si-OH
+H
-
Si-0
+ H,
Indeed, the observed value of (a’, 0.2, is consistent with the fraction of surface covered with OH groups as reported by others. The contribution of additional hydrogen recombination through surface OH groups, q‘ exp(-E’IRT), appears in the right side of eq 8 y = P exp(-E/RT) + q‘ exp(-E’/RT) (11) Values for q‘ and E’ are taken from the analysis of H T data (Table 11). With the same values for (a, E, 6, and p as in the 0 and N analysis (Table III), the least-squares fit for the H atom recombination data is shown in Figure 3. The fit is good but the value of the one adjustable parameter, E d , for H is 37 f 2 kJ mol-l, which is lower than that for 0 and N. The value of E d can be estimated from the interaction energy between these atoms and the silica solid by using the Kirkwood-Muller formula.61*62Like the values obtained in the least-squares fit (Table 111), 0 and N atoms give a larger E d (ca. 15 kJ mol-’) than H (ca. 8 kJ m ~ l - l ) . ~ ~ However, these calculated values are smaller than those obtained from the least-squares fit by about a factor of 4. This can be understood in terms of the partial coverage of OH groups on the silica surface. de Boer and Custers63 point out that the interaction energy of an atom physisorbed in a “pocketlike” adsorption site interacts with the site with an interaction energy 4 times larger than with the flat surface. Pocketlike adsorption site is meant by the site due to surface roughness of the atomic level. The partial coverage of OH groups on the silica surface may increase the interaction energy by providing pocketlike adsorption sites. Comparison with Literature Values. Apparent atom recombination probability on silica glasses, such as Pyrex, Vycor, and silica, has been reported in the literature. For comparison, the same roughness is assumed ( F = 2.4) for all sample surfaces. The value of y at RT varies between 4 X 10-6 and 1 X for oxygen recombination,l-15 while for n i t r ~ g e n l and ~ - ~h~y d r ~ g e n , y~ varies ~ - ~ ~from 2 x 10-6 and from 1X to 1X respectively. The to 2 X scatter in the literature data may be due to number of factors such as inhibition by water on the surface,42surface roughness, composition and structure of the surface, and the experimental method used. Marshalllg used electron spin resonance spectroscopy to measure relative concentration of N atoms in flowing Nz gas inside a silica tube. As shown in Figure 5, his value of y on silica measured in this flow system is in a good agreement with our results. This confirms the reliability (59)Schneider, M.Ph.D. Thesis, University of Heidelberg, Germany, 1962. (60)Davydov, V. Y.;Kiselev, A. V.; Zhuravlev, L. T. Trans. Faraday SOC. 1964, 60,2254. (61)Polanyi, M.; London, F. Naturwissenschaften 1930,18, 1099. (62)Muller, A. Proc. R. SOC.London, A 1936, A154, 642. (63)de Boer, J. H.; Custers, J. F. H. Z. Phys. Chem. 1934, B25,225.
Langmuir, Vol. 7,No. 12, 1991 3005
Recombination of 0, N, and H Atoms on Silica
of our measurements of relative atom concentration profile. Rosner et al.,Z7 who worked at higher temperatures than we did, reported y for N atom recombination decreasing with increasing temperature at those higher temperatures. Berkowitz-Mattuck12 reported the result of oxygen recombination by using NO titration. Vycor shows y an order of magnitude smaller than our results on silica (Figure 6). The overall temperature dependence of y, however, is the same, so Vycor may have a lower number density of active sites than silica. The Pyrex surface used by Tsu and Boudart31for hydrogen recombination (Figure 3) seems to contain more active sites than silica. It should be noted that their highest experimental temperature (439 K) was much less than ours. Atom Recombination on the Surface of Space Shuttle Orbiter. Rakich et al.35reported data obtained during an SSO flight. Thermocouples were affixed just under, and in contact with, the RCG coating material on the insulating tiles. After extensive fluid mechanics and heat transfer calculations, the recombination probability of atoms was estimated from the surface temperatures recorded during the reentry flight. By assuming the recombination of oxygen atom to be the main reaction taking place on the surface of insulation tiles, the authors around estimated a recombination probability of 5 X 1000K as shown in Figure 6. Similar experiments carried in subsequent flights of SSO confirm that figure, as reported by Zoby et al.36 Given the approximations involved in the calculation of the recombination probability from flight experiments, these data on RCG are not significantly different from the results on silica.
Conclusion The complex temperature dependence of the recombination probability of oxygen, nitrogen and hydrogen atoms on silica is explained in terms of the surface diffusion of atoms on a surface with a small fraction of active sites, as first proposed by Tsu and B ~ u d a r t .The ~ ~ kinetic significance of surface diffusion is that surface potentials for diffusion are generally shallow, and thus the rate of surface diffusion is large as compared to the rate of desorption. This means that adatoms may sample many surface sites during their residence time a t the surface and thus increase their probability of finding active sites. Indeed, due to surface diffusion, y at low temperature is increased by 2 orders of magnitude. At high temperature, where the rate of desorption is close to the rate of surface diffusion, the diffusion is negligible and y is proportional to the number density of the active sites. The number density should be decreased in order to make less active coating materials for space vehicles. The falloff of due to the desorption of the chemisorbed atoms at very high temperatures suggests a way to decrease the surface activity for recombination of atoms. Reduction of the binding energy of atoms to the active sites would lower the falloff temperature. In fact, the falloff temperature for silica-based tile material RCG mentioned in the introduction was found to be 1000 K, lower than that for pure silica, 1250 K. Elements other than
silica in RCG may interact with the active sites and decrease the binding energy of atoms to the sites.
Glossary A A, a b
D D, D(A-B) E E' ED Ed
F
FP R Ro
T T' AT Td U U va vd
X XD
atom (0,N, and H) reversibly adsorbed atom hop distance of diffusing atom, cm half distance between active sites, cm diffusion coefficient of a gas phase atom, cm2 s-l surface diffusion coefficient of an adsorbed atom, cm2 s-1 dissociation energy of A-B, k J mol-' activation energy for recombination at active site, kJ mol-l activation energy for H recombinationwith surface hydroxyl group, kJ mol-' activation energy for surface diffusion, kJ mol-' activation energy for desorption, kJ mol-1 roughness factor of silica tube additional roughness factor due to silica powder gas constant radius of sample tube, cm probe temperature before atom generation, K probe temperature after atom generation, K temperature rise (T- T),K thermal desorption temperature, K slope in the plot of In AT vs XIRo, defined in eq 2 gas kinetic velocity, cm s-1 rate of adsorption, cm2 s-1 rate of desorption, cm-2 s-1 distance from the atom generator, cm mean surface diffusion distance before desorption, cm
P
=(vD/ vd)
Y
true recombination probability of atom striking the surface apparent recombination probability frequency factor for surface diffusion, s-l frequency factor for desorption, s-1
YA VD Vd
P
=(&/Ed)
7
mean residence time of reversibly adsorbed atom,
cp
fraction of surface covered by active sites, (a/b)2 fraction of surface covered by surface hydroxyl groups
S
9'
Acknowledgment. Funds for the support of this study were allocated by the NASA-Ames Research Center, Moffett Field, CA, under Interchange No. NCA 2-158. Special thanks are due to R. L. Altman for assistance in building the experimental system at NASA-Ames. D. G . Loffler and J. Xu for help in the interpretation and fitting of the data are also acknowledged. Registry No. Hydrogen atom, 12385-13-6;nitrogen atom, 17778-80-2;oxygen atom, 17778-88-0;silica, 7631-86-9.