J . Phys. Chem. 1991, 95, 4028-4033
4028
the lower FEvalues in our falloff analysis because they can only be rationalized if SK- Senis assumed to be larger than the %ormal" upper limit value of 2. As mentioned above, the "true" value for klJ298K) remains uncertain by a t least a factor of 2. Heats of reaction for the X + NO2 XNOz reactions are summarized in Table VI. While the X-N02 bond strengths decrease monotonically in the order F > CI > Br, the BrN02 bond strength determined in this study is identical within experimental uncertainty to the I N 0 2 dissociation energy determined by Troe and c o - ~ o r k e r s . ~ 'It. ~should be noted that the potential contribution of an I O N 0 species to the iodine recombination experiments has yet to be adequately addressed.
-
Summary The kinetics of the Br(2P312)+ NO2 association reaction have been investigated as a function of temperature (259-432 K), pressure (12.5-700 Torr), and bath gas identity (He, Ar, H2, N2, C02,CF4, SF6). At temperatures below 350 K, the association reaction is irreversible on the (-30 ms) time scale of the experiment. The 21 rate coefficients obtained with N2 as the buffer gas were fit to the expression recommended by the NASA panel for chemical kinetics and photochemical data evaluation for use in parametrizing the pressure and temperature dependences of association reaction rate coefficients for atmospheric modeling the best-fit parameters are A = 4.24 X 10-31(T/300)-2.4 ~ )B = 2.66 X T/300)0,0 cm3 cm6 molecule-2 s-'( ~ k ' ,and molecule-' s-l ( - k l , J . At temperatures above 350 K, reversible addition has been observed. Rate coefficients for BrN02 formation and decomposition have been determined over the temperature range 374-432 K. Second- and third-law analyses of the data yields somewhat (44)Hippler, H.; Luther, K.;Teitelbaum, H.; Troe, J. Inr. J . Chem. Kinet.
1977. 9. 917. -.
(45)?Danis, F.; Caralp, F.; Masanet, J.; Lesclaux, R. Chem. Phys. Lett. 1990, 167,450. (46) Herzberg, G. Molecular Spectra and Molecular Structure. Ill. Electronic Spectra and Electronic Structure of Polyatomic Molecules; Van Nostrand Reinhold: New York, 1966. (47) Donovan, R. J.; Husain, D. Chem. Reu. 1970, 70,489. (48) Hippler, H.; Troe, J.; Wendelken, H. J. J . Chem. Phys. 1983, 78, 6709. (49) Mourits, F. M.;Rummens, F. H. A. Can J . Chem. 1977,55, 3007. (50) Reid, R. C.: Shenvood, T. K. Properties of Gases and Liquids; McGraw-Hill: New York, 1958.
different thermochemical parameters. The major uncertainty in the calculated third-law entropy change appears to be the electronic entropy of BrN02. Averaging the second- and third-law results and choosing error limits to encompass all reasonable possibilities, we report the following thermochemical parameters for reaction 1: p H 0 2 9 8 = -19.6 f 1.7 kcal mol-', AHoo = -18.6 f 2.0 kcal mol-', = -29.3 f 4.2 cal mol-' K-I, AHf0298(BrN02) = 17.0 f 1.8 kcal mol-' (uncertainties are 2u estimates of absolute accuracy). Our experimental value for AHoo has been employed to calculate k , , F , the low-pressure limit rate coefficient in the strong collision limit, using the method of Troe.33-35Calculated values for k,,oSCare inconsistent with experimental results unless - A P 0 is assigned a value near the lower limit derived from the hightemperature data, Le., 16.6 kcal mol-'. Systematic errors in the calculations could result from the assumptions that (a) reaction occurs entirely on the ground electronic state potential energy surface and (b) formation of the BrONO isomer is unimportant; experimental data are available that suggest that assumption b is valid,'"14 but no information is available to validate assumption a. The procedure developed by Troe and C O - W O ~ ~ ~has ~ S ~ been employed to extrapolate experimental falloff curves to the low- and high-pressure limits. Derived values for l~',~(M,298K) in units of cm6 molecule-2 s-' range from 2.75 for M = He cm6 to 6.54 for M = C02. Values for kl,0(N2,T)in units of molecule-2 s-' are 5.73 at 259 K, 4.61 a t 298 K, and 3.21 at 346 K; the temperature dependence of kl,o(N2,T)is consistent with the theoretical temperature dependence for j3ckl,oSC.Values for k,,JT) in units of 10-l' cm3 molecule-' s-I are 2.86 at 259 K, 3.22 at 298 K, and 3.73 at 346 K. Uncertainties in derived values for k,,oare estimated to be f20%, whereas uncertainties in derived values of k I , J 7') are considerably larger-a factor of 2 or more; however, the derived small positive activation energy for kIJ T ) is probably correct. Experimental data up to pressures of several hundred atmospheres and ab initio calculations that characterize the low-lying bound electronic states of BrN02 are needed before the magnitude of kI,- can be considered well established.
Acknowledgment. We thank Drs. G. Yarwood and H. Niki for communicating their unpublished results to us. This work was supported by NSF Grant ATM-88-02386 and NASA Grant NAGW- 1001.
Protectlon of Rh/Ai203 Catalysts by Potassium Functlonallzatlon of the A1,03 Support Mohamed I. Zaki? Todd H. Ballinger, and John T. Yates, Jr.* Surface Science Center, Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 (Received: July 16, 1990)
It has been found that the exchange functionalization of isolated AI-OH groups with K+ to produce AI-OK groups provides
a method for the stabilization of Rh/AI2O3 catalysts against Rh' formation. K2C03was employed for functionalization, and the thermal decomposition of the C032-anion (T< 870 K)was studied along with the effective and preferential removal of isolated AI-OH groups, producing AI-OK species. Rh/K-A1203 catalysts are shown to be stable against Rhoxdegradation in the presence of CO(g). The AI-OK functionalization procedure is effective in preventing AI-OH re-formation under extreme hydrolysis conditions up to 770 K. This procedure may lead to effective practical methods for the protection of Rh environmental catalysts against loss of catalytic activity due to Rhoxdegradation.
I. Introduction Supported rhodium a t a l y s t ~have been extensively investigated by using a number of surface spectroscopies including IR,'+ EXAFS,'*I2 and NMR spectro~copies.'~*'~ This work has pro*To whom correspondence should be addressed. 'On leave of absence, Minia University, El-Minia, Egypt.
0022-36S4/91/209S-4028S02.50/0
duced a reasonably deep understanding of Rh surface chemistry and structure in connection with its catalytic performance in (1) Yann. A. C.: Garland. C . W. J . Phvs. Chem. 1957. 61. 1504.
(2) Cavinagh, d. R.; Y a k , J. T., Jr.-J. Chem. Phys. 1981, 74, 4150. (3) Solymosi, F.; Pasztor, M.J. Phys. Chem. 1985, 89, 4789. (4) Solymosi, F.;Pasztor, M. J . Phys. Chem. 1986, 90,5312.
Q 1991 American Chemical Society
~ - ~ ~
The Journal of Physical Chemistry, Vol. 95, No. 10, 1991 4029
Protection of Rh/A1203 Catalysts syngas-related industrial reactions15J6and in automobile exhaust gas converters." One of the issues that has been thoroughly investigated due to a close relevance to the catalyst durability is the oxidative disruption of A1203-supportedRhoxparticles into highly dispersed Rh' species in a CO atmosphere. It has been postulated that CO molecules erode Rhoxmetal particles,'oJl which then interact with support AI-OH groups and convert subsequently to oxidized catalytically inactive Rh1(C0)2species: (1 /x)Rho,
+ AI-OH + 2CO(g)
-
(Al-O-)Rh'(CO), + (1 /2)H2(g) (1)
This general type of interaction between OH groups and transition metals on oxide supports was postulated earlier by Brenner et al.'&mand confirmed later by Solymosi and Pasztor,' Zaki et al.? and Basu et aL6v7 Moreover, Basu et al.6.7 provided direct IR spectroscopic evidence indicating that the Al-OH species involved are the isolated hydroxyl species on yA1203(and on SO2). None of these experiments have actually monitored the production of H2(g) in the postulated oxidation Accordingly, research efforts to find methods to prevent the oxidative disruption of Rhoxparticles have primarily focused on determining the most adequate and effective ways of eliminating active Al-OH groups. The earliest attempts3 chose to eliminate the Al-OH groups by heating Rh/A1203catalysts to high temperatures. Complete dehydroxylation was only possible after prolonged evacuation above 1100 K. Metal particle sintering and subsequent metal penetration into an A1203 upp port^'-^^ as well as support structural modification^^^ have recently been observed in the course of the high-temperature treatment of Rh/A1203. The undesirable effect of wering up Rh sites by the support can render the thermal dehydroxylation process technologically unfavorable at elevated temperatures. In order to avoid such undesirable side effects, in previous work dehydroxylation of Rh/A1203was effected by a low-temperature (473 K) in situ exposure to hexamethyldisilazane(HMDS) vapor, followed by evacuation for 1 h.5 HMDS reacts readily with acidic AI-OH groups, leading to the formation of surface silyl groups:% 2Al-OH (5)
-
+ HN(Si(CH3)3)2
2Al-O-Si(CH3)3
+ NH3
(2)
Zaki, M. I.; Kunzmann, G.; Gates, B. C.; Knijzinger, H. J . fhys.
Chem. 1987, 91, 1486. (6) Basu, P.; Panayotov, D.; Yates, J. T., Jr. J . Phys. Chem. 1987, 91, 3133. (7) Basu, P.; Panayotov, D.; Yates, J. T., Jr. J . Am. Chem. Soc. 1988,110, 2074. (8) Zaki, M. I.; Tesche, B.; Kraus, L.; Knbzinger, H. Surf. Interjuce Anal. 1988,12,239. (9) Robbins, J. L. J . fhys. Chem. 1986, 90, 3381. (10) van't Blik, H. F. J.; van Zon, J. B. A. D.; Huizinga, T.; Vis, J. C.; Koningsberger, D. C.; Prins, R. J . fhys. Chem. 1983,87, 2264. van't Blik,
H. F. J.; van Zon, J. B. A. D.; Huizinga, T.; Vis, J. C.; Koningsberger. D. C.; Prins, R. J . Am. Chem. Soc. 1985, 107, 3139. (1 1) The concept involving hydroxyl group reaction in the production of Rh'(CO)2 is derived from earlier studies of &(CO)16 behavior. See: Smith, A. K.; Hughes, F.; Thcolicr, A.; Basset, J. M.; Ugo, R.; Zanderighi, G. M.; Bilhou, J. L.; Bilhou-Boughol, V.; Graydon, W. F. Inorg. Chem. 1979. 18, 3104. Thcolicr, A.; Smith, A. K.; Ltconte, M.; Basset, J. M.; Zanderighi, G. M.; Psaro, R.; Ugo, R. J. Orgunomer. Chem. 1980, 191,415. (12) Koningsberger, D. C.; Martens, J. H. A.; Prins, R.; Short, D. R.; Sayers, D. E. J. Phys. Chem. 1986, 90,3047. (13) Duncan, T. M.; Root, T. W. J . Phys. Chem. 1988,92,4426. (14) Thayer. A. M.; Duncan, T. M. J . Phys. Chem. 1989, 93,6763. (15) Poels, E. K.; Ponec, V. Curlrlysis-Specialist Periodicul Reporr, Vol. 6; Royal Society of Chemistry: London, 1983; pp 196-221. (16) Lee, G. V. D.; Ponec, V. Cutul. Rev.-Sci. Eng. 1987, 29, 183. (17) Taylor, K. C. In Cutulysis-Science und Technology,Vol. 5; Anderson, J. R., Boudart, M.. Us.; Springer-Verlag: New York, 1984; pp 119-170. (18) Brenner, A.; Burwell, R. L., Jr. J . Curd. 1978, 52, 353. (19) Brenner, A.; Hucul, D. A. J . Curd. 1980, 61, 216. (20) Hucul, D. A.; Brenner. A. J . Phys. Chem. 1981,85, 496. (21) Wong, C.; McCabe, R. W. J . Curd. 1989, 119, 47. (22) Chen, J. G.; Colaianni, M. L.; Chen. P. J.; Yates, J. T., Jr.; Fisher, G. B. J . Phys. Chem. 1990, 94, 5059. (23) Ballinger. T. H.; Yates, J. T., Jr. J . Phys. Chem. 1991, 95, 1694. (24) van Rawrmalen, A. J.; Mol, J. C. J . Phys. Chem. 1978,82,2748. van Roosmalen, A. J.; Hartmann, M. C. 0.;Mol, J. C. J . Curd. 1980, 66, 112.
IR spectra taken, following CO adsorption on the silated catalyst, indicated a marked suppression in the formation of the oxidized Rh1(C0)2species. However, no attempt was made to monitor consequent changes in the surface-OH population. Moreover, the possible molecular coordination of HMDS, and the NH3 produced, to Lewis acid sites on the support was not experimentally controlled. Adopting a similar approach, recent s t ~ d i e shave ~ ~ ,succeeded ~ in specifically eliminating isolated Al-OH species, by exposing a Rh/Al2O3 catalyst at 450 K to (CHJ3SiCI, as shown in reaction 3. The removal of these OH groups was confirmed spectroAl-OH
+ C1-Si(CH3),
-
-
Al-O-Si(CH3)3
+ HCl
(3)
scopically, and this silation treatment strongly suppressed the tendency for the Rhox(CO) Rh1(C0)2conversion. It was also demonstrated that the CO chemisorption capacity of Rhoxsites is maintained following the surface s i l a t i ~ n .Subsequent ~~ experiments26indicated that Al-O-Si(CH3)3 can withstand hydrolysis up to at least 430 K. When heated in O2 at 475 K, however, the CH3 moieties are observed to oxidize, but residual A1-O-Si linkages are postulated to remain intact, and these residual species have been found to be protective against Al-OH re-formation under extreme hydrolysis conditions. The present study is consistent in concept with the earlier ~ t u d i e s , 2 'for ~ ~it~ also adopts a chemical modification approach. However, here we have developed an inorganic surface modification p d u r e . K2CO3 is used as the modifier, with the objective of replacing the isolated Al-OH groups by AI-OK groups. A Rh/K-AI2O3 catalyst is prepared and compared with Rh/A1203, using IR spectroscopy and CO as a probe. The chemical consequences of the hydrolysis of Rh/K-A1203 at high temperatures are also investigated. The choice of K2C03 as an alumina modifier was made on the basis of promising work in the literature. For instance, there is general agreement that K2C03 strongly interacts with A1203 surface^?^-^^ and it has been suggested that K2C03reacts with the relatively basic Al-OH groups to form Al-OK.'O A further reaction with the alumina lattice was considered unlikely by Stork and Pott?* and other^,^^,^^ because of the large size of the potassium ion. 11. Experimental Section
The stainless steel ultrahigh-vacuum IR cell used in these studies has been described previo~sly.~~ It is equipped with CaF2 optical windows allowing IR measurements in the 4000-1000-cm-' spectral range. The cell is attached to a bakeable all-metal gas handling system equipped with a liquid Nz cooled zeolite sorption pump, a 30 L/s ion pump, a Baratron capacitance manometer, and a quadrupole mass spectrometer for gas measurements. The base pressure in this system is typically 1 X lo-* Torr. The high area catalytic precursor materials are deposited by a spraying technique onto a tungsten grid (70% transmittance) which is held rigidly by nickel clamps. Electrical heating power may be used to control the grid temperature, during spraying and also under vacuum, using an electronic ~ o n t r o l l e r . The ~ ~ temperature of the grid, and hence the supported catalyst, is measured by a chromel/alumel thermocouple (0.077-mm diameter) spotwelded to the top central region of the grid. For infrared measurements, the grid support is held in the center of the stainless steel cell. (25) Paul, D. K.; Ballinger, T. H.; Yates. J. T., Jr. J . Phys. Chem. 1990, 94, 4617. (26) Paul, D. K.; Yates, J. T.. Jr. J . fhys. Chem. 1991, 95, 1699. (27) Levy, R. M.; Bauer, D. J. J . Cord 1967, 9. 76. (28) Stork, W. H. J.; Pott, G. T. J . fhys. Chem. 1974, 78, 2496. (29) Krupay, B. W.; Amenomiya, Y. J. Cutul. 1981, 67, 362. (30) Kantschewa, M.; Albano, E. V.; Ertl, G.; Knbzinger, H. Appl. Curd. 1983, 8, 7 1. (31) Levy, R. M. J . Curd 1967, 9, 87. (32) Basu, P.; Ballinger, T. H.; Yates, J. T., Jr. Rev. Sci. Instrum. 1988, 59, 1321. (33) Muha, R. J.; Gates. S. M.; Yates, J. T.. Jr.; Basu, P. Rev. Sci. Instrum. 1985, 56, 613.
Zaki et al.
4030 The Journal of Physical Chemistry, Vol. 95, No. 10, 1991
Degussa y-aluminum oxide C (1 04 m2/g) and J.T. Baker anhydrous K2C03 (99.98% pure) were used to prepare the K2C03-impregnated y-A1203 (denoted K2CO3/AI2O3). The amount of K2CO3 required to obtain 5 wt % K was dissolved in an appropriate volume (3 mL/g of AZO3)of distilled water. Then, the appropriate amount of yAI2O3was sprinkled slowly into the carbonate solution while being stirred, until a homogeneous paste was formed. The paste was subsequently dried at 393 K for 24 h in an electric oven. The dried material, K2CO3/AI203,was calcined in air at 973 K for 5 h to obtain the K-modified A1203 (denoted K-A1203). K-A1203 was not analyzed for the actual potassium content. For comparison purposes, a sample of K-free y-A1203was prepared following the above procedure. The catalyst preparation procedure involved slurrying the required amounts of Aldrich RhCl303H20and K-modified or K-free alumina to produce 2.2 wt % Rh/K-A203 or 2.2 wt % Rh/Al,O,, respectively. The slurrying was effected in a liquid consisting of nine volumes of acetone and one volume of distilled water, by ultrasonic agitation for 45 min. The K-containing suspension required a longer period of ultrasonic agitation (1.5 h) to yield a reasonable slurry. The slurry was then uniformly sprayed, by a N2-pressurizedatomizer, onto the entire exposed grid area (5.2 an2).The grid was electrically heated during spraying to 323-333 K to flash evaporate the liquid phase.32 The net weight of the material deposited onto the grid varied, depending on the final state of gelation in the slurry. It was highest (28.4 mg) with the K-free materials and lowest (12 mg) with the K-containing materials. Spectroscopic measurements were in some cases normalized to an identical mass of the supported catalysts. The supported Rh catalyst was then prepared by reducing the catalyst parent material, following 12 h outgassing at 473 K. The reduction was achieved at 473 K with four successive exposures of 400 Torr of H2 (99.9995%pure, Matheson) for 15-60 min (after each exposure, the cell was evacuated for 30 min), followed by outgassing at the reduction temperature for 12 h. The carbon monoxide used with adsorption experiments was 99.9%pure obtained from Matheson Gas Roducts in a break-seal glass storage bulb. The water atmosphere used with the hydrolysis experiments was obtained by expanding the vapor of doubly distilled water which was purified prior to application by freeze-pumpthaw cycles. Transmission IR spectra were measured for the catalyst and for the adsorbed species in a purged double-beam Perkin-Elmer Model 580B infrared grating spectrometer coupled with a Model 3500 data station for data storage and manipulation. Spectra were signal averaged for data acquisition times of 2.2 s/cm-l (3900-3200 cm-I), 3.7 s/cm-' (2400-1700 cm-I), and 2.1 s/cm-l (17OC!-1000 cm-I), acquired at 1 point/cm-'. The spectra shown in the hydroxyl region and the carbonate region have not been corrected for the background. All other spectra have been obtained by using a background subtraction procedure involving the starting catalyst material. X-ray powder diffractograms were obtained with a Model XRD 700 Diano diffractometer. A Model CA-8L Diano generator operated at 50 kV and 32 mA provided a source of Ni-filtered Cu Ka radiation (A = 1.54051 A). The diffractometer was operated with 3.0° diverging and 0.2O receiving slits at a Scan rate of 2 deg/min and produced a continuous trace of diffracted X-ray intensity as a function of 20. The samples for diffraction studies were obtained by scraping the material off the tungsten grid and pressing the powder onto a glass substrate. Diffraction patterns (Illovs d spacings (A)) were derived from the diffractograms and matched subsequently with those filed as ASTM standards.M 111. Results and Discussion A. Potassium-Modiijied Alumina (K-AlZO3). 1 . Observation of OH- and COj2- Decomposition on K-AI2O3. The K-Al2O3
support material employed in this study was prepared by c a h nation of K2CO3/AI2O3at 973 K for 5 h. In order to study the (34) ASTM Powder Diffraction File, Joint Committee on Powder Diffraction Standards, Philadelphia, PA, 1967.
TsCon= 300 K U
13:s
p y
-+Resolution Maximum Heolinq Temperature
Q
u C
0
e s n a
,
I
,
I
I
I
\,(dl
I d l
,
,
,
3 00 3700 3500 3300 1800 1600 1400 Wovenumber ( c m - ' )
F i i 1. infrared spectra of K,CO, decomposition on A1203following heating under vacuum at the temperature indicated: (A) VOH region, using a 19-point smoothing function to smooth the displayed spectra; (B) vc0,z- region. Measurement temperature was 300 K.
thermochemical processes occurring during calcination, separate experiments were performed in which IR spectra were taken from K2CO3/AI2O3mounted inside the IR cell, following 48-h outgassing at 473 K and 5 X 10" Torr. To mimic calcination, sequential heating experiments were carried out under vacuum at higher temperatures (up to 973 K) for 30 min followed by IR measurements. The spectra were taken at 300 K, in the VOH (4000-3000 cm-'), vc4* (1800-1300 cm-I), and b*(1200-1000 cm-l) spectral ranges. The vOH and vc0,i- spectra are shown in Figure 1A,B. The spectra obtained for K2C03/A1203after outgassing at 473 K are similar to those of the unheated material (at 300 K) and differ significantlyin the VOH region from that of untreated Al203. For K2C03/A1203,the YOHspectra display bands at 3718, -3550, and 3445 CUI-'. The 3718- and -3550-cm-' bands are due respectively to small amounts of isolated and to associated AI-OH group^.^^.^^ For the carbonate species, the spectra (Figure 1B) monitor several vco Z- bands at 1560, 1395, and 1348 cm-'. In addition, a very weak 6 ~ 0 ~band 1 - was observed at 1097 cm-'(not shown). According to Busca and Lorenzelli,3' these bands indicate coexisting monodentate (1395 (v,), 1348 (v,), and 1097 cm-' (a); v, - v, = 47 cm-I) and chelating bidentate (1 560 (vu) and 1348 cm-I (us); vu - v, = 212 cm-l) C O P species bound to K+ ions. It is worth mentioning that potassium hydroaluminocarbonate (K2O.AI2O3.2CO2.2H20)is reported38to exhibit a largely similar (s), 1544 (vs), 1412 set of OH and C O l - IR bands at 3-3412 (vs), and 1108 cm-I (w). The resemblance of the IR spectra obtained here to the spectra of potassium hydroaluminocarbonate seems to indicate that there is a strong interaction between K2C03 and the alumina surface, as proposed In addition, the strong VOH band at 3445 cm-'(Figure lA), which is not typical of alumina s u r f a d H band^,)^.^^ may be due to AI-OH groups associated with potassium. Following heating at 570 K,the IR spectra (Figure 1B) exhibit an appreciable weakening of the monodentate uco,h bands, and the band due to the bending mode at 1097 cm-I has disappeared. Concomitantly,as shown in Figure lA, the VOH band at 3445 cm-' decreases significantly in absorbance. A mass spectrometric analysis of the gas phase during the thermal treatment at 570 K detected the production of C 0 2 and H20. These results indicate that the surface carbonate begins to decompose at 570 K. The persistence of the v-2- infrared bands at 1560 and 1348 cm-' may suggest that the remaining carbonate species are preferably bidentate bound to potassium. (35) Peri, J. B. J. Phys. Chcm. 1%5,69, 220. (36) KnBzinger, H.; Ratnasamy, P.CuruJ. Rev.-Sei. Eng. 1978,17, 31. (37) Bum, G.; Lorenzelli, V. Murrr. Chcm. 1982, 7, 89. (38) Berger, A. S.;Tomilov. N. P.; Vorsina, I. A. Russ. J . Inorg. Chcm. (Engl. Trunsl.) 1971, 16, 42.
The Journal of Physical Chemistry, Vol. 95, No. 10, 1991 4031
Protection of Rh/AI2O3 Catalysts
I
28 (degreer) 70
80
50
45
I
I
I
.
I
TScan = 300K
2.2% Rh/K-AI,O,
200 Counts
1
30
35
40
I
3680
3185
pScon = 5x
TW!
dk Resolution 5 . 3 cm-‘
K-AI,O,
c .a
% C
(47310
-
U
C
I
a.
I
(47310
(47310
100 3700 3500 3300
1.4
1.6
1.8
2.0
d
2.2
2.4
2.6
2.8
3.0
A
Figure 2. X-ray powder diffractograms obtained from (a) 973 K calcined y-A1203,(b) 973 K calcined K-treated A1203,and (c) 473 K reduced 2.2% Rh/K-A120>
Increasing the temperature to 770 K resulted in IR spectra (Figure 1A) displaying very weak bands due to AI-OH groups (at 3718 and -3500 cm-’) and bidentate C032-species (at 1560 and 1348 cm-l). Further heating to 870 K causes both the OH and the C032- bands to disappear. This behavior indicates that K2C03/A1203decomposes extensively in the temperature range (-570-870 K). It is of interest to note that, in this temperature range, bulk K2C03is thermally stable. Bulk K2CO3 is reported to commence thermal decomposition only at temperatures above 900 K.B Similar results showing low-temperature K2CO3/AI2O3 decomposition were obtained by temperature-programmed decompositionm and thermal analysis28studies of K2C03/A1203 containing 12 and 9 wt % K, respectively. Therefore, all of these experimental results suggest that a strong interaction occurs between K2C03and y-A1203,resulting in a lower C032-decomposition temperature. It is interesting to note that the thermal stability of the A I 4 H groups in the presence of K2C03is also diminished considerably. By 870 K, almost all of the AI-OH species have disappeared, whereas on A I 2 0 3 which has not been treated with K2CO3, Al-OH species persist up to 1100 K.23 It is therefore likely that the reactions leading to K2C03decomposition on A1203involve AlOH groups. Consistent with our observations of the synergistic decomposition behavior of AI-OH and surface carbonate species, it has been reported that water vapor plays an important role in enhancing the kinetics of K&03 decomposition on A1203.39 2. State of Porussium on K-A1203. Figure 2 compares XRD diffraction patterns of K-AI2O3 and the similarly treated pure A I 2 0 3 (Le., calcined at 973 K). The patterns are similar in showing strong diffraction peaks due to y-AI2O3(seeASTM 10-425) and much weaker peaks ( d values 2.58 and 2.70 A) possibly indicative of a minority S-A1203(ASTM 4-877). No characteristic diffraction peaks of potassium oxides (K20, ASTM 10-235; a-KOz, 8-35 1; KO;, 11-526) or aluminates (e.g., KA102,ASTM 2-0897; /3-K2AI2,O3,, 1-1301) are detectable in the pattern of K-Alz03. The XRD analysis results indicate that the impregnation of y-Al2O3 by K+, followed by calcination at 973 K, does not form additional bulk phases at the level of detection of XRD. This
-
(39) Lee, W. H.; Ladd, M. F. C. In Mellor’s Comprehensive Treaties on Inorganic and Theoretical Chemistry;Vol. 11, Suppl. 111; Wiley: New York, 1963; p 1888.
Wavenumber (cm-‘ I Figure 3. Infrared spectra in the uOH region comparing untreated and K-treated A1203 after 473 K vacuum heating. Spectra a and b were normalized to spectrum c by sample weight deposited onto the support grid. suggests that the K resides mainly at the A1203surface. Similar results were obtained through transition-metal phosphorescenceB and infrared reflection” spectroscopicstudies of variously loaded K-modified alumina samples prepared by calcination at similarly high temperatures. Another study2’ employed N2 adsorption isotherms to reveal that K adsorption occurs preferentially at A I 2 0 3 pore entrances, leading to an appreciable pore blockage. More recent work40 found that K does not change the amount of exposed Rh as judged by CO adsorption capacity. 3. Surface Hydroxyl Groups on K-A1203. Figure 3 compares IR spectra in the hydroxyl region obtained from untreated A1203 (spectrum a) and K2C03-impregnated (spectrum b) y-Al2O3 following 48-h outgassing at 473 K. In addition, the OH spectral region for Rh/K-AI2O3 is also shown. The comparison shows the following. (i) The net surface hydroxylation of A 1 2 0 3 decreases considerably as a result of the carbonate impregnation. (ii) While isolated hydroxyl groups on the clean alumina are associated with three strong absorptions appearing at 3732, 3680, and 3585 cm-1135,36 the carbonateimpregnated alumina shows only one very weak band due to isolated OH groups at 3718 cm-l. (iii) The K2C03/A1203surface shows (similar to untreated AlzO3) a broad OH absorption (at -3500 cm-’) due to the associated OH groups but differs in displaying a discernible absorption band at 3445 cm-l which is believed to be due to OH groups associated with potassium.’O This 3445-cm-’ OH band is not present on the reduced Rh/K-A1203 catalyst (reduction at 473 K) as shown in Figure 3c, and this behavior is probably associated with the presence of Rh on the oxide support. These results indicate that the isolated AI-OH groups are largely removed by reaction during the impregnation procedure with K2C03. When coupled with the XRD analysis results, discussed above, these results suggest that K+ ions replace the protons of isolated hydroxyl groups on the alumina surface, as proposed previously by Stork and PottZ8and adopted later by Krupay and Amen0miya.2~The general reaction is shown in (4). K2C03 + 2Al-OH
+
C02 + H 2 0 + 2A1-OK
(4)
The replacement of OH groups by OK groups is consistent with expectations from colloidal chemical principles of the impregnation method applied to prepare K2C03/A1203.41The aqueous sus(40)Mori. Y.; Mori, T.; Hattori, T.; Murakami, Y. J . Phys. Chem. 1990, 94, 4515.
Zaki et al.
4032 The Journal of Physical Chemistry, Vol. 95, No. 10, 1991 I
1
I
I
l
I
I
I
TBcon= 300 K pBCan = 5 x IO-~TOU
T,con = 300 K pIcan = 5 x ~ O - ~ T O ~ I
35?5
+Resolution 5 . 3 cm-l
-+Rcrolution 3676 3 6 0 q
5.3 cm-1
Hydrolysis ( H 2 0 ) Trsolment Tempcroture
(dl
770 K
al
u E 0
e
s
570 K
n
a
300 - 4 7 0 K
2
2
I
( a ) Rh/AI,O, I
i
I
i
untreated
A
00 3700 3500 3300
Wavenumber (cm-') Figure 4. Infrared spectra (A) in the um region after CO adsorption on K-treated and untreated 2.2% Rh/A1203catalysts and (B) in the uOH region before CO adsorption using a 19-pointsmoothing function on the spectra. Spectrum a was normalized to spectrum b by sample weight.
-
pension of the y-A1203particles (IEP at pH = 7.5-8.242) in the carbonate impregnating solution (pH 10) would facilitate the following amphoteric dissociation of the surface hydroxyl groups:
DO 2000 (800 3900 3700 3500 3300
Wovenumber (cm-' ) Figure 5. Infrared spectra showing the stability of 2.2% Rh/K-AI2O3 under hydrolysis conditions: (A) um region and (B) UOH region, treated at the indicated temperature for 30 min under 2 Torr of H20(g). The IR spectra were obtained at 300 K.
sites, leading to the formation of Al-OK groups. It has been suggested that the exchange process is very much enhanced during the drying procedure.43 The stability of the Al-OK groups during chemical treatment of Rh/K-A1203 catalysts is of interest in the work to be described in this paper. Figure 3c shows the IR spectrum in the OH region for a reduced Rh/K-A1203 catalyst prepared a t 473 K by reduction of Rh"' to Rhox using H2. It is seen that the major OH absorption band occurs at about 3500 cm-I, corresponding to associated AI-OH groups. Thus, the AI-OK functionalization prevents the reestablishmentof large cowrages of isolated AI-OH groups during several extreme treatments used to produce the Rh/K-A1203 catalysts, as follows: (1) suspension in aqueous acetone medium, (2) outgassing a t 473 K, (3) reduction in H2 a t 473 K including exposure to H 2 0 products formed during reduction. In addition, the Al-OK functional group is stable at high temperatures under oxygen since it was originally prepared by calcination of K2C03/A1203a t 973 K in air. B. Stability of Rhox/K-A1203 Catalysts: IR Measurements of CO Adsorption. The stability of Rh/A1203and Rh/K-A1203 catalysts toward dispersion according to reaction 1 may be easily determined by IR spectnwoopy. In the case of extensive production of Rh'(C0)2 from Rhox,the characteristic IR doublet due to the symmetric and antisymmetric coupling of the CO vibration in Rh1(CO)2will be observed a t -2100 and 2030 cm-1.1*2Alternatively, under conditions where metallic Rhoxsites are preserved, 2060 cm-I) and bridged CO (vco both terminal CO (uc0 1820 cm-I) will be observed as dominant IR features. In Figure 4A, the remarkable difference between Rh/A1203 and Rh/K-A1203 is shown, following CO adsorption (2 Torr) at 300 K. Spectrum A.a for Rh/A1203 shows the strong doublet due to Rhi(C0)2 and a weak band a t 1843 cm-I due to some population of bridged-CO species on Rhoxsites. In contrast, in spectrum A.b for Rh/K-A1203, strong absorption bands due to
bridged-CO (1816 cm-l) and terminal-CO (2058 cm-') on Rhox sites are present, along with Rh'(CO)2 species which exhibit a small shift of 12-1 3 cm-' to lower wavenumbers. Similar small effects of K on the spectrum of chemisorbed CO on Rh have been reported by others.40 The spectral absorbances in Figure 4A,B have been normalized to identical weights of catalyst, but it should be emphasized that this normalization procedure is only a crude method for comparison of absolute spectral intensities. A comparison of the normalized hydroxyl spectra for each catalyst prior to C O adsorption is shown in Figure 4B. For Rh/A1203 (spectrum B.a) the presence of isolated OH groups is indicated by OH bands at 3724 and 3676 cm-' prior to CO adsorption. These bands disappear upon the formation of Rhl(CO), species in accordance with the findings of earlier investigation~.~~' For the Rh/K-A1203 catalyst (spectrum B.b) the isolated O H species are strongly diminished in absorbance prior to CO treatment. Since Rh1(C0)2formation requires isolated-OH species, less degradation of Rhoxcan occur in the presence of CO. These results therefore indicate that the functionalization of AI-OH groups to produce Al-OK groups is effective in reducing the ability of C O chemisorption to produce a uniformly high dispersion of Rh. The removal of Rhox sites by CO treatment is suppressed considerably when the Rh is supported on the Kfunctionalized A1203,as indicated by the enhanced absorbance of terminal- and bridged40 species on Rhox sites. However, considerable amounts of Rh'(CO), appear to be present also as judged by the Rhl(CO), spectral intensity. Other experimental results have indicated that alkali-metal treatment of supported transition-metal catalysts leads to enhanced dispersion of the metal."*45 The fact that our XRD studies (Figure 2) do not detect the presence of metallic Rh (-4-nm sensitivity6) on Rh/AI2O3 catalysts indicates that while the ultimate dispersion to Rh'(C0)2 species does not occur to a high degree on K-functionalized AI2O3,neither does extensive sintering to large Rh crystallites occur on any of the catalytic preparations studied here. It should be noted that the functionalization of the A1203 support with K+ prior to Rh deposition does not produce Rhox sites modified by K+, as has been re rted by others using a different K+ impregnation procedure4PONormal CO IR spectra
(41) Delmon, B., Grange, P., Jacobs, P. A,, Poncelet, G.,Eds. frepururion ojCutulysts II; Elsevier: Amsterdam, 1979; p 233. (42) Parks, G. A. Chem. Rev. 1965, 65, 177. (43) Che, M.; Bonneviot, L. 2.fhys. Chem. (Munich) 1987, 152, 113.
(44)A i h , K.; Simazaki. K.; Hattori, Y.;Ohya, A.; Ohshima, S.; Shirota, K.; Ozaki, A. J . Curd. 1985, 92, 296. (45) Miyake, T.;Inoue, M.; Inui, T. J . Carol. 1986, 99, 243. (46) Wanke, S.E.; Dougharty, N.A. J . Card 1972, 24. 367.
Al-QH s A1-0-
+ H+
(5)
As a result, K+ ions would be expected to adsorb strongly on Al-0-
-
-
4033
J. Phys. Chem. 1991, 95,4033-4037 characteristic of the pure Rhox surface@ are obtained by using our method of K modification of the A1203. C. Behavior of Rh/K-A1203 under Extreme Hydrolysis Conditions, A series of experiments shown in Figure 5A,B were designed to determine whether Rh/K-AI2O3 catalysts, subjected to extreme hydrolysis treatment, might re-form Rh1(C0)2 if Al-OH groups are re-formed on the support. This experiment involved monitoring the CO and O H IR spectra following treatment of the CO-covered catalyst for 30 min with 2 Torr of water vapor pressure in the temperature range 300-770 K. In Figure 5, the IR spectra for the various hydrolysis treatments are shown as a function of the treatment temperature. It is observed that the Rh1(C0)2 species are actually destroyed during these hydrolysis experiments. Spectrum A.a of Figure 5 represents the starting point for these measurements. Both Rho, and Rh' sites are present. Upon heating to 470 K under H20(g) (spectrum b), Rh1(C0)2has started to disappear and both bridged- and terminal-CO species undergo enhancement of their absorbance. Continued H20(g) treatment to 570 K resulted in complete loss of CO absorbance (not shown; due to expected thermal desorption of CO); the spectra shown in Figure 5A.c,d were obtained by additional CO adsorption at 2 Torr (following heating under H20(g) and evacuation at 300 K). It is noted that a partial recovery of Rh'(C0)2 absorbance is obtained at 570 K, but this trend is reversed again upon heating to 770 K. The hydroxyl spectra (Figure 5B) indicate that hydrolysis cam a significant enhancement of the associated-OH absorbance but only a small enhancement of the isolated-OH absorbances above 3680 cm-*. Thus, AI-OK groups are stable under extreme hydrolysis conditions. Mass spectroscopic studies of the gas-phase composition during hydrolysis indicate that C 0 2 is a major product, along with H2. The H2 produced above the catalyst (water gas shift reaction: CO + H 2 0 C 0 2 H2) may participate in reduction of Rh' to Rho, causing Rh1(C0)2to be consumed as has been observed previously for pure H2
-
+
(47) Kesraoui, S.;Oukaci, R.; Blaclunond, D. G. J. Catal. 1987,105,432. (48) Dubois, L. H.; Somorjai. G. A. Surf. Sci. 1980, 91, 514.
IV. Summary of Results The results of this study are summarized below: 1. K2C03may be used to convert isolated AI-OH groups to Al-OK groups on yA1203catalyst support material. 2. High-temperature treatment of K2C03-treatedcatalysts in the range 570-870 K leads to C03* decomposition at temperatures below the decomposition temperature for bulk K2CO3. 3. The K-A1203 catalyst support material does not exhibit new bulk phases observable by XRD analysis, following calcination at 973 K, suggesting that AI-OK groups are produced on the surface of the A1203. 4. Rh/K-A1203 catalysts exhibit a limited tendency to produce Rh1(C0)2species in the presence of CO(g). However, IR results indicate that Rho, sites remain abundant. This is due to the replacement of isolated Al-OH groups by Al-OK groups, removing the AI-OH as an agent for (l/x)Rho, Rh'(C0)2 formation. 5 . Rh/K-A1203 catalysts are stabilized against hydrolysis processes to produce Rh'(C0)2 at temperatures up to 770 K in H20(g) at 2 Torr pressure. Only a small regeneration of isolated AI-OH groups is observed under these extreme hydrolysis conditions. 6. The results, when compared to other studies involving the removal of Al-OH groups by heat treatment, or by functionalization to produce A14-SiR3 species, strongly support the concept that isolated Al-OH groups are necessary for the degradation of Rho, species in the presence of CO(g), producing Rh1(C0)2. 7. These experiments suggest that oxidative degradation of supported Rh, automotive exhaust catalysts, to produce the catalytically inactive Rh' site, may be extensively diminished alkali-metal functionalization of isolated AI-OH species on the A1203support, producing Al-OK surface species adjacent to Rho, metal particles.
-
Acknowledgment. We thank the Department of Energy, Offce of Basic Energy Sciences, for support of this work. M.I.Z. acknowledges support of a sabbatical visit by the Fulbright Foundation. Registry NO. A1203, 1344-28-1; KZCO,, 584-08-7; Rh, 7440-16-6; CO, 630-08-0.
Surface Assisted Xe-Xe Bonding? Roald Hoffmann,* Meinolf Kersting, and Zafiria Nomikou Department of Chemistry and Materials Science Center, Cornell University, Nhaca, New York 14853 (Received: July 17, 1990)
An idea for forming noble gas atom-noble gas atom bonds on a surface is explored theoretically,for Xe on Pd. The notion is that interaction with the surface should withdraw electrons from the top of the forming Xe 5p band, thus allowing Xe-Xe bonding, truly compressed structures. Molecular orbital studies of Xe overlayers and Xe2 pairs on Pd( 100) and Pd( 11 1) are not encouraging; the X e P d interaction appears insufficient to produce the desired effect.
Surfaces often bind molecules. And the involvement of a surface may lower barriers to chemical reactions. From a chemical point of view, the most interesting surface phenomena occur when a molecule is formed on the surface that has little kinetic or thermodynamic stability in the gas phase (e.g. CCH, CCH3),or when the catalyzed reaction is one that is difficult normally, therefore a desired transformation. It is interesting then to think of molecular species that could exist on surfaces, perhaps only on surfaces. This contribution, speculative and, as it emerges, not all that encouraging, addresses such a problem. 0022-3654191/2095-4033$02.50/0
The context is the formation of bonds between inert or noble gas atoms. The potential energy curve between He (or Xe) atoms is, of course, repulsive, except for the van der Waals minimum.' The reason for this is obvious; the Is orbitals interact as shown in 1. The antibonding uu* combination is raised in energy more than the bonding us one is lowered. Filling of both levels results ( 1 ) See, inter alia: (a) Patil, S. H. 1.Phys. B: AI. Mol. Phys. 1987, 20, 3075. (b) Krauss, M.; Regan, R. M.; Konowalow, D. D. J . Phys. Chcm. 1988, 92,4329. (c) Dias da Silva, J.; BrandXo, J.; Varandas, A. J. C. J . Chem. Soc.. Faraday Tram. 2 1989,85, 1851.
0 1991 American Chemical Society