NOTES Oxygen Chemisorption on Supported Gold

JOURS.\I, 08 <'.\T.\LYYIS 57, 177-182 (1979)

The absorption of O2 on Au has been 4. Studies on high arca supported Au the subject of several investigations, which have not been carried out and it is not covered various forms of Au and widely known how the cffcct of Au particle size, different experimental conditions. A sum- support, and prcparativc condit’ions in- mary of these studies is reported in Table 1. fluence 0, adsorption. From these investigat’ions, the following conclusions emerge : Earlier investigations on the oxygen transfer reaction between CO and CO, 1. Despite the diversity in the experi- showed that some supported Au prepara- mental conditions and the possibility that t.ions displayed high cat,alytic activit,y at in some inst,ances physical and not chemical 330°C (8). More recent studies on the adsorption was observed, there was clear reduction of ?JO by H, catalyzed by sup- indication that a surface Au-oxygen intrr- ported Au indicated a range of the reaction action was present. The lack of detectable selectivity to IX2 from 4 t.0 90% at. 350°C adsorption apparent in some cases (4~) (9). It was suggested that the effect was a was a reflection of the experiment’al condi- consequcncc of an interaction b&wren the tions (low virtual oxygen pressure). The Au particles and the support. degree of surface cleanliness was not In view of the complex behavior of always specified in these studies and may massive Au and t,he lack of information on have contributed to the different chemi- supported Au in oxygen chcmisorption and sorption behavior recorded. of the ready activit)y of supported Au 2. The nature of the adsorption was not’ in catalytic reactions involving oxygen bear- definitively established. There remains a ing molecules, it seemed interesting to in- controversy over whether oxygen adsorp- vcstigatc t’he adsorption of molecular tion is endothermic (6) or exothermic (2). oxygen on supported Au. Such a study, in 3. The adsorption stoichiometry was in- addition to providing confirmation of some vestigated at 12O’C. From measurements of the patterns discovered at surfaces of on t,hc adsorption rate, it was concluded massive Au, would be helpful in assessing that at low surface covrragc four surface the potential of oxygen adsorption as a sites per 02 molecules were involved (a). characterization method for the size of Assuming that Au atoms are the adsorption the supported Au particlcs. sites, a stoichiometry corresponding to Au20 was derived. However, as the surface Following these considerations WC have reached saturation, a restructuring of the conducted a series of measurements on adsorbate took place and a stoichiometry the adsorption of Au supported on MgO, corresponding to Au0 was found to be more SiOs, and A1203 in the temperature range representative of the experimental results. 170 to 450°C and partial pressures of

177 0021-9517/79/040177-06$02.00/O Copyright Q 1979 by Academic Press. Inc. All rights of reproduction m any form reserved. 17s NOTES

TABLE 1 Oxygen Adsorption on Au

Type of Au Pressure of 02 Temperature Characteristic behavior Ref. (Torr) (“C)

Powder 100 to 760 98 to 157 Amount adsorbed increasing with temperature 0) Powder 0.2 120 Adsorption heat decreasing with increasing (.2) surface coverage Powder 1 -89 to 450 Isobar shows maxima at -50, 210°C (3) Powder 3 x 10ea 295 to 380 No detectable adsorption (44 Wire 0.2 50 to 400 Bulk incorporation at T > 200°C (5) Foil 10-T 25 to 700 Low surface coverage; isobar shows maximum (6) at 427°C Film 10-2 25 to 405 Irreversible chemisorption ; no incorporation VI Film 10-z to 10” - 183 to 0°C Physical adsorption; no chemisorption (4b)

02 in the range of 0.4 to 6.0 Torr (1 used were : SiOZ (400 mz/g), A1203 (180 Torr = 133.33 N m-2). m”/g), and MgO (12 m”/g). Preparative procedures, composition and thermal treat- EXPERIMENTAL ments prior to adsorption are summarized Materials. Reagent grade HAuCL, in Table 2. Au preparations were charac- KAu(CN)z, and Au(en)&ls (IO), wereused terized by wide-angle X-ray scattering as precursor salts. Ultra high purity Hz and (WAXS) and transmission electron micros- 02 (> 99.995y0) were employed. Supports copy (TEM). WAXS determinations were

TABLE 2 Supported Au : Composition, Preparation, and Adsorption Pretreatment

Support Preparative method Adsorption pretreatment

SiOg Impregnation with HAuCl,, 3OO”C, 1 hr, in air, 70 Torr; Hz reduction, 4OO”C, 2 hr 4OO”C, 1 hr, in HZ, 70 Torr; evacuated at 45O”C, 30 min SiOg 1.24 Cationic exchange with Same Au(en)z& Hz reduction, 1 atm, 4OO”C, 2 hr SiO9 1.30 Same as 1.24 wt% Same SiOf 3.11 Impregnation with KAu(CN)z, Same 35O”C, 2 hr, 0.1 Torr

MgO 2.00 Impregnation with HAuClr, 1O-4 Torr, 35O”C, 8 hr, reduction with oxalic acid, cooled to 300°C 350°C 2 hr, 0.1 Torr MgO 3.46 Impregnation with HAuClr, Same as Au/SiOz reduction with HZ, 1 atm, 2 hr, 300°C; 2 hr, 400°C

Al203 4.25 Impregnation with KAu(CN)z, Same as Au/SiOz 350°C 2 hr, 0.1 Torr

L1By atomic absorption and neutron activation. NOTES 179

soaked in amylacetate to dissolve the parlodion film prior to examination. Magni- fication was calibrated using the (002) planes of graphitized carbon black. Mea- surements of Au particle sizes and size distribution were made with micrometer eye- piece from enlarged prints. Size distribu- tions had the typical skewed shape (Fig. 1). Oxygen adsorption was measured in a conventional, all glass, volumetric system with a Televac thermocouple manometer calibrated on oxygen gas. Linear adsorption isotherms were found on the pure supports; upon extrapolation to zero pressure, there was no evidence of chemisorption. There- fore no correction for support was necessary when calculating the amount chemisorbed on the supported Au. The accuracy of the adsorption technique was estimated at fO.l pmole/g sample. Particle diameter from oxygen adsorption was calculated 4-o 80 120 160 using t’he relation d = G/Sp, where S is 6) the surface area and p the density.

FIG. 1. Sixe distribution of supported Au particles RESULTS in 1.24 wt% Au-SiOn by Transmission Electron Microscopy. Since preliminary runs indicated that no detectable adsorption took place on carried out in an X-ray diffractomet’er, hav- each sample investigated at temperatures ing a resolution of 0.005”. The crystallite <17O”C (in the course of 24 hr of observa- size, d, was calculated by means of the tion), all subsequent measurements were Scherrer expression aft’er correction for the carried out at temperatures >17O”C. instrument contribution (11), cl = SOX/aB+ Typical results for SiO, supported Au are cost9 [A]. (la), where X is t’he CuKa wave- reported in Fig. 2 for various initial 02 length, 0 the diffraction angle, and B+ the part,ial pressures. In all instances, the width of half the height of the diffraction adsorption showed a fast, initial period peak, usually those from the (200) planes. follow-cd by one of slew or negligible ad- Elect,ron microscopy size measurements sorption. At 200°C and in the lower range were conducted in a high-resolution trans- of O2 prcssurcs (2.94 Torr), initial rapid mission Siemens Elmiskop Electron Micro- adsorption led to surface sat’urat,ion and no scope. Catalyst samples for the microscope further adsorption could be detected even after long periods of time. At higher 02 mere ground in a mullite mortar and pestle; pressures (4.78 and 5.19 Torr) rapid ad- the fine powder was sprinkled onto a dry sorption was followed by a much slower, prepared parlodion substrate on microscope but easily detectable, adsorption. After grids. The film was then placed in a vacuum 2 hr at 200°C, an additional 50y0 pressure evaporator and coated with about 100 A decrease was observed with no indication of evaporated carbon. The sample grids of a ready saturation. It was assumed that were then placed on adsorbing paper the first rapid process was an indication of lS0 NOTES

monolayer chemisorption, while the second one corresponded to diffusion onto the ceramic support. By extrapolation to zero time of the linear portion of the isotherms, the oxygen uptake corresponding to the fast initial period was calculated. A typical plot of the oxygen uptake versus Po2 is reported in Fig. 3 as a function of Po2 for 1.24 wt$!& Au-SiOz. By means of the values of the saturation O2 uptake, calculated with the aid of the above procedure, the average diameter of the Au particles was computed and it is reported in Table 3, together with the results of size measurements from WAXS and TEM. In these calculations an atomic surface density in polycrystalline Au of 1.15 X 1015atoms/cm2 was employed, and the assumption of a spherical particle shape was used. Inspec- Hours tion of Table 3 shows that for adsorption FIG. 2. Adsorption of 02 on 1.24 wt% Au-SiOt, at 200°C the Au/O = 2 stoichiometry yields 200°C; 0, PO,, 2.94 Tom; 0, Paz, 4.78 Torr; A, PO,, results more consistent with particle size 5.19 Torr; 0, 1.30 wt’% Au-SiOs, 17O”C, Pal, calculated from WAXS and TEM. As 3.1 Torr. mentioned previously, the Au/O = 2 stoichiometry was also found in earlier in- sional “rafts” of a few atoms are imbedded vestigations to better represent the adsorp- in the support. Since in these cases O2 tion on unsupported Au powder (2). Upon chemisorption indicates an average value increasing the temperature of adsorption of particle size that would not represent on supported Au to 3OO”C, the Au/O = 1 realistically the state of the Au, a direct stoichiometric ratio gave results more con- comparison between Au prticle size derived sistent with the WAXS and TEM size from oxygen chemisorption with that from WAXS and TEM would be difficult to calculations. justify. The agreement between size measure- From the results reported in the litera- ments from WAXS, TEM, and 02 adsorp- ture (Table 1) the impression is generated tion is considered satisfactory, since the that the heat of O2 adsorption on Au may highest discrepancy between them is by a vary between wide limits. If the adsorption factor of 1.5. The only exception is 4.25 wt% Au-A1203, which shows a discrepancy by a factor of 4.2. The techniques employed in this study for particle size deter- mination involve several assumptions. Fur- z 200 t 0 thermore, some supported Au samples were found to display a complex morphological 2 0 0 3.150 and chemical behavior (I$), including the simultaneous presence of a dispersed phase :M of widely different ranges of particle sizes Torr (from > 30 to > 2000 A) and a “dissolved” FIG. 3. Oxygen uptake versus po,, on 1.24 wt% phase in which single atoms or two-dimen- Au-SiOn, 200°C. NOTISS 1Sl

TABLE J Diameter of Supported Au Particles

Sample Temperature Au particle diametera composition (“C) (Q (wt”/ol 02 ads TEM WAXS

Au Au -=l -=2 0 0

0 .v54 Au-SiO 2 200 1340 670 - 450 300 500 250 1.24 AuSiOn 200 176 88 200 122 61 60 40 200 172 86 200 144 72 1.30 Au-SiO 2 200 118 59 85 - 170 144 72 3.11 Au-SiOz 200 2600 1300 900 2.0 Au-MgO 200 84 42 80 55 300 72 36 3.46 Au-MgO 200 174 87 83 4.25 Au-A1203 200 1280 640 220 150

a &lOy, O2 adsorption, f2O7, TEM, WAXS.

heat is impurity controlled, it should be could not be desorbed at the same tempcra- quite temperature dependent. To assess the ture; rather reacting it with Hz was neces- extent of irreversibly adsorbed versus sary to free the surface (1). On Au films at reversibly adsorbed oxygen, adsorption- temperatures as high as 405”C, the adsorp- desorption experiments were carried out on tion of 02 was found to be irreversible (7). 2.0 wt% Au-MgO. It was found t,hat at The adsorption isobar showed a maximum 200°C all of the OS was irreversibly ad- (Fig. 4) at about 340°C. Variation in sur- sorbed, while at 400°C all of the O2 could face stoichiometry and/or endothermic dis- be desorbed. This is reminiscent of earlier solution of oxygen, as discussed above, may investigations on unsupported Au powder be postulated to explain the observed showing that O2 chemisorbed at 130°C maximum but., clearly, other explanations may be suggested. More extended investi- 500 gations of the effect are necessary for its t definition. Earlier, a maximum was observed at 210°C in the 02 adsorption on Au powder (3). Our results are more consistent with the maximum at 427°C ob- tained on unsupported Au powder (6) and we concur with these authors that cndo- thermic adsorption may be a determining factor in the maximum and a fundamental FIG. 4. 02 adsorption isobar for 2 wt% Au-MgO link between supported and nonsupported as a function of temperature; poz, 27 Torr. Au surfaces. 182 NOTES

CONCLUSIONS Chem. Sot. 78, 4533 (1956); (b) Trapnell, B. M. W., PTOC. Roy. Sot. A 218, 566 (1953). The adsorption of oxygen at 200°C and 6. Kulkova, N. P., and Leveshenko, L. P., Kinet. initial oxygen pressure of 3 to 6 Torr is a Catal. 6, 688 (1965). suitable method to assess the surface area 6. Endow, N., Wood, B. J., and Wise, H., J. Calal. of supported Au. Under these conditions 15, 316 (1969). 7. Richardson, P. C., and Rossington, D. R., J. the correct stoichiometry corresponds to Catal. 20, 420 (1971). Au/O = 2. There are indications that the 8. Cha, D. Y., and Parravano, G., J. Catal. 18, 200 extent and strength of the adsorption are (1970). sensitive to surface contamination of the 9. Galvagno, S., and Parravano, G., J. Catal. 55, Au. As the temperature is increased from 178 (1978). 200 to 4OO”C, the adsorption becomes re- 10. Block, B. P., and Bailar, J. C., Jr., J. Amer. Chem. Sot. 73, 4722 (1951). versible, while the isobar shows a maximum 11. Lipson, H., and Steeple, H., “Interpretation of at 340°C. These effects are qualitatively X-Ray Powder Diffraction Patterns.” St. consistent with the adsorption behavior Martin’s Press, New York, 1970. recorded at surfaces of massive Au. ld. Neff, H., “Grundl. Anwendung der Rontgen- feinstrukturanalyse,” Oldenbourg, Miinchen, 1962. ACKNOWLEDGMENT IS. Bassi, I. W., Lytle, F. W., and Parravano, G., J. Catal. 42, 139, (1976) ; Cocco, G., et al., We wish to express our appreciation to Dr. V. unpublished results. Mormino for the results on Au-MgO. The work was carried out with the financial support of the T. FUKUSHIMA National Science Foundation Grant Eng 75-14193 S. GALVAGNO’, 2 and Montedison S.P.A. This support is gratefully acknowledged. G. PARRAVANO~ Department of Chemical Engineering REFERENCES The University of Michigan Ann Arbor, Michigan 48109 1. Benton, A. F., and Elgin, J. L., J. Amer. Chem. Sot. 49, 2426 (1927). Received February 17, 1978 2. Dobrovol’skii, N. N., Ostrovskii, V. E., Raba- shov, A. M., and Temkin, M. I., Dokl. Akad. 1 On leave from Montedison S.P.A., Novara, Nauk SSSR, 183, No. 5, 1120 (1968). Italy. S. McDonald, W. R., and Hayes, K. E., J. Caatal. 2 To whom inquiries about this note should be 18, 115 (1970). addressed. 4. (a) Gonzalez, 0. D., and Parravano, G., J. Amer. 3 Deceased on April 1, 1978.