J . Chem. SOC., Faraday Trans. I , 1987,83, 1369-1 380 Aspects of Temperature-programmed Analysis of some Gas-Solid Reactions Part 2.-Hydrogen Temperature-programmed Desorption from Silica-supported Platinum Mariana S. W. Vong and Paul A. Sermon" Department of Chemistry, Brunel University, Uxbridge, Middlesex UB8 3PH Well resolved temperature-programmed desorption spectra have been ob- tained for hydrogen preadsorbed upon silica-supported platinum; these reveal three maximum rates of desorption (B1 at Tmaxl, p2 at Tmaxz and p3 at Tmax3). The total extent of desorption of hydrogen exceeds the amount of hydrogen observed to chemisorb in extrapolation of adsorption data at ambient temperature to zero pressure. As the Pt dispersion increases, the amount of hydrogen associated with the p3 peak becomes smaller than the sum of the (D1+p2) peaks.This may contradict suggestions that integral enthalpies of hydrogen chemisorption on supported platinium are indepen- dent of its average particle (Sen et al., J. CataE., 1986, 101, 517). Therefore the temperature required for desorption of a given fraction of adsorbed hydrogen is observed to decrease as the Pt surface area increases and the average Pt particle size decreases, possibly indicating an increase in the concentration of sites binding hydrogen with lower energy. The p, peak may arise from desorption from distinct spillover sites upon the support, but the and p2 peaks appear associated with Pt-held hydrogen, and interdiffusion between these means that and p2 peaks cannot be taken to indicate the populations of different surface Pt sites; indeed Pt surface areas estimated from the sum of the (D, +p2) peaks can exceed those determined by hydrogen chemisorption.However, Pt surface areas estimated from hydrogen t.p.d. to the temperature (here 573 K) to which catalysts are evacuated prior to volumetric measurement of chemisorption do agree well with chemisorption measurements. There is no consistent trend of desorption activation energies with Pt dispersion (as reported by Takasu et al., J . Chem. SOC., Chem. Commun., 1983, 1329) or t.p.d. peak temperature (as reported by Foger and Anderson, J. Catal., 1978, 54, 318 and Anderson et al., J. Catal., 1979, 57, 458). A recent consideration' of temperature-programmed bulk reduction (t.p. b.r.) of oxide- supported hexachloroplatinic acid and subsequent temperature-programmed desorption (t.p.d.) of chemisorbed hydrogen has suggested that diffusion may be important in defining the profiles and maxima seen.Thus in t.p.b.r. the rate of anion diffusion in the bulk solid particles to be reduced may be too slow to allow good resolution of t.p.b.r. peaks arising from the reduction of distinct homogeneous phases. In t.p.d., surface diffusion of adsorbate prior to desorption may allow significant randomization and make it difficult to associate particular t.p.d. peaks with the occupancy of distinct surface sites. This point is considered using detailed analysis of temperature-programmed desorption of hydrogen preadsorbed upon silica-supported platinum. In addition attention is given to the relationship between the extents of hydrogen desorption in different t.p.d. peaks and the amounts of hydrogen found to chemisorb upon these catalysts, and how the size of such t.p.d.peaks varies with the Pt dispersion. 13691370 Hydrogen T.P.D. from Supported Platinum Table 1. Properties of silica supports used most pore volume frequent /cm3 g-l pore SN2 radius supporta /m2 g-l Hg H2O /nm catalysts SiO, (D70) 321.9 1.1 1.28 4.4 A, E, H, J SiO, (D923) 534.9, 517.0 out of 1.10 3.75 D range 6 K Sorbsil AQU30 307.0 0.41 - ~~ ~~~~~ a D70 and D923 denote Davison 70 and 923 silicas; Sorbsil AQU30 denotes a silica support. Table 2. Supported adsorbents and catalysts wt % method of conditions of samples metal preparationa drying/reduction A 5.0 (4.4) IMP 393 K, 24 h/373 K, 2 h D 5.0 (4.3) IMP 393 K, 24 h/373 K, 2 h E 5.0 (3.9) IMP 393 K, 24 h/373 K, 2 h H 2.4 (1.5) IE sintered J at 873 K/6 h in H, J 2.4 IE 295 K (vac)/573 K, 3 h K 6.2 (2.4) CE 378 K, 16 h/673 K, 0.5 h a IMP, IE and CE denote samples prepared by impregnation and ion exchange methods and loaned by the Council of Europe group on catalysis.wt% Pt are those predicted from the preparation, and data in brackets are concentrations estimated by atomic absorption. Experimental Materials Flow desorption measurements used hydrogen (99.9 % purity, BOC) and nitrogen (white spot 99.9 % purity, BOC). The hydrogen stream was further purified by passage through 1% Pd/H,WO, (5 g) and then a trap at 195 K, thereby removing both 0, and H,O. The nitrogen stream was similarly purified by passage through 44.8% MnOJCelite (10 g) and then a trap at 195 K, leaving residual concentrations of < 1 ppb 0, and 0.87 ppm H,O.Table 1 indicates the supports used together with their principal properties. Pt was introduced to these (except H, J and K) as indicated in table 2 using aqueous solutions of H,PtCl, (Specpure, Johnson Matthey) to produce supported catalysts which have been studied here., Methods Total surface areas of supports and catalysts were estimated by application of B.E.T. theory to N, adsorption measured at 77 K. Pore volumes and pore size distributions were measured using a Carlo Erba porosimeter. Absorption spectrophotometry (Perkin-Elmer 303) was used with an air-ethyne flame to analyse for platinum at 265.9 nm after calibration with standard solutions (0.1-100 ppm) prepared by dissolving platinum black (98.84% purity) in aqua regia, filtering and making up to 0.1 dm3.Oxide supports alone before and after adsorptionM. S. W. Vong and P. A . Sermon 1371 Table 3. Platinum surface areas and average particle sizes estimated by chemisorption and physical methods2 sample -_____ 8.27 27.56 27.60 44.70 54.40 166.0 average of all chemsn data2 S /m2 g,: ____ 16.93 52.50 53.60 203.6 249.5 256.1 dpt/nm by TEMa dp,/nm by XRDb d, 16.62 5.38 5.33 1.35 1.15 1.09 7.30 11.20 12.35 5.37 8.37 10.10 5.02 8.14 9.45 ND ND ND ND ND ND ND ND ND 10.5 9.4 5.8 6.8 5.3 5.9 ND ND ND ND ND ND a TEM yielded mean number, mean surface and mean volume diameters of Pt particles dn, ds and dv, X-Ray camera and diffractometer measurements of 11 1 reflection line broadening.ND indicates that Pt particles in this sample were not detected by this technique. and t.p.d. experiments showed no significant Pt concentration detectable by AA (i.e. their platinum content was < 2 ppm). Transmission electron microscopy was applied to the silica supports and catalysts using a Jeol lOOC with microtomed sections of samples embedded in epoxy resin. The sizes of several hundred Pt particles were estimated in each sample. X-Ray diffraction line broadening (XRDLB) was followed by camera-microdensitometer (PW 1024/ 10 camera, Cu K, radiation, PW 1008 generator, Mk3CS microdensitometer) and horizon- tal diffractometer methods. The extents of H,, CO and 0, chemisorption were measured at 295 K volumetrically.2 Temperature-programmed desorption (t.p.d.) of H, was followed with samples (0.5 g) of catalysts flushed with N, (101 kPa, 60 cm3 min-l, 15 min, 293 K) and then H,.T'ae sample temperature was then raised (LMVS 100 Stanton Redcroft low-mass vertical furnace and LVPCA4R Stanton Redcroft temperature programmer) from 295 to 423 K, where it was held in H, for 1 h. The sample was then purged with N, (60 cm3 min-l, 1 h, 423-573 K). It was then allowed to adsorb H, at 101 kPa and 423 K and finally 273 K, and the sample was then flushed with N, (60 cm3 min-l, 30 min) at 273 K until no gaseous H, was detected when samples of the gas stream were injected into an HWD gas chromatograph (Perkin-Elmer F17 calibrated with samples of 6% H, in N,, fitted with a molecular-sieve column at 323 K through which flowed an N, carrier gas using a 5 mm3 gas-tight syringe Hamilton 1750 RN).The N, flow rate was then set and held at 15 cm3 min-l and t.p.d. commenced by raising the temperature of the samples at 4 K min-l from 273 to 900 K while the concentration of H, in the exit gas stream was monitored and plotted as a function of time and temperature. Sample temperatures were measured using an adjacent chromel- alumel thermocouple. Results Characterisation of Supported Platinum Table 2 shows that the Pt concentrations extracted from catalysts were less then the amounts introduced during preparation, especially in samples H and K. However, even after extraction such samples were greyish, indicating the retention of traces of platinum.In all calculations the platinum concentration estimated from the support impregnation1372 Hydrogen T.P.D. from Supported Platinum Fig. 1. Histograms of the sizes of Pt particles supported upon silica in samples A (-), D(----) and E (--.--.--. ) measured from transmission electron micrographs. (catalysts A, D and E) or the concentration supplied by others (catalysts H and I) or the concentration determined2 by atomic absorption (catalyst K) was used. The extents of H, chemisorption on each sample estimated by volumetric methods2 are given in table 3, together with average surface areas and average particle sizes of Pt calculated2 from CO, 0, and H, chemisorption. The adsorption stoichiometry H : Pt, = 1 and the average number of surface Pt atoms adsorbing an H, molecule irrespective of the extent of adsorbate dissociation nH2 = 2 were assumed.Catalysts were assumed to contain spherical crystallites of Pt of uniform size dpt, where the Pt surface area is given by S = 6000/dptppt bPt = density of platinum). Platinum catalysts H, J and K (with an average platinum crystallite size dpt < 2 nm estimated by chemisorption) did not show any X-ray diffraction patterns as a result of excessive line broadening (XRDLB). However, a platinum crystallite size of 1.34 nm determined by XRDLB in catalyst K has been rep~rted.~ The width of the x.r.d. peaks corrected for instrumental broadening using the Warren correction and an aluminium standard with > 200 nm crystallites were converted to average platinum crystallite sizes dpt, using the Scherrer equation (d = KA/pcos8).Table 3 shows that for catalyst A Pt particle sizes obtained by XRDLB are in reasonable agreement with those from electron microscopy. However, for samples D and E the Pt particle sizes obtained by XRDLB agree well with those from hydrogen chemisorption. It appears that gas adsorption and XRDLB (where applicable) give estimates of the average crystallite size of supported Pt which are in moderate agreement for the series of supported Pt catalysts studied here. However, it must be remembered that most catalysts contain a relatively broad particle-size distribution, revealed by transmission electron micrographs of all catalysts except those of greatest Pt dispersion (i.e. H, J and K). Although electron microscopy of catalyst K has shown3 that 78 % of Pt particles were < 2 nm, micrographs of the other catalysts showed the Pt particles to be almost spherical or hemispherical and of a wide range of sizes (see fig.1). The calculated number-average Pt diameter d, (= Xnidi/ni>, the surface-average Pt diameter d, (= Cnidf/nidf) and the volume-average diameter dv (= Cnidf/nid:, where ni is the number of Pt crystallites with diameter di) were calculated from micrographs and are given in table 3. Average diameters of Pt crystallites determined by chemisorption and XRDLB are also tabulated. The values of as and &, from electron microscopy were higher than the d values determined by hydrogen chemisorption and XRDLB.A 1373 H ri D P2 00 0 12 ri 4 0 0 0 0 8 2 0 0 P3 E 0 0 0 0 "a 0 0 0 0 ; 0 I I 0 0 0 0 0 ri 0 0 0 0 0 I 01 I 40 80 120 160 tlmin 0 p1 z (32 0 0 0 0 0 0 O O O O b 0 ' 0 O 1 0 O 0 0 J 0 ?J% 0 1 1 1 1 3 2 1 ,8.P1 K tlmin Fig. 2. Profiles of temperature-programmed desorption for hydrogen preadsorbed on different silica-supported Pt samples (A, D, E, H, J and K) as their temperature is raised at a constant rate dT/dt (4 K min-l) from 273 to 900 K. The rate of desorption is given as dn/dt or ri in lo1' H atom per g catalyst per min.1374 Hydrogen T.P.D. from Supported Platinum Table 4. Results of temperature-programmed desorption of H, preadsorbed on silica-supported Pt samples H atoms ( x desorbed per g catalyst sample 6 573 K PI P2 P 3 total A 2.2 1.04 2.39 5.79 9.22 D 2.5 1.87 2.13 1.61 5.02 E 3.4 2.80 2.39 7.68 12.78 H 4.9 4.49 0.91 0.80 6.20 J 7.6 7.10 3.10 4.80 15.00 K 14.33 10.80 4.60 14.13 2.10 4.35 20.60 H atoms ( x per g Pt S/m2 per g Pt chemi- t.p.d.t.p.d. chemi- sample (Dp,+P,) d 573 K sorption (D1+P2) d 573 K sorption A 68.6 44.0 19.8 54.88 35.20 15.94 D 80.0 50.0 66.1 64.00 40.00 53.14 E 102.0 66.8 66.7 8 1.60 53.46 53.60 H 220.0 199.6 223.5 176.0 159.7 179.6 J 425.0 316.7 272.0 340.0 253.3 218.5 K 261.8 231.1 321.6 209.4 209.4 261.1 198.7 Overestimation by TEM could arise from aggregation of small metal particles into larger ‘apparent’ crystallites and also because Pt particles smaller than 2 nm which chemisorb hydrogen may not be visible in TEM. For these catalysts hydrogen chemi- sorption probably measures the total platinum surface areas satisfactorily.Temperature-programmed Desorption of Hydrogen Temperature-programmed desorption (t.p.d.) profiles (273-900 K) obtained for hydro- gen preadsorbed at 273 K and 101 kPa for 30 min on silica-supported platinum (see fig. 2) show the rate of hydrogen dn/dt or n desorbed as a function of time and hence temperature. In general, three different maximum rates of chemisorbed hydrogen on platinum were indicated by peaks appearing in the t.p.d. spectra at specific temperatures (Tmax) at ca. 347-397, 573-633 and 693-786K. These peaks were designated as /I1, /Iz and /I3, respectively. The quantity of hydrogen desorbed was estimated by integrating the area under each of the peaks, and the results so obtained are given in table 4. Overlapping peaks were separated by dividing these vertically at minimum rates of desorption; uncertainties in t.p.d.peak areas were & 15%. In the measurements of the extent of hydrogen chemisorption,2 samples (see tables 3 and 4) were pretreated in that they were reduced in H, at 423 K and 13.33 kPa and then evacuated at 573 K and 1.33 mPa before the isotherms were measured at room temperature. By assuming that the hydrogen desorbing above 573 K in t.p.d. (i.e. associated with /Iz and /I3 peaks) was not removed by evacuation in adsorption experiments, the amount of hydrogen taken up by the sample in the subsequent adsorption process should be comparable to the amount of hydrogen desorbed from the catalyst between 273 and 573 K (i.e. associated in part with the /II t.p.d. peak). Table 4M .S. W. Vong and P. A . Sermon 1375 Table 5. Extents of hydrogen chemisorption at ambient temperature and desorption" per g platinum H atoms ( x 1019) desorbed per g Pt % (D1+82) adsorbed per g Pt H atoms ( x of total desorbed 0 kPae 10kPa 100kPa sampleb +P2 total - - 22.58 (22.58) 22.58 30.15 112.55 28.00 110.40 A 68.60 184.40 37.2 D 80.00 100.40 79.7 66.40 85.68 259:15 66.50 156.37 965.21 (67.46 157.33 966.1 7 E 102.00 255.6 39.9 - - 74.95 110.73 432.75 F 33.49 L 30.37 G 64.55 51.97) - 60.14 H 220.00 258.30 J 425.0 625.00 K 248.39 322.3 261.80 - 215.64 - 228.38 263.20 - - 89.60 - 87.1 223.35 225.40 272.02 68.0 (274.07 74.8 325.38 78.8 323.60 - 330.25 - 314.79 - 322.62 305.87 117.67 323.70 325.75 419.1 1 421.16 - 433.33 441.41 424.41 370.30 1226.87 1228.92 1742.92 1744.97 - 1500.22 1508.00 1491.00 a 'T.p.d. results for samples L, F and G have been reported.' previously.2 Determined by extrapolation of chemisorption data.Samples have been described shows that the agreement with chemisorption is fair and far better than any other t.p.d. assessment previously used (i.e. areas, p1 +p2 areas etc.). This is novel, important and potentially very useful. Discussion The hydrogen t.p.d. results obtained here are more detailed and cover a wider range of Pt particle sizes than previously. Here /I1 and p2 in the present work may be assumed to result from hydrogen atoms chemisorbed on the platinum because (i) extents of chemisorption (after evacuation at 573 K) at ambient temperature and 'zero pressure' and the fraction of the @,+p2) t.p.d.peak areas desorbed up to 573 K are in fair agreement, and (ii) the B1 and Bz t.p.d. peaks appeared at the same temperatures as the y and 6 t.p.d. peaks detected4 on Pt black, and are also in reasonable agreement with t.p.d. on Pt single crystals5 and B2 peaks from Pt film.6 Although strict comparison of the present t.p.d. results with those in the literature is difficult because of differences in methods of preparation, Pt dispersion, supports and the adsorption-t.p.d. conditions, the present p3 t.p.d. results also agree well with those reported on supported platinum catalyst^.^? Thus T,,, for the p3 peak was very close1376 Hydrogen T.P.D. from Supported Platinum to that for the desorption peak (at 753 K)g of hydrogen from platinum-free alumina exposed to dissociated hydrogen atoms in a high-frequency discharge.It is also close to that identified for hydrogen on platinum supported upon silica, alumina and zeolites.l'-12 Furthermore, table 5 shows that the total size of the (al+p,+/3,) peaks is too large to be accounted for in terms of desorption of Pt-held hydrogen alone, although this quantity does vary significantly with adsorption pressure. However, no desorption peak was detected in the t.p.d. of Pt-free supports alone, indicating that the b3 peak was not dehydroxylation or dehydration of the support surface. The concentration of OH groups on the silica surface has been estimated13 to be 0.2 OH per nm2 or 0.6 x 1020 OH per g Davison 70 silica. If significant dehydroxylation had occurred the amount of hydrogen desorbed would have been insignificant.Hence it is concluded that the p3 peak detected in the present work is associated with spilt-over hydrogen? which migrates back from the silica support to platinum and then desorbs therefrom. Differences in the p3 size with various samples might be due to the varying extent of interaction between Pt and the oxide support. Future studies on the effect of adsorption parameters on the t.p.d. profiles may provide more information on the effect of metal and support interaction. T.p.d. profiles of catalysts E, A and D, which had relatively large Pt particles (i.e. 5.3, 16.6 and 5.4 nm, respectively) exhibited a singlet B1 peak (between the doublet p1-p2 states of catalysts H, J and K) and then a larger peak (or peaks) at higher temperature.T.p.d. profiles of highly dispersed catalysts K, J and H (dpt of 1 . 1 ? 1.2 and 1.4 nm, respectively) showed the peak to be split into a doublet with the first maxima at 373,363 and 347 K and the second maxima at 443,432 and 410 K, respectively. As dpt decreases, the fraction of Pt-held hydrogen released as PI increased. The results show a less clear trend for the temperature of the profile maxima ( Tmax) with dpt, but certainly do not show an increase in Tmax with decreasing dpt as lo Furthermore, the relative contribution of the second peak in the doublet became more significant as dp, decreased. With t.p.d. from these silica-supported Pt samples at the low heating rates used it is not believed that mass-transfer 1imitationsl4 are significant. Nor is there any evidence that desorption is from a non-metallic active phase (e.g.formed uiu Pt-Si-0 linkages15), especially since only very modest temperatures of pretreatment and desorption have been used; indeed the porous supports may have reduced contamination of the supported Pt surfaces.16 Nevertheless, with silica at high temperatures or with other supports interactions might be far more significant. Conclusions The decrease in the temperature of the desorption maxima as the platinum particle size decreased could be due to a decreasing presence of higher-energy binding sites for hydrogen. However, an increase in the proportion of relatively low coordination number platinum atoms as the size of the particle becomes smaller might be lo There is evidence that platinum atoms at corner, edge and kink1' sites are relatively electron- deficient.In addition, theoretical calculationsls of the hydrogen-binding energy have shown that these sites adsorb hydrogen more strongly. This suggests an increase in Tmax with decreasing dpt should be expected (and seen8, lo for supported Pt); this is the reverse of the t.p.d. results here. Nevertheless, for these samples the enthalpy of chemisorption of hydrogen at 5-6 kPa has been found1 to decrease from 91.99 to 60.60 kJ mol-1 H, as dpt decreased from 3.59 to 1.09 nm. The causes of these different trends with dpt remain uncertain, but it is clear that the extent of adsorption-desorption measured here for sample K agrees with previous estimatesll and that the fraction of preadsorbed hydrogen which is subsequently desorbed at a particular temperature increases as the average dpt decreases.Possibly as a result of the use of lower rates of temperature-programming than previously (i.e. 20 K min-1)8, lo, l9 rather than stepped-temperature desorption, theM. S. W. Vong and P . A . Sermon 2.0 3.0 4.0 1 1 1 1 1 1 1 ~ 1 ~ ' ~ ~ - "? - PP i: '0 - \ 5 - - t Oe * 3.0 ~ j - - \ $$P* '> 00 4 " 0 3, - -Og$"Q: " 0 d\$ - 0; 0 Q 0 e i 0 o f - p o r! or: ip? " 0 @,, CI 2.0- ,! ""t, , Q 0 ' - .: - 9 0 ' :a 0: - 9 '4 8 'e - e: '>QQ 0 \ ""0 O b '\ 0 Gb '\,O B 0 QB? '\ O 8 - - , - - Q 0 ' " 8 - I , \ - %i - 9 - O'\\, - l o"',, @a - *4 O\'$, "\ Q "b 1.0 - I - "0 - I @? Oo \\ - 0 ) 4 1 1 9 1 I 1 1 I I I I I I . - 2.0 3.0 4.0 1377 3.0 'L -2.0 E 1.0 resolution of the hydrogen desorption profiles shows more detail and not a rather broad t.p.d. peak8? lo, lS with poorly resolved shoulders which are difficult to separate even by deconvolution.1° Instead, at least three different moderately well resolved t.p.d.peaks are seen ; and b2 tentatively associated with dissociated hydrogen atoms chemisorbed on platinum and P3 with hydrogen spilt over on to the silica oxide support. Insufficient evidence exists to conclude whether the B1 and b2 peaks arise from hydrogen on binding states whose energetics are defined by surface sites of different local crystallographic properties.12 Nevertheless, it must be remembered that none of the silica-supported Pt samples studied here contained a narrow range of Pt particle size, hence to look for particle-size effects of their t.p.d.profiles cannot strictly be warranted. There can be many binding states of adsorbed molecules on a given surface and AH,,, may differ by as much as 82 kJ mol-l. The existence of several binding states indicates that even a single crystal plane is heterogeneous when viewed by the adsorbed species. These sites of different adsorptive strengths are filled with increasing coverage or with changes in other experimental variables. Clearly the choice of programming rate is very important, as suggested.l T.p.d. peak breadths did not appear dependent upon dpt as predicted elsew here. Consider for a moment whether B1, B2 and P3 peak areas really correspond to the 46 FAR 11378 Hydrogen T.P.D. from Supported Platinum Table 6.Activation energiesa of desorption of hydrogen from supported platinum measured from plots in fig. 3 when dt+O and [S] or hydrogen coverage is essentially constant E,/kJ mol-1 (at T,,,/K) sample El for Dl E2 for 8 2 E3 for 83 A 19.38/408.2 77.321582.7 36371692.2 D 13.761393.2 32.111594.2 - 1739.2 E 21.63/388.2 71.501597.2 26.581708.5 H 38.45/347.2 - /410.2 - /786.2 J 20.11/363.7 50.471432.2 - 1694.2 K 21.901373.2 - 1443.2 - 1755.2 a Desorption activation energies for H2 from Pt/Si02 were previously measured19 over a wide range of temperature. populations of energetically different surface sites. The rate of hydrogen desorption from i surface sites will be given by if readsorption is negligible and the rate constant of desorption does not vary with surface-site occupancy20 at small dt.Hence i can be 1, 2 or 3 (i.e. for #I1, p2 or p3 here) and [S], is the concentration of the particular discernible occupied surface sites with specific energetics of adsorption corresponding to one t.p.d. peak; s is the kinetic order with respect to this surface site concentration. Therefore for any one t.p.d. peak i a plot of In (dnldt,) (or Inn,) us. l / T at essentially constant site occupancy and [S], for a particular surface site [i.e. for a small increment of time (dt + 0) or temperature] should allow Ei for desorption of that species to be deduced. Fig. 3 shows such plots for the hydrogen desorption peaks #I,, #Iz and p3 for Pt/SiO, used here. Activation energies El, E2 and E3 so derived are given in table 6.The average activation energy (35.80 kJ mol-l) for hydrogen desorption in B1, #I2 and p3 peaks from silica-supported platinum shown in table 6 is fortuitously close to Edesorption (37.66 kJ mol-l) seen previously for H, from Pt.,l As expected, these are far greater than the activation energy for diffusion of hydrogen on the Pt surface (18.85 kJ mo1-1).22 It is clear that such activation energies do not decrease in the sequence p3 > a, > #I1 (or its reverse) for all samples, nor do they change consistently with dpt for one t.p.d. peak for samples of different Pt dispersions. Previously it has been reported19 that Ei decreases from 50 to 34 kJ mol-' as dpt decreases from 4.3 to 1.6 nm; here samples have values of dpt lower and higher than previously19 and also have equally wide particle-size distributions as previo~s1y.l~ This underlines the difficulty of associating particular t.p.d.peaks with the populations of distinct surface sites i and therefore that surface interdiffusion may be important.', 22? 23 With such complex and overlapping t .p.d. peaks complete profile analysis20 is not generally possible. Thus surface site occupancies deduced from t.p.d. profiles of hydrogen12 and subsequent correlations with catalytic activity* may not be useful. Compensation effects24 between A, and Etpd, upon n have not been considered here. If temperature-programmed desorption of hydrogen on silica-supported platinum releases more hydrogen in total than measured in chemisorption from the zero-pressure intercept (with part of this arising from the support and part from the Pt, and only partial differentiation of hydrogen adsorbed in sites with different adsorption energies), it is possible that temperature-programmed titration of adsorbed hydrogenz5 could differen-M.S. W. Vong and P. A . Sermon 1379 tiate hydrogen readsorbed in these different Pt, surface sites at lower temperatures and with slower and less extensive interdiffusion than is involved in t.p.d. at higher temperatures. Nevertheless, it has been shown here for the first time that the fraction of the hydrogen @,+p,) t.p.d. area up to the temperature used to evacuate samples in chemisorption experiments (here 573 K) measured under the present t.p.d. conditions (i.e. a low heating rate dT/dt of 4 K min-l, giving better t.p.d.peak resolution than previously reported for oxide-supported Pt) can be used to estimate the Pt dispersion with an accuracy comparable with chemisorption of hydrogen. This finding is useful; it could allow the characterisation of the surface coverage of hydrogen at any selected reaction or chemisorptive-pretreatment temperature. There is also much uncertainty about the effect of average Pt particle size on the stoichiometry26 and enthalpy2' of hydrogen chemisorption. Recent results27 suggest no effect of Pt dispersion on integral enthalpies of hydrogen chemisorption (although the values are lower than many reported2*) and this would not be expected from the present t.p.d. results. Clearly temperature-programmed such as t.p.d. complement direct enthalpy measurement~~~t 28 and this discrepancy must be resolved.Nevertheless, for the present the summed (P,+P2) t.p.d. peak areas for hydrogen measured under present conditions do appear to estimate supported Pt surface areas and dispersions moderately well. The (P, +/I2) results in table 4 for sample K agree with those recently reported30 from laboratory A1 on the sample catalyst after adsorption of H, (101 kPa, flowing, ambient temperature) during desorption at 300-700 K [i.e. (1 1.5-1 1.9) x 1019 H atoms desorbed per g catalyst in B'+C' t.p.d. peaks] with only a small fraction remaining to be desorbed at higher temperature [i.e. (0.8-1.5) x 1019 H atoms per g catalyst] and possibly arising from spilt-over hydrogen. The advantages of temperature-programmed titrations of preadsorbed hydrogen over desorption will be reported.The provision of a studentship for M. S. W. V. by S.E.R.C. is gratefully acknowledged. References 1 N. D. Hoyle, P. H. Newbatt, K. Rollins, P. A. Sermon and A. T. Wurie, J. Chem. Soc., Faraday Trans. 2 A. R. Berzins, M. S. W. Vong, P. A. Sermon and A. T. Wurie, Adsorption Sci. Technol., 1984, 1, 51. 3 Council of Europe (unpublished data). 4 S. Tsuchiya, Y. Amenomiya and R. J. Cvetanovic, J. Catal., 1970, 19, 245; 1971, 20, 1 . 5 K. Christmann and G. Ertl, Surf. Sci., 1976, 60, 365; K. E. Lu and R. R. Rye, Surf Sci., 1974, 45, 6 J. J. Stephan, V. Ponec and W. M. H. Sachtler, J . Catal., 1975, 37, 81. 7 R. A. Dalla Betta and M. Boudart, Proc. 5th Znt. Congr. Catal. 1973, vol. 2, p. 1329; J. C. Vedrine, 8 K . Foger and J.R. Anderson, J. Catal., 1978, 54, 318. 9 R. Krarner, Naturwissenschaften, 1977, 64, 269; Originalmitteilungen Physikalische Chemie der 1, 1985, 81, 2605. 677. M. Dufaux, C. Naccache and B. Imelik, Proc. 7th Int. Conf. Solid Surfaces 1977, 1, 481. Universitat Innsbruck 1977, 64, 269; R. Kramer and M. Andre, J . Catal., 1979, 58, 287. 10 J. R. Anderson, K. Foger and R. J. Breakespeare, J . Catal., 1979, 57, 458. 1 1 Y. Amenomiya, J . Catal., 1971, 22, 109. 12 J. P. Candy, P. Fouilloux and A. J. Renouprez, J. Chem. Soc., Faraday Trans. I , 1980,76, 616. 13 P. A. Sermon, J . Chem. Soc., Faraday Trans. I , 1980, 76, 885. 14 E. Tronconi and P. Forzatti, J . Catal., 1985, 93, 197; Y. J. Huang and J. A. Schwarz, J. Catal., 1986, 99, 249; J. S. Rieck and A. T. Bell, J . Catal., 1984, 85, 143. 15 S. M. Levine and S. H. Garofalini, SurJ Sci., 1986, 177, 157; K. E. Keck and B. Kasemo, Surf. Sci., 1986, 167, 313. 16 S. Ichikawa, Surf. Sci., 1986, 176, L853. 17 Y . W. Tsang and L. M. Falicov, J . Phys. C, 1976,9, 51 ; K. Besocke, B. Krahlurban and H. Wagner, 18 D. J. M. Fassaert and A. V. D. Avoird, Surf. Sci., 1976, 55, 313. 19 Y. Takasu, M. Teramoto and Y. Matsuda, J . Chem. Soc., Chem. Commun., 1983, 1329. Surf. Sci., 1977, 68, 39. 46-21380 Hydrogen T.P.D. from Supported Platinum 20 R. J. Cvetanovic and Y. Amenomiya, Adv. Catal., 1967, 17, 103; E. Habenschaden and J. Kuppers, 21 K. Christmann, G. Ertl and T. Pignet, Surf. Sci., 1976, 54, 365. 22 G. Ehrlich and D. 0. Hayward, Discuss. Faraday SOC., 1966, 41, 102; R. Lewis and R. Gomer, Surf. 23 D. D. Beck and J. M. White, J. Phys. Chem., 1984, 88, 174. 24 P. J. Estrup, E. F. Greene, M. J. Cardillo and J. C. Tully, J. Phys. Chem., 1986, 90, 4099. 25 M. S. W. Vong and P. A. Sermon, J. Chem. Soc., Chem. Commun., 1983,660. 26 S. Ladas, Surf. Sci., 1986, 175, L681; A. Coma, M. A. Martin and J. Perez Pariente, Surf. Sci., 1984, 27 B. Sen, P. Chou and M. A. Vannice, J. Catal., 1986, 101, 517; M. A. Vannice, L. C. Hasselbnng and 28 E. R. A. Mills, P. A. Sermon and A. T. Wurie, Proc. 8th Znt. Congr. Catal. (Verlag, Berlin, 1984), Paper 29 A. Jones and B. McNicol, Temperature-Programmed Reduction for Solid Materials Characterization 30 A. Frennet and P. B. Wells, Appl. Catal., 1985, 18, 243. Surf. Sci., 1984, 138, L147. Sci., 1969, 17, 333. 136, L31. B. Sen, J. Catal., 1985, 95, 57. 111-131 and ref. (23H25) therein. (Marcel Dekker, New York, 1986). Paper 6/451; Received 5th March, 1986