J. CHEM.SOC. FARADAY TRANS., 1994, 90(5), 787-795 DRIFT and Mass Spectrometric Experiments on the Chemistry and Catalytic Properties of Small Ir Clusters at the Surfaces of Polycrystalline a-Al,O, L. Basini* and A. Aragno Snamprogetti SPA.Research Laboratories, Via Maritano 26,20097San Donato Milanese (MI), Italy The DRIFT and mass spectra collected during eight high-temperature, high-pressure experiments are described and discussed. Relationships between the occurrence of aggregation-disaggregation reactions of Ir species on a low-surface-area polycrystalline a-Al,O, and the formation of gaseous reaction product have been defined. Reactions in which carbidic oxidic and carbonyl species are interconverted have been produced and their role in CO and CO, hydrogenation is discussed.The results are compared with those previously found for Rh on a-alu m i na. This work is part of an extensive research programme dedi- cated to the high-temperature, high-pressure (HTHP) surface chemistry of small aggregates and/or monatomic species of Rh, Ir and Ru. Previously' we have described the chemistry of Rh'(CO), surface species. These were formed uia a reaction between Rh,(CO)', and the active surface sites of M,O, a-Al,O, and (30,(Rh content 0.1%); no other ligands, with the exception of the surface atoms and of the CO groups, were present. Here we present a study on Ir-containing a-A1203samples (Ir content 0.25 wt.%) which have been pre- pared oia a reaction between Ir,(CO),, and the surface of a-Al,O, .The anisotropic conditions and the 'degree of disorder' of the extremely mutable environment of the surfaces are responsible for the difficulties in describing the chemistry of the noble metals at the surfaces of polycrystalline oxides. However, many surface complexes have been successfully and selectively produced and characterized2-14 by organometallic chemistry methods, spectroscopic experiments and molecular mechanics modelling. Many small metal clusters appear to be stabilised in the cages or supercages of zeolite^,^*'^*'^ by con- finement in the cages and/or by mechanisms such as proton anchoring.''-' However, extensive aggregation-disaggregation reactions occur when the noble metal cluster species are anchored on low-surface-area polycrystalline oxides.Some clusters are formed only in reaction conditions as a result of a delicate balance between thermodynamic and kinetic processes. This work does not aim to produce, stabilize and charac- terize well defined surface species, but, (i) to investigate the HTHP formation and transformation processes of Ir surface species; (ii) to relate the molecular features discovered at the surfaces to the products of the solid-gas interactions and (iii) to discuss the analogies and the differences between Ir and Rh chemistry.' It also attempts to describe, at the molecular level, the elementary steps of complex reactions such as the hydrogenation of CO and CO, . Experimental Sample Preparation An anhydrous THF solution of Ir4(CO)12 (Strem) under CO atmosphere was dropped into a slurry of a-Al,O, (Aldrich 99.999 wt.%) dispersed in the same solvent.The purity and crystalline structure of a-Al,O, was checked by optical emis- sion arc spectroscopy and XRD analysis. The surface area measured by the BET method was ca. 10 m2 g-'. After a few hours the solid species was filtered in a CO atmosphere and dried under vacuum at 25°C. The preparation procedures and the insertion of the samples into the DRIFT cell were performed in moisture-free and C0,-free environments. The samples contained 0.25% Ir ; this amount corresponds approximately to a monolayer of Ir, carbonyl clusters with volume 3905 A3 which is the volume of the +P(CH2C6H,XC6H,),] [HIr,(CO),,] -Cluster.' Conse-quently, less than a monolayer of [HIr,(CO), ,]-anions [which are formed at the surface after the chemisorption of the Ir,(CO),,] is present at the surface.The DRIFT spectra of the materials obtained, used at point A (see Fig. 1, later) of each experiment, showed CO stretching bands at 2065 (sh), 2024 (s), 1990 (sh), 1958 (vw) and 1731 (m) cm-l. These spectra indicate that the Ir4(CO)1, clusters have been modi- fied after interaction with the surface as will be discussed below. Apparatus The reactivity experiments were performed in an HTHP cell equipped with two ZnSe windows, one for the incoming radi- ation and the other for the diffuse reflected radiation. The equipment has been described previously'*'s and allows the collection of DRIFT spectra in flowing gaseous environments between 25 and 500°C and between 0.1 and 5 MPa. The present experiments were performed at 0.1 MPa.The HTHP cell was inserted into the sample compartment of a Nicolet 2OSXC spectrometer equipped with an MCT detector. Spectra were recorded with a resolution of 4 cm-'.The cell exit line was linked to a quadrupole mass spectrometer (UTI) through a two-orifice pressure-reduction sampling stage with differential pumping between the orifices. The composition of the exit line was examined with selected peak monitoring with both time and with temperature, and with repeated scans between 1 and 60 u. High-purity (Strem electronic grade) He, H,, 0,, CO, CO, gases were used and the He and H, streams were further purified with OM1 filters (Li and Co compounds which react with H,O, CO, CO,, 0, impurities). Each gas line was equipped with dedicated mass- flow-meters and controllers. A laser Raman spectrometer (Dilor triple monocromator spectrometer) equipped with a multichannel detector with 512 diodes (Omars) was used to record spectra of the A samples (see Fig.1, later). The best spectra were achieved in a backscattering configuration with a slit width of 80 pm and with a confocal microscope. Under these microsampling con- ditions the S/N ratio was significantly improved, probably owing to the elimination of stray light coming from outside Experiment 1 A B co, C 25-500°C 25-500-c He H coAcExperiment 2 A 2 ~ B 2-oc C -D 25-50 "C He Experiment 3 A -He co2 __c CExperiment 5 A 2zc25-500°C He H co -BExperiment6 A c D 25-500°C 25-!XO"c 25-500s He 02 coExperiment 7 A --cdD 25400% ~500~25-5oO"c He H2-c COZ+H& Experiment8 A -25500°C * 25-500"c 25-500"c Fig.1 Scheme of the main steps of the experimental sequences the focal volume. The 514.5 nm line of an Ar' ion laser (Spectra Physics, Model 2020) with a power ranging from 5 to 50 mW was used to excite the Raman scattering. Experimental Sequences The eight experimental sequences are shown schematically in Fig. 1. The first step in each experiment was a thermal treat- ment in flowing He. This was performed both to investigate the temperature-induced reactions in an inert atmosphere and to obtain, at point B, the same type of bare Ir cluster.The materials obtained at point B were reacted with CO or with CO-H,, Experiments 1-4, or CO, and CO2-HZ, Experiments 5-8, after reducing (H,) or oxidising (0,) thermal treatments. The heating was always carried out in a flow of He; in each run the temperature was stabilised for 10 min at 25, 100, 200, 300, 400,500°C to collect the DRIFT spectra and to introduce pulses of gaseous reactants. Mass spectra were recorded continuously during the experiments. Results and Discussion The Ir4(CO)12 cluster is transformed into other cluster species after liquid-solid chemisorption. The tetrametallic cluster has only linearly bonded CO groups whose stretching vibrations absorb IR radiation at the frequencies reported in Table 1.The peak maxima of the species obtained at point A (see Fig. 2), particularly the peak at 1730 cm-', indicate the formation of clusters with bridged CO groups such as HIr,(CO);, ,' 7T24i25 or the Ir6(CO)16 red isomer with CO face-bridged groups26 and the Ir6(CO)Ii 27,28 anion. These species were interconverted by the reactions (1)-(4) both in Table 1 CO stretching frequencies of Ir4(CO)12 (cm-') THF solution' 2110 (w), 2067 (vs), 2026 (mw) KBr pellet and crystalline formb 2111 (w), 2086 (sh), 2056 (s),2040 (sh), 2040 (sh), 2020 (m), 2004 (sh) y-Al,O, Ic SiO, ,d Na-Ye 2086 (sh), 2054 (s), 2008 (sh) 'Ref. 19. Ref. 20. 'Ref. 21 and 22. Ref. 23. Ref. 8. J. CHEM. SOC. FARADAY TRANS., 1994, VOL.90 4.01 I Y r' 0.5 0.0 V I1 2100 2000 1900 1800 wavenum ber/cm- ' Fig. 2 (a)Spectrum of the species obtained at point A; (b) magnifi-cation of the multilplet in the 2100-1900 cm-' range homogeneous solution27 and at the surface of polycrystalline materials.**' 9*27--2 Ir4(CO)12+ OH--+ Ir,(CO),,CO,H-(1) OH-+ Ir,(CO),,CO,H -+ HIr,(CO);, + HCO, (2) 3HIr,(CO)r1 + 30H-+21r6(CO):; + 3c0, + 3H2 (3) Ir6(CO):5 + CO + OH-+Ir6(CO)16+ H, + 0,-(4) Reactions (1)-(4) require a basic environment in homoge- neous solution. They have been reported to occur also at the surfaces of MgO" and in the cages of the Na-X and Na-Y zeolite^.'*^^-^^ The formation of clusters with bridged-bonded CO species suggests that the ol-Al,O, used in the present work has OH-containing surface sites able to activate this chemistry.The comparisons between the DRIFT spectrum of species A and the IR spectra of the three candidate species indicate strongly that the anion HIr,(CO);, has been formed. However, the broadness of the IR bands (which is consistent with the interaction of the metal complex with slightly differ- ent surface sites) prevents a definitive assignment of the IR bands. A laser-Raman spectrum of a sample containing 3 wt.% Ir was also recorded (the higher loading of Ir was necessary because of the lower intensity of the Raman signals compared with IR signals). The sample was prepared by the same procedure by depositing Ir4(CO)12 on the surface of the a-Al,O,.The spectrum is shown in Fig. 3 and shows sharp peaks that are assigned to carbonyl stretching bands of HIr,(CO);, at 2110, 2035, 2010 and 2001 cm-'.9 Conse-727 1 I cnc h 1 127 1 2200 2100 2000 1900 wavenum ber/cm -' Fig. 3 Carbonyl Raman scattering of a sample prepared by depos- iting Ir4(CO)12 at the surface of a-Al,O, J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 quently, the species obtained at point A will be represented by this tetrametallic anion. The Ir carbonyl species were highly stabilised at the surfaces and could not be extracted with a THF solution of [PPNl[Cl]. We assume that a strong anchoring is achieved because the low amount of Ir4(CO)12 molecules are allowed to react selectively with the more sta- bilising surface sites.The low frequency of the band at 1730 crr-' is consistent with a strong ion-pairing effect between surface A1 atoms and the bridged carbonyl groups of the anionic clusters.24 Experiment 1 The clusters obtained at point A were oxidized and disaggre- gated during the first heating in helium. This is inferred by the formation at 200°C of a doublet at 2055 and 1981 cm-'. The band positions (see Fig. 4) indicate that IrI(CO), mono- metallic surface species were formed (the two bands are assigned to symmetric and anti-symmetric CO stretching). The band shapes and the intensity ratios, also suggest that the low-frequency peak gains contributions from the vibra- tions of other carbonyl cluster species. The intensity of the doublet at 2055 and 1981 cm-' is strongly reduced at temperatures higher than 300°C but no other absorption bands, with the exception of a new peak at 1720 cm-', were formed.We assign this new peak (which is also present at 500°C when all the other absorption features in the 1600-2000 cm-' range have disappeared) to stretching vibrations of a hydrogencarbonate species. It is also worth noting that heating in He produced the species responsible for the 1720 cm-' band only on the Ir/a-A1203 sample; the same or an analogous band was not detected on the Rh/a- A1,0, sample or on pure a-Al,O, . Eqn. (5) is chosen to rep- resent the temperature-induced oxidation reaction of the species produced at point A. Other indications on the chem- istry which occurred during the first experimental step will be given after the discussion of Experiment 2.HIr,(CO);, + 30H--+ 4Ir'(co), + 3HCO; (5) The formation of gaseous CO,, revealed by mass spec- trometry as well as the reduction of the OH stretching band intensities and the formation of a peak at 1720 cm-' in the DRIFT spectra, sustains the formalism, eqn. (5). After cooling, a new thermal cycle was performed by intro- ducing 5 min CO pulses (flow rate 60 ml min-') into the cell at 50, 100, 200, 300,400 and 500°C. Peaks at 2055 and 1981 cm-' assigned to IrVCO), and a new band at 2045 cm-' were detected after the first two CO pulses at 50-100°C (see Fig. 4 and Table 2). This new band is assigned to vibrations of CO groups linearly bonded to Ir clusters and indicates the 2400 2000 1600 2400 2000 1600 wavenumber/cm -' wavenumber/cm-' Fig.4 DRIFT spectra in the co-stretching range recorded during Experiment 1. (a)-@) recorded during the first thermal cycle, in flowing helium, at (a) 25, (b) 100, (c) 200, (d) 300, (e) 400 and (f)500 "C and (9)on cooling to 50 "C.(h)-(n) recorded during the second heating cycle, with 5 min CO pulses, at (h) 50, (i) 100, 0')200, (k)300, (0 400 and (m)500"C and (n) on cooling to 50 "C. occurrence of CO-induced aggregation reactions of the monometallic species. At increasing temperatures the doublet assigned to IrVCO), disappeared, while the peak assigned to linearly bonded CO groups remains up to 500°C and a shoulder at 1960 cm- can also be distinguished at increasing temperatures. The formation of gaseous CO, was also detected.To summarise: the reaction between the II-,(CO)~, and the surface of the a-Al,O, produced HIr,(CO);, species, reac- tions (1)-(4) are suggested to be involved in this process. During the heating to 500°C in He, the surface clusters were first disaggregated to form IrYCO)2 species, gaseous CO, and Table 2 CO stretching frequencies measured at the peak maxima obtained in the last step of Experiments 1-3 Experiment 1 Experiment 2 Experiment 3 T/"C CO pulse He flow CO pulse He flow CO pulse He flow 50 2070, 1995 2060, 1995 2080 (sh), 2057 2046 2063, 1995 2062, 1995 100 200 2070, 1995 2070, 1995 2055, 1994 2070, 2050, 1995 2080 (sh), 2057 2052 2049 2046 300 2070,2050, 1995 2070, 2036, 1995 2048 2040 400 500 50 2040, 1995 (sh) 2035, 1974 (sh) 2067 2030, 1975 (sh) 2025, 1970 (sh) 2049, 1963 (sh) 2046 2032 2060 2035 2030 2050 carbonate surface species [see eqn.(5)]. Subsequently the monometallic species were decarbonylated. During the second heating the interactions between the bare clusters and gaseous CO molecules caused the reaggregation of the mono- metallic Ir complexes into new clusters. Experiment 2 Thermal treatment in He was followed by a second heating of the decarbonylated sample in a flowing H, atmosphere. At temperatures between 100 and 300°C a new weak band, centred near 2050 cm-', was detected and this then disap- peared at temperatures higher than 400°C while CH, and H,O were desorbed.The experiment was repeated by using deuterium instead of hydrogen and the same results were achieved (see Fig. 5). The absence of any clearly detectable isotopic shift excludes the possibility that the band could be related to Ir-H or Ir-D stretching vibrations. The purifi- cation system (see Experimental) and the analysis of the com- position of the D, and H, gases also excluded the possibility that the bands could be originated by CO impurities. Analo- gous indications were obtained while studying the reactivity of Rh species at the surfaces of a-Al,O, or MgO' during Experiments 1 and 2. We interpret these findings according to ref. 1, assuming: (i) the formation of carbidic and oxidic species by thermal dissociation of the CO ligands of the HIr,(CO);, clusters during the first experimental step and (ii) the H, (D,)-induced high-temperature reaggregation of the carbidic and oxidic species to produce new hydridocarbonyl clusters.Examples of well defined carbidic carbonyl clusters come from the studies on 0s surface chemistry which demon- strate the possibility of selectively synthesizing the [Os,C(CO), ,I2-and the [Osl,C(CO),,]2-carbidic clusters from OS,(CO),, under CO or H,-CO atmo~pheres.~~Evi-dence for the production of reactive oxidic species, comes from IR and XPS experiment^^'*^' which demonstrated the formation of Rh-oxygen adducts during the oxidative , , i' d 3800 2800 1800 21'00 1900 waven u m be r/cm -' wavenumber/cm-' Fig.5 Spectra, showing the OD and CO stretching ranges, re-corded in Experiment 2 at (a) 50°C in He flow and at (b) 50, (c) 100, (6)200, (e) 300, (f)400 and (9)500"Cin D, flow J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 decarbonylation of RhYCO), surface species. However at 400°C the carbonyl stretching bands disappeared and CH, and H,O were detected. In the third thermal cycle the recarbonylation reactions were carried out by introducing into the cell CO pulses as in Experiment 1. At 50"C, after the first CO pulse, a broad and asymmetric band at 2050 cm-' formed (see Fig. 6);this band became sharper and gradually shifted to 2030 cm-' with increasing temperatures up to 500 "C. The sharpening also reveals a shoulder at 1960 cm-(see Fig.6 and Table 2). The high-frequency band moves back to 2050 cm-' after cooling to 50°C. The spectrum produced at the end of the thermal cycle can be overlapped with the spectrum obtained at 50°C at the beginning of the cycle; this suggests that during the thermal cycle the CO pulses caused only reversible modifi- cations of the Ir aggregates. Comparison between Experi- ments 1 and 2 also indicates that carbonyl bands with the same shapes and positions were detected at temperatures above 300°C. This suggests that the same or a similar 'state of aggregation' is reached and stabilized at high temperatures in the presence of CO on both hydrogen-treated and untreated materials. When Experiments 1 and 2 were per- formed on the Rh-containing a-Al,O, the evolution of the spectral features indicated a different, continuous and irre- versible modification of the state of aggregation of the Rh clusters.' In Schemes 1 and 2 of Fig.7 we show the reactivity fea- tures that can be inferred from the results of Experiments 1 and 2 performed on the Ir- and Rh-containing a-Al,O, . Experiment 3 The Ir surface species obtained at point B were treated under CO atmosphere as described earlier. Initially, the spectra of the IrYCO), and of other carbonyl clusters described in Experiment 1 were produced and subsequently (after 30 min) i i d'i 2400 2000 1600 wavenumber/cm-' Fig. 6 Spectra recorded in Experiment 3 after CO pulses at (a) 50, (b) 100, (c) 200, (6)300, (e) 400 and (f)500°C and after cooling to 50 "C J.CHEM. SOC. FARADAY TRANS., 1994, VOL 90 Experiment 1 1r4(c0)1Z Rh4(C0)12 I CO a-A1,03 t 25-7 00"C.( C 25-500 "C, CO ii r Rh'jCO), + C02 Ir,(CO),+CO~ 20@100 "C 1 Rh,(CO), + COz I CO I a-A1203 species A Rhi (C O), Ir'(CO)2+ [Ir,C(CO)Y] + co, Ir,C + [Osl Rh,C + [OJ [HlrJ + CH, + H20J 1 [HRh,,,?] + CH4 + tiLO 25-500 "C, CO 25-500 "C. C<l1 1 (I Irxw), species D [HRh"p(co),J Fig. 7 Scheme 1: reactions detected during Experiment 1 on Ir on r-A1203(this work) and Rh on r-A1203,ref. 1. Scheme 2: reactions detected during Experiment 2 on Ir and Rh on r-Al,O I two bands at 2050 (s) and 1995 (m) cm- revealed the forma- tion of 1r4(C0)12 species (see Fig.8). The aggregative effect of the CO atmosphere on Ir1(C0), producing Ir4(CO)l ,has been observed already at the surfaces of ;,-A1,0,33 and in the cages of Na-Y zeolite' and this reaction was expected also !n this case. The subsequent heating at 100°C in 0, sharpened these two bands and the high-frequency peak was snifted slightly up to 2063 cm-'. Desorption of CO, was also seen in the exit line during the oxidative treatment. The sharpening of the bands is consistent with an 0,-induced dispersion of aggregates of Ir4(CO)12 molecules as already described in ref 23. The simultaneous desorption of CO, suggests that the carbidic species are oxidised. A CO pulse introduced into the cell, after cooling at 50°C did not modify the absorption fe't-tures of the DRIFT spectra.A different reaction was revealed when the experiment was performed on the RhIr-Al,O, sample. The Rh carbonyl clus- 79 I __-___~ ---.-.--/' 2400 2200 2000 1800 1600 wavenumber cm ' Fig. 8 Spectra recorded during Expenment 3. (a)Freshly prepared sarnple at 2S'C in He flow; (b) at 50 C after a decarbonylative thermal treatment in He and 30 min of reaction in CO atmosphere, (c)-( f)in flowing 0, at 1oO*C,(9)at 25 C after a S min ('0pulse ters were completely disaggregated at 50 -C in oxygen into Rb'(CO), and at 100"C a complete decarbonylation was achieved to produce a rhodium-oxygen adduct which in CO at 50 C gave again Rh'(CO), and gaseous CO, .' In Fig. 9 the reactions obtained with the Ir- and Rh-containing x-Al,O, art.composed. 1r4 CO),, + no rem on! 5o-1 00 "C.0,2 + RhACO), 4\rHspecies E Rh'(C;O), Fig 9 Reaction scheme detected for Experiment 3 on Ir on r-Al,O, (thi, work) and Rh on r-A1203samples (ref 1) J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Experiment 4 The results obtained in the first two steps of Experiment 2 were reproduced here. In the third step 10 min CO-H, pulses (flow rate 60 ml min-') were introduced into the cell at 25, 100, 200, 300, 400 and 500°C. Broad bands were detected in the DRIFT spectra at each temperature investigated both in a flowing CO-H, mixture and in an He flow (see Fig. 10 and Table 3). The position of the peaks was always shifted 2400 2000 1600 2400 2000 1600 wavenumber/cm-' wavenumber/cm-' Fig.10 Spectra recorded during Experiment 4. (a) The decarbony- lated and hydrogen-treated sample at 50°C (b),(4,(f),(h)and (j)in CO-H, flow at 50, 100, 200, 300 and 400°C; (c), (e), (g), (i), (k) and (I) after switching to an He flow. Table 3 CO stretching frequencies revealed during Experiments 4 and 8 ~ ~~~ Experiment 4 Experiment 8 CO + H, CO, + H, T/"C pulse He flow pulse He flow 50 2010(b) 2010(b) 2003 2000 100 2010(b) 2010(b) 2017 2003 200 2032 (b) 2025 (b) 2023 2027 300 400 2033 (b) 2026(b) 2021 (b) 1993 (b) 2025 2025 2027 2025, 1970 (sh) 500 2020 (b) 1985 (b) 2026, 1978 (sh) 2028, 1966 CO pulse He flow CO pulse He flow 50 2061 2040 2052, 1965 (sh) 2050, 1965 (sh) upwards in the presence of the CO-H, mixture.The broad- ness of the bands indicates that many different aggregates are formed, and the upward frequency shift is interpreted as a consequence of the increased CO coverage of large Ir aggre- gates which cause : (i) enhancement of the dipole-dipole inter- actions between chemisorbed molecules ;34-36 (ii) decrease in the Ir-CO bond strength and increase in the IrC=O bond strength.37 The reaction between CO and H, produced CH,, H,O and CO, at temperatures >300"C. The presence of CO, is due to the occurrence of the water-gas shift reaction. Analogous results were achieved when Experiment 4 was per- formed with Rh/a-Al,O,, however, in this case, the formation of CH, ,H,O and CO, was first detected at 200 "C. Experiment 5 The interaction between CO, (5 min pulses, flow rate 60 ml min-') and the decarbonylated material obtained at point B, produced a chemical reaction at 300 "C.At this temperature a broad peak, centred near 2030 cm-', was detected in flowing CO, (see Fig. 11). We observed that on switching the CO, flow to He the peak was increased in intensity and sharpened. This result was also found in Experiment 7 (see later) where we observed that at 300°C a carbonyl band was stabilized after the CO, pulse only in an He atmosphere and disap- peared on restoration of the CO, flow. To explain these effects we tentatively propose, in accord with ref. 1, that Ir- oxygen adducts Irz(0), , Ir carbonyl clusters Ir-CO and Ir bare clusters Ir" are produced and interconverted as a result of temperature-induced aggregation reactions and oxygen- induced disaggregation reactions.The reaction cycle in which atomic surface oxygen species [O,] are also involved, is rep- resented by the following scheme. co2 The position, the intensity and the shape of the IR bands, recorded at 400 and 500°C are reported in Table 4 and Fig. 11. After cooling to 50°C absorptions at 2031 cm-' with shoulders at 2045 and 1962 cm-' were detected. The experi- ment was concluded by introducing, at 50"C, a CO pulse into the cell. This produced species with one IR band at 2054 cm-' and a shoulder at 2003 cm-' while a small amount of Table 4 CO stretching frequencies measured at the peak maxima obtained in the last step of Experiments 5-7 v(CO)/cm - Experiment 5 Experiment 6 T/"C C0,pulse He flow CO, pulse He flow 50 - - - - 100 - - - - 200 - - - - 300 - 2030, 2003 (sh) 2032, 1970 (sh) 2022, 1967 (sh) 400 2008 2023 2029, 1969 (sh) 2029, 1069 (sh) 500 2027 2023 2024, 1971 (sh) 2007 (w) CO pulse He flow CO pulse He flow 50 2056, 2005 2054,2003 2070 2043 Experiment 7 CO, pulse He flow 2060 (vw), 1995 (vw) 2060 (vw), 1995 (vw) 2060 (vw), 1995 (vw)2060 (vw), 1995 (vw) 2060 (vw), 1995 (vw)2025 2029 2029 2029 2028 2024 2022 CO pulse He flow 2072, 1998 2062, 1998 J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 2600 2400 2200 2000 1800 2600 2400 2200 2000 1800 wavenumber/cm-' wave n umber/cm -' Fig. 11 Spectra recorded during experiment 5.(a) The thermally decarbonylated sample; (b), (4,(f), (h), (j)and (I) during 5 min CO, pulses at 50, 100, 200, 300, 400 and 500°C; (c),(e), (g), (i), (k) and (m) in He flow after the CO, pulses; (n)the cooled sample; (c)during the final CO pulse; (p) after the CO pulse. CO, was desorbed. The formation of the CO, is interpreted by assuming that reactive oxygen species remained at the sur- faces together with Ir carbonyl clusters after the breaking of the one of the CO, carbon-oxygen bonds. Experiment 5 was performed' on Rh/a-Al,03 which showed a much lower reactivity towards the CO,; in that case a weak carbonyl stretching band formed only at 500°C and quickly disappeared under flowing He.The treatment with the CO pulse of the cooled sample produced gaseous CO, and the RhVCO), species. We concluded that inter- action with the CO, caused the oxidative disaggregation of the surface Rh clusters. Experiment 6 A decarbonylated sample was produced at point C as already described in Experiments 2 and 4. The following interaction with the CO, pulses at 300°C and 400°C produced an asym- metric peak at 2030 cm-' with a shoulder near 1971 cm-' (see Fig. 12 and Table 4). The peak was broadened and weakened at 500°C by switching the CO, flow to He. We stress that the band intensities were not enhanced in He flow as found in Experiments 5 and 7 (see later). After cooling at 50"C,a very weak absorption feature remained, centred at 2020 cm-'.This band was enhanced in intensity and shifted to 2043 cm-l when a CO pulse was introduced into the cell at 50°C and the formation of some gaseous CO, was detected. Comparison with the results of the previous experi- ment indicates that the CO, reductive chemisorption reac- tions were enhanced on the hydrogen pre-treated samples (see Table 4 and Fig. 11 and 12) but this effect is much less relevant for Ir than for the Rh/or-Al,O, sample.' In the last 2600 2400 2200 2000 1800 2600 2400 2200 2000 1800 wavenumber/cm-' wavenumber/cm-' Fig. 12 Spectra recorded during Experiment 6. (a) Thermally decarbonylated and hydrogen-treated sample; (b), (4,(f),(h), (j)and. (I) during 5 min CO, pulses at 50, 100, 200, 300,400 and 500°C; (c), (e),(g), (i), (k)and (m)in He flow after the CO, pulses; (n) the cooled sample; (0)during the final CO pulse; (p) after the CO pulse.case, at temperatures as low as 50"C, hydridocarbonyl com- plexes were detected. The enhancement in the CO, disso-ciative chemisorption on the hydrogen pre-treated samples can be interpreted by a metal-mediated reaction between a surface hydride species and CO,. This mechanism is consis- tent with the CO, insertion reactions into the M-H bonds of hydridocarbonyl complexes which occur in homogeneous liquid solution^.^^-^^ Solymosi and co-workers have shown that the CO, chemisorption reactions are enhanced on Pt, Pd, and Rh single crystals by reducing the work function of the metals with a chemical deposition of K.45-48The same effect should not be involved in the observed hydrogen- induced reactivity enhancement since H, adsorption on the same metals increases their work fun~tion.~~-~~ Experiment 7 The oxidative treatment in flowing oxygen at 100°Cdid not inhibit the high-temperature reactivity with the CO, pulses.A broad CO stretching band at 2025 cm-'was first detected at 200°C by switching the CO, flow to He (see Fig. 13 and Table 4). This phenomenon, noted in Experiment 5, is more evident when the interaction with CO, follows oxidative treatment. At higher temperatures, up to 500 "C, the carbonyl stretching bands remain in both flowing CO, and He. A sub-sequent CO pulse introduced at 50°C into the cell produced gaseous CO, and clusters with CO stretching at 2062, and 1998 (sh) cm-'.Bands with the same position and shape were detected during Experiment 3 and were assigned to Ir4(CO)1, species. In these same conditions the Rh surface species' showed a greater mobility and the carbonyl clusters were stabilized between 200 and 500°C only after switching the CO, flow to He. 2600 2400 2200 2000 1800 2600 2400 2200 2000 1800 wavenumber/cm-' waven u m ber/cm -' Fig. 13 Spectra recorded during Experiment 7. (a) Spectrum of the thermally decarbonylated and oxygen-treated sample; (b),(4,(f),(h), (j)and (0 during 5 min CO, pulses at 50, 100, 200, 300, 400 and 500 "C;(c),(e),(g), (i), (k)and (m)in He flow after the CO, pulses; (n) the cooled sample; (0) during the final CO pulse; (p) after the CO pulse.J---i 2400 2200 2000 1800 1600 2400 2200 2000 1800 1600 wavenumber/cm-' wavenumber/cm-' Fig. 14 Spectra recorded during Experiment 8. (a) Spectrum of the thermally decarbonylated and hydrogen-treated sample; (b),(4,(f), (h),(j),(0 and (n) at 50, 100, 200, 300,400 and 500°C in C0,-H,; (c), (e),(g) (i),(k),(m)and (0) at the same temperatures in He; @) during the final 5 min CO pulse; (4) the CO pulse. J. CHEM.SOC. FARADAY TRANS., 1994, VOL. 90 Experiment 8 A broad carbonyl stretching band centred near 1981 cm-' originated at 50°C during a 10 min pulse of the CO2-H, atmosphere and persisted in He flow. At increasing tem- peratures the bands were sharpened and one peak at 2025 cm-'and shoulders at 1970 and 2060 cm-' could be distin- guished up to 500°C.The bands detected in flowing C0,-H2 atmosphere and in He atmosphere had the same shape, posi- tion and intensity. The sharpening at increasing temperatures and the absence of any shift of the IR bands is interpreted by the selective formation of carbonyl clusters. After cooling, the peak became even sharper with a maximum at 2045 cm-' and the shoulders were shifted to 1961 and 2070 cm-' (see Fig. 14 and Table 4). When a CO pulse was introduced into the cell at room temperature desorption of CO, was observed, but the IR spectra were not modified. The same effects were also produced when Experiment 8 was performed on the Rh/a-Al,O, sample.' The occurrence of the CO, che- misorption reaction at 50°C with the formation of carbonyl clusters was not observed in Experiments 5,6 and 7.This can only be the consequence of the presence of H, in the gaseous environment. We assume that the reductive chemisorption of CO, occurred via a metal-mediated interaction with a hydrido surface species; this mechanism should be more effec- tive in the presence of gaseous hydrogen since the oxygen species produced by the breaking of the C-0 bonds can react to form OH bonds and water molecules. Conclusions DRIFT and mass spectra gathered during eight experiments gave insight on some undisclosed aspects of Ir surface chem- istry. In the following we summarize the main conclusions of the work.Formation and Reactivity of Carbidic and Oxidic Species and Metbanation Reactions Several assume that during CO hydrogenation processes, such as Fischer-Tropsch and methanation reac- tions, carbidic, oxidic and carbonyl species are simulta-neously present at the surfaces and are interconverted in stationary conditions. Our results sustain this hypothesis since there is a clear experimental evidence for the following steps in the CO hydrogenation process represented by eqn. (6)-(9), i.e. (i) the formation of carbidic and oxidic species from the thermal dissociation of C-0 bonds of carbonyl clusters; (ii) the H, (D,)-induced reaggregation of the carbidic and oxidic species into new carbonyl clusters stabilized up to 300°C; (iii) the hydrogenation (deuteriation) of the new clus- ters into bare Ir aggregates and gaseous CH, (CD,) and H20 (Dm Ir-(CO) +Ir-C + 0, (6) Ir-C + 0, + H, +HIr-(CO) (7) HIr-(CO) + H, +Ir + CH, + H,O (8) Ir + CO +Ir-(CO) (9) Surface, highly reactive, oxygen species, 0,, were also formed during CO, reductive chemisorption and were detected because of their ability to oxidize gaseous CO mol- ecules at 25 "C.We consider that these surface oxygen species were able to disaggregate and to oxidize carbonyl Ir clusters during Experiments 5 and 7. This disaggregative and oxida- tive effect should be counter-balanced by a temperature-induced reductive and aggregative effect in a flowing He environment (see Scheme 1 in Fig. 7). J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 The hydrogen pretreatment moderately enhanced the C02 dissociative chemisorption which was first observed at 200 "C. However, carbonyl clusters were detected at 50 "C in a C0,-H, environment, much earlier than the products of the methanation reaction which were detected at 300°C. The low-temperature reactivity is explained by assuming a metal- mediated insertion of a surface hydrido species into a CO bond of the CO, followed by a reaction between atomic oxygen and hydrogen to produce surface OH groups and water. The hydrogenation of the surface carbonyl groups has a higher activation energy and is the rate-limiting step in the CO, methanation reaction. Ir Cluster Species at the Surfaces Tetrametallic Ir clusters were highly stabilized and strongly anchored to the surface of a-Al,O, .HIr4(CO)F1 is the candi- date species formed at point A after the chemisorption of Ir4(CO)12. The anion could not be extracted with a THF solution of PPN Cl. 1r4(C0)12 surface species were repro- duced after the interaction of a thermally decarbonylated sample with CO and were stabilized at 100°C in a flowing 0, environment. In many cases the IR bands of the carbonyl species were enhanced in intensity and sharpened at increas- ing temperatures thus indicating the selective formation of several types of small carbonyl clusters. The only exceptions to this general behaviour have been found in Experiment 4, when the interaction with H,-CO mixture was studied.In this case very broad bands were originated and were related to the unselective formation of large Ir aggregates. We are grateful to Dr. M. Marchionna and Dr. A. Vetere for helpful and stimulating discussions. References 1 L. Basini, M. Marchionna and A. Aragno, J. Phys. Chem., 1992, 96,9431. 2 Metal Clusters in Catalysis, ed. B. C. Gates, L. Guczi and H. Knozinger, Surface Science and Catalysis, Elsevier, Amsterdam, 1986, vol. 29. 3 H. H. Lamb and B. C. Knozinger, Angew. Chem., Znt. Ed. Eng., 1988,27, 1127. 4 Surface Organometallic Chemistry: Molecular Approaches to Surface Catalysis, ed. J-M. Basset, B. C. Gates, J. P. Candy, A. Chaplin, M. Leconte, F. Quignard and C. Santini, Kluwer, Dor- drecht, 1988. 5 L-F. Rao, A.Fukuoka, N. Kosugi, H. Kuroda and M. Ichikawa, J. Phys. Chem., 1990,94,5317. 6 P. Dufour, C. Houtman, C. C. Santini, J-M. Basset, L. Y. Hsu and S. G. Shore, J. Am. Chem. SOC., 1992,114,4248. 7 H. F. Van't Blick, J. B. A. Vanzon, T. Huizinga, D. D. Konis-berger and R. Prins, J. Phys. Chem., 1985,89,4783. 8 S. Kawi, J. R. Chang and B. C. Gates, J. Am. Chem. SOC., 1993, 115,4830. 9 G. Hestl, N. D. Trian Trafillou, H. Knozinger and B. C. Gates, J. Phys. Chem., 1993,97,666. 10 G-D. Lei and W. M. H. Sachtler, J. Catal., 1993,140,601. 11 Z. Zhang, B. Lerner, G-D. Lei and W. M. H. Sachtler, J. Catal., 1993,140,481. 12 Z. Zhang, H. Chen and W. M. H. Sachtler, J. Chem. SOC., Faraday Trans., 1991,87, 1413. 13 L. Xu, Z. Zhang and W. M. H. Sachtler, J.Chem. SOC., Faraday Trans., 1992,88,2291. 14 Z. Zhang, T. T. Wang and W. M. H. Sachtler, J. Catal., 1991, 128,13. 15 P-L. Zhou, S. D. Malony and B. C. Gates, J. Catal., 1991, 129, 315. 16 T. J. Lee and B. C. Gates, Card Lett., 1991,8, 15. 17 R. Bau, M. Y. Chang, C-Y. Wei, L. Garlaschelli, S. Martinengo, T. F. Koetzle, Inorg. Chem., 1894, 23, 4758. 18 L. Basini, A. Aragno and A. Raffaelli, J. Phys. Chem., 1991, 95, 211. 19 S. Kawi and B. C. Gates, Znorg. Chem., 1992,31,2939. 20 D. M. Adams, I. D. Taylor, J. Chem. SOC.,Faraday Trans., 1982, 78, 1573. 21 S. Kawi, J-R. Chang and B. C. Gates, J. Phys. Chem., 1993, 97, 5375. 22 K. Tanaka, K. L. Watters and R. F. Howe, J. Catal., 1982, 75, 23. 23 R. Psaro, C. Dossi, A. Fusi, R.Della Pergola, L. Garlaschelli, D. Roberto, L. Sordelli, R. Ugo and R. Zanoni, J. Chem. Soc., Faraday Trans., 1992,88,369. 24 D. H. Vandemberg, T. Chin-Choy and P. C. Ford, J.Organomet. Chem., 1989,366,257. 25 L. Malatesta and G. Caglio, J. Chem. SOC., Chem. Commun., 1967,420. 26 L. Garlaschelli, S. Martinengo, P. L. Bellon, D. Demartin, M. Manassero, M. Y.Chang, C-Y. Wei and R. Bau, J. Am. Chem. SOC.,1984,106,664. 27 M. Angoletta, L. Malatesta and G. Caglio, J. Organomet. Chem., 1975,94,99. 28 S. Kawi and B. C. Gates, J. Chem. SOC., Chem. Commun., 1992, 702. 29 S. Kawi, J-R. Chang and B. C. Gates, J. Catal., 1993,142, 585. 30 H. H. Lamb, A. S. Fung, P. A. Tooley, J. Puga, T. R. Krause, M. J. Kelley and B. C. Gates, J. Am. Chem.SOC., 1989,111,8367. 31 H. E. Fischer and J. Schwartz, J. Am. Chem. SOC., 1989, 111, 7644. 32 K. C. Kannon, S. K. Jo and J. White, J. Am. Chem. Soc., 1989, 111,5064. 33 S. Kawi, J-R. Chang and B. C. Gates, J. Phys. Chem., 1993, 97, 5375. 34 G. D. Mahan and A. A. Lucas, J. Chem. Phys., 1979,68, 1344. 35 B. N. J. Persson and A. Liebisch, Surf: Sci., 1981, 110, 356. 36 D. P. Woodruf, B. E. Heiden, K. Prince and A. M. Bradshaw, Surf. Sci., 1982, 123, 397. 37 H. Ueba, Surf: Sci., 1987,188,421. 38 R. Eisemberg and D. E. Hendrichson, Adu. Catal., 1979,28,79. 39 J. Darensbourg and C. Ovalles, J. Am. Chem. SOC., 1984, 106, 3750. 40 J. Pugh, M. R. M. Bruce and B. P. Sullivan, Znorg. Chem., 1990, 30,86. 41 P. Braunstein, D. Hatt and D. Nobel, Chem.Reu., 1988,88,747. 42 D. J. Darensbourg, H. P. Wiengraffe and P. W. Wiengraffe, J. Am. Chem. SOC., 1990,112,9252. 43 D. A. Palmer and R. van Eldik, Chem. Rev., 1983,83,651. 44 D. J. Darensbourg and R. Kudaraski, Adv. Organomet. Chem., 1993, 22, 129. 45 A. Berko and F. Solymosi, Surf: Sci., 1987,187,359. 46 Z. M. Liu, Y. Zhan, F. Solymosi and J. M. White, Surf. Sci., 1991,245,289. 47 J. Kiss, K. Renesn and F. Solymosi, Surf: Sci., 1988,207, 36. 48 A. Berko and F. Solymosi, Stud. Surf. Sci. Lett., 1986,171, L498. 49 R. J. Behm and K. Christmann, Surf. Sci., 1990,99,320. 50 R. J. Behm, V. Fenka, M. G. Cattania, K. Christmann and G. Ertl, J. Chem. Phys., 1983,78, 7486. 51 M. Ehsasi and K. Christmann, Surf: Sci., 1988,194, 172. 52 K. Christmann, M. Ehsasi, W. Hirschwald and J. H. Block, Chem. Phys. Lett., 1986,136, 192. 53 G. Lauth, E. Schwartz and K. Christmann, J. Chem. Phys., 1989, 91, 3729. 54 S. Fujita, H. Terunuma, M. Nakamura and N. Takezawa, Ind. Eng. Chem. Res., 1991,30, 1146, and references therein. 55 C. H. Bartholomew, Stud. Surf: Sci. Catal., 1991,64, 158. 56 F. Solymosi, A. Erdoheli and T. Basagi, J. Catal., 1981, 68, 371. 57 H. Shultz, K. Bek and E. Erick, Methane Conversion, ed. D. M. Bibby, C. D. Chang and S. Yurchak, Elsevier, Amsterdam, 1988, and references therein. 58 L. Basini, Ind. Eng. Chem. Res., 1989, 28, 659, and references therein. Paper 3/05914F;Received 1st October, 1993