J. CHEM. SOC. FARADAY TRANS., 1994, 90(2), 355-362 Thermogravimetry-FTIR Study of the Surface Formate Decomposition on Cu, CuCI, Cu,O and CuO Correlations between Reaction Selectivity and Structural Properties Jianyi Lin* Department of Physics, National University of Singapore, Singapore 0511 Koon Gee Neoh and Wah Koon Teo Department of Chemical Engineering, National University of Singapore, Singapore 0511 In this work three types of the surface formate, ionic, monodentate and bidentate species, are identified to be the main products of formic acid adsorption on Cu, CuCI, Cu,O and CuO at room temperature. Combined thermal analysis [differential thermal analysis (DTA), thermogravimetry (TG) and evolved gas analysis (EGA)] and FTIR studies have shown that the decomposition selectivity and activity are highly dependent on the electronic and structural properties of the surface formate.The ionic formate which gives the characteristic infrared bands at 1600 and 1320 cm-' is found to exist on all the surfaces studied in the lower temperature range (<490 K) and is easily (with an activation energy of 5-19 kcal mol-') decomposed at elevated temperatures, forming water as the main product. Bidentate formate with infrared bands at 1630 and 1350 cm-' is observed on the Cu, Cu,O and CuO surfaces and is decomposed at high temperatures (around 500 K) with an activation energy of 1539 kcal mol-' , producing mainly carbon dioxide. The adsorbed formate on CuCl at high temperatures demonstrates only an asymmetric OCO band at 1610 cm-' and is considered to be the monodentate formate, whose pyrolysis favours the production of water, with an observed activation energy of 19 kcal mol-'.The reaction scheme and the origin of the structural dependence of the surfaceformate decomposition are discussed. Formic acid has been shown to be adsorbed mainly as the surface formate on a variety of metals and oxides at room temperature. The decomposition of the adsorbed for-mates has been often used as a model system of catalytic selectivity studies since only one reactant is involved and the products of the reactions (both dehydration and dehydrogenation), CO,, H,, CO and H,O, can be easily monitored. Although due attention has not been paid to studies on copper oxides, a study on copper metal has re- cently been considered particularly attractive as the surface formate species is often thought to be a pivotal intermediate in several catalytic processes including methanol synthesis and water-gas shift.' 7-44 The correlation between the catalytic selectivity and physi- cal properties of the solids which adsorb formic acid has been the subject of systematic studies.'-' Our recent secondary ion mass spectrometry (S1MS)-X-ray photoelectron spectroscopy (XPS) study of formic acid adsorption on copper metal and copper(1) chloride surfaces led to the conclusion that the cata- lytic selectivity of the decomposition of formic acid is closely related to the structure of the surface formate species.45 This paper intends to verify the above assumption by using FTIR and thermal analysis techniques. FTIR of the adsorption layers can provide information on the structure of the surface formate, while thermal analysis allows for a kinetic study of the surface formate pyrolysis by a combination of simulta- neous DTA, TG and EGA.As it will be shown later, the catalytic selectivity and kinetic activity of the surface formate decomposition are indeed dependent on the structure of the surface formate over Cu, CuCl, Cu,O and CuO, with the ionic and monodentate formates favouring dehydration, whereas the bidentate covalent species favour dehydroge- nation. Experimental Preparation and Characterization of the Surface Formates The catalyst samples used in this experiment were all in the powder form as received (Cu, H&W, GR; CuCl, Merck, GR; Cu,O, Merck, GR; CuO, Fluka, AG).Formic acid adsorp- tion was carried out in a gas uptake+volution system where the sample could be evacuated by a rotary pump and the adsorption-desorption could be monitored by a mercury barometer. Normally the sample maintained at a low vacuum was first heated to 320-370 K for 1 h in order to pump off the contaminant gases. Then, after the sample was cooled to room temperature, formic acid vapour (ca. 40 mmHg), which was in equilibrium with liquid formic acid (Merck, GR) at room temperature, was introduced into the system and allowed to come into contact with the sample at room tem- perature for 1 h.The pressure of the system with the vapour was observed to drop, indicating that adsorption had occurred. The adsorbed layers thus prepared were characterized by FTIR on a Shimadzu FTIR-8101, using the pellet technique which involved mixing the finely ground sample and pot- assium bromide powder, and pressing the mixture in an evac- uable die at 8 tonne rn-, of pressure to produce a transparent disk. Strong bands associated with the OCO stretching were observed in the FTIR spectra obtained from these samples, indicating that at room temperature the formic acid was adsorbed mainly as surface formate on all the samples studied. These observations agreed well with those obtained from the earlier vibrational studies under ultrahigh vacuum conditions,".' demonstrating that the surface '~'~9~' formates were fairly stable in air at room temperature.In fact, it was found that samples after being exposed to air for a short period of time and then stored in capped sample bottles for several weeks still produced infrared spectra identical to that of the freshly prepared sample. Therefore, it was possible to study the changes of the surface formate layers during the course of pyrolysis by preparing samples in the gas uptake- evolution system at room temperature, heating to different elevated temperatures in the STA 409 instrument (vide infra) and testing on the FTIR-8101 spectrometer after the samples were cooled to room temperature. 40 scans were taken for each spectrum to obtain a good signal-to-noise ratio.Kinetic Studies of the Surface Formate Pyrolysis The kinetic study of the pyrolysis of the adsorption layers was performed in a simultaneous thermal analysis apparatus, Netzsch STA 409. Ca. 100-200 mg of the sample prepared as described above was placed in an open crucible and was heated from room temperature to CQ. 700 K at a linear rate of 5°C min-' in nitrogen which was used as the carrier gas through the furnace at a flow rate of 100 ml min-'. The change in weight during the experiment and the differential temperature, AT, which was the temperature difference between the sample and the reference material, were contin- uously recorded as a function of the sample temperature, giving TG and DTA curves.EGA by FTIR was performed on the Shimadzu FTIR-8201. The FTIR gas cell was directly connected to the outlet of the STA 409 so that a stable plug- flow state was reached, allowing the outlet gas to be contin- uously tested. As will be shown below, the combination of simultaneous DTA-TG-EGA allows kinetic and catalytic studies of the surface formate decomposition, providing infor- mation on the reaction selectivity, activation energy, calorim- etry and so on. Blank experiments using the solids without the adsorbed overlayer were carried out in parallel to the above described experiments, including both the thermal analysis and FTIR measurements, to make sure that the information obtained was truly from the adsorbed formate overlayers.Results and Analysis Decomposition of the Surface Formate on Cu Thermal Analysis: DTA, TG and EGA Results Fig. 1 displays the DTA-TG results obtained from pyrolysis of the formate-covered copper. At temperatures around 360 K, DTA evidently shows an endothermic heat flow while the TG data exhibit a weight loss. Further increasing the sample temperature to 440 K results in the second weight loss, accompanied by exothermic heat flow. Since the dehydration of the formic acid is known to be endothermic while the dehydrogenation is exothermic,' the TG-DTA data in Fig. 1 clearly demonstrate that the pyrolysis of the surface formate species on Cu favours dehydration at temperatures around 360 K, whereas dehydrogenation becomes predominant at higher temperatures (above 440 K).This conclusion can be also deduced from EGA by FTIR. Fig. 2 displays the infrared spectra of the evolved gas at various temperatures. The infra- red bands at 3500-4000 and 1250-2000 cm-' are known to 0.5 1.5 1.o 0 0.5 7$ -0.5 & 0% i I'd -DTA , -0.5 -1 .o -1 .o -1.5 -1.5 300 320 340360380400420440460480 500 TIK Fig. 1 DTA-TG data for the formate-covered Cu in N, at 5°C min-'. W stands for the total mass loss due to formate decomposi- tion at elevated temperatures and AW/W is the fraction of the formate decomposed. AT is the temperature difference between the sample and reference in mV (thermocouple reading) and represents the heat flow due to formate pyrolysis. J. CHEM.SOC. FARADAY TRANS., 1994, VOL.90 4000 2000 1500 wavenurnber/crn-' Fig. 2 Infrared spectra of the evolved gas from the formate-covered Cu at 310 (a),340 (b),360(c),410 (4,450 (e)and 480K cf) be due to the water vapo~r,~~ whereas the band at 2340-2361 cm-' is due to CO, . Hydrogen, which is a homopolar mol- ecule, is not infrared active. CO is observed normally as a few lines between 2000 and 2250 cm-'. Though the molar absorbability of CO is found to be lower than those of H,O and CO,, note that the amount of CO produced from decomposition of surface formate is lower than that expected based on the stoichiometry of the dehydration of formic acid, which should give rise to equimolar amounts of CO and H,O. Since the ability for the instrument to detect CO has been proven by injecting a trace of CO into the outlet of the STA 409 and obtaining a signal in the corresponding wave- number region, this fact may suggest a reaction scheme for the surface formate decomposition where some of the CO produced from the dehydration may react further with adsorbed OH, adsorbed oxygen or lattice oxygen, as will be shown below. It is obvious from Fig.2 that H,O is the major component in the evolved gas at low temperatures (300-380 K). The water signal becomes weaker at 410 K and then retains high intensity up to 480 K. The CO, signal increases more rapidly at higher temperatures and becomes very pro- nounced at temperatures between 440 and 460 K. At 480 K, the CO, intensity decreases, indicating that the decomposi- tion of the surface formate is almost completed.The conser- vation of the water signal at temperatures above 410 K may result from some side reactions including the proportionation reaction of the surface OH group, which may take place at high temperatures. It is possible to estimate the activation energy of the surface formate pyrolysis using the TG data obtained. Decomposition of the adsorbed formate is known to be a first-order reaction., It has been reported47 that for a first-order reaction the fraction of the reactant decomposed at temperature T, AW/W,and the activation energy of the reac- tion, E, ,have the following relationship: log[-log(l -AW/W)/T2]= log[(AR/aE,)(l -2RT/E,)] -EJ2.3RT where A is the frequency factor of the reaction, a the linear heating rate and R the molar gas constant. In Fig.3, Y = log[ -log(l -AW/W)/T2], calculated using the TG data in Fig. 1, is plotted against 1/T. The data points are best fitted with straight lines, L1 and L2, yielding an activation J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 -5.2 - \ 1 -5.4 -5.6 .1 'rcL2 I -2% I -5.8 Y -6.0 -6.2 -6.4 -6.6 -6.8 0.0021 0.0023 0.0025 0.0027 0.0029 KIT Fig. 3 Y us. 1/T plot for the formate-covered Cu: Y = log[-log(1 -AW/W)/T2],where AW/W is the fraction of the formate decomposed and T the pyrolysis temperature. Data (H)are obtained based on the TG results in Fig. 1 ;the data are best-fitted by two straight lines (L1 and L2), giving the activation energy E = 7 kcal mol-' at low temperatures and E = 15 kcal mol-' at high temperatures.energy of 7 kcal mol-' for the formate pyrolysis over Cu at 340-410 K and 15 kcal mol-' at 410-460 K. Note that the activation energy obtained as described above is based on the loss in sample weight (i.e. the consumption of reactant) and is therefore the apparent activation energy for the overall reac- ti~n.~* FTIR Spectra of the Adsorbed Layers Fig. 4 shows the infrared spectra for the formate-covered Cu surfaces at different temperatures. Included at the top of the figure, Fig. 4(a) is obtained from the bulk copper@ formate in the wavenumber region between lo00 and 2000 cm-'. Similar to the known spectrum for the bulk sodium formate which has IR bands at 1567, 1377 and 1366 cm-', the bands at 1600, 1380 and 1320 cm-' may be assigned to v,,(OCO), 6(CH) and vs(OCO), respecti~ely.'~*~~Fig. 4(b) is for the formate-covered Cu surface at room temperature. The most c 4000 2000 1500 1000 wavenumber/cm-' Fig.4 Infrared spectra of the formate adsorbed on Cu at 300 (b), 380 (c), 480 (d) and 720 K (e).Spectrum (a) is obtained from bulk copper(1) formate and included for comparison. 357 prominent bands in this spectrum are the asymmetric OCO stretching vibration, va,(OCO),in the 1600-1640 cm-' region and the symmetric vibration, v,(OCO), at 1320-1380 cm-'. Note also that the v,,(OCO) band is composed of two well resolved peaks at 1600 and 1630 cm-' while the v,(OCO) band, which is peaked at 1350 cm-',is asymmetric in shape, having a shoulder a 1320 cm-'.Besides these peaks there is also a signal observable at 1380 cm-'.This appears to imply that there may exist more than one chemisorption state for surface formate on polycrystalline copper surfaces. The bands at 1600, 1380 and 1320 cm-' can be ascribed to the ionic complex according to Fig. 4(a)for the bulk copper@) formate. On the other hand, the bands at 1630 and 1350 cm-'appear to be due to the bidentate surface formate since previous vibrational studies have observed bands at 1650 and 1348 cm-' for formate-Cu( 110) surface13 and at 1640 and 1330 cm-' for formate-Cu(lOO),'O respectively, and the geometric structure for the formate on these two well defined surfaces is either a chelating bidentate on Cu(ll0) or a bridging biden- tate on Cu(100),as known from studies using extended X-ray absorption fine structure (EXAFS),'-'*'' see Fig.5(a) and (b). Further support may be found from the study of the formate- covered GaAs(ll0) surface, where the bands at 1640 and 1350 cm-' are also attributed to the bidentate formate.26 When the sample temperature is increased to 380 K, the OCO infra-red band at 1600 cm-', which is assigned to the ionic species, evidently decays while the 1635 cm-' band remains fairly strong [see Fig. 4(c)J.This appears to indicate that the ionic surface formate is less stable and decomposed in the lower temperature range than the bidentate formate which is mostly decomposed at temperatures higher than 380 K.At ca. 500 K, almost all of the signals due to the formate dimin- ish, which is indicative of the completion of the formate pyrolysis and is in good agreement with that reported from temperature-programmed desorption (TPD)' and ultraviolet photoelectron spectroscopy (UPS)4 studies. The bands at 3453,2920, 2360, 1070, 760,670 and 620 cm-', which can be assigned to v(OH), V(CH), v(CO,), n(CH), s(OCO), 74C0,) and v(Cu'-0), respectively, according to the literat~re,'~*~~ may be observed at some temperatures but will not be dis-cussed in detail. three-electron H one-electron donor 0 R I cu-0' monatomic chelate monodentate with additional bridging (9) (h) Fig.5 Structural models proposed in the literature' 7,4531-61 for the bidentate bridging formate on Cu(100) (a), bidentate chelate formate on Cu( 1 10) (b),hydrogen-bonded monodentate formate on CuCI(11 1) (c), and for the carboxylato copper complexes, (d)-(i). Surface Formate Decomposition on CuCl DTA, TG and EGA Results The TG profile of the formate-CuC1 in Fig. 6 shows two weight losses at the temperatures around 440 and 510 K, respectively;while DTA displays two endothermic heat flows in the corresponding temperature ranges. In agreement with the DTA data, EGA by FTIR shows that water vapour is the main product from the pyrolysis of the formate adsorbed on CuCl in the whole temperature range studied, with some CO, being detected mainly at temperatures between 460 and 500 K (see Fig. 7).The activation energy, estimated using the same procedures as described above, is 5 kcal mol-' between 380 and 490 K, and 19 kcal mol-' at 490-560 K. FTIR Spectra of the Adsorbed Layers Fig. 8 illustrates the FTIR spectra of the formate-covered CuCl at a number of different temperatures. Between room temperature and 390 K, the asymmetric OCO stretching vibration band is peaked at 1600 crr-', while in the sym- metric stretching vibration region there are two peaks at 1380 and 1320 cm-', respectively [see Fig. 8(a)]. Following the assignment for the formate-Cu system, these infrared bands 1 10.5oi II 1 -1.5 I --1.57 300 350 400 450 500 550 600 T/K Fig. 6 DTA-TG profiles for the formate-covered CuCl 4000 3000 2000 1750 1500 1250 wavenumber/cm-' Infrared spectra of the evolved gas from the formate-covered Ckl at 310 (a),340 (b),370 (c),470 (4,500 (e)and 520 K cf) J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 4000 2000 1500 wavenumber/cm-' Fig. 8 Infrared spectra for the formate adsorbed on CuCl at 300 an2 390 K (a),510-K(b) are due to the ionic surface formate. After heating to 510 K the infrared bands at 1320 and 1380 cm-', which are due to the OCO stretching and the CH in-plane deformation, disap- pear and the peak at 1600 cm-', which is assigned to the asymmetric OCO stretching, is now shifted to higher wave- number, located at 1610 cm-'. These facts suggest that the adsorbed formate on the CuCl surface at high temperatures may exist mainly in the monodentate configuration [Fig.5(c)], as previously suggested in our SIMS-XPS study.45 Note that the OH band in Fig. 8 appears to have a tail at its lower wavenumber side. The tail seems to be indicative of the presence of hydrogen bonding since the stretching vibration of the Cl-H bond is located in this energy range (ca. 2991 cm-l ).46 Close values for the asymmetric OCO vibration of the monodentate surface formate have been reported on Cu/SiO, (1640 ~m-'),~~ Cu(100) at 100 K (1640 cm-'),lo Ru(OO1) (1600 cm-1),22*24 A120, (1621 and Pt(ll1) (1620 cm")." Note that there is no band in the region between 1320 and 1380 cm-' as expected for the symmetric OCO vibration. The absence of the symmetric OCO vibration (1320-1380 cm-') has been reported on the formate-covered Cu-deposited KBr surface.It is attributed to the roughness of the surface and explained in terms of the selection rule for dipolar excitations. l4 Alternatively, the absence of the asym- metric OCO vibration in the region between 1560 and 1680 cm-' for a bidentate formate was reported on Cu(l10),'3 Cu(100)" and R~(00l),~~ and was interpreted on the basis of the selection rule. It appears that the change of the orienta- tion of the dipole moment with respect to the electric field may also be the most probable consideration for the absence of the symmetric OCO band in the case of formate-CuC1. Nevertheless, since the CuCl used in the present experiment was in powder form and the infrared light was unpolarized, the selection rule may not be applicable and the reason for the absence of the symmetric vibration is still not very clear.Surface Formate Decomposition on Cu,O DTA, TG and EGA Results The TG data in Fig. 9 show a weight loss occurring at tem- peratures between 380 and 450 K. This is an endothermic process as shown by the DTA profile in the same figure. EGA by FTIR indicates that the main product of the formate pyrolysis in this temperature range is water vapour. A signifi- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 $ -0.5-d -1.5 300 350 400 450 500 -0.5 550 T/K Fig. 9 DTA-TG data of the formate-covered Cu,O cant weight loss follows immediately at temperatures between 450 and 500 K, with an evident heat release.EGA shows that in this temperature range carbon dioxide is the main product from the pyrolysis of the formate adsorbed on Cu,O. The activation energy is estimated to be 9 kcal mol-I at tem- peratures between 380 and 450 K and 24 kcal mol- ' at 450- 520 K. FTIR Spectra of the Adsorbed Layers The FTIR spectra of the formate-covered Cu,O are displayed in Fig. 10. The Cul-0 stretching vibration is observed at 620 cm-' as a huge band. It does not change during pyrol- ysis and is therefore not presented in the figure. Similar to the spectra of the formate-Cu, the v,,(OCO)band consists of two rather weak and can be observed only as a small bump. When the temperature is higher than 510 K all the bands due to the formate ion disappear, indicating that the decomposi- tion is complete.Surface Formate Decomposition on CuO DTA, TG and EGA Results The DTA-TG data of the formate-covered CuO are illus- trated in Fig. 11 where a small weight loss occurs around 350 K. This is an endothermic process as shown by the DTA profile in the same figure. EGA by FTIR indicates that the main product of the formate pyrolysis in this temperature range is water vapour. A significant weight loss, with an evident exothermic heat flow, is observed at temperatures between 460 and 530 K. EGA shows that carbon dioxide is the main component in the evolved gas at high temperatures. The activation energy estimated based on the TG data is 14 kcal mol-' at 320-360 K and 39 kcal mol-' at 460-530 K.FTIR Spectra of the Adsorbed Layers Fig. 12 illustrates the FTIR spectra of the formate-covered CuO. At room temperature, the formate-CuO surface gives a va,(OCO)band centred at 1600 cm-'and an overall v,(OCO) band with three peaks at 1380, 1360 and 1330 cm-'.Increas-ing the temperature to 420 K results in the decay of the t3 peaks at 1600 and 1635 cm-' while the v,(OCO)band is composed of the peaks at 1350 and 1380 cm-',respectively. The bands at 1600 and 1380 cm-' may be attributed to the -0.5 QO::DTAionic complex, and those at 1635 and 1350 cm-' to the bidentate formate. With increasing temperature, the 1600 cm-' band decreases faster than the 1635 cm-' band as -1 .o shown in Fig.lqb). At 470 K, only the peak at 1630 cm-' is observed [see Fig. lqc)]. The 1350 cm-' band becomes -1.5 300 350 400 450 500 550 600 TIK Fig. 11 DTA-TG profiles of the formate-covered CuO 4000 2000 1500 4000 2000 1500 wavenumber/cm-wavenumber/cm-Fig. 10 Infrared spectra of the formate adsorbed on Cu,O at 300 Fig. 12 Infrared spectra of the formate adsorbed on CuO at 300 (a), (a),400 (b) and 470 K (c) 420 (b) and 520 K (c) 360 formate bands, with the absorption band at 1600 cm-' decreasing more than that at 1630 cm-' [see Fig. ll(b)]. Based on the assignments for the formate-Cu surface, the band at 1600 cm-' may be due to the ionic surface complex while the 1630 cm-' band is attributed to the bidentate formate.The ionic complex is decomposed at relatively low temperatures, while the bidentatic formate may be decom- posed mostly at higher temperatures. The Cu"-O stretching vibration is observed at 530 cm-' but is not presented in this figure. Discussion As summarized in Table 1, three types of adsorbed formate species, ionic, monodentate and bidentate formate, have been identified by the infrared study as the main products of formic acid adsorption on Cu, CuC1, Cu,O and CuO surfaces at room temperature. The ionic formate, which gives the characteristic infrared bands at 1600 and 1320 cm-',is found to exist on all the surfaces studied in the low-temperature range (<490 K) and is easily decomposed at elevated tem- peratures, forming water as the main product.The bidentate formate, with infrared bands at 1630 and 1350 cm-I, is the most abundant species observed on Cu, Cu,O and CuO sur- faces, based on the TG data, and is decomposed at high tem- peratures (around 500 K), producing mainly carbon dioxide. The adsorbed species observed on the CuCl surface at high temperatures exhibits only an asymmetric OCO band at 1610 cm-' with little infrared absorption in the energy region for the symmetric OCO vibrations and is considered as the monodentate formate whose pyrolysis favours the production of water. These observations suggest that the selectivity of the surface formate decomposition may not really depend on the type of catalyst (i.e. whether metal, transition metal oxide or other metal oxide) as often proposed,' but, may actually depend on the structural properties of the formate formed on these materials, i.e.whether the surface formate is ionic or covalently bonded, and whether it is monodentate or biden- tate in geometric configuration. Note that in Table 1 the acti- vation energy of formate pyrolysis is lowest (ca. 5-14 kcal mol-') for the ionic formate, intermediate (19 kcal mol-') for the monodentate and highest (15-39 kcal mol-') for the bidentate surface formate. In particular, for the bidentate formate, the pyrolysis activation energy is shown to be the highest on CuO (39 kcal mol-') and lower on Cu,O and Cu (24 and 15 kcal mol-', respectively). Since the electron charge transfer from the formate ligand to the copper atom/ion takes place to an increasing extent in the fol- lowing sequence: ionic < monodentate (one-electron donor)" < bidentate (three-electron donor)' ' formate and Cuo < Cu' < Cu", it appears that the kinetic activity of the pyrolysis decreases (i.e.the activation energy increases), with increasing electron charge transfer from the formate to the metal atom/ion. This agrees well with our previous SIMS- XPS study4' where the greater electron charge transfer is J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 considered to be responsible for the more stable OCO bonding on Cu compared with that on CuCl. UPS studies4** and theoretical molecular orbital calc~lations~~ indicated that the highest occupied molecular orbitals of the formate ion are la,, 6a,, 4b,.These orbitals are mainly due to the oxygen atoms and are most strongly involved in bonding to metal, donating electron charge to the surface. In particular, the la, orbital, which has .n symmetry, is essentially the anti- symmetric combination of the 0 2p, orbitals and is thus antibonding with respect to the 0-C-0 bond. Therefore, the removal of electron density from this orbital would result in the relative strengthening of the OCO bonding.l3 In con- trast, less electron charge transfer is involved in the ionic bonding interaction or in the monodentate configuration, leading to a less stable 0-C-0 bond. This conclusion can find support both from the previous SIMS results4' and from the present FTIR studies.In the SIMS studies the OCO adspecies is observed to be more stable on Cu than on CuCl, while the FTIR results show that the 0-C-0 stretching frequencies (and thus the OCO bonding) are reduced from 1630/1350 for the bidentate formate to 1610/-for the mono- dentate species and 1600/1320 cm-' for the ionic complex. According to the above conclusion, the decomposition of a bidentate formate would favour dehydrogenation owing to its relatively stable OCO bonding, while a further splitting of the OCO bond may occur with smaller activation energy for the ionic or monodentate formates (because of the less stable OCO bonding), resulting in the predominant dehydration. This can be best described in the following reaction scheme, Scheme 1, which is modified from a reaction mechanism sug- gested by.1glesia and Boudart :'3 /H.* CO, -+CO,(g) + H(ads) HCO,(ads) He .(CO).. .O + CO(ads) + OH(ads) CO(ads) CO(g) CO(ads) + O(ads or lattice) eCO,(g) OH(ads) + CO(ads)e CO,(g) + H(ads) OH(ads) + OH(ads)e H,O(ads) + O(ads) H(ads) + OH(ads) eH,O(ads) H,O(ads) eH,O(g) H(ads) + H(ads) ~t H,(g) Scheme 1 The proposed reaction scheme involves an activated complex as the reaction intermediate, in which the C-H bond is weakened through coordination to the catalyst Table 1 Kinetic activities for surface formate decomposition on Cu, CuCI, Cu,O and CuO surfaces" ionic (1600, 1 320)b monodentate (1610, -)* bidentate (1630, 1350)b AH E.3 AH Ea AH Ea surface T/K /kcal mol-' /kcal mol-' T/K /kcal mol-' /kcal mol-' T/K /kcal mol-' /kcal mol-' cu 340-430 endo 7 ---430-480 exo 15 CuCl 380-490 endo 5 490-560 endo 19 ---CU,O 380-460 endo 9 ---460-520 exo 24 CUO 320-360 endo 14 ---460-530 exo 39 ~~ T,decomposition temperature; AH, heat flow; E,, activation energy; endo, endothermic; exo, exothermic.* IR bands in cm- '. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 361 surface. If the HCO,(ads) adopts a bidentate configuration, the OCO bonding is relatively stable and the decomposition proceeds mainly through reaction (1) which leads to the for- mation of gaseous CO,. On the other hand, the ionic/ monodentate HCO,(ads) in which the OCO bonding is relatively unstable would follow reaction (2) forming CO(ads) 6 7 8 J.Stohr, J. L. Gland, W. E. Eberhard, D. Outka, R. J. Madix, F. Sette, R. J. Koestner and U. Dobler, Phys. Rev. Lett., 1983, 51, 2414. A. Puschmann, J. Haase, M. D. Crapper, C. E. Riley and D. P. Woodruff, Phys. Rev. Lett., 1985,54, 2250. Th. Lindner, J. Somers, A. M. Bradshaw and G. P. Williams, Surf: Sci., 1987, 185, 75, and references therein. and OH(ads) which then further react giving water as the main product. In this scheme the C-H cleavage is viewed as the rate-determining Since the O-H bond of the bidentate HCO,(ads) is normally thought to be in a molecu- lar plane perpendicular to the surface and is away from the ~urface,~~,~",~~a tilting of the HCO,(ads) species may be 9 10 11 P. Hofmann, C. Mariani, K. Horn and A. M. Bradshaw, in Proc. 4th Int.Conf: on Solid Surfaces and 3rd European Cont on Surface Science, ed. D. A. Degras and M. Costa, Paris, 1980, p. 541. B. A. Sexton, Surj Sci., 1979,88, 319. L. H. Dubois, T. H. Ellis, B. R. Zegarski and S. D. Kevan, Surf: Sci., 1986, 172, 385. required in order for the C-H bond to approach the surface. It appears that less energy may be needed for the decomposi- tion of a monodentate formate, which has only one oxygen atom bound to the surface, allowing for the free rotation of the C-H bond toward the surface, than of a bidentate formate with two metal-oxygen Thus, it is under- 12 13 14 15 P. Sen and C. N. R. Rao, Surf. Sci., 1986,172,269. B. E. Hayden, K. Prince, D. P. Woodroof and A. M. Bradshaw, Surf: Sci., 1983, 133, 589. M. Ito and W. Suetaka, J. Phys.Chem., 1975,79, 1190. R. J. Madix, in Vibrations at Surfaces, ed. C. R. Brundle and H. Morawitz, Elsevier, Amsterdam, 1983, p. 25; Surf: Sci., 1979, 89, 540, and references therein. standable that a higher pyrolysis activation energy may be found for the surface formate in the bidentate configuration than for the ionic/monodentate form. Cu, Cu,O and CuO are known to have very different crystal structures, but the surface formates on these catalysts are all dominated by the bidentate species as inferred from 16 17 18 P. Hofmann, S. R. Bare, N. V. Richardson and D. A. King, Surf: Sci., 1983, 133, L459. D. A. Outka and R. J. Madix, in Chemistry and Physics of Solid Surfaces, ed. R. Vanselow and R. Howe, Springer Verlag, Berlin, 1986, vol. 6, p. 133, and references therein.B. A. Sexton, A. E. Hughes and N. R. Avery, Surf. Sci., 1985, 155, 366. infrared studies. This similarity may be derived from the pres- ence of many possible structural configurations for the biden- tate formate ligand. On well defined copper surface^'^,'^ the formates are known to have either chelating or bridging bidentate configurations [Fig. 5(a) and (b)].In most of the 19 20 21 F. C. Henn, J. A. Rodriquez and C. T. Campbell, Surf: Sci., 1990, 236,282. C. Egawa, I. Doi, S. Naito and K. Tamaru, Surf: Sci., 1986, 176, 491. S. W. Jorgensen and R. J. Madix, J. Am. Chem. SOC., 1988, 110, 397. carboxylato Cu"-Cu' complexes, whose crystal structures 22 F. Solymosi, J. Kiss and I. Kovacs, J. Phys. Chem., 1988,92, 796; have been determined by X-ray or neutron diffraction, the carboxylate ions may exist as chelating bidentate ligands, i.e.as a bridging bidentate ligand in a syn-syn, anti-anti, anti-syn configuration [see Fig. 5(d)-(l1)].~"-~"The Cu-Cu distance within those carboxylato copper complexes ranges from 2.64 to 5.77 A. The above features found for the carboxylato copper complexes appear to provide flexibility, allowing a surface formate to adopt a bidentate configuration among several possible coordination geometries and to fit the surface structures over the Cu, Cu20 or CuO catalysts. Among the carboxylato Cu"-Cu' complexes with known crystal struc- tures there are a few exceptions where the carboxylato group 23 24 25 26 27 28 29 30 Surf. Sci., 1987, 192, 47. F. Solymosi and I. Kovacs, Surf: Sci., 1991,259,95.M. Chtats, P. A. Thiry, J. P. Delrue, J. J. Pireaux and R. Caudano, J. Electron Spectrosc. Relat. Phenom., 1983, 29, 293. B. H. Toby, N. R. Avery, A. B. Anton and W. H. Weinberg, J. Electron Spectrosc. Relat. Phenom., 1983, 29, 3 17. R. Matz and H. Luth, Surf. Sci., 1982,117, 362. M. A. Karolewski and R. G. Cavell, Surf: Sci., 1989, 210, 175; 219, 261. B. F. Lewis, M. Mosesman and W. H. Weinberg, Surf. Sci., 1974, 41, 142. J. Klein, A. Leger, M. Berlin, D. Defourneau and M. J. L. Sang- ster, Phys. Rev. B, 1972, 7, 2336. K. Yamashita, S. Naito and K. Tamaru, J. Catal., 1985,94,353. is bound in the monodentate configuration [Fig. 5(i)].65 The fact that the surface formate adsorbed on CuCl is observed to function as a monodentate ligand [Fig.5(c)], different from those over Cu, Cu,O and CuO, may be related to the rela- tively high ionic character of the bonding in CuCl and the 31 32 33 K. Lui, M. Vest, P. Berlowitz, S. Akhter and H. H. Kung, J. Phys. Chem., 1986,90,3183. J. M. Vohs and M. A. Barteau, Surf: Sci., 1986, 176, 91; 1988, 197, 109. Y. Noto, K. Fukuda, T. Onishi and K. Tamaru, Trans. Faraday SOC., 1967,63,2300; 3072. presence of the highly electronegative anion in the vicinity of 34 A. Ueno, T. Onishi and K. Tamaru, Trans. Faraday SOC., 1970, the copper ions, which may lead to the formation of hydro- gen bonding, favouring the monodentate configuration. 35 66,756. M. Bowker, H. Houghton and K. C. Waugh, J. Chem. SOC., Faraday Trans., 198 1,77,3023. We thank Prof. E.I. Solomon of Stanford University and Dr. J. Y. Lee for critical reading and valuable comments prior to the submission of this paper. Experimental assistance from Mr. K. W. Wong is gratefully acknowledged. 36 37 38 D. G. Walmsley, in Vibrational Spectroscopy of Adsorbates, ed. R.F. Willis, Springer-Verlag, Berlin, 1980, p. 67. A. Deluzarche, J. P. Hindermann, R. Kieffer and A. Kienne- mann, Rev. Chem. Int., 1985,6, 255, and references therein. P. A. Taylor, P. B. Rasmussen and I. Chorkendorff, J. Phys.: Condens. Matter, 1991,3, s59. 39 K. Waugh, Catal. Lett., 1990, 7, 345. 40 G. C. Chinchen, K. C. Waugh and D. A. Whan, Appl. Catal., 1986, 25, 101. References P. Mars, J. J. F. Scholten and P. Zwietering, in Advances in Catalysis, ed. D. D. Eley, H. Pines and P.B. Weisz, Academic Press, New York, 1963, vol. 14, p. 35, and references therein. J. Fahrenfort, L. L. Van Reyen and W. M. H. Sachtler, in Proc. Symp. Mechanism of Heterogeneous Catalysis, ed. J. H. Boer, 41 42 43 44 R. G. Herman, in New Trends in CO Activation, ed. L. Guczi, Elsevier, Amsterdam, 1991, p. 265, and references therein. V. Ponec, in ref. 41, p. 117, and references therein. K. Klier, Adv. Caral., 1982,31,243. G. W. Bridger and M. S. Spencer, in Catalyst Handbook, ed. M. V. Twigg, Wolf, London, 2nd edn., 1989, p. 446, and refer- ences therein. Elsevier, Amsterdam, 1960, p. 23. 45 J. Lin, A. T. S. Wee, A. C. H. Huan and K. L. Tan, Surf: Sci., P. Mars, in ref. 2, p. 49. R. W. Joyner and M. W. Roberts, Proc. R. SOC.London, A, 1976, 350, 107. D.A. Outka, R. J. Madix and J. Stohr, Surf. Sci., 1985,164,235. 46 47 1993, 285, 31. K. Nakamoto, in Infrared and Raman Spectra of Inorganic and Coordination Compounds, 4th edn., and references therein. A. W. Coats and J. P. Redfern, Nature (London), 1964,201,68. 362 J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 48 M. A. Newton, S. M. Francis and M.Bowker, in Catalysis and Surface Characterisation, ed. T. J. Dines, C. H. Rochester and J. 58 59 D. A. Langs and C. R. Hare, Chem. Commun., 1967,890. B. W.Skelton and T. N. Water, J. Chem. SOC., Dalton Trans., 49 Thomson, Royal Society of Chemistry, Cambridge, 1992, p. 165. G. J. Miller, C. H. Rochester and K. C. Waugh, J. Chem. SOC., Faraday Trans., 1991,87, 1491. 60 61 1972,2133. G. A. Barclay and C. H.L. Kennard, J. Chem. SOC., 1961,3289. G. B. Drew, D. A. Edwards and R. Richards, J. Chem. SOC., 50 51 N. R. Avery, Appl. Surf: Sci., 1983,14,149. F. A. Cotton and G. Wilkinson, in Advanced Inorganic Chem- 62 Chem. Commun., 1973,124. D. A. Edwards and R. Richards, J. Chem. SOC., Dalton Trans., 52 istry, Wiley, New York, 4th edn.,1980, p. 63. S. D. Peyerimhoff and R. J. Buenker, J. Chem. Phys., 1969, 50, 1846. 63 1973,2463. K. Okada, M. I. Kay, D. T. Cromer and I. Almodovar, J. Chem. Phys., 1966,44, 1648. 53 E. Iglesia and M. Boudart, J. Phys. Chem., 1991,9!5,7011. 64 A. G. Massey, in The Chemistry of Cu, Ag and Au, Pergamon 54 J. Catterick and P. Thornton, Adv. Inorg. Chem. Rudiochem., 1977,20, 291. 65 Press, Oxford, 1973, p. 1. K. Anzenhofer and L. N. A. Ten Rouwelaar, Rev. Trav. Chim., 55 C. Oldman, Progr. Inorg. Chem., 1968,10,223. 1967,86,801. 56 57 G. B. Deacon and R. J. Philips, Coord. Chem. Rev., 1980,33,227. J. N. Van Niekerk and F. R. L. Schoening, Acta Crystallogr., 1953,6, 227. Paper 3/02662K; Receioed 1lth May, 1993