J. Chem. SOC., Faraday Trans. 1, 1986, 82, 747-758 Surface Reduction in Low-temperature Formation of Nickel Oxide from Different Nickel Salts Thermal Analysis, Surface Area, Electron Microscopy and Infrared Studies Wayne R. Pease,? Robert L. Segall, Roger St. C. Smart" and Peter S. Turner School of Science, Grifith University, Nathan, Queensland 41 11, Australia The influence of source material, decomposition and annealing temperatures on the reactivity of nickel oxide surfaces has been studied. The process of thermal decomposition in air of four nickel salts (acetate, oxalate, carbonate and hydroxide) was followed using thermogravimetric and differential thermal analysis, differential scanning calorimetry, B.E.T. surface area determination and infrared spectroscopy. Nickel acetate and oxalate gave products consisting of nickel metal and nickel oxide.The reduction mech- anisms are described. The carbonate and hydroxide gave only nickel oxide, but the martensitic decomposition of the carbonate (AHt = 28.8 kJ mol-l) is considerably more energetic than that of the hydroxide (AHt = 2.8 kJ mol-l). The loss of surface area in the decomposition is correspondingly larger for the carbonate. Infrared studies of the degassing process from 25-500 "C in uacuo produced evidence for reduction on the ex-carbonate NiO surface which correlated with reactivity in dissolution kinetics studies. An absorption at 2193 cm-l was observed, with maximum intensity after 300 "C decomposition, assigned to an Ni . . . (O-C-0)- species resulting from CO interaction with surface 0- ions.Other carbonate and hydroxyl surface groups were characterised for each oxide at different decomposition temperatures. The influence of source material, decomposition temperature and procedures (e.g. in air or in vacuum) and annealing temperature on the reactivity of nickel oxide is well known [see, for example, ref. (1)-(3)]. Our work has been concerned with a particular form of reactivity through the study of the kinetics of dissolution in dilute acidic ~olution.~-~ The role of surface (and bulk) defects in this process has been central to the definition of mechanisms controlling dissolution rates in nickel oxides prepared at different annealing temperatures. The defects involved may be described variously as metal ion vacancies and excess oxygen ions in a p-type semiconductor;' low-coordination (i.e.< five-fold) sites in an ionic oxide;s or reactive Ni3+ and 0- sites in a localised adsorption model.*. Recently it has also been shown that surface reduction to produce defects in the form of metal atoms can result from reaction of surface carbon with lattice oxygen at high temperatures.lO7 l1 Since defect concentrations are largely controlled by the decomposition process and subsequent a n n e a l i ~ ~ g , ~ > ~ a study of nickel oxides formed from different nickel salts under different conditions may provide some insight into the correlation between defect formation and dissolution kinetics. In this paper we describe studies of the thermal decomposition processes in nickel acetate, oxalate, carbonate and hydroxide using thermogravimetric analysis (t.g.a.), differential thermal analysis (d.t.a.), differential scanning calorimetry (d.s.c.) and B.E.T.surface area measurements. Surface adsorbed species were characterised by i.r. spectro- t Present address : Government Chemical Laboratories, William Street, Brisbane, Queensland 4000, Australia. 747748 Surface Reduction of Nickel Oxide scopy during degassing of the partially decomposed salt. Alterations in crystal structure, surface structure and dislocation densities have been followed with X-ray diffraction and high-resolution transmission electron microscopy. Both bulk and surface reduction processes have been identified and a basis for comparison of defect concentrations based on the energy of the decomposition process is established.Experiment a1 Materials Nickel acetate and nickel carbonate were supplied (J. T. Baker Chemical Co.) as (CH,COO), Ni - 4H,O (maximum impurity 0.006% Co) and NiCO, (maximum impurities 0.01 % nitrogen and 0.01 % sulphate), respectively. Nickel hydroxide was supplied (Ajax Chemicals) as Univar nickel oxide (maximum impurity 0.01% Fe). The process of manufacture of this compound produces the ' hydrated oxide' by purification of the nickel solution, precipitation of the hydroxide followed by low-temperature drying. l2 Nickel oxalate was prepared in our laboratory by precipitation from a nickel nitrate solution (Ajax Unilab, maximum impurity 0.17; SO:-) with oxalic acid (Ajax Univar, maximum impurity 0.005 % SO:-). The precipitate was extensively washed with doubly distilled, deionised water, dried and stored under vacuum.The purity of the product was checked by X-ray diffraction and i.r. spectroscopy. No trace of sulphate or other impurity was detected . Procedures Thermal Analysis T.g.a. and d.t.a. analyses were carried out on a Rigaku TG/DTA standard temperature unit M8075 using Al,O, as the reference material. All materials were heated from room temperature to 720 "C at 10 "C min-l. Masses used were normally ca. 20 mg. D.s.c. measurements were made on a Perkin Elmer DSC-2 calorimeter. The reference material was A1,0, and the mass of source material was varied according to the size of the exothermic or endothermic peak(s) to be investigated. The calorimeter was calibrated against a known mass of indium.Sample Decomposition Techniques Nickel oxide powders used in surface area measurements were prepared in sintered A1,0, Alsint boats in a 2.4 kV Pyroco tube furnace fitted with a Pythagoras tube. The Pythagoras ceramic (Haldenwanger Ceramics) is made from mullite (A1,0, + SiO,), is impervious to gases, resistant to temperature change and is normally used at temperatures up to 1450 "C. The sample was heated gradually in air from 23 "C to the required annealing temperature over ca. 1 h and held at the final temperature for 4 h. The sample was then removed from the furnace and cooled to ca. 100 "C in < 10 min in air. Surface Area Determination Surface areas were measured with a Perkin Elmer 212D sorptometer using N, adsorption in He carrier gas and assuming the cross-sectional area of N, to be 1.627 x m2 at - 195.8 "C.The apparatus was purged for > 12 h with helium before each run and calibrated with IUPAC standards in the range 0.2-500 m2 g-l. Accuracy varies from - +20% at 0.2 m2 g-l to &2.5% at 200 m2 g-l.W. R . Pease et al. 749 Infrared Studies The material for i.r. study was prepared by decomposing the hydroxide or carbonate salt in air for 4 h at 550 "C. Coherent pellets were used without support material for all i.r. studies. The end-pieces of a Specac 13 mm diameter die were coated with PTFE, impregnated with the oxide from a particular nickel salt at low pressure loads (i.e. <3 tonne) until the exposed surface was fully covered with the pressing material, and then pellets were produced (ca.0.5 mm thick) at the same low pressures. They generally had an i.r. transmittance of cu. 10% at 2000 cm-l. Contamination of the NiO pellet surfaces prepared in this way were monitored using X-ray photoelectron spectroscopy. Average levels of ca. 8 atom % carbon and ca. 0.4 atom % fluorine were found, repre- senting relatively low levels of surface contamination for pellets pressed in air. The pellets were mounted in an oxidised nickel metal holder in a conventional vacuum cell similar to that of Cant and Little13 in which variation of temperature and pressure could be controlled. A Perkin Elmer 377 spectrometer was used normally in slow-scan mode (52 min) with resolution of 3 cm-l at 3000 cm-' and 1.2 cm-1 at 1250 cm-l. Diflraction and Electron Microscopy Studies The products from t.g.a./d.t.a.and d.s.c. runs, as well as oxides prepared at 550, 700 and 1100 "C were examined by conventional X-ray diffraction and by high resolution electron microscopy in a JEOL lOOC microscope operating at 100 kV. Diffraction patterns from the powder and from individual crystals were thus obtained. The surface structure and dislocation structure were monitored using methods described previ~usly.~ Results Thermal Analysis Nickel Acetate The t.g.a./d.t.a. runs in air showed an endothermic peak at 120 "C with 32% mass loss and a broad, exothermic peak near 400 "C with 44% mass loss. The d.s.c. runs in air gave a similar pattern to the t.g.a./d.t.a. analysis with the endothermic peak at 120 "C and the larger exothermic peak at 405 "C.However, a d.s.c. run in an inert N, atmosphere gave a markedly different pattern. An endothermic peak at 120 "C was found, but this was followed by a second broad endothermic region with two maxima, at 379 and 387 "C. Comparison of the exothermic d.s.c. (air) peak with the two-maxima endothermic d.s.c. (N,) peak shows that the processes associated with the exothermic (air) peak are ten times more energetic than those associated with the endothermic (N,) peak. X-ray diffraction (X.r.d.) of the product from thermal decomposition in air showed a mixture of Ni metal and NiO. Nickel 0 xala t e T.g.a./d.t.a. analysis in air gave an endothermic peak at 260 "C with 30% mass loss and a large exothermic peak at 380 "C with 41 % mass loss. The d.s.c. analysis in air gave similar peaks.In inert N, atmosphere, the d.s.c. analysis again gave an endothermic peak at 260 "C, but a two-maxima, broad endothermic region with peaks at 377 and 402 "C. The exothermic (air) peak had four times the integrated intensity of the endothermic (N,) region. X.r.d. again gave a product in air consisting of both Ni metal and NiO.750 Surface Reduction of Nickel Oxide Nickel Carbonate The t.g.a./d.t.a. runs in air gave a very broad, endothermic peak with a maximum near 150 "C (80-200 "C; 22% mass loss) and a second endothermic peak at 380 "C with 33% mass loss. D.s.c. analysis gave the same two peaks in air or N, atmospheres. The transition energies for the two peaks, assuming NiCO, 2H20 as the source material (see Discussion later) were AH, = 15.5 kJ mol-l(I 50 "C) and AHt = 28.8 kJ mol-l (380 "C).No evidence for Ni metal was found in X.r.d. or electron diffraction patterns of the NiO product. Nickel Hydroxide The t.g.a./d.t.a. runs in air gave a very broad endothermic peak with a maximum near 120 "C (90-1 60 "C ; 8 % mass loss), a second endothermic peak at 320 "C with 10 mass loss, and a third, small exothermic peak at 380 "C. The d.s.c. analyses in air and N, were closely similar to the t.g.a./d.t.a. analysis showing all three peaks. The transition energies for the three peaks, assuming Ni(OH), - NiO - H,O (see Discussion later) as the source material, were 1.2, 2.8 and -0.18 kJ mol-l, respectively. The product in air gave X.r.d. and electron diffraction patterns for NiO with no trace of nickel metal.Surface Area Measurements Surface area measurements were made on specimens of NiO, both ex-carbonate and ex-hydroxide, preannealed in air for 4 h at temperatures in the range 150-1 100 "C. The results are summarised in table 1 . Infrared Studies The degassing sequence of oxides, at temperatures up to 400 "C, prepared at 550 "C from the carbonate and hydroxide samples have been followed in the i.r. spectra. The simpler NiO ex-hydroxide sequence is described first. Fig. 1 shows representative spectra from NiO ex-hydroxide after degassing for 1 h at ( a ) 22, (b) 100 and (c) 300 "C. The spectra were also monitored at 75,200 and 400 "C. In fig. 1 (a), a large broad band is evident with maximum near 3550 cm-1 and a small, sharp peak superimposed at 3660 cm-l.There is also a broad band at 1640cm-l. Exposure of the disc to 25 Torr'f of D,O at 22 "C for 30 min, followed by evacuation, shifted ca. half of the intensity of the 3550 cm-l absorption to 2600 cm-l, leaving a broad featureless band in the original position, and shifted almost all of the intensity of the 1640 cm-l band to 1180 cm-l. The sharp feature at 3660 cm-l is wholly replaced by a new sharp band at 2710 cm-l. After degassing at 75 "C, the sharp 3660 and the broader 1640 cm-l bands are lost, but the spectrum is otherwise unchanged. Through 100, 200, 300 and 400 "C degassing temperatures, the broad 3550 cm-l absorption decreases steadily in intensity and broad plateaux in the 3000-2700 cm-1 and 1800-1200 cm-1 region are revealed in the background spectrum.These plateaux are characteristic of i.r. spectra from NiO annealed in air above 700 "C. Spectra from pelleted NiO ex-hydroxide prepared in air at 700 "C show the broad 3550 cm-l band and weak absorptions due to adsorbed carbonate at 1580, 1360, 1040 and 860 cm-l. At annealing temperatures >900 "C, no carbonate bands are observed. Fig. 2 combines i.r. spectra from NiO ex-carbonate after degassing for 1 h at (a) 22, (b) 150 and ( c ) 300 "C. Spectra were also recorded after degassing at 50, 100, 200, 350 and 400 "C. At 22 "C there is a broad band in the region 362&3500, a very weak band at 2193 and maxima at 1520, 1380, 1280, 1040,965 and 860 cm-l. D 2 0 exchange t 1 Torr = 133.3 Pa.W. R. Pease et al. 75 1 Table 1. Variation of surface area (m2 g-l) with annealing temperature for samples ex-carbonate and ex-hydroxide T/"C source material 150 400 550 700 1100 ex-carbonate 114 143 50 10.5 4.8 0.46 ex-hydroxide 72 99 41 10.2 2.6 0.54 w aveleng t h/pm 2.5 3.0 4.0 5.0 6.0 7.0 8.0 9.010 12 14 16 20 30 40 -'80 - 6 0 I I I I I I I I I I 1 1 4000 3000 2000 1600 1200 806 400 200 Fig.1. 1.r. spectra from NiO ex-hydroxide, decomposed at 550 "C for 4 h in air, after degassing for 1 h at: (-) 22, (----) 100 and (.--.-) 300 "C. w avenum ber/cm -' wavelength/pm 2.5 3.0 4.0 5.0 6.0 7.0 8.0 9.010 12 14 16 20 30 40 I I I I I I I I 100 -80 -60 -40 -20 2 I I I I I I I I I I 1 I 4000 3000 2000 1600 1200 800 400 200 wavenumber/ em-' Fig. 2. 1.r. spectra from NiO ex-carbonate, decomposed at 550 "C for 4 h in air, after degassing for 1 h at: (-) 22, (----) 150 and (-.-.) 300 "C.using the same procedure as for the NiO ex-hydroxide, produced much less marked effects on the spectra. Ca. 10% of the broad 3550 band was replaced by a new broad absorption at 2600 cm-l. At 50 and 100 "C the peaks at 1280 and 965 cm-l decreased steadily in intensity whilst that at 2193 cm-l increased. At 150 "C the bands at 1280 and 965 em-l are missing, the 2193 cm-l band has markedly increased in intensity and the bands at 1520, 1380, 1040 and 860 cm-l are noticeably weaker. A new shoulder is apparent near 1590 cm-l. The broad 3620-3500 em-l band appears largely unchanged. Between 200 and 300 "C, the 2193 cm-l band continues to grow, but there is a change in the752 Surface Reduction of Nickel Oxide lower-frequency bands, which are now observed at 1580 and 1360 cm-l with an intermediate maximum near 1450 cm-l.The 1040 and 860 cm-l bands are weaker. At 350 and 400 "C the 2193 cm-l band has significantly decreased in intensity, the 1580, 1450 and 1360 cm-' peaks have sharpened and the 3550, 1040 and 860 cm-l peaks have further decreased in intensity. 1.r. spectra of pelleted NiO ex-carbonate prepared in air at 700 "C still show weak carbonate bands at 1580, 1360, 1040 and 860 crn-l, but there is no evidence of the 2193 cm-l band. At annealing temperatures >900 "C, no trace of these bands is found. Discussion We will first consider the thermal decomposition processes arising from the thermal analysis and then correlate these processes with the i.r. studies.Thermal Decomposition Processes The decomposition of simple acetate salts of calcium13 and magnesium14 have been studied previously and provide a useful basis for the interpretation of the nickel acetate decomposition. Three processes are found for calcium acetate monohydrate decomposition : l3 (1) (2) (3) with the final decomposition of CaCO, occurring at 850 "C, whereas for magnesium acetate monohydrate,14 the carbonate is not separately stable and the last two processes are combined, i.e. (4) Ca(CH,COO), .H,O -+ Ca(CH,COO), + H,O Ca(CH,COO), -+ CaCO, + (CH,),CO CaCO, -+ CaO + CO, Mg(CH,COO), + MgO + (CH,),CO + CO,. MgCO, is stable to 350 "C, whereas our results show that nickel carbonate is stable to 380 "C. Similarly, the thermal decomposition of nickel oxalate can be interpreted through previous studies of calciumP5-l8 and magnesiuml99 2o oxalate monohydrates.Thus, for the calcium salt CaC,O, * H 2 0 -+ CaC,O, + H,O ( 5 ) (6) (7) where the process (6) was found to be endothermic in inert atmosphere and extremely exothermic in In similar fashion, for the magnesium salt, the last two processes are indistinguishable : (8) CaC,O, -+ CaCO, + CO CaCO, -+ CaO + CO, MgC,O, + MgO + CO + CO,. On the basis of this work, we are able to assign the processes observed in our d.s.c. (N,) analysis as shown in table 2. The water molecules are clearly more strongly bound in the oxalate than in the acetate structure, since they are not released until 260 "C. In both cases, however, this step leads to a significant (> 25% ) increase in surface area (table 1).Wiedemann and Nehring20 have analysed the thermal decomposition of NiC,0,.2H20 and reported the same series of decompositions described here. On the basis of mass changes, they ascribe an intermediate compound to a basic carbonate NiCO, * 20Ni0, but this is not specifically suggested in our work. Mass losses in the loss of water of hydration step are higher than those predicted for the monohydrate,W. R. Pease et al. 753 Table 2. Decomposition processes in nickel acetate and oxalate in inert N, atmosphere mass lossc processa T/"Cb (% (9) Ni(CH,COO), - H,O + Ni(CH,COO), + H,O (1 1) NiCO, + NiO + CO, (1 3) NiC,O, + NiCO, + CO (1 4) NiCO, + NiO + CO, 120 10 (32) 387 260 12 (39) 377 (10) Ni(CH,COO), + NiCO, + (CH,),CO 379J 39 (44) (1 2) NiC,O, - H,O --+ NiC,O, + H,O 402) 34(41) a All endothermic.indicates the maximum. The peaks are relatively broad in most cases, i.e. & 50 "C; the temperature Figures in brackets represent measured mass losses. suggesting that our compounds had between three and four water molecules per nickel atom. The apparent decomposition temperature of the NiCO, in each case is above its true temperature of 380 "C. This is caused by the kinetics of its formation from the initial salt coupled with the 10 "C min-l heating rate used here. Different rates of heating alter the observed decomposition temperature of the carbonate. In air, the process of decomposition of either oxalate or acetate obviously results in the formation of carbon monoxide [cf. eqn (6)] and the exposure of the defective NiO to this product produces a significant proportion of nickel metal in the product.Several competing reactions are likely to be occurring in a complex, predominantly exothermic sequence, e.g. on acetate surfaces : (15) (16) (CH,),CO + $0, -+ CO + CH,CH,OH CH,CH,OH + 30, -+ 2C0, + 3H,O On both surfaces: CO + 0, -+ CO, + Oads (17) from (13) and (15) The presence of adsorbed CO and 0- ions is strongly supported by the i.r. studies discussed below. Reduction of NiO prepared at low temperature (below 500 "C), to nickel metal atoms via CO adsorption or reaction with surface carbon impurities is well known.'? lo, 21 The presence of large proportions of excess oxygen' as 0- ions in NiO surfaces is also now well established, both theoreticallys and e~perimentally.~-~~ Thus this reaction sequence is in accord with known phenomena on NiO surfaces.The observed mass losses for processes (10) + (1 1) and (1 3 ) + (14) are larger than predicted, as would be expected if reduction to the metal [process (19)] is accompanying these reactions and reoxidation [process (20)] is less significant, as is apparently the case. The oxide product from nickel oxalate or acetate in air is clearly not suitable for study as NiO unless extensively annealed at temperatures > 1200 "C. It is then difficult to relate defect concentrations to the starting material and decomposition process and it is therefore less useful for reactivity comparisons. Thermal decomposition of nickel carbonate in air or nitrogen occurs in two endothermic steps at 150 and 380 "C: NiCO, -2H,O -+ NiCO, + 2H,O AHt = 15.5 kJ mol-l (21) NiCO, 7 NiO+CO, AHt = 28.8 kJ mol-l.(22)754 Surface Reduction of Nickel Oxide Mass spectra of the gas phase during decomposition at 350-380 "C showed O,, CO and CO, in the ratios of roughly 1 : 2: 4. During decomposition of the carbonate group, carbon and oxygen atoms diffuse to the surface. These reactive adatoms can combine at the surface to give 0,, CO or CO, molecules. The mass spectra are consistent with this view and show that the two reactions Cads+20ads = CO, and COads+Oads = CO, occur more frequently than the Cads + Oads = CO and Oads + Oads = 0, reactions. Measured mass losses (22 and 33%) correspond well with those expected from these processes (22 and 30%, respectively).A previous study2, of a basic nickel carbonate, NiCO, - NiO, found elimination of water below 300 "C, transformation to the oxide between 300 and 350 "C and indications that the loss of CO, may continue up to 850 "C. Our thermal analysis accords with this work and our i.r. studies strongly support the last observation. The hydroxide or 'hydrated oxide' gave endothermic peaks at 120 and 320 "C which are likely to have arisen from similar decomposition steps of a Ni(OH);NiO.H,O material : Ni(OH), NiO * H,O -+ Ni(OH), * NiO + H,O AHt = 1.2 kJ mol-1 (23) (24) Measured mass losses (8 and 10%) correspond closely to those expected (9 and lo%, respectively) and this sequence accords with the preparation method.12 Cabannes-OtP reported that nickel hydroxide loses water of hydration below 115 "C and is stable to 260 "C, but that pure NiO may not be found until 600 "C. These results are in accord with our studies.The exotherm at 380 "C is likely to relate to a lattice rearrangement resulting from the collapse of a pseudomorphic structure formed during decomposition at low temperature. This structure, fully characterised by Larkins et al.,24 consists of very thin flakes that have the overall morphology of the parent hydroxide (brucite) platelets with an orientation of [ 1 111 planes parallel to the surface of the flake. The NiO flakes contain many irregularly shaped channels separating domains of the oxide, but there does not appear to be any misorientations of the oxide structure across the channels. The mechanism of decomposition of the hydroxide is p r o p o ~ e d ~ ~ - ~ ~ to be topotactic, i.e.lateral removal of H,O molecules without major disruption of the layer structure. The [00 13 plane of the lamellar (brucite) hydroxide structure becomes the [ 1 1 11 plane of the (rock salt) NiO structure. Reorientation of the flakes occurs to give crystallites of NiO near 400 0C,24 This type of relatively gentle topotactic decomposition and subsequent rearrangement is not observed in the carbonate decomposition. Niepce and c o ~ o r k e r s ~ ~ - ~ ~ have compared the martensitic decomposition, i.e. transformation without diffusion, of cadmium hydroxide and carbonate, which have the same structures as nickel hydroxide and carbonate, respectively, and have defined the structural mechanisms involved.In the carbonate, the [OOl] plane does not become the [l 1 11 plane of the oxide, but rather the [I401 axis of three disjointed sets of oxide crystallites at 120" orientation to each other. Consequently, rearrangement of this [I401 triad axis of the oxide pseudomorph to the final NiO crystallites is much more severe and results in much more disorder in the final oxide. This is evidenced by comparison of the 20 half-peak breadths of different crystallographic planes from the X.r.d. patterns of oxide produced at the same temperature. Floquet and Niep~e,~ defined k, = 28[ 1 1 1]/28[200] and k, = 28[ 1 1 1]/20[220] and found values for CdO ex-Cd(OH), of k, = 0.89 and k, = 0.72 compared with values for CdO ex-CdCO, of k, = 0.99 and k, = 0.90. This marked difference in the oxides formed is reflected in our studies with NiO in three ways: (i) the transition energies of AH, (28.8 kJ mol-l) for the carbonate us.AH, (2.8 kJ mol-l) for the hydroxide; (ii) the considerably larger loss of surface area for the carbopate during the NiO crystallite Ni(OH), .NiO -+ 2Ni0 + H,O AHt = 2.8 kJ mol-l.W. R. Pease et al. 755 growth (from 150 to 400 "C); and (iii) the considerably higher reactivity of the ex-carbonate Ni0.30 We should note here that the surface area changes primarily reflect changes in particle size distribution and not changes in surface ~rdering.~ Therefore, these results are consistent with the k, and k, comparisons of Floquet and Niepce in that the area (and particle size) of the NiO ex-carbonate is ca.25% larger than that of the NiO ex-hydroxide after preparation at 400 "C. Infrared Studies From the i.r. spectra of the NiO ex-hydroxide during degassing (fig. 1) the following conclusions are evident. Roughly half of the OH groups are accessible to exchange with D,O in 30 min at room temperature. The retention of the pseudomorphology of the hydroxide to 380 "C suggests that, after decomposition at 550 "C, readsorption and reabsorption of H,O and CO, from air is enhanced by residual structure in the form of defects with the correct orientation and form to accommodate these molecules directly or by reaction. This result is consistent with the original structure of the Ni(OH), - NiO * H,O compound suggested by thermal analysis, since any residual or reabsorbed OH groups in the lamellar Ni(OH), structure, as well as water of hydration, would exchange much more slowly than surface OH groups in a diffusion-controlled process.The loss of the 3660 and 1640 cm-l bands at 75 "C represents loss of water molecules, some of which were likely to be surface-adsorbed as indicated by the sharp 'free' hydroxyl stretching vibration at 3660 and bending vibration at 1640 cm-l exchangeable by D,O to 2710 and 1180 cm-l, respectively ( i e . a frequency ratio of 1.35, close to that normally observed for such proce~ses).~~ The remaining exchangeable OH groups, i.e. those represented in the partial shift of the broad 3550 absorption to 2600 cm-l, are hydrogen-bonded and probably represent readsorbed water, water of hydration and some surface OH groups of any 'reformed' hydroxide structure.The elimination of H,O is then reflected in the systematic reduction in intensity of this broad 3550 cm-l band relative to background and might be expected to follow the two steps (23) and (24), but there is no obvious division between the two processes in spectra at 100 and 400 "C. This is probably due to the difficulty of measuring absolute v(0H) intensities in these spectra when the background spectra and beam attenuation are changing. However, at 400 "C v(0H) is still evident in the spectrum and the decomposition process is not complete as suggested by Cabanne~-Ott.,~ Our spectra and assignments are closely similar to those reported by Guglielminotti et al.,,l who have studied the thermal decomposition of nickel hydroxide under vacuum at 200 "C.1.r. spectra from the degassing of NiO ex-nickel carbonate show the following features. There is no evidence of the sharp band due to 'free' hydroxyls near 3660 cm-l or of the H,O bending vibration near 1640 cm-l. This result, with the relatively small amount of D,O exchange, indicates that OH and H,O species are predominantly formed by reabsorption into the bulk of the crystallites and are hydrogen-bonded. The broad 3550 cm-l band decreases in intensity systematically from 50 to 400 "C, again with no obvious discontinuities or sub-structure that might be associated with process (21) in any reformed carbonate structure. The intensity decrease appears, however, to be less marked than that observed for the NiO ex-hydroxide and this supports the view that this band arises predominantly from entrapped hydrogen-bonded hydroxyl groups in the bulk of the material.Fig. 3 illustrates the variation of the 2193 cm-l peak relative to background as the degassing temperature is increased from 22-350 "C. It is evident that it has its largest intensity after degassing at 300 "C. The frequency of the band lies above that for gaseous CO (2143 cm-l) and below that for gaseous CO, (2350 cm-I). It is correlated with the appearance of both CO and CO, in the gas phase above the carbonate sample. The presence of an absorption in this frequency region for CO adsorbed on several different756 Surface Reduction of Nickel Oxide 801 I- T I I I I 1 2.500 2400 2300220021002000 wavenumberlcm-' Fig. 3. Variation of the intensity of the 2193 cm-l absorption from NiO ex-carbonate (fig.2) with degassing temperature. Temperatures: (-) 22; (----) 100; (---) 200; ( e . . . ) 300; and ( a * * ) 350 "C. metal and metal oxide surfaces is well It has been variously assigned to Ni-(0-C-0) specie^:,^^ 34 CO weakly attached to 0,- lattice oxygen3, or absorbed CO+;359 39 and CO adsorbed by dipolar attra~tion.,~ On the basis of our results here and recent work on NiO surfacesSvg-l1 it is now possible to assign this band with some confidence to the v, asymmetric stretching vibration of a Ni-(0-C-0)- species. The surface of defective NiO has been showngT9-11 to consist of both 0- and 0,- species in varying ratios depending on pretreatment of the oxide. The 0- species is associated with a defective structure for oxygeng7 l1 and is, therefore, likely to dominate the adsorbent surface. The interaction of CO with this species will directly produce an adsorbed CO; species.This species has been previously proposed by Horgan and King40 as the intermediate in the catalytic oxidation of CO to CO, over thin nickel films supported on silica and is consistent with the presence of 0- species in defective nickel films in contact with oxygen-containing species. It is also consistent with the p-type semiconductivity of NiO. If CO adsorbs as a CO+ species, the depletion layer limits the surface coverage to less than lo-, of a monolayer and this limited coverage will not be observed in i.r. spectra. Similar arguments apply to adsorption of CO by dipolar attraction.Formation of COT species is only limited by the surface availability of 0- ions since it has also been shown that less defective NiO surfaces only physically adsorb CO, i.e. in interaction with the 02- lattice ions.41 The lowering of the CO, frequency in the COT species has been observed previously by Cotton and K r a i h a n ~ e l ~ ~ ? ~ ~ and is thought to be due to occupancy of the antibonding TC orbital in the CO, structure. There is no evidence of the v, bending vibration probably near 650 cm-l or the v, symmetric stretching vibration expected below 1388 cm-l. The lack of evidence for the former vibration is due to NiO absorption in this region and, for the latter, probably indicatesW. R. Pease et al. 757 a relatively weak effect on the symmetry of the strongly bonded CO; species by adjacent nickel sites and, hence, a low absorption intensity.This species could presumably desorb as either CO or CO, species in accord with gas analyses. As the CO, desorbing species, the effect on the surface is reductive leading to further hole depletionll and formation of Nio surface atoms.l0? l1 Activated reduction of Nil1 to NiO by CO adsorption at 300 "C has previously been directly demonstrated for NiO prepared by annealing at 700 "C for 4 h.lo7 l1 The observation of optimal intensity of the 2193 cm-1 band after degassing at 300 "C is consistent with this activated reduction process in competition with desorption of the CO, thus formed. The complex series of absorption bands in the 1600-800 cm-l region of fig. 2 is due primarily to carbonate groups.31* 34 There is a considerable amount of literature on the assignment of these absorptions to simple carbonate, unidentate carbonate, bidentate carbonate and binuclear carbonate species.We will not attempt a detailed analysis of the spectral changes, but will note the following important features. The initial spectra (fig. 2) do not show a symmetrical CO, with D,, symmetry, but indicate a species with unidentate character and absorptions at 1520 cm-l (asymm. stretch), 1380 (symm. stretch) 1040 (mainly C=O stretch) and 860 cm-l (out-of-plane bend). The 1280 and 965 cm-' bands appear to be associated with extensive interaction of the carbonate species with H,O [supported by the v(0H) spectra] since they are lost in degassing below 150 "C, and may represent bicarbonate or similar species.Much of the unidentate carbonate may be within the bulk of the crystallites at degassing temperatures below 200 "C, as suggested by the d.t.a./d.s.c. results, if the same process of readsorption/re- absorption into pseudo-structural defect sites is followed. There is then a pronounced shift of the spectral features between 200 and 400 "C to a more bidentate species with absorptions at 1580, 1360, 1040 and 860 cm-l. This may be due to a shift to predominance of surface species over bulk. The frequencies are intermediate between those expected for unidentate and bidentate species in other and it is not possible to be more explicit in assignment. The 1450 cm-1 intermediate maximum in fig. 2(c) is intriguing and may represent some 'free' or simple CO, in which the symmetry is very little distorted by interaction with neighbouring sites.The presence of carbonate groups remaining on NiO ex-carbonate annealed at temperatures over 700 "C in air has also been demonstrated. W. R. P. gratefully acknowledges the support of the Director of the Government Chemical Laboratories and valuable technical assistance from Colin F. Jones (Griffith University). The work was financially supported by the Australian Research Grants Scheme. References 1 2 3 4 5 6 7 8 9 10 11 P. C. Gravelle and S. J. Teichner, Adu. Catal., 1969, 20, 167. E. R. S. Winter, Ado. Catal., 1968, 10, 165. C. F. Jones, R. L. Segall, R. St. C . Smart and P. S. Turner, J . Chem. Soc., Faraduy Trans. I , 1977, 73, 1710.C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, J . Chem. Soc., Faraduy Trans. I , 1978, 74, 1615. C. F. Jones, R. A. Reeve, R. Rigg, R. L. Segall, R. St. C. Smart and P. S. Turner, J . Chem. Soc., Faraday Trans. 1, 1984, 80, 2609. C. F. Jones, R. L. Segall, R. St. C. Smart and P. S. Turner, J . Muter. Sci. Lett., 1984, 3, 810. Z. M. Jarzebski, Oxide Semiconductors (Pergamon Press, Oxford, 1973). E. A. Colbourn and W. C. Mackrodt, Surf. Sci., 1982,117, 571 ; E. A. Colbourn and W. C. Mackrodt, SolidStute Ionics, 1983,8,221; E. A. Colbourn, J. Kendrick and W. C. Mackrodt, Surf. Sci., 1983,126, 550. M. W. Roberts and R. St. C. Smart, J . Chem. Soc., Faraday Trans. 1, 1984, 80, 2957. S. J. Cochran and F. P. Larkins, personal communication. M. W. Roberts and R.St. C. Smart, Surf. Sci., 1980, 100, 590.758 12 R. K. Barton, personal communication. 13 L. Walter-Levy and J. Laniepce, C.R. Acad. Sci., 1960, 250, 332. 14 F. Burriel-Marti, S. Jiminez-Gomes and C. Alvarez-Herrero, An. Edafol. Fisiol. Veg., 1955, 14, 221. 15 C. D. Villamil, An. Soc. Espan. Fis. Quim., 1924, 22, 74. 16 E. Moles and C. D. Villamil, An. SOC. Espan. Fis. Quim., 1924, 22, 174. 17 M. Aumeras, C.R. Acad. Sci., 1925, 181, 214. 18 P. Waters, Nature (London), 1956, 178, 324. 19 K. Kawagaki, J . Chem. Soc. Jpn, 1951, 72, 1079. 20 H. G. Wiedeman and D. Nehnng, Z . Anorg. Chem., 1960,304, 137. 21 E. Guglielminotti, L. Cerruti and E. Borello, Gazz. Chem. Ital., 1977, 107, 44. 22 J. Francois-Rossetti, M. T. Charton and B. Imelik, Bull. SOC. Chem. Fr., 1957, 614. 23 C. Cabannes-Ott, C.R. Acad. Sci., 1955, 240, 68. 24 F. P. Larkins, P. J. Fensham and J. V. Sanders, Trans. Faraday Soc., 1970, 66, 174. 25 P. J. Anderson and R. F. Horlock, Trans. Faraday Soc., 1962, 58, 1993. 26 C. Cronan, F. J. Micale, H. Leidheiser, M. Topic and A. C. Zettlemoyer, Croat. Chem. Acta, 1976, 48, 27 J. C. Niepce and G. Watelle, J. Muter. Sci., 1978, 13, 149. 28 J. C. Niepce, M. T. Mesnier and D. Louer, J . Solid State Chem., 1977, 22, 341. 29 N. Floquet and J. C. Niepce, J . Muter. Sci., 1978, 13, 766. 30 W. R. Pease, R. L. Segall, R. St. C. Smart and P. S. Turner, J. Chem. SOC., Faraday Trans. I , 3 1 L. H. Little, Infrared Spectra of Adrorbed Species (Academic Press, New Y ork, 1966). 32 R. P. Eischens and W. A. Pliskin, Adv. Catal., 1957, 9, 662. 33 C. E. O’Neill and D. J. C . Yates, Spectrochim. Acta, 1961, 17, 953. 34 R. P. Eischens and W. A. Pliskin, Adv. Catal., 1958, 10, 1. 35 R. A. Gardner and R. H. Petrucci, J. Am. Chem. SOC., 1960,82, 5051. 36 H. L. Pickering and H. C. Eckstrom, J. Phys. Chem., 1959, 63, 512. 37 R. P. Eischens, S. A. Francis and W. A. Pliskin, J. Phys. Chem., 1956, 60, 194. 38 J. H. Taylor and C. H. Amberg, Can. J . Chem., 1961,39, 535. 39 L. H. Little and C . H. Amberg, Can. J . Chem., 1962, 40, 1997. 40 A. M. Horgan and D. A. King, Trans. Faraday Soc., 1971, 67, 2145. 41 M. W. Roberts and B. R. Wells, Faraday Discuss. Chem. Soc., 1966, 41. 162. 42 F. A. Cotton and C. S. Kraihanzel, J . Am. Chem. Soc., 1962,84,4432. 43 C . S . Kraihanzel and F. A. Cotton, Inorg. Chem., 1963, 2, 553. Surface Reduction of Nickel Oxide 521. 1986, 82, 759. Paper 5/590; Received 9th April, 1985