Magnetic and Optical Studies of Chromium Oxides Part 3.-Calcination of Coprecipitated Chromium and Aluminium Hydroxide Gels BY ALAN ELLIsoN*t AND KENNETH s. w. SING$. School of Science, Hull College of Higher Education, Kingston upon Hull HU6 7RT Received 16th December, 1977 Magnetic and optical studies were performed on coprecipitated chromium and aluminium hydroxide gels after calcination in air at temperatures between 50 and 1150°C. The results show that magnetic diIution was not achieved and that surface clusters of chromium oxide were formed. The magnetism, e.s.r. and optical spectra of oxidised samples are discussed in terms of a mixed- valency phase of partially-oxidised chromium containing Cr3+ and Cr6+ species. This new model ofy-phase chromium is shown to be consistent and an acceptable part of the extensive mixed-valency chemistry of chromium.This work continues a systematic study of the magnetic and spectral properties of chromium-alumina catalysts. It was shown in Parts 1 and 2 that the state of dispersion of chromium on the alumina support influences the stability of Cr6+, the occurrence and stability of y-phase chromium and the production of an " oxidised- phase " of chromium. There is some evidence that a well-dispersed state of chromium can be obtained using coprecipitation techniques and it is to be expected that on calcination of coprecipitated gels the chromium ion has the most favourable chance of occupying cation sites of well-defined symmetry in the host lattice. This paper is concerned with a study of the thermal decomposition of chromium and ahminium hydroxide gels coprecipitated from aqueous solution.EXPERIMENTAL The coprecipitated gels were prepared by allowing aqueous solutions of chromium (III) nitrate, aluminium nitrate and ammonium hydroxide to react under controlled conditions of pH, concentration and flow-rate in an apparatus similar to that of Harris and Sing,3 but modified to include a constant-head device. Three precipitates were prepared, containing 1.3, 6.1 and 12.3 wt. % Cr, respectively. The methods of calcination, analysis and measurement of magnetic susceptibility, e.s.r. and diffuse reflectance spectra are described in Part 1.' Samples are designated according to their chromium content (wt. %) and calcination temperature; thus, sample CAC 1.3 (50) contains 1.3 wt.% Cr and was calcined at 50°C for 5 h. RESULTS The mass susceptibility of Cry xz, measured at room temperature for samples CAC 1.3, CAC 6.1 and CAC 12.3 initially dried at 50°C, is plotted against the t Present address : School of Science, Faculty of Combined Studies, Hull College of Higher Education, Kingston upon Hull. $ Present address : School of Chemistry, Brunel University, Uxbridge, Middlesex. 1-89 28072808 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS I l l l l l ~ ~ l l ' l 200 600 600 800 1000 1200 T/"C FIG. 1.---YCm' at room temperature plotted against calcination temperature T for the samples x CAC 1.3 (50) ; 0, CAC 6.1 (50) ; A, CAC 12.3 (50). temperature of calcination in fig. 1. Between 50 and 200°C the observed decrease in xz is accompanied by a change in colour from blue or green to yellow or brown depending upon chromium content.The mass susceptibilities of samples CAC (200), CAC (400) and CAC (600) are substantially magnetic-field dependent, see table 1. A field-independent set of measurements is included for comparison and it should be noted that the original gels dried at 50°C had field-independent susceptibilities. 3500 3000 2500A . ELLISON AND K. S. W. SING 2809 to O L 500 3500 3000 2500 (b) H/kG gain 1000 n - - 1 1000 500 1 I 3 500 3000 (4 H/kG FIG. 2.-Qualitative e.s.r. spectra at 77 K for the samples (low field, gain 1000, high field, gain 40). (a) CAC 1.3 (200); (b) CAC 1.3 (400); (c) CAC 1.3 (600). Qualitative e.s.r. spectra, obtained at 77 K are shown in fig.2 for samples CAC 1.3 (200), (400) and (600). Resonances of markedly different intensities are shown, recorded at different signal gains. The predominant resonance in each case is a strong narrow and symmetric signal at gain 40, peak to peak width (p.p.w.) 50 G, centred at ~3400 G. This is identified as the y-resonance of supported chromium." At a gain of 1000, &resonance is detected for the 200 and 400°C samples, but not for the 600°C sample. The fine structure is well resolved showing distinct lines at 11002810 MAGNETIC PROPERTIESOF COPRECIPITATED CHROMIUM HYDROXIDE GELS and 1700 G. The e.s.r. spectrum at 77 K for sample CAC 6.1 (400), consists of an intense y-resonance; P-resonance is also detected but is either of low intensity or, at this temperature, has been dipole-dipole broadened almost beyond detection.Fig. 3 shows molecular field plots, (xA)-'/m3 mol-1 against T/K, for samples CAC 1.3 (200), (400) and (600). These plots are curves convex to the temperature axis, giving extrapolated intercepts on this axis of between +90 and + 100 K. TABLE 1 .-MAGNETIC FIELD DEPENDENCE OF MAGNETIC SUSCEPTIBILITY 10: XF/rn3 kg-1 sample ( n ) (b) C) CAC 1.3 (50) 13.75 13.78 13.70 CAC 1.3 (200) 5.14 4.98 4.80 CAC 1.3 (400) 5.52 5.03 4.74 CAC 1.3 (600) 5.98 5.71 5.54 Field strength (a) 2.1 (b) 4.0 and (c) 6.5 kG. Quantitative e.s.r. spectra, for those samples showing only y-resonance at 3400 G, were determined over the temperature range 77 to 573 K. The resonance intensity I, calculated as the first moment of area of the y-resonance for samples CAC 1.3 (200), (400), (600) and CAC 6.1 (400) is shown in table 2 as a function of temperature.Examples of the spectra obtained are shown in fig. 4. The widths of the y-resonances 600 5 00 2 Y A00 53 L 300 d \ 4 8 ?3 200 100 1 1 I 0 100 200 300 TlK FIG. 3 . - 1 / ~ ~ plotted against temperature of measurement, T, (XA is the atomic susceptibility of Cr) for the samples (a) CAC 1.3 (200), x ; (b) CAC 1.3 (400), 0 ; (c) CAC 1.3 (W), A. remain constant within 1 G, over the whole temperature range and are independent of Cr content. Molecular field plots derived from these data are plotted in fig. 5 as I-1 against T. The plots show pronounced curvature, convex to the temperature axis, giving extrapolated intercepts close to those obtained from the xE-' against T plots for the same samples, see fig.3.281 1 A . ELLISON A N D K . S . W . SING TABLE 2.-E.S.R. SPECTRAL PARAMETERS FOR y-RESONANCE (T is the measurement temperature ; I is the relative number of spins ; p.p.w. is the peak to peak width) sample CAC 1 . 3 (200) CAC 1 3 (400) CAC 1.3 (600) CAC 6.1 (400) key T/K gain p.p.w./G 10-171 gain p.p.w:/G 10-17 Z gain p.p.w./G lO-17Z gain p.p.W./G 104’1 fig. 4 77 125 313 160 133 160 153 160 173 193 160 223 233 160 273 298 160 323 348 373 398 423 448 473 498 523 548 573 60 29.5 100 60 42.1 250 75 29.1 100 70 83.1 60 16.3 100 60 23.2 320 75 16.7 125 70 44.1 60 14.1 60 12.5 320 75 8.7 125 70 34.9 60 10.2 320 75 7.1 125 70 28.7 60 8.8 320 75 5.5 125 70 25.2 320 75 5.2 125 70 23.0 60 7.2 100 60 10.0 320 75 6.6 100 60 16.5 100 60 12.9 500 75 500 75 500 75 500 75 500 75 630 75 630 75 630 75 630 75 630 75 630 75 5.7 5.3 5.1 3.9 2.9 2.5 2.5 2.9 3.7 4.8 5.5 F~o.4.-Quantitative e.s.r. spectra of y-resonance for sample CAC 1.3 (600). Key for spectra (a) to ( I ) is given in table 2.2812 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS 30 25 2 0 rl I 0 + 0 P( CI 1 E 1 5 - 5 1 0 L 5 - 35 - - - I 30 25 10 5 1 I I J 0 100 200 300 T/K FIG. 5.-Quantitative e.s.r. spectra ; I-' plotted against temperaturelof measurement, T. ( I is the relative e.s.r. intensity ofy-phase Cr). x , CAC 1.3 (200) ; 0, CAC 1.3 (400) ; A, CAC 1.3 (600)A. ELLISON AND K. S . W. SING 281 3 Values of 1 for the y-resonance of CAC 1.3 (600) are plotted against T in fig.6. The intensity at low temperatures is seen to increase with decreasing temperature at a much greater rate than expected for a paramagnet obeying the Curie or Curie- Weiss laws. The data was collected from two separate series of measurements, from 77 to 298 K and from 298 to 573 K causing the obvious discontinuity at 298 K. However, the increase in intensity with increasing temperature after x 470 K is anomalous and indicative of a transition temperature. DISCUSSION Between the calcination temperatures 50 and 200°C there is a decrease in ~2, accompanied by a change in sample colour, due to oxidation of Cr3+ ions. xE decreases to give a minimum centred at ~ 4 0 0 ° C for each sample, independent of Cr content, behaviour similar to that observed for CrC1,-Al,O, samples.2 The breadth of the minima and the temperature at which xg begins to increase, uiz.at 600°C for CAC 1.3 and at 400°C for CAC 6.1 and CAC 12.3, depends upon Cr content. This refiats the manner in which the stability of oxidised chromium depends upon the degree of dispersion of chromium on the support.l* Chromium at low concentration is more stable to oxidation than chromium at higher concen- tration. Indeed fig. 1 shows that even at 400°C some proportion of Cr3+ remains unoxidised. Samples CAC 6.1 and CAC 12.3 have identical Xz-minimum values; in the case of CAC 1.3 the position and value of the Xz-minimum is uncertain due to the marked field-dependent susceptibilities observed. Nevertheless the xE values all lie within a very narrow range. After calcination at 1120°C all of the chromium has been reduced to the Cr3+ state. The xE values are larger than the corresponding values in related systems suggesting that a more dispersed form of chromium has been achieved.This idea is confirmed on observing that the low-field derivative spectra in fig. 2 for CAC 1.3 are much more well defined and narrower than similar &resonances reported else- where.2 This width is usually interpreted in terms of a range of low-symmetry, zero-field terms, D and E, in the spin Hamiltonian. However an appropriate operator representing dipole-dipole broadening would provide an acceptable alter- native interpretation which is certainly more realistic for the planar clusters of chromium discussed in Part 2.2 On this model the 8-spectra of fig.2 suggest weaker dipole-dipole coupling due to greater dispersion of chromium in the support. The nature of the chromium responsible for the XC,'-minirna is worthy of some discussion. It has been shown that supported Cr3+, in both impregnated and coprecipitated conditions, is incompletely oxidised within a small temperature range, independently of concentration. This is not the behaviour of unsupported Cr3+. Cathers and Wendlandt have shown that all isomers of hydrated CrC1, decompose on heating to give Cr203, with loss of HCl and HzO, without passing through oxida- tion states higher than 3+. Similarly a-Cr,O,, does not undergo oxidation on heating in this temperature region. In addition, over the same temperature range, it is observed that supported Cr6+ is reduced and although the extent of reduction depends upon chromium content, e.s.r.and susceptibility measurements show that considerable reduction of low chromium content samples has occurred at 200"C.1 It is significant that not only do the xg-rninima occur in this region of coincident oxidation and reduction but also that the predominant paramagnetic phase, from e.s.r. evidence, is y-phase chromium independent of loading and of type of supported2814 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS system. In addition susceptibilities are magnetic field-dependent and the values of Xz-minima show very little variation with chromium content. The chromium oxidation states whose identities have been firmly established in this region, by optical spectra and e.s.r.techniques, are Cr3+ and Cr6+. y-phase chromium, the predominant paramagnetic phase, has been interpreted as magnetically- isolated Crs+ ions which are not coupled by exchange interactions.6 If one accepts this model, for the sake of argument, the situation arises that in the calcination temperature range 200-600°C and at the Xz-minima, one must consider that there are three states of chromium present, viz., Cr3+, Cr6+ and Cr5+. x:, iteff and 6 depend markedly upon the relative amounts of the different chromium species and upon their individual degrees of dispersion. In fact we have observed that, in this calcination temperature range, xE remains remarkably constant even though the rates and magnitudes of oxidation and reduction and the values of 6 and of peff of each chromium species are very different and, moreover, depend upon chromium content.Further, the proportion of y-phase chromium produced on calcination of Cr3+ or Cr6+ depends markedly upon the surface area of the chromium, the nature of the support, the chromium content and the calcination temperature, resulting in modification of petf, 8 and ~2.'. x-' / .- / I /A . ELLISON AND K. S . W. SING 2815 of the mobile carrier spin during a " hop " from one lattice site to the next. The mobile carrier contributes to the binding energy of the system provided that the spins on neighbouring sites are parallel. The electric and magnetic properties of such compounds depend strongly upon their composition, x. Thus when x = 0 or 1 the oxides are good insulators and antiferromagnetic, while for a range of intermediate compositions the electric conductivity is several orders of magnitude greater and the materials are ferromagnetic. The theoretical temperature dependence of the magnetic susceptibility is shown in fig.7 and exactly describes the curved x-' against T plots reported here. Provided that the upper transition temperature is antiferromagnetic, a discon- tinuity in slope is observed at TN, yielding a negative, high-temperature Weiss constant 8. Low-temperature extrapolation yields a positive-temperature intercept, T,.' * The oxidised phase of chromium, 02--Cr3+-02--Cr6+-02- possesses magnetic properties described by the Zener model; moreover the existence of collective Zener electrons is shown by the colours, spectra and p-type semiconductivity of this phase.At the extremities of the series (Cr:' Cr;Lx)Oi- are Cri+03 (x = 0, n-type semi- conductor, antiferromagnetic) and Cr6+03 (x = 1, insulator, temperature independent paramagnetic). The stable Zener phase, when x > 0, occurs at the surface of chromium clusters where the antisotropic ferromagnetic coupling responsible for stability is maximised. It is important to realise that the existence of mixed-valency compounds is well established and in particular is characteristic of the chemistry of chromium. For example, the reviews of Allen and Hush l4 and of Robin and Day cite totals of 150 and 820 references, respectively. It is not possible to give here a complete review of the incidence of mixed-valency species in chromium chemistry.Neverthe- less this review does reveal that Crrrr-Crvr, Crrrr-Crrr, Crrrr-Crrv and Crv-Crvl mixed-valency compounds occur as stable stoichiometric and non-stoichiometric compounds. Most importantly Wilhelmi and co-authors 16* have shown that compounds of the type MCr308 and M2Cr309 (M = alkali metal ion)16 and intermediate oxides of chromium, including Cr205, Cr308, CrsOIz and Cr,OI5 l7 are black crystalline substances containing CrrIr-CrVr mixed-valency species alone. It is maintained that the magnetic and optical properties of the oxidised-phase of chromium are more completely explained on the basis of mixed-valency species of the type CrIrr-Crvr without recourse to the requirement of CrV ions stabilised at specific sites in the support l a t t i ~ e .~ Indeed the observed data are typical of those expected for a Zener double-exchange system. ~2 for samples at susceptibility minima is often magnetic field dependent. Molecular field plots (xi1 against T) show pronounced curvature towards the tempera- ture axis with intercepts of w -k90 IS. This unusual behaviour cannot be explained on uncoupled systems of Cr6+ or Cr3+ ions. Cr6+ ions have a temperature- independent susceptibiIity which would produce high-temperature deviation from linearity not observed here. Octahedral Cr3+ has a weak field 4A2, ground state with minimal orbital angular momentum contribution to peff or 6 through spin-orbit coupling. The amorphous Cr3+ phases in the temperature range 77 to 298 K are paramagnetic producing P-phase e.s.r., giving negative 6 values but no curvature in these plots.Ferrimagnetic structures could produce such curves but require well- ordered, three-dimensional interpenetrating-ferromagnetic sub-lattices. In, for example, C4" symmetry Cr5+ ions with a split 2T2 ground state can produce low- temperature deviation from Curie-Weiss behaviour but the deviation is away from the temperature axis, not as observed here. The Zener theory, however, does predict curved molecular field plots.2816 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS Visible-u.v. reflectance spectra show Cr3+ and Cr6+ bands with, in addition, enhanced absorbance over the whole wavelength range, 200-1000 nm. Most signifi- cantly, these brown-black samples absorb radiation at wavelengths where the postulated individual species (Cr3+, Cr6+, Cr5+) are transparent.These observations are consistent with the existence of Zener mobile-electrons in a class 111 mixed- valency system. The predominant e.s.r. signal is y-resonance. Both 6- and P-resonances are often not detected even though the presence of Cr3+ species is confirmed from reflectance spectra. This suggests that the normal resonances of Cr3+ electron-spins have been obliterated through strong exchange forces. attaining a maximum intensity for alumina samples near 2 wt. % Cr where the proportion of small and thin, platelike chromium clusters is the greatest. Indeed, y-resonance appears at the expense of /?-resonance whilst &resonance is still often retained. y-resonance achieves greatest intensity for silica supported chromium, the most active catalysts in ethylene polymerisation.It has been asserted that &phase chromium has never been observed in silica systems, although after achieving a greater dispersion of chromium by thermal decomposition of mixed oxalato pre- cipitates, a small but unstable &phase signal has been reported.lg Cr5+ ions, often said to be responsible for y-resonance, need to occupy high symmetry support-lattice sites to acquire the necessary stabilisation energy. However, not only is a dispersed phase of CrS+ (the product of &phase) unlikely but also there is much evidence that chromium is insoluble in silica even at the liquidus temperature.' It would appear that although Cr3+ species occur in the oxidised chromium their normal magnetic properties are obliterated. Instead the oxidised phase itself exhibits different behaviour which is not characteristic of Cr5+ ions isolated in the support lattice.Moreover molecular field plots for y-resonance, I-' against T, in all of the supported systems studied show pronounced curvature with positive temperature intercepts, behaviour consistent with direct exchange interactions between ions of different oxidation state. For alumina-supported chromium the width of the y-resonance, from 77 to 593 K, is completely temperature-independent. The usual argument invoked is that spin- lattice relaxation is slow. However, if this is the case, and if as is observed the resonance does not show signs of saturation then another predominating relaxation mechanism must be efficiently operative.Thus Adrian 2 o maintains that spin-lattice interactions can become so inefficient that they are ineffective as a relaxation process. It has been estimated 21 that the contribution to the line width made by spin-lattice interactions of Cr3+ in A1203 is only 1 at room temperature. Spin-spin mechanisms must therefore provide the necessary route for the efficient release of magnetic energy during resonance absorption. This again implies that the origin of y-resonance lies in a magnetically-exchange-coupled system of spins rather than in magnetically-isolated species. Where temperature-dependent spectra are observed, in SO2 samplesY6 they are anomalous showing line-splitting at high temperatures and positive-temperature deviation from the Curie-Weiss law.This behaviour is not satisfactorily explained by these authors. In conclusion, a surface mobile-electron Zener phase creates the ideal environment in which catalytic activity is maximised, the current model for the polymerisation active sites requiring both ease of oxidation and reduction and a quality of coordinative unsaturat io n. y-phase resonance arises from the surface of clustered chromiumA. ELLISON AND K. S. W. SING A. Ellison, J. 0. V. Oubridge and K. S . W. Sing, Trans. Faraday Suc., 1970, 66, 1004. A. Ellison and K. S. W. Sing, J.C.S. Faraday I 1978. 74, 2017. M. R. Harris and K. S. W. Sing, J. Appl. Chem., 1957, 7, 397. B. E. O’Reilly and D. S. MacIver, J. Phys. Chem., 1962, 66, 276.R. E. Cathers and W. W. Wendlandt, J. Inorg. Nuclear Chem., 1365,27, 1015. L. L. Van Reijen and P. Cossee, Disc. Faraday SOC., 1966, 41, 277. M. P. McDaniel and R. L. Burwell Jnr., J. Catalysis, 1975, 36,404. (a) F. S. Baker, J. D. Carruthers, R. E. Day, K. S. W. Sing and L. J. Stryker, Disc. Furaday SOC., 1971, 52, 173; (b) 3. D. Carruthers, IS. S. W. Sing and J. Fennerty, Nature, 1967, 213, 66; (c) J. D. Carruthers, J. Fennerty and K. S. W. Sing, 6th Int. Symp. Reactivity of Solids, 1968, ed. J. W. Mitchell, R. C. de Vries, R. W. Roberts and P. Cannon (J. Wiley and Sons, 1969), lo (a) R. L. Burwell Jnr., G. L. Haller, K. C. Taylor and J. F. Read, Adv. Catalysis, 1969, 20, 1 ; l1 C. Zener, Phys. Rev., 1951,82,403. l2 P. W. Anderson and H. Hasegawa, Phys. Rea, 1955, 100, 675.l3 P. G. de Gennes, Phys. Rev., 1960, 118,141. l4 G. C. Allen and N. S. Hush, Progr. Inorg. Chem., 1967, 8, 357. l5 M. P. Robin and P. Day, Ado. Inorg. Chem. Radio Chem., 1967,10, 247. l6 (a) K. A. Wilhelmi, Chem. Comm., 1966,437 ; (b) K. A. Wilhelmi, 0. Jonsson and E. Lagervall, l7 K. A. Wilhelmi, Actu Chem. Scand., 1965,19,165. l9 D. E. O’Reilly, F. E. Santiago and R. G. Squires, J. Phys. Chem., 1969, 73, 3172. ‘O F. J. Adrian, J. ColloidInterface Sci., 1968, 26, 317. 21 J. S. Thorp and H. P. Buckley, J. Material Sci., 1974, 9, 1499. ’’ Yu. Ermakov and V . Zakharov, Adu. Catalysis, 1975, 24, 173. 2817 ’ A. Ellison, review to be published. p. 127-135. (b) M. P. McDaniel and R. L. Burwell Jm., J. Catalysis, 1975, 36, 394. Acta Chem. Scand., 1969, 23, 1074.C. P. Poole Jnr. and D. S. MacIver, Adv. Catalysis, 1967, 17,223. (PAPER 7/2201) Magnetic and Optical Studies of Chromium Oxides Part 3.-Calcination of Coprecipitated Chromium and Aluminium Hydroxide Gels BY ALAN ELLIsoN*t AND KENNETH s. w. SING$. School of Science, Hull College of Higher Education, Kingston upon Hull HU6 7RT Received 16th December, 1977 Magnetic and optical studies were performed on coprecipitated chromium and aluminium hydroxide gels after calcination in air at temperatures between 50 and 1150°C. The results show that magnetic diIution was not achieved and that surface clusters of chromium oxide were formed. The magnetism, e.s.r. and optical spectra of oxidised samples are discussed in terms of a mixed- valency phase of partially-oxidised chromium containing Cr3+ and Cr6+ species.This new model ofy-phase chromium is shown to be consistent and an acceptable part of the extensive mixed-valency chemistry of chromium. This work continues a systematic study of the magnetic and spectral properties of chromium-alumina catalysts. It was shown in Parts 1 and 2 that the state of dispersion of chromium on the alumina support influences the stability of Cr6+, the occurrence and stability of y-phase chromium and the production of an " oxidised- phase " of chromium. There is some evidence that a well-dispersed state of chromium can be obtained using coprecipitation techniques and it is to be expected that on calcination of coprecipitated gels the chromium ion has the most favourable chance of occupying cation sites of well-defined symmetry in the host lattice.This paper is concerned with a study of the thermal decomposition of chromium and ahminium hydroxide gels coprecipitated from aqueous solution. EXPERIMENTAL The coprecipitated gels were prepared by allowing aqueous solutions of chromium (III) nitrate, aluminium nitrate and ammonium hydroxide to react under controlled conditions of pH, concentration and flow-rate in an apparatus similar to that of Harris and Sing,3 but modified to include a constant-head device. Three precipitates were prepared, containing 1.3, 6.1 and 12.3 wt. % Cr, respectively. The methods of calcination, analysis and measurement of magnetic susceptibility, e.s.r. and diffuse reflectance spectra are described in Part 1.' Samples are designated according to their chromium content (wt. %) and calcination temperature; thus, sample CAC 1.3 (50) contains 1.3 wt.% Cr and was calcined at 50°C for 5 h. RESULTS The mass susceptibility of Cry xz, measured at room temperature for samples CAC 1.3, CAC 6.1 and CAC 12.3 initially dried at 50°C, is plotted against the t Present address : School of Science, Faculty of Combined Studies, Hull College of Higher Education, Kingston upon Hull. $ Present address : School of Chemistry, Brunel University, Uxbridge, Middlesex. 1-89 28072808 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS I l l l l l ~ ~ l l ' l 200 600 600 800 1000 1200 T/"C FIG. 1.---YCm' at room temperature plotted against calcination temperature T for the samples x CAC 1.3 (50) ; 0, CAC 6.1 (50) ; A, CAC 12.3 (50).temperature of calcination in fig. 1. Between 50 and 200°C the observed decrease in xz is accompanied by a change in colour from blue or green to yellow or brown depending upon chromium content. The mass susceptibilities of samples CAC (200), CAC (400) and CAC (600) are substantially magnetic-field dependent, see table 1. A field-independent set of measurements is included for comparison and it should be noted that the original gels dried at 50°C had field-independent susceptibilities. 3500 3000 2500A . ELLISON AND K. S. W. SING 2809 to O L 500 3500 3000 2500 (b) H/kG gain 1000 n - - 1 1000 500 1 I 3 500 3000 (4 H/kG FIG. 2.-Qualitative e.s.r. spectra at 77 K for the samples (low field, gain 1000, high field, gain 40).(a) CAC 1.3 (200); (b) CAC 1.3 (400); (c) CAC 1.3 (600). Qualitative e.s.r. spectra, obtained at 77 K are shown in fig. 2 for samples CAC 1.3 (200), (400) and (600). Resonances of markedly different intensities are shown, recorded at different signal gains. The predominant resonance in each case is a strong narrow and symmetric signal at gain 40, peak to peak width (p.p.w.) 50 G, centred at ~3400 G. This is identified as the y-resonance of supported chromium." At a gain of 1000, &resonance is detected for the 200 and 400°C samples, but not for the 600°C sample. The fine structure is well resolved showing distinct lines at 11002810 MAGNETIC PROPERTIESOF COPRECIPITATED CHROMIUM HYDROXIDE GELS and 1700 G. The e.s.r. spectrum at 77 K for sample CAC 6.1 (400), consists of an intense y-resonance; P-resonance is also detected but is either of low intensity or, at this temperature, has been dipole-dipole broadened almost beyond detection.Fig. 3 shows molecular field plots, (xA)-'/m3 mol-1 against T/K, for samples CAC 1.3 (200), (400) and (600). These plots are curves convex to the temperature axis, giving extrapolated intercepts on this axis of between +90 and + 100 K. TABLE 1 .-MAGNETIC FIELD DEPENDENCE OF MAGNETIC SUSCEPTIBILITY 10: XF/rn3 kg-1 sample ( n ) (b) C) CAC 1.3 (50) 13.75 13.78 13.70 CAC 1.3 (200) 5.14 4.98 4.80 CAC 1.3 (400) 5.52 5.03 4.74 CAC 1.3 (600) 5.98 5.71 5.54 Field strength (a) 2.1 (b) 4.0 and (c) 6.5 kG. Quantitative e.s.r. spectra, for those samples showing only y-resonance at 3400 G, were determined over the temperature range 77 to 573 K.The resonance intensity I, calculated as the first moment of area of the y-resonance for samples CAC 1.3 (200), (400), (600) and CAC 6.1 (400) is shown in table 2 as a function of temperature. Examples of the spectra obtained are shown in fig. 4. The widths of the y-resonances 600 5 00 2 Y A00 53 L 300 d \ 4 8 ?3 200 100 1 1 I 0 100 200 300 TlK FIG. 3 . - 1 / ~ ~ plotted against temperature of measurement, T, (XA is the atomic susceptibility of Cr) for the samples (a) CAC 1.3 (200), x ; (b) CAC 1.3 (400), 0 ; (c) CAC 1.3 (W), A. remain constant within 1 G, over the whole temperature range and are independent of Cr content. Molecular field plots derived from these data are plotted in fig. 5 as I-1 against T. The plots show pronounced curvature, convex to the temperature axis, giving extrapolated intercepts close to those obtained from the xE-' against T plots for the same samples, see fig.3.281 1 A . ELLISON A N D K . S . W . SING TABLE 2.-E.S.R. SPECTRAL PARAMETERS FOR y-RESONANCE (T is the measurement temperature ; I is the relative number of spins ; p.p.w. is the peak to peak width) sample CAC 1 . 3 (200) CAC 1 3 (400) CAC 1.3 (600) CAC 6.1 (400) key T/K gain p.p.w./G 10-171 gain p.p.w:/G 10-17 Z gain p.p.w./G lO-17Z gain p.p.W./G 104’1 fig. 4 77 125 313 160 133 160 153 160 173 193 160 223 233 160 273 298 160 323 348 373 398 423 448 473 498 523 548 573 60 29.5 100 60 42.1 250 75 29.1 100 70 83.1 60 16.3 100 60 23.2 320 75 16.7 125 70 44.1 60 14.1 60 12.5 320 75 8.7 125 70 34.9 60 10.2 320 75 7.1 125 70 28.7 60 8.8 320 75 5.5 125 70 25.2 320 75 5.2 125 70 23.0 60 7.2 100 60 10.0 320 75 6.6 100 60 16.5 100 60 12.9 500 75 500 75 500 75 500 75 500 75 630 75 630 75 630 75 630 75 630 75 630 75 5.7 5.3 5.1 3.9 2.9 2.5 2.5 2.9 3.7 4.8 5.5 F~o.4.-Quantitative e.s.r. spectra of y-resonance for sample CAC 1.3 (600). Key for spectra (a) to ( I ) is given in table 2.2812 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS 30 25 2 0 rl I 0 + 0 P( CI 1 E 1 5 - 5 1 0 L 5 - 35 - - - I 30 25 10 5 1 I I J 0 100 200 300 T/K FIG. 5.-Quantitative e.s.r. spectra ; I-' plotted against temperaturelof measurement, T. ( I is the relative e.s.r. intensity ofy-phase Cr). x , CAC 1.3 (200) ; 0, CAC 1.3 (400) ; A, CAC 1.3 (600)A.ELLISON AND K. S . W. SING 281 3 Values of 1 for the y-resonance of CAC 1.3 (600) are plotted against T in fig. 6. The intensity at low temperatures is seen to increase with decreasing temperature at a much greater rate than expected for a paramagnet obeying the Curie or Curie- Weiss laws. The data was collected from two separate series of measurements, from 77 to 298 K and from 298 to 573 K causing the obvious discontinuity at 298 K. However, the increase in intensity with increasing temperature after x 470 K is anomalous and indicative of a transition temperature. DISCUSSION Between the calcination temperatures 50 and 200°C there is a decrease in ~2, accompanied by a change in sample colour, due to oxidation of Cr3+ ions. xE decreases to give a minimum centred at ~ 4 0 0 ° C for each sample, independent of Cr content, behaviour similar to that observed for CrC1,-Al,O, samples.2 The breadth of the minima and the temperature at which xg begins to increase, uiz.at 600°C for CAC 1.3 and at 400°C for CAC 6.1 and CAC 12.3, depends upon Cr content. This refiats the manner in which the stability of oxidised chromium depends upon the degree of dispersion of chromium on the support.l* Chromium at low concentration is more stable to oxidation than chromium at higher concen- tration. Indeed fig. 1 shows that even at 400°C some proportion of Cr3+ remains unoxidised. Samples CAC 6.1 and CAC 12.3 have identical Xz-minimum values; in the case of CAC 1.3 the position and value of the Xz-minimum is uncertain due to the marked field-dependent susceptibilities observed. Nevertheless the xE values all lie within a very narrow range.After calcination at 1120°C all of the chromium has been reduced to the Cr3+ state. The xE values are larger than the corresponding values in related systems suggesting that a more dispersed form of chromium has been achieved. This idea is confirmed on observing that the low-field derivative spectra in fig. 2 for CAC 1.3 are much more well defined and narrower than similar &resonances reported else- where.2 This width is usually interpreted in terms of a range of low-symmetry, zero-field terms, D and E, in the spin Hamiltonian. However an appropriate operator representing dipole-dipole broadening would provide an acceptable alter- native interpretation which is certainly more realistic for the planar clusters of chromium discussed in Part 2.2 On this model the 8-spectra of fig.2 suggest weaker dipole-dipole coupling due to greater dispersion of chromium in the support. The nature of the chromium responsible for the XC,'-minirna is worthy of some discussion. It has been shown that supported Cr3+, in both impregnated and coprecipitated conditions, is incompletely oxidised within a small temperature range, independently of concentration. This is not the behaviour of unsupported Cr3+. Cathers and Wendlandt have shown that all isomers of hydrated CrC1, decompose on heating to give Cr203, with loss of HCl and HzO, without passing through oxida- tion states higher than 3+. Similarly a-Cr,O,, does not undergo oxidation on heating in this temperature region.In addition, over the same temperature range, it is observed that supported Cr6+ is reduced and although the extent of reduction depends upon chromium content, e.s.r. and susceptibility measurements show that considerable reduction of low chromium content samples has occurred at 200"C.1 It is significant that not only do the xg-rninima occur in this region of coincident oxidation and reduction but also that the predominant paramagnetic phase, from e.s.r. evidence, is y-phase chromium independent of loading and of type of supported2814 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS system. In addition susceptibilities are magnetic field-dependent and the values of Xz-minima show very little variation with chromium content.The chromium oxidation states whose identities have been firmly established in this region, by optical spectra and e.s.r. techniques, are Cr3+ and Cr6+. y-phase chromium, the predominant paramagnetic phase, has been interpreted as magnetically- isolated Crs+ ions which are not coupled by exchange interactions.6 If one accepts this model, for the sake of argument, the situation arises that in the calcination temperature range 200-600°C and at the Xz-minima, one must consider that there are three states of chromium present, viz., Cr3+, Cr6+ and Cr5+. x:, iteff and 6 depend markedly upon the relative amounts of the different chromium species and upon their individual degrees of dispersion. In fact we have observed that, in this calcination temperature range, xE remains remarkably constant even though the rates and magnitudes of oxidation and reduction and the values of 6 and of peff of each chromium species are very different and, moreover, depend upon chromium content. Further, the proportion of y-phase chromium produced on calcination of Cr3+ or Cr6+ depends markedly upon the surface area of the chromium, the nature of the support, the chromium content and the calcination temperature, resulting in modification of petf, 8 and ~2.'.x-' / .- / I /A . ELLISON AND K. S . W. SING 2815 of the mobile carrier spin during a " hop " from one lattice site to the next. The mobile carrier contributes to the binding energy of the system provided that the spins on neighbouring sites are parallel. The electric and magnetic properties of such compounds depend strongly upon their composition, x.Thus when x = 0 or 1 the oxides are good insulators and antiferromagnetic, while for a range of intermediate compositions the electric conductivity is several orders of magnitude greater and the materials are ferromagnetic. The theoretical temperature dependence of the magnetic susceptibility is shown in fig. 7 and exactly describes the curved x-' against T plots reported here. Provided that the upper transition temperature is antiferromagnetic, a discon- tinuity in slope is observed at TN, yielding a negative, high-temperature Weiss constant 8. Low-temperature extrapolation yields a positive-temperature intercept, T,.' * The oxidised phase of chromium, 02--Cr3+-02--Cr6+-02- possesses magnetic properties described by the Zener model; moreover the existence of collective Zener electrons is shown by the colours, spectra and p-type semiconductivity of this phase.At the extremities of the series (Cr:' Cr;Lx)Oi- are Cri+03 (x = 0, n-type semi- conductor, antiferromagnetic) and Cr6+03 (x = 1, insulator, temperature independent paramagnetic). The stable Zener phase, when x > 0, occurs at the surface of chromium clusters where the antisotropic ferromagnetic coupling responsible for stability is maximised. It is important to realise that the existence of mixed-valency compounds is well established and in particular is characteristic of the chemistry of chromium. For example, the reviews of Allen and Hush l4 and of Robin and Day cite totals of 150 and 820 references, respectively.It is not possible to give here a complete review of the incidence of mixed-valency species in chromium chemistry. Neverthe- less this review does reveal that Crrrr-Crvr, Crrrr-Crrr, Crrrr-Crrv and Crv-Crvl mixed-valency compounds occur as stable stoichiometric and non-stoichiometric compounds. Most importantly Wilhelmi and co-authors 16* have shown that compounds of the type MCr308 and M2Cr309 (M = alkali metal ion)16 and intermediate oxides of chromium, including Cr205, Cr308, CrsOIz and Cr,OI5 l7 are black crystalline substances containing CrrIr-CrVr mixed-valency species alone. It is maintained that the magnetic and optical properties of the oxidised-phase of chromium are more completely explained on the basis of mixed-valency species of the type CrIrr-Crvr without recourse to the requirement of CrV ions stabilised at specific sites in the support l a t t i ~ e .~ Indeed the observed data are typical of those expected for a Zener double-exchange system. ~2 for samples at susceptibility minima is often magnetic field dependent. Molecular field plots (xi1 against T) show pronounced curvature towards the tempera- ture axis with intercepts of w -k90 IS. This unusual behaviour cannot be explained on uncoupled systems of Cr6+ or Cr3+ ions. Cr6+ ions have a temperature- independent susceptibiIity which would produce high-temperature deviation from linearity not observed here. Octahedral Cr3+ has a weak field 4A2, ground state with minimal orbital angular momentum contribution to peff or 6 through spin-orbit coupling. The amorphous Cr3+ phases in the temperature range 77 to 298 K are paramagnetic producing P-phase e.s.r., giving negative 6 values but no curvature in these plots.Ferrimagnetic structures could produce such curves but require well- ordered, three-dimensional interpenetrating-ferromagnetic sub-lattices. In, for example, C4" symmetry Cr5+ ions with a split 2T2 ground state can produce low- temperature deviation from Curie-Weiss behaviour but the deviation is away from the temperature axis, not as observed here. The Zener theory, however, does predict curved molecular field plots.2816 MAGNETIC PROPERTIES OF COPRECIPITATED CHROMIUM HYDROXIDE GELS Visible-u.v. reflectance spectra show Cr3+ and Cr6+ bands with, in addition, enhanced absorbance over the whole wavelength range, 200-1000 nm.Most signifi- cantly, these brown-black samples absorb radiation at wavelengths where the postulated individual species (Cr3+, Cr6+, Cr5+) are transparent. These observations are consistent with the existence of Zener mobile-electrons in a class 111 mixed- valency system. The predominant e.s.r. signal is y-resonance. Both 6- and P-resonances are often not detected even though the presence of Cr3+ species is confirmed from reflectance spectra. This suggests that the normal resonances of Cr3+ electron-spins have been obliterated through strong exchange forces. attaining a maximum intensity for alumina samples near 2 wt. % Cr where the proportion of small and thin, platelike chromium clusters is the greatest. Indeed, y-resonance appears at the expense of /?-resonance whilst &resonance is still often retained.y-resonance achieves greatest intensity for silica supported chromium, the most active catalysts in ethylene polymerisation. It has been asserted that &phase chromium has never been observed in silica systems, although after achieving a greater dispersion of chromium by thermal decomposition of mixed oxalato pre- cipitates, a small but unstable &phase signal has been reported.lg Cr5+ ions, often said to be responsible for y-resonance, need to occupy high symmetry support-lattice sites to acquire the necessary stabilisation energy. However, not only is a dispersed phase of CrS+ (the product of &phase) unlikely but also there is much evidence that chromium is insoluble in silica even at the liquidus temperature.' It would appear that although Cr3+ species occur in the oxidised chromium their normal magnetic properties are obliterated.Instead the oxidised phase itself exhibits different behaviour which is not characteristic of Cr5+ ions isolated in the support lattice. Moreover molecular field plots for y-resonance, I-' against T, in all of the supported systems studied show pronounced curvature with positive temperature intercepts, behaviour consistent with direct exchange interactions between ions of different oxidation state. For alumina-supported chromium the width of the y-resonance, from 77 to 593 K, is completely temperature-independent. The usual argument invoked is that spin- lattice relaxation is slow.However, if this is the case, and if as is observed the resonance does not show signs of saturation then another predominating relaxation mechanism must be efficiently operative. Thus Adrian 2 o maintains that spin-lattice interactions can become so inefficient that they are ineffective as a relaxation process. It has been estimated 21 that the contribution to the line width made by spin-lattice interactions of Cr3+ in A1203 is only 1 at room temperature. Spin-spin mechanisms must therefore provide the necessary route for the efficient release of magnetic energy during resonance absorption. This again implies that the origin of y-resonance lies in a magnetically-exchange-coupled system of spins rather than in magnetically-isolated species. Where temperature-dependent spectra are observed, in SO2 samplesY6 they are anomalous showing line-splitting at high temperatures and positive-temperature deviation from the Curie-Weiss law.This behaviour is not satisfactorily explained by these authors. In conclusion, a surface mobile-electron Zener phase creates the ideal environment in which catalytic activity is maximised, the current model for the polymerisation active sites requiring both ease of oxidation and reduction and a quality of coordinative unsaturat io n. y-phase resonance arises from the surface of clustered chromiumA. ELLISON AND K. S. W. SING A. Ellison, J. 0. V. Oubridge and K. S . W. Sing, Trans. Faraday Suc., 1970, 66, 1004. A. Ellison and K. S. W. Sing, J.C.S. Faraday I 1978. 74, 2017. M. R. Harris and K. S. W. Sing, J. Appl. Chem., 1957, 7, 397. B. E. O’Reilly and D. S. MacIver, J. Phys. Chem., 1962, 66, 276. R. E. Cathers and W. W. Wendlandt, J. Inorg. Nuclear Chem., 1365,27, 1015. L. L. Van Reijen and P. Cossee, Disc. Faraday SOC., 1966, 41, 277. M. P. McDaniel and R. L. Burwell Jnr., J. Catalysis, 1975, 36,404. (a) F. S. Baker, J. D. Carruthers, R. E. Day, K. S. W. Sing and L. J. Stryker, Disc. Furaday SOC., 1971, 52, 173; (b) 3. D. Carruthers, IS. S. W. Sing and J. Fennerty, Nature, 1967, 213, 66; (c) J. D. Carruthers, J. Fennerty and K. S. W. Sing, 6th Int. Symp. Reactivity of Solids, 1968, ed. J. W. Mitchell, R. C. de Vries, R. W. Roberts and P. Cannon (J. Wiley and Sons, 1969), lo (a) R. L. Burwell Jnr., G. L. Haller, K. C. Taylor and J. F. Read, Adv. Catalysis, 1969, 20, 1 ; l1 C. Zener, Phys. Rev., 1951,82,403. l2 P. W. Anderson and H. Hasegawa, Phys. Rea, 1955, 100, 675. l3 P. G. de Gennes, Phys. Rev., 1960, 118,141. l4 G. C. Allen and N. S. Hush, Progr. Inorg. Chem., 1967, 8, 357. l5 M. P. Robin and P. Day, Ado. Inorg. Chem. Radio Chem., 1967,10, 247. l6 (a) K. A. Wilhelmi, Chem. Comm., 1966,437 ; (b) K. A. Wilhelmi, 0. Jonsson and E. Lagervall, l7 K. A. Wilhelmi, Actu Chem. Scand., 1965,19,165. l9 D. E. O’Reilly, F. E. Santiago and R. G. Squires, J. Phys. Chem., 1969, 73, 3172. ‘O F. J. Adrian, J. ColloidInterface Sci., 1968, 26, 317. 21 J. S. Thorp and H. P. Buckley, J. Material Sci., 1974, 9, 1499. ’’ Yu. Ermakov and V . Zakharov, Adu. Catalysis, 1975, 24, 173. 2817 ’ A. Ellison, review to be published. p. 127-135. (b) M. P. McDaniel and R. L. Burwell Jm., J. Catalysis, 1975, 36, 394. Acta Chem. Scand., 1969, 23, 1074. C. P. Poole Jnr. and D. S. MacIver, Adv. Catalysis, 1967, 17,223. (PAPER 7/2201)