J. MATER. CHEM., 1991, 1(4), 511-515 511 X-Ray Diffraction Characterization of Iridium Dioxide Electrocatalysts Alvise Benedetti,*" Stefan0 Polizzi," Pietro Riello," Achille De Battistib and Andrea Maldottib a Dipartimento di Chimica Fisica, CaIle Larga S. Marta 2137, 1-30123 Venezia, Italy Dipartimento di Chimica dell'Universita and Centro di Studio sulla Fotochimica e Reattivita degli Stati Eccitati C.C. del C.N.R., via L. Borsari 46, 1-44100 Ferrara, Italy An X-ray diffraction (XRD) line-broadening analysis of coatings of pyrolytic iridium dioxide supported on amorphous silica microbeads is reported. The influence of the temperature of pyrolysis and of the annealing treatments on the average crystallite sizes of the IrO, and Ir phases have been studied in detail.The values obtained for IrO, are lower than those obtained from an analgous RuO, system. However, the analysis suggests that in both cases the crystalline phase grows within an amorphous microcrystalline environment where impurities are mostly segregated in the amorphous phase and then released to the atmosphere. The lattice parameters are also calculated. Compared with the reference values, slightly greater values are obtained both for IrO, and for Ir. Keywords: X-Ray diffraction ; Line broadening; Iridium dioxide; Electrocatalyst The industrial interest of pyrolytic oxides of ruthenium and iridium has stimulated many works on the electrochemistry of these synthetic materials since the early 1970s. The literature on fundamental and industrial aspects has been thoroughly discussed in several and the influence of annealing temperatures of ruthenium oxide electrodes on the microstruc- tural features was recognized by Pizzini et a!.in one of the first papers.' The same aspect was implicitly taken into account in other pioneering where the electrochemi- cal charging mechanism of this material has been studied. Analogous investigations were carried out later on iridium dioxide, for which precursor paths and microstructural fea- tures have also been discu~sed.~-'~ Generally, the microstruc- tural texture of RuO, and IrO, affects their electrochemical properties in different ways. The surface area or the roughness factor substantially reflects the crystallite size which is related to the 'degree of dispersion' of the material.In this sense, not only the catalytic activity, but also the resistance to corrosion under applied potential may be modified by changing the microstructure of the oxides. This aspect, directly correlatable with the preparation path, is probably the most widely considered in the literature. 192,9 Moreover, the presence of lattice distortions can also exert a basic effect on the chemical behaviour of the material: a large number of microstructural defects can be related, in principle, to a larger number of active sites. In this context the analysis of the line-broadening of XRD peaks has proven to be extremely useful for the microstructural characterization of Crystalline phases.However, the simplified approaches generally used in the characterization of Ru02 and IrOz films,'T6 which are based on the measurement of the apparent crystallite size calculated by the Scherrer equa- tion, have limited the potential of this approach. In the present work a careful XRD line-broadening analysis on coatings of pyrolytic iridium dioxide supported on amorph- ous silica microbeads is presented. In a previous study, the formation of iridium metal, together with the expected oxide was discussed on the basis of XRD phase identification and thermoanalytical data." In the same work an analysis of the effect of temperature of pyrolysis and temperature and dur- ation of the annealing treatments was also carried out. The complexity of the transformations that the film undergoes during the different thermal treatments"." has prompted a more detailed XRD investigation.In particular, a Fourier line-broadening analysis of XRD peak profiles, and, if possible, the Warren-Averbach (W-A) method" allow a careful microstructural analysis of the different phases in order to obtain average crystallite sizes, crystallite size distributions and lattice distortions possibly present. Usually this approach, for which a complete knowl- edge of the peak profiles on a large portion of reciprocal space is necessary, is limited by the presence of very broadened peak envelopes and/or of an important or complex back- ground. In order to overcome this problem a 'profile fitting technique' has been used, in which one fits the experimental envelopes of peaks with analytical functions [one for the background and two for each profile (Kal +KQ,)]utilizing a minimization routine. Previous tests on different materials 16v1 showed that it is possible to obtain very reliable results in spite of some simplifications and theoretical assumptions.Experimental and Methodology Samples Preparation The samples investigated were prepared by painting amorph- ous silica microbeads (spherosil) with isopropyl alcohol solu- tions of hydrated iridum(Ir1) chloride, followed by evaporation of the solvent under controlled conditions." The final coatings were obtained by pyrolysis at 673 K for 2 h, and by successive annealing at 673, 773 and 873 K for periods of 15 h and at 673 K for 30 and 58 h, respectively.Both pyrolysis and annealing were performed under dry oxygen. In such a way we were able to analyse the influence of the temperature and of the time of annealing on the microstructural features of the developed components. Wide-angle X-Ray Scattering (WAXS) Profiles were collected using a Philips vertical goniometer connected to a highly stabilized generator. Cu-Ka Ni-filtered radiation, a graphite monochromator and a proportional counter with a pulse-height discriminator were used. A step- by-step technique was employed; steps were of 0.05"with an accumulation counting time of 100 s per angular abscissa. Line-broadening Analysis For each peak, two pseudo-Voigt functions (linear combi- nation of a Gaussian profile with a Cauchian one) related by the well known crystallographic constraints on peak intensit- ies, angular positions and half-width at half-maximum were used in order to take account of the Ka,-Ka2 doublet.16 For the background, a straight line or a polynomial of the third degree was used, depending on the angular range under consideration.The minimization was carried out using a modified version of the Simplex method.18 The optimized pseudo-Voigt functions relevant to the Ka, of the experimental broadened profile were deconvoluted from the instrumental and spectral effects in order to obtain the corrected Fourier transforms A(L)(Lis the variable in direct space).Ig According to Warren and Averbach the coefficients A(L)(or the Fourier transforms) are the products of size coefficients A&) and distortion coefficients Ad&): 4L)=As(ww) where A&) is independent of the peak order and A&) is dependent upon the order of the diffraction peak.If at least two orders of reflections of the same plane family are known, by means of the following expression: In A (L,l/dhkr2)=In A&) -27r2(~2(L))L2/dhk12(1) where hkl are the Miller indices, (c2(L))is the squared microstrain averaged over all distances L, and dhkr is the interplanar spacing, it is possible to separate the crystallite size contribution from that of the lattice distortion. Further- more, it has been shown that the volume-weighted crystalline size distribution, p,(L), obeys the following relation P,(L) LCd2As(L)/dL2I (2) while the volume-weighted average crystallite size (D),calcu-lated perpendicular to the (hkl)planes is given by m W (3) IJ 0 J. MATER.CHEM., 1991, VOL. 1 Only the peaks in the range between 10 and 50" in 28 and the 202 Ir02 peak at ca. 73" were examined. For the other profile envelopes it was not possible to obtain an unambiguous and stable solution for the very broadened constituent reflec- tions. Moreoever, owing to the variable values of the unit cell parameters of the crystalline phases considered (see Table 3, later), it was not possible to fix, a priori, some 'known' parameters in the fitting procedure, at the position of the maximum of the peaks, in order to diminish the number of parameters to minimize.In order to describe the decreasing background scattering of the amorphous silica, a polynomial of the third degree was considered for the 28 range ca. 24-45". This simplification was necessary in the fitting procedure because no suitable scaling factor of the 'true' experimental amorphous back- ground, scattered by the microbeads, could be found. Even if this fact could be explained by interference effects, the contri- bution of a non-crystalline phase containing Ir cannot be excluded. For narrower ranges a straight line was found sufficient to reproduce the background. Results The spectra of the five different samples are characterized by a big halo centred at about 28=22", due to the amorphous silica, and by the presence of two different crystalline phases, Ir02 (rutile structure) and Ir.As already described in our previous paper, the higher the annealing temperature the smaller the amount of iridium present, whereas the amount of iridium was almost constant for different times of isothermal treatment. lo In Fig. 1 we report, as an example, the X-ray pattern of the sample annealed at 673 K for 30 h. From the figure, and inset, it is evident that the presence of the background and the overlaps makes the use of the 'profile-fitting technique' necessary in order to apply Fourier X-ray analysis. In this way, it has been possible to study the 110, 101, 200 and 202 reflections relevant to the Ir02 phase and the 111 and 200 reflections relevant to the Ir phase.In Fig. 2 the fitting of the 10 20 30 40 50 2e/o Fig. 1 XRD pattern of the sample annealed at 673 K for 30 h. The inset shows the best-fitted envelope of the 200 IrO, and the 111 Ir profiles with the corresponding residuals J. MATER. CHEM., 1991, VOL. 1 i6 N e X u) -C 3 $4.L .-u) Q,c. .-C 2 24 30 36 42 261" Fig. 2 Experimental data (dots) and fitting (continuous line) of the envelope of the amorphous component; the 110, 101 and 200 Ir02 and 11 1 Ir profiles relevant to the sample annealed at 673 K for 58 h. Residuals are reported below IrOz 110, 101, 200 and the Ir 111 profiles, together with the amorphous component, is shown for the sample annealed at 673 K for 58 h.The index used was Rwp ={Twi[Ifit(28i)-zobs(28i)12 /icwizobs(28i)2 Yl' where wi= 1/[Iob&?8i)]is the weight assigned to the ith observation, Zfit(20i)and IObs(2Oi)are the calculated and observed profile intensities at 28,. The obtained R,, values (<2.0%), indicate that the fitted analytical envelope repro- duces the experimental X-ray pattern in a satisfactory way. The W-A method has been applied to the 101 and 202 IrOz peak profiles. In Fig. 3 the relevant Fourier coefficients for the sample annealed at 873 K for 15 h are shown. The inset shows the corresponding Warren-Averbach diagram. The slopes of the straight lines in this diagram are related to the strain parameter, according to eqn. (1): negative values indicate the presence of strain.The slopes shown in Fig. 3 are 0.3 .. h-A 0.9. -q 011.5:- I 1. 0.15 0.3 0.45 1ldkl 0 40 80 LIA Fig. 3 A(L) Fourier coefficients for the 101-202 pair of reflections for the sample annealed at 873 K for 15 h. Inset: W arren-Averbach diagram in which the logarithms of A(L) are reported as functions of l/dhkr2(AL=3.56 A) nearly zero within the errors involved in this procedure, so that this kind of disorder can be considered negligible. Similar results have been obtained for the other samples. This confirms that no strain is present in the Ir02 crystalline phase. Table 1 shows the values of the volume-weighted average crystallite sizes obtained from each peak. The Fourier analysis of profiles 110 and 200 is further evidence of the absence of strain.In fact, for these two profiles, although all the broaden- ing has been attributed only to the crystallite size, the resulting average crystallite sizes are similar to the ones obtained by applying the W-A method. The fact that the crystallite sizes calculated from the different analysed profiles of the same sample are very similar, indicates that Ir02 crystallites can be considered equiaxial in three-dimensional space. The last column of Table 1 reports the average of the volume-weighted average crystallite sizes of the four profiles, D.The duration of the annealing treatment has no effect on the Ir02 crystallite size. As far as the effect of the annealing temperature is concerned, the increase of (D), from 673 to 873 K can be considered larger than the relevant estimated error (10-15%).The role of the different pyrolysis temperature is shown in Fig. 4. The volume-weighted crystallite size distribution obtained according to eqn. (2) for the 101 reflection for the sample annealed at 873 K [curve (a)] and that obtained directly by pyrolysis of the precursor salt at 873 K [curve (b)] are reported. The different pyrolysis temperature affects the crystallite size quite significantly. The behaviour of the iridium crystalline phase under heat treatment is rather different. In Table2 the (D), values are reported. Both the time of isothermal annealing and the annealing to higher temperature have the effect of increasing the iridium crystallite size.The difference obtained for the two considered crystallographic directions could be due to a preferential direction of growth of the crystallites. However, owing to the lack of further peaks for analysis and to the large amorphous contribution the problem has no definite solution. The application of the fitting procedure also has the advan- tage that it supplies the values of the position of the peak maximum with very high precision, allowing one to obtain accurate lattice parameters. Table 3 shows the values of the IrOz and Ir unit-cell parameters, obtained with a computer J. MATER. CHEM., 1991, VOL. 1 Table 1 Average crystallite sizes calculated from different hkl profiles relevent to the IrO, phase average crystallite size (D>,/A annealing temperature/K time/h W-A pair 101-202 110 200 D/A 673 15 37 41 43 40 673 30 40 42 36 39 673 58 38 48 43 43 773 15 41 46 51 46 873 15 54 55 45 51 thermal and the annealing treatment, neither crystalline phase is easily reproducible owing to the presence of substitutional ions in the structure.Discussion and Conclusions The relatively low (D),values of the IrO, crystallites yield a larger dispersion of the crystalline part of the IrO, catalyst compared to that obtained for an analogous RuO, system." In this case, an average crystallite size up to 1500 A was found under the same preparation conditions. However, in order to propose a possible explanation of this different behaviour, it is necessary to consider other aspects of anal- ogous IrO, systems reported in previous papers.Chemical analysis," for instance, indicates that larger amounts of residual water and chlorine species are present in L/A pyrolitic iridium oxide films compared with ruthenium oxide Fig. 4 Volume-weighted crystalline size distributions p,(L),calculated films prepared at the same temperature from the respective from the 101 reflection, relevant to the sample annealed at 873 K for hydrated chlorides." Thermal analysis confirms this result. lo 15 h [curve (a) (D), =51 A] and to the sample obtained directly by Hydrogen profiles obtained by the 1H('5N,ay)'2C nuclear pyrolysis of the precursor salt at 873 K [curve (b) (0),=80 A] reaction for RuO, and IrO, films" indicate that hydrogen species are stored in larger amounts in the IrO, samples. The Table 2 Volume-weight average crystallite sizes calculated from the average oxygen stoichiometry, found by Rutherford backscatt- 111 and 200 profiles relevant to the Ir phase ering spectrometry was 2.5 for IrO, and 1.8 for RUO,~~~~~.average crystal-Moreover, in these systems, the RuO, and the IrO, crystalline lite size phases grow from a pre-existing amorphous phase,14 and for (D>,lA the RuO, with lattice parameters closely related to the tem- perature of annealing and chloride ~ontent:~." the higher the annealing temperature/K time/h 111 200 temperature the closer the lattice parameters are to the refer- ence values.On the basis of these results, it seems reason- 673 15 94 72 able to assume that the crystalline rutile phase, common to 673 30 115 72 673 58 124 87 both RuO, and IrO,, grows within an amorphous microcrystal- 773 15 113 79 line environment where impurities are mostly segregated in 873 15 153 133 the amorphous phase and then released into the surround- ings. This stage seems to be quite slow in the case of the iridium dioxide films and powders, thus preventing further Table 3 Unit-cell parameters relevant to the tetragonal IrO, and growth of IrO, crystallites, e.g. during the annealing treat- cubic Ir phases ments. Only direct pyrolysis at higher temperatures allows IrO, the achievement of larger crystallite size as shown in Fig.4. annealing For this system, the different values of lattice parameters are temperature/K time/h a,/A c,/A Ir a,/A influenced by the temperature and the annealing time, how- ever, no direct influence of the chloride ions has been detected. 673 15 4.544(3) 3.175(5) 3.859(3) As has already been outlined, as far as the Ir phase is 673 30 4.505(3) 3.183(5) 3.853(3) concerned, we have not been able to evaluate the disorder 67 3 58 4.497(4) 3.178(2) 3.846(2) 4.533(6) 3.165(3) 3.853(4) influence, nor consequently to clarify the role of the impurities. 773 15 873 15 4.522(6) 3.162(3) 3.850(3) In any case, the variation of the unit-cell parameters seems Lit.20 4.4983 3.1544 3.8394 to suggest a 'true' decrease of a, as a function of the annealing treatment.The results obtained on the microstructure of IrO, films refinement programme for the different samples. In general, are also interesting from the electrochemical point of view. In compared with the reference values," slightly greater values fact, as mentioned in the introduction, the electrocatalytic are obtained both for the IrO, and for Ir. In any case, when activity of oxide electrodes, and their instability in relation to the time or the temperature are increased the values approach corrosion processes taking place under polarization, are linked the standard ones. This fact could suggest that, after the by the ion-insertion characteristics of these materials. Accord- J. MATER. CHEM., 1991, VOL. 1 ing to the evidence obtained in this work, the IrO, crystallites show no strain, which should minimize ion-exchange processes across them, so limiting the ions’ mobility.The intergranular region, on the other hand, is likely to be quite large and, owing to its defective nature, could allow larger diffusion coefficients for small ions such as protons, that are involved in the charge-storage mechanism. IrO, electrodes prepared by the above-mentioned method in fact show large charge- storage capacity, compared with RuOz e1ectr0des.l~ We are grateful to Professor G. Fagherazzi for his constructive criticism and to Mr. L. Bertoldo for assistance during exper- imental work. This research work was supported by CNR (Progetto Finalizzato Materiali Speciali per Tecnologie Avanzate) and by Minister0 dell’universita e della Ricerca Scientifica (MURST 40%).References 1 S. Trasatti and G. Lodi, in Electrodes of Conductive Metal Oxides, ed. S. Trasatti, Elsevier, Amsterdam, 1980, part A, p. 301. 2 S. Trasatti and G. Lodi, in Electrodes of Conductive Metal Oxides, ed. S. Trasatti, Elsevier, Amsterdam 1981, part B p. 521. 3 A. Nidola, in Electrodes of Conductive Metal Oxides, ed. S. Trasatti, Elsevier, Amsterdam, 1981, Part B p. 627. 4 D. M. Novak, B. V. Tilak and B. E. Conway, in Modern Aspects of Electrochemistry, ed. B. E. Conway and J. Bockris, Plenum Press, New York, 1982, vol. 14, p. 195. 5 S. Pizzini, G. Buzzanca, C. Mari, L. Rossi and S. Torchio, Mnter. Res. Bull., 1972, 7, 449. 6 G. Lodi, G. Zucchini, A.De Battisti, E. Sivieri and S. Trasatti, Mater. Chem., 1978, 3, 179. 7 S. Trasatti and G. Buzzanca, J. Electroanal. Chem., 1971, 29, 1. 8 G. Lodi, E. Sivieri, A. De Battisti and S. Trasatti, J. Appl. Electrochem., 1978, 8, 135. 9 C. Angelinetta, S. Trasatti, L. J. Atanososka, Z. S. Minevski and R. T. Anatososki, Mater. Chem. Phys., 1989, 22, 231. 10 G. Lodi, A. De Battisti, A. Benedetti, G. Fagherazzi and J. Kristof, J. ElectroanaI. Chem., 1988, 256,441. 11 G. Lodi, A. De Battisti, G. Bordin, C. De Asmundis and A. Benedetti, J. Electroanal. Chem., 1990, 277, 139. 12 G. Battaglin, A. Carnera, G. Della Mea, G. Lodi and S. Trasatti, J. Chem. SOC., Faraday Trans. 1, 1985,81, 2995. 13 G. Lodi, G. L. Wucchini, A. De Battisti, A. Giatti, G. Battaglin and G. Della Mea, Sur-Sci, submitted. 14 I. D. Belova, T. V. Varlamova, B. Sh. Galyamov, Yu. E. Rogin- skaya, R. R. Shifrina, S. G. Prutchenko, G. I. Kaplan and M. A. Sevostyanov, Mater. Chem. Phys., 1988, 20, 39. 15 B. E. Warren, and X-Ray Difraction, Addison- Wesley, Reading, MA. 1969, p. 264. 16 S. Enzo, S. Polizzi and A. Benedetti, 2. Kristallogr., 1985, 170, 275. 17 A. Benedetti, G. Fagherazzi, S. Enzo and M. Battagliarin, J. Appl. Crystallogr., 1988, 21, 543. 18 J. A. Nelder and R. Mead, Comput. J., 1965, 7, 308. 19 S. Enzo, G. Fagherazzi, A. Benedetti and S. Polizzi, J. Appl. Crystallogr., 1988, 21, 536. 20 Joint Committee on Powder Diffraction Standards, 1988, Powder Diffraction File (Swarthmore, Pennsylvania: International Centre for Diffraction Data) no. 6-0598 and 15-870. 21 G. Lodi, C. Bighi and C. De Asmundis, Mater. Chem., 1976, 1, 177. Paper 0105257D; Received 22nd November, 1990