Inorganic Oxide Aerosols of Controlled Submicronic Dimensions H. MOZZANEGA A. THEVENET BY F. JUILLET F. LECOMTE S. J. TEICHNER AND P. VERGNON Institut de Recherches sur la Catalyse (C.N.R.S.) DCpartement de Chimie Physique 69-Villeurbanne France Received 4th December 1972 Metallic oxides aerosols are prepared by decomposition of anhydrous chlorides in the diffusion flame of a hydrogen-oxygen reactor. The flow rate of the chloride vapour the temperature of the flame and the residence time of the reagent in the flame determine the shape (spherical or polyhedral) the dimensions (in the range from 60A to 2000 A) and in some cases the crystalline structure of the particles which in all cases are non-porous. These highly divided oxides exhibit unusual photocatalytic properties which are not encountered with aerosols in the micron range or with porous particles prepared in a conventional way.Titanium dioxide in particular enables the catalytic photo-oxidation (in the u.-v. range) at room temperature of organic and inorganic compounds. Paraffins and olefins are oxidized partially and/or totally whereas ammonia yields N20and NZ,carbon monoxide yields carbon dioxide and hydrogen sulphide yields sulphur dioxide and sulphur. Industrial smokes inject into the atmosphere submicronic particles of metallic oxides which are often in contact with industrial gases containing hydrocarbons ammonia carbon monoxide hydrogen sulphide sulphur dioxide and oxides of nitrogen. Since the natural sedimentation of these aerosols is a slow process their reaction with the above gases in the presence of oxygen and of u.-v.irradiation is of interest in the study of environmental problems. Although the smokes of metallic oxides are usually prepared in the laboratory by electric arc or plasma methods we preferred to generate the submicronic aerosols by the flame reactor method 1* because it permits a good control of the size and the shape of the oxides and also because it is more closely correlated to the industrial " involuntary " generation of particles in the smokes. A special interest was attached to the shape of oxide particles generated in the flame reactor because previous studies have shown that surface properties of anatase aerosols are different for polyhedral or spherical particles.Indeed spherical particles exhibit a statistical abundance of all crystallographic planes whereas polyhedral particles may have some privileged planes developed. Moreover the number of discontinuities at the surface (corners edges and steps) is also greater in the poly- hedral particles and therefore point defects because of poorly coordinated surface ' ions seem to be more abundant in these particles. The size of particles may also control the shape and the defect structure of the surface of particles because when only a small quantity of ions with a normal co-ordination number is present in a particle then only the most stable planes are likely to be de~eloped.~ EXPERIMENTAL The aerosol particles are obtained by decomposition of anhydrous metallic chloride vapour in the hydrogen-oxygen flame of a diffusion multitubular burner.The experimental 57 INORGANIC OXIDE AEROSOLS device has been described in dctail.2* The flow rate of reacting gases the concentration of the chloride in the feed and the temperature of the flame may be varied over a large range. For a given temperature of the flame (obtained by varying the Hz/02ratio) an increase of the concentration of the chloride vapour carried out into the burner by the oxygen feed increases the diameter of particles of the aerosol collected in a electrostatic precipitator. The particles are non-porous and practically monodispersed for each preparation. The photocatalytic properties at room temperature of the aerosols were studied in a differential reactor already described.6 A u.-v.source was used in some cases with a mono- chromator or filters such that the selected wavelength could pass through a silica window and irradiate the aerosol deposited on a porous film in the reactor in the form of a thin layer. The reaction products were analysed by gas chromatography. RESULTS AND DISCUSSION MORPHOLOGY OF AEROSOLS Particles of AI2O3?TiO, SiO, ZrO, Fe203 Crz03 V205,SnO and GeO, were prepared as required from corresponding volatile chlorides (or oxychlorides). The relationship between the concentration of the chloride in the feed and the diameter of particles for each temperature of the flame has been given else~here.~ In the present paper particular attention is attached to the relationship between the shape and the diameter of particles with the surface activity of TiO aerosols prepared in flames whose tempxature is in the range of 1500 to 3000 K.Typical flow rates of gases into the burner for a flame at 3000K are of the order of 8 x mol/s for hydrogen 4x mol/s for oxygen I .2 x mol/s for nitrogen. For a cold flame (e.g. 1500 K) a proportion of the hydrogen is substituted by nitrogen. Titanium tetrachloride vapour flow rate may vary over the range of to 5 x mol/s. The residence time of the particles in the flame which depends on the flow rate of the carrier gas and the cross-sections of the burner tubes may vary between 0.3 x lo- and 15 x lo- s per 1 cm length of the flame. It is assumed that the rate of transformation of the chloride vapour in the flame into the oxide is much higher then the rate of growth of the oxide droplets or particles.Consequently the residence time of “ initial ” molecules of the oxide in an elementary volume of the flame depends on the flow rate of reagents into the burner. On the other hand the concentration of “ initial ” molecules of the oxide or its vapour pres- sure is determined by the concentration of metallic chloride vapour in the carrier (oxygen) gas. An attempt is then made to determine the conditions of the formation in the flame of a particle of the oxide in connection with the concentration of “ initial ” molecules of the oxide and their residence time in the flame. The electron micrograph of fig. la shows a titania aerosol obtained in a cold flame (1700 K) whereas fig.lb shows the aerosol obtained in a hot flame (3000 K). The overall rate of flow of feed gases (H, O, N,) was 1.3 x lo- mol/s for both prepara- tions (residence time 0.05 s/cm) and the flow of titanium tetrachloride was also identical at the low rate of 0.4~ lo-’ mol/s for both aerosols. These results show that for any flame temperature and for a low concentration of reacting species the aerosol particles present have dimensions below 200A and their shape is that of a polyhedral type exhibiting facets. In this range of concentration of TiC14 the influence of the residence time is negligible. When the titanium tetrachloride flow rate is increased almost 100 times (25 to 30 x mol/s) fig. lc (1700 K) and ld(3000 K) show a remarkable difference in the morphology of particles.For the cold flame (lc) the shape of aerosol particles is of the same type as previously shown (la) though their diameter is now increased to FIG. la-Aerosol of titania prepared in a FIG.1h.-Aerosol of titania prepared in a hot cold flame (1700 K). TiCI flow rate 0.4 x flame (3000 K). TiCi flow rate 0.4 x 10-5 mol s-I. Ill01 5-I. Fw. Ice.-Aerosol of titania prepared in a cold FIG. I(/.-Aerosol of titania prepared in a hot flame (1700 K). TiCll flow rate 3Ox flame (3000 K). TiCI flow rate 25 x 10 ' niol s-I. mol s '. ((1) (h) FIG.2-(tr) and (0) Formation of spherical particles of titania from polyhedral particles in a flame of intermediate temperature (2100 K) TiCI flow rate 25 x mol s-l.JUILLET LCCOM TE MOZZANEGA,TEICHNER THEVENET VERGNON 59 360-400A. The variation of the residence time only modifies the size of particles and not their morphology. For the hot flame (Id) the particles now present a perfectly spherical shape of a diameter of the order of 1500 A. It is supposed that this spherical morphology results from the condensation of " initial " molecules of Ti02 into liquid droplets (melting point of Ti02 = 2200 K) which after cooling and quenching give solid particles with the initial shape of droplets. In this latter case the proportion of spherical particles in the aerosol increases when the residence time increases. The question now arises why in the case of a low concentration of titanium tetrachloride and hence of "initial " molecules of Ti02 (fig.lb) the liquid droplets of a smaller diameter are not formed in the hot flame. The polyhedral shape of the particles seems indeed to indicate that condensation of "initial " molecules of Ti02 proceed directly into a solid state in the same manner as a for a cold flame (fig. la) below the melting point of Ti02. Because the flame reactor enables one continuously to vary the temperature of the flame and the residence time in the flame it was possible to set the boundary conditions between the spheres and the polyhedral particles. Fig. 2 shows the micrographs from which the mechanism of the formation of a spherical particle may be deduced. It must also be recalled that spherical particles of diameter smaller than 300& have never been observed for any flame temperat~re,~ which seems to show that the liquid state cannot be formed below some critical diameter of particles.Fig. 2a and 2b seem to indicate moreover that the liquid droplet is not obtained directly from the con- densation of" initial "molecules of TiOz but only by the melting of a group or cluster of small solid polyhedral particles initially condensed. For an intermediate flame temperature (2100K) and a high flow rate of titanium tetrachloride (30x mol/s) a sufficiently high concentration of small (polyhedral) particles is present in the flame to allow the formation of aggregates of particles which at this temperature will just be able to melt-resulting in a spherical cluster. This behaviour should be correlated with two observations (i) the vapour pressure increases when the radius of particles decreases and (ii) the vapour pressure in equilibrium with the condensed phase is smaller for solid than for liquid hence a critical radius of curvature may exist below which only the solid phase is stable.The crystalline structure of titanium dioxide seems also to depend on the dimen- sions and shape of particles. In cold flames for polyhedral particles anatase is principally formed. However for hot flames polyhedric particles may contain up to 30 % of rutile whereas spherical particles contain almost 100 % of anatase. It is therefore not surprising that particles of different morphology and structure exhibit different catalytic properties and also photo-catalytic properties as shown in the next section.PHOTOCATALYTIC OXIDATION IN THE PRESENCE OF SOME OXIDE AEROSOLS It has been already shown that alumina titania or zirconia aerosols (diameter of particles below 300 A)may be reduced on their surface in vacuum at 500°Cgiving non- stoichiometric oxides." * For titania this reduction may be achieved at room temperature in vacuum if the solid is simultaneously irradiated in u.-v. (2000-3600 Moreover titania aerosols exhibit at room temperature photo-catalytic behaviour in the partial and/or total oxidation of hydrocarbon^.^ For this reason a study of their behaviour in the photocatalytic oxidation of inorganic molecules was also undertaken. It must be recalled that the photocatalytic activity e.g. in the oxidation of iso- butane into acetone expressed as a number of micromoles of acetone formed per INORGANIC OXIDE AEROSOLS minute per gram of catalyst spread out on the porous support in the differential reactor is a linear function of the weight of the catalyst up to some critical limit.It has been suggested that this behaviour is related to the surface nature of theprocess and to the need for the u.-v. radiation which is unable to reach the catalyst particles at the bottom of the bed if the thickness of the bed exceeds some critical limit. Further-more all the tests of photocatalytic activity were performed with the mass of the aerosol not exceeding the critical mass. In a typical test of CO oxidation the composition of reacting feed was 12.5 % of 02,25 OJO of CO in 62.5 % of He as a carrier gas with a flow rate of 1 L/h onto 10 to 33 mg of aerosol uniformly deposited on a porous support (fiberglass) in the reactor.Table 1 gives the results of the photocatalytic oxidation of CO at room temperature onto titania aerosols of various surface areas prepared in the flame reactor. TABLE ACTIVITY OF TiOz AEROSOLS 1.-PYOTOCATALYTIC surface area/ m*g-1 morphology total mass in the bed/ mg conversion % activity a pmol COz min-1 tn-2 activity/pmol COz min-1 9-1 temperature o the flame/K st NCt me % rutilc 23.5 32.5 41.O 68.O 70.0 98.0 140.0 spheres spheres spherespoiyhedr. polyhedr. polyhedr. pol yhedr. 20 35 13 19 13 15 10 0.71 2.39 1.51 1.48 1.36 1.73 1.73 72 118 202 135 181 200 350 3.20 3.70 4.90 1.99 2.56 2.04 2.49 3000 3000 2700 1900 1900 1700 1700 a All experiments are performed with a constant intensity of u.-v.beam If the photocatalytic qctivity in micromoles of CO per min and per m2Qmol min-' m-2) is plotted as a function of the surface area (fig. 3) two distinct plots are observed for spheres and polyhedral particles. The activjty per unit surface should be t 6- 5-CI 4--E IC .-f 3-I 0 A b \ 2-I A A I I I I I-I I I I c * 0 50 ! GO I50 S/m2 g-' FIG.3.-Photocatalytic activity for CO oxidation of spherical and polyhedral aerosols. 0,spheres ; A polyhedra JUILLET LECOMTE MOZZANEGA TEICHNER THEVENET VERGNON 61 independent of the extent of the surface if the quality of this surface in catalysis remains constant.This is the case for polyhedral particles whereas for spherical titania the quality of the surface in photocatalysis seems to increase with the specific surface area i.e.,when the particle size decreases. It is difficult to ascribe this behav- iour to a different rutik content (table 1) of aerosols because their surface content is not known. However if the photocatalytic activity depends on surface defect structure (point defects),10 the polyhedral particles have some chance to exhibit the same surface concentration of defects whereas for spherical (melted) particles the organization of the surface may be more difficult to achieve when the dimensions of the particles (for small particles of higher surface area) are close to the crystallo- graphic distances in various planes of the lattice.As previously observed in the photocatalytic oxidation of hydrocarbons,6 titania (anatase) obtained in a conventional manner by hydrolysis of TiCI4 does not exhibit any activity in the oxidation of CO. It must be recalled that this sample is porous and therefore not convenient for a surface photo process. TABLE 2.-PHOTOCATALYTIC ACTIVITY IN THE OXIDATION OF co OF VARIOUS AEROSOLS activity E//imol C@ nature surface area/ml g-1 min-1m-2 70 0.46 37 0.016 220 0 39 0 100 0.017 54 0.03 34 0 13 traces 2 0.45 aThe conditions of irradiation were different from those used for data in table 1 (decreased intensity of u.-v. radiation). Among all the aerosols prepared in the flame reactor titania exhibits the highest photocatalytic activity in the oxidation of CO.Table 2 gives the comparative values of the activity for some aerosols for which the morphological study was not under- taken. The same oxides prepared in a conventional way by precipitation in aqueous media and calcining do not exhibit any measurable photocatalytic activity. In contrast with the photocatalytic oxidation of hydrocarbons where only titania aerosols were active various oxides exhibit some not negligible activity in the CO oxidation. Photo-oxidation of other inorganic substances was mainly studied on polyhedral titania (70 m2 g). A mixture of ammonia (20 %) oxygen (40 %) and helium (40 %) was passed with a flow rate of 1.2 I./h through the differential photoreactor and a conversion of 5 "/o was registered.The reaction products are N (85 %) and N20 (15 %) apart from water. Nitrous oxide is neither photo-oxidized in the same conditions nor can it be used as a source of oxygen in the photo-oxidation of hydrocarbons or CO. Blyholder and coworkers l2 have however observed a phot o-oxidation of CO by nitrous oxide in the presence of ZnO obtained by a conventional method. But this catalyst is also able to oxidize CO with N20 in a thermal process at low temperature. Finally hydrogen sulphide in a mixture of H2S (20 %) O2(30 %) and He (50 %I INORGANIC OXIDE AEROSOLS with a flow rate of 1.2 l./h was photo-oxidized on titania (70 m2/g) with a conversion of 6 %. In the exhaust gases sulphur dioxide and water vapour were identified but sulphur was deposited simultaneously onto the catalyst bed.Experiments to determine the activity of aerosols other than titania in the photo- oxidation of NH3 and H2S are still to be attempted. However it may be already concluded that the state (dimensions morphology) of the oxide particles in industrial smokes is of paramount importance in their surface activity. In conclusion it has been shown that it is not possible to extrapolate and compare data obtained for less divided oxides or for oxides prepared in a conventional way (mainly in aqueous media) so far as their photocatalytic activity is concerned. R. Caillat J. P. Cuer J. Elston F. Juillet R. Pointud M. Prettre and S. J. Teichner Bull.SOC. Cltim. France 1959 152. J. Long and S. J. Teichner Rev. Int. Hautes Temp. Rifract. 1965 2 47. J. Herrmann S. J. Teichner and P. Vergnon J. Catalysis to be published. R. Van Hardeveld and F. Hartog Surface Sci. 1969 15 189. M. Formenti F. Juillet P. Meriaudeau S. J. Teichner and P. Vergnon in Aerosols and Afmos- spheric Chemistry ed. G. M. Hidy (Academic Press N.Y. 1972) p. 45. M. Formenti F. Juillet P. Meriaudeau and S. J. Teichner Chem. Techn. 1971 1 680 and 5th Intern. Congr. Catalysis (Palm Beach 1972). 'B. Arghiropoulos J. Elston P. Hilaire F. Juillet and S. J. Teichner in Reacticity of Solids ed. J. H. de Boer (Elsevier Pub. Company Amsterdam 1961) p. 525. J. Long F. Juillet and S. J. Teichner Rev. Int. Hautes Temp. Rkfract. 1965 2 163. M. Formenti H. Courbon F. Juillet A. Lissatchenko J. R. Martin P. Meriaudeau and S. J. Teichner J. Vac. Sci. Techn. 1972 9 947. lo Ph. Roussel and S. J. Teichner Catalysis Review 1972 6 133. 0. M. Poltorak V. S. Boroninet and A. N. Mitrofanova Proc. 4th Intern. Congr. Catalysis Moscow 1968 ed. J. W. Hightower (Houston 1970) p. 1235. Ken-lchi Tanaka and G. Blyholder J. Chem. SOC.D 1971 14 736.