Condensation and Evaporation of Metallic Aerosols BY E. R.BUCKLE AND K. C. POINTON Department of Metallurgy The University Sheffield S1 3JD Received 22nd January,1973 The heat-pulse cloud chamber has been used to study the condensation of metallic aerosols in the presence of purified argon. Multiplication growth and evaporation of Ca Cd Pb and Zn particles vary with the background temperature in the chamber. The volatile Cd and Zn resemble the alkali halides in that growth occurs readily in suspension when a sufficient vapour pressure is maintained and particles are formed that settle out at appreciable speeds. These particles fall from the cloud independently. With Pb the vapour pressure in the vicinity of the melting point is much lower and nucleation in the vapour at the high temperatures close to its point of generation is followed by the rapid arrest of growth and evaporation as the particles move away into the chamber.This results in the freezing- in of large numbers of minute particles and a smoke is formed in which there is little evidence of further change. When the temperature of the chamber is reduced to room temperature the particles are exceedingly fine and numerous when first condensed but the smoke thins out apparently by agglomeration. Observable motion in the smoke apart from the Brownian motion is dependent on convection in the supporting gas ; the particles move by streaming and do not fall out. The present cloud chamber is modelled on the original design of Buckle and Ubbelohde,' with certain adaptations necessary for its use with metals and for the general improvement of operation.A substantial advantage is obtained by the use of probes which obviate the need to dismantle the chamber for sampling the fall-out and renewing the metal supply. The technique is basically as before and involves the repeated production of aerosols as the temperature of the background is slowly varied. Highly supersaturated vapour is produced by passing current through a coil in a supersaturator probe so as to flash-heat the metal sample above the back- ground temperature. After formation the vapour rapidly cools to form a suspension of droplets or solid particles by condensation. The suspension is viewed telescopically under intense illumination and the behaviour of the particles observed.Other probes are used to sample the fall-out and control the motion of the aerosol. EXPERIMENTAL DESIGN OF CLOUD CHAMBER The chamber is assembled inside a horizontal tube furnace with a Pt-Rh winding. This provides a steady background temperature controllable to 1 K between 400 and 1800 K. The chamber is a muffle of refractory alumina 76 cm long and of 52 mm bore extended at one end via a water-cooled brass head by a Pyrex manipulation section (fig. la 2). The free ends of the extended muffle terminate in brass heads each of which is fitted with a window for viewing the interior of the chamber and two probe carriers. The head mounted on the alumina muffle is also water-cooled. Nine recrystallized alumina crucibles are assembled end-to-end in the muffle dividing tke chamber into a further nine sections and reducing heat loss from the interior.The middle section functions as the generating chamber and com- municates with the heads by means of two Vitreosil pipes passing through axial holes drilled in the floors of the alumina crucibles. Additional holes carry alignment rods and probes. 78 BY E. R. BUCKLE AND K. C. POINTON I' ? f Q (6) FIG. 1.-(a) longitudinal section of cloud chamber ; (6) radial section of generating chamber a supersaturator probe; 6 substrate probe; c exhaust probe; d thermocouple; e chamber window ;f water cooling ; g sample feeder ; h observation window ; i muffle containing crucibles ; j glass section ; k generating chamber ; I viewing pipe ; myprobe recess ; n alignment hole.FIG.2.-Glass manipulation section. 0,supersaturator access turret ; p substrate access turret. CONDENSATION AND EVAPORATION The central crucible is also modified by reducing the bore to match the pipe-section with alumina cement leaving grooves in which the probes are recessed. This construction avoids the formation of turbulent eddies in the aerosols. The central section is observed and illuminated through the pipes which are coated internally with carbon from a sooty flame to reduce their reflectivity. The temperature is probed in the generating section by means of a sheathed Pt/Pt-Rh thermocouple. Gas inlets controlled by needle valves are connected to the end-heads and allow oxygen- and water-free argon to be passed into the chamber from either end after the initial evacuation of air.An exhaust probe extends into the central chamber for local pumping during observations on aerosols. The design of an efficient supersaturating device has involved considerable experiment- ation. A simple design consists of a twin-bore alumina tube carrying leads of 1 mm thick Ni welded to a small heating coil of tungsten. The coil is supplied with current from a Variac. To minimize its effect on the chamber temperature the power dissipated in the coil must be sufficiently low in comparison with the average power input to the furnace windings. At the same time to avoid the shorting of the coil and its subsequent failure it is necessary to protect it from the test metal which otherwise spreads along the coil when molten.A satisfactory design is shown in fig. 3a. The coil is of Mo tightly wound and closely spaced and supported on a former of alumina. It fits into the lower bore of a piece of twin-bore alumina tubing and the upper bore is exposed over the middle 1 cm of its length by grinding to form a slot. The ends of this bore are sealed with alumina cement. The metal sample is held in the slot and the whole assembly attached to the probe after welding the coil to the leads by a tightly fitting sleeve of Pt-Rh. (b) FIG.3.-(a) supersaturator probe; (b) substrate probe (exploded view). q Mo coil; Y Pt-Rh sleeve; s Nitrile rubber gasket ; t insulating compression disc ; u Ni lead ; u thermocouple ; w,coolant gas circuit ; x Pt-Rh connecting sleeve not shown.The substrate probe (fig. 3b) consists of ashort piece of alumina thermocouple sheathing on which a flat surface has been cut. This is connected to a 4-bore capillary probe again with a Pt-Rh sleeve to provide rigid support at high temperatures. A thermocouple in contact with the under surface of the flat gives the temperature of the upper surface with reasonable accuracy. The substrate can be cooled below the chamber temperature by a gas stream conducted via the other two bores of the probe. OPTICAL TECHNIQUE The viewing technique is the same as before.' A large converging lens is used to collect as much light as possible from the diffuse source of a 250-W Hg arc-lamp and to focus it to BY E.B. BUCKLE AND K. C. POINTON form a secondary light source of about 2 ,nun diam on an iris diaphrap. The illuminated aperture is focussed with a converging doublet of short focal length on to ;t second diaphragm consisting of two blades with V-notches enabling the aperture to be narrowed down to give an extremely small tertiary point-source. A final doublet of weak convergence collects the light from this source into a very narrow beam which enters the cloud chamber after passing through a filter to select the 5461 A mercury line. The use of silica windows to isolate the central section was abandoned because of the obscuring effect of metallic condensate strongly illuminated by the incoming light beam. Confinement of clouds to the generating aection was achieved by a new technique of gas flow described below.The advantage to visibility was substantial because the Airy patterns of the cloud particles could be viewed with the telescope directed at a lower angle to the light beam. PROCEDURE To prepare for a condensation run the furnace is set to heat the chamber to a steady temperature about 300 K below the melting point of the metal (table 1). The chamber is evacuated to a pressure between 1 and 10 N m-2 and filled with purified argon to atmospheric pressure. The argon stream is continued during the loading of the supersaturator probe. The slot of the supersaturator is positioned in the manipulation section directly below the vertical turret. Test metal in the form of wire is then fed through the seal in the turret to the heated probe until the slot is full of molten metal.The probe is then pushed into thegener-ating chamber and the gas flow discontinued leaving the whole cloud chamber under a slightly positive pressure. TABLE 1 .-VAPOUR PRESSURES OF THE METALS 4 AT SIGNLFICANTTEMPERATURES (Tf = m.p. Tb = b.p.) metal 0.8 Tt/K TdK ~'(0.8T;)/Nm-2 t Ca 893 1756 7 Zn 554 1180 8x Cd 475 1038 6x Pb 480 2020 3 x lo-' Al 746 2740 3 x lo-" t value for liquid extrapolated from Tf Using the Variac the metal is flash-heated to several 100 K above the background temp- erature until a suspension of condensed particles appears. The particles show a tendency fo move out of the central chamber into the viewing pipes where the temperature is unknown.This is prevented by a slow flow of argon along the pipes. The gas enters at the inlets at each end and is drawn into the central section by the exhaust probe. By careful setting of the needle valves the gas flow may be tuned and the metallic vapour and suspended particles caused to circulate slowly in the central section. Particles can be held almost stationary for periods of up to 30 s depending on the volatility if they occupy positions toward the centre of the rotating cloud. A temperature scanning procedure ' is used to establish the properties of the metallic aerosol that depend on the growth evaporation and physical state of the particles in them. RESULTS Before commencing work on the metals the performance of the apparatus and in particular the new design of supersaturator was tested on gne of the salts studied by Buckle and Ubbelohde.' KI was chosen as a representative salt with a suitable vapour pressure curve and clear-cut cloud phenomena.The critical solidification temperature Tswas reported as 799 K. As the temperature of the cloud chamber was raised from 770 to 800 K marked changes in the behaviour of the KI clouds were seen. Dense clouds of tiny particles CONDENSATION AND EVAPORATION which responded slowly to changes in the supersaturator current gave way at higher temperatures to clouds in which the particles evaporated or grew rapidly as the supply of vapour was varied. The particles in clouds showed twinkling when the background was at 795 K but the effect had vanished when the temperature reached 797 K.T therefore lies between these values as judged from the readings on a thermocouple recessed in the wall of the open-ended generating chamber. The small discrepancy with the value previously reported was not considered to be serious enough to warrant further refinements to the method at this stage and work on the metals was begun after expelling the residual salt under vacuum at high temperature. The correlation T,-0.8 Tf was used as a guide in the search for critical effects in metallic clouds.2* GENERAL OBSERVATIONS ON METALLIC AEROSOLS The work so far has been restricted to telescopic observations. The Airy patterns formed by the particles are similar to those seen with salt aerosols the brightness of the central disc and the number of concentric haloes indicating the relative size.Depending on conditions particles appear thinly in small numbers when the size is large and densely in large numbers when the size is small. Large particles that grow quickly are lost by sedimentation. Such particles may be seen to move independently. When small particles occur in more or less per- sistent streams or " curtains "? and show the eclipsing effect ' as well as Brownian motion within the curtain. The curtains move by streaming in various ways depend- ing on the pattern of convection in the chamber. When the gas flow is suitably tuned the appearance of the curtain in motion is suggestive of the rotation of a stellar nebula. Such conditions of motion are optimum for the observation of single particles in a cloud which is not too dense.The effect known as " twinkling " has not been observed to be a general property of metallic aerosols formed by condensation. With Zn and also with Cd the vast numbers of particles that are formed at temperatures below 0.8 Tf and which may be attributed with confidence to condensation show an indistinct flicker but it has not been possible to detect a sudden onset of this effect. Large twinkling particles are sometimes seen in small numbers during the early stages of formation of a cloud. It is believed that these are ejected along with the vapour as the sample on the super- saturator becomes finally molten. The ejection is not visible with the present design the probe being recessed in the roof but it was clearly observed in the work on salts.Such particles drop out very quickly and may be distinguished from the condensate if turbulence is avoided. A shower of particles has also been observed after tapping the pro be. ZINC At temperatures in the vicinity of 0.8 Tf (table l) large particles are easily grown in large numbers. The aerosol motion is controllable by gas-flow tuning and the lifetimes are substantial (at least 60 s). These particles flicker when seen in isolation. At 500 K the numbers are greater and at 475 K there are small as well as large particles present. The proportion of small particles increases on further cooling and at 390 K stable dense suspensions of minute particles are formed. CADMIUM From 0.8 Tf down to 430 K the aerosols are densely populated with particles that fall out rapidly (in a few seconds).Growth is difficult to induce and the continued BY E. R. BUCKLE AND K. C. POINTON operation of the supersaturator merely produces more particles. At the same time the particles appear to diminish in size by evaporation. Flickering is also observed with this metal. There is no change down to 400 K beyond an increase in number density and a decrease in the size of the particles. With the chamber at room temperature the particles formed by condensation are exceedingly faint but stable. Quantities of larger particles are also formed that possibly originate from unmelted Cd expelled from the probe. These show unusual behaviour in apparently shrinking in size as they fall directly and rapidly towards the floor of the chamber.Similar properties are shown by the particles which fall when the probe is tapped and as the size diminishes so does the speed of descent. There is apparently a connexion between these effects and the presence of residual metallic vapour. If the chamber is flushed with argon and probe particles again dislodged without passing current the effects are not observed. LEAD Aerosols of Pb behave differently from those of Zn and Cd. At 0.8 T' ejected particles appear first then curtains of minute particles formed by condensation. The fine particles are very persistent but do not grow in the vapour. Even above the melting point (601 K) growth is too slow to relieve the supersaturation when the probe is kept hot.Instead the number of particles increases. As the temperature is lowered from 0.8 Tfthe concentration of particles formed in a cloud is increased and the particle size is decreased. The clouds are also less persistent. At room temperature a smoke is formed in which the Airy discs are initially barely visible. The smoke slowly thins out and the particles that remain become brighter indicating growth. Brownian motion continues and there is no loss by sedimentation. The impression is that the process of enlargement is visible at room temperature because of the high density of the initial smoke whereas at higher temperatures the process is still operative but the brightening of the Airy disc cannot be discerned. It is difficult to compare by eye the brightness of the discs when the particles are thinly dispersed.CALCIUM The behaviour of Ca in the chamber also has unique features. A complication is the low vapour density. This leads to the stratification of the particles and inhibits circulation. The same tendency possibly accounts for their persistence at high temperatures when they might be expected to evaporate more readily. The data of table 1 suggest that Ca should behave as a volatile metal like Zn and Cd but sub-stantial growth of the particles could not be induced even at 0.8 Tf. Another problem was reaction of Ca with the alumina of the probe. This inter- fered as the chamber temperature approached T, and when the melting point of Ca was reached (1 116 K) the reaction became self-sustaining and generated curtains of condensate even when the current was off.It is possible that at these high temper- atures A1 is vaporized and condenses along with the Ca. It would be expected (table 1) that pure A1 vapour would condense only to minute particles. DISCUSSION In the work on salts 6* it was established that the cloud lifetimes always tended to decrease as the background temperature was increased. The rise of the temperature through the twinkling threshold T could in many cases be correlated CONDENSATION AND EVAPORATION with a sharp fall in the lifetime. The effect was attributed to the increased evapora- tion of particles which remain liquid throughout the period of observation. It was also observed with salts that when the chamber temperature was much lower than T the clouds formed were persistent and composed of multitudes of minute particles.This may be explained as follows. Assuming that the test material is always heated to the boiling point by the supersaturator the saturation ratio p/p" where po is the vapour pressure at the chamber temperature T can approach very high values whan Tis low (see e.g, table 1). The result is a high concentration of nuclei which have little prospect of growth. From the few results we have obtained so far it would appear that there is an essential difference between the properties of aerosols of metals and salts. If the metal is involatile at the melting point (Pb; table 1) the growth-rate of particles even when liquid is so slow that at high temperatures one merely generates increasing numbers of them without effecting much enlargement.At low temperatures (Pb at room temperature) the number density of particles is much greater so great in fact that even in the first faint smoke agglomeration takes place. It is tentatively pro- posed that it is this that leads to the brightening of the images observed through the telescope. This interpretation will be tested by examination of fall-out. It would be in keeping with microscopical observations on metallic condensate sampled from various other sources such as exploding wires.' Particle aggregation in the fall-out from fine smokes has not been observed with the halides of the metals,'* 6* but it has with oxides,8 which again are often relatively involatile compounds.On theoretical ground^,^ coltision leading to fusion between particles in a volatile aerosol is a rare event in comparison with growth. As defined in this way therefore coagulation should not contribute to the relief of supersaturation by providing a short cut to the aggregation of molecules. It was also argued that under uniform conditions of supersaturation growth should be severely limited. If this conclusion is valid the observed formation of micron-size particles in metallic aerosols is to be attributed to their nucleation and growth under conditions of steep temperature and concentration gradients near the supersaturator. The possibility that they are heterogeneously nucleated on foreign particles already of appreciable size is unlikely if these do not also originate at the supersaturator.We are grateful to the Science Research Council for support including a mainten-ance award to K. C. P. E. R. Buckle and A. R. Ubbelohde Proc. Roy. Suc. A 1960,259 325. 'D. Turnbull and R. E. C& J. Appl. Phys. 1950 21,804. E. R. Buckle Nature 1960 186 875. 0. Kubaschewski E. L1. Evans and C. B. Alcock Metalhrgicaf Thermochemistry (Pergamon Oxford 4th ed. 1967). J. F. Elliott and M. Gleiser Thermochemistry fur Sfeelmaking (Addison-Wesley Reading Mass. 1960) vol. 1. E. R. Buckle and C. N. Hooker Trans.CFaraday Suc. 1962,58 1939. 'E. R. Buckle Condensation and Euworotion ofSofids,ed. E. Rutner et af. (Gordon and Breach New York 1964) p. 537. J. Harvey H. 1. Matthews and H. Wilman Discuss. Faraday SOC.,1960 30 113. E. R. Buckle this Discussion.