standard LMIS theory. However, discussions of the LMIS make use of the concepts of a &&viscous-drag-free source''(for which the cone base pressure is zero), and &&viscous-drag-limited'' sources (for which the cone base pressure isnon-zero).

Liquid}metal ion source operation has been modelled, both analytically and numerically. Progress in modellingthe steady-state-#ow mode has been substantial and successful. Attempts to model time-dependent EHD behav-iour have run into fundamental computational di$culties.

REFERENCES

Forbes, R. G. (1997) Understanding how the liquid}metal ion source works. <acuum 48, 85.

CLASSIFICATION OF THE MODES OF EHD SPRAYING

A. Jaworek and A. Krupa

Institute of Fluid Flow Machinery, Polish Academy of Sciences, P.O. Box 621, 80-952 Gdansk, Poland

Di!erent modes of electrohydrodynamic (EHD) spraying and their classi"cation according to geometric criteria,based on the forms of meniscus and jet, are discussed. The reason for interest in this classi"cation is that the dropsizes and geometrical forms of the aerosol, particularly important from a practical point of view, are di!erent fordi!erent modes of spraying. It is well known that the jet can disintegrate into droplets, while issuing froma capillary maintained at high potential, in many di!erent ways. Several attempts, based on di!erent criteria, havebeen undertaken to classify the modes of EHD spraying, however, without referring to a clear de"nition. Toclassify the modes of EHD spraying the following de"nition of the modes of spraying is proposed in the paper: Themode of electrohydrodynamic spraying is the way the liquid is dispersed into droplets, and is characterized by twocriteria:

(1) The geometrical form of the liquid at the outlet of the capillary (drop, spindle, jet),(2) The mechanism of the disintegration of the jet into droplets (type of instability).

In general, the spraying modes can be divided into two groups. The "rst group comprises the dripping,microdripping, spindle, multi-spindle and rami"ed-meniscus modes, i.e. the modes in which only fragments ofliquid are ejected from the capillary. The second group includes: the cone-jet, precession, oscillating-jet, multijetand rami"ed-jet modes, which are characteristic in that the liquid issues a capillary in the form of a long continuousjet which disintegrates into droplets after some distance, usually a few mm, from the outlet of the capillary. The jet,and the meniscus, can be stable, can vibrate, rotate spirally around the capillary axis or whip irregularly.

Usually, the spraying at negative polarity di!ers from that at positive for the same voltage value. At positivepolarity the droplets are usually "ner. It is also easier to establish the regular modes of spraying when positiveexcitation voltage is applied to the capillary.

The characteristics of some of the modes based on the experimental studies are as follows: (1) In themicrodripping mode the generated droplets are usually monodisperse and have the diameters ranging from a fewhundred of micrometers down to a fraction of micrometers. The stream of droplets is linear, and they #ow alongthe capillary axis. (2) In the spindle mode only elongated fragments of liquid are ejected from the capillary insteadof regular drops. (3) In the multi-spindle mode two or more spindles are ejected periodically at the circumference ofthe capillary. Next, the spindles can disintegrate into a few smaller droplets. The droplets are polydisperse and aresmaller than 100mm in diameter. (4) The oscillating-jet mode has been de"ned as the one in which the cone and thejet oscillate in the plane of the capillary axis. This mode allows to generate polydisperse aerosol of the dropletssmaller than a few hundreds of mm in diameter. The aerosol is sprayed in a plane or in an oblate cone of a #atellipsoidal base. (5) In the precession mode the jet rotates around the capillary axis. The generated droplets are ofa few tens of mm in diameter. The droplets are sprayed in a regular cone of circular base. This mode seems to bevery promising in practical applications. (6) In the cone-jet mode the jet is straight linear, and the droplets aregenerated due to varicose or kink instabilities. The droplets are a few tens of mm or smaller. (7) The multijet modeis generated when two or more jets "ner than a few tens of micrometers in diameter issue simultaneously from thecapillary at its circumference. The aerosol is "ne and is sprayed in distinct streams of droplets.

ELECTROSPRAYING OF GLASSES*PREPARATION OF GLASS COATINGSON GLASS

S. Rosenbaum and R. Clasen

Universitaet des Saarlandes, Lehrstuhl fuer Pulvertechnologie von Glas und Keramik, Postfach 151150,66041 Saarbruecken

Glass coatings on ceramic or metallic substrates are prepared by two di!erent methods. In a two-step processthe glass powder is deposited on the glass substrate by methods such as airless spraying of enamel or glazesuspensions, dipping in enamel suspensions, electrospraying or electrophoretic deposition of enamel or glazesuspensions, and electrostatic powder spraying. The particle size of the glass powders which are generally prepared

Abstracts 975

by milling, ranges from 500 nm to 100km. Therefore, these layers of glass powder have to be densi"ed by melting ina second step. As for glass substrates this processing temperature should not exceed 500}7003C to prevent bendingof the substrate, low melting glass powders have to be applied in this case. Alternatively, the glass powders can beheated to melting temperatures while spraying (e.g. #ame or plasma spraying) to reduce the thermal load of thesubstrate (one-step process). Unfortunately, the molten glass particles freeze very quickly when they hit on the coldsubstrate and the density of the deposited glass layer is lower compared to the two-step process due to trappedresidual pores. It is the objective of our work, to improve the coating methods of glass by combining these di!erentmethods and to reduce the processing temperatures by applying nanosized glass powders. Due to the high surfaceactivity, these powders can be sintered to transparent glass below the critical crystallization temperature.Therefore, the processing temperature can be signi"cantly reduced compared to the melting technique.

In this paper the eletrospraying of glass melts and glass suspensions is presented which are new methods forpreparing glass coatings on glass substrates. The process of electro-melt spraying is similar to the LMIS process,but instead of molten metal a glass melt is atomized under vacuum. The electrical conductivity and surface tensionof these glass melts are higher than that of water but lower than that of molten metal. The droplet diameter ofatomized glass melts lies between 100 nm and 10km. The molten droplet build up a dense coating on a hotsubstrate and are sintered in situ. In our case a #at glass substrate is mounted on a metal counter electrode. Due tothe high electric "elds which are needed for forming a Taylor cone of a glass melt, a vacuum of about 10~2mbar isnecessary to prevent electrical break-down. At this pressure most technical glasses show foaming. Thus, a new typeof crucible was developed, where the glass melt is under normal pressure and the capillary tip is under vacuum.Additionally, this crucible enables a continuous production of glass powders.

For the electrospraying of glass suspensions a new process was developed, the so-called electro-#ame spraying.In contrast to the classical electrospraying, a hydrogen}oxygen #ame is used as a counter electrode to build up theelectric "eld for creating the Taylor cone at the tip of a capillary tube. In this #ame the dispersing liquid of thesuspension is vaporized and the glass particles are molten. Due to the high surface charge, the molten glass dropssplit up in several smaller droplets which are deposited on a substrate forming a dense glass layer.

Finally, the synthesized glass powders and deposited glass coatings are characterized.

EXPERIMENTAL STUDY OF THE JET BREAK UP FOR EHDA OF LIQUIDSIN THE CONE-JET MODE

D. M. A. Camelot,* R. P. A. Hartman,* J. C. M. Marijnissen,* B. Scarlett* and D. Brunners

*Particle Technology group, Department of Chemical Technology & Material Science,Delft University of Technology, Julianalaan 136, 2628 BL, Delft, The Netherlands

s Institute of Process Engineering, ETH-Federal Institute of Technology, Zurich, Switzerland

The scaling laws developed by FernaH ndez de la Mora (1994, J. Fluid Mech. 260, 155}184) and by Gan8 aH n-Calvo(1994, J. Aerosol Sci. 25S, S309}S310) have been veri"ed since then through various experiments. Chen and Pui(1997, Aerosol Sci. ¹echnol. 27, 367}380) investigated the dependence of the current and droplet diameter with thepermittivity of the liquid. Hartman et al. (1997) developed a physical model to describe the spraying of liquids inthe cone-jet mode. These two approaches compare remarkably concerning the current produced by the cones.Nonetheless some di!erences have appeared concerning the jet break up. Namely, the scaling of the jet diameterwith the #ow rate and the conductivity is di!erent in both approaches.

Thus, we present here an original experimental study to investigate these di!erences. It was conducted usinga High-Speed Spray Imaging System (HSSIS) purchased from Oxford Laser. This system consists of a digitalcamera (KODAK) connected to a computer which is equipped with a frame grabber. A long-distance microscopiclens is "xed to a camera. The illumination of the subject is done by an infrared laser. A control box synchronizes thecamera and the laser. Pictures can be made with an illumination time down to 0.5 ms and the separation betweentwo following pictures can be down to 15ms. The optical system allows us to see objects down to a fewmicrometers but only objects bigger than 10 mm can be accurately measured. The experiments were done using "vedi!erent liquids namely, ethanol, butanol, isobutanol, 2-butanone and ethylene glycol. For each liquid theconductivity and the #ow rate were varied. For each situation photos were taken to determine the jet size, thebreak up of the jet, the droplet size and the droplet velocity. Moreover, for every situation the spray current wasmeasured.

Our results show that the jet diameter for the di!erent liquids studied exhibits a dependence on the #ow rate ata power &0.6. This indicates "rstly, that the model developed by Hartman is correct in its calculation of the jetdiameter and secondly, that the break up of the jet cannot be assimilated to the one of an unchanged jet.Nevertheless our results must be put in perspective with the fact that the jet diameter is measured at a "xed distancefrom the cone, while the jet length varies with the #ow rate. Further, our study brought some other interestingresults.

First, concerning the jet break-up mechanism. It is a known fact that within the range where IkQ1@2 the jet canbreak up due to varicose or kink instabilities. The results showed that for a given liquid the break up was goingfrom varicose to kink with the #ow rate going up. Second, they also showed that the number of satellites producedbetween two main droplets increases with the #ow rate.

Moreover, we have tried to de"ne, as a function of liquid properties, a limit which separates the two types of jetbreak up mechanism (varicose or kink). This is of primary importance because it is known that the size distributionof the main droplets is much narrower when the jet breaks up due to varicose instability.

976 Abstracts