ELSEVIER PowderTechnology86 (1996)41-67 Charge distribution and state of agglomeration after tribocharging fine p'miculate materials K. Sch~nert, K. Eichas, F. Niermfller Institarf~r AuJbereiarng, Univemtdt Clmrslhul-Zellerfefd.D~3867a Cfaasrkal-Zelterfeld. Germany Received21 March1995;revised19July 1995 Abstract Particle reixtarcsof different noncouductivc:"~ciesneedto be treated in a tribochargingunit so that theycanbe separatedin an electrostatic field. Such u unit producesan aerosolwith panicles earryingdifferentpositive and negativecharges. An efficientseparationdemandshigh charges; however,high chargessupportagglomerationand agglomeratesworsenthe separationprocess.This problemleasbeon studi~ using quartz/calcite mixturesin the size range 20-200 tira usinga vibrationchamberand a brushaerosolgeneratoras tM:trilx)dtatgingunit 11~ particle chargesvary over a wide range up to almost the theoretical maximumvalue, Tim charge distributiondepends on the particle size, mixture ratio arid atmospherichumidity such that the median value decteasosas the particles becomesmaller, the humidity rises and tim fraction in the mixtureincreases.Accordingto Coehn's mIe, quartz carries mainlynegativeand calcite mainly[msiti'cecbarg'eLElectnx~.tatic agglomeratescontainiag particlesabove 10 p,ra can be brokendownby the Con]ombicfarce in the field,but agglomerateswith tinct"particles have to be destroyed before enteringthe sepa~tor. The most effective methodis to use a high air velocity inside a slit nozzle. Even a feed with highlycharged mierofineparticlescould be deglomeratedcompletely. geywords: Electrostaticsgparatlon;Paniclecharging;Deglonmratioa 1. Introduction Particle mixtures of different noncnnducfive species need to be treated in a tribocharging unit so that they can be separated in an electrostatic field. The particle charges differ with resp*et to polarity and charge density. The particle stream is split into a positive and a negative product leaving the field near the electrodes. Also, so-calledmiddlings can be obtained that contain particles of low charge. The field is usually oriente,d horizontally and the particle stream is fed from the top. The particles should not meet the electrodes and be deposited on them, as otherwise sparking effects could be caused. The application of these free-fall separators is limited to particle sizes between 50-100/.Lm and several millimeters. The lower limit is due to agglomeration effects and particle deposition on the electrodes; the upper limit follows from an insufficient residence time owing to the fast seuling velocity, Big installations of free-full separators are successfully used for proeessiag salt minerals [ 1]. The first investigations on particle tribocharging were published by Szantho in 1939 [21, and extensive work was done by Carta and his group during the 1980s (see, for example. Refs, [31 and [4]). Extending the application of this separation method into the particle range below 100 #m requires the production of 0032-5910196/$15.00 © 1996EIs~vi*rScience$.A.Alltightsmsm'~ml $8DlO032-591 O(95 )03036-9 an aerosol with charged particles that feeds into a scpm'ator. A new type of such a separator, called a triboelectfic drum separator, was developed recently and invostigated [5,6]; see Fig. I. The species are separated in the gap I~tstama two rotating polycarbonate drums. Two imbedded electrodes build up a high dectrostatic field, The charged particles are deposited onto the drums, transported to the back and removed there from the drums by an electric force caused by oppositely charged back electrodes and bmsbes. A slightly modified brush aerosol generator [71 is used for tribucl~glag and is mounted above tim separator. Binary mixtur~ of quartz, calcite, flnorspar, barite and anthrucit, have successfully ~ e n separated in the size range l-t00 p,m. Recoveries and purifies above 90% could be achieved in one step. Detailed results are given in Refs. [5] and [6L To understand the laoccss and optimize the OlXa'ationconditions, knowledge of the charge distribution and the state of agglomeration is needed. 2. Charge distribution 2.1. Experimental The charge distribution of binary quartz/calcite mixtures of three narrow fractions between 20 and 200 ~m have been 42 if. Schiinerret at./Powder Technology ,~6 (1996) 41~47 6 ~ 8 ~i 4 < ), / 1i,.,( ) 5 9 5 Fig. I. Sketch of the tnbaelcctnc drum scparalor [51: (l) electrodes; 12.) rotating polycarl~nate drums; ( 3 ) back aiectmdes; (a ~tJru.shes;( 5 ) product chambers; (61 powder sample; (7) rotating brush; (8) air inlet; (9) air uniter with a fill. measured in two different devices. The first one. shown in Fig. 2, comprises a vibration chamber as the tribocharging unit, the usual free-fall separator and a particle collector with ten sections [ 8]. The top and bottom of the vibration chamber consist of metallic frits so that the inserted sample can be conditioned with respect to the moisture content and the temperature of the air. An opening, covered with a screen at the bottom, lets the particles flow into a 2 mm metallic capillary robe leading to the separator. The unit is grounded. The acceleration of the linear oscillation can be adjusted up to 20g. The particle collector is made of PTFE with ten small brass boxes conntected to a very sensitive electrometer (Keithley 617). After the experiment, the charge and the particle mass in each box are measured in order to calculate the charge distribution. Two methods have been used, in which the fractions of both species are either equal in size (method A) or different (method B). In the second case, the coarser fraction, called here lhe partner fraction, is kept back by the screen covering the opening, and the finer fraction, called the object fraction, settles into the separator. This device can be used down to a fineness of 4 0 - 8 0 / L m depending on the material. More details can be found in Rcf, 18]. The 20125 ~ m fraction was investigated by measuring the particle trajectories in a horizontal electrostatic field. Fig. 3 shows the arrangement. The particles are triboeharged in the brush aerosol generator ~hown in Fig. l. The aerosol jet enters a large cylinder with a diameter of 300 mm and a height of 300 ram, where the air flow is calmed. An opening of 0.9 mm in the bottom allows the particles in the center to enter the settling chamber with two electrodes. The trajectories are recorded with a video camera. As long as Stokes" law can be Fig. 2, Sketch of the electros'taticseparator for measuring particle charge distribution [g]:l I ) vertical vibrating chamber for uibocharging; (2) electrQdcs; t3) particle collector with ten se..:ti,~r,zl; (4) capifla6cs for feeding particles into the eluctrostaricfield; (~;) air inlet and outlet far conditioning the sample: (6) box with sensors for temperature, humidity and volnmetde flow. applied, the charge density or can be calculated from a, the angle with respect to thc vertical, and wr, thc settling velocity. according to Eq. ( 1 ): 7 $ 9 $ 5 Fig. 3. Sketch of the sedimentation chamber aid ~plica| an'angemeat for measuring particle charge dist~ibotioas [5 ]: ( I ) inlet for 11~ aerosol with the charged particles; (2) chamber for calming the air flow; (3) air outlet; (4) capillary far feeding particles into the ¢lecuoslalic field; (5) settling chara~r; (6) ¢lcclct',,tes;(7) light source; (8) lenses; (9) apertures; it0) vi¢~¢OCamors. K. Schllnert ¢t a! / PowderTechnMogy 86 (1996)41.-47 ~ [g~p~12(p~-pf) l'/~'[w~ r~ tan odE1 (i) where ~ is the known viscosity, p~ the particle density, pt the air density, E the field strength and g the gravitational acceleration. The species cannot be distinguished; therefore a mixture is inserted consisting of the object fraction 20/25 p,m and the partner fraction 40150 #m so that the species are characterized by the settling velocity. At least 120 particle trajectories were always evaluated in determining the charge distribution. For general considerations, the charge density should be normalized by a characteristic value o',. It seems reasonable m evaluate o'~ with the general equation relating charge density to field strength and introducing the breakdown field strength EB: o'~=c~En; eo=8.854X10-12AsV-im t (3) This value corresponds to about 160 elementary charges over 1 /.~m2 or to an average spacing of about 800 A in a square arrangement. The ratio of the actual charge density to o', is called the saturation grade ~. The samples were washed in deionized water, dried for about 10 h and cooled with dry air. The triobeharging unit was always conditioned with a flow of air at the required temperature and humidity before introducing the sample. The mixture ratio was varied and is characterized by the surface and mass ratios, which are related to each other by SIIS:= (x21x~)(~lp~) (MtlMD n 2 4 e e 10 12 14 !6 18 20 Fig 4. Median~omraaongrade ~ndoscil1~g accclcraion.Quart I~I 200 pm, t= i0 min,tel. humidity2%. S~IS~ = I. 50 ~ .... J~ (2) The only question is what value should be used for EB, which depends on the particle size and shape. In principle Ea increases as the particles become finer and decreases for nonspherical shapes. For the following, En is set equal to 3 X l 0~ V m - ~, which is approximately the value of the breakdown field in dry air between plates [9]. It then follows that OL:2.656× 10 -'~ A s m-" 45 (4) where x is the particle size, p the density, S the surface, and M the mass. 2,2, Results The particle charges are expected tu depend on the two species uf the sample, the mixture ratio, the particle size, the moisture, the material of the tfibocharging unit, and the mechanical action characterized by oscillating ~caleration, oscillating time and brush speed. The essential findings are discussed in the following section; more details are given in Refs. [5] and [g]. The charge distributions are expressed using either the distribution functions H(~) or the median values ~so. Quartz is always raainly negatively charged and calcite mainly positively, according to the well-known Coehn's rule, because the dielectric constants of quartz and calcite are reported in the ranges 2.6--6.5 and 7,6-8.9, respectively. o I 2 3 4 o~,g 5 6 7 t ~ t , mm 8 9 10 Fig. 5. Median mtu.mdon grade and oscillating lime. Qtzanz 1601200 pro, ~¢¢elcr~iort 12g,tel. humidity 2%, 5 ~ 1 ~ - I. The acceleration and time of oscillation influence the saturation grade as shown in Figs. 4 and 5, with the quartz fraction being 160/200 pro. For a constant oscillating time of I0 min the median value rises from 26% at 38 to 48% at 12g. Higher accelerations cause much abrasion of the calcite, and the saturation grade decreases. The charging occurs within a few minutes, and further increases a~ small. Most experiments, therefore, were done at 12g and with an oscillating dine of lO min. The brush speed does not show any influence between 500 and 1500 rpm. The corresponding circumferential velocities are 0.92 and 2.76 m s - I The influence of the material was studied using a brush aerosol generator with both the wall and the brush made of steel or polyamide. Either a quartz or a calcite fraction was fed; the results for both are shown in Fig. 6 in the form of histograms. Both species are negatively charged, but since palyamid¢ causes much higher charges than steel, steel was used for all experiments. This result also indicates that the charging could be optimized for a particular application by selecting favoumble materials for the walls arid the brush. A typical example of how the mixture ratio influences the charge distribution is sh~.wn in Fig. 7 for the size fractiou 1601200/an vibrated together with the partner fraction 4001 500 pm according to method B. The mixtureswere composed such that the surface ratio of the finer to the coarser fracdon was 1:1 or 9: L At a ratio of 9: l the particles arc less charged than at a ratio of t:l, because fewer panr~r particles are available for contacting. The median satmation grade increases from 36% to 44% for quartz and from 14% to 34% for calci~e. A very small f."action is oppositely charged with a 1ow charge density. The negative charges on calcite particles could be caused by contact with the metallic walls, as can be 44 K. Schihzer~ et at, / Powder Technology g6 (I 996) 41-4 7 ~'~151rttl'''h''l'''i(~Jar tZ (-) I+) ....... i ,++ _ AI _P ~I+I,S,I,ITTTr, + ~. . . . . . e~,,i,,, satarntloe,'/= Fig. 6+Chargedistribution with different matefialsfor the brush andchamber wall: left, steel; right, polyaraide,Qnarlznr calcitu 20[25 #m fractiontreated separately, tel. humidity 2%. i tlU~tZ 1-) t+l "1 N ,+ ]+!.,.Pp'~,. 1.......... ~ s 0calctie~ l I-YOI+1 ~o 40 20 o 2u 40 ao qutarlz (q (.) I .... I,HI~,, sat~nl~. t~) saMa~n, Fig. 8. Charge distribution of separately treated quartz or calcite 160/'200 ~m fraction, Accelerurion 12g. t - l0 rain, =eLhumidity 2%. 16 LI LJ=~h t= I,LLL~LL±~L etmrJtz (-) (+) quartz ( l t+l N ~s ,+ !!! ,re.m.+. ,~++~r.+. ~'~i~cn~clte i=11*1 no 40 20 o 20 4o 6o utl~ntJarL,~+ sal~nllon, Fig 7, Charge distribution of quartz/calcire mixture.~ofdifferentmiles::left, 3o+/Sr,m ~ 9; fight, So~/Srm. = 1. Object fraction 160/200 ,u.m,partner truelion 400/500/.tin, acceleration 12g, t - 10 rain, rot, humidity 2%, seen in Fig. 8, which shows the charge distribution of a separately vibrated quartz or calcite fraction. About 40% of the calcite particles arc negatively charged, The positive charges on quartz could be caused by local impurities on the surface or by adhesion of abraded calcite dust to the quartz surface. The increase in the oppositely charged fraction of quartz with increased calcile content, decreased particle size and at a high acceleration (over 12g) may be taken as an indication that the second-memioned reason is the more likely. Fig. 9 demonstrates the influence of the atmospheric humidity: the charges decrease and the fraction of oppositely charged particles rises. At higher humidity, more water is adsorbed onto the particle surface, causing an increase in the surface conductivity. The triboeharging effect, therefore, decreases. The increase in the oppositely charged fraction could be explained by the difference between the electronic properties of the surfaces of the two species decreasing as more water molecules are adsorbed, TEen the materialdependent fluctuations of charges at the contact can be super- ].,o+,; ~'i;, i+J, "..... 0 60 4O 2O o ~o 40 60 saturation,1%) jIITTT r P 60 40 20 O ao 40 60 saturM&~n,('/01 Fig, 9 Chargedistributionof quartz/calcite mixturnsat different hnraidities: left, 2~.; right. [2%, Ot~iectfraction 160/200/zm. partner Imction 400/500 p.m, S,,t+/$r== I, acceleration 12g, ,' = I fl rain. 50 0e.27 Qt 4O 00.5 ~ ~'9#~':/" °Ct'P- ==9 U~ Fig+I0. Median saturationgradeof(1) quartz and tit calcite for different p~icle sizesand surfaceratios. Object fraction and partner fraction: 20/25 and 40/50; 8C,/lO0and 200/250; 160/20g.and 400/500. Acceleration 12g. t= I0 rain, brash speed 1500 tpm. rel. humidity 2% The numbers on the figure indicate the tmxtur¢ ratios iraterms or ( S,,h/Srm); the lines am interpoloriom of the gumvalues for ($,.j/Sr~ . ) = I, K. SchOnert er aL / Powder Techm~logy 86 ( I 9~) 4 I--47 imposed by stochastic processes, It is known from dust explosions that the particles of a uniform dust carry positive and negative charges after being blown up. In Fig. 10, the size influence is demonstrated by plotting the median saturation grade over the particle size. At a constant mixture ratio, the ~so value decreases as the panicles become smaller. Quartz is always charged more than calcite; this difference increases with decreasing particle size. The influence of the mixture ratio is demonstrated using the 20/ 25 and 160/200 p,m fractions. It is remarkable that even the very fine quartz particles can be charged highly if they are imbedded in a feed containing mainly calcite, as is the ease with the mixture ratio S,,~/S~a~0.27. An decrease in the S,~/Sv,~, ratio shifts the ~distribution to higher q~'values.The experiments have shown the following ~ma, and ff',n,nvalues for the 20/25 pin quarlz particles: s,,,/s,~ 2 t 0.5 027 ~m~ Io ~ , ~ t~12% 1-72% 6-70% 17-95% 3. State of agglomeration In a highly bipolar charged particle system, agglomerates are formed. Impressive simulations of this phenomenon have recently been published [ I 0, l 1]. The agglomeration lowers the performance of a tribe-electro.static separation because the particles in a cluster cannot be charged sufficiently, and agglomerates may not be broken down completely by the Coulombic forces in the field; therefore the recovery and the quality of the products become worse. The deglomeration of an electrostatic agglomerate in an electric field can be understood by considering a two..partJcl¢ cluster and assuming a uniform charge distribution on the particle surface. This ideal approach does not take into account eilher any redistribution of the charge due to the contact of the particles or pohrization effects caused by the electric field. The Cgalombie adhesion force F,~ between two nonconductive spheres of diameters x~ and x: and with surface charge densities o I and or2 is given by ,5 45 10 50 for F2<F~ (9) E>-EB~.~2/(I + ~ 2 for Fi <F2 (I0) E>-EBq~/(I+~ ~" The upper limits of these relations are E = EB and E= EB/ 4, respectively, Therefore almost any two-particle agglomerate can be broken down, and also each multiparticle agglomerate, if it stays long enough in the field to turn around into the right position alter a particle has been drawn off. For microfine particles the van der Waals attraction can exceed the electrostatic adhesion. Some calculations for spheres are possible with the Lif>hitz equation, and then the ratio of bolh forces reads [ 12] FJF~dw = [ 16¢r2ato'.,x,zal [so( 1 + ~ 1 / [ ~,t~ff,( 1 + ~+ 2z~lx,)21 Fd/F~,~w=[167r'-crlo',.xlz2~l/|eofid(l+~] (5) For ~= I and o-~= a-2= a, (6) Fel/ Fvdw = 8"n'~o'-xlz~" ,"~fi05 and F2=,trx~a~g (7) Assuming the adhesion is caused only by the Coulombic force, then the particles are separated if the minimum of F~ and F, exceeds Fa: Min(F~,F,) ->Fa (8) Introducing the charge saturation grade ¢-- or/eoEn, it follows ilia[ (I I ) where 05is the Lifshitz-van der Waals constant and z is the distance between the spheres; the other symbols have already been introduced. For an ideal smooth surface, z is assumed to be 4 A; however, z is estimated to increase up to 20 A for ~'ealparticles of 1-10/zm [ 12,13 ]. Toe term 2z~/-¢1 is always small enough to be neglected, so it follows that Fa = [ ¢~.xio-2x~/~] [ I~/( 1 + ~ I a ~=x2/xl Fl~x~o'~E 5~ Fig I I Ratio of Coulornbic1o van dec WoolsaaracliOn,accofditlgt~ Eq. (l~). Fa = [ ~ralcr~l ~] [x~x~l (xl + x~) 2] The larger diameter should always be denoted xt, so ~:___1 by definition. The field acts on each particle with the Couturebic force lqO (I2) (13) Eq. (13) is plotted in Fig. 11 with o'=2.66× 10 -5 A s m-'- (q,= 1) and fi~3--9× 10--'a N m (quartz/quastz contact). The electrostatic adhesion is equal to the van dcr Wools force for 100 gm at z = 9 A and for 10 #m at z = 30 h,. Taking into account the irregular particle sna~, it is to he ¢:~p~cted that the Coulombic force can break down agglomerates of particles larger than about l0 jzm, However, mierofine particles on very fine particles, e.g., a 3 p.m particle on a20 bun particle, may h~,rdly be separate:. Therefore deglomeration has to be performed before the aerosol enters the separator. This was the reason for using the brush aerosol generator as the tr/bocharging unit. K.Sch#n¢,letal./ Po'~derTechnatogy86(1996)41~t7 46 3,1. Experhne.tal 3.2. Results In order to determine the state of agglomeration after the tribocharging unit, the aerosol jet was blown directly into the la~er beam of the Halos 12 LA instrument to measure the in situ particle size distribution. The result was compared with the reference size distribution of the particular sample measured with the same instinmeni after careful dispersion by an ultra.sound treatment in an aqueous NaIP:O7 solution. Six RI0 sieve fractions (20/25/xm up to 63180 gin) and four air-classifier fractions ( 1,4, 2/7, 4/12, 5/25 #m) of quartz and calcite were used separately or as binary mixtures. The grade of agglomeration was evaluated according to Koglin [ 14] and Leschonski et al. [ 15], defined with the moments of the distribution densities as follows: "l'he state of agglomeration is expected to depend on the particle size, mixture ratio, air velocity, brush speed and moisture content, which was kept constant at 2% because tribocharging demands dry air. The brush speed does not show any influence between 500 and 900 rpm, which is the reasonable range for good separation performance. Fig. 12 shows some examples with the air classifier fractions at a mixture ratio of 1:1 and an air velocity of 100 m s- L The distributions after the aerosol generator are much coarser than the reference; the corresponding agglomeration grades/3 and (3') are: 0AS ( 0.11 ), 0,57 (0.15), 0.64 ( 0.33 ), 0.74 (0.42). The deglomeration can be enhanced by increasing the air velocity, as shown with the sieve fraction 25/32 /zm and the air classifier fraction 4/12 am in Figs. 13 and 14. The very line fraction needs a much higher air velocity, but nevertheless a complete deglomeration can be achieved at 200 m s - i In Fig. 15, the influence of the mixture ratio is shown for example with the sieve fraction 25/32 ,am at the low air velocity of 20 m s- ~. The maximum for ,& in the range 0.4--0.6 wt.% quartz, corresponds to the maximum of [3= 1 - ( M ~ o l M . , ) = i - (M ~.flM*-~3) (Koglin) y= 1 (Sv/ST;) -1-(M:.o/M~,)(M~/M*-.~.3) y = (14) (Leschonski) (15) 1 - (M,_JM~Io) ( 1 - [3) (16) 0.4 The asterisk indicates the reference distributi(m, and the moments are delined as known: 0.3 (17) Mko= !Xkqll(x) dx Mk.3= !;f*q3(x) dX qo(x) represents the number frequency and q ~ ( x ) the mass frequency of the particle assembiagc. The program included a variation of the brush speed (n = 500, 750,900 rpm corresponding in u ~ 0.92, 1.38> 1,65 m s-~), the air velocity in the inlet slot nozzle of 40×0.3 mm (t,t. = 18 to 200 m s- ~) and the mixture ratio (cm = fl.I0.9 wt,% quartz). 10@ 9080- -4112,~,~ I 0,4 0.0 la 24 28 2B ~0 air v~ogtty, m Z s 32 34 36 2 5 1 3 2 / t i n , b r u s h speed 9 0 0 t o m , 1,e ~ /P .f 22 Fig. 13. A g g l o m e r a t i o n g r a d e a n d air wlt..xity. I:1 q u a r t z / c a l c i t e mixltlle /T/Tf 7" L¢/ 20 / L • , , , , , , , , i , i , i , 0.8 I4 ~r~ P 40, 30 0.4- O~OSOi 20I0- ol 8 10 port;tie 5~ze, / ~ 50 Fig, 12. P;u'(ide size dislributions o f quarlzfc;~lcile tnixt u~s aflerlribocharging (u~m~ol) and references. Mixtun¢ ratin l : l, o.irvelo~;ity l O0 In s- I, brush speed 9 0 0 r p m . 20 40 60 $0 tO0 !20 140 180 180 200 el v r,~c~ity, ~ / $ Fig. 14. Aggloillcrati0B grade and air velocity. I:1 quarlz/calcile mixtur~ 4/t2/~m, bnJL~h~p~ed 900 rpm. K. 3chiinert et aL /Powder Technology 86 (1996) 41-47 OA ~ , L ~ l ~ t , 5. List of symbols electric field strength EB breakdown field strength F Coulombic force acting on a particle FA, F,~ electrostatic adhesion force F~,aw van der Waals for,'e g gravitational acceleration H(~b) charge distribution in terms of the saturation grade ~, M sample mass Mk,~ moment of the size distribution q(x) size d:.stribution frequency S particle surface wf particle settling velocity x particle size z eontart distance between two particles E 0.3 ~ o,2. Q,t O.O 47 O~, fl.4 ft.6 per,~ent quartz fl,S 1,O Fig. IS, Agglomeration grade and mixlure z-alia, Quartz/calcite rnixluzes :25132 p.m, brush speed 900 rpm the tribocharging activity. More details are re~rted in Ref. [51. Greek letters a 4. Conclusions Binary mixtures of quartz and calcite fractions in the size range 20-200 p.m can be charged highly with specially doraaped Iribocharging units. A vibration chamber oscillating ~t 12g performs well for particles coarser than about 100 #m, but fails/or finer feed because of agglomeration effects. [n this lower range, down to 1 pan, a brush aerosol generator accomplishes effective trihocharging. For general considerations the dimensionless c barge saturation grade ~'is helpful. This is defined as the ratio of tile actual charge density or to the characteristic value or,, represe acing the maximum charge density o',,~, However, the teal value of o',,,,~ for fine, iITegalarly shaped particles is unknown and cannot be evaluated, and therefore this value is calculated using the bleakdown field strength between two smooth plates. The charge is distributed widely and depends on a variety of parameters, dominated by the moisture and the mixture ratio. The best results have been achieved with dry materials and a composition such that both species have the same surface area. The particles are charged up to Offi60%; however, a very small fraction is oppositely charged. At other mixture ratios, the particles of the major species are less charged but the median charge of the minor species increases depending on the material (quartz is affected more than calcite). Electrostatic agglomerates oo~sisting of only a few particles [urger than about 10 p,m can be broken up in the electrostatic field, if the field strength is high enough. The necessary field strength depends on the charge density but is always below the breakdown strength. In agglomerates of smaller particles, van der Waals adhesion exceeds the electrostatic adhesion. Such agglomerates have to be broken before entering the electrostatic field. This can be done by forcing the aerosol through a narrow slit nozzle at a high velocity. In this way, very fine feed, 4/12 p,m, could be deglomerated completely even after being charged highly. angle of the trajectory of a charged pahicle in the electric field /3 agglomeration grade after Koglin y agglomeration grade after Leschonski r/ air viscosity particle size ratio Ps, Pr density of the particle and tbe air, respectively a surface charge density ~b saturarion grade, normalized charge density ~o Lifshitz-van der Waals constant Acknowledgements The authors wish to thank the German Research Foundation (DFG) for the support of this research. References [ 1] A. Singewald and 13. Frieke, Chem.-Ing,.-Tech., 55 (1983) 39. 121 E. ~on Szantho, Brauakohlc, 3~ t 1939) 803. [-;I M. Carla, G. Alraclo, P. Caz'oini, R. Ciccu and C. Del Fa', J. Electroslatics. 10 ( 1981 ) 177. [4] C. Alfano, P. Carbini, R. Ciccu, M. Ghiani, R. Perreli and A, Zueca, Proc. 16th h~t, Mineral Prucessing Congr., Stock.holm. Sweden. 1988, Part A, pp. 833-g44. [5] K- Eichas. Ph.D. The.~i,~,University of Clattsthal. Germany. 1993_ [ 6] K. Eichas and K. Schiiaea, Proc. 18th Int. Mineral Procez ,ins CoMgr.. Sydney. Auslralla. 1993, Vol. 2, pp. 417-423, [7] A. Zahradaicek. Staub-ReinhUr. Lufr. 35 (t975) 226. [gl F, Nierm61ler,Ph.D. Thesi,¢. Universily of Clausthal, Oerrnany. 1988 [9] L.B. Loeb, Static Electrificalion, $1ainger, Bedin, 1958. [ I0] A. Gutseh and F. Lflller, Proc. 6th Ira. Syrup, Agglomeration, Nagoya. }apart, 1993, pp. 46-52. [ I 1] A. Gulsch. $.E. Pratsinis and F. L6fffer. J. Aerosol Sci. 26 (1995) 187, [12] H. Krupp.Adv, Coll~idlnterfuceSri,,l (1967), Ill. [ 13] H. Rumpf. Chern,-Ing, Tech.. 46 (197d) I. [ 14] B, Koglin, P#oc, t,~zInt. Coal: Particle Techmdogy, Chicago. IL, USA. 1973, pp, 273-278. [ 15] K. Lesehonski, S. R/hhele and U~ManTx:l,Particle Characrerizatioa. I (1984) 161.