Solvent sublation for the removal of hydrophobic chlorinated compounds from aqueous solutions

Wat. Res. Vol. 20, No. 9, pp. 1161-1175, 1986 Printed in Great Britain

0043-1354/86 $3.00+0.00 Pergamon Journals Ltd

SOLVENT SUBLATION FOR THE REMOVAL OF HYDROPHOBIC CHLORINATED C O M P O U N D S FROM AQUEOUS SOLUTIONS KALLIAT T. VALSARAJ*,JIMMY L. PORTER, E. KRISTINA LILJENFELDT? and CHARLES SPRINGER* Department of Chemical Engineering, University of Arkansas, Fayetteville, AR 72701, U.S.A. (Received July 1985)

Akin'act--Solvent sublation, a surface chemical technique, was used to remove mono-, di-, tricldorobenzenes and a chlorinated pesticide (DDT) from aqueous solutions, Considerable improvement in efficiency of removal as compared to conventional fine bubble aeration was observed when bubbles of very small size (<0.5 mm alia) were used. The materials were solvent sublated (levitated using fine air bubbles) into mineral oil and lauryl alcohol. The experiments were conducted on a laboratory batch scale. The larger the hydrophobicity of the compound, the better the removal efficiency by solvent sublation was found to be. The removal rate was somewhat enhanced by higher airflow rates and was also more or less independent of the volume of the organic solvent floated on top of the aqueous column. Low aqueous/organic solvent interfacial tension, immiscibility with water and high affinity for the clflorinated compounds were essential characteristics of a good solvent that can be used for sublation purposes. A theoretical model was tested but agreement with experiment was found to be only satisfactory. The effects of non-hydrophobic organics (e.g. ethanol), electrolytes (e.g. sodium nitrate) and long chain alkyi surfactants (e.g. sodium lauryl sulfate) upon the process were also studied. Key words--fine bubble aeration, solvent sublation, wastewater treatment, effects of co-solutes

INTRODUCTION Diffused aeration has long been recognized as an efficient and economical method for the removal of volatile organic compounds from wastewater (Trussel and Trussel, 1980). On the other hand, compounds of low aqueous solubility and low vapor pressure (or low Henry's constants) are not readily removed by diffused bubble aeration. Moreover, a National Research Council report (1978) noted that trihalomethanes (THMs) removal by diffused aeration had the following disadvantages: (i) the THMs were released to the atmosphere, (ii) the THMs were regenerated upon storage of the aerated water and (iii) a very large air/water ratio was required for satisfactory removal. Nevertheless, of all the treatment technologies available (granular activated carbon' adsorption, packed tower aeration, steam stripping, biodegradation, photolytic oxidation, etc.), Trussell and Trussell (1980) suggets that aeration is probably the

*To whom correspondence should be addressed. ?Present address: Royal Institute of Technology, Kemisektionen, Studie Vaglednigen S-10044, Stockholm, Sweden.

least adequately evaluated technique. Chlorinated organic compounds, e.g. chlorobenzenes, chlorinated pesticides, polychlorinated biphenyls etc., are known to be major pollutants in wastewater (Symons, 1975); some of them are known to be refractory (how biodegradable). They are of low aqueous solubility and low vapor pressure and are therefore not easily removable by diffused bubble aeration except for compounds of low chlorine content and molecular weight, since diffused aeration depends primarily on the favorable partitioning of the material into the interior of the air bubbles. On the other hand most of these compounds are hydrophobic, i.e. in aqueous solutions they tend to have high activity coefficient (Mackay, 1981) because of their inability to compete with the strong hydrogen bonding forces between water molecules. These compounds therefore tend to prefer the air/water interface of the rising bubbles rather than the aqueous phase. This tendency makes them "surface active" and these characteristics make them amenable to the so-called "solvent sublation" process. Other terms used to describe the same phenomena are "extractive sublation" and "flotoextraction". This process which originated with Sebba (1962) for ionic-surfactant complexes was later made use of extensively for removing hydropbobic molecular compounds and ion-pair complexes by Wilson and co-workers (Clarke and Wilson, 1983).

1161

1162

KALLIATT. VALSARAJet al.

Solvent sublation is one among the several adsorptive bubble separation techniques wherein a hydrophobic compound is levitated on a bubble surface to the top of an aqueous column where they encounter a solvent layer (e.g. mineral oil, octanol, anisole, lauryl alcohol) to which the material is transferred as the bubbles move through the solvent layer. The floating organic liquid on top of the aqueous column is called the "solvent" and the material levitated by the bubble is called the "sublate". The process is different from solvent extraction in that, because of the continuous mass transfer accompanying the unidirectional motion of bubbles across the aqueous/organic solvent interface, the process is not limited by any equilibrium conditions. Gas bubble levitation of PCBs (polychlorinated biphenyls) have been proposed to be a mode of transfer from the sediment to the "surface microlayer" by Sodergren and Larsson (1981). Mackay (1985) also suggests that gas bubble transport should be considered as a possible route for hydrophobic compounds from the sediment to the atmosphere. When fine gas bubbles (of very small radii in the range 0.01--0.05 crn and in laminar flow), are passed through an aqueous column (with an overlying organic layer) containing hydrophobic compounds, because of their inherent tendency to concentrate at the air-water interface, these hydrophobic materials collect on the bubble surface ,by diffusion through the thin boundary layer surrounding the air bubble. Hydrophobics which are partly volatile will be carried simultaneously in the vapor phase within the bubbles and also in the adsorbed phase on the surface of the bubble. These two processes, viz. volatilization into the interior of the bubble and adsorption of hydrophobics on the bubble surface may be concerted in nature. As the bubble transits the aqueous column and moves through the organic solvent layer the adsorbed phase gets stripped into the organic phase. At the same time if equilibrium between the vapor (inside the bubble) and organic liquid phase is established rapidly, the volatile material present in the interior of the bubble may also partition into the organic layer. If the organic phase remains relatively non-agitated, it has been observed that mass transfer from the organic layer to the atmosphere is a slow process (Springer et al., 1985). Therefore, presumably emissions of these compounds to the atmosphere are also mitigated. Thus, solvent sublation not only helps improve the efficiency of a bubble aeration column, but may also help to overcome, at least partly, the undesired air pollution problem accompanying a simple bubble aeration process. Moreover, the presence of the organic solvent also reduces the eventual redispersion of the material into the aqueous column upon bubble bursting which usually occurs in conventional bubble aeration columns. Wilson and co-workers have shown that the technique has a great deal of promise in removing diehlorobenzenes, chlorinated pesticides (Valsaraj, 1983; Carter et aL, 1985) nitrophenols

(Valsaraj and Wilson, 1983) polynuclear aromatic hydrocarbons (Huang et al., 1983) and alkyl phthalate esters. Valsaraj (1983) studied two different models for solvent sublation along with their experimental verification. A rather comprehensive discussion on the use of solvent sublation for water reclamation is also available (Wilson and Pearson, 1984)i The usefulness of sublation lies in its ability to concentrate hydrophobic organics in a small volume of an organic solvent which can then be destroyed using current practices such as incineration. It combines and intensifies the advantages of both air stripping and solvent extraction with minimum dissolution of the solvent in the aqueous phase. In this paper we present our results on the solvent sublation of a series of chlorinated benzenes and a pesticide (DDT) in a small lab scale batch column. Our goals in conducting these experiments were: (a) to compare the efficiencies of solvent sublation and conventional fine bubble aeration for a series of compounds of varying degrees of hydrophobicity, (b) to test a proposed chemodynamic model for solvent sublation, (c) to ascertain the effects of parameters like air flow rate and aqueous to organic solvent volume ratio which will be important considerations in a scaled-up operation. (d) to ascertain those properties of an organic solvent that makes it most suitable for solvent sublation and (e) to understand the effects of certain co-solutes commonly found with hydrophobic organics in actual wastewaters. This project was an outgrowth of some of our recent work on the control of volatile organic chemical emissions from stagnant surface impoundments (for detaiis see Springer et al., 1985 and Springer et aL, 1984).

THEORY

Mode~for a single stage batch solvent sublation (rate controlled process )--aqueous phase model Let us visualize a scenario of air bubbles rising through a homogeneous volume VL of aqueous solution topped with an immiscible organic layer (see Fig. 1). We shall use a modified version of an earlier model (Huang et al., 1983), the only difference being the assumption of a completely mixed aqueous phase. It has been shown (Wilson and Valsaraj, 1982; Valsaraj and Wilson, 1983; Huang et al., 1983) that the rate of mass transfer from the aqueous phase to a rising bubble is controlled by the concentration gradient across a thin boundary layer around fine bubbles (of average diameter ~ 0.05 cm). It is assumed that the bubble is not in equilibrium with the liquid phase surrounding it and that mass transfer through the boundary layer is rate-limiting. Then

Removal of hydrophobic chlorinated compounds A i r OUt

Torgonic

::::::::::::::::::::: o

o 0

o

o

0

0 o

0

o

solvent section h 0

o

o o o

0

0

0

o

o

o 0

0

0 o

0

0

0

o

o 0

Aqueous section h

Risin 0 air bubbles o

o

0

o

0 o

0

o o

0

0 o

dm dt

0 0

0

--

o

0

o o Oo

° o 0

0 0

o o

~t=47rr2kw

0 o

0

o

A i r in

Fig. 1. Schematic diagram of a solvent sublation process. the following relationship can be written for mass transfer across a bubble of radius r. dm = kw47rr2[C - Cb] dt

--

(1)

where m = total number of moles associated with the bubble (adsorbed on the air-water interface of the air bubble, plus vapor phase inside the air bubble) at time t k w = b o u n d a r y layer mass transfer coefficient, cm s -I C = aqueous phase (bulk) concentration of the trace contaminant, mol cm -3 Cb = concentration of solute in the inner (aqueous) side of the bubble boundary layer immediately adjacent to the air-water interface, mol cm -3. Since equilibrium is assumed only at the inner side of the boundary layer, Cb is related to m through an equilibrium relationship as follows (for details see Huang et al., 1983) =

-

(3)

tim

and

fl=kw

/ (Kr+ )~ H

.

0

- - - Fritted porous disc

m

~C

where

Oo

0

o 0

....

=

0

o

0

The first term on the RHS in equation (2) gives us the mass of solute on the surface of the bubble (assumed to be given by a Langmuir adsorption isotherm) and the second term, the mass of the solute in the vapor phase (assumed to be given by a Henry's law). Since the solute concentrations are extremely small, we have used a linearized Langmuir isotherm in equation (2). This assumption seems justified by previous work (Valsaraj and Wilson, 1983; Huang et al., 1983; Tamamushi and Wilson, 1985). The ratio I~/C~/2 which can be evaluated using a procedure described in a previous publication (Valsaraj and Wilson, 1983) is designated K. Using equation (2) in (1) we obtain

o

o 0

1163

7tr 2

-4-

lrr3H

(2)

where l ~ x = L a n g m u i r parameter for the adsorbed phase, mol cm -2 C,/2=Langmuir parameter for the adsorbed phase, mol cm -3 H = Henry's law constant (dimensionless).

We next make the approximation that there is no change in C during the time interval required for a single bubble to transit the aqueous column. This approximation has been used successfully by Munz and Roberts, (1982) and Matter-Muller et al. (1981) in modeling bubble aeration columns and by Wilson and Valsaraj (1982) and Wilson and Pearson (1984) in modeling multistage batch solvent sublation columns. We can then treat C as a constant for the rise time of a bubble which is given by z = h / U where h is the height of the aqueous column and U is the rise velocity of a bubble of radius r. This then allows the integration of equation (3) between limits of t = 0 and t = z to obtain ~t

m ( z ) = ~ C[1 - exp(-/~z)]

(4)

where re(O), the mass of solute associated with a bubble at the bottom of the column (entry point for the bubble) is taken to be zero and m(~) is the mass of solute associated with a bubble as it exits the aqueous column and enters the organic solvent layer. It may be noted that the factor within the brackets on the RHS of equation (4) determines the approach to equilibrium for the bubble. F o r a single stage batch process, the rate of change of C with time due to the levitation of the solute by the rising bubbles is given

by dC

V L -~

= -- Nbm(z )

(5)

where Nb = number of bubbles produced per second, given by Q./]~r 3 Qo = air flow rate, cm 3 s -l. In obtaining equation (5) we have neglected the changes in pressure and bubble volume along the length of the column. Separation of variables and

KALLIATT. VALSARAJet al.

1164

integration of equation (5) [after substituting for m(z) from equation (4)] then gives In

= -

3Q, :t 4--~r3 (~-V-~L)[l-exp( -- 13x)]t

]]

krem.~.(~L)(~g-~-U)[l-ex p (K+3H)UIj

(7)

Experimental removal rate constants (obtained from the slopes of plots of In C/C(O), with time) can then be compared with the theoretical values from equation (7) if the parameters Q,, vc, r, K, H, kw, h and U are evaluated. The parameters Q,, VL and h are adjustable parameters and are easily obtained. The average bubble radius r in our column was evaluated photographically to be 0.025 cm. We next enumerate the calculation of H, K, kw and U.

Calculation of H Henry's constant, which gives the equilibrium relationship between solute concentrations in the vapor and aqueous phases, can be expressed in a variety of ways. In our case we define the Henry's constant by the following equation

H =

P~M S(760 x 82.05 T)

(8)

where

P~av= M = S = T=

16ax = A~ I

vapor pressure of the solute, mm Hg molecular weight of the solute aqueous solubility, g m l absolute temperature, K.

An excellent compilation of H is given by McCarty (1983). For a more detailed treatment of Henry's constants for compounds of environmental significance see Thibodeaux (1979).

(10a)

exp([31/0) Vm

(6)

where C(0) = initial solute concentration in the aqueous phase, mol cm -3. The first order rate constant for the removal of the material from the aqueous phase is then given by

=kw

parameters as follows

C~,2 = -

(10b)

where A,, = area occupied by a solute molecule at the air-water interface V,, = effective molecular volume of the solute [3 = 1/kT, the Boltzmann factor V0 = adsorption energy, ergs/molecule. The binding (adsorption) energy V0 (which is a negative value) can be estimated using our earlier approach (Valsaraj and Wilson, 1983): Vo = [y~. - yo,. - o : , w ] A . ,

( l l)

where Gs, Yaw, and L., are the air-solute, air-water and solute-water interfacial tensions. These values are obtained from Jasper's compilation of interfacial tensions (Jasper, 1972); the solute-water interfacial tension is usually obtained using the well established Girifalco-Good equation (Girifalco and Good, 1957), 7~, = Yaw+ G~ - 2~b(GwY,~)'12

(12)

where ~ is a parameter, 0.7 for chlorinated organics. Mention should be made of the fact that for the chlorinated compounds of environmental interest the values of Y,w are frequently not available in the literature and use of equation (12) provides only an estimate. This un0ertainty in Ysw causes a disconcerting amount of error in V0 and hence gets magnified as an exponential error in C1/:. This is the most disturbing aspect of our model and constitutes a major weakness of all the models presently available for solvent sublation. Since our isotherm is linearized, as noted earlier, K = Ir~C,/: can thus be estimated. The larger this value, the more surface active (hydrophobic) the solute.

Calculation of kw Calculation of Ig~x and C~/2 The value of I ~ and Cl/: are constants involved in a non-linear Langmuir isotherm of the type Ir

I~C C~/2 + C

(9)

which was derived earlier by Valsaraj et al. (1983) from a simple statistical mechanical treatment of the adsorption of molecular hydrophobic compounds (assuming a cell model for adsorption) at the air-water interface. The assumption of a Langmuir isotherm seems justifiable in light of the arguments presented earlier (Valsaraj and Wilson, 1983; Leja, 1982). The two constants are related to molecular

kw is the boundary layer mass transfer coefficient which can be obtained by solving the diffusion equation in spherical co-ordinates with the appropriate boundary conditions as was shown before (Huang et al., 1983). This method of obtaining kw is necessitated by the special boundary condition involving adsorption (binding) of hydrophobics at the air-water interface of rising bubbles. The final solution is an eigen value equation, namely,

tan

~ = 2

1

D

~r

(13)

Removal of hydrophobic chlorinated compounds where 6 = boundary layer thickness, era D -- solute diffusion coefficient in aqueous solution, OTI2 5-1

rH ct ='-~-+ K,

cm.

The above equation can be solved graphically (Huang et aL, 1983). The first eigen value 21 is related to the mass transfer coefficient as (Huang et al, 1983) kw = 21 ~.

117.3E - 18(~bM,) 1/2T /~s V°6

(15)

where Ms= T = gs = V=

molecular weight of water, kg m o l absolute temperature, K viscosity of water (kg m s - 1) solute molal volume at normal boiling point (m 3 k m o l - 1).

The value o f 6, the boundary layer thickness for fine air bubbles in aqueous solutions for which the Reynolds number (Re=2rUp/Iz) is ~<800 can be obtained using Levich's formula (Levich, 1962) /zr 1/2

where /z -- viscosity o f the solution, poise p = density o f the solution, g crn -3. M o r e sophisticated values of 6 can be obtained by using Schlichting's calculation o f boundary layers on solid spheres as shown by Valsaraj and Wilson (1983). U, the rise velocity o f the bubble can be obtained using Oseen's formula (Valsaraj, 1983) by iteration

2gpr 2 [ U=--~

l (prU'~ 112 1+4\2U

]

+

0.34~,~11 1-2--~ _]

(17)

where g = acceleration due to gravity, era s -2. The values of U thus calculated for various bubble radii are available elsewhere (Kiefer and Wilson, 1980). The value of 6 calculated thus was 0.007 em for an average bubble of radius 0.025em. Equation (13) may therefore be rewritten as X tan (0.007 X) = X2

-1 a(0.025)

where

X f ( ~ ) 1/2.

The above equation can be solved for the lowest eigen value XI by plotting the R H S as functions of X for the four compounds of interest which have different a values. The mass transfer coefficients kw thus calculated are given in Table 1 (see Porter, 1985 for details). The knowledge of all these parameters then helps us to estimate a theoretical value for krm from equation (7) which can then be compared to the experimental value o f kre m obtained from a plot o f ln[C/Cto)] vs time.

(14)

In order to use equation (13) the value of 6 and D have to be estimated. D can be estimated using the Wilke--Chang correlation (Reid et al., 1979) O =

1165

ffi F(X, u)

(18)

EXPERIMENTAL

The column was made of Pyrex glass tubing with a fine fritted glass disk which generated bubbles of average radii 0.025 cm in distilled water. The height of the column was 100 cm and it was 5.0 cm in outer diameter. Three stopcocks were arranged at 10, 50 and 90 cm from the bottom to allow liquid to be collected at various heights in the column. The temperature of the solution was monitored by a thermometer inserted through a large rubber stopper sealing the top of the column. A piece of glass tubing was inserted through the same stopper and connected to a soap film flowmeter. The flow of house air was controlled by a micrometer needle valve and passed through a humidifier before reaching the fritted silica disk. Flow rates were measured with a soap film flowraeter and a stopwatch and was monitored continuously during runs using a rotometer. The experimental arrangement is shown in Fig. 2. Standard solutions of monochlorobenzene (120 mgl -l Matheson Company), 1,4-dichiorobenzene (103mgl-a; Fisher), 1,2,4-trichlorobenzene (97 nag 1- i Aldrich) and p, p'-DDT (414/zg l-1, Polyscience) were prepared in pesticide grade hexane for GC calibrations. To prepare saturated aqueous solutions of each of these compounds separately, a known amount of the respective compound was kept vigorously stirred in distilled water and any excess material was filtered or decanted. The aqueous solutions so prepared were 379 mg 1-1 in monocblorobenzene, 35 mg 1-~ in dichlorobenzene, 20mgl -~ in tri-chlorobenzene and 1/~g1-1 in DDT. The individual solutions were then calibrated against standard hexane solutions, after extracting the solutes into hexane. The co-sohites used were ethanol (Fisher), sodium nitrate (Fisher) and sodium lauryl sulfate (Fisher). The concentrations of the chlorinated organics were measured using a Hewlett Packard 5890 gas chromatograph equipped with a Flame Ionization Detector (for the analysis of monochiorobenzene) and a Ni 63 Electron Capture Detector (for the analyses of di- and tri-chlorobenzenes and DDT.) Helium was the carrier gas for FID and Nitrogen for ECD. The column used for analysis of monochlorobenzene was a 6 ft steel column with 1/8 in. i.d. and packed with Chromosorb 102, while for the analysis of di- and trichlorobenzenes and DDT a 25 m capillary column with 0.31 mm i.d. and coated with Carbowax 20 M was used. Peaks were recorded and integrated using a Hewlett Packard 3392A Integrator. Hamilton 701N series 10/~l syringes were used for injection into the gas chromatograph. To commence a sublation run the column was first rinsed with distilled water, filled with 21. distilled water and the flow rate adjusted to the desired value. The water was then drained off and the column filled with the experimental solution (viz. saturated aqueous solutions of the respective compounds) with or without co-solutes. On top of this was added the required volume of the organic solvent; mineral oil (a commercial U.S.P. grade sample of viscosity 52 cP and average molecular weight 200 + 30) or I octanol (Fisher), or lauryl alcohol (Fisher) as the case may be; the timer started when the first sample was withdrawn from the middle tap

KALLIAT T. VALSAKAJ et al.

1166

Table i. Physical properties of chlorobenzenes and DDT Property

MCB

DCB

TCB

112.56 1.107 12" 488* 2.83~

147.01 1.25 0.70** 80** 3.37*

181.45 1.454 0.25t"t 37§ 4.23~

32.6~:

31.4~t

38.54t'~

DDT

Literature values

Molecular weight Density (g ml- t ) Vapor pressure (mmHg) Aqueous solubility (rag I-I) log (octanol/water partition coefficient) Surface tension (dyncm ~)

354.50

1.55~t~

1.0 x 10 Y** 1,2 x 10 -3.* 6,36¶ ~40

Estimated values

Diffusion constant (cm2s -t) Solute-water interfacial tension (dyn cm -I) Adsorption energy (ergsmol -~) I ~ x (molcm -2) Ci/2 (molcm -3) Henry's constant, H (dimensionless) ( H + 3/r K) (dimensionless) A m (A °2 tool i ) 0m (em s tool -t) k~(cms -t)

8.41 x 10 6

7.71 x 10 6

37.19

37.26

7.28 x 10 6 37.18

1.0 × I0 s (estimate) ~37.2

-2.87 x 10 ~3 4.49 x 10 l0 9.43 x 10 -4 0.148

-3.20 × 10 ~3 4.08 × 10 -l° 3.57 x 10 6 0.069

- 3 . 0 2 x 10 J~ 3.92 x 10 10 5.21 x t0 6 0.070

--5.17 x 10 i~ 2.25 × 10 -1° 1.50 × 10 -s 0.00159

0.154

0.083

0.079

1.80

36.94 101,68 5,02x t0 -4

40.69 117.61 5.13× 10 4

42.31 124.8 2.49x 10 4

73.9:~:~ 228.7 2.45 × 10 3

References are given in brackets. *Goldstein, (1982). "fU.S. EPA 440/5-80-039, (1980). ~/Jasper (1972). §U.S. EPA 440/5-80-028, (1980). ¶Chiou, (1981). **Mackay, (1981). "ti'Kirk-Othmer Encyclopaedia, (1979). ~/:l:Delacy and Kennard, (1971). ~Miller et aL, (1984).

on the column. The aeration runs were made w i t h o u t any organic solvent o n top o f the a q u e o u s solution. At prescribed intervals samples o f approx. 4.0 ml vol were w i t h d r a w n from the middle tap into a 5 mi screw cap scptum vial (supplied by Pierce). F o r analysis of m o n o e h l o r o benzene the a q u e o u s solutions were injected directly into the G C ( F I D ) whereas for di- and tri-ehlorobenzem~ and D D T these solutions were extracted into I ml o f hexane and 2 pl hexane solution injected into the G C (ECD). The air flow rate t h r o u g h the c o l u m n was maintained constant during the run by adjusting the micrometer valve as necessary. After the experiment was over, the organic liquid was pipetted off and the solution drained f r o m the bottom. The c o l u m n was rinsed with alconox solution, rinsed several times with water and finally with acetone and left to dry.

RESULTS AND DISCUSSION

Comparison of efficiencies of solvent sublation and aeration (without floating organic solvent)for the removal of chlorinated benzenes and DDT The compounds chosen for study were monochlorobenzene (MCB), di-chlorobenzene (DCB), trichlorobenzene (TCB) and DDT; the reasons for their choice being two-fold--apart from the fact that these are all environmentally significant contaminants in groundwater and wastewater they also display strong hydrophobic character. Their aqueous solubility and TO

Put dhosa n

SOClp f i l m

flowmerer

T:cm

® 100 cm

Fig. 2. Solvent sublation apparatus: l - - P r e s s u r e regulator, 2 - - P r e s s u r e guage, 3 - - M i c r o m e t e r needle valve, 4---Saturator, 5 - - F r i t t e d disk, 6----Sublation column, 7 - - T h e r m o m e t e r , 8 - - S a m p l i n g port, 9 - - I n f l u e n t tank.

Removal of hydrophobic chlorinated compounds vapor pressure decrease in the order MCB > DCB > TCB > D D T while their hydrophobic nature increase in the order MCB < DCB < TCB < DDT as display by their octanol-water partition coefficient (see Table 1) which has been suggested to be a very good indicator of the degree of hydrophobicity (Valvani et aL, 1981). The wide range in properties for the selected compounds is evident from the fact that MCB is a volatile hydrophobic while DDT is a non-volatile hydrophobic. The properties of all these compounds are shown in Table 1. Details of property estimations arc given by Porter (1985). The increasing hydrophobic nature as we go from mono- to di- to tri-chlorobenzene to DDT has been attributed to the increasing size of the relatively non-polar molecules which renders them less and less able to compete with the strong hydrogen bonding forces between water molecules. Consequently they are "squeezed" out of the interstitial water structure (for details see ErdyGruz, 1974). As the hydrophobicity increases so does the tendency of the compound to aggregate at the air-water interface of the rising bubbles, contributing to the increasing "surface active" nature of the compound. At the same time since their vapor pressure decrease in the order M C B > D C B > TCB > DDT, the partitioning into the interior of air bubbles becomes less and less favorable. Hence, one should anticipate that the ratio of removal rates (sublation to aeration) should increase as we go from MCB to DDT. This was exactly what we observed in our experiments as is evident from Fig. 3-6 and Table 2. The experimental points and the best fit straight line (as for a first order removal) are shown in the figures. The experimental rate constants and the values predicted by our aqueous phase model [equation (7)] are also shown in Table 2 (see Porter,

@

8 ~,° t,

1C

0

I

I

I

I

I

I

30

60

90

120

150

180

Time (min)

Fig. 4. Aeration and solvent sublation ofp-dichlorobenzcne from aquous solutions. Solvent used for sublation--mineral oil, solvent height = 3.8cm, aqueous height = 92 cm, air flow rate -- 1.2 ml s - i. process

Rate constant

SE of k

k (s-') x 104

(s-')x 104

0.75 1.82

0.06 0. I 1

1 Aeration 2 Sublation

1985 for details of calculations). Air flow rates were maintained constant at 1.2 ml s -1 in all cases. It is evident from the rate constants that aeration (without any mineral oil on top of the aqueous column) b ~ o m e s less feasible as we go from MCB to DDT. But the relative removal by sublation to aeration increases as displayed by the ratios of rate constants in Table 2. Comparing rate constants seems more appropriate because the initial concentrations in both cases (aeration and sublation) were kept the same. Alternatively one may compare the half lives for the lOO

7o

70 o

50

2

o ~

70

lOO

5o

1167

% 3O

30

1C

I 30

I 60

I 90

Time

I 120

I 150

I 30

I 60

I 90

Time

height ffi 92cm, air flow rate = 1.2ml s-I.

1 Aeration 2 Sublation

0

(min)

Fig. 3. Aeration and solvent sublation of monochlorobenzene from aqueous solutions. Solvent for sublation---minerai oil, solvent height = 3.gcm, Aqueous

Process

10

Rate constant

SE of k

k (s - I ) x 104

(s - I ) x 104

1.12 1.72

0.03 0.16

[ 120

I 150

(min)

Fig. 5. Aeration and solvent sublation of 1,2,4-trichlorobenzene from aqueous solutions. Solvent used for sublation--mineral oil, solvent height=3.8cm, aqueous height = 92cm, air flow rate = 1.2rnl s-L Process 1 Aeration 2 Sublation

Rate constant k (s - l ) x 104

SE of k (s -I) x 104

0.36 1.22

0.08 0.11

1168

KALLIATT. VALSARAJet al. ~oo~r

l

~

(e)

m 1

e,

70

L

\

"\. tO

10

I

o

3o

I 60

1 90

I 120

I 1.50

Time (rain)

Fig. 6. Aeration and solvent sublation of p,p'-DDT from aqueous solutions. Solvent used for sublation--mineraloil, solvent height = 3.8 era, aqueous height = 92 era, air flow rate = 1.2 ml s-t. Process 1 Aeration 2 Sublation

Rate constant k (s -l) × I04

SE of k (s -I) x 104

0.04 2.48

-0.26

removal to obtain similar conclusions. This then suggests that for each of these solutes the relative improvement in sublation over aeration is in agreement with their relative hydrophobieities. These experiments as well as earlier results (Valsaraj and Wilson, 1983; Huang et al., 1983) and those of Wilson et al. (1984) show that in fine bubble aeration of compounds of low Henry's constants and high hydrophobicity a considerable amount of the material is adsorbed on the air-water interface of the rising bubbles that do not partition into the air phase within bubbles and it is this material that is levitated into the organic solvent during sublation, in addition to the material that is carried in the air phase within the bubbles. During aeration (without mineral oil) the adsorbed material simply remixes and redistributes with the aqueous section as the bubbles burst at the top of the column whereas in sublation these materials are trapped by the mineral oil solvent floating on top and thus prevents the remixing and thereby improving the removal rates.

It may be noticed that there is considerable discrepancy between the experimental and predicted rate constants. This is understandable in light of the error involved in the determination of V0, the binding energy which is the most difficult parameter to estimate. Notice from equation (10b) that C~/2 has an exponential dependence on V0 and hence any slight error in the estimation of V0 will be greatly magnified and show up in the value of K (see also an earlier work in Huang et al., 1983). On the other hand, the Henry's constant is fairly well understood as explained by McCarty (1983). Choice of a single average bubble radius is also suspect because of the non-uniform bubble sizes in the column. Above all, the major drawback of the model is the assumption of a completely mixed aqueous section. This assumption is likely invalid at low air flow rates when axial dispersion is not enough to completely mix the aqueous section. Under such circumstances the process should be modelled as a bubble fractionation column where the rising bubbles enrich the upper sections of the aqueous layer with the hydrophobic (from where it is extracted into the organic solvent) and lower layers of the aqueous section are being depleted of the hydrophobic solute. Clarke and Wilson (1983) and Huang et al. (1983) give excellent accounts of such models. It should also be noted that the improvement in the removal rates by solvent sublation as compared to conventional fine bubble aeration is made possible in part by the very fine bubbles produced which provide for a high surface to volume ratio for mass transfer and because of their very small rise velocities, the smaller bubbles provide for a longer contact time with the aqueous solution in the column. This means that the two most important considerations that have to be met for the success of solvent sublation are fine bubble generation and tall columns. These factors have been elaborated upon in earlier publications (Valsaraj and Wilson, 1983; Huang et aL, 1983). A recent innovation in the area of fine bubble generation is the so called "gas-aphrons" suggested by Sebba (1985) which provide bubbles of micron size diameter. In light of what we have observed here, Sebba's technology may be of tremendous importance in this field in the future. Never-the-less, use

Table 2. Rate constantsfor removalof chlorobenzenesand DDT* (solvent used for sublationis mineraloil Experimental Theory----equation(7) k~m(aeration) kr=m(sublation) k~m(Sublation)~ k,,m (sublation) (s I) (s-l) k~cm(Aeration) log Kowt (s-i) Monochlorobenzene 1,4-diehlorobenzene 1,2,4-trlchlorobenzene p,p'-DDT

1.12 0.75 0.36 0.04

x x x x

10 -4 10 -4 10 -4 10 -4

1.72 1.82 1.22 2.48

× x x x

10 -4 10 -4 10 -4 10 -4

1.53 2.42 3.39 62.00

2.83 3.37 4.23 6.36

1.25 1.52 0.62 1.37

x x x x

10 -4 10 -4 10 -4 10 3

*The standard error of estimate in rate constantswere calculatedand are givenin Figs 3~5. Detailsof calculationare describedby Porter (1985). tKow is the octanol-waterpartitioncoefficient. ~tk,~m(sublation) (t,/2) aeration k~m(aeration) (t~/~)sublation"where t,/2 is the half life of a first order process.

Removal of hydrophobic chlorinated compounds of fine bubbles of micron size would in fact severely limit the hydraulic loading rate for continuous operations, but that is the price we pay for an improvement in the efficiency. It should also be noted that with larger bubbles, the bubbles will no longer remain spherical and their motion will also be turbulent which would offset the improvements gained by the use of solvent sublation. It is also evident from Figs 3-6 and Table 2 that the improvement in the removal rates by sublation over simple aeration is greater for compounds which are more hydrophobic, i.e. compounds of low aqueous solubility, low vapor pressure, relatively nonpolar and having high activity coefficients in aqueous solution. Whereas, for compounds of high volatility and low surface activity (hydrophobicity), as for example, trihalomethanes, the sublation process may not provide any real improvement in the removal efficiency as compared to aeration but may presumably reduce the air emission problem because of the presence of the mineral oil on top of the aqueous solution. This is an area which needs further investigation. Another class of compounds which are highly hydrophobic (low vapor pressure and low aqueous solubility) and are of considerable environmental significance are the chlorinated pesticides. A typical pesticide which we chosen to investigate was DDT (p,p'-dichlorodiphenyl trichloroethane). The removal rates by sublation into mineral oil and simple aeration are shown in Fig. 6, the rate constants are given in Table 2. Considerable improvement in efficiency is observed. As indicated in Table 1, its Henry's constant is extremely small and hence simple aeration is insufficient to affect its removal from aqueous solution. At the same time its hydrophobicity is quite large as indicated by its octanolwater partition coefficient and the adsorption isotherm parameters, therefore its removal efficiency by solvent sublation is good as shown in Fig. 6. In fact, of all the compounds investigated in this paper the relative improvement in removal by solvent sublation as compared to simple aeration is largest for DDT. The percentage removal of DDT by solvent sublation is 82% in 2 h which shows that almost all of the material is carried into the mineral oil layer in the adsorbed phase on the air-water interface of the rising bubbles. Most chlorinated pesticides show much the same characteristics as DDT, viz. high molecular weight, low volatility, low aqueous solubility, relatively non-polar and extremely hydrophobic. These characteristics, therefore, indicate that solvent sublation should prove to be an effective method especially for the removal of these compounds from aqueous solutions.

Comparison of different air flow rates for sublation The solvent sublation of tri-chlorobenzene into mineral oil was investigated for three different air flow rates (0.6, 1.2 and 3.0mis-~). We observed

100,

1169

g~

70 50 0

%

10 0

I 50

I 60 Time

I 90

I 120

I 150

(min)

Fig. 7. Effect of air flow rate on the solvent sublation of trichlorobenzvne from aqueous solutions into mineral oil. All other parameters as in Fig. 5. Air flow rate (mls -l)

Rate constant k (s -I) x 10~

SE of k (s-') x 104

1

0.6

0.59

0.03

2 3

1.2 3.0

1.22 1.70

O.11 0.04

increased removal rates as the air flow rate was increased as shown in Fig. 7. The increase in removal rates were almost proportional to the increase in air flow rates for 0.6 and 1.2 ml s-~ whereas it was not quite so for the 3.0 ml s-~ run. This is probably due to the increased mean bubble radius which we observed with increasing air flow rates which would decrease the inteffacial area per unit volume of air (which is given by 3/r) and would also decrease the bubble residence time since larger bubbles have higher rise velocities. Also, axial dispersion would certainly increase with increasing air flow rate. Neverthe-less, the results do suggest that increased flux of air through the column would improve the performance of the solvent sublation process provided the bubble sizes are kept small. Increasing air flow rates may be the only way to increase the performance of a severely overloaded counter current column (Clarke and Wilson, 1983).

Comparison of aqueous to solvent volume ratio for sublation Sebba (1962) has shown that in the case of the solvent sublation of ion-surfactant complexes from aqueous solution into 2-octanol, the amount of material removed from the aqueous solution is independent of the volume of 2-octanol. These conclusions were checked further by Kargcr et aL (1967) and Cervera et al. (1982). We report here our results on the same effect for the removal of tri-chlorobenzene from water into mineral oil. The results are shown in Fig. 8. No significant improvement in efficiency was observed for aqueous to solvent (mineral oil) volume ratios varying from 94:1 to 44:1. The slight fluctuations we observed are attributable to fluctuations in air flow rates from run to run.

1170

KALLIAT T. VALSMtAJet al.

100 I •

.

70

1

50

3o

~j

10

I 30

I 60

1 90

Time

I 120

I 150

I 100

( rain )

Fig. 8. Effect of aqueous to solvent height (volume) ratio on the solvent sublation of trichlorobenzene from aqueous solutions into mineral oil. Air flow rate was kept at 1.2 ml s -1. Curve 1 displays the result for aeration without mineral oil. Legend 2 3

Aqueous: solvent volume ratio 94 1 44 1

These results are not surprising and do show that sublation is more or less independent of the organic volume. This can be explained as follows: since for the most part mass-transfer occurs from gas bubbles crossing the aqueous-solvent interface and not from diffusion of solute across this interface, the amount of material transferred should depend only on the amount of air crossing the interface and not on the organic volume. In liquid-liquid extraction, on the other hand, the relative volume of the two immiscible phases can be a very important parameter in determining the relative amount extracted. Herein lies the important difference between liquid-liquid extraction and solvent sublation. If the organic volume used in solvent sublation is too low (e.g. in our experiments if aqueous-organic ratios more than 94:1 were used) the mineral oil-water interface would be drastically disrupted at higher flow rates and the process would lose its efficiency; reverse mass transfer of solute from the organic phase to the aqueous phase would occur and solvent volume dependence will become significant (Valsaraj and Springer, 1986). Hence in designing experiments both air flow rates and solvent volume must be so chosen that the disruption of the interface is minimal. On an industrial scale, sublation should prove quite useful in that the aqueous trace contaminant can be concentrated into a very small volume of an organic solvent (like mineral oil, a good choice from economic considerations) which can then be incinerated so as to destroy the toxic compounds as suggested by a recent U.S. EPA study (1984).

(aqueous-organic solvent) is an important criterion for the success of solvent sublation. It then becomes apparent that the aqueous-organic solvent interracial tension would also be a deciding factor. If the bubble encounters a high inteffacial tension, then it will tend to coalesce with other bubbles reaching the interface, becoming large and then move across the interface. This would reduce the interfacial area/unit volume of air moving across the interface at any time and would therefore reduce the removal rate from aqueous solution. In order to study this effect we chose two alcohols--lauryl alcohol and 1-octanol apart from mineral oil as organic solvents for the solvent sublation of tri-chlorobenzene from aqueous solutions. The results using lauryl alcohol and mineral oil (the same volume was used in both instances) are shown in Fig. 9, which shows a slight improvement in performance using lauryl alcohol. The interfacial tensions measured using a Rosano Surface Tensiometer (Biolar Corporation Model LG) are given in Table 3. It was observed that for lauryl alcohol-water system where the interfacial tension was very small the bubbles crossed the interface without much coalescence whereas for mineral oil-water system the bubbles stopped momentarily at the interface, coalesced and then moved sideways and up along the walls of the column along with some smaller bubbles rising through the center of the column. During the lauryl alcohol experiment, the solvent layer was not observed to get cloudy (unlike the mineral oil experiment), nor was water entrainment, channelling or bubble hold-up observed. Our results using octanol were counter productive since in less than 30 min of operation the octanol because of its high solubility ( ~ 5 0 0 m g l -~ at 25°C--Seidell, 1941) caused additional water contamination and the analysis of aque-

100 70 5O O

o

~g 30 k~

1C

I 30

I 60

I 90

Time

1 120

I 150

(rain)

Fig. 9. Effect of different organic solvents on the solvent sublation of trichlorobenzene from aqueous solutions. Air flow rate = 1.2 ml s -l, Aqueous height = 92 cm, organic solvent height = 3.8 cm (in both cases).

Comparison of different organic solvents

Solvent used

As discussed in the previous paragraphs, the unhindered motion of the bubbles across the interface

1 Mineral oil 2 Lauryl alcohol

Rate constant SE of k k (s-I) x 104 (s -1) × 104 1.22 1.63

0.11 0.09

Removal o f hydrophobic chlorinated compounds Table 3. Surface and interfaciai tensions of aqueous/organic solvent systems* Surface or interfacial tension System (dyn cm -')

1171

Mineral oil-air Mineral oil-water

36.4 33.3

Table 4. First order removal rate constants of 1,2,4-trichlorobenzene from aqueous solutions in the presence of various co-solutes at an air flow rate = 1.2 + 0.2 mi s - ' * Standard error of estimate Rate constant, kt in k, Amount of co-solute (s-') x 104 (s -l) x 104

Lauryl alcohol-air Lauryl alcohol-water

29.4 7.8

No co-solute (aeration)~ No co-solute (sublation)

l-Octanol--air 1-Octanol-water

26.5 9.6

Co-solute: ethanol

(mol fractions) 0.0001 0.0003 0.01 0.04 0.10

*Determined using a Rosario Surface Tensiometer.

ous solution showed large octanol peaks and hence the experiments with octanol were abandoned at that stage. This backmixing of octanol was quite evident from the bubble sizedistributionin the column which was observed to become very dense with swarms of very small bubbles giving a milky appearance to the solution. Octanol also caused some odor problems in our laboratory. It should also be noted that both the mineral oil and the lauryl alcohol wc used had insignificant solubility in water (e.g. lauryl alcohol solubility ~ 2 m g l -m at 25°C--Kinoshista et al., 1958). These solubilityconsiderations were reported in detail in an earlier publication of Valsaraj and Wilson (1983). However one drawback of using the long-chain alcohols as solvents is their expense as compared to petroleum-based heavy oils such as mineral oil. These results show that apart from the low aqueous-organic solvent interracialtension, the organic solvent should also have very low aqueous solubility,but should have a high solubilityfor the toxic contaminant of interest.Moreover, the solvent should be non-toxic, and non-volatile so as not to cause additional environmental contamination.

0.36 1.24

0.08 0.11

2.12 2.67 3.14 2.13 0.97

0.15 0.1 I 0.37 0.05 0.07

1.99 2.94

0.12 0.33

3.18 4.52

0.26 0.25

Co-solute: sodium nitrate

(% by weight) 2% 20% Co-solute: sodium lauryl sulfate (molar)

10-4 1 x 10-3 1 ×

*Mineral oil was used as the solvent for sublation. The ratio of aqueous to mineral oil volume was 18:1. tRate constan, were determined from the slopes of Figs 10-12. :[:Run without mineral oil on top of the aqueous section.

Since this s e p a r a t i o n p r o c e s s d e p e n d s o n t h e d e g r e e of hydrophobicity of a compound, any other cosolute w h i c h influences its h y d r o p h o b i c i t y (altern a t e l y a n y c o - s o l u t e w h i c h effects t h e h y d r o g e n b o n d ing in w a t e r ) w o u l d also effect t h e s e p a r a t i o n efl~ciencies b y t h e s u b l a t i o n p r o c e s s . It h a s b e e n widely r e c o g n i z e d t h a t c o - s o l u t e s like a l c o h o l s , i n o r g a n i c salts a n d s u r f a c t a n t s c a n influence t h e h y d r o p h o b i c i t i e s o f molecules. W e , t h e r e f o r e , felt it necessary t o s t u d y t h e s e effects o n s o l v e n t s u b l a t i o n b e f o r e a full scale pilot p l a n t i n v e s t i g a t i o n is a t t e m p t e d o n a c t u a l w a s t e w a t e r samples.

O lO ~5o72O,2

30

Ib

o Io O.O0-ooro.o. • ~ O.O0-s,,~,o,oo • 2o.ooo~ • 30.0003

10

o

A

o

~

, 0.04

6

0.101 30

\

",. \

\

\

\

"q..\ _\ \ • \ \

\

\ ~4 ,

.

\\

\

%3

I

Iv

IN

so

9o

120

",~\2 I

15o

~ u

18o

T i m e (rain)

Fig. 10. Effects o f various concentrations o f ethanol upon the solvent sublation o f TCB into mineral oil. W.R.20/9--G

1172

KALL]AT T. VM..SAR.,Uet al.

Effects o f ethanol as co-solute

The influence of various concentrations of ethanol ranging from 0.0001 to 0.10mol fraction upon the removal rates of TCB are shown in Fig. 10. Table 4 lists the first order removal rate constant. Two distinct effects are clear. At low mol fractions ( < 0.04) enhanced removal rates are observed whereas at mol fractions 0.04 and higher the removal rates start to decrease, the effect becoming quite predominant at 0.10 mol fraction. Similar effects were noted recently by Carter et al. (1985) in their work on the solvent sublation of aldrin, a pesticide. The enhancement in removal rates at low moi fractions may be due to two factors. Firstly, we noticed that addition of ethanol changed the bubble properties considerably, the number of very small bubbles were much larger than when ethanol was absent. This is a well known effect arising from the lowering of surface tension of water which prevents the bubbles from growing to larger sizes at their point of formation on the fritted disk (Adamson, 1982; Carter et al., 1985). These smaller bubbles provide a very large surface area per unit volume of air (since the ratio of area to volume increases with decreasing bubble radius) which contributes to enhanced mass transfer from the liquid phase to the bubbles. Smaller bubbles also have slower rise velocities (Levich, 1962) and hence contribute to longer residence time within the aqueous section for mass transfer to occur. Secondly, according to Ben-Naim (1980), mol fractions of up to 0.03 of ethanol would, in fact, increase the hydrogen bonding of water thus making the aqueous phase "less comfortable" for organic molecules. This means that the TCB molecules would find the aqueous phase less favorable as a medium and hence would prefer the air-water interface of the rising bubbles instead. The combined effect of these two mechanisms enhance the removal rates considerably. Ben-Naim's classic work (Yacobi and Ben-Naim, 1973) also pointed out that ethanol mol fractions above 0.03 tend to disrupt the water structure considerably and makes the aqueous phase "more comfortable" for other organics, i.e. the phase behavior of ethanol-water mixtures are more organic-like. This makes the TCB more soluble in aqueous solution and hence make it more difficult to remove them by solvent sublation. The rate constants in Table 4 show this dramatic effect. One should recognize that such an effect at high ethanol mol fractions is seen in spite of the reduced bubble sizes in the column. The solubilizing effects described above are better understood through phase diagrams of ternary systems involving the hydrophobic (TCB), water and co-solute (ethanol). Another simpler approach to quantify these effects is the use of the so-called two-suffix Margules equation (Hala et al., 1967) as was done by Mackay et al. (1982). The ratio of the solubilities of the hydrophobic in the presence of

ethanol to that in pure water is given by S in the following equation log S

=

X3(AI2

--

A~3 + A23)

=

X3A

(19)

where A~2, AI3 and A23 are called interaction parameters with 1 referring to TCB, 2 to water and 3 to ethanol. The significant conclusions from the above equation are that: (i) The greatest solubilizing effect will be shown by co-solutes which tend to reduce Yl rapidly as X 3 increases which occurs when .413 is very small, i.e. the co-solute has organic character similar to those of the hydrophobic solute and thus miscible with it. (ii) The same effect is possible when .423 is very large, i.e. when there is relatively high non-ideality between co-solute and water. It should then be expected that higher molecular weight organic cosolutes will be more effective solubilizers (even at low concentrations) than low molecular weight compounds. This also provides an explanation for the impairment in performance observed for solvent sublation of polychiorinated biphenyls (PCBs) when l-octanol (of aqueous solubility = 5 0 0 m g l -l at 25°C) was used as the organic solvent instead of mineral oil (Valsaraj and Wilson, 1983; Valsaraj, 1983). Effects o f sodium nitrate as co-solute

The effects of increasing ionic strength (i.e. increasing amounts of an inorganic salt, sodium nitrate) are shown in Fig. 11. It can be observed that the presence ofionorganic salts increases the removal rate by sublation of TCB from aqueous solutions. This is due to the so-called "salting out" effect whereby the presence of salts tend to decrease the aqueous solubility of a hydrophobic organic com-. pound like TCB. This effect is due to the "tying up" of the water molecules in the hydration shells of the ions and thereby reducing the number of "free" water molecules available for solubilizing the hydrophobic in solution. Therefore, the removal of the hydro-

10o \ ...." 70

0 o 50

" ~

30

'~"

Na__N03(_%w/x ) u

0

~

\o

-;o

lC

N~O

I 30

I I 60 90 120 Time (min)

'

I 150

I 180

Fig. 11. Effects of various concentrations of sodium nitrate upon the solvent sublation of TCB into mineral oil.

Removal of hydrophobic chlorinated compounds phobic on the air-water interface of the rising bubbles also increase. This effect increases with increasing ionic strength (increasing salt concentration) as shown in Fig. 11. These effects can also be explained quantitatively using the McDevit-Long theory of "salt-effects" as was shown by Aqua-Yeun et al. (1979). Considerable amounts of sodium and potassium salts can be present in wastewatcr that have been subjected to the primary and secondary stages of wastewater treatment. In a tertiary treatment (such as solvent sublation) these salts can seriously effect the removal of hydrophobics, but hopefully in a positive way as shown by our lab-scale results. Effects o f sodium lauryl sulfate as a co-solute

The presence of surfactants in waste water can arise from secondary and certain tertiary operations like foam flotation (ion flotation) for the removal of heavy metal ions (Clarke and Wilson, 1983). The surfactants can therefore be an interference in other tertiary treatments like solvent sublation of low solubility contaminants. In diffuse bubble aeration operations these have been known to improve the performance of the column just as in the case of packed tower aeration columns. A series of experiments were conducted to study the influence of a commonly found surfactant, viz. sodium lauryl sulfate (SLS) upon the solvent sublation of TCB. Two SLS concentrations were tried, viz. 1 x 10-4M and 1 x 10 -3M. The results arc shown in Fig. 12 and first order rate constants arc given in Table 4. Spectacular improvements were observed in both cases. The presence of surfactants at the air-water interface have been known to reduce the diffusion constant of solutes through the "skin" around the air bubble (Bird et al., 1960; Yousef and McCoy, 1983). However, surfactants tend to reduce the surface tension of the aqueous solution quite drastically depending upon their concentration. This decreases the size of the bubbles generated at the

1173

sparger. As mentioned earlier, as the population density of small bubbles increase,they provide a very large interracialarea per unit volume of air, which apparently more than offsetsthe effectsof decreased mass transfer coefficient,and so increase the overall transfer rate (see also Eckenfelder, 1966). Moreover the reduced interracialtension at the water-mineral oil interfaceas a resultof the presence of surfactants in the aqueous phase helps the bubbles to cross the interface easily without coalescence. All of these factors would tend to increase the transfer to T C B into the mineral oil layer. The increased efficiencyof the process can thus be explained. However, at very high SLS concentrations as the criticalmicelle concentration is approached, there is the probability that the self-aggregatesof the surfactant molecules (viz micelles) can trap hydrophobic molecules within theirhydrophobic core. Our attempts to investigate this were frustrated for two reasons. Firstly,the high concentration of SLS ( C M C of SLS = 8 x 10 -3 M ) formed an emulsion of mineral oil with water and the aqueous layer was therefore found difficultto analyze. Moreover, at high SLS concentrations, the formation of a large foam at the top of the column was a major problem. However, such high surfactant concentrations in waste water samples are unlikely and hence we hope that the study of such effectswould not be very critical. The effectsof ionic surfactants described here are important from another consideration, that of the contemplated use of "micro-gas dispersions" or "colloidal gas aphrons" which are micrometer sized bubbles produced by entraining air in soap films (Sebba, 1985). These small bubbles (10-50 # m dia) provide several fold increase in interfacial area per unit volume of air than the fritteddisk method used here and they risethrough the solution at very low velocities and arc therefore expected to increase the efficiency'of a sublation process tremendously.

CONCLUSIONS lOO

From the laboratory bench-scale studies of the air bubble transport of hydrophobic organics from aqueous phase to immiscible solvents we conclude that:

sLs,mo,,-,, \ Ic O

* 0.O _~ ~ o -4 °'x; °'3 30

\ \ , \~ 60

,\

90

\

\

"X~

,

120

150

, 180

Time (rain)

Fig. 12. Effects of various concentrations of sodium laury|

sulfate (SLS) upon the solvent sublation of TCB into mineral oil.

(1) Solvent sublation using fine bubbles is a marked improvement over conventional fine bubble aeration of mono-, di- and tri-chlorobcnzenes and also of pesticides like DDT. Mono-, di- and tri-chlorobcnzcnes are transported both in the adsorbed phase and in the vapor phase of the bubble, while DDT is transported almost exclusively in the adsorbed phase on the bubble surface. (2) Solvent sublation is m o ~ effective for relatively non-polar compounds of high hydrophobic character, low aqueous solubility and low vapor pressure, precisely the same properties which make them less amenable to simple bubble aeration. Examples of

1174

KALLIATT. VALSARAJet al.

such compounds are chlorinated insecticides and pesticides for which conventional bubble aeration is ineffective in removing them for aqueous systems. Tall columns and fine bubbles are essential for the success of the process. Sebba's technology of generation of micron sized bubbles should be investigated as an alternative to sparging using fritted disks. (3) Considerable discrepancy between the aqueous phase model and the experimental results was observed but seems reasonable in light of the uncertainties involved in the estimates of some of the model parameters. Further refinements in the estimation of K are called for. (4) Increased air flow rates enhance the performance of the process somewhat provided care is taken to see that bubble sizes are kept small. Otherwise, at higher air flow rates, generation of larger bubbles would compromise the efficiency of sublation. (5) Solvent sublation of hydrophobic compounds is somewhat independent of the organic volume provided the organic solvent volume is larger than a critical value. But in the design of the process for large scale applications, the volume of the organic solvent and the air flow rate have to be so chosen as to have minimal disruption of the organic-aqueous interface. (6) The organic solvent used should have low aqueous-solvent interfacial tension, very low aqueous solubility, but should have a high affinity for the toxic contaminants. It should also be non-toxic and non-volatile. From both economic considerations as well as post-disposal considerations mineral oil (or similar petroleum-based hydrocarbons) seem to be an excellent choice for chlorinated compounds. (7) Small amounts of a highly soluble organic co-solute (e.g. ethanol mol fractions <0.04) would improve the rate of removal while larger concentrations (>0.04) would decrease the removal rate. Very large mol fractions (>>0.1) may make the process useless as a separation process. (8) Increasing electrolyte concentrations (e.g. sodium nitrate) would improve the separation efficiency by decreasing the aqueous solubility of the hydrophobic organic. (9) Even small amounts of added surfactants (e.g. sodium lauryl sulfate) would improve the separation remarkably. Certain problems to be addressed in the future are: the effectiveness of the organic solvent as a control method for release of toxic contaminants to the atmosphere during solvent sublation, further investigations into the organic phase model, and analysis of the air released from the process. Preliminary work in a countercurrent mode also shows some success. Results thus far show sufficient promise for investigations of this technique on a pilot-plant scale as a tertiary wastewater treatment process for low solubility hydrophobics (non-polar or slightly polar)

and possibly as a retrofit for existing diffuse bubble aeration columns and quiescent aeration tanks or surface impoundments. Acknowledgements--This work was supported by the Department of Chemical Engineering, University of Arkansas. EKL was supported through a co-operative Summer Student Research Program between the Royal Institute of Technology, Sweden and the University of Arkansas. REFERENCES

Adamson A. W. (1982) Physical Chemistry of Surfaces, 4th edition, pp. 65-70. Wiley, New York. Aqua-Yeun M., Mackay D. and Shiu W. Y. (1979) Solubility of hexane, phenanthrene, chlorobenzene and p-dichlorobenzene in aqueous electrolyte solutions. J. Chem. Engng Data 24, 30-34. Ben-Naim A. (1980) Hydrophobic Interactions, Section 2.4, p. 24. Plenum Press, New York. Bird R. B., Stewart W. E. and Lightfoot E. N. (1960) Transport Phenomena, pp. 541-542. Wiley, New York. Carter K. N. Jr, Foltz L. K. and Wilson D. J. (1985) Removal of refractory organics by aeration VII. Solvent sublation of indene and aldrin. Sep. Sci. Technol. 21, 57-78. Cervera J., Cela R. and Perez-Bustamante J. A. (1982) Analytical solvent sublation of metallic dithizonates. Part I. Solvent sublation of copper. Analyst 107, 1425-1430. Chiou C. T. (1981) Partition coefficientand water solubility in environmental chemistry. In Hazard Assessment of Chemicals--Current Developments (Edited by Saxena J. and Fisher F.), Vol. 1. Academic Press, New York. Clarke A. N. and Wilson D. J. (1983) Foam Flotation-Theory and Applications, Chap. 6. Marcel Dekker, New York. Delacy T. P. and Kennard C. H. L. (1971) X-ray Crystal Structure of DDT: 1,1, Bis-(p-chlorophenyl)-2,2,2-trichloroethane. Chem. Comm. 1208-1209. Eckenfelder W. W. (1966) Industrial Water Pollution Control, pp. 66-67. McGraw-Hill, New York. Erdy-Gruz T. (1974) Transport Phenomena in Aqueous Solutions, pp. 59--64. Wiley, New York. Girifalco L. A. and Good R. J. (1957) A theory for the estimation of surface and interfacial energies. J. phys. Chem. 61, 904-908. Goldstein D. J. (1982) Air and Steam Stripping of Toxic Pollutants, Vols I and II. U.S. Environmental Protection Agency Report No. 68-~3-002. Hala E., Pick J., Fried V. and Vilim O. (1967) Vapor-Liquid Equilibrium. Pergamon Press, New York. Huang S. D., Valsaraj K. T. and Wilson D. J. (1983) Removal of refractory organics by aeration V. Solvent sublation of polynuclear aromatic hydrocarbons. Sep. Sci. Technol. 18, 941-968. Jasper J. J. (1972) The surface tension of pure liquid compounds. J. phys. Chem. Ref. Data 1, 841-1010. Karger B. L., Caragay A. B. and Lee S. B. (1967) Studies in solvent sublation: extraction of methyl orange and rhodamine B. Sep. Sci. Technol. 2, 39-64. Kiefer J. E. and Wilson D. J. (1980) Electrical aspects of adsorbing colloid flotation XI. Adsorption isotherm, particle detachment and differential capacitance. Sep. Sci. Technol. 15, 57-68. Kinoshita K., Ishikawa H. and Shinoda K. (1958) The solubility of alcohols in water determined by surface tension measurements. Bull. Chem. Soc. Japan, 31, I081-1085. Kirk-Othmer Encyclopaedia of Chemical Technology (1979) 3rd edition, Vol. 5. Wiley-Interscience, New York. Leja J. (1982) Surface Chemistry of Froth Flotation, pp. 394-395. Plenum Press, New York. Levich V. G. (1962) Physicochemical Hydrodynamics, pp. 434-471. Prentice-Hall, Englewood Cliffs, NJ.

Removal of hydrophobic chlorinated compounds Mackay D. (1981) Environmental and laboratory rate of volatilization of toxic chemicals from water. In Hazard Assessment of Chemicals--Current Developments (Edited by Saxena J. and Fisher F.), Vol. 1. Academic Press, New York. Mackay D. (1985) Air-water exchange coefficients. In Environmental Exposure from Chemicals (Edited by Neely W. F. and Blau G. E.), Vol. I, Chap. 5. CRC Press, Boca Raton, Fla. Mackay D., Shiu W. Y., Bobra A., Billington J., Chan E., Yeun A., Ng C. and Szeto F. (1982) Volatilization of Organic Pollutants from Water. Environmental Protection Agency Report No. 600/3-82-019, Environmental Research Laboratory, Athens, Ga. Matter-Muller C., Gujer W. and (3iger W. (1981) Transfer of volatile organics from water to the atmosphere. Wat. Res. 15, 1271-1279. McCarty P. L. (1983) Removal of organic substances from water by air stripping. In Control of Organic Substances in Water and Waste Water (Edited by Berger B. B.), U.S. Environmental Protection Agency Report No. 60a/s-83-_~_ ~_T Miller M. M., Ghodbane S., Wasik S. P., Tewari Y. B. and Matire D. E. (1984) Aqueous solubilities, octanol/water partition coefficients and entropies of methlg of chlorinated benzenes and biphenyls. J. Chem. Engng Data 29, 184-190. Munz C. and Roberts P. V. (1982) Transfer of Volatile Organic Contaminants to a Gas Phase During Bubble Aeration. Technical Report No. 262, Department of Civil Engineering, Stanford University, Calif. National Academy of Sciences (1978) Chloroform, Carbon Tetrachloride and Other Halo-methane: an Environmental Assessment. Washington, D.C. Porter J. L. (1985) Aeration and solvent sublation of aqueous chlorinated aromatics. M.Sc. thesis, Department of Chemical Engineering, University of Arkansas, Fayetteville. Reid R. C., Pransnitz J. M. and Sherwood T. K. (1979) Properties of Gases and Liquids, 3rd edition, pp. 567-568. McGraw-Hill, New York. Sebba F. (1962) Ion Floation, pp. 567-568. Elsevier, New York. Sebba F. (1985) An improved generator for micron-sized bubbles. Chem. Ind. 91-92. Seidell A. (1941) Solubilities of Organic Compounds, Vol. 2, p. 622. Van Nostrand-Reinhold, New York. Sodergren A. and Larsson P. (1981) Transport of PCBs in aquatic laboratory model ecosystems from sediment to the atmosphere via the surface microlayer. Ambio 10, 41-45. Springer C., Valsaraj K. T. and Thibodeaux L. J. (1985) The use of floating oil covers to control volatile chemical emissions from surface impoundments: laboratory investigations. Haz. Was. Haz. Mater. 2, 487-501. Springer C., Lunney P. D., Valsaraj K. T. and Thibodeaux L. J. (1984) Emissions of Volatile Organic Chemicals fiom Surface and Near-Surface Impoundments to Air. Draft

1175

Final Report to U.S. Environmental Protection Agency on Project No. 808161-02. Symons J. M. (1975) National organic reconaissance survey for halogenated organics. J. Am. Wat. Wks Ass. 67, 634--644. Tamamushi K. and Wilson D. J. (1985) Removal of refractory organics by aeration VI. Solvent sublation of alkyl phthalate ester. Sep. Sci. Technol. 19, 1013-1018. Thibodeanx L. J. (1979) Chemodynamics-Environmental Movement of Chemicals in Air, Water and Soil, Chap. 2. Wiley, New York. Trussel R. R. and Trussel A. R. (1980) Evaluation and treatment of synthetic organics in drinking water supplies. J. Am. Wat. Wks Ass. 72, 458-470. U.S. EPA (1980) Ambient Water Quality Criteria for Chlorinated Benzenes. U.S. Environmental Protection Agency Document No. EPA 440/5-80-028, National Technical Information Service, Va. U.S. EPA (1980), Ambient Water Quality Criteria for Dichlorobenzenes. U.S. Environment Protection Agency Document No. EPA 440/5/80-039, National Technical Information Service, Va. U.S. EPA Project Summary (1984) Environmental Characterization of Disposal of Waste Oils by Combustion in Small Scale Commerical Boilers. Environmental Protection Agency Report No. 600/$2-84-150. Valsaraj K. T. (1983) Surface chemical methods for the removal of toxic compounds from water: solvent sublation and precipitate flotation-theory and experiment. Ph.D. thesis. Vanderbilt University, Nashville. (Available through University Microfilms International, Ann Arbor, Mich.) Valsaraj K. T. and Wilson D. J. (1983) Removal of refractory organics by aeration IV. Solvent sublation of chlorinated organics and nitrophenols. Colloids Surf 8, 203-221. Valsaraj K. T. and Springer C. (1986) Removal of traces of pentachlorophenol from aqueous solutions by solvent extraction and solvent sublation. Sep. Sci. Technol. In press. Valvani S. V., Yalkowski S. H. and Roseman T. J. (1981) Solubility and partitioning IV. Aqueous solubility and octanol-water partition coefficients of liquid nonelectrolytes. J. Pharm. Sci. 70, 502-507. Wilson D. J. and Valsaraj K. T. (1982) Removal of refractory organics by aeration II. A fast algorithm for modelling solvent sublation columns Sep. Sci. Technol. 17, 1387-1396. Wilson D. J. and Pearson D. E. (1984) Solvent sublation of Organic Contaminants for Water Reclamation. O.W.R.T. Report No. Ru-84/5, Department of Interior, Washington, D.C. Yacobi M. and Ben-Naim A. (1973) Hydrophobic interactions in water-ethanol mixtures. J. Soln. Chem. 2, 425-443. Yousef A. and McCoy B. J. (1983) Diffusion of suffactants to an interface: effect of a baxrier of oriented water. J. Colloid Interface Sci. 94, 497-501.