. . I OFI. ORNLP 2961 - - EEEEEEEE 11:25 11.4 16 MICROCOPY RESOLUTION TEST CHART NATIONAL BUREAU OF STANDARDS - !963 í heim. ORNL-P. 2961 !ONA -7036 1 MASTER Com PRICES HC. $2.00; HNK . HYPERFILTRATION WITH DYNAMICALLY-FORMED MEMBRANES . .. Kurt 1. Kraus, Arthur J. Shor, and James S. Johnson, Jr. . . . : . . Oak Ridge National Laboratory Oak Ridge, Tennessee, U.S.A. . . . . .. -... --- RELEASED FOR ANNOUDCADI - . IL NUCLEAR SCIENCE ABSTRACTS . . -- - ... ------ LEGAL NOTICE - - . . . This report was prepared u an account of Government sponsored work. Neither the United States, nor the Commission, nor any person acting on behalf of the Commission: A. Makes may warranty or representation, expressed or implied, with respect to the accu- racy, completenes, or usefulness of the information contained in his report, or that the use ol way information, apparatus, method, or process disclosed la this report may not infringo printoly owned righua; or B. Asamo vay liabilities with roup.ct to the use of, or for damages resulting from the va of any information, apparatus, moinod, or procesa disclosed in this report. . A. vuod in the abovo, "person acting on bolall of the Commission includes any om- · ployee or contractor of the Commission, or omployee of such contractor, to the except that such employee or contractor of the Commission, or employot of such contractor prepares, dienominates, or provides accomo to, any information pursuant to his employmeat or contract with the Commission, or his employment with such contractor. . llyperfiltration Studies X. Research sponsored by the Office of Saline Water: U, S. Depart- ment of the Interior, under Union Carbide Corporation's contract with the U. S. Atomic Energy Commission. In hyperfiltration or reversc osmosis salts are removed from salino -- - * ! PS - wators by forcing the solution under pressure through appropriate membranes. -. y7 . The membrane material of choice at present is celluloso acetato whose salt. -. N SS -' ..YAT: 5 . . -. .- - .-..-.... . IRENE . . . NA rejecting properties were discovered by Reid and Breton. M) When cast ac. cording to a special recipe - i.e., that of loob and Sourirajan - (2) reason- abiy high water transmission rates can be achieved together with excellent salt rejection. Much of the optimism regarding the future of hyperfiltration as a desalination technique rests with this remarkable membrane. . . From current great emphasis on cellulose acetate one might get the mis- taken impression that salt-filtering properties of materials are rare or una usual. Reid and others, many of them working under the sponsorship of the Office of Saline Water, have, however, found(3) that many synthetic organic membranes have salt-filtering properties; it now appears that many inorganic materials also reject salts. While we also made membranes by synthetic organic techniques, (4) we have more recently concentrated on the dynamic method of forming them. This study began with observations, first made by A. E, Marcinkowsky of this Labora- tory, that high-flux, salt-rejecting membranes can be formed on porous silver frits by passing solutions containing Th (IV) over them. Formation of salt. rejecting layers on porous supports from feed solutions, containing various "additives" has since been shown to be rather general. Bolo) The method. . . - RES : -- . Z:: - - -***: - - --- .. mais -- - - - - - - . - - - . . --- - - + : is simple and by-passes conventional, time-consuming, membrane casting procedures. It permits rapid testing of rejection characteristics of mate- rials and thus helps to screen materials suitable for membranes which can supplement modol solution studies. Mols) Probably of greatest interest how- 1 : cver is the observation that hy the dynamic method salt-rejecting mombranes with very high transmission rates can frequently be prepared. Theso membranes can also be maintained dynamically by retaining in the solution a small concen. . . - -- --- - - - - - -- - - - - - - - tration of the additive, which presumably covers areas in which weaknesses, might develop. In this sense the dynamic membranes can be considered "self- healing". While we have not been successful in demonstrating it to our : satisfaction, the technique might also permit control of the thickness of membranes and hence their transmission characteristics through control of the hydrodynamics of the system. Many materials seem capable of forming dynamic membranes with salt.. rejecting properties; Table I lists typical examples. Some of these can be present in natural feed waters and one wonders how far deposition of these in thin layers on, e.g., cellulose acetate membranes, affects the performance characteristics of these membranes. The number ind variety of porous sup- port structures on which dynamic membranes can be developed also is very large; 'a representative list is given in Table II. These lists are only to indicate the breadth of the phenomenon, not its details. Not all support structures have been found equally satisfactory nor have the various additives shown equal or high desalting abilities, Much has yet to be learned about the mechanism of dynamic membrane formation. Indeed we are very puzzled as to why some of these membranes form. Formation of mem- brane-like layers on porous supports is perhaps expected for colloidal addi. tives or particulates, but the pore diameters of the supports are very much larger than the molecular size of many of the additives. For example, with Th (IV) we have observed formation of salt-rejecting membranes even with very slightly hydrolyzed solutions (*) where the polymers are small. One thus won- ders if there may" not be a distinction between "plugging" pores by unidentified - - - - - .............. ............................. trace colloidal constituents and presumed "activation" of this "filter cake" through adsorption of low molecular weight solutes. With the organic polymers and polyelectrolytos the process of forming the membrane might begin with a modest retardation of the polymeric additive at the interface which eventually leads to a thin, rolatively concentrated polymer film of high viscosity, or per- haps even to precipitation if the solubility at the interface should be exceed- ed. At present these are speculations and, as mentioned, much remains to be donc to determine the mechanism of formation and nature of the film and to develop the desired control over thickness and water flux. . - - - - -.. . . .- . -.-.- General Considerations on Hyperfiltratior: Before discussing some of the more . -.- - -. -. -. -.-. . - --..- TY quantitative aspects of salt rejection with dynamic membranes, a review of some requirements of membrancs for salt rejection is desirable. Only a brief discussion is possible here; for details and derivation of the appropriate equations, see reference (8). According to a simple, though, we believe, reasonable. theory, salt re- jection R can be described in terms of an equilibrium distribution coefficient D* (moles solute per kg water in membrane/moles solute per kg water in aqueous phase): a coefficient ß , presumed to be between zero and one, describing coupling of salt and water fluxes; and a flow parameter, o - 112/02, which is the dimensionless ratio of water flux Ja, membrane thickness l and a special salt diffusion coefficient b, in the menibrane. . :: If we define rejection R as ' 31. mail the where in is the salt concentration of the product and ma the concentration at the entrance (a) interface of the membrane, these quantities are related by the equation . ... . . . . .. --*. ES rrivremen R/(1-B) = (1-0-80) ((1/2)-1) (2) (3) ----..-- .- . .-,.-* - . ... By Eq. (2), good rejection is dependent on a kinetic torm involving Bo, which should be large, and an equilibrium term, BD", which should be smail. As prossuro increases, J, and o increase, influence of the kinetic term diminishes, and rejection approaches a limiting value, hoe : 1.BD dependent on coupling and equilibrium distribution coefficient at the high pressure interface. For a menbrane with completely uncoupled flow, B=0, rejection " approaches unity at high pressuros; uncourled flow see is to be a.good approximation for cellulose acetate, (9).(10) However, if Bes is less than one, as it is for the dynamically-formed membranos so far studied by us, at least a degree of cou. pling is implied, we shall set. Bol in the present discussion, since (in ab. sence of independent determination of B, in principle possible but difficult with this class of membranes) this extreme seems more reasonable to us for these membranes than B=0. Equation (3) then becomes Bar olo Da The correlation of high limiting rejection with small values of dis. tribution coefficients implies that with neutral membranes of the cellulose acetate type, activity coefficients of salt in the membrane phase should be greater than in the aqueous phase.. i.e., the activity coofficient ratio, r , should be larger than one. Ordinarily, this occurs with membrane materials of low dielectric constant, (8), (11) and of low water content(?). (the salt- rejecting layer must also be essentially homogeneous). These requirements imply low water flux, except for membranes with very thin salt-rejection layers, such as those (order of 0.1 micron) attained with cellulose acetate cast by the Loeb-Sourirajan (L.5.) recipe.. . (3a) .asringa s 4.2.- - . .-.-saios.com.raamimisse - -. .-. .- - . ... .-. . . . pe. . . .. ... - . f . . . . . . . . . ... - .-. - Distribution coefficients less than unity can also be achieved by making the membranes charged - 1.0., from ion-exchange-active materials. Here the activity coofficient ratio r need not be particularly high because salt re. jection occurs by salt exclusion through the "Donnan equilibrium". With ion- exchange membranes the concentration of the counter-ions (ions having charge sign opposite to thet of the fixed charge of the membrane) in the membrane is large and hence the concentration of co-ions (ions of same charge sign as mem- brana), which determines salt invasion and the distribution coefficient, is usually small. Thus for a salt MX at concentration m in the contacting solution phase. and an anion exchange & Oven b+ 1 262) () 9 . (4a) . - . . ---- TE where b'is the charge of the co-ion; c* the ion-exchange capacity (moles per kg water) and r the appropriate activity coefficient ratio; Q is the ratio of counter-ion concentration in the solution to the ion-exchange capacity. To make D* small, should also be small. Ion-exchange membranes may thus be effective for salt-rejection even at high water contents and with pore diameters of many tens of angstroms provided Q < 1. (12) However, they would not be expected to be effective at the pore diameters, of 0.1 to su - - - . ' - - . - .. - - - - - - - - which we use in our support structures. - - . S u VA. .um.'. k - - - ' k * - - Concentration Polarization: To demonstrate the salt rejection inherent in a membrano, salts which accumulate at the feed interface must be removed. If this is not done, salt build-up or concentration polarization at the interface - in - - * .-R 27. errea ws11 rapidly roach such high values that the product solution will have the same composition as the foed, as if the membrane had failed. A remedy is to pump feed solution past the membrane as rapidly as possible. For prescrit pur- poscs we shall assume that we pump the feed solution past the membrane under turbulent conditions, nunca ..-...a If we consider turbulent flow in a tubes) and define observed re- jection - . -- Pobs 1. (5) -. (where feed concentration, me, is taken as the concentration in the turbulent - : . .ovo. core), the relationship between Robs and R is given by (14) . 108[(1-Bobs)/Robs! • k v/u•* . log[(1-R)/R] Here v is the transmission rate of the membrane, u the circulation rate, and . .- .-.. K a constant determined by geometrical factors, the Schmidt number, and con- - S.-..-. stants from the Chilton-Colburn analogy and the Blasius equation. To evaluate intrinsic rejection of a membrane & from Robs; it is thus necessary to extrapolate data obtained as a function of circulation velocity to infinite velocity - e.g., by plotting log (1-Robs)/Robs against v/u'. Appropriate data may be obtained if the porous support is in the form of a tube - e.g., a carbon tube - and the equipment (see Fig. "1) contains both pressurizing and circulating pumps with appropriate valves to control the linear flow velocity ū through the tube. Typical results, shown in Fig. 2, were obtained with a hydrous zirconium oxide membrane dynamically deposited on the inside of a porous carbon tube. Equation (6) predicts a straight-line relationship between log (1-Robs)/(Robs) and v/u•75. Although this holds here well, the slope is not that expected from the equation because rejection by this membrane varies with concentra. ' . ' , • A.-..-. li cooooooooooooo - ----- . . . -- -- -- rion. However, if this is taken into consideration, satisfactory agreement ----- -- - between theoretical and observed slopes is obtained. (14) - - - - According to Eq. (6), in order to maintain a given ratio of Robs/R (i.e., a riven extent of concentration polarization for different permeation rates) the velocity ratio v/uº' should be kept constant. With low flux membranes (small values of v) relatively small circulation rates are sufficient to keep concentration polarization within bounds. However, with high flux membranes, with which we shall be largely concerned here, correspondingly larger cir. culation velocitics should be used.. Very large values of u are required when v is of the order of 1 cm/min and concentration polarization is to be kept small. In such cases use of turbulence promoters may be desirable. We have tested a "detached promoter" .. (developed by D. G. Thomas) (15) in the form of a spiral with only point con- tacts with a carbon tube. The results were encouraging. Figure 3 compares. results obtained with a single tube which had on the downstream half . of its length the promoter. Concentration polarization at a given value of v/u was considerably less in presence of the promoter than in its absence. In addition the transmission rate v was substantially larger in the promoted section than in the unpromoted one; we have frequently seen v to be 20 to 50% larger in the promoted section. - ..:- .. Dynamic Hydrous Zirconium Oxide Membranes: We shall discuss in some detail hydrous Zr(IV)-oxide membranes to demonstrate some of the properties of dyna- ------...---- ) I mic ion-exchange membranes. This is done not so much because this additive is particularly suitable or uniquely attractive for desalination but rather because the properties of this material are relatively well understood, the ---------- membranes are readily formed, and the opportunity presented itself to check . -8- out some of the basic theoretical considerations. As we had mentioned, ion-exchange membranes might be of interest to le- salination because they might permit relatively high water fluxos when many . TU .. ... won c-motor..d . ~--. - .- -... microns thick while for noutral membranos (uncharged) high fluxes would re- quire much thinner films. We have attempted preparation of ion-exchange mem- branes for several years and demonstrated (4) that membranes of very high in- trinsic permeability can be produced synthetically, but the actual membranes were of the order of 100 microns thick and showed permeabilities of only a few gpdifta. Very much highor fluxos have now been observed with aynamically deposited ion-exchange membranes. . For example, hydrous zirconium oxide membranes (prepared from boiled zirconium oxychloride solution) may show water fluxes of the order of 1 cm/ min (1 cm/min = 355 gpd/ft?), while showing significant salt rejection. Some typical observed rejections: (not extrapolated to v/u - 0) and flow rates are shown in Fig. 4 (sce also reference ( 5 )). Intrinsic salt rejection R by ion-exchange membranes should have some special characteristics if it is largely determined by the equilibrium dis- tribution coefficient p* (Eqs. (3a);(4)). Salt rejection should increasc as the concentration of electrolyte in the surrounding medium decreases. The effect should be more marked - 1.e.,, the slope greater - with salts contain- ing multivalent co-ions, and the relationship should be less steep for salts . with multivalent counter-ions. One expects for most exchangers, but not nec- cessarily for all, that rejection is poorer when polyvalent counter-ions are present. Values of B as a function of feed concentration and charge type of solute may be used to detect coupling and to identify rejection by an lon- exchange mechanism..: com videre . Figure 5 gives extrapolated NaCl rejections as a function of M Naci ob- • ...:: . .' 'I in. ny: . L -- C o o o o º o o o • 0 0 0 0 0 - - - . - - . - "-9. . . . . . .-.-.-.-.-....---... tained with a hydrous Zr(IV) membrane dynamically deposited on a carbon tube. Transmission rates at 400 psi were ca. 0.25 cm/min. (At this pressure the extrapolated values of R are only slightly less than Rs.) The increase in re- jection with decreasing Naci concentration is approximately as expected from an ion-exclusion argument with variations of activity coefficients in the aqueous phase taken into consideration (solid line, Fig. 5). As illustrated in Table III, in acidic solutions MgCl2, BaCl2 and LaC13 are much better rejected than NaCl- by hydrous zirconium oxide as expected; under these conditions the oxide is an anion exchanger and M9**, Ba**, and La*** are polyvalent co-ions, Sodium sulfate (with a divalent counter-ion) is negligibly rejected. In basic solutions, the riverse is the casc : (Table III). Sodium sulfate is better rejected than hariun chloride. In the intermediate pH range, salt rejection should follow the varia. tion of capacity with pil. As shown in Fig. 6 this is at least qualitatively the case. In this figure the lines represent ion-exchange capacities (moies/ kg solid) obtained from titration curves of hydrous zirconium oxide. (26) The points are capacities (moles/kg H20) computed with Eq. (4) (for r . lip* - 1-B) from values of R obtained by extraolation to infinite circulation velocity. The ordinates were adjusted to bring capacitj of the solid ex. changer and the membrane (different units) to the same value at one ph. The --- .........--.--.-. ----------- -- - -- --- --- remarkable parallel between ion-exchange properties of solid hydrous zirconium . - . oxide and hyperfiltration properties of a membrane dynamically-formed with .. . -.. colloidal dispersions of this material confirms that rejection is by an ion- exchange mechanism, and that fq. (3a) applies. The capacity of the membrane is iro. less than that of the solid exchanger. Properties of Other Dynamic Membranes: We listed in Table I a few of the mate- ...... . .......:: - . IM -10- - - -- -- - - rials which when used as additives yield dynamic membranes with salt-rejecting properties. While detailed discussion of these materials is not possible ·'hore we shall give a few typical results: (a) Hydrous Ferric Oxide: This material would be expected to be a general contaminant in hyperfiltration work carried out in steel-containing equipment. To avoid extensive contamination most of the work reported in this paper was carried out in equipment where even stainless steel was avoided' as much as possible. The pumps were either titanium or Hastelloy C and connect- ing tubings were of titanium or reinforced rubber tubing. while we established in preliminary experiments in titanium equipment that "colloidal" hydrous ferric oxide (prepared by boiling Fećiz solutions). causes development of a salt-rejecting membrane, further studies were carried out in stainless steel equipment with a carbon tube as substrate. We may summarize these experiments as follows. (the experiments were performed by Dr. H. 0. Phillips of the ORNL Chemistry Division; we wish to thank him for per- mission to quote them): . Salt rejection by hydrous Fe2ºz is pH sensitive, as is rejection by hydrous Zr02. The rejection minimum (isoelectric point?) is located near pH 8 in .08 M Naci solutions (10-4 M Fe(111). · In one series of experiments Robs was .56 near pll 3 for .08 M Naci, was negligible noar pll 8 and became .2 near pil 12. (Measurements were at 600 psi; at a circulation velocity u a 15 ft/sec and transmission rate væ0.1 cm/min.) lobs became essentially in- dependent of pressure near 500 psi; it was significantly lower at the lowest pressure used (100 psi). With increasing temperature v increased approxi. mately as expected from the change in fluidity 9 of water (1.6.; V/was approximately constant) and Robs was hardly affected. SU ET - -- - -- - - - - - - - - - - -- - - - - - - --- - - - - - - - - - . TE tore .. 1 . ... 0 0 0 0 0 0 0 0 0 0 0 0 0 a - - - - - - - - - - - · - - - .-11. . () Organic Polyelectrolytes: We have been able to prepare dynamic hyperfiltration membranes from a large number of organic polyelectrolytes in- cluding materials with strongly acidic, weakly acidic, strongly basic and weakly basic functional groups. Some typical examples have been given in a previous publication. A typical set of results, taken from this paper, is shown in Fig. 7. - - - 4 :: 4: (c) Humic Acid and Bentonite: Humic acid, as used by us, is a dark. colored extract obtained from decaying oak leaves (our stock was prepared by D. J. Nelson of the ORNL Health Physics Division). We have used it without further purification and found it to form dynamic membranes with interesting rejection properties. A typical set of results is given in Fig. 8. . The "humic acid" membrane was developed on a 0.24 flat porous silver filter by circulating past it a solution of 250 mg/& humic acid content for sevural hours. The results of Fig. 8 were obtained over a 2-week period; the feeds contained 25 ppm of humic acid. By comparing optical density of "bleed" and product solutions, from time to time, the additive was found to be better than 90% rejected. Salt rejection increases with decreasing feed concentration and is high- er for the salt with the divalent co-ion, sulfate, than for Naci at a given sodium concentration, qualitatively as expected of a cation-exchange membrane. The membrane rejects MgCl2 (a salt with a divalent counter-ion) to an appre- ciable extent. The permeation rate seems less in the presence of MgCl2; this probably reflects a change in membrane properties with time (the order of ex- periments was Naci, Na2S04, MgCl2). After the MgCl2 series, pormoation of 0.017 M NACI was 25 gpd/ft? and rejection 71%, values somewhat changed from . - . . . -. -- - - - - . those given in the figure. TI --... --- - . -12. Concentration of humic acid considerably lower than 25 mg/d sufficed to 'maintain rejection, though just what minimum concentration is necessary is yet doubtful. In one test, with 7 mg/l, no loss of rejection was observed; at 2-5 mg/l, a decrease (from 65% to 40%) was noted during about 10 hours operation. Rejection was readily restored by subsequent exposure to 25 mg/l. In a later test, after the MgC1, series, there was no rejection loss in 17 hours operation at 1 mg/l. Bentonite, a montmorillonite-type clay, when used as an additive in low concentration, has at times appeared to form dynamically salt-rejecting mem- branes, as one might expect from its cation-exchange characteristics. Un- fortunately our results with this additive have been relatively irreproduci. ble, and have cast doubt on whether membranes formed from this clay alone ro- ject salt. Combination of humic acid and clays, as mixed additives, however, have generally shown rejection properties which were superior to thosc shown by either additive alone. A typical set of results is given in Fig. 9. The membrane was prepared by circulating past a 0.84 silver-frit a 0.02 M NaCl solution containing 45 mg/l of finely divided bentonite. When flow rate had decreased to 170 gpd/ft2 at 500 psi, salt rejection was still negligible in this case. But on exposure to a 0.02 M NaCl solution containing 250 mg/l of "humic acid" the layer ac- quired rejection properties (Robs * .63): permeation rate had decreased to 10 gpd/ft? at 500 psi. The results (points) of Fig. 9 were obtained with this membrane at 2000 psi. The curves are the rejection data copied from Fig. 8. While the results were more erratic with the combination membrane than with humic acid alone, it is apparent that rejections are substantially higher for the combination membrane than for the "simple" one. Permeation rates (and pormeabilities) were substantially lower. ..- - - - -- -- - - - - - --- -- - -- -- 1 . . ! A . . - . . - - . . . - - - - -13- - - - . . • The observation that dynamic membranes may he formed from such "natural" . . - . . . - .. additivos raises the question whether some natural waters may not contain mato- - - - - . - -- -- - - rials suitable for salt-filtration and other purifications. One expects that, - . - - . -- . . .- .-. . at least under some circumstances, use of special additives may not be re- . . ... . . quired to effect water purification. One also wonders how widely nature may . - . -.- be utilizing "dynamic membranes" deposited or formed on porous formations for . . ... . . .. changing relative salinities of waters and for carrying out other water, purifications or compositional changes. (d) Polyvinyl Pyridine: This material as an additive for dynamic memo branes is of particular interest because it should represent, a class of ora ganic compounds whose charge density (ion-exchange characteristics) should be controllable hy the solution pll. One expects the material to form an anion- exchange membran in acidic solutions (where the pyridinium ion forms) anela . - - . "neutral" (uncharged) membranc in neutrai and basic solutions. Several series of experiments with a polyvinylpyridine membrane carried :: ANA :.- --- out on a membrane "aged" by several days prior operation) are summarized in .--..! Fig. 10. The membrane was formed on a flat disk of letrice! VM6 with nominal ---- . pore size 0.450. Initial concentration of the additive was 100 ppm; during the rejection experiments the solutions contained i ppm of the additive. The membrane showed 70% rejection of 0.05 M Naci near pil 3. This rose to WS 85-90% above pll 4. Magnesium chloride (.025 M) was 85% rejected at pl! 3; rc- jection was somewhat better at higher ph. Sodium sulfate retection (0.011) rose from 43% at pll 3 to better than 80% at high pll. If we consider the rema branc in acidic solutions to be an anion exchanger the relative rejection Rmoci, > Ryacı > RNasso, is as expected. llowever, sodium sulfate with its di- valent counter-ion is remarkably well reiccted. Also in the few experiments so far carried out, the concentration dependence of reiection of NOCI Os mienn's 14. use . . . . . . . . . . . . - - - - less than expected from the usual ion-exclusion arguments. Thus, the usc- fulness of describing the properties of this membrane in acidic solutions in terms of ion exchange is in doubt; it seems to behave "better" than antici. pated. · Examination of the permeation properties of the membrane as a function of pll, : also given in Fig. 10, supports the contention that there isng trinsi. tion from an ion-exchange membrane to an uncharged one as the pil is raised. Flow rate through this membrane decreases almost a factor of ten ?s the pil .. .. . . - - - - - - - -- of the solution is raised. Presumably in acidic solutinns the membrane is in -... ... - - - -- - the pyridinium form and resembles a swollen (hydrated) polyelectrolytc. At high pll, the pyridinium groups are neutralized and a low water content mem- hrane forms. If the amount of polymer ner unit area were the same in the two pll ranges, this loss of water would be suivalent to a drastic decrease in the equivalent pore size and hence would account for the very large diccrcasc 11 WAT - - . WLOT . -..- . .. in transmission rate, While flow rates in the high pll range were low (4.7 gpd/ft? at 1500 psi). the experiments are nevertheless of considerable interest since they reoree sent the best results so far of our attempts to fomn dynamically salt-rejecting membrancs. As shown in the figure, salt rejection in neutral solutions can be high. Also in Fig. 10 are results of tests with a solution ap; roximatins a natural water ("Coalinga-type" . 0.0028 !! NalıCO3, 0.0996 N N.22504. 0.0012 CaCl2, 0.002 N MgC12). Total anion rejection was 93%; it was essent:ally the same after acidification to pl 5.8 but dropped to 63% 27 pl! 2.9. Chave found that this dynamic me:brane is capable of significant rejection of alts from seawater (obtained at Pawley's Island, South Carolina). In one 18-hour test whore chloride rejection was 90%, we observed 99% Mg and Ca rejection and 99% sulfate rejection. Flow rate was 3 mod at 1500 psi. From earlier model soiution studies !10) we had anticipated that piridine * menya m TV-IW ... ..-.in - - 1 -- - - - - -- - - - - -- - - - - - - -15 membranes and cellulose acetatc membranes would resemble each other. Our re- sults confirm this expectation. If one assumes that cellulose acetate imem- - ... TURLIN. branes and dynamically prepared polyvinyl pyridine membranes have similar kinetic properties, one concludes that our dynamic membranes are only a few times thicker than the active layer of the L-S membranes, We hope that with ! .. further work they might be made with comparable thickness. rejection by ion-exchange membranes; separation of some ions from each other - should be possible by hyperfiltration. With hydrous oxide membranes this should be particularly effective when the ions to be separated are co-ions of - - - - - different charge. - - - n e S ne i li jection with Cači2, MgCl2 and LaCl, rejections for various CaCl2-NaCI, NpC12- NaCl and LaC13-NaC1 mixtures, all containing 10-3 M HCl. A dynamic Zr02 xH20 membrane deposited on a carbon tube was used. The results listed were oo- tained from observed rejections by extrapolation to v/u•75 * 0. Pressure was 800 psi, which in this case was sufficient to make § almost equal to Res. The polyvalent salts are much better rejected than Naci. The effect is particularly marked for the trivalent salt, LaCiz; Reaci, is almost 0.9 even in : 0.5 M Naci. Lower, but still interesting, rejection occurs with the alkaline earth chlorides. With CaCl2 rojection is still almost 50% in 1 Naci. The implication is thus strong that with dynamic membranes it should be possi- A- inutes to the other one m some .... .. rathi. ble to carry out significant water softening. Unfortunately the hydro:is Zroz membrane used here probably would not be satisfactory for this purpose because mpe: 19.12.2017 :ters which one micht want to soften contain a him enoleri sı:1- - . ' fate concentration to affect seriously rejection proncrtins for all sn?. .. - - . -- - ..in . 0 0 0 0 0.0 · OOOOooo 0.0 -16. -- - With high flux (dynamic) membranes salt separations (or water softening) -- ...... can be enhanced by operating at lower pressures, (With conventional membranes this might also be possible in principle, though actually less attractive be- cause fluxes may become too low at low pressures.) We pointed out with Eq. (2) that salt rejoction is strongly dependent on BO the kinetic term eº (or e" for b=1). When:o is of the order of unity at a given flux, rejection will decrease rapidly with decreasing flux' (or pres- sure). The parameter o involves the inverse of the diffusion coefficient on tho salts in the membrane; these would differ for different salts. Thus at a given flux and membrane thickness values of o for different solutes may di fer substantially from each other. The opportunity thus presents itself to select a (low) pressure where for one salt, A, RA is still near Baco, while for the other onc, B, RR is much less than Rods This kinetic tochniquo' of separation should be particularly useful if Bac is also larger than RR.- i.e., if salt A is intrinsically better rejected than salt B. With a dynamic hydrous Zr0, membrane this situation seems to apply for LaCl3-Naci mixtures. Fig. 11 shows that preci, is insensitive to pressure, while Ryacı decreases, and the separation of Laciz from Naci improves as the pressure decreases. SSIO Rejection of Organic Materials: While ono expects rejection of salts by hydrous oxide membranes, it might be considered surprising that they reject orqanic materials as well. Our experiments on rejection of organic solutes have not been particularly extensive; but a few typical results with hydrous zirconiam oxide are given in Table V. Whilo low molecular weight organic solutos are at least partially re- jected, these membranes seem more effective with high molocular weinit main. . -17 rials. Thus in the ethylene glycol-polyethyleneglycol (PEG) series, rejec- tion increased from ca. 20% to 90% as the nominal molecular weight was in- . - - - - - creased to 6000. Since rejection seemed to level off at 90% with this particu- lar membrane, one wonders whether the permeating solute is from the low - . . . . tion), or if the residual organic content measures imperfections in the mem- . . . TAL . ... - .. . - . . -.. intenties -.-. . t -.- - - - ... - - traiteme -- - de - e - - -. - brane. We believe that both effects occur. Turbidities of solutions prepared from the effluent of the PEG-4000-6000 solutions were a little more than half those made of the original feed, measured at the same organic concentration. But rejection of a PEG with average molecular weight greater ihan 100,000 was also only 90%. Further, as discussed below, rejection of various sugars can be increased by exposure of the membrane 1:0 hig'er pressures. . Some rejection results for the series glucose, sucrose, raffinose are given in Fig. 12. These results were obtained by G, H. Gizinski, L. L. Gasner, and C. K. Neulander of the Massachusetts Institute of Technology Practice. School at ORNL (S, M, Fleming, Director) with a new, and presumably slightly different, Zr(IV)-oxide membrane, formed on a 0.2w silver frit; the solutions contained 1004 M Zr(IV). Before experiment (1), fig. 12, the membrane had been exposed for ca. 2 hours at 400 psi to .04 M NaCl and salt rejection was Robs - 0.5. In a series of experiments at 400 psi (experiments 1-4) lobs in- creased in the order glucose-sucrose-raffinose; Robs was in the range 15 to 35%. On increasing pressure to 1000 psi, sucrose rejection rose gradually to 48% and flux decreased to 60 gpd/ft?; permeation rate per atmosphere became much lower. After 3 hours, rejection of the sugars was in the range 35 to 60% depending on molecular weight (experiments 6-8, Fig. 12). After exposure to 1000 psi, rejections at 400 psi were essentially the same as at the higher pressures. Exposure to the higher pressure seemed to have improved rejection an! decreased! .. - . - - . sitiven si misteritiewicemina concessidad .. - - ... - . - - - - - --.. ..- - . . . - todas the rest international del e Sine :,*-**7 tritie site rencontr ce panorama N . -18. : . . . - U17- permeability; the changes are apparently only slowly reversed on returning to lower pressure.. Rejection of Nači did not closely reflect the changes in organic rejection: Brace was in the range 44 to 57% from before experiment i to after experiment 12. Since rejection of neutral organic solutes appears to depend primarily on size, we presume that the effect of pressure on sugar re- jection is the result of compaction of the colloidal hydrous oxide making up the membrane. Since Aracı did not vary much, one may conclude that "pores" or "interstices" removed by compacting were originally small enough not to affect significantly the volume charge density of the membrane, which presum- ably determines salt rejection. We have done so far very little on testing other dynamic membranes for rejection of organics. However, we found that a bentonite-humic acid membrane previously formed on a 0.84 silver frit rejected 75% of sucrose from a 50p/l solution. Permeation rate was 20 gpd/ft? at 2000 psi. ... We believe that the main significance of these experiments is that they suggest that dynamic membranes may find use not only in salt removal but also in more general pollution control, where removal of organic contaminants may be the principal objective. . .. min.- IT .- : : .-. - - r -. : . - . - . Self-Rejecting Dynamic Membranes: Perhaps one of the most fascinating aspects of dynamic membranes is the ability of certain colloidal materia ls to form dynamic membranes which then prevent further significant penetration of the same material from the feed stream. In this case we possibly are dealin? more . with ultrafiltration with dynamically-formed membranes rather than hyperfiltra- tion or salt filtration, although both of course can occur simultaneously. We may illustrate this effect with the following cxamples, Over a period of time, Naci rejection of a Th (IV) membrane formed from trening - . -. . - -. . -18- permeability; the changes are apparently only slowly reversed on returning to lower pressure.. Rojection of NaCl did not closely reflect the changes in organic rejection: fracy was in the range' 44 to 57% from before, experiment i to after experiment 12. Since rejection of neutral organic solutes appears to depend primarily on size, we presume that the effect of pressure on sugar re- jection is the result of compaction of the colloidal hydrous oxide making up the membrane. Since Bijacı did not vary much, one may conclude that "pores" or "interstices" removed by compacting were originally small enough not to affect significantly the volume charge density of the membrane, which presum- ably determines 'salt rejection. We have done so far very little on testing other dynamic membranes for rejection of organics, llowever, we found that a bentonite-humic acid membrane previously formed on a 0.84 silver frit rejected 75% of sucrose from a 50g/8. -- - - - - - - . . . . - - We believe that the main significance of these experiments is that they - - · suggest that dynamic membranes may find use not only in salt removal but also .. in more general pollution control, where removal of organic contaminants may be the principal objective. ... .. . .. Solf-Rejecting Dynamic Membranes: Perhaps one of the most fascinating aspects of dynamic membranes is the ability of certain colloidal materials to form dynamic membranes which then prevent further significant penetration of the ........ - - - - same material from the feed stream. In this case we possibly are dealino more · ** - - with ultrafiltration with dynamically-formed membranes rather than hyperfiltra- tion or salt filtration, although both of course can occur simultancously. We may illustrate this effect with the following examples. Over a period of time, Naci rejection of a Th (IV) membrane formed from - * - -. . . . .. . . - - - - - --- composer __ . : .. movement . . 4 -19. - --- a 0.02 M NaC1-0.002 M THC14 solution reached ca 70, but Th(IV) was eventually rejected better than 99%!5) . A similar effect may be observed with solutions containing Fe (III). Within a few minutes of passing a yellow Fe(111)-containing solution across a porous support, flow through it decreases substantially and iron rejection begins. Apparently an Fo(III) hydrous oxide membrane is formed which retards further passage of Fe(III). High values of Fe(III) rejections (more than 95%) - - - - - . - .. - . .. - - have been observed. . - -- Another example was mentioned in our discussion of salt rejection by . .. so.... -- -- - humic acid: Judging from the optical densities of feeds and products, the humic acid was better than 95% removed. -- - - - - - - - . . . A perhaps more dramatic case of formation of a self-rejecting membrane is given by some experiments we have carried out on the removal of organic pollutants from paper mill wastes. (We are indebted to Drs. A, J. Wiley and B. F. Luock of the Pulp Manufacturers Research league for supplying us with - - - - - - - - samples of paper mill wastes. The experiments were carried out by J. J. Perona, ORNL Chemical Technology Division and Mr. F.' 11. Butt, International Atomic Energy Agency Fellow at ORNL.) - --- -- . - .. - - They examined a "spent sulfito liquor" which was obtainod as a highly - -- - - -- colored concentrate containing 50% solids. The dissolved solills are largely calcium lignosulfonate although there is also a considerable fraction of low . --- . ---.. - . molecular woight carbohydrates, down to hexoses and pentosos. Our studies -- -. - - were with solutions prepared from this stock diluted to 0.5 to 5%. These solu- - . - - ... tions were passed through carbon tubes in the hope that calcium limnosulfonates - or other materials of the food would form a self-rejecting membrane. In addi- - - tion, we have carried out cxneriments with addcil hverous Zr!TV) oxicle. I?10 ro. sults have not been very reproducibir so far but we have ohtaincil relatively abilit ed the favorable results in some cases, Thus, with a paper mill waste solution containing 1% solids and no further additive, a membrane giving 80% color rejection was formed. Product flux was 87 gpd/ft2 at 400 psi. At the high circulation rates used, rejec- tion and flux remained relatively constant over a 7-hour period. · Color re- jection was obtained through optical density measurements at 281 mu where an absorption maximum is located. In this case we almost surely are dealing with a self-rejecting membrane. Presumably the lignosulfonates become concentrated at the interface and retard penetration by further lignosulfonates and other solutes of the feed. It is not clear why the upper limit of rejection in these experiments were in the vicinity of 80% though we suspect that more homogeneous supports than the carbon tubes rised might yield more favorable results. We suspect that the lignosulfonate membrane has a relatively low yield strength with respect to pressure and that penetration of little- . altered solution occurs through imperfections in the support. in support of . win- Ve-. -. ... ...we L-s, vivem- se a termine the insta this, we found in later experiments that the permeability (flux per unit pressure) increases rather abruptly near 400 psi pressure when similar car- - --- - -- -- -- - - in bon tube supports were used; simultaneously with this increase in permeability, there was rapid deterioration in the quality of the product. Color rejection could be improved through addition of small amounts of colloidal hydrous zirconium oxide. Thus, in the experiment described above enough colloidal Zr(IV) oxide was added to make the solution 10-3 11 in Zr(IV); color rejection rose to 93% although flux dropped to slightly less than 5n apd/ft2, in another experiment a Zr(IV) oxide membrane was first slynamically formed on a new carbon tube by circulating a .02 M NaC1-10-3!1 Zr (IV) solution through it. The chloride rejection became 60% and flux 34 zpd/ft2. 19en the solution was replaced by a 1% spent sulfite liquor, color relection of 98,5 ... - . - com*** -21- was obtained at a flux of 10 grd/ft? at 400 psi pressure. WS Conclusion: le have found that salt-rejecting membranes can be formod dynamically on porous supports from appropriate additives in the food solina tions. Since the choice of additive and support scoms almost unlimited, this technique may find use for the study of membrane materials and, hone- fully, in all types of water treatment applications. In viow of the fact that self-rejecting membranes may also be formed dynamically, there may also be applications in many other typos of problems, including food processins, where combination of hyperfiltration, ultrafiltration, and ordinary filtra- tion of particulates is desirable. One, of course, wonders if membranes of this type are also formed in nature or if the technique can eventually be applied to natural formations to control their general filtration and purifi. cation properties. re Acknowledgment: We are greated indebted to Warren G. Sisson, C. Gary' Westmoreland, and Neva Harrison for valuable technical assistance; and to Dr. Willis H. Baldwin for vaiuable discussions concerning organic addi- tives and their preparation and purification. Tablo I . 2 Typical Additives which Can form Salt-Rejecting Membranes - . . . - . us r Suu 0 Hydrous Oxides (Al(111), Fe(111), si(IV), Zr(IV), Rh (IV), U(VI)) Fincly Ground Low Cross-Linked Ion-Exchange Rosins Clays (Bentonite) Ilumic Acid Poly (Styrone Sulfonic Acid) (a) Poly (Vinyl Benzyl Trimethyl Ammonium Chloride) A) Cellulose Acotate Hydrogen Phthalatec) Cellulose Acetate N,N-Diothylamminoacetate() Poly (Mothyl Vinyl Ether/Maleic Anhydride) (Gantrez AN) . Poly (4-Vinyl Pyridine) i · Poly (4-Vinyl Pyridinium Butyi Chloride) (c) Poly (Vinylpyrrolidone) per intentional there are basic onlin W (A) we are indebted to Dr. R. E. Anderson, now Chemical Co., for thesc ma crials, (b) we are indebted to Dr. R. M. Fuoss, Yale University, for supplying us with : this material. (We arc indebted to Dr. W. H. Baldwin, ORNL Chemistry Division, for synthusis and/or purification of thesa materials, . Sr -.-.,.· Table II - . Typical Porous Supports for Dynamically formed Membranes Trade Name or Supplier Selas Corp. - - - - - . . - Nominal Pore Material Diameter (Microns) Silver 0.2 to 5 Porcelain 0.5 Sintered glass ultrafine Carbon 0.2 - 0.5 Polyvinyl chlorido 0.45 (on nylon) 0.45 Vinylidene fluoride 0.45 Cellulose acetate 0.1. 0.5 Porous filerglas tubes Union Carbide Com Metricel Vito (Gelmar Vinyl Acropor VNM-450 16c!man) Metricel VF-6 (German) Metricel GA-1 (Gelman) Millipore American Standard (6) Havens Industries (6) General Electrice) Nuclepore 0.5 ; .. . ... ... - - - (a) We are indebted to Dr. L. M. Litz of the Union Carbide Corporation for supplying us with a large variety of carbon tubos. (We are indebted to Dr. Alfred Shaines of American Standard and Dr. G Havens of Havens Industries for samples of their glass-fiber tubes. (C)We are indebted to Dr. of General Electric Corporation for samples of Nuclepore membranes, i . . . - - - - ht . . . Tablo ?!! Salt Rejection by Hydrous Zirconium Oxide Mombrancs (Porous Carbon tubos. 'V: 0.05 to 0.4 cm/min. 400-600 psi. 25°c) Salt Conc. (M) PH ell Naci MgCl2 BaCl2 LaCl3 0.05 0.027 0.025 0.029 3-4 3-4 3-4 3-4 Naci Na2504 BaCl2 0.05 0.025 0,025 w11,5 ~11.5 11,5 - - Table IV . : · Separation of Salts by Hyperfiltration (Dynamic Zr02 ·XI120 Membrane, 25 °C, 10-3 M HC1, 10-4 y Zr(IV), 800 psi) - - - - --- -- Water Flux (cm/min) - - .98 .19 - .88 .15 -- .19 Solution BNaI LAICIx .017 M LaC1z - .05 M Naci .70 .017 | LaCiz..5 M Naci .. .35 .054 M MgC1, • .12 M Naci .52 .87 .052 M MgCl2 • .52 M NaCl ..048 M CaCl2 - .11 N Naci: .57 .046 M. Caci, -1.02 M Naci .14 .045 M Rac12 - .01 M HCI 03(HCI) .96 --- .27 .65 .84 .49 - - .24 - . i .48 -- - - -- - - - - - - Tabic V . . . . . . .. .. 50 Rejection of Some Organic Materials by Hydrous Zr(IV)-Oxiile Membranes [25 °C, 35 at., 0.2u-Silver Frits, ca. 10-4 M Zr(IV) 1 Concentration Pressure Organic Solute grams/liter (at) gnd/ft? wobs Phenol 105 230 .10 n-butanol 105 150 Ethylene Glycol 60 Diethylene Glycol 60 PEG-300(a) 50 PEG-1000 PEG-4000 PEG-6000 50 ..... 23 .. -. • 26 - ... .52 - 45 40 - - .89 .. . (a)PEG refers to'polyethylene glycol", polymeric ethers of etilylene plyco!. .. . .. . ... -.. - .... .. .. .. . - - - -- -- -- - - Bibliography.. --- - - --- -- - - - Do - . - (1) C, F, Reid and E. J. Breton, J. Appl. Polymer Sci. 1, 133 (1959). (2) S. Loeb and S. Sourirajan, University of California (Los Angeles) Dept. Eng, Rpt. 60-60,(1960). (3) Annual Reports on Saline Water Conversion , Office of Saline Water, W. S. Department of the Interior, (4) W. H. Baldwin, D. L. Holcomb, and J. S. Johnson, J. Polymer Sci. A3, '833 (1965). - . - - - ... - - - - -- - - © A, Ei, Marcinkowsky, K. A. Kraus, H, O, Phillips, J. S. Johnson, and A. J. Shor, J. Am. Chem. Soc. 88; 5744 (1966). K. A. Kraus, 11. 0. Phillips, A. F. Marcinkowsky, J. S. Johnson, and A. J. Shor, Desalination 1, 225 (1966). K. A. Kraus, R. J. Raridon, and W. H. Baldwin, J. Am. Chem. Soc. 86, 2571 (1964). - - - Ĉ -.-.- -.. --- ---- --- (8) - - J. S. Johnson, L. Dresner, and K. A. Kraus, Chapter 8 in "Principles of Desalination", edited by K. S. Spiegler, Academic Press, New York (1966). - - --- --- - - - mi - . . . ::? - ... - - Ni - -- (9) H, K, Lonsdale, U. Merten, and R. L. Riley, J. Appl. Polymer Sci. 9, 1341 (1965). (10) R. J. Raridon, L. Presner, and K. A.' Kraus, Desalination 1, 210 (1966). (11) G. Scatchard, J. Phys. Chem. 68, 1056 (1964). :: (12) L. Dresner and K. A. Kraus;" J. Phys. Chem. 67, 990 (1963); L. Dresner J. Phys. Chem. 69, 2230 (1965). (13) T.: K. Sherwood, P. L. T. Brian, R. E. Fisher, and L. Dresner, Ind. Eng. Chem. Fundamentals 4, 113 (1965). (14) A. J. Shor, K. A. Kraus, J. S. Johnson, and W. T. Smith, submitted to Ind. Eng. Chem. Fundamentals. (15) D, G, Thomas, Ind. Eng. Chem. Fundamentals 6, 97 (1967). (16) R. J. Raridon and K. A. Kraus, unpublished. - - --- - - . . ..' - - - - . - - - - - -.- - . - - - cca. - " m t . . .. .. ... . . ORNL-DWG. 67 - 1168A PRESSURE GAUGE HI-LOW PRESSURE SWITCH .. ... . .- - -- - --.-- - -- . - • . - - - - - AOW CONTROL VALVE HEAT ! EXCHANGER L O TEMPERATURE NOICATOR PRESSURE CONTROLLER I ROTAMETER BYPASS HIGH PRESSURE ROTAMETER : RECIRCULATING , PUMP na tech Pumps Cooling Water. TUBULAR TEST SECTION CICI -. REUEF VALVE -PRODUCT COLLECTION PAN :. DRAIN . . . RESERVOIRO II ....... PRESSURIZING PUMP ..... - - ? 1 PROOUCT ROTAMETER Hyperfiltration System Schematic Fig. 1 . ---- ---- - -- - - . -. . - - - - - ... .. --. . ... . .._ -... --., ..- - ... . . - - Ximena ORNL-DWG. 67-2656 . .8 $904 / 18904-1). Robs 0.0001 0.0002 (V/.75) (cm/sec).25 EFFECT OF CIRCULATION VELOCITY. ON OBSERVED REJECTION (Hydrous Zro, Membrane, 0.013 M NaCI, 800 psi, v~0.33 cm/min, 25°C) - - - .. Fig. 2 - - -- - - - - ...------- .... ... ........ . ............. .... M . io.. e Sansowani *** - e * r a t -- -- - --- - -- - - - - - - .-. . . . ;-.. - . . ... - . - . . . . . - . . . . ----- - - - - -- - - .. . 1 - Rosa Rois TTTTTT R. OBSERVED REJECTION (%) 1001 TO NO TURBULENCE PROMOTER O DETACHED TURBULENCE PROMOTER 0 2 4 6 8 10 12 14 I 16 I 20 18 2 t CNROA X 109 Fig. 3. Effect of Detached Turbulence Promoter on Rejection of 0.01M MgCl2 at 400 psi. v- (No Turbulence Promoter) = 0.43 cm/min = 150 gal/ft2 day v- (Turbulence Promoter) = 0.53 cm/min = 190 gal/ft2 day Ciri. - OLUN 14 SIC . P ORNL- OWG. 66-5662A . --- - - - --- - * - i. . o _ + . -- V AV oitotilitian N A TIL A - S CHLORIDE,REJECTION 3%, --- ... - - Vatra 1 . A D Ure 1. . 11. 1. 1 u . T buyen- minem 0.01 0.02 0.05 0.10 - . - - LION in the demandes sobre todo MEATION 2013 . come 2 adol VT . wil..'1 " at Hyperfiltration by a Dynamically created Hydrous Zr(IV) Oxide, Membrane. (Base filter: silver 0.8 micron nominal pore diameter. Pressure 35 ATM.) Curve Calculated for Ideal Ion Exclusion of Ion-Exchanger of capacity: 0.09 Equivalents/kg Water" in Membrane Phase 2 3 O PO . D ........... - - - - - - -- --- --- -- -- - - -- ORNL-DWG. 66-8991 A +0.25 -0.50 0.5 o o : : 0.75 70.90 (1-R) REJECTION (R) . -THEORETICAL CURVE FOR CAPACITY 0.066 moles /kg H2O 40.95 . .001 .01 05 . LuwL0.99 MOLALITY OF NOCI IN FEED Rejection of NaCl by a Dynamically- Formed Zr(W) Hydrous Oxide Membrane. Ion-Exchange Capacity Assumed: 0.066 moles/kg *** H2O. Computed Curve Includes Allowance for Variation of Activity Coefficients of NaCl in Aqueous Phase. . Fig. 5 . - - - - --- - - - - - - - - - - - -.-. - - . - - - - - --- - - - - - - - ... . . . . . . . . . . . NISLAV ORNL-DWG. 66-9243 TT 0.70 .. -CI UPTAKE TL. . REJECTION (18) ION UPTAKE ( moles/ kg solid) FIR)R(2-R)/(4- Not UPTAKE- 0.30 0.20 0.10 2 3 4 5 6 7 8 9 10 11 12 13 . . PH COMPARISON BETWEEN SALT REJECTION (0.05 m NaCI) BY MEMBRANE AND ION EXCHANGE PROPERTIES OF HYDROUS ZIRCONIUM OXIDE (f(R) - C/m for Ideal Donnan Distribution ) Fig. 6 1 .. - - -- ----- - - - - - - - - - - - -- --- - - - - - - - - - -- - - - - - - ---- - ... ------ - , . -.- .., .. . - T - 1: - - -- - - - - .. . - - - - . - - -- -- -- - - .-.- . .- . . . - - - . - . --.- . r . . - - - .... ...... · · - . - . .. . .. ... .. . . . . .. .. - . . .. .. . ·---.-nou, acon. - ---- . ORNL-OWG. 66-6276A . :PMP : Om -- - REJECTION, % corto-- X 0001 11200544 do 0.05 0.005 0.01 FEED CONCENTRATION, moles anion / liter Ta * PERMEATIO cm /min الامارا O oºooo . cº. GPD / 112 للللللل · HYPERFILTRATION BY MEMBRANE DYNAMICALLY FORMED ON POROUS SILVER (0.2 micron Nominal Pore Diameter) WITH BENZYLTRIMETHYL AMMONIUM CHLORIDE. PRESSURE, 34 Atm. SOLID CURVES COMPUTED IDEAL REJECTION BY ION-EXCHANGER OF CAPACITY OF 0.04 EQUIVALENTS/kg WATER IN MEMBRANE PHASE (Coupled Flow). • LoCl3, 0.6 ppm Additive o MgCl2: 06 ppm Aaditive O Naçi, 0.6 ppm Additive O NOCI, 2.5 ppm Additive : Noci, 25 ppm Additive + HCI, 0.6 ppm Additive X NOCIO4 ,0.6 ppm Additive Fig. 7 . -. . . . . . . . ... . .. ... - - - ---. . - -.- . - - ... - - - ... - -. - . Kon weine UiA. . tention rise- me promene -- --- ORNL-DWG. 66-8303B Na2SO4 REJECTION (%) NaC1 MgCla 0.005 0.01 0.05 FEED MOLARITY, MONOVALENT ION 0.1 0.5 . PERMEATION (cm/min) o o gal per day/f72 - Hyperfiltration Properties of Dynamically formed Humic Acid Membrane. 0.2M Ag Filter, 1000 psi, ~ 25 ppm Humic Acid. Fig. 8 . ... ... ... ... ... ... . . . XIN 2. ORNL-DWG. 66-8302B No2S04 REJECTION (%) Naci : Na2SO4 20 • NOCI cm MgCl2 A MgCl, Inn 0.005 0.5 0.01 0.05 0.1 FEED MOLARITY, MONOVALENT ION PERMEATION (cm/min) gal per day/ft? Hyper filtration Properties of Dynamically formed Bentonite - Humic Acid Membrane (Full Points). Curves Are for Humic Acid Membrane. 0.8 M Ag Filter, 2000 psi, ~25 ppm Humic Acid. Fig. 9 - -,. ... - ...---. -.--- ---- ----- Y- - - - 2- . ORNL-DWG. 67-2654 WR4YOTA JORNL - DWG. 67-2654 --- tot . A -Naci Sot ------ REJECTION, Robs Na, so: ....... .. - - ... - . ... ----------... FLUX, V, cm/min TTT FLUX, gpd/ft2 --!* wie 10 inimeserva SALT REJECTION BY A DYNAMICALLY FORMED POLY (4-VINYLPYRIDINE) MEMBRANE ( 25°C, 1500 psi, PVC Support - 0.454) o 0.05 M Naci --- 0.01M Na2SO4 .1 0.025 M MgCl2 + "Coalinga-Type" Brackish Water Fig. 10 . . . . . .. .. .. - .- .. . -- --- -- - - -- - ---- - ORNL-DWG. 67-2655. B Y 4 . >.017M LACIz-:05M NOCI .017M LaClz-50M NOCI SALT REJECTION, R RNO | - - . . . PERMEABILITY cm min-1 at.-4 200 400 600 800 od morning • PRESSURE (psi). cm ww:: 16 SEPARATION OF Lat++ AND Na+ BY HYPERFILTRATION ..wi k ia . ... .............. ;;,?, ,,,, 1.., :. . ..:. ORNL-DWG. 67-231 À 1000 psi O 400 psi A LKUCHY B - - -- - - i - - - -- - REJECT Ol. -- . ... .-- _ - .. - - - . -.- .. . ------- - - - -- --- GPD /ft?.. 0.01 F cm min' PERMEATION cm min' atm' 0.008 0.006- $ 0.004+ o N 0.002 Rejection of Glucose, Sucrose, and Raffinose by hiydrous Zr(V) Oxide Membrane on 0.2 M Ag Filter. Solutions Contain About 50 gms / l of Sugar and Are 10°5 M in Zr(V). Numbers Indicate Order of Experiments. Fig. 12 END DATE FILMED 5 / 26 / 67 **!