GENERAL ARTICLES Molecular Weight Distributions of Cationic

Jun 1, 1981 - of the active carbon as though a hole is being dug by the catalyst crystallites. Such a contact mechanism/pitting mechanism is commonly ...
3 downloads 0 Views 617KB Size
Id.Eng.

O

as shown in Figure 9, complete conversion was easily achieved at lower temperatures such as 180 and 350 "C, respectively, without any carbon consumption. Therefore, this catalyst system could be recommended for practical applications such as emission control of small-scale sources or vehicles by the catalytic surface reaction or the gasaolid reaction.

u 100

200

TEMPERATURE

Literature Cited

300

("C)

Figure 9. Catalytic reduction of NO with Hzor CO over the composite catalyst: SV 20000 h-l; (0) 1.8%NO + 4.3% H,; (A)1.8% La2034.7%Pt. NO + 4.3% CO; catalyst, 4.6% ( 2 ~ 2 . 5 %

This fact indicates that the reaction proceeds from the surface in contact with the catalyst crystallite into the bulk of the active carbon as though a hole is being dug by the catalyst crystallites. Such a contact mechanism/pitting mechanism is commonly accepted in the catalytic oxidation of graphite, which was studied applying in situ electron microscopy technique (Baker et al., 1976). When this composite catalyst, Co-La,O,-Pt supported on active carbon, was used for the steady-state surface reaction of NO with H, (reaction IV) or CO (reaction V) NO + Hz -+ '/2Nz + HzO (IV)

NO + CO -+ '/2N2

59

Chem. Prod. Res. Dev. 1982,21, 59-63

+ COZ

Adihart, 0. J.; Hindin, S. G.; Kenson. R. E. Chem. Eng. frog. 1971, 67(2), 73. Baker, R. T. K.; France, J. A.; Rouse, L.; Wake, R. J. J. Catal. 1976, 41, 22. Dwyer, F. G. Catal. Rev. 1972, 6, 261. Inui. T.; Funabiki, M.; Suehiro, M.; Sezume, T. J. Chem. Soc. Faraday Trans. 11979a, 75, 787. Inui, T.; Ueno, K.; Funabiki, M.; Suehiro, M.; Sezume, T.; Takegami, Y. J. Chem. Soc. Faraday Trans. 11979b, 75, 1495. Inui, T.; Funabiki, M.; Takegami, Y. Ind. Eng. Chem. Prod. Res. Dev. I980a, 19, 385. Inui, T.; Funabiki, M.; Takegami, Y. J . Chem. SOC.Faraday Trans. 1 IQeOb, 76. . - , 2237. .. Inui, T.; Otowa, T.; Takegami, Y. J . Chem. SOC. Chem. Commun. 1980~. 94. Katzer, J. R. "The Catalytic Chemistry of Nitrogen Oxides"; Klimisch, R. L.; Larson, J. G., Ed.; Plenum Press: New York-London, 1975; p 133. Moser, W. R. "The Catalytic Chemistry of Nitrogen Oxides";Kiimisch, R. L.; Larson, J. G., Ed.; Plenum Press: New York-London, 1975; p 33. Shelef, M. Catal. Rev. 1975. 11, 1. Voohoeve, R. J. H. "Advanced Materials in Catalysis"; Burton, J. J.; Garten. R. L., Ed.; Academic Press; New York, 1977; p 129.

___

Received for review June 1, 1981 Accepted November 6, 1981

(V)

GENERAL ARTICLES Molecular Weight Distributions of Cationic Polymers by Aqueous Gel Permeation Chromatography Irvln J. Levy and Paul L. Dubln' Clairol Research Laboratoty, Sbmford, Connecticut 06922

Aqueous size exclusion chromatography on nonionic semirigid gels (PW columns, Toyo Soda Corp.) has been successfully applied to a variety of industrially significant cationic polymers. Ionic effects, including polyelectrolyte expansion and Donnan equilibrium "salt peaks", were controlled with mobile phases of moderate ionic strength (0.1 or 0.2 M NaCI). Commercial poly(ethy1enimine) samples exhiblt broad bimodal distributions. Other samples characterized display symmetrical chromatograms and normal elution behavior. Comparisons of molecular weights measured by gel permeation chromatography (GPC) with expected values based on other methods suggest that separations are controlled chiefly by molecular exclusion and not strongly influenced by adsorption or partition.

Introduction Cationic polyelectrolytes have become increasingly valuable industrial chemicals in such important areas as water clarification, sewage sludge dewatering, paper processing, and other applications which involve flocculation. Current usage trends from the water treatment industry lead to a projected market of $120 million for these poly-

* Department of Chemistry, Indiana University-Purdue University at Indianapolis, Indianapolis, IN 46223.

mers by 1990 (Chem. Eng. News, 1981). Over 26 years ago Michaels and Morelos (1955) noted that increased polyelectrolyte chain length corresponded to greater flocculation of clay particles. The results of more recent investigations tend to support this observation. Tilton et al. (1972) found that flocculation of algae occurred with greater efficiency as the molecular weight of the flocculant, poly(ethy1enimine) (PEI), increased from 800 to 2000; however, no appreciable effect was noted at higher molecular weights. Anthony et al. (1975) reported a strong inverse relationship between molecular weight and 0 1982 American Chemical Society

60

Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 1, 1982

optimum polymer dosage for poly(acry1amide) in the flocculation of domestic sewage. Lindquist and Stratton (1976) found that adsorption and flocculation of colloidal silica was enhanced with increasing molecular weight of PEI and Treweek and Morgan (1977) reported a similar effect in the aggregation of E . coli by this polymer. Gregory (1973) found that the flocculation rate of latex particles increased with the molecular weight of quaternized poly(dimethylaminoethy1 methacrylate). Dubin (1980) observed that increased molecular weight of poly(vinylamine hydrochloride) reduced the optimum polymer dosage for the flocculation of silica suspensions. Clearly, strong evidence exista linking polymer efficacy to increased molecular weight. In principle, the results of such experiments should guide in the development of theories for polyelectrolyte flocculation; indeed the various models for polymer flocculation, e.g., polymer bridging (Kitchener, 1972) and patch formation (Gregory, 1976), differ appreciably in their prediction of molecular weight effects. Unfortunately, a lack of detailed and accurate molecular weight data has obstructed such interpretations. Absolute molecular weight methods such as osmometry and light scattering are time consuming and difficult (especially for polyelectrolytes) and molecular weights reported by suppliers are often tenuous. Also, the broad polydispersity of many commercial samples, not detected by these methods, can complicate the results. In contrast, liquid exclusion chromatography (GPC) is an experimentally simple technique which rapidly yields not only average molecular weights but, in principle, also provides molecular weight distributions (Yau et al., 1979). It should be noted that GPC is not an absolute molecular weight method and is valid only when separations caused by nonsize exclusion mechanisms are negligible. Also, a careful analysis of GPC results must give special attention to the relationship between the molecular volume and molecular weight of the polymer. No class of polymers has been more problematic in aqueous GPC than polycations. A wide range of solutes may be chromatographed on polyacrylamide or crosslinked dextran gels (Determan, 1969),but the low packing efficiency and poor mechanical strength of these gels preclude high-speed, high-resolution analyses. Inorganic substrates such as porous glass are pressure insensitive and can be packed with moderate efficiency (ca 500 plates ft-'), but untreated glass absorbs proteins, polycations, and even nonionic hydrogen bonding polymers such as poly(ethy1ene glycol) and poly(viny1pyrrolidone) (Frenkel and Blagrove, 1975; Regnier and Noel, 1976; Hiatt et al., 1977; Dubin et al., 1977). Substrates prepared from soluble silica (e.g., SEC columns (Du Pont), Lichrospher (Merck)) exhibit far greater efficiency than glass but are even less suitable for aqueous chromatography because of their adsorptive behavior and gradual dissolution (Barker et al., 1979). Surface treatment, based mainly on versatile and well understood silane coupling reactions, has provided a means of reducing these adsorptive effects (Engelhardt and Mathes, 1977). Good separations have been reported using commercial packings of this type-e.g., Aquapore (Chromatix), SW (Toyo Soda), Glycophase G (Pierce)-for nonionic synthetic polymers, proteins, and polysaccharide derivatives. However, they appear to be unsuitable for the analysis of polycations, possibly because of the presence of residual underivatized acidic silanol groups (Crone and Dawson, 1976). Two reports have appeared on GPC of polycations using anion-exchange substrates-either quaternized Styragel (Butler, 1976) or porous glass or silica covalently bonded

to quaternary amine functional groups (Talley and Bowman, 1979). Such packings may be especially sensitive to ionic effects and, at any rate, are not commercially available. In this report we present findings with PW columns, (Hashimoto et al., 1978a) the packing for which is a cross-linked polyether gel (Hashimoto et al., 1978b). In some respects PW packing may be comparable to other hydrophilic cross-linked semirigid gels based on poly(2hydroxyethyl methacrylate) (Coupek et al., 1973) or poly(acryloylmorpho1ine) (Epton et al., 1974), but published data for these packings focus on nonionic polymers. Furthermore, these latter packings appear to lack the mechanical strength that enable PW columns to exhibit over 5000 plates ft-', at flow rates above one mL m i d and back pressures of several hundred psi (Hashimoto et al., 1978a,b;Toyo Soda Manufacturing Co.) PW columns have previously been used to elute, with minimal adsorptive effects, a wide variety of water soluble polymers including poly(ethy1ene glycol), poly(viny1 alcohol), poly(viny1pyrrolidone), poly(acrylamide), proteins, and chitosan (Toyo Soda Manufacturing Co.), all of which exhibit adsorptive behavior on one or more of the commercial columns described above. In this report we discuss the GPC analysis of several industrially important polycations, including poly(dimethyldially1ammonium chloride), poly(ethylenimine) and poly(methacryloxyethyltrimethy1ammonium methosulfate). Experimental Section Materials. Poly(oxyethy1ene)(PEO) was obtained from two sources, Dow (poly(ethy1ene glycol)) or Aldrich (poly(ethy1eneoxide)). Since these samples are reasonably well characterized and of narrow molecular weight distributions, they are suitable for column calibration. Polymeric cations were from a variety of sources. Poly(ethylenimine)(PEI) samples of varying molecular weights were from Cordova Chemical Co. Poly(dimethyldially1ammonium chloride) (PDMDAAC) was a commercial product, Merquat 100 (Calgon Corp.), or lower molecular weight homologues from the same source. A commercially available ionene polymer, Onamer M, was obtained from Onyx Chemical Co. Poly(methacryloxyethyltrimethy1ammonium methosulfate) (Q-100) and polymer T-100, a homopolymer of a tertiary amine hydrochloride, were kindly provided by R. Stratton (Institute of Paper Chemistry, Appleton, WI). Methods. The chromatography hardware consisted of a Milton Roy Minipump, a Rheodyne Model 7010 injector, a Rheodyne Model 7030 switching valve, and a Waters R401 differential refractometer. In initial investigations a single 30-cm G3000 PW (kindly provided by Bio-Rad Corp.) column was utilized; however because of the relatively low exclusion limit (ca. 60000 MW) of this column all later work was performed using a G5000 PW column in series with a G3000 PW (thus increasing the exclusion limit to ca. 1 million). Polymer samples were typically dissolved directly in the mobile phase, 0.1 or 0.2 M NaC1, and filtered through 0.45-c~mMillipore filters. In some cases the polymer solutions were dialyzed against excess mobile phase overnight. At a flow rate of ca. 30 mL h-' the pressure across each column was about 100 psi. Results and Discussion Selection of Mobile Phase. A supporting electrolyte in the mobile phase is essential to suppress polyion expansion which would otherwise make the GPC elution volume overly sensitive to polymer charge density and salt impurities. Careful selection of this electrolyte is necessary

Ind. Eng. Chem. Prod. Res. Dev., Vol. 6 4 10

5

6

I

I

7 I

8

9

I

I

21,

No. 1,

1982

61

10

MPE1:

1104

Io3

Figure 2. Chromatograms of commercial poly(ethy1enimine) samples Corcat P-12, P-18, and P-600; 1X G5000 PW + 1 X G3000 PW columns, 0.2 M NaCl.

10

10

12

14

16

18

20

Ve,ml

Figure 1. Poly(oxyethy1ene) GPC calibration curves. (0) 1X G5000 PW + 1 X G3000 PW columns, 0.2 M NaC1; (0)1 X G3000 PW column, 0.1 M NaC1. MPEo:

in order to avoid Donnan equilibrium effects which can lead to complex interfering salt peaks (Dubin and Levy, 1981). Since most of the polymers studied were chloride salts, sodium chloride was employed despite its corrosive action on the in-line fritted filter disks. A salt concentration of 0.1 M was sufficient to suppress Donnan effects if the sample load was low, e.g. 20 p L at 1 wt 9%. Calibration. The calibration curves obtained using PEO samples appear in Figure 1. In contrast to the other standards, the 100OOO molecular weight PEO sample was a broad-distribution product with = 1.0 X 105 reported by the supplier (Aldrich Chemical Co.) from light-scattering. The elution volume corresponding to MW-obtained by curve summation (Yau et al., 1979) from the chromatogram and a tentative calibration curve based on the other standards-was used to plot the data point for this sample in Figure 1. All subsequent molecular weights are reported based on these calibration curves as Mppm,the apparent molecular weight a t the peak elution volume; it is appropriate to comment on the relative nature of such values. Even if separations are controlled only by steric factors, GPC is not an absolute molecular weight method because retention volume is related to molecular size rather than mass. A standard calibration will be valid for any particular sample only to the extent that the sample and standard have the same relationship of molecular weight to molecular volume. To illustrate the variability of this relationship, one can contrast compact globular proteins with highly extended polyelectrolytes. Similar effects exist between branched and linear polymers of nearly identical chemical composition. A second complicating factor may be the dissimilarity in the molecular weight per unit contour length. Presumably because of this second effect, we find that MPm values obtained for other polymer samples, e.g., PEI, etc., are t y p i d y smaller by a fador of 2-4 than

I

I

lo6

lo5

I lo4

I

lo3

I '01

Figure 3. GPC fractionation of PEI Corcat P-600: A, Starting material, B, GPC fractions; 1 X G5000 PW + 1 X G3000 PW columns, 0.2 M NaCl.

expected values. GPC may be used to obtain accurate molecular weight data by calibrating with narrow molecular weight fractions of the polymer of interest or by monitoring the GPC effluent with a light-scattering molecular weight detector (Ouano and Kaye, 1974). In the absence of such methods useful comparisons may still be made. Characterization of Polycations. Commercially available PEI samples (Corcat P-12, P-18 and P-600) were injected as the free base at pH e 10. Measurements of the pH of the GPC effluent confirmed that the samples remained in this essentially un-ionized state throughout the chromatography. The chromatograms of Figure 2 are all bimodal, the effect being especially striking for P-600. To confirm that this bimodality is not an artifact resulting from adsorptive effects, a large sample (200 p L of 10 wt 9% solution) of P-600 was injected and the collected fractions were analyzed. The resultant peaks as shown in Figure 3 were seen to correspond to those of the initial chromatogram. This constancy of retention volume after 25-fold dilution is added evidence of the absence of partition or adsorption and thus provides further proof of the true bimodality of these samples. Narrow distribution PEI fractions, characterized by ultracentrifugation (Lindquist and Stratton, 1976) were used to calibrate the GPC columns in the molecular weight range 103 < MW < 105, and the corresponding Wnvalues are shown in the lower MW axis of Figure 2. The PEO calibration curve apparently underestimates the true h4W of PEI by as much as a factor of ten for Mpm > lo4. The actual difference in MW between the two observed peaks

62

Ind. Eng. Chem. Prod. Res. Dev., Vol. 21, No. 1, 1982

Table I. Expected Molecular Weights and Apparent GPC Molecular Weights (Relative to Poly(oxyethy1ene))

sample

a

mobile phase [NaCl], M

column(s)

MW (method)a

M,PEO

PEI P-12 0.2 1 x G5000 PW + 1 x G3000 PW 0.2 PEI P-18 1 x G5000 PW + 1 x G3000 PW PEI P-600 0.2 1 x G5000 PW + 1 x G3000 PW PDMDAAC-9 0.1 1 X G3000PW 0.1 PDMDAAC-24 1 X G3000PW PDMDAAC-30 0.1 1 X G3000PW LS = low angle light-scattering; VPO = vapor phase osmometry. 6,900

1600,1000 2000,1000 64000, 2900 1400 2800 4000

1200 (VPO) 1800 (VPO) 60000 (-) 9 x 103 (LS) 2.4 x 104 (LS) 3 x 104 (LS)

2800

;%

I

3-

d’---V L

rvl PEO

I

: to6

to5

lo4

o3

loz

Figure 4. Chromatograms of retention aid polymers (A) Q-lo0and (B)T-100; 1 X G5000 PW + 1 X G3000 PW columns, 0.2 M NaCl.

for P-600 is thus very great. The striking difference in calibration for PEI and PEO is almost certainly a consequence of the dense branching of the former-especially at high MW-which increases its molecular compactness and hence its elution volume. Commercial samples of PEI have been used to examine the role of molecular weight in flocculant efficacy (Tilton et al., 1972); clearly polymers with distributions that are broad and asymmetric, and overlap considerably, may fail to provide a firm basis for studies that focus on molecular weight effects. Variations in branch density of PEI samples may further obscure results, especially if branch density itself is a function of molecular weight. The analysis of molecular weight effects is more straightforward for the retention aid polymers &-lo0 and T-100which are linear high charge density polyions. As shown by the chromatograms in Figure 4, their molecular weight distributions are symmetrical and well separated. The retention efficacy of 0.05 % polymer in handsheets of beaten bleached softwood kraft pulp, as measured by loss of TiOz filer, increased by 50% when the higher molecular weight polymer was used (Stratton, 1981). Molecular weight, therefore, appears to be a factor in the efficacy of these polymers. Commercial polycationic flocculants are available from a range of synthetic routes with a corresponding variety of chemical structures. Thus the cyclopolymerization of dimethyldiallylamine affords PDMDAAC, while condensation coupling reactions yield low molecular weight ionenes such as Onamer M. As seen in Figure 5, separation of these polymers was accomplished with no evident adsorption. It should be noted (Table I) that apparent molecular weights differ considerably from expected molecular weights in the case of the PDMDAAC samples. These low values of MpPEomight be attributed to the adsorption of PDMDAAC, but one may establish through the “Universal Calibration” procedure (Grubisic et al., 1967) that separations are based on exclusion alone. Accordingly, we have shown elsewhere (Dubin and Levy, 1981) that the hydrodynamic volume, as represented by the product of intrinsic viscosity and molecular weight,

Figure 5. Chromatograms of low molecular weight cationic polymers: A, commercial ionene polymer, Onamer M; B, low molecular weight poly(dimethy1diaUyla”oniw.u chloride) (MW = 30000, ---; MW = 24000, -; MW = 9OO0, 1 X G3000 PW, 0.1 M NaCl. -e-.-);

governs the elution of PDMDAAC and dextran in a uniform manner. Strong adsorption of PDMDAAC thus appears unlikely, and the low MppEovalues of this polymer are better ascribed to its high MW per unit contour length vis-a-vis PEO. Conclusions The molecular weight distribution of cationic polymers can be determined by aqueous GPC using nonionic semirigid gel packings, commercially available as PW columns (Toyo Soda Corp.). Ionic effects, including intramolecular polyelectrolyte expansion and Donnan equilibrium “salt peaks”, can be controlled if the quantity of injected polymer is small relative to solvent ionic strength (i.e., 0.2 mg in 0.1 M NaC1). Comparisons of molecular weights measured by GPC with expected values based on other methods suggest that separations are controlled by molecular exclusion and not strongly influenced by adsorption or partition. Literature Cited Anthony, A. J.; King, P. H.; Randall, C. W. J . -1. Pdym. Sei. 1975, 19, 37. Barker, P. E.; Hiatt, B. W.; Holding, S. R. J . chrometogr., 1979, 174, 143. Butler, 0. B. US. Patent 3962208, June 8, 1976. Own. Eng. News, 1981, 59(22), 12. Gwpek, J.; W a k o v a , M.; Pokorny, S. J . po3rtn. Scl. C , 1973, 42, 185. Crone, M. D.; D a m , R. M. J . Chromntogr., 1976, 129, 91. Detem\ann. M. “Gel Chromatography", 2nd ed.;Sprhger-Verlag: New York. 1969. DUM, P. L.; Koontz. S.; Wright, K. L. J . f d y m . Sci., Chem. Ed.. 1977. 15. 2047. Dubin. P. L. US. Patent 4217214, Aug 12, 1880. Dubh, P. L.; Levy, I. J. po3rtn. Prepr. 1B81, 22(1), 132. En-, H.; MameS,D. J . chromatogr., 1977, 143, 311. Epton, R.; Ho#oway, C.; McLaren, J. V. J . Chromatogr., 1974, 90. 249. Frenkd. M. J.: BlEgrOve, R. J. J . Chromafog., 1975, 1 1 1 , 397. Gregory, J. J . CokM InWace Scl.. 1973, 42. 448. Gregory, J. J . CokM Interface Scl., 1976, 55, 35. 2.; Rempp, P.: Bendt, H. J . polvm. Scl. B , 1967, 5 , 753. HasMmoto,T.; Sasekl. H.; Alura, M.; Koto, Y. J . Cht‘otna~ogr..1976, 180,

m.

m, mi

Ha&-&, T.; &saki, H.; Alura, M.; Koto, Y. J . polym. Scl., f d y m . phvs. Ed.. 1978, 18, 1789. Hlatt, C. W.; Shebkov, A.; Rosenthal, E. J.; GaUimore. J. M. J . Chromarogr., 1977. 56. 362. Khchenir, J:A. Br. Pot)”. J . , 1972. 4 , 217.

03

Ind. Eng. Chem. Prod. Res. Dev. 1982, 21 63-68 I

Lindqulat, (3. M.; Stratton, R. A. J . cdkld Inferface Sei., 1076, 55, 45. Michads, A. S.; Morelos, 0. Ind. Eng. Chem., IOS5, 57, 1801. Ouano. A. C.: Kave. W. J . Pobm. Sd., &iytn. chem.Ed., 1074, 12, 1151. Regnk, F. E.; Nbel, R. J . chnmulfogr. S d . , 1076, 14, 316. Stratton, R. A., m a t e communlcatkn, Institute of Paper Chemistry, Appleton, WI, 1981. Tallev. C. P.; Bowman, L. M. Anal. Chem., 1070, 51, 2239. &, R. C.; Murphy, J.; Dlxon, J. K. Watsr Res., 1072, 8 . 155. co.,Lw., 4'TSKQELpw Type Columnsp,Technical soda Data Brochure, 1980. Treweek. Q. P.; Morgan. J. J. J . cdldd Inferface Scl., 1077, 60, 258.

Yau, W. W.; Kirkland, J. J.; Bly, D. D. "Modern Size-Exclusion Liquid Chromatography"; Wiley: New York. 1979; pp 318-322.

Received for review July 2, 1981 Accepted October 9, 1981 Presented at the 181st National Meeting of the American Chemical Society, Atlanta, Ga., Mar 29-Apr 3, 1981, Division of Polymer Chemistry, Symposium on Aqueous Polymer Systems.

Reducing Energy Requirement in Latex Concentration by Ultrafiltration Joseph J. S. Shen' and Leon Mlr Abcor, Inc., 850 Maln Street, Wllmlngton, Massachusetts 02173

This paper discusses the industrial application of ultrafiltration technology for removing water from both the in-process and waste latex streams. Due to the high level of permeate flux attainable with stable latices, the ultrafiltration process for latex concentration is becoming economical and practical on the commercial production scale. Recent process and product developments for latex ultrafiltration are presented, with particular reference to using tubular ultrafiltratlon membranes. Several technical parameters of interest such as flux characteristics, concentration polarization, latex stability, and membrane cleanlng are discussed. Performance data with PVC and SBR latices collected in laboratory and pilot plant studies are presented. Finally, a comparison of operating energy costs among ultraflltratlon and evaporative methods for PVC latex dewaterlng are tabulated. It is shown that the energy cost for the ultrafiltration process is only a fraction of that for the more energy-intenshrethermal evaporation processes.

Introduction The concentration of latex emulsions via ultrafiltration has long been recognized as a possible separation process by the workers in the field. There are two major industrial applications for this process. The first is the concentration of dilute latex from 0.5% to 25% solids or more. It is generally used as a pollution control measure for direct sewage discharge, but frequently for waste latex recovery for reformulation. The principal latices in this category are styrene butadiene rubber (SBR) and poly(viny1 acetate) (PVAc). The second major application is the concentration of in-process latex streams from 30% to above 50%, replacing the evaporator by the ultrafilter. Poly(viny1 chloride) (PVC) is the most important latex in this category. Under ideal conditions, latex particles in the range of 0.05 to 0.5 pm would be completely rejected by the ultrafiltration membrane, and high permeate fluxes would be obtained. However, efforts to commercialize ultrafiltration systems for industrial latex streams were, to a large extent, frustrated in the early years of ultrafiltration technology. Two major difficulties encountered in this endeavor were latex instability and membrane life. Latex emulsions must be stable for acceptable flux performance. However, the shear induced by the pumping action required in an ultrafiltration system may deleteriously affect an otherwise stable latex emulsion. The high solids concentration found in the gel layer on the membrane surface due to concentration polarization may also be detrimental to latex stability. When latex emulsions become unstable, coagulation of latex particles takes place and a compact foulant layer is formed on the membrane surface. Consequently, flux will fall off appreciably as the 0196-4321/82/1221-0063$01.25/0

foulant layer offers the limiting hydraulic resistance to permeation. In the severe cases of latex instability and membrane fouling, an entire flow passage in the membrane system may be plugged by the coagulated latex. Thus, it was necessary to be able to clean and reuse a fouled membrane. The tenacious fouling layer, once formed on the membrane sufrace, wils not easily amendable by ordinary cleaning techniques; however, the early generation ultrafiltration membranes (cellulose acetate type) could not tolerate the harsh chemical environment necessary for effective membrane cleaning. Consequently, no significant progress was made in the commercialization of the ultrafiltration process for industrial latex streams until the emergence of improved ultrafiltration membranes. Over the past few yeam, significant product and process developments in latex ultrafiltration have been made to overcome these two difficulties. A chemically inert polymeric (non-celluloseacetate) membrane (designated HFM membrane by Abcor, Inc., Wilmington, MA) was developed. Furthermore, a solvent cleaning technique was devised so that fouled membranes can be effectively cleaned and reused. In this paper, the performance data of HFM membranes with several industrial latex streams are presented. Several technical parameters of interest, such as flux characteristics, latex stability, and membrane cleaning are then discussed. Finally, the operating energy cost of applying ultrafiltration in PVC latex dewatering is estimated and compared to the costs of using the conventional thermal evaporative methods. Latex Ultrafiltration Performance In this section, some of the technical parameters of interest in latex ultrafiltration are examined. Some inter@ 1982 American Chemical Society