A linear relation between the cloud point and the number of oxyethylene units of water-soluble nonionic surfactants valid for the entire range of ethoxylation

A linear relation between the cloud point and the number of oxyethylene units of water-soluble nonionic surfactants valid for the entire range of ethoxylation

Journal of Colloid and Interface Science 260 (2003) 219–224 www.elsevier.com/locate/jcis A linear relation between the cloud point and the number of ...

108KB Sizes 2 Downloads 38 Views

Journal of Colloid and Interface Science 260 (2003) 219–224 www.elsevier.com/locate/jcis

A linear relation between the cloud point and the number of oxyethylene units of water-soluble nonionic surfactants valid for the entire range of ethoxylation Hans Schott School of Pharmacy, Temple University, Philadelphia, PA 19140, USA Received 6 August 2002; accepted 8 November 2002

Abstract The following linear equation correlates the cloud point (CP) of water-soluble polyoxyethylated nonionic surfactants (NSs) with the average number p of oxyethylene units per molecule: (p − p0 )/CP = a + b(p − p0 ). Here p0 is the smallest value of p that confers solubility in cold water: In a homologous series of NSs, it belongs to the surfactant with CP = 0 ◦ C. Plots of CP versus p for five representative homologous series of NSs consist of three segments: A steeply ascending, nearly straight line, a transition region that ranges from p = 15–22 to p = 20–28, and a nearly horizontal plateau that approaches asymptotically the CPs of polyethylene glycols with molecular weights between 30,000 and 4400. These CPs range from 113 to 130 ◦ C. Most CPs for NSs were taken from the literature or measured on commercially available samples; eight CPs above 100 ◦ C were measured on newly synthesized surfactants. Previously published linear equations correlating CP with p cover only NSs with p < 16 and CPs < 100 ◦ C: They apply only to the ascending segment of the CP versus p plots. Our equation covers the entire plots and applies to the full range of NSs, including extensively polyoxyethylated NSs with p  100. It can be used for selecting specific NSs for high-temperature applications. The hydrophile– lipophile balance of the surfactant with p = p0 oxyethylene units, namely, HLB0 , is a novel quantitative measure of the hydrophobicity of the hydrocarbon moiety of the relevant homologous NS series. Its value reflects the size, composition, and structure of the hydrocarbon moiety.  2003 Elsevier Science (USA). All rights reserved. Keywords: Cloud point versus number of oxyethylene units per molecule; Degrees of ethoxylation; Hydrophobicity of hydrocarbon moieties of nonionic surfactants; Limiting hydrophile–lipophile balance; Lower solubility limit of nonionic surfactants; Water-soluble nonionic surfactants

1. Introduction The cloud point (CP) of aqueous solutions of polyoxyethylated nonionic surfactants (NSs) has theoretical and practical importance. Due to their low critical micelle concentrations and high micellar molecular weights, NSs in solution at use levels of a few percent exist mainly as micelles, which resemble cross-linked polymer molecules. By analogy with polymer solutions, the CP is a lower consolute temperature approaching the Θ temperature. Reversible phase separation occurs at and above the CP because NSs are more soluble in cold than in hot water. Consequently, solid–water suspensions stabilized with NSs flocculate or coagulate when heated above the CP, oil–water emulsions break, and foams of NSs collapse. Therefore,

the CP is an important property for selecting NSs in the formulation of such disperse systems. To raise the CP within a category or homologous series of NSs, the hydrophile–lipophile balance (HLB) must be increased. This is accomplished by increasing the number p of oxyethylene units per molecule or, within limits, by selecting a hydrocarbon moiety with fewer carbon atoms n [1–3]. It is important to know by how much p must be increased within a NS category with a given n value to attain or exceed a required CP value. The purpose of this study is to correlate quantitatively p with the CP for four representative categories of NSs. For the sake of simplicity, the resulting relation should be linear. The following CP–p relations have been published: CPs were correlated in a linear fashion with log p [4,5] and with n [5] and in a nonlinear fashion with p [6] for various ho-

0021-9797/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0021-9797(02)00183-2

220

H. Schott / Journal of Colloid and Interface Science 260 (2003) 219–224

mologous series of NSs. A fourth study [7] used topological indices to characterize the hydrocarbon moieties of various surfactant series and combined these indices with p to arrive at an empirical universal relation for predicting CPs. The common shortcoming of these studies [4–7] is that they are limited to NSs with p < 16 and CPs  100 ◦ C. In that range, CPs increase steeply and monotonically with increasing p at constant n. However, when p values are increased beyond ∼16, the CPs rise above 100 ◦ C but level off below 128 ◦ C and approach the CPs of polyethylene glycols (PEGs) asymptotically (see Refs. [8–10] and the following). NSs with p values up to 100 are commercially available and find applications in industrial formulations. Extrapolating the equations of Refs. [4–7] to the range of p values from 16 to 100 leads to unrealistically high CPs. For instance, the CP of Igepal CO-997 (nonoxynol 100, actual p = 94) extrapolated according to Eq. (7) of Ref. [5] is 328 ◦ C, compared to the experimental value of 114 ◦ C (see the following). Many CP values of NSs have been published [1–10]. However, because of experimental difficulties, few were determined above the normal boiling point of the aqueous NS solutions. Two papers that extend the CP measurements well beyond 100 ◦ C [8,9] are completely overlooked in the literature [1–10]. In one of these papers, Matell described a simple technique: He sealed NS solutions inside meltingpoint glass capillaries and measured their CPs with a hot stage microscope [9]. Unlike measurements with larger volumes [11], he observed a slight hysteresis: The clearing temperature on cooling was usually 1–2 ◦ C lower than the clouding temperature on heating [9]. For the present study, published CP values [1–10] were supplemented by measuring the CPs of extensively ethoxylated commercial NSs as well as the CPs of eight NSs prepared in our laboratory to fill some gaps.

A molecularly distilled sample of Triton X-35 (p = 3) was further ethoxylated to p = 43, 74, and 90 in a stirred autoclave at 160–180 ◦ C and 2–3 atmospheres with a 0.7% NaOH catalyst. The values of p are generally calculated from the weight increase during ethoxylation and/or from the hydroxyl number of the products. To the extent that small amounts of PEGs contaminate the NSs, the p values calculated from hydroxyl numbers represent lower limits because these PEG impurities have generally lower molecular weights than the NSs and have two terminal hydroxyl groups compared to the surfactants’ one. A few percent of low molecular weight PEGs leave the CPs of the NSs largely unaffected [8]. Polyoxyethylated p-nonylphenols or nonoxynols were mainly supplied by Rhodia Inc. under the tradename Igepal CO [13] and by Berol AB [9]. The nonyl chain is a propylene trimer. Polyoxyethylated dodecyl alcohols were manufactured by Berol AB [9] and by ICI Americas Inc. under the tradename Brij [14]. We ethoxylated n-dodecyl alcohol to p = 15, 30, and 40 to fill gaps in that homologous series, using the conditions previously described. Polyoxyethylated tridecyl alcohols were supplied by Jefferson Chemical Co. [15], Cognis Corp., and Berol AB [9]. We further ethoxylated two polyoxyl 3.0 tridecyl ether samples to p = 30 and 40. Tridecyl alcohol is an oxoalcohol consisting mostly of tetramethyl-nonyl alcohols. The oleyl alcohol used for ethoxylation was 99.7% pure [16]. Two commercial ethoxylates were based on technical grade oleyl alcohol, which contains significant amounts of cetostearyl alcohol. Because their CPs differed by 11 and 14 ◦ C from the CPs of the corresponding ethoxylates of pure oleyl alcohol, they were omitted.

3. Methods 2. Materials All NSs studied were normally distributed because homogeneous NSs are only available with p < 9 and therefore have a limited CP range. Moreover, homogeneous NSs have higher CPs than normally distributed analogs with the same average p values, even though these CP differences become smaller with increasing p [1,2]. Two of the four surfactant categories investigated were polyoxyethylated alkylphenols because of their extensive use in research and manufacturing, and because they are available in a wide range of p values. Most polyoxyethylated p-octylphenols or octoxynols were supplied by Union Carbide Corp. under the tradename of Triton X [12], by the Swedish company Berol AB [9], and by Nippon Chemical Co. [8]. The terms “octoxynol” and “nonoxynol” (see the following), followed by the value of p, are used by The National Formulary and The Cosmetic, Toiletry and Fragrance Association. The octyl chain is an isobutylene dimer.

CPs were determined visually as the temperatures at which clear or slightly hazy surfactant solutions suddenly turned opaque on heating and clear on subsequent cooling. Each sample was subjected to three successive heating and cooling cycles. With the proper experimental precautions [11], the six CP values agreed within 0.2 ◦ C. Below 10%, CPs depend only slightly on surfactant concentration [17]. Our measurements were made at 4 or 5%, based on the weight of water. Two methods were used to measure the CPs of NS solutions that exceeded the normal boiling points. The first consisted in sealing aliquots in 5-ml glass ampules used for injectable dosage forms. The sealed solutions were heated and cooled at a rate of 10 min/◦ C in the vicinity of the CP. No hysteresis was observed. To ensure the absence of leakage, the ampules were weighed before and after the CP measurements. The second method consisted in lowering the CP below the normal boiling point by adding increasing amounts of

H. Schott / Journal of Colloid and Interface Science 260 (2003) 219–224

221

and nearly straight line, followed by a transition region and plateau. To linearize these curves, they were fitted to an equation analogous to the linear or reciprocal form of these two types of isotherms, namely, x = a + bx, y

Fig. 1. Cloud points of octoxynols as a function of the number p of oxyethylene units per molecule. Solid symbols indicate newly synthesized surfactants.

Fig. 2. Cloud points of polyoxyethylated dodecyl (circles) and tridecyl (triangles) alcohols as a function of the number p of oxyethylene units per molecule. Solid symbols indicate newly synthesized surfactants.

(1)

where x is the “excess” degree of ethoxylation p − p0 and y is the CP: Lower surfactant homologs, which are insufficiently ethoxylated to be soluble even in cold water, have no CP, i.e., their presumptive CP is below the freezing point of water. To correlate the CP with p, the lowest average number of oxyethylene units per molecule conducive to solubility in ice-cold water, p0 , is subtracted from the actual average number p. The difference p − p0 is the effective degree of ethoxylation, which determines the value of the CP. p0 is estimated by extrapolation of the initial, linear portion of the CP versus p plots of Figs. 1 and 2 as the number of oxyethylene units of the homologous surfactant possessing the lowest possible CP, namely, CP0 = 0 ◦ C. Equation (1) becomes     p − p0 p − p0 = = a + b(p − p0 ). (2) CP CP − CP0 The extrapolation to estimate p0 is illustrated for nonoxynols, for which nearly 50 CPs have been published [1,2, 5,7,9,13,19]. For the lower homologous nonoxynols, there are linear relations between p and the CP up to p = 10 and between log p and the CP up to p = 11.5, respectively. The equations for the two corresponding least-squares regression lines are CP = −109.5 + 17.88p (n = 10, r = 0.998)

(3)

and NaNO3 to NS solutions and extrapolating the CP as a function of the molality of NaNO3 to zero salt concentration. NaNO3 lowers the CP through the salting-out propensity of the Na+ ions; the NO− 3 ions neither increase nor decrease the CPs of NSs [18]. The CP decreases linearly with increasing molality of NaNO3 up to 2–3 m [19], which facilitates the extrapolation to zero molality. CP values obtained by the two methods agree within ±0.1 ◦ C.

4. Results The CP versus p curves for three of the five categories of NSs are shown in Figs. 1 and 2. The nonoxynol and oleyl alcohol data are not plotted because no new CPs were measured. Unpublished CPs of the other three surfactants are shown as solid symbols. The curves resemble Langmuir adsorption isotherms as well as the second-order swelling kinetics of semicrystalline polymers [20]. They consist of an initial, steeply ascending

CP = −268.0 + 341.2 log p

(n = 11, r = 0.996).

(4)

Both equations extrapolate to p0 = 6.1 for CP0 = 0 ◦ C. The CP vs p data for nonoxynols are fitted to Eq. (2) by the method of least squares, setting p0 = 6.1. The result is   p − 6.1 = 0.02859 + 0.008166(p − 6.1) CP (n = 48, r = 0.998). (5) These data are plotted in Fig. 3. The literature contains 48 CPs of water-soluble, normally distributed nonoxynols, which belong to 33 different homologs [1,2,5,7,9,13,19]. Their average p values range from 7 to 94. Equation (5) was compiled from all published values: None were discarded. The CPs of NSs are determined by the size, structure, and isomerism of the hydrocarbon moiety and by the average p value of the polyoxyethylene moiety and its statistical distribution. The high degree of self-consistency of the CP data of nonoxynols is demonstrated by the small scatter of

222

H. Schott / Journal of Colloid and Interface Science 260 (2003) 219–224

Fig. 3. Cloud point data for nonoxynols (circles) and polyoxyethylated oleyl alcohols (triangles) plotted according to Eq. (2). For the latter, the ordinate scale is (p − 5.45)/CP and the abscissa p − 5.45.

points in Fig. 3 and by a linear correlation coefficient, which, at 0.998, is very close to unity. This indicates that the 33 nonoxynols belong to a homologous series in which the only significant variable is the average p value. The distribution of the pvalues (approximately Poisson) and the hydrocarbon structure, both of which also affect CPs strongly [4], must be practically constant throughout the series. This consistency is all the more remarkable because the 33 nonoxynol samples were produced on three continents over a period of 50 years! Similar calculations were made for the other four surfactant series. The CP vs p data, linearized according to Eq. (2), are plotted in Fig. 4 for octoxynols and in Fig. 5 for polyoxyethylated dodecyl and tridecyl alcohols. The small scatter of the individual points about the regression lines of Figs. 3–5 and linear correlation coefficients exceeding 0.99 validate the empirical Eq. (2). The parameters of the regression lines of the five categories or homologous series of NSs investigated are recorded in Table 1. Since all linear correlation coefficients exceeded 0.99, they were omitted. Intercepts and slopes were compared in pairs by the small-sample T test when at least one surfactant category contained fewer than 25 homologs and by the largesample Z test when both categories contained more than 25 homologs [21,22]. Comparing the values of the five NS categories two at a time yields 10 combinations. Among the 10 pairs of intercepts, those of nonoxynols/ polyoxyethylated dodecyl alcohols and dodecyl/oleyl alcohols show no statistically significant difference at the 5% probability level while the intercepts of nonoxynols and polyoxyethylated oleyl alcohols are identical. According to an F test [22], these three intercepts do not differ signifi-

Fig. 4. Cloud point data for octoxynols plotted according to Eq. (2).

Fig. 5. Cloud point data for polyoxyethylated dodecyl (circles) and tridecyl (triangles) alcohols plotted according to Eq. (2). Table 1 Linear regression parameters according to Eq. (2) for the five categories of nonionic surfactants Hydrophobic moiety

p0 a

Intercept a (number of oxyethylene units/◦ C)

Slope b (◦ C−1 )

i-Octylphenol i-Nonylphenol n-Dodecyl alcohol Tridecyl alcohol Oleyl alcohol

6.5 6.1 4.1 4.2 5.45

0.02025 0.02859 0.02458 0.04194 0.02823

0.008446 0.008166 0.007259 0.007311 0.007810

nb Plateau HLB0 c CP (◦ C) 27 48 15 15 10

112 115 128 ∼118 115

11.6 11.0 9.8 9.6 9.4

a Smallest average number of oxyethylene units for borderline solubility in ice-cold water, pertaining to the homolog with extrapolated CP = 0 ◦ C. b Population in the surfactant category. c Hydrophile–lipophile balance for the homolog with borderline solubility in ice-cold water (p = p 0 ).

H. Schott / Journal of Colloid and Interface Science 260 (2003) 219–224

cantly. The other seven pairs of intercepts differ significantly at the 5% level. Among the 10 pairs of slopes, those of nonoxynols/ polyoxyethylated oleyl alcohols, polyoxyethylated dodecyl/tridecyl alcohols, and polyoxyethylated tridecyl/oleyl alcohols do not differ significantly at the 5% probability level: The three regression lines in Figs. 3 and 5 are parallel. The other seven pairs of slopes differ significantly at the 5% level. Since the regression lines of the nonoxynols and polyoxyethylated oleyl alcohols have practically identical intercepts and slopes, they are coincident (see Fig. 3). No other regression lines are coincident. The regression lines for octoxynol (Fig. 4) and nonoxynol (Fig. 3) cross over at a point whose coordinates correspond to the values of poctoxynol = 36.3, pnonoxynol = 35.9, and a common CP of 109.6 ◦ C. Of the five limiting or plateau CPs in Table 1, the four lowest values, which average 115 ◦ C, are indistinguishable from one another because their reproducibility is ±2 ◦ C. Only the polyoxyethylated dodecyl alcohols have a significantly higher plateau CP, namely, 128 ± 3 ◦ C. As p increases, the NSs increasingly resemble polyethylene glycols (PEGs). The most extensively ethoxylated homologs of the five surfactant categories of Table 1 are octoxynol 90, nonoxynol 101, polyoxyl 80 dodecyl ether, polyoxyl 51 tridecyl ether, and polyoxyl 51 oleyl ether. Their HLBs are 19.0, 19.1, 19.0, 18.4, and 17.9, respectively, which corresponds to 95.1, 95.3, 95.0, 91.8, and 89.3% ethylene oxide and only 5–11% alkylphenol or alcohol. Consequently, their limiting CPs, which range from 112 to 128 ◦ C, are essentially those of PEGs. The CPs of PEGs decrease with increasing molecular weight but depend less on concentration: At a 5% concentration, CPs of 130 and 113 ◦ C correspond to molecular weights of 15,000 and 30,000 for unfractionated PEGs [23, 24] and to molecular weights of 4400 and 9600 for fractionated PEGs [25], respectively. These two CP values bracket the plateau CPs of the NSs of Table 1. In a given homologous series of NSs, the CP decreases as the number p of oxyethylene units decreases. The homolog with CP = 0 ◦ C and p = p0 has borderline solubility in water: Homologs with fewer than p0 oxyethylene units are insoluble at all temperatures. The hydrophile–lipophile balance of such marginally soluble homologs, HLB0 , is a novel parameter for characterizing NSs. It quantitates the hydrophobicity of the hydrocarbon moieties. Since the HLB of NSs equals one-fifth of their percentage of ethylene oxide, HLB0 =

(100)(44.05p0) 881p0 = , 0 5(44.05p + M) 44.05p0 + M

(6)

where M is the molecular weight of the alkylphenol or alcohol. Higher HLB0 values indicate more polar, less hydrophobic, or less water-insoluble hydrocarbon moieties.

223

The HLB0 values of the five NS series are listed in the last column of Table 1. The two homologous series of polyoxyethylated alkylphenols have higher HLB0 values than the two polyoxyethylated alcohol series because the alkylaromatic hydrocarbons are less hydrophobic than the purely aliphatic hydrocarbons. The higher HLB0 of octoxynols compared to nonoxynols is ascribed to two factors: The aliphatic chain of octoxynols has one less carbon atom than that of nonoxynols and is more highly branched and compact [26]. The iso-octyl chain has two tertiary carbon atoms while the iso-nonyl chain has none. Increased branching of the hydrocarbon moiety tends to reduce the insolubility of organic compounds in water. The HLB0 values of the homologous series of polyoxyethylated dodecyl, tridecyl, and oleyl alcohols decrease by small but statistically significant amounts in that order. This is also the order of increasing size and, therefore, increasing hydrophobicity of the hydrocarbon moieties. A further large increase in hydrophobocity, accompanied by a large decrease in CP, occurs when the hydrocarbon moiety of NSs is replaced by fluorocarbon: The CP of the homogeneous surfactant n-C6 F13 CH2 (OC2 H4 )6 OH is 11 ◦ C [27]. The CP of the homogeneous hydrogenated counterpart is 74 ◦ C [2], i.e., 63 ◦ C higher. It would be interesting to measure the CPs of less extensively ethoxylated fluoroheptanols to determine the p0 and HLB0 of that homologous series and compare them with those of Table 1.

5. Discussion Of practical importance is the observation that the CPs of NSs cannot be raised above a 112–128 ◦ C range merely through more extensive ethoxylation. The upper limit of the surfactants’ CPs are the CPs of PEGs. Pharmaceutical, food, cosmetic, and industrial suspensions and emulsions are subjected to high temperatures during processing or end use, e.g., steam sterilization by autoclaving at  121 ◦ C, ultrahigh-temperature pasteurization at  132 ◦ C, dyeing of synthetic fibers, and lubrication. If the pertinent disperse systems are stabilized solely with NSs, they will coagulate or break when the surfactants’ CPs are exceeded, the upper limit being 112–128 ◦ C. However, conferring a slightly ionic character to nonionic micelles boosts their CPs substantially. For instance, adding 0.0005 m sodium dodecyl sulfate or 0.005 m decylammonium chloride to 2.00% (ca. 0.033 m) solutions of octoxynol 9 raises the CP by 26 ◦ C through the formation of mixed micelles [28]. The majority of drugs are cationic. In their usually watersoluble salt form (e.g., as hydrochlorides), they form mixed micelles with NSs. When present as the usually waterinsoluble but oil-soluble free bases, they are solubilized by the micelles of the NSs. Both processes confer a slight positive charge to the nonionic micelles and raise their CPs considerably, including their plateau CPs.

224

H. Schott / Journal of Colloid and Interface Science 260 (2003) 219–224

References [1] H. Schott, J. Pharm. Sci. 58 (1969) 1443. [2] M.J. Rosen, Surfactants and Interfacial Phenomena, 2nd ed., Wiley, New York, 1989, Chap. III.B. [3] N.M. van Os, J.R. Haak, L.A.M. Rupert, Physico-Chemical Properties of Selected Anionic, Cationic and Nonionic Surfactants, Elsevier, Amsterdam, 1993, Chap. 3. [4] W. Denk, W. Frank, Fette Seifen Anstrichm. 65 (1963) 46. [5] T. Gu, J. Sjöblom, Colloids Surf. 64 (1992) 39. [6] M. Meerbote, B. Koch, Tenside Surf. Det. 31 (1994) 39. [7] P.D.T. Huibers, D.0. Shah, A.R. Katritzky, J. Colloid Interface Sci. 193 (1997) 132. [8] M. Aoki, Y. Iwayama, Yakugaku Zasshi 79 (1959) 516. [9] M. Matell, in: Proc. 2nd Scand. Symp. Surface Activ., Stockholm, 1964, p. 127. [10] M. Donbrow, R. Hamburger, E. Azaz, J. Pharm. Pharmacol. 27 (1974) 160. [11] H. Schott, J. Colloid Interface Sci. 189 (1997) 117. [12] Triton® X-Series Surfactants, Union Carbide, Danbury, CT, 1997. [13] Igepal® Nonionic Surfactants, Rhone–Poulenc Surfactants and Specialties, Cranbury, NJ, 1992. [14] ICI Surfactants, Products for Personal Care, ICI Americas, Wilmington, DE, 1996.

[15] Surfonic® Technical Brochure, Jefferson Chemical, Houston, TX, 1978, presently Huntsman Corp., Austin, TX. [16] H. Kunieda, G. Umizu, Y. Yamaguchi, M. Suzuki, J. Jpn. Oil Chem. Soc. 47 (1998) 879. [17] H. Schott, S.K. Han, J. Pharm. Sci. 65 (1976) 979. [18] H. Schott, A.E. Royce, S.K. Han, J. Colloid Interface Sci. 98 (1984) 196. [19] H. Schott, Colloids Surf. 186 (2001) 129. [20] H. Schott, J. Macromol. Sci. Phys. B 31 (1992) 1. [21] W.J. Youden, Statistical Methods for Chemists, Wiley, New York, 1951, Chap. 5. [22] D.G. Kleinbaum, L.L. Kupper, Applied Regression Analysis and Other Multivariable Methods, Duxbury Press, North Scituate, MA, 1978, Chap. 8. [23] Y.C. Bae, S.M. Lambert, D.S. Soane, J.M. Prausnitz, Macromolecules 24 (1991) 4403. [24] F.E. Bailey Jr., R.W. Callard, J. Appl. Polymer Sci. 1 (1959) 56. [25] S. Saeki, N. Kuwahara, M. Nakata, M. Kaneko, Polymer 17 (1976) 685. [26] H. Schott, J. Pharm. Sci. 73 (1984) 790. [27] G. Mathis, P. Leempoel, J.-C. Ravey, C. Selve, J.-J. Delpuech, J. Am. Chem. Soc. 106 (1984) 6162. [28] H. Schott, A.E. Royce, J. Pharm. Sci. 73 (1984) 793.