e, t m ox S. ille, ille, a r t i c l e i n f o a b s t r a c t microemulsions containing photoresponsive anionic surfactants [3–5]. Our group has focused upon the use of acid-cleavable 1,3- dioxolane alkyl ethoxylate surfactants [6–9]. The pH-degradable alkyl ethoxylate, 4-CH3O (CH2CH2O)7.2-CH2, 2-C13H27, 2-C2H5, 1,3-dioxolane or ‘‘cyclic ketal” surfactant, CK-2,13-E7ave, with ‘‘7.2” representing the average degree of polymerization for the emulsion systems that can be used for the purification of proteins [10]. Winsor-III systems consist of a surfactant-rich middle phase (typically possessing bicontinuous microemulsions) in equilibrium with ‘‘excess” oil and aqueous phases. Through combining an aque- ous solution containing a dissolved protein mixture and oil con- taining surfactant, a Winsor-III system forms that possesses a middle phase highly concentrated in proteins that interact with surfactant, resulting in a selective bioseparation method. Recent work has shown that measuring the partitioning of alkyl or alkylphenyl ethoxylates between the Winsor-III oil and water excess phases is valuable for predicting the ‘‘optimal” formulation ⇑ Corresponding author at: Department of Biosystems Engineering and Soil Science, 2506 E.J. Chapman Drive, Knoxville, TN 37996-4531, USA. Fax: +1 865 974 4514. Journal of Colloid and Interface Science 352 (2010) 424–435 Contents lists availab Journal of Colloid an .co E-mail address:
[email protected] (D.G. Hayes). 1. Introduction ‘‘Triggerable surfactants”, defined as amphiphiles whose surface activity can be changed by environmental factors, such as pH, tem- perature, light, or chemical reagents [1,2], may have many poten- tial applications in microemulsion systems, particularly to release nanoencapsulated solutes for drug delivery or recover products from multiphasic reactions hosted in microemulsions. Triggerable surfactants have been employed mainly as micelle-forming aggre- gates in aqueous systems [1,2]. Eastoe and co-workers have dem- onstrated that ultraviolet radiation can trigger the destruction of ethoxylate group, synthesized in our laboratory (depicted in Fig. S1, Supporting material), undergoes hydrolysis in acidic media, leading to a loss of surface activity [6,8]. CK-2,13-E7ave, produces water-in-oil (w/o-) microemulsions that possess phase behavior and physical properties similar to common alkyl ethoxylates, ex- cept for lower surfactant efficiency (e.g., high critical microemul- sion concentration) and strong attractive interactions. In order to improve the efficiency of CK-2,13 surfactant, our group has recently investigated the utilization of binary surfactant mixtures that contain CK-2,13 as a component. Specifically, the goal is to produce thermodynamically stable Winsor-III micro- Article history: Received 14 May 2010 Accepted 28 August 2010 Available online 28 September 2010 Keywords: Aerosol-OT Alkyl ethoxylate surfactants Alkyl glucoside surfactant Cleavable surfactants HPLC analysis of surfactants Microemulsions Nonionic surfactants Partition coefficients Small-angle neutron scattering Thermodynamic model 0021-9797/$ - see front matter � 2010 Elsevier Inc. A doi:10.1016/j.jcis.2010.08.076 Partition coefficients for a pH-degradable 1,3-dioxolane alkyl ethoxylate surfactant, 4-CH3O (CH2CH2O)5.6- CH2, 2,2-(CH2)12CH3, 2-(CH2) CH3, 1,3-dioxolane or ‘‘cyclic ketal” surfactant, CK-2,13-E5.6,ave, between isooctane- and water-rich phases of 2- and 3-phasemicroemulsion systems (Kn) were determined as func- tions of the ethoxylate size, n, and temperature for the neat surfactant and its binary surfactant mixtures, to understand the partitioning of alkyl ethoxylates possessing a broad distribution of ethoxylate size and to determine conditions required for formation of 3-phase microemulsion systems at an optimal temper- ature where phase separation occurs rapidly, important for protein purification via proteins’ selective partitioning to the middle phase, driven by affinity to the second surfactant of the binary mixture. A semi-empirical thermodynamic mathematical model described the partitioning data well, provided opti- mal temperature values consistent with phase diagrams and theory, and demonstrated that the tail region of CK-2,13-E5.6,ave ismore polar than the hydrophobes of fatty alcohol ethoxylates. The addition of Aerosol- OT (AOT) removed the temperature sensitivity of CK-2,13-E5.6,aves partitioning, producing 3-phase microemulsion systems between 20 �C and 40 �C. Analysis of the bottom phases of the 2- and 3-phase microemulsion systems formed by CK-2,13-E5.6,ave via small-angle neutron scattering demonstrated the presence of spherical, monodisperse oil-in-water microemulsions. � 2010 Elsevier Inc. All rights reserved. Partitioning behavior of an acid-cleavabl surfactant in single and binary surfactan microemulsion systems according to eth Javier Gomez del Rio a, Douglas G. Hayes a,b,⇑, Volker aDepartment of Chemical and Biomolecular Engineering, University of Tennessee, Knoxv bDepartment of Biosystems Engineering and Soil Science, University of Tennessee, Knoxv cOak Ridge National Laboratory, Chemical Sciences Division, Oak Ridge, TN 37831, USA www.elsevier ll rights reserved. 1,3-dioxolane alkyl ethoxylate, ixtures for 2- and 3-phase ylate head group size Urban c TN 37996, USA TN 37996, USA le at ScienceDirect d Interface Science m/locate / jc is oid a Nomenclature A coefficient used to express Dhn as a function of n in Eq. (4) A0 coefficient used to express Dsn as a function of n in Eq. (5) Javier Gomez del Rio et al. / Journal of Coll at which oil and water solubilization in the middle phase is maxi- mized and the volume of oil and water ‘‘excess” phases are approx- imately equal [11–13]. The cited work applies to commercial surfactant preparations, where the surfactants possess different ethoxylate sizes with distribution described by Poisson’s equation, similar to the ethoxylate distribution of CK-2,13-E7ave, [7]. For the protein purification work that occurs in our laboratory, ‘‘optimal” conditions lead to the most rapid equilibrium formation of Win- sor-III systems, which maximizes protein partitioning to the mid- AOT aerosol-OT [sodium bis(2-ethylhexyl) sulfosuccinate] surfactant B coefficient used to express Dhn as a function of n in Eq. (4) B0 coefficient used to express Dsn as a function of n in Eq. (5) Bincoher Incoherent background scattering for SANS data (Eq. (6)) C coefficient used to express Dhn as a function of n in Eq. (4) C0 coefficient used to express Dsn as a function of n in Eq. (5) Cjn concentration of CK-2,13-En in phase j (wt.%) C8bG1 octyl b-glucoside surfactant CK-2,13-En alkyl ethoxylate surfactant containing alkyl tail groups of 2 and 13 carbon atoms in length, a hydrophile of n ethylene glycol monomeric units, and a 1,3-dioxo- lane (‘‘cyclic ketal”) pH-degradable group between the hydrophile and alkyl tails: 4-CH3O (CH2CH2O)n-CH2, 2,2-(CH2)12CH3, 2-(CH2) CH3 clc critical microemulsion concentration HLB hydrophilic–lipophilic balance I(Q) scattering intensity from SANS analysis (cm�1) Kn partition coefficient for CK-2,13-En surfactant of ethoxy- late size n between aqueous and apolar phases of 2- and 3-phase Winsor microemulsion systems K0,HLB y-intercept of a semilog plot of Kn vs. n, reflecting the contribution of the hydrophobe to an alkyl ethoxylate surfactant’s partitioning m slope of a semilog plot of Kn vs. n, reflecting the relative polarity of ethoxylate monomeric units at the HLB tem- perature MPEG poly(ethylene glycol) monomethyl ether MWi molecular weight of component i (g mol�1) n number of ethoxylate monomeric units in a molecule of CK-2,13-En surfactant Np number concentration of scattering bodies (oil-in-water microemulsions); Eq. (6) olE slE mole ratio of oil (isooctane) to surfactant in oil-in-water microemulsions; Eq. (7) P(Q) form (or shape) factor, employed in the mathematical modeling of SANS data (Eq. (6)) Q momentum transfer vector, SANS analysis (Å�1) R ideal gas law constant (J K�1 mol�1) Rc+s,ave average radius for the core plus shell of an oil-in-water microemulsion (Å) S(Q) interparticle structure factor, employed in the mathe- matical modeling of SANS data (Eq. (6)) SGi specific gravity of component i (g mL�1) T temperature (K) THLB apparent hydrophilic–lipophilic balance, or HLB, tem- perature (�C or K) T�HLB hydrophilic–lipophilic balance, or HLB, temperature (�C or K) for a surfactant concentration equal to cc nd Interface Science 352 (2010) 424–435 425 dle phase and minimizes denaturation of protein [10]. For commercial alkyl and alkylphenyl ethoxylates and CK-2,13-E7ave, the average ethoxylate size of surfactants in the bottom phase of a Winsor-III system is significantly larger than for the top phase due to the tendency of surfactants with larger ethoxylate groups to partion to the more hydrophilic, aqueous phase [7,11–13]. This trend in partitioning occurs even for optimal formulations, at which the partion coefficient for the overall surfactant is approxi- mately 1.0. Vp volume of a scattering body (oil-in-water microemul- sion); Eq. (6) (Å3) WI Winsor-I microemulsion system (oil-in-water-, or o/w-, microemulsions in equilibrium with an ‘‘excess” oil phase) WII Winsor-II microemulsion system (w/o-microemulsions in equilibrium with an ‘‘excess” aqueous phase) WIII Winsor-III microemulsion system (surfactant-rich mid- dle phase in equilibrium with ‘‘excess” oil and aqueous phases) WIV Winsor-IV microemulsion system (1-phase microemul- sion solution) Dhn enthalpy at standard state for the transport of CK-2,13- En from the water-rich phase to the oil-rich phase of a Winsor microemulsion system (J mol�1) cc ‘‘Critical” surfactant concentration derived from a ‘‘fish” phase diagram (surfactant concentration vs. tempera- ture), determined from the intersection of the 1- and 3-phase microemulsion system phase boundaries Dsn enthalpy at standard state for the transport of CK-2,13- En from the water-rich phase to the oil-rich phase of a Winsor microemulsion system (J K�1 mol�1) Dq difference in neutron scattering length density, or con- trast, for SANS analysis in Eq. (6) Å�2 k wavelength of incident neutrons employed for SANS Å) /j volume fraction of subcomponent j in a microemulsion solution rR�1c;ave polydispersity index for Rc, based on the Schulz distribu- tion h scattering angle employed for SANS analysis Dlw!o;n standard chemical potential for the transport of CK- 2,13-En from the water-rich phase to the oil-rich phase of a Winsor microemulsion system (J mol�1) Superscripts o oil-rich phase of 2- or 3-phase Winsor microemulsion system w water-rich phase of 2- or 3-phase Winsor microemul- sion system Subscripts ave average c (oil-rich) core of an oil-in-water microemulsion n number of ethoxylate monomeric units in CK-2,13-En surfactant s (surfactant-rich) shell of an oil-in-water microemulsion solv solvent id a The partition coefficient for a CK-2,13-En alkyl ethoxylate sur- factant molecule of ethoxylate head group size n between aqueous and apolar phases of 2- and 3-phase Winsor microemulsion sys- tems, Kn, is defined as: Kn ¼ C w n Con ð1Þ where Cwn and C o n are the concentration of surfactant in the water and oil phase, respectively. For Winsor-III systems, the two concen- trations refer to the excess aqueous (bottom) and apolar (top) phases, respectively. Thus, a partition coefficient of 1.0 (log(10) Kn = 0) represents a CK-2,13-Ens equal partitioning into both phases, with values >1 partitioning in favor of the aqueous phase and 99%) and purchased from either Fisher Scientific, Pittsburgh, PA, or Sigma–Aldrich. Deionized water was used throughout. O-[(2-tridecyl, 2-ethyl-1,3-dioxolan-4-yl) methoxy]–O-meth- oxy poly(ethylene glycol), or ‘‘cyclic ketal” alkyl ethoxylate surfac- tants, having average ethoxylate monomeric units of 5.6 and 3 (CK- 2,13-E5.6,ave and CK-2,13-E3, respectively), depicted in Fig. S1 of the Supporting material, were synthesized employing a previously published procedure by our group using 3-hexadecanone, glycerol, and MPEG mesylate as starting materials, with the latter produced from mesyl chloride and MPEG3 or MPEG7 [7,14]. The CK-2,13 sur- factants were >95% pure, with 1H NMR employed to verify the structure of the surfactants and the intermediate products pre- pared during the synthesis and HPLC and FTIR spectroscopy em- ployed to determine the extent of impurities present (unreacted MPEG and ketone), which was which consisted of subtracting scattering contributions from the to the solvents, weighed according to their respective volume frac- top Mol 3 0.04 0.23 0.47 0.11 0.33 0.09 0.72 0.14 0.63 0.18 0.39 0.10 0.50 0.09 0.33 0.14 . oid a steps were repeated until the top and the bottom phases were clear. Aliquots were removed from the bottom and top phases for HPLC analysis. Replicate experiments were performed for sev- eral different conditions to determine the standard error in parti- tion coefficient values. 3.2.3. HPLC analysis Molecules of the surfactant CK-2,13-E5.6,ave were separated according to the size of their ethoxylate group, n, via reversed- phase high performance liquid chromatography (RP-HPLC), per- formed using a dual-pump analytical system from Varian, Inc. (Walnut Grove, CA) with evaporative light scattering (Model MKIII from W.R. Grace, Deerfield, IL) and refractive index (Varian) detec- tors, and a reversed-phase Microsorb-MV 4.6 mm � 250 mm, 5 lm C18 column from Varian, maintained at 25 �C. An isocratic solvent �1 Table 1 Ethoxylate distribution of CK-2,13-E5.6,ave in single and binary surfactant mixtures for Temperature (�C) Phase MPEG-350a CK-2,13-E5.6,ave (original surfactant)b 7.5 wt.% CK-2,13-E5.6,ave 20 Top 20 Bottom 50 Top 50 Bottom 3.75% CK-2,13-E5.6,ave/3.75% CK-2,13-E3 20 Top 20 Bottom 50 Top 50 Bottom 3.75% CK-2,13-E5.6,ave/3.75% C8bG1 20 Top 20 Bottom 50 Top 50 Bottom 3.75% CK-2,13-E5.6,ave/3.75% AOT 20 Top 20 Bottom a Data taken from [10]; other mole fractions: n = 2:0.004; n = 11:0.06; n = 12:0.05 b Other mole fractions: n = 11:0.02; n = 12:0.02. c Mole fraction is id a 4. Results and discussion 4.1. ‘‘Fish” phase diagram for heptane/CK-2,13-En/water A partial ‘‘fish” phase diagram for water/CK-2,13-E5.6,ave/isooc- tane is depicted in Fig. 1. For this diagram, temperature is plotted against the weight percent of surfactant, with the water/oil weight ratio held constant at 1:1. This diagram is typical of alkyl ethoxy- lates, for which at low-to-moderate amounts of surfactant the transition Winsor-I (WI)?WIII?WII occurs as temperature is in- creased, due to the increased hydrophobicity of the ethoxylate group. Winsor-I, II, and IV systems refer to an aqueous surfac- tant-rich solution (typically possessing oil-in-water-, or o/w-, microemulsions) in equilibrium with an ‘‘excess” oil phase, surfac- tant-rich oil solution (typically possessing w/o-microemulsions) in equilibrium with an ‘‘excess” aqueous phase, and 1-phase micro- emulsions systems, respectively. Similar to the ‘‘fish” diagrams of other alkyl ethoxylates, the 3-phase region reduced to a single point as the weight fraction of surfactant was increased. The tem- perature corresponding to this point is referred to the hydrophilic– lipophilic balance, or HLB, temperature (T�HLB), while the point’s weight fraction (cc) is a measure of the surfactant’s efficiency. At this temperature the surfactant partitions equally between the Fig. 1. Phase diagram: mass fraction of surfactant vs. temperature for the water/CK- 2,13-E5.6,ave/isooctane microemulsion system. Water: isooctane ratio held constant at 1:1 g g�1. Dotted lines represent approximate positions of phase boundaries. 428 Javier Gomez del Rio et al. / Journal of Collo oil and water phase. For surfactant concentration equal to cc, as temperature increases from just below to just above T�HLB, the microemulsion type changes from o/w- to w/o-microemulsions. The value of T�HLB and cc for water/CK-2,13-E5.6,ave/isooctane, �40 �C and 15 wt.%, respectively, are higher and lower, respec- tively, compared to previously published values by our group for water/CK-2,13-E7,ave/isooctane, 24 �C and 40 wt.% [6]. The differ- ence is attributed to the lower ethoxylate size and higher degree of purity for the surfactant employed in this study, obtained through the employment of hexane extraction in the purification. In the previous study [6], the surfactant contained �15–20% MPEG, a starting material for the surfactant synthesis; herein, hexane was employed to remove nearly 100% of MPEG. In this paper, microemulsion systems containing 5–10% surfac- tant were examined between 20 and 50 �C. According to Fig. 1, systems formed at 20 �C are of the WI type, while for 30–50 �C, WIII occurs. For the latter, the middle phases were highly viscous when below 47 �C; above this temperature the middle phase behaved as a fluid. Polarized light microscopy analysis of the mid- dle phases did not yield any signatures corresponding to anisotropic structures for all temperatures, suggesting the middle phases were isotropic, probably consisting of bicontinuous microemulsions. 4.2. Partitioning of CK-2,13-En in single surfactant microemulsion systems The ethoxylate distribution of CK-2,13-E5.6,ave in oil and water- rich phases for 2- or 3-phase systems formed by heptane/water 1:1 w/w are given in Table 1. Although the average ethoxylate group size, n, for the surfactant is 5.6, the average n value for oil- and water-solubilized surfactant at 20 �C is 4.1 and 5.9, reflecting that molecules with larger ethoxylate groups favorably partition to the aqueous phase, due to their higher polarity. As the temperature is increased to 50 �C, the n value for oil-solubilized surfactant in- creases to 4.6, reflecting the increased partitioning of surfactants with larger ethoxylate groups to the oil phase, due to the decreased hydrophilicity of poly(ethylene glycol) upon an increase of temper- ature. Water–oil partition coefficients for CK-2,13-E5.6,ave surfac- tants as a function of the number of ethoxylate monomeric units, n, and temperature are given in Fig. 2A. As expected the coeffi- cients increased sharply with an increase of n and a decrease of temperature, both of which increase the hydrophilicity of alkyl ethoxylates. Thus, for the CK-2,13-E5.6,ave surfactant mixture, mol- ecules with low ethoxylate group size partition to the oil phase (i.e., log Kn < 0) while those with higher ethoxylate group size par- tition to the aqueous phase (log Kn > 0), reflecting the overall sur- factant’s broad distribution between the two phases, as described previously [7]. The temperature at which equal partitioning of surfactant be- tween oil and water occurs (i.e., log Kn = 0 for the average n value of the surfactant) at a specific surfactant concentration, defined as THLB, is the apparent HLB temperature denoting the transition from water-soluble to oil-soluble microemulsions. Thus, T�HLB equals THLB for a surfactant concentration equal to cc. THLB is approximately equal to the median temperature of the WIII region in a ‘‘fish” diagram at a specified surfactant concentration. For homogeneous surfactants, THLB � T�HLB at all surfactant concentra- tions < cc; moreover, the WIII region is bisected symmetrically by the horizontal line T = T�HLB. For alkyl ethoxylates possessing a broad ethoxylate distribution, the WIII region moves to higher tem- perature as the surfactant concentration decreases starting from cc [19]. Moreover, for surfactant concentrations just below cc, the relationship THLB � T�HLB holds true; but, as the surfactant concen- tration decreases further, THLB increases [19]. The underlying cause is the decreased partitioning of molecules with short ethoxylate chains to liquid–liquid interfaces with decreasing surfactant con- centration, due to their increased fraction among the monomeric surfactant population. This leads to an enhanced interfacial con- centration of surfactants with large ethoxylate groups and thus a higher THLB as the surfactant concentration decreases [19]. For 7.5% and 10% CK-2,13-E5.6,ave, THLB (at n = 5.6), obtained by interpo- lating between the data of Fig. 2A, approximately equals 42 �C, �T�HLB, consistent with theory. For 5% surfactant, THLB is calculated to be 51.2 �C using a mathematical model described below, consis- tent with the increase of THLB with a decrease of surfactant concen- tration as described above. Through interpolation of the 5% surfactant/50 �C data of Fig. 2A, the average n value that yields Kn = 0 is �7.1, which is higher than the average value for the sur- factant, 5.6, consistent with the increased interfacial concentration of surfactants with large ethoxylate groups as discussed above. Furthermore, Fig. 2A demonstrates that as the CK-2,13-E5.6,ave con- centration is decreased, the log Kn values decrease for surfactants with ethoxylate chain length < 6, reflecting their increased parti- tioning to the isooctane-rich phase, where they exist in monomeric form [7], as surfactant concentration is decreased, consistent with the trends discussed above [19] . nd Interface Science 352 (2010) 424–435 Salager and co-workers demonstrated for ‘‘optimal” WIII sys- tems (at T = THLB) that Kn increased linearly with n for both alkyl phenol and alkyl ethoxylates when plotted on semilog coordinates B D h, n, C8b urfa olid s be oid and Interface Science 352 (2010) 424–435 429 [11–13]. As demonstrated in Fig. 2A, the log Kn vs. n profile for 7.5% and 10% surfactant at 20 �C curves downward,” i.e., its second derivative is negative. The profile at 30 �C is less curved and that C Fig. 2. Semilog plot of partition coefficients for CK,2-13-En vs. ethoxylate chain lengt 2,13-E5.6,ave, (B) 50 wt.% CK-2,13-E5.6,ave, 50%. CK-2,13-E3, (C) 50% CK-2,13-E5.6,ave, 50% phase. Outlined and unfilled symbols: 5 wt.% surfactant; filled gray symbols: 7.5% s 30 �C, squares: 40 �C, and diamonds 50 �C. Curves represent model fits to the data. S n > 8 could not be determined due to the concentration of these surfactant molecule are within 6.6%, based on standard deviation from replicate experiments. A Javier Gomez del Rio et al. / Journal of Coll for 40 �C is even less curved, approaching linearity. In contrast, the profile for 7.5% and 10% surfactant at 50 �C curves upward, i.e., its second derivative is positive. This suggests the straight-line log Kn vs. n profile for 7.5% and 10% surfactant must occur at a tem- perature (equal to THLB) that is slightly above 40 �C and to account for the increase of intensity with Q in the low-Q region (Fig. 3). Ellipsoidal and cylindrical form factors were also tried, but it was found that introducing shape anisotropy in this manner did not improve the fit quality compared to polydisperse spherical form factors, particularly at low Q (Fig. S3 of Supporting material). Values of the average radius and the polydispersity index, Rc+s,ave and r R�1c;ave, respectively, are given in Table 3. The aggregates pos- sessed radii of 38–54 Å, with the size decreasing upon an increase of dilution factor. The change of Rc+s,ave is attributed to a slight per- turbation of aggregate structure upon dilution. The polydispersity indices were low, 0.16–0.18, indicating the aggregates were rea- sonably spherical. In contrast, it was shown previously that in the absence of oil, CK-2,13-E5.6,ave forms micelles possessing pro- late ellipsoidal shape, with the short semi-axes of 35–39 Å and as- pect ratios being between 2.8 and 3.5 [7]. SANS data of such micellar shapes cannot be described by a spherical model with low polydispersity; therefore, the aggregates contained in the bot- tom phases of the WI and WIII systems are not believed to be mi- celles. Since it is well known that the encapsulation of oil into a micellar core promotes the transition in shape from ellipsoids to spheres [20], the aggregates formed in the bottom phases should Momenturm Transfer, Q, A-1 0.001 0.01 0.1 1 I(Q ), cm -1 0.001 0.01 0.1 1 10 100 1000 20oC, 1.0 20oC, 0.8 20oC, 0.6 20oC, 0.2 30oC, 0.4 30oC, 0.2 Fig. 3. SANS droplet contrast data for the bottom, aqueous, phase of Winsor-I and III microemulsions systems formed at 20 �C and 30 �C, respectively, formed by heptane/CK-2,13-E5.6,ave/D2O 39.5/5.0/55.5 w/w/w (51.3/4.6/44.1 v/v/v), after dilu- tion with further D2O. Prior to dilution, bottom phase recovered at 20 �C and 30 �C contained 5.1 and 4.3 wt.% CK-2,13, respectively. Legend provides temperature and 430 Javier Gomez del Rio et al. / Journal of Colloid and Interface Science 352 (2010) 424–435 a dilution series for each. The aqueous excess phase aliquot ob- tained at 30 �C from the WIII system, and the most concentrated samples from its dilution series for which the D2O/aliquot ratio was 0.67:1 lL lL�1 or less, formed macroemulsions shortly after their removal from the 30 �C bath and exposure to room tempera- ture, and thus were not analyzed further. The resultant deuteration scheme for the samples consisted of a ‘‘droplet” contrast between a non-deuterated heptane-rich core (if o/w-microemulsions formed) plus surfactant shell and a deuterated D2O-rich bulk phase. the volume of original bottom phase per volume of original bottom phase + added D2O. Curves represent form factor-structure factor model fits to the data. Param- eters for the model fits are given in Table 3. I(Q) values for (20 �C, 0.8 lL lL�1), (20 �C, 0.6 lL lL�1), (20 �C, 0.2 lL lL�1), (30 �C, 0.4 lL lL�1), and (30 �C, 0.2 lL lL�1) were divided by 4, 8, 16, 40, and 40, respectively, to improve visualization. The data were analyzed by form factor-structure factor model- ing based on spheres that included polydispersity of the radius based on the Schultz distribution and a structure factor based on repulsive interactions of the Hayter–Penfold type, as described above. The fits to the data by this approach are good, with the repulsive interaction-based structure factor being the best choice Table 3 Results from form factor-structure factor fitting of small-angle neutron scattering data emp the system Heptane/CK-2,13-E5.6,ave/D2O.a Temperature (�C) 20 Dilution factor�1c 1.0 Volume fraction of core + shell, /c+sd 0.131 Volume fraction of core + shell (Guinier), /c+sd,e f Volume fraction of CK-2,13, /CKg 0.0590 Heptane – CK ratio of o/w-microemulsions, olEslE , mol mol �1h 6.0 Mean core + shell radius, Rc+s,ave, Åi 54.0 Mean core + shell radius (Guinier), Rc+s,ave, Åe,i f Polydispersity index, r R�1c;ave d 0.182 a Composition of samples described in Fig. 3. Form factor: spherical core, shell of con factor: interdroplet repulsions of the Hayter–Penfold type, unless noted otherwise. Scatte for the dispersed aggregates to that of the solution’s CK-2,13-En population, calculated t b Use of a hard sphere structure factor rather than interdroplet repulsions provided an Rc,ave. c Volume of bottom phase aliquot per volume of bottom phase aliquot + added D2O. d Error limits: ±0.01. e From Guinier analysis; see Supporting material for further information. f Guinier analysis was not performed, due to the prominence of the repulsive interac g Calculated from the concentration of CK-2,13-E5.6,ave, determined by HPLC analysis, h Calculated using Eq. (7). i Error limits: ±1.0. be o/w-microemulsions. The presence of oil in the bottom phase was confirmed by the use of lipophilic Sudan IV dye, in the micro- emulsion systems, where its red color was observed in the aqueous bottom phases. Preliminary SANS data for a WI bottom phase taken at an earlier time, formed at 20 �C using 4.5 wt.% CK-2,13-E5.6,ave overall rather than 5.0% employed to obtain the data of Fig. 3 and Table 3, strongly overlaps with SANS data for a WI/20 �C dilution series sample sharing the same surfactant concentration and provides similar values of Ro,ave, r R�1c;ave, and, /c+s via S(Q) � P(Q) modeling (Figs. S4 and S5 of the Supporting material), demonstrating the repeatability of the experimental approach. Of interest, the preli- minary SANS investigation employed a larger Q range, 0.006–0.5 Å�1. The fact that the parameter values obtained via modeling for the preliminary investigation and the data of Table 3 are similar suggests that the use of a smaller Q range for the latter (0.005– 0.2 Å�1) did not significantly affect ability of the model to fit the data. The formation and analysis of a dilution series for the WI and WIII bottom phases allows for a second estimate of the dispersed phase (core + shell) volume fraction, /c+s, and for the relative loying droplet contrast collected for the bottom phases of Winsor-I and III systems for 20 20 20 30 30 b 0.8 0.6 0.2 0.4 0.2 0.098 0.072 0.017 0.027 0.013 f f 0.017 f 0.009 0.0474 0.0307 0.0115 0.0198 0.010 5.4 5.6 6.8 2.1 2.1 50.9 50.1 40.3 43.1 37.9 f f 40.8 f 46.2 0.175 0.161 0.148 0.175 0.197 stant thickness, with the core radius described by a Schultz distribution. Structure ring length density values: solvent was equated to that of D2O (6.37 � 10�6 Å�2) and o be 2.43 � 10�7 Å�2and 2.57 � 10�7 Å�2 at 20 �C and 30 �C, respectively. equally effective fit to the data, and yielded the same values of /c+s, Rc+s,ave, and r tion structure factor at low Q. and the density of all components, neglecting any volume change due to mixing. readily forms temperature-insensitive microemulsions phases with potential applicability to drug delivery [22]. Temperature- insensitive microemulsions were also reported for the water/ C10bG1/C6E2/n-octane system when an equal mass of each surfac- tant was employed [23,24]. The relationship between ethoxylate group size of the former surfactant, temperature, and water–oil partitioning is depicted in Fig. 2C. The conditions employed pro- duce WIII systems, except at 50 �C, which resulted in a WII system, in agreement with phase diagrams reported by our group [22]. The middle phase volume fraction for the system formed at 40 �C was quite small. The addition of C8bG1 to CK-2,13-E5.6,ave did not affect the temperature sensitivity of the latter, evidenced by similar dis- tances between the log Kn vs. n isotherms of Figs. 2A–C as a func- tion of temperature. Furthermore, C8bG1 increased the extent of partitioning for CK-2,13-En into the aqueous phase, as observed by the higher log Kn values given in Fig. 2C. C8bG1 also partitioned strongly to the aqueous phase for all experiments depicted in Fig. 2C. The aqueous phase concentration of CK-2,13-E5.6,ave was in- creased in the presence of C8bG1,with concentrations ranging from 4.6–6.8% compared to 3.3–5.1% for pure CK-2,13-E5.6,ave at compa- rable overall concentrations (Table 2). The surfactant mixture of the top phase was depleted of molecules with n > 6, Table 1. Sim- oid and Interface Science 352 (2010) 424–435 431 amount of heptane and CK-2,13-E5.6,ave in the microemulsions to be determined. The fact that the /c+s values given in Table 3, deter- mined from the S(Q) � P(Q) modeling, decrease linearly with the inverse of the dilution factor (volume of bottom phase aliquot per volume of aliquot plus added D2O), as plotted in Fig. S6 of the Supporting material, supports the accuracy of the model-de- rived /c+s values. To further support the accuracy of the /c+s values, Guinier analysis (described in the Supporting material) was ap- plied to the SANS data for the most dilute sample of each dilution series, since these samples had the least influence of the repulsive interaction-based structure factor in the low-Q region employed for the Guinier analysis. (To support the hypothesis that repulsive interactions are negligible for the most dilute sample of the 20 �C/ WI dilution series, a hard sphere structure factor was equally effec- tive for the S(Q) � P(Q) modeling of the data, and yielded the same parameter values as obtained using a repulsive interaction-based structure factor, Table 3.) The Guinier analysis also provides a mea- sure of the average radius of the scattering bodies in the dispersed phase. The Guinier plots yielded the anticipated straight-line rela- tionship between log[I(Q)] and Q2 (Fig. S7 of the Supporting mate- rial). As demonstrated in Table 3, good agreement exists between the values of Rc+s,ave and /c+s obtained via Guinier analysis and S(Q) � P(Q) modeling. Therefore, it is concluded the volume frac- tion values obtained by S(Q) � P(Q) modeling are accurate. Also, the number of adjustable parameters used for the model fitting was reasonably small (e.g., only Rc+s,ave and r R�1c;ave for P(Q)]). The mole ratio of heptane to CK-2,13-En in the microemulsions, olE slE , for the WI/20 �C andWIII/30 �C bottom phases was calculated as follows: olE slE ¼ /cþs /CK � /CK;clc � 1 ! SGo SGCK MWCK MWo ð7Þ where SGi andMWi refer to the specific gravity andmolecular weight of species i, respectively. (MWCKwas estimated to be 583 and 613 for the 20 �C and 30 �C bottomphase samples, respectively.) The volume fraction of surfactant, /CK (listed in Table 3), was estimated using the surfactant concentration measured using HPLC and appropriate val- ues of specific gravity, neglecting volume change due to mixing. The parameter /CK;clc, the volume fraction of monomeric surfactant, or equivalently, the aqueous phase critical microemulsion concentra- tion (clc) expressed as a volume fraction, was determined from the intercept of a plot of /CK vs. /c+s (Fig. S6 of the Supporting mate- rial). Values of /CK;clc obtained for the 20 �C and 30 �C systems were 0.0051 and 0.0014, respectively; equivalently, clc values equal 0.43 wt.% and 0.12%, respectively. As shown in Table 3, the o/w-micro- emulsions existent in the WI excess phase possessed a significantly larger oil-surfactant ratio, 6.0 ± 1 mol mol�1, compared to the ratio for the WIII aqueous phase’s microemulsions, 2.1 mol mol�1. This occurrence is probably related to the larger concentration of surfac- tant for the WI/20 �C, subphase, Table 2, and the smaller concentra- tion of the surfactant molecules with short ethoxylate groups, suggested by Fig. 2A. The latter are less effective surfactants for the solubilization of oil into water compared to molecules with larger ethoxylate groups. In conclusion, it is clear that o/w-microemulsion form in the ‘‘excess” aqueous phase of WIII systems at temperatures below THLB. Thus, the Kn values provided herein are not true water- isooctane partition coefficients, but are apparent ones [21]. 4.4. Partitioning of CK-2,13-En in Binary Surfactant Microemulsion Systems CK-2,13-E5.6,ave has been applied in binary surfactant mixtures Javier Gomez del Rio et al. / Journal of Coll by our group to improve and modify its interfacial behavior for applications such as protein extraction and drug delivery. One such binary system is CK-2,13-E5.6,ave/alkyl b-glucoside (C8bG1), which ilar to C8bG1, the addition of AOT to CK-2,13-E5.6,ave at AOT/CK ra- tios of 4:1 and 1:1 g g�1 increased the partitioning of CK-2,13-E3 and -E4 into the aqueous phase (Figs. 2A–C and 4). For all Fig. 4 experiments except for the 1.5%/6.0% CK/AOT system, WIII systems formed. The middle phase contained >90% of the system’s AOT. The remaining AOT partitioned preferentially to the bottom phase, evi- denced by the AOT concentration in the top phase being below detection limits according to HPLC analysis. In other words, parti- tion coefficient values for AOT, KAOT, were�1.0. In contrast, in sin- gle-surfactant systems AOT partitions to isooctane compared to water and forms WII systems [25]. The binary 1.5%/6.0% CK- E5.6,ave/AOT system produces WII systems at 50 �C (Table 2) due to the increased hydrophobicity of CK at the higher temperature. The more significant change imparted by AOT onto the parti- tioning behavior of CK-2,13-E5.6,ave is the reduction of temperature sensitivity, to a level at which the effect of temperature on the val- ues of Kn between 20 and 50 �C is within experimental error (Fig. 4). This result strongly suggests that AOT strongly influences the partitioning behavior of CK-2,13-E5.6,ave. AOT is well known Fig. 4. Semilog plot of partition coefficients for CK,2-13-En vs. ethoxylate chain length, n, as a function of temperature for the binary CK-2,13-E5.6,ave/AOT system. (Filled gray symbols) 6 wt.% AOT and 1.5% CK-2,13-E5.6,ave; (unfilled symbols): 3.75% of each AOT and CK-2,13-E5.6,ave. Symbols represent temperatures, as provided in Fig. 2. Solid and dashed lines represent model fits to the 6 wt.% AOT/1.5% CK-2,13- E5.6,ave; and 3.75% AOT/3.75% CK-2,13-E5.6,ave data, respectively. For these systems Kn values for n > 8 could not be determined due to the concentration of these surfactant molecules being below detection limits in the isooctane excess phase. plots, representing the enthalpy and entropy for the transport of CK-2,13-En, Dhn andDsn, respectively, have a straight-line relation- ship with n [11–13]. Since the systems examined herein were not necessarily at optimal conditions, the relationship between Dhn and Dsn, and n were not linear, but could be described reasonably well by second-order polynomials, Eqs. (4) and (5) (and Figs. 5 and 6), respectively. The polynomial equations are empirical, moreover, having no theoretical basis. Values of the polynomial coefficients employed in Eqs. (4) and (5) for the different surfactant systems present at 7.5 wt.% are given in Table 4. The relationships for Dhn and Dsn vs. n are similar in trend, with maximum values of the two thermodynamic parameters occurring for n = 5–7, agreeing roughly with the average degree of polymerization of CK-2,13- E5.6,ave’s ethoxylate group, 5.6. The addition of a second surfactant significantly lowered both Dhn and Dsn for CK-2,13-E5.6,ave. Eqs. (4) and (5), along with the polynomial coefficient values provided in Table 4, were inserted into Eq. (3) to provide a predic- tive model of log Kn vs. n for the single and binary surfactant sys- tems investigated herein. As depicted in Figs. 2A–D, the models described the surfactant partitioning data well. The model also cor- rectly predicts the curvature of the log Kn vs. n profiles. To assist the discussion, Eq. (3) is first modified to account for deviations of log Kn at T compared to log Kn at THLB: log10 ½Kn T ½Kn T¼THLB ! ¼ Dhn 2:303 � R � 1 T � 1 THLB � � ð8Þ In other words, for a temperature below THLB, log Kn will in- crease relative to log Kn at THLB since the right-hand side of Eq. (6) will be positive and proportional to Dhn. Therefore, the similar- ity of the shape of the log Kn vs. n profile (Fig. 2) and the Dhn vs. n id and Interface Science 352 (2010) 424–435 to induce microemulsion formation in the absence of co-surfactant, with only a very small fraction of surfactant existing in the mono- meric state (clc of 1–2 mmol L�1 [26,27]). Therefore, the majority of CK-2,13 surfactant should co-partition with AOT (into the mid- dle phase), reducing the surfactant concentration in the excess phases. In agreement, the addition of AOT to CK-2,13-E5.6,ave re- duced the bottom phase surfactant concentration as the AOT/CK ratio was increased from 0 to 4:1 g g�1 (Table 2). The increase of the AOT/CK-2,13-E5.6,ave ratio from 1:1 to 4:1 g g�1 did not strongly affect Kn values (Fig. 4). Other groups have also reported that AOT/ alkyl ethoxylate surfactants form temperature-insensitive micro- emulsion systems [28–31]. A recent review summarizes the im- proved ability of binary surfactant systems to form microemulsion systems of increased size or stability compared to single-surfactant systems [32]. The addition of salt to microemulsion systems increases the hydrophobicity of ionic surfactants via the Debye shielding of their charged head groups, but has relatively little effect on the behavior of alkyl ethoxylates [33]. In agreement, the addition of 1 wt.% (0.172 mol L�1) NaCl to the aqueous phase had no effect on the par- titioning behavior of CK-2,13-E5.6,ave for the single-surfactant sys- tem (data not shown). Marquez et al. demonstrated that the inclusion of 8%NaCl in the aqueous phase led to a slight and uniform decrease of Kn for ethoxylates which reflected a decrease of the crit- ical micelle concentration; thus, employment of 1% NaCl would be expected to have a minimal impact [34,35]. On the other hand, the addition of 1% salt did affect the partitioning behavior of the binary CK-2,13-E5.6,ave/AOT systemwhen the surfactants were present at a ratio of 1:1 g g�1 (Fig. 2D).While the partitioning at 20 �Cwas nearly identical to that for the same system in the absence of salt, at in- creased temperature Kn values were decreased slightly further, sug- gesting a slight increase of temperature sensitivitywith the addition of salt (Figs. 2D and 4). The presence of salt will decrease the activity of AOT, thus reducing its influence on the partitioning of CK-2,13-En. This effect may be the underlying cause for the formation ofWI sys- tems, which occurred between 20 and 40 �C, compared to WIII sys- tems formed in the absence of salt (Table 2). However, the reduced activity of AOT did not reverse its weak partitioning to the excess oil phase. In contrast, for an AOT/CK-2,13-E5.6,ave ratio of 4:1 g g�1 in the presence of 1% NaCl(aq), WII systems formed; moreover, both AOT and CK-2,13-E5.6,ave partitioned strongly to the isooctane-rich top phase and the concentration of both surfactants in the aqueous phase was negligible (data not shown). 4.4.1. Thermodynamic model Marquez et al. developed a theoretical approach for interpreting partition coefficient measurements for alkyl and alkyl phenol eth- oxylates at optimal conditions (i.e., maximum solubilization of water and oil in the middle phase, �THLB) [11,12,36], as outlined above in the Theory section. They were able to build a predictive model to calculate the partitioning of alkyl ethoxylate surfactant using as inputs the alkyl and ethoxylate chain lengths, tempera- ture, oil chain length, salinity, and the concentration of n-alkanol co-surfactant. We applied their modeling approach using a simpli- fied set of conditions: uniform alkyl group size for surfactant, iso- octane as oil, and low or no salinity. For their experiments, they mixed a series of commercial surfactant homologues together to produce an ‘‘optimal” WIII system. Herein, the ethoxylate surfac- tant (mixture) employed was identical between experiments, sug- gesting that systems were not (necessarily) at optimal conditions. The linear relationship between log Kn and the inverse of abso- lute temperature, T�1, predicted by Eq. (3) was observed for the partitioning of CK-2,13-E5.6,ave in single and binary surfactant mix- 432 Javier Gomez del Rio et al. / Journal of Collo tures, despite the fact that the plots combined data from 2- and 3- phase systems (Figs. S8 and S9 of the Supporting material). For sys- tems at ‘‘optimal” conditions, the slope and y-intercepts of the A B Fig. 5. Plot of enthalpy of partitioning for CK,2-13-En, Dhn, as a function of ethoxylate chain length, n, for: (A) (h, j) 100 wt.% CK,2-13-E7,ave at 5 and 7.5 wt.%, respectively, and (N) 50% CK,2-13-E7,ave, 50% CK,2-13-E3 (7.5% surfactant), (B) (j) 50% CK,2-13-E7,ave and 50% C8bG1 and (N) 50% CK,2-13-E7,ave, 50% AOT in the presence of 1 wt.% NaCl in the aqueous phase (7.5% surfactant). Eqs. (4) and (5) are substituted into Eq. (3), the following equation is obtained, upon rearrangement: ½log10ðKnÞ T¼THLB ¼ 1 2:303 � R A THLB � A0 � � n2 þ B THLB � B0 � � n � þ C THLB � C0 � �� ð10Þ Comparison of Eqs. (9) and (10) yields the following three relationships: THLB ¼ � A A0 ð11Þ A Javier Gomez del Rio et al. / Journal of Colloid and Interface Science 352 (2010) 424–435 433 B profile (Fig. 5) for T < THLB is supported theoretically. For T > THLB, the right-hand side of Eq. (6) will be negative; therefore, the log Kn vs. n profiles will mirror-Dhn vs. n, thus yielding a downward curving profile (Fig. 2). The model derived herein can be used to predict THLB given that log Kn vs. n is linear at optimal conditions, as discussed above, or: ½log10ðKnÞ T¼THLB ¼ mnþ log K0;HLB ð9Þ where m and log K0,HLB are the slope and y-intercept of log Kn vs. n, respectively. log K0,HLB reflects the contribution of the lipophile (i.e., the n-C13H27 and C2H5 tails and the 1,3-dioxolane group for CK- 2,13-E5.6,ave) toward water–oil partitioning at THLB while m reflects the relative polarity of the ethoxylate monomeric unit at THLB. If Fig. 6. Plot of entropy of partitioning for CK,2-13-En,Dsn, as a function of ethoxylate chain length, n, for: (A) (h, j) 100 wt.% CK,2-13-E7,ave at 5 and 7.5 wt.%, respectively, and (N) 50% CK,2-13-E7,ave, 50% CK,2-13-E3 (7.5% surfactant), (B) (j) 50% CK,2-13-E7,ave and 50% C8bG1 and (N) 50% CK,2-13-E7,ave, 50% AOT in the presence of 1 wt.% NaCl in the aqueous phase (7.5% surfactant). Table 4 Coefficients for the polynomial curve fit relationship for enthalpy (Dhn) and entropy (Dsn) v single and binary surfactant systems containing 7.5 wt.% surfactant (unless otherwise not Surfactant system Dhn, kJ mol�1 A B CK-2,13-E5.6,ave (5 wt.%) �1.0 ± 0.3 15.0 ± 4.1 CK-2,13-E5.6,avea �1.7 ± 0.3 23.1 ± 4.1 CK-2,13-E5.6,ave/CK-2,13-E3 1:1 g/g �0.7 ± 0.3 14.6 ± 3.9 CK-2,13-E5.6,ave/C8bG1 1:1 g/g �2.4 ± 2.1 29.7 ± 21.1 CK-2,13-E5.6,ave/AOT 1:1 g/g b b CK-2,13-E5.6,ave/AOT 6.0:1.5 g/g b b CK-2,13-E5.6,ave/AOT 1:1 g/g, 1% NaCl �2.4 ± 0.8 28.7 ± 9.3 a Polynomial curve fits applied to thermodynamic values derived from data obtained b For CK-2,13-E5.6,ave/AOT systems in the absence of NaCl, Dhn was assumed to be zer c A linear relationship was employed for Dsn vs. n for CK-2,13-E5.6,ave/AOT systems. m ¼ 1 2:303 � R B THLB � B0 � � ð12Þ logK0;HLB ¼ 1 2:303 � R C THLB � C 0 � � ð13Þ Estimates of THLB, m, and log K0,HLB for single and mixed surfac- tant systems that contain CK-2,13-E5.6,ave calculated using Eqs. (9)– (12) are given in Table 5. The THLB value obtained for 7.5% surfac- tant, 40.2 �C, agrees with T�HLB obtained from the ‘‘fish” phase dia- gram (Fig. 1), 41 �C. The increase of THLB to 51.2 �C for a decrease of surfactant concentration to 5%, is consistent with the literature [19] and with the surfactant partitioning trends described above (Fig. 2). The value of the slope of the log Kn vs. n profile, m, for the single-surfactant system, 0.21 ± 0.08, is lower than the value obtained for octylphenyl ethoxylates and linear alkyl ethoxylates at optimal conditions (25 �C), 0.45, reflecting the lower polarity of the ethoxylate group as the temperature increases [11,36]. The value of the y-intercept (log K0,HLB), �1.5 (for 7.5 and 10% surfac- tant; �1.1 for 5% surfactant), is larger than the values reported for octylphenyl and linear alkyl ethoxylates, which are near �4.0 [11,36]. The difference reflects the greater polarity of the surfactant molecule minus its ethoxylate chain, attributable to the 1,3-dioxo- lane ring, the occurrence of branching in the alkyl tail region [21], and the methoxy terminal moiety of the head group. The addition of a second surfactant to CK-2,13-E5.6,ave signifi- cantly affected the values of the three parameters. The addition of CK-2,13-E3 increased THLB to 74.5 �C, reduced m to 0.14, and in- creased the absolute value of log K0,HLB (Table 5). All three results confirm that the CK-2,13-E3/-E5.6,ave mixture is more hydrophobic than neat CK-2,13-E5.6,ave, as would be expected. A higher THLB is needed for the more ethoxylated members of CK-2,13-E5.6,ave to be- come hydrophobic, so that the partitioning of CK-2,13-E3 and - E5.6,ave are more consistent with each other. The decrease of log K0,HLB with the addition of CK-2,13-E3 suggests the latter surfac- tant’s role as a hydrophobic co-surfactant, leading to an increased partitioning of CK-2,13-E5.6,ave to the isooctane phase, consistent with the lowering of Kn for alkylphenyl ethoxylates by the co-sur- s. ethoxylate monomeric units of CK-2,13-E5.6,ave (n), Eqs. (4) and (5), respectively, for ed). Dsn, J mol�1 K�1 C A0 B0 C0 �4.3 ± 10.6 �3.1 ± 1.0 42.2 ± 11.9 14.6 ± 33.3 �24.6 ± 11.6 �5.3 ± 1.1 69.7 ± 12.9 �57.8 ± 36.1 �6.4 ± 10.8 �2.1 ± 0.9 39.4 ± 11.4 19.6 ± 31.7 �41.3 ± 49.8 �7.3 ± 6.5 83.1 ± 65.9 �93.4 ± 156 b c 6.71 ± 0.44 �18.1 ± 2.4 b c 6.88 ± 0.5 �19.2 ± 2.8 �46.4 ± 24 �7.3 ± 2.8 81.5 ± 31.2 �119 ± 80 for 7.5 + 10.0%. o, indicating that surfactant partitioning was independent of temperature. ous phase; in turn, CK-2,13-E5.6,ave increased the partitioning of for the purification of proteins by liquid–liquid extraction. 00OR22725. We thank Dr. Guangming Luo for help with SANS data n temperature, calculated from the mathematical model. id a factant n-pentanol [35]. The decrease ofm for the CK-2,13-E5.6,ave/- E3 mixture compared to that for the single CK-2,13-E5.6,ave surfac- tant system reflects a decrease in the hydrophilicity of the ethoxy- late groups at the high THLB value of 74.5 �C. The addition of C8bG1 and AOT increased THLB slightly, to 57.9 �C and 54.4 �C, respectively (Table 5), which may reflect the increased driving force for CK-2,13-E5.6,ave solubilization into the aqueous phase by the addition of the second surfactant. Consistent with this trend is the slightly higher aqueous phase concentration of CK- 2,13-E5.6,ave (Table 2). In agreement, a recent thermodynamic study concluded for the nonylphenyl ethoxylate/sodium dihexyl sulfo- succinate binary system that the latter increases hydration of the ethoxylate groups, which would increase their polarity, hence pro- moting their partition to the aqueous phase [37]. In contrast, the contribution of CK-2,13s hydrophobic moiety toward partitioning was not strongly affected by C8bG1 or AOT, evidenced by log K0,HLB values for the binary surfactant systems and CK-2,13-E5.6,ave single-surfactant system being similar (Table 5). Also, the value of m is significantly higher when C8bG1 or AOT is added, in contrast to the anticipated decrease of m due to the slight increase of THLB. The increase ofm is perhaps caused by the increased hydrophilicity of the ethoxylate groups’ nano-environment imparted by the other surfactant. The increased value of m for the binary surfactant sys- tems expands the range of log Kn values: 2.5 order of magnitude (Fig. 2C), compared to two orders of magnitude for the CK-2,13- E5.6,ave single-surfactant system, for 3 6 n 6 8 (Fig. 2A). The approach described in this paper was useful for us to deter- mine the optimal temperature, T�HLB, forWIII system formation using mixed surfactants systems containing CK-2,13-E5.6,ave at a given composition. Knowledge of T�HLB was important for us since at T � HLB an optically clear and stable WIII system formed most rapidly. The latter criterion was important for purifying proteins through their selective partition into the middle, bicontinuous microemulsion phase, to prevent the denaturation of proteins invokedby emulsions forming at a liquid–liquid interface. This approach required a simple Surfactant concentration THLB (�C)a mb log K0,HLBc CK-2,13-E5.6,ave (5%) 51.2 ± 22.8 0.21 ± 0.08 �1.5 ± 4.9 CK-2,13-E5.6,ave (7.5%) 40.2 ± 11.5 0.21 ± 0.05 �1.1 ± 0.8 CK-2,13-E5.6,ave/CK-2,13-E3 (3.75%/ 3.75%) 74.5 ± 46.5 0.14 ± 0.05 �2.0 ± 4.6 CK-2,13-E5.6,ave/C8bG1 (3.75%/3.75%) 57.9 ± 72.0 0.34 ± 0.37 �1.6 ± 3.4 CK-2,13-E5.6,ave/AOT (3.75%/3.75%)d 54.4 ± 28.4 0.33 ± 0.16 �1.2 ± 1.0 a HLB temperature, calculated using Eq. (11). b Slope of log Kn vs. n at THLB, calculated using Eq. (12). c y-Intercept of log Kn vs. n at THLB, calculated using Eq. (13). d 1% NaCl present in aqueous phase. Table 5 Values of HLB temperature and slope and y-intercept of log K vs. n at the HLB 434 Javier Gomez del Rio et al. / Journal of Collo experimental protocol: formation of four or more microemulsion solutions of equal composition, with each solution incubated at dif- ferent temperatures. Subsequently, aliquots of the bottom and top phases were withdrawn and analyzed by reversed-phase HPLC for separation as a function of the surfactants’ ethoxylate size. TheHPLC data were then analyzed as described above. The alternative ap- proach would be to perform an exhaustive temperature scan of a microemulsion system, which may require the formation of several microemulsion samples and several hours to days of time for equil- ibration to occur. The approach described herein may be useful for analysis of other alkyl ethoxylate-based microemulsion systems. 5. Conclusions Water–oil partition coefficients for CK-2,13-E5.6,ave span more than three orders of magnitude as a function of n for 3 6 n 6 10, collection and Dr. J.S. Lin for technical assistance. We acknowledge the support of the National Institute of Standards and Technology, US Department of Commerce, and ORNL, US Department of Energy, in providing SANS facilities used in this work. Appendix A. Supplementary material Further information on the HPLC methodology employed, preli- minary SANS data, form factor selection and Guinier analysis of the SANS data, a plot of volume fraction of the o/w-microemulsions (obtained from the form factor-structure factor modeling of the SANS data), and van’t Hoff plots of the partition coefficients, Kn, vs. inverse of temperature, are provided. Supplementary data asso- ciated with this article can be found, in the online version, at doi:10.1016/j.jcis.2010.08.076. References [1] A. Tehrani-Bagha, K. Holmberg, Curr. Opin. Colloid Interface Sci. 12 (2007) 81. [2] M. Stjerndahl, D. Lundberg, K. Holmberg, in: K. Holmberg (Ed.), Novel Surfactants, Surfactant Science Series, vol. 114, Marcel Dekker, New York, 2003, p. 317. [3] J. Eastoe, M.S. Dominguez, H. Cumber, P. Wyatt, R.K. Heenan, Langmuir 20 (2004) 1120. [4] J. Eastoe, M. Sanchez-Dominguez, H. Cumber, G. Burnett, P. Wyatt, R.K. Heenan, Langmuir 19 (2003) 6579. [5] J. Eastoe, Prog. Colloid Polym. Sci. 133 (2006) 106. [6] M. Iyer, D.G. Hayes, J.M. Harris, Langmuir 17 (2001) 6816. [7] M.E. Rairkar, M.E. Diaz, M. Torriggiani, R.L. Cerro, J.M. Harris, S.E. Rogers, J.A. Although applied to a 1,3-dioxolane alkyl ethoxylate, the modeling approach described herein may be applicable to alkyl ethoxylates and alkylphenyl ethoxylates that possess broad ethoxylate distributions. Acknowledgments This work was supported by the National Science Foundation grant BES-0437507. The research performed at Oak Ridge National Laboratory’s Center for Structural Molecular Biology (CSMB) was supported by the Office of Biological and Environmental Research, using facilities supported by the US Department of Energy, man- aged by UT-Battelle, LLC under Contract No. DE-AC05- AOT into the aqueous phase, resulting in WI and WIII systems rather than WII systems. The addition of octyl-b-glucoside in- creased the partitioning of CK-2,13-E5.6,ave into the aqueous phase. A semi-empirical thermodynamic model based in part on the mod- el derived by Salager et al. [11,12,36], described the data well. The model effectively predicted the temperature required for WIII for- mation as a function of the relative amounts of surfactants in the binary surfactant systems described above and the salt concentra- tion, which in turn has been employed to find the compositions and conditions for forming WIII systems near room temperature partially explaining the surfactant’s poor efficiency compared to commercial alkyl ethoxylates [7]. Partition coefficients for CK- 2,13-E5.6,ave are higher than those for ethoxylated fatty alcohols and alkyl phenyl ethoxylates, presumably due to the higher rela- tive polarity of the former’s tail group. The addition of CK-2,13- E3 lowered the partition coefficients for CK-2,13-E5.6,ave, leading to the formation of Winsor-III microemulsion systems at lower, near-ambient, temperatures. The addition of AOT removed the temperature sensitivity of CK-2,13-E5.6,ave between 20 and 50 �C and increased the solubilization of CK-2,13-En, n < 7, into the aque- nd Interface Science 352 (2010) 424–435 Eastoe, J. Eastoe, J.A. Gomez del Rio, D.G. Hayes, Colloids Surf. A 301 (2007) 394. [8] M.E. Rairkar, D.G. Hayes, J.M. Harris, Biotechnol. Lett. 29 (2007) 767. [9] M.H. Alkhatib, pH-Degradable 1,3-Dioxolane Surfactants: Synthesis and Application in Mixed Surfactant Systems, PhD Dissertation, University of Alabama in Huntsville, Huntsville, AL, 2006. [10] J.A. Gomez del Rio, Extraction of Proteins by Winsor-III Microemulsion Systems, PhD Dissertation, University of Tennessee, Knoxville, TN, 2008. [11] N. Marquez, R. Anton, A. Graciaa, J. Lachaise, J.-L. Salager, Colloids Surf. A 100 (1995) 225. [12] J.-L. Salager, N. Marquez, A. Graciaa, J. Lachaise, Langmuir 16 (2000) 5534. [13] N. Marquez, B. Bravo, G. Chavez, F. Ysambertt, J.L. Salager, Anal. Chim. Acta 405 (2000) 267. [14] M. Iyer, Synthesis and Characterization of pH-Degradable Nonionic Surfactants and Their Applications in Microemulsions, PhD Dissertation, University of Alabama in Huntsville, Huntsville, AL, 2002. [15] G.W. Lynn, W. Heller, V. Urban, G.D. Wignall, K. Weiss, D.A.A. Myles, Phys. B 385–386 (2006) 880. [16] S.R. Kline, J. Appl. Crystallogr. 39 (2006) 895. [17] M. Kotlarchyk, S.H. Chen, J. Chem. Phys. 79 (1983) 2461. [18] J.B. Hayter, J. Penfold, Mol. Phys. 42 (1981) 109. [19] M. Ben Ghoulam, N. Moatadid, A. Graciaa, J. Lachaise, Langmuir 20 (2004) 2584. [20] J. Israelachvili, Colloids Surf. A 91 (1994) 1. [21] N. Marquez, R.E. Anton, A. Graciaa, J. Lachaise, J.-L. Salager, Colloids Surf. A 131 (1998) 45. [22] M.H. Alkhatib, D.G. Hayes, V.S. Urban, J. Surfactants Deterg. 12 (2009) 277. [23] L.D. Ryan, K.-V. Schubert, E.W. Kaler, Langmuir 13 (1997) 1510. [24] L.D. Ryan, E.W. Kaler, Colloids Surf. A 176 (2001) 69. [25] T.K. De, A. Maitra, Adv. Colloid Interface Sci. 59 (1995) 95. [26] K. Mukherjee, S.P. Moulik, D.C. Mukherjee, Langmuir 9 (1993) 1727. [27] K. Murali Monoj, R. Jayakumar, S.K. Rakshit, Langmuir 12 (1996) 4068. [28] H. Kunieda, K. Hanno, S. Yamaguchi, K. Shinoda, J. Colloid Interface Sci. 107 (1985) 129. [29] B.P. Binks, P.D.I. Fletcher, D.J.F. Taylor, Langmuir 13 (1997) 7030. [30] B.P. Binks, P.D.I. Fletcher, D.J.F. Taylor, Langmuir 14 (1998) 5324. [31] C. Tongcumpou, E.J. Acosta, L.B. Quencer, A.F. Joseph, J.F. Scamehorn, D.A. Sabatini, S. Chavadej, N. Yanumet, J. Surfactants Deterg. 6 (2003) 191. [32] J. Eastoe, A. Bumajdad, in: S.G. Pandalai (Eds.), Recent Research Developments in Physical Chemistry, Part II, Transworld Research Network, vol. 4, Trivandrum, India, 2000, p. 337. [33] R. Aveyard, B.P. Binks, J. Chen, J. Esquena, P.D.I. Fletcher, R. Buscall, S. Davies, Langmuir 14 (1998) 4699. [34] N. Marquez, B. Bravo, G. Chavez, F. Ysambertt, J.L. Salager, Anal. Chim. Acta 452 (2002) 129. [35] N. Marquez, A. Graciaa, J. Lachaise, J.-L. Salager, Langmuir 18 (2002) 6021. [36] N. Marquez, B. Bravo, F. Ysambertt, G. Chavez, N. Subero, J.L. Salager, Anal. Chim. Acta 477 (2003) 293. [37] E.J. Acosta, A.S. Bhakta, J. Surfactants Deterg. 12 (2009) 7. Javier Gomez del Rio et al. / Journal of Colloid and Interface Science 352 (2010) 424–435 435 Partitioning behavior of an acid-cleavable, 1,3-dioxolane alkyl ethoxylate, surfactant in single and binary surfactant mixtures for 2- and 3-phase microemulsion systems according to ethoxylate head group Introduction Theory Materials and methods Materials Methods Phase diagrams Partitioning behavior of CK-2,13-E5.6,ave HPLC analysis Small-angle neutron scattering Results and discussion “Fish” phase diagram for heptane/CK-2,13-En/water Partitioning of CK-2,13-En in single surfactant microemulsion systems Aggregation state of CK-2,13-En in the aqueous phase of Winsor-I and -III systems Partitioning of CK-2,13-En in Binary Surfactant Microemulsion Systems Thermodynamic model Conclusions Acknowledgments Supplementary material References