Nonionic Surfactants with Linear and Branched Hydrocarbon Tails

Thus, the real surface observed in the Porod region is increased. An empirical approach based on the description of fractal scattering by Beaucage23 a...
0 downloads 0 Views 175KB Size
6526

Langmuir 2007, 23, 6526-6535

Nonionic Surfactants with Linear and Branched Hydrocarbon Tails: Compositional Analysis, Phase Behavior, and Film Properties in Bicontinuous Microemulsions Christian Frank, Henrich Frielinghaus, and Ju¨rgen Allgaier* Institut fu¨r Festko¨rperforschung, Forschungszentrum Ju¨lich GmbH, 52425 Ju¨lich, Germany

Hartmut Prast Zentralabteilung fu¨r Chemische Analyse, Forschungszentrum Ju¨lich GmbH, 52425 Ju¨lich, Germany ReceiVed December 22, 2006. In Final Form: March 27, 2007 Nonionic alcohol ethoxylates are widely used as surfactants in many different applications. They are available in a large number of structural varieties as technical grade products. This variety is mainly based on the use of different alcohols, which can be linear or branched and contain primary, secondary, or tertiary OH groups. Technical grade products are poorly defined as they are composed of alcohol mixtures being different in chain length and structure. On the other hand, monodisperse alcohol ethoxylates are commercially available; however, these surfactants exist only with primary and linear alcohols. In the field of microemulsion research the monodisperse alcohol ethoxylates are widely used. The phase behavior and film properties of these surfactants were studied intensively with respect to the size of the hydrophilic and hydrophobic moieties. Due to the lack of appropriate model surfactants until now, there is little information on how the structure of the hydrocarbon tail influences the microemulsion behavior. To examine structural influences, we synthesized a series of surfactants with the composition C10E5 and having different linear and branched hydrocarbon tails. The surfactants were monodisperse with respect to the hydrocarbon tail but polydisperse with respect to the ethoxylation degree. However, a detailed characterization showed that they were similar concerning the average ethoxylation degree and EO chain length distribution. The phase behavior was investigated for bicontinuous microemulsions, and the film properties were analyzed by small-angle neutron scattering (SANS). Our results show that the structure of the hydrocarbon tail strongly influences the microemulsion behavior. The most efficient surfactant is obtained if the hydrocarbon tail is linear and the hydrophilic group is attached in the C-1 position. Surfactants having the hydrophilic group bound to the C-2 or C-4 position or which contain a branched hydrocarbon tail are less efficient and exhibit in most cases visibly lower phase inversion temperatures. Both the efficiency and temperature behavior mainly can be explained on the basis of increased bulkiness of the branched structures compared to the fully linear version. The phase behavior results are largely confirmed by the SANS investigations. Those results show that the fully linear surfactant exhibits the most rigid interfacial film. In additional experiments this surfactant was compared with its monodisperse analogue. According to the phase diagrams, the surfactant having the polydisperse hydrophilic moiety is drastically more efficient although the film stiffnesses are almost identical.

Introduction Alcohol ethoxylate surfactants are widely used in fundamental research for investigations of microemulsions. They are available as chemically pure monodisperse compounds and contain exclusively linear hydrocarbon tails. The phase behavior, microstructure, and film properties of these surfactants were studied extensively with respect to the lengths of the hydrocarbon tails and the ethylene oxide (EO) chains.1-9 On the other hand, technical grade alcohol ethoxylates play an important role in cleaning and laundering applications. The alcohols used for the synthesis of those alcohol ethoxylates can originate from natural or synthetic sources.10 Depending on the origin, they can be linear or branched and contain primary, secondary, or tertiary (1) Kahlweit, M. J. Colloid Interface Sci. 1982, 90, 197. (2) Kahlweit, M.; Strey, R. Angew. Chem., Int. Ed. Engl. 1985, 24, 654. (3) Kahlweit, M.; Strey, R.; Firman, P.; Haase, D.; Jen, J.; Schoma¨cker, R. Langmuir 1988, 4, 499. (4) Aveyard, R.; Binks, B. P.; Fletcher, P. D. I. Langmiur 1989, 5, 1210. (5) Lindman, B.; Shinoda, K.; Olsson, U.; Anderson, D.; Karlstro¨m, G.; Wennerstro¨m, H. Colloids Surf. 1989, 38, 205. (6) Kahlweit, M.; Strey, R.; Busse, G. J. Phys. Chem. 1990, 94, 3881. (7) Strey, R. Colloid Polym. Sci. 1994, 272, 1005. (8) Sottmann, T.; Strey, R. J. Chem. Phys. 1997, 106, 6483. (9) Sottmann, T.; Strey, R.; Chen, S.-H. J. Chem. Phys. 1997, 106, 8606. (10) McCoy, M. Chem. Eng. News 2005, 83 (4), 21.

OH groups.11 Technical grade alcohol ethoxylates usually contain mixtures of different alcohols and exhibit a more or less broad distribution of different ethoxylation degrees. Despite their wide use in applications and usefulness in microemulsions, there are few reports where the technical alcohol ethoxylates were examined systematically with respect to their microemulsion behavior.12-14 The main outcome of these investigations is that the phase behavior with respect to the chain lengths of the hydrophobic and the hydrophilic moieties is qualitatively the same as what was found before for monodisperse model surfactants. However, all these reports contain no information on how the architecture of the hydrocarbon tails influences the phase behavior. Technical grade alcohol ethoxylates are a potential surfactant source for such a study as they are available in different linear and branched versions. However, they are poorly defined with respect to composition. Therefore, we decided to synthesize our own model alcohol ethoxylates with differently branched C10 (11) See, for example: Noweck, K. Fatty Alcohols. In Ullmann’s Encyclopedia of Industrial Chemistry, 73rd ed.; John Wiley & Sons: New York, 2006; electronic release. (12) Wormuth, K. R.; Geissler, P. R. J. Colloid Interface Sci. 1991, 146, 320. (13) Do¨rfler, H. -D.; Grosse, A. Tenside, Surfactants, Deterg. 1999, 36, 29. (14) Sottmann, T.; Lade, M.; Stolz, M.; Schoma¨cker, R. Tenside, Surfactants, Deterg. 2002, 39, 1.

10.1021/la0637115 CCC: $37.00 © 2007 American Chemical Society Published on Web 05/10/2007

Nonionic Surfactants with Hydrocarbon Tails

Langmuir, Vol. 23, No. 12, 2007 6527

Scheme 1. Chemical Structure of Alcohol Ethoxylates with Differently Branched Hydrocarbon Tails

hydrocarbon tails. The structure of these surfactants is shown in Scheme 1. They contain chemically pure hydrocarbon tails but are polydisperse with respect to the ethoxylation degree. After synthesis, the average ethoxylation degrees were analyzed by 1H NMR and the chain length distributions of the EO chains were monitored by size exclusion chromatography (SEC) and highperformance liquid chromatography (HPLC). After the analysis was carried out for the freshly synthesized surfactants, the same procedure was applied to determine the ethoxylation degrees and chain length distributions of the oil-soluble surfactant fractions in the microemulsions. These analysis results allowed exact determination of the EO composition of the interfacially active surfactant fractions. This extended characterization is of crucial importance because not only the average ethoxylation degree but also the EO chain length distribution influences the interfacial properties. Both quantities have to be identical for all surfactants to attribute possible differences in their behavior to the different structures of the hydrophobic unit. After having ensured that the surfactants only differ in the structure of their hydrophobic units, the phase behavior was examined with water and decane at symmetric water/oil volume fractions. By small-angle neutron scattering (SANS) experiments the structure of the microemulsions was obtained. The dependence of the domain size as a function of the surfactant content revealed the membrane thickness. The Gaussian random field theory derived an expression for the bending rigidity. All parameters obtained from phase diagrams and SANS were compared to obtain detailed information about the different surfactants. In another series of experiments the fully linear surfactant containing the polydisperse hydrophilic group was compared with its monodisperse analogue. Experimental Section Surfactant Synthesis. All synthetic manipulations except the surfactant synthesis were carried out at a high-vacuum line (pressure 10-4 to 10-5 mbar) or in a glovebox, filled with argon (M Braun, Unilab). The flasks for the manipulations were equipped with Teflon stopcocks, which allowed transfer of materials between the vacuum line and glovebox without air contamination. The flasks which were exposed to overpressure were pressure tested to 4-12 bar, depending on the size of the flask. Surfactant syntheses were carried out in a double mantle steel reactor (Buchi ecoclave 075). The steel reactor could be connected to the vacuum line but also could be flushed with argon. Temperature control was achieved by using a Huber Unistat 380w thermostat.

Potassium metal (Fluka, >98%) was used as received. 1-Decanol (lot purity 99.2%), 3,7-dimethyl-1-octanol (lot purity 99.5%), 2-decanol (lot purity 99.0%), and 4-decanol (lot purity 98.8%) were degassed at the vacuum line prior to use. Ethylene oxide (Fluka, >99.8%) was condensed into a flask at liquid nitrogen temperature and degassed. Alcohol ethoxylates were synthesized by first partially reacting the alcohols with potassium metal. For primary alcohols the metalation degree was 25%, and for secondary alcohols it was 33%. The metalation reaction was carried out by filling alcohols and potassium into flasks equipped with Teflon stopcocks in the glovebox. The flasks were degassed at the vacuum line, closed, and then warmed to 90 °C overnight. In the case of 4-decanol the reaction was continued for 3 h at 120 °C. The amounts of materials were calculated such that only a slight hydrogen overpressure could develop. The alcoholalcoholate mixtures were transferred to the glovebox and filled into gastight syringes which were equipped with steel valves. The mixtures were filled into the steel reactor under argon flow. Then the reactor was cooled to -80 °C, and argon was pumped off via the connection to the vacuum line, which also was used to condense in EO. After that step was finished the reactor was pressurized with argon, stirred, and slowly heated to 80 °C. In the case of 4-decanol the final temperature was 100 °C. The heating rate was adjusted such that the pressure did not exceed 7 bar. After having reached the maximum temperature, the reaction was continued until no pressure drop was visible. This took usually 1-2 h. After that time the reactor was emptied and the product neutralized with acetic acid. Precipitated potassium acetate was eliminated by centrifugation. As only part of the potassium acetate precipitated, the surfactants were dissolved in a 20-fold excess of freshly distilled heptane, which caused more salt to precipitate. After the solutions were filtered, the heptane was distilled off and the surfactants were stirred under high vacuum for approximately 4 h to remove residues of solvent, acetic acid, and water. The samples synthesized with 2-decanol and 4-decanol were additionally heated under high-vacuum conditions at 50 °C to partially remove the residual alcohols. The obtained ethoxylation degrees for the different surfactants are listed in Table 1. Characterization. The NMR spectra were obtained on a Varian Inova 400 MHz spectrometer. All samples were measured at 295 K in CDCl3 with a 5 mm PFG AutoX DB probe. The ethoxylation degrees were calculated by comparing the signal intensities of the alkyl protons of the alcohols between 0.8 and 1.6 ppm with the signal intensities of the EO protons between 3.5 and 3.7 ppm. Some NMR spectra of the surfactants were recorded in fully deuterated decane, extracted as excess phases from microemulsion samples. In addition to the determination of the ethoxylation degrees, the NMR results of samples 1-C10E5 and 4-C10E5 in d-decane were used to calculate the contents of free alcohol. In the case of sample 1-C10E5 this was done by comparing the signal intensities of the CH2 group next to the OH group in the free alcohol at 3.45 ppm and the CH2 group next to the EO chain in the ethoxylated molecules at 3.35 ppm. For 4-C10E5 the CH2 group next to the OH group in the free alcohol was hidden by the signal of the EO protons. In that case the intensity of the protons at the carbon atom next to the EO chain in the ethoxylated molecules was compared with the intensity of the CH3 end group. The missing intensity of the protons at the carbon atom next to the EO chain was assigned to free alcohol. NMR was also used to determine the absolute surfactant concentrations in excess

Table 1. Alcohol Ethoxylates with Different Hydrophobic Moieties: Raw Products after Synthesis and Intermediate Products after Partial Removal of Free Alcohol raw product

intermediate product

sample

alcohol

ethoxylation degree of overall producta

concn (mol %) of free alcoholb

ethoxylation degree of ethoxylated alcohol fractionc

concn (mol %) of free alcoholb

1-C10E5 DM-C10E5 2-C10E5 4-C10E5

1-decanol 3,7-dimethyl-1-octanol 2-decanol 4-decanol

5.56 5.62 4.59 4.01

5.2 7.0 24.4 31.4

5.80 5.92 5.94 5.77

19.1 18.9

a

Measured by 1H NMR. b Measured by SEC. c Calculated from 1H NMR/SEC results.

6528 Langmuir, Vol. 23, No. 12, 2007

Frank et al.

decane. This was done by comparing the signal intensities of the surfactant signals and the solvent signals. The intensities of the solvent signals were calibrated in a separate measurement with a 1-decanol sample of known concentration. SEC experiments were carried out using a Polymer Laboratories PL-GPC 50 instrument in combination with two Polymer Laboratories Oligopore columns and a Viscotec model TDA 300 triple detector. The solvent was THF at a flow rate of 1 mL/min; it was degassed using a Viscotec model VE 7510 instrument. The content of free alcohol was calculated by comparing the refractive index (RI) signal intensities of the alcohol signals in the surfactant samples with the signal intensities of alcohol samples of known concentrations. In addition to SEC, HPLC was used to determine the EO distribution of the ethoxylates. Due to the absence of chromophoric groups the samples were derivatized by mixing equal amounts of alcohol ethoxylate and phenyl isocyanate (Fluka, 99%) and heating the mixtures at 60 °C for 30 min before separation by HPLC. The HPLC instrument consisted of a Merck Hitachi L6220 gradient pump, a Merck Hitachi L4550 A DAD detector, a Merck Hitachi 5025 column oven, and the Merck Hitachi D6000 data system. Injection was performed manually with a Rheodyne 9125 injection valve. The injected sample volume was between 50 and 100 µL. UV detection was performed at 230 nm. For the separation a Hypersil APS 3 µm, NH2-bonded phase with the dimensions 125 mm × 4.6 mm was used. To separate the sample mixtures, a method was developed which was based on an existing HPLC method. That method was used for the determination of alkylphenol ethoxylates (APEO) in environmental samples.15 The new method used as the mobile phase mixtures of n-hexane (A) and n-hexane/2-propanol (80:20 (v/v)) (B). The applied gradient is given in the following table:The new method was first verified with APEO with different time (min)

A:B

flow rate (mL/min)

time (min)

A:B

flow rate (mL/min)

0.0 1.0 11.0 19.0 20.0

97:3 97:3 75:25 50:50 50:50

1.0 1.0 1.0 1.0 1.5

21.0 22.0 31.9 32.0

50:50 97:3 97:3 97:3

1.5 1.5 1.5 1.0

EO chain lengths and then with known derivatives of surfactants at high concentrations. Finally the method was modified to fit with the typical concentrations of the investigated system. Phase Diagrams. Decane (Aldrich) with a purity of >99% and water were used for the preparation of the microemulsions. All investigations were carried out at a constant water to oil volume ratio of 1. Phase diagrams were recorded by adding appropriate water and oil quantities to the initial water-oil-surfactant mixture to lower the surfactant concentration. In a thermostated water bath at each surfactant concentration the mixtures were investigated visually in transmitted light using glass tubes of 15 mm diameter. Lamellar phases were characterized visually with the help of crosspolarizers. Turbidity indicated the coexistence of the lamellar phase with a bicontinuous phase. For some samples SANS was used for this determination. This is described in the next section. In addition, three-phase systems with all surfactants were produced using fully deuterated decane as the oil. The surfactant concentrations were 3-6 wt % below the fishtail points. Excess decane was separated and used for 1H NMR and HPLC measurements to analyze the surfactant fractions being solubilized in the oil excess phases. SANS Measurements. SANS experiments were performed at the diffractometer KWS-2 at the research reactor FRJ-2 in Ju¨lich. The white beam was monochromated with a velocity selector by Dornier. The neutron wavelength was λ ) 6 Å with an fwhm bandwidth of ∆λ/λ ) 0.1. The collimation aperture was fixed to 3 × 3 cm2 at a distance of 20 m to the sample. The sample of 1 mm thickness was irradiated on an area of 1 × 1 cm.2 The detector with (15) Schmitt, T. M. In Analysis of Surfactants; Schick, M. J., Fowkes, F. M., Eds.; Surfactant Science Series, Vol. 40; Marcel Dekker: New York, 1992; Chapter 6.

an active area of 50 × 50 cm2 was adjusted for the right Q range at distances between 8 and 1.25 m. Thus, the scattering vector Q spanned a range of 5 × 10-3 to 0.3 Å-1. The collected data were corrected for the detector efficiency and dark current background; this intensity was calibrated in absolute units by a secondary standard plexiglass and radially averaged to obtain the macroscopic cross section dΣ/dΩ(Q) of an isotropic sample. The batch samples were heated in a water bath before being filled in Hellma quartz cells. To the actual sample in the cell was added further surfactant through a capillary hole in the Teflon stopper. The masses of the batch sample and further surfactant were determined with a precision of better than 1 mg. Every new sample was stirred by the remaining air bubble in the rotating cell, and the phase stability was checked in the temperature bath. At high surfactant contents slight turbidity and a scattering pattern with two peaks indicated the coexistence of a microemulsion phase with a lamellar phase. The agreement of the temperature of the SANS furnace and the water bath was better than 0.1 °C.

Results and Discussion Surfactant Synthesis and Characterization. The aim of this work is to investigate alcohol ethoxylates with respect to the different architectures of their hydrocarbon tails. This requires a series of surfactants having linear and differently branched hydrophobes of the same chain length, whereas the hydrophilic parts have to be identical with respect to ethoxylation degrees and EO chain length distribution. Technical grade alcohol ethoxylates are not useful for such investigations as the hydrocarbon moieties usually consist of product mixtures. In addition, it is difficult to find products with identical hydrophilic parts. For this reason a series of alcohol ethoxylates were synthesized using the following C10 alcohols: 1-decanol (1-C10OH), 3,7-dimethyl-1-octanol (DM-C10OH), 2-decanol (2C10OH), and 4-decanol (4-C10OH). The alcohols were partially transformed into their potassium salts and conventionally ethoxylated. The chemical structures of the surfactants are shown in Scheme 1. In anionic ethoxylation reactions the reactivity of primary hydroxyl groups is higher than the reactivity of secondary hydroxyl groups. Therefore, at a given ratio of EO to alcohol, the amount of unreacted, free alcohol is higher if a secondary alcohol is used than is the case for a primary alcohol. As a result, the average ethoxylation degree of the ethoxylated alcohol fraction is higher for the secondary alcohol. For this reason the strategy for the surfactant synthesis was to obtain in the first step products with the same ethoxylation degrees of the ethoxylated alcohol fractions. Therefore, higher ethoxylation degrees were chosen for the surfactants 1-C10E5 and DM-C10E5 synthesized with primary alcohols than for 2-C10E5 and 4-C10E5 synthesized with secondary alcohols. In the second step, the different contents of free alcohol were adjusted. The surfactants were analyzed with different techniques. 1H NMR was used to determine the average ethoxylation degrees. EO chain length distributions were monitored by SEC and HPLC. In SEC the separation depends on the molecular size, whereas in HPLC it is based on the adsorption properties of the analyte. In principle both methods are useful to obtain chain length distributions. For our purposes it turned out that the combination of SEC and HPLC allowed a maximum of information about the surfactants’ composition to be obtained. The ethoxylation degrees of the raw products were measured by 1H NMR. The values for the synthesized samples are summarized in Table 1. The free alcohol contents were analyzed by SEC. Figure 1a shows the SEC trace of sample 2-C10E5. The rightmost signal at an elution volume of 16.8 mL corresponds to the free alcohol. It is almost fully separated from the signals

Nonionic Surfactants with Hydrocarbon Tails

Langmuir, Vol. 23, No. 12, 2007 6529

Figure 1. (a) SEC and (b) HPLC spectra of the raw product of 2-C10E5.

originating from the ethoxylated chains at lower elution volumes. The free alcohol content in the surfactants was determined by comparing the signal intensities of free alcohol in the surfactant samples with the signal intensities of alcohol samples, both of known concentration. The values found are collected in Table 1. As expected the surfactants synthesized with secondary alcohols contain much more free alcohol than the surfactants synthesized with primary alcohols. With the knowledge of the free alcohol contents it was possible to calculate the ethoxylation degrees of the ethoxylated alcohol fractions in the samples. The values given in Table 1 show that the ethoxylation degrees are between 5.77 and 5.94 EO units per chain and differ little. In the next step alcohol was partially removed from the surfactants 2-C10E5 and 4-C10E5 under high-vacuum conditions and moderate warming. From the comparison of the SEC traces before and after alcohol removal it was found that under these conditions almost uniquely alcohol was removed. With regard to the signals from the ethoxylated fractions of the products, the spectra were indistinguishable, except for the molecules containing only one EO unit. Here a small decrease was visible after the removal of the alcohol. The free alcohol contents of the intermediate products after the removal processes are shown in Table 1. The surfactants 2-C10E5 and 4-C10E5 contained almost the same mole fractions of free alcohol but still significantly more than those of 1-C10E5 and DM-C10E5, where no free alcohol

was removed. Therefore, in the final step small quantities of the corresponding alcohols were added to samples 1-C10E5 and DMC10E5 to obtain the same level of free alcohol in all samples. Table 2 shows the weight fractions of alcohol ethoxylate and added alcohol used for the final samples. Together with the values given in Table 1, it was possible to calculate the overall mole fractions of free alcohol and ethoxylation degrees of the final samples. The results given in Table 2 clearly show that with respect to these quantities all samples are almost identical. To have surfactants with identical hydrophilic moieties, they should not only contain the same ethoxylation degrees and the same contents of free alcohol but also have the same EO chain length distribution. This is difficult to analyze by SEC because of the low separation efficiency for the higher ethoxylation degrees. Therefore, the samples were additionally analyzed by HPLC. The HPLC spectrum for the raw product of sample 2-C10E5 is shown in Figure 1b. This time the signals corresponding to the different ethoxylation degrees are well separated. The highest detectable ethoxylation degree is E9. From the SEC trace in Figure 1a it can be concluded that even higher ethoxylation degrees are present in the sample. For other samples the highest ethoxylation degree detected was E10 or E11, but again from comparison with SEC spectra the conclusion could be drawn that the high molecular weight side of these spectra was underestimated in the HPLC analysis. This finding is at least to some large extent related to the different detection methods in SEC and HPLC. For the HPLC analysis a UV detector was used. Because of the end group functionalization with phenyl isocyanate, each molecule was equipped with one UV-vis chromophore. This means that in the HPLC spectrum the signal intensity is proportional to the molar concentration of that compound. The RI detector used in SEC is a mass detector, and the signal intensity is proportional to the mass concentration. Therefore, the molecules with high ethoxylation degrees, which are present in small molar concentrations, are much more visible in the RI detector than in the UV detector. In the HPLC spectrum, the signal at 2.5 min corresponds to free alcohol. However, HPLC could not be used to quantify the free alcohol contents. This might be related to the functionalization of the HPLC samples with phenyl isocyanate. We assume that the different reactivities of the hydroxyl group in ethoxylated molecules on the one hand and hydroxyl groups in the free alcohols on the other hand caused different functionalization degrees, which prevented a quantitative data analysis. For differently ethoxylated molecules different reactivities with respect to the functionalization reaction are not expected. Therefore, the EO chain length distributions of the ethoxylated sample fractions were calculated from the HPLC spectra. The free alcohol contents were determined by SEC, where no sample functionalization was necessary. Figure 2 shows the EO chain length distribution plots for the raw products, which were obtained from the HPLC analysis. The free alcohol is ignored in this diagram. It is visible that the distributions are almost similar. The differences between the distribution curves at high ethoxylation degrees might be related

Table 2. Alcohol Ethoxylates with Different Hydrophobic Moieties: Final Products overall composition in final sample

a

sample

concn (wt %) of alcohol ethoxylate

concn (wt %) of added alcohol

concn (mol %) of free alcohol

ethoxylation degree

1-C10E5 DM-C10E5 2-C10E5 4-C10E5

94.0a 94.3a 100.0b 100.0b

6.0 5.7

18.4 19.4 19.1 18.9

4.73 4.77 4.81 4.68

Raw product. b Intermediate product after partial removal of free alcohol.

6530 Langmuir, Vol. 23, No. 12, 2007

Figure 2. EO chain length distribution plots of the ethoxylated fractions of raw products 1-C10E5, DM-C10E5, 2-C10E5, and 4-C10E5.

to the difficulties in the HPLC detection described before. This is supported by the SEC measurements. From overlaying spectra of different samples, it was found that the spectra were visually almost identical. In conclusion the results document that for all surfactant samples the EO chain length distributions are almost similar. In microemulsions, alcohol ethoxylate surfactants not only are located at the water-oil interface but are also visibly soluble in the oil phase. If the microemulsions are examined in a narrow temperature and composition range, the absolute surfactant concentration in oil can be regarded to be constant. The solubility in water is much smaller than the oil solubility. For linear monodisperse C10E4, for example, the solubility in aliphatic hydrocarbon oils is 1-2 orders of magnitude larger than in water.16,17 With increasing EO chain length the water solubility increases slowly; however, the fractions of the higher ethoxylated chains in the surfactants decrease strongly above EO6 so that water solubility was not considered in this work. Under these conditions, the interfacially active surfactant fraction can be calculated from the simple relation

γi ) γ - γo Here γ is the overall mass fraction of surfactant in the microemulsion, γi is the mass fraction of the interfacially active surfactant, and γo is the mass fraction of the oil-soluble surfactant. If one assumes the oil solubility to be constant, γo increases and γi decreases with increasing oil to overall surfactant ratio. The situation gets more complex if one considers that the oil solubility of alcohol ethoxylates decreases with increasing EO chain length. This first means that the EO chain length distributions of the overall surfactant, the interfacially active surfactant, and the oilsoluble surfactant are different. Another consequence is that the EO chain length distribution of the interfacially active surfactant cannot be regarded as being constant. With increasing oil to overall surfactant ratio the ethoxylation degree increases. This scenario is important because a useful way to characterize the phase behavior of water-oil-surfactant mixtures is to vary the surfactant concentration at a constant water to oil ratio. The composition of the interfacially active surfactant is not constant for a phase diagram obtained by this procedure. It depends on the composition of the mixture. To quantify this effect, the oil solubilities of the surfactants were determined. For this purpose, (16) Burauer, S.; Sachert, T.; Sottmann, T.; Strey, R. Phys. Chem. Chem. Phys. 1999, 1, 4299. (17) Rosen, M. J. Surfactants and Interfacial Phenomena; John Wiley & Sons: New York, 1989.

Frank et al.

Figure 3. HPLC spectrum of the decane-soluble fraction of 4-C10E5. Table 3. Oil-Soluble Surfactant Fractions sample

ethoxylation degree

concn (mol %) of free alcohol

concn in oil (mg/g of oil)

1-C10E5 DM-C10E5 2-C10E5 4-C10E5

2.21 2.30 2.27 2.34

24

52 45 55 55

30

mixtures were produced which contained equal volume fractions of water and deuterated decane as well as small quantities of surfactant. The surfactant quantities only allowed to partial solubilization of water and decane as a microemulsion. After phase separation excess decane was removed and investigated by 1H NMR. Table 3 shows the ethoxylation degrees of the oil-soluble surfactant fractions, calculated from the NMR results. For samples 1-C10E5 and 4-C10E5 it was possible to determine additionally the fractions of free alcohol. In the case of the other surfactants this was not possible because of the low quality of the NMR spectra, which resulted most likely from impurity accumulations in the excess decane. Additionally, 1H NMR was used to determine the absolute surfactant concentrations in the oil phases. These values are also listed in Table 3. The details for the NMR analysis are given in the Experimental Section. As expected the ethoxylation degrees of the oil-soluble surfactant fractions are significantly lower than the ethoxylation degrees of the original surfactant samples. Thus, free alcohol and lower ethoxylated chains are considerably enriched in the oil. The other important result from this analysis is that the surfactant compositions with respect to the ethoxylation degree and solubility in decane differ little. The discrepancy between the fractions of free alcohol in samples 1-C10E5 and 4-C10E5 should not be overinterpreted because of the larger error bars resulting from small signal intensities. In addition, the decane excess phases were examined by HPLC. Figure 3 shows the HPLC trace of the oil-soluble fraction of 4-C10E5. In agreement with the last mentioned results, the lower ethoxylation degrees are enriched in the oil whereas the higher ethoxylation degrees are visible only marginally. The signal up to E7 could be clearly identified. Because of additional signals resulting from impurities, it was unclear whether even higher ethoxylates were present to some very small extent, and therefore, this part of the spectrum was not considered for the further examinations. In the later investigations phase diagrams were recorded by varying γ at a water to decane ratio of exactly 1. Therefore, the γ dependence of the EO chain length distribution of the interfacially active surfactant was examined. Figure 4a exemplarily shows the EO chain length distributions of the overall surfactant (squares) and the oil-soluble surfactant (triangles) for 4-C10E5 at γ ) 20%. For a better comparison the results are

Nonionic Surfactants with Hydrocarbon Tails

Langmuir, Vol. 23, No. 12, 2007 6531

Figure 5. Schematic phase diagram for monodisperse alcohol ethoxylates at a water to oil ratio of 1.

Figure 4. (a) EO chain length distribution plots at γ ) 20% for the original surfactant 4-C10E5, the decane-soluble fraction of 4-C10E5, and the interfacially active fraction of 4-C10E5 in absolute concentrations. (b) EO chain length distribution plots for the original surfactant 4-C10E5 and the interfacially active fractions of 4-C10E5 at γ ) 30% and at γ ) 19% in relative concentrations.

given in this case in absolute concentrations. The stars represent the values calculated for the interfacially active surfactant at the same γ value. The distribution curves for the overall surfactant and for the interfacially active surfactant are similar at high ethoxylation degrees; only for the lowest EO chain lengths and the free alcohol do significant differences exist. In Figure 4b chain length distributions are given for the overall surfactant 4-C10E5 (theoretically γ ) 100%) and the interfacially active surfactant at γ ) 30% and γ ) 19% in terms of relative concentrations. This γ range is representative of the later reported investigations. Because of the better oil solubility of the lesser ethoxylated chains, the distribution curves are shifted slightly to higher ethoxylation degrees with decreasing γ. However, it can be stated that in the γ range under investigation the EO chain length distribution changes only marginally. The average ethoxylation degree of the interfacially active surfactant increases from γ ) 30% to γ ) 19% by about 0.2 EO unit. This scenario is the same for the other surfactants. Therefore, it can be assumed that with respect to the EO chain length distribution the surfactants only differ very little. Phase Behavior. In this part of the work phase diagrams were recorded for all surfactants. To determine the influence of the structure of the hydrophobic group on the interfacial properties, it is essential that all surfactants contain similar hydrophilic groups. This applies not only to the average ethoxylation degree but also to the EO chain length distribution. Using mixtures of monodisperse surfactants with different ethoxylation degrees or polydisperse surfactants, this topic was investigated especially at the air-water interface.18-20 From these investigations it can (18) Staples, E. J.; Thompson, L.; Tucker, I.; Penfold, J. Langmuir 1994, 10, 4136. (19) Penfold, J.; Staples, E.; Tucker, I.; Thomas, R. K. J. Colloid Interface Sci. 1998, 201, 223. (20) Folmer, B. M.; Holmberg, K. Colloids Surf., A 2001, 180, 187.

Figure 6. Phase behavior of water-decane-1-C10E5 and waterdecane-monodisperse C10E5 at a water to decane volume ratio of 1.

be concluded that only the EO chain length distribution at similar average ethoxylation degree significantly influences the interfacial properties. However, the surfactant analysis described in the last section showed that the surfactants under investigation in this work are similar within experimental error with respect to the properties discussed above. Therefore, they only differ in the structure of the hydrocarbon tails. The phase diagram of a monodisperse alcohol ethoxylate at water to oil volume ratios of about 1 has the shape of a fish if temperature is plotted against γ. This is shown schematically in Figure 5. At higher γ the fishtail represents the one-phase region (1). The fishtail point is defined by γ˜ , the lowest surfactant concentration where a one-phase microemulsion is formed, and T˜ , the corresponding temperature. The fish body at surfactant concentrations lower than γ˜ represents a three-phase system (3), where the microemulsion coexists with excess water and excess oil. The two-phase regions below the fish (2) and above (2h) contain microemulsions coexisting with excess oil and excess water, respectively. The shape of the fish is symmetrical with respect to T˜ . A more detailed description of the phase behavior of water-oil-alcohol ethoxylate systems is given, for example, in ref 2. In this work we uniquely focus on the one-phase region. First, the comparison of 1-C10E5 with its monodisperse analogue is shown. Thereafter, results are presented which were obtained with the differently branched surfactants. Figure 6 shows the one-phase regions of 1-C10E5 and of monodisperse C10E5 with a water to oil volume ratio of 1 and decane as the oil. Both surfactants are equipped with the same hydrophobe, a linear C10 chain; however, the hydrophilic parts are different. The hydrophilic part contains only E5 units in the case of monodisperse C10E5. For 1- C10E5 it is polydisperse and equipped with 4.73 EO units on average. While the fishtail point of the 1-C10E5 system is located at γ˜ ) 15.5%, it is shifted to γ˜ ) 20.5% in the case of monodisperse C10E5 (see Table 4). This finding is in contrast to another report, where technical grade alcohol ethoxylates were compared with their monodisperse analogues.14

6532 Langmuir, Vol. 23, No. 12, 2007

Frank et al.

Table 4. Characterization of the Surfactants at the Fishtail Points

sample monodisperse C10E5 1-C10E5 DM-C10E5 2-C10E5 4-C10E5

ethoxylation degree of the interfacially active γ˜ γ˜ I T˜ surfactant (mass %) (mass %) (°C) 5.00 5.25 5.14 5.19 5.07

20.5 15.5 17.2 20.0 19.0

19.8 13.7 15.6 18.1 17.1

51.0 57.5 52.5 59.0 45.0

The lower efficiency of the technical grade surfactants in that report was explained by the high oil solubility of the lower ethoxylated surfactant fraction. This reduces γi and increases the ethoxylation degree of the interfacially active surfactant compared to the overall surfactant. Both effects should result in lower efficiency of the polydisperse surfactant. With the knowledge of the oil solubility of 1-C10E5, the ethoxylation degree of the interfacially active surfactant at the fishtail point was calculated to be 5.25. This value is visibly above the ethoxylation degree of monodisperse C10E5 and should consequently lead to a slightly lower efficiency of the polydisperse surfactant. γ˜ i for 1-C10E5 was calculated to be 13.7%. In the case of monodisperse C10E5 an oil solubility of 2.0 mass % was assumed.16 This leads to a value for γ˜ i of 19.8%. In other words, this means that on the basis of γ˜ i the difference in efficiency between 1-C10E5 and monodisperse C10E5 is even more prominent than on the basis of γ˜ . These results on the one hand confirm the arguments mentioned in ref 14 to explain the lower efficiency of the polydisperse surfactants, found in that work. On the other hand, in our study not the polydisperse but the monodisperse surfactant is considerably less efficient. One explanation for the different results might be the absence of a detailed analysis of the technical grade surfactants in ref 14. The composition of those surfactants was only roughly known. In any case this result highlights the importance of the EO chain length distribution on the surfactant properties. On the temperature scale the one-phase region of the microemulsion containing 1-C10E5 is found at higher temperatures than the system with monodisperse C10E5. At the fishtail points, for example, the difference is 6.5 °C (see Table 4). This can be explained by the different ethoxylation degrees. At the fishtail point the ethoxylation degree of 1-C10E5 is 0.25 EO unit higher compared to that of monodisperse C10E5. For alcohol ethoxylates the addition of one EO unit leads to a temperature increase of the one-phase region of about 20 °C if a water to oil volume ratio of 1 is considered.16 The addition of 0.25 EO unit should lead to an increase of 5 °C, which is in good agreement with the temperature difference seen in our case. Besides the different locations of the fishtail points, there are also differences with respect to the shape of the one-phase regions. Especially the upper phase transition boundary shows weaker temperature dependence for 1-C10E5 than for monodisperse C10E5. This scenario was seen before for other technical grade alcohol ethoxylates.14 It was explained by the increasing ethoxylation degree of the interfacially active surfactant fraction with decreasing γ. In the case of 1-C10E5 in the γ range between 30.5% and 15.5% the increase of the ethoxylation degree of the interfacially active surfactant was calculated to be 0.3 EO unit. This compensates for 6 °C of the expected temperature decrease in the γ range under investigation. Together with the measured temperature drop of 3 °C in that γ range, one would expect for a system without compositional change of the interfacially active surfactant a temperature reduction of 9 °C between 30.5% and 15.5%. For the same ∆γ, monodisperse C10E5 shows a temperature drop of about 8 °C, which is in good agreement with the expected

Figure 7. Phase behavior of water-decane-1-C10E5, waterdecane-DM-C10E5, water-decane-2-C10E5, and water--decane4-C10E5 at a water to decane volume ratio of 1.

value for 1-C10E5 at a constant surfactant composition. At higher γ there is a lamellar phase (LR) in coexistence with the bicontinuous microemulsion. With respect to the fishtail point, the location of that region is similar for both surfactants. Figure 7 shows again the phase diagram of 1-C10E5 together with the phase diagrams of the other polydisperse surfactants DM-C10E5, 2-C10E5, and 4-C10E5. All phase diagrams are qualitatively similar. To determine the influence of the hydrophobe’s structure on the phase behavior quantitatively, the hydrophilic moieties must be similar, not only for the surfactant originally used to prepare the microemulsions but also for the interfacially active surfactant fractions. From the surfactant analysis described before two conclusions can be drawn. First, for the original surfactants the average ethoxylation degrees as well as the EO chain length distributions are almost similar. Second, the same scenario is valid for the oil-soluble surfactant fractions. Taking into account the measured values from the surfactant analysis, the ethoxylation degrees presented in Table 4 were calculated for the interfacially active surfactant at the fishtail points. The highest ethoxylation degree was calculated for 1-C10E5. This originates from the highest amount of oilsoluble low ethoxylated surfactant, as a consequence of the lowest γ˜ . However, the ethoxylation degrees of the other polydisperse surfactants are only marginally lower, which again underlines that there is little difference between surfactants with respect to the hydrophilic moiety. From the values for γ˜ listed in Table 4 it is visible that there are significant differences with respect to the location of the one-phase region in the phase diagram. 1-C10E5 clearly is the most efficient surfactant. The small structural changes in the other polydisperse surfactants shift the one-phase regions to higher γ values. The shift is less expressed for DM-C10E5, where the linear C10 chain is replaced by the more bulky dimethyloctyl group. Interestingly, the small structural change from 1-C10E5,

Nonionic Surfactants with Hydrocarbon Tails

Langmuir, Vol. 23, No. 12, 2007 6533

the Hamiltonian of the system to three leading contributions, which describe a binary liquid with low surface tension and a domain-stabilizing term. This description leads to a domaindomain correlation function, which in real space is interpreted by an alternating sequence of domains (with spacing d) and a relatively low correlation length ξ. Overall, the macroscopic scattering cross section dΣ/dΩ|TS of the Teubner-Strey theory reads

dΣ dΩ

Figure 8. Typical macroscopic scattering cross section of a microemulsion as a function of the scattering vector Q for 4-C10E5 at γ ) 0.39 close to the lamellar phase. The Teubner-Strey fit is shown as a gray line, and the extended formula including a Beaucage term is shown as a black line.

|

TS )

8π〈ν2〉/ξ Q4 - 2(k02 - ξ-2)Q2 + (k02 + ξ-2)2

A detailed analysis including the SANS instrument resolution yields the two structural parameters d and ξ with a high reliability. The factor ν2 of the numerator determines the amplitude of the fluctuations times the scattering contrast. At slightly higher Q the real Porod region is observed compared to the estimation of the Teubner-Strey formula. The TeubnerStrey theory is based on an expansion of the free energy with long-range fluctuations, neglecting the always-present shortrange undulations of the membrane surrounding the domains. Thus, the real surface observed in the Porod region is increased. An empirical approach based on the description of fractal scattering by Beaucage23 allows for the extension of the model fitting over a large Q range. The original theory consists of a Guinier term describing the overall size at large wavelengths and an empirical term according to the fractal behavior. The latter term we use here to describe the additional domain surface with a concrete Porod Q-4 power law. The thus obtained formula reads

where the hydrophilic unit is fixed at the end of a linear C10 chain, to 2-C10E5, where it is located at the C-2 position, causes a strong shift of γ˜ . In the case of 4-C10E5 the γ˜ shift is less expressed although the structural change compared to that of 1-C10E5 is stronger. If the oil-soluble fractions of the surfactants are taken into consideration and γ˜ is replaced by γ˜ i, the scenario is approximately the same. The influence caused by the slightly different ethoxylation degrees can be neglected. For monodisperse C10 ethoxylates the increase of one EO unit reduces γ˜ by about 3%.16 Therefore, for 1-C10E5, a reduction of the ethoxylation degree by 0.1-0.2 EO unit, to have the same composition as the other surfactants, would even lower γ˜ by 0.3-0.6% and increase the differences in γ˜ i. Besides the influence of the hydrophobe’s structure on γ, it also causes considerable changes in the temperature behavior. 1-C10E5 and 2-C10E5 show the highest values for T˜ (see Table 4), whereas for 4-C10E5 and DM-C10E5 the one-phase region is shifted considerably to lower temperatures. Again these differences cannot be explained by the different ethoxylation degrees. Taking 1-C10E5 as a reference and assuming that the addition of one EO unit increases T˜ by 20 °C, the same EO chain length as in 1-C10E5 would increase T˜ slightly for the other surfactants. For 2-C10E5 the increase would be 1.2 °C, and for DM-C10E5 2.2 °C and for 4-C10E5 3.6 °C were calculated. Finally, the position of the lamellar phase with respect to the fishtail point changes little for the different surfactants. A detailed discussion of the phase behavior will be presented later together with the results from the SANS measurements. SANS Results. The method of analyzing SANS experiments on microemulsions has been discussed extensively in ref 21; thus, we only refer briefly to theoretical aspects here. A typical SANS pattern is shown in Figure 8. At low scattering vectors Q the intensity is rather high, indicating strong long-range fluctuations, which is typical for microemulsions. Then a correlation peak is observed. The peak position k0 ) 2π/d indicates the domain size d, and the peak width is proportional to the reciprocal correlation length ξ-1. These two features of the scattering pattern are well described by the Teubner-Strey (TS) formula.22 The origin of this theory is a Landau approach describing long-range fluctuations of the domains. The symmetry of the system due to the identical volumes of oil and water reduces

The radius of gyration Rg lies in the range of single domains and thus describes the additional fluctuations on the surface of the domains correctly. The overall Porod constant P ) 8π〈ν2〉/ξ + G/(1.5Rg4) describing the high-Q power law dΣ/dΩ ) PQ-4 depends on both the Teubner-Strey formula parameters and the amplitude G of the fractal Beaucage term. An additional overall Gaussian factor was introduced to describe the conformal roughness (σ being typically 2 Å). Ideally the ratio of the overall Porod constant P and the surfactant volume fraction should tell about the surfactant film thickness. Unfortunately, our data were so noisy that no reliable value was obtained in this way. The reason is the small Q range of the Porod region, limited by the correlation peak and the incoherent background bg. In other words, the domains are rather small, and the additional surface by strong fluctuations is rather important. Thus, the real surface is not well defined in the limited Q range. In the following we describe the intensity by the TS formula only. In Figure 9 the parameters d and ξ are summarized for the surfactant 1-C10E5. The two parameters are plotted as a function of the surfactant membrane volume fraction ψ. ψ is equivalent to γi except that γi represents not the volume but the mass fraction of surfactant being localized at the water-oil interface. With increasing surfactant amount both d and ξ are decreasing, since both scale with the surface to volume ratio S/V ≈ ψ. For the domain size d the expression of Roux24 is used to fit the full curve and to obtain a single parameter, the film thickness δ:

(21) Byelov, D.; Frielinghaus, H.; Holderer, O. O.; Allgaier, J.; Richter, D. Langmuir 2004, 20, 10433. (22) Teubner, M.; Strey, R. J. Chem. Phys. 1987, 87, 3195.

(23) Beaucage, G. J. Appl. Crystallogr. 1996, 29, 134. (24) Roux, D.; Nallet, F.; Freyssingeas, E.; Porte, G.; Bassereau, P.; Skouri, M.; Marignan, J. Europhys. Lett. 1992, 17, 575.

(

dΣ dΣ ) dΩ dΩ

|

TS

+

)

G erf12(1.06QRg/x6) 1.5Q4Rg4

exp(-σ2Q2) + bg

6534 Langmuir, Vol. 23, No. 12, 2007

Frank et al.

Figure 9. Structural parameters directly obtained from SANS experiments: the domain size d and the correlation length ξ as a function of the membrane volume fraction ψ. The line represents the fitted theory by Roux with only one parameter, the film thickness δ.

d)

(

( ( ) ))

κR 2δ 1 kBT 1+ ln c ψ 4π κR kBT

1/2

d 2a

The ψ dependence of the bending rigidity κR is used from considerations discussed below. The cutoff variable c ) 1.84 describing the high wavelength number cutoff is a constant. The length a is the square root of the area occupied per surfactant molecule in the membrane. For monodisperse C10E5 a was determined to be 7.8 Å.25,26 The line through the data points of ξ represents the fitted theory by Roux. It should be stressed that a single fitting parameter δ is enough to achieve such a good agreement between experiment and theory. κR is another number which can be obtained from the scattering experiments. Through the Gaussian random field theory27 the two structural parameters d and ξ are connected with the bending rigidity κR/kBT ) 0.85ξ/d. The result of these considerations is plotted in Figure 10. For all four surfactants one obtains a linear increase of the bending rigidity with increasing ψ (which is used in the Roux expression). This is due to the renormalization28 of the elastic constants with the membrane volume fraction or more precisely with the considered length scale of the correlation peak. A fluctuating membrane appears floppier on larger length scales than one would locally assume. Thus, the renormalized elastic constant reads

κR ) κ0 + (3/4π) ln ψ whereby κ0 is the intrinsic bending rigidity. Within the limited ψ range considered here this dependence looks linear. Comparison of the Phase Behavior and Scattering Results. It has been approved how to extract information about the elastic constants of the membrane from scattering experiments and phase diagram measurements.21 It is assumed that the elastic energy of the film dominates the free energy of a microemulsion. The emulsification failure boundary (i.e., the fishtail point) tells about changes of the intrinsic saddle splay modulus κj0 ) κjR + (25) Sottmann, T.; Strey, R.; Chen, S. H. J. Chem. Phys. 1997, 106, 6483. (26) Sottmann, T. Dissertation, Georg-August-Universita¨t zu Go¨ttingen, Go¨ttingen, Germany, 1997. (27) Pieruschka, P.; Safran, S. A. Europhys. Lett. 1993, 22, 625. Pieruschka, P.; Safran, S. A. Europhys. Lett. 1995, 31, 207. Endo, H.; Mihailescu, M.; Monkenbusch, M.; Allgaier, J.; Gompper, G.; Richter, D.; Jakobs, B.; Sottmann, T.; Strey, R.; Grillo, I. J. Chem. Phys. 2001, 115, 580. (28) Peliti, L.; Leibler, S. Phys. ReV. Lett. 1985, 54, 1690. David, F. In Statistical Mechanics of Membranes and Surfaces; Nelson, D., Piran, T., Weinberg, S., Eds.; World Scientific: Singapore, 1989; p 157.

Figure 10. Renormalized bending rigidity as a function of the membrane volume content for the four different surfactants. The straight lines are just empirical fits used in the Roux expression. The reason for this dependence is the actually logarithmic renormalization of the elastic moduli κ and κj. Table 5. Elastic Moduli KR and K j 0, Film Thickness δ, and Changes of the Intrinsic Spontaneous Curvature ∆c0 for All Surfactants sample monodisperse C10E5 1-C10E5 DM-C10E5 2-C10E5 4-C10E5

κR/kBT at ψ ) 0.26

κj0/kBT

δ (Å)

∆c0 (Å-1)

0.581 ( 0.001 -0.43 ( 0.01 13.3 ( 0.1 -0.0107 ( 4 × 10-4 0.590 ( 0.001 0.572 ( 0.001 0.535 ( 0.001 0.488 ( 0.001

-0.53 ( 0.01 -0.50 ( 0.01 -0.46 ( 0.01 -0.47 ( 0.01

14.2 ( 0.1 0.0000 (set) 13.4 ( 0.1 -0.0071 ( 4 × 10-4 14.2 ( 0.1 0.0021 ( 4 × 10-4 15.2 ( 0.1 -0.0178 ( 4 × 10-4

[(10/3)/4π] ln ψ, which is obtained by the assumption that the renormalized saddle splay modulus κjR is zero at the fishtail point. The so-obtained bending moduli are summarized in Table 5. Focusing on the self-synthesized surfactants first, one observes decreasing absolute values of κR in the order 1-C10E5, DMC10E5, 2-C10E5, and 4-C10E5. κj0 runs parallel with κR, except that the values for 2-C10E5 and 4-C10E5 are approximately similar for κj0 if the error bars are considered. This scenario is expected as a floppier membrane has more fluctuations or a higher surface to volume ratio, i.e., a lower efficiency. Monodisperse C10E5 exhibits the smallest value for κj0 due to the lowest efficiency. In contrast its value for κR is close to the one measured for 1-C10E5. Here the SANS and the phase behavior measurements clearly are contradictory. The other parameter extracted from the scattering experiments is the film thickness δ. It is obtained from the precise description of the domain size d as shown in Figure 9. The results are listed in Table 5. δ of 1-C10E5 is 0.7 Å larger than for DM-C10E5. Because of the branched tail, DM-C10E5 is two carbon atoms shorter, which translates into a length difference of 2.5 Å if an all-trans conformation is considered. However, alcohol ethoxylates are tilted within the membrane. Linear C10E5 has a molecule length of roughly 35 Å, while the film thickness is 14 Å. This fact explains why the film thickness of DM-C10E5 is smaller by only a fraction of the length difference. The two other surfactants, 2-C10E5 and 4-C10E5, have the branching points directly at the hydrophilic/hydrophobic junction. As the structural changes from 1-C10E5 to 2-C10E5 are minimal, it is understandable that the δ values are similar. For 4-C10E5 one would intuitively expect a considerably smaller δ value due to the fact that the hydrophobic chain is split into two short units. Interestingly, our results reveal that the membrane thickness for 4-C10E5 is more than 1 Å larger than for the fully linear structure. This could be explained under the assumption that 4-C10E5 is less tilted because of the different

Nonionic Surfactants with Hydrocarbon Tails

geometry of the hydrocarbon tail. The monodisperse C10E5 forms a thinner film than 1-C10E5, and what is even more astonishing is the film thickness is the same as that of DM-C10E5. However, this result is supported by another study, where monodisperse and polydisperse alcohol ethoxylates were examined at the airwater interface.20 In that investigation it was found that the area per head group is larger for the monodisperse homologue. The change of the intrinsic spontaneous curvature ∆c0 was obtained from the relative temperature changes of the fishtail points. For nonionic surfactants the intrinsic spontaneous curvature is simply linearly dependent on temperature changes according to ∆c0 ≈ µH∆T. The coefficient µH was estimated from experiments on C10E4.23,26 For the ∆c0 values shown in Table 5 1-C10E5 was chosen as a reference point. Therefore, ∆T is equivalent to the T˜ difference between 1-C10E5 and the other surfactants. ∆c0 is visibly influenced by the structure of the hydrophobic tail. It is lower the more bulky the hydrophobic surfactant tails are because a bulkier hydrophobic part bends the surfactant membrane more toward water. 2-C10E5 seems to behave exceptionally. Its slightly positive spontaneous curvature compared to 1-C10E5 is not in agreement with the increased bulkiness of the hydrophobic unit. The situation is different if one assumes that the methyl group next to the EO chain is directed toward the water phase. In this scenario the effective length of the hydrocarbon group comprises only nine C atoms. From the examination of monodisperse alcohol ethoxylates it is known that the reduction of the hydrocarbon chain length by one C atom increases T˜ by 5-10 °C.16 Therefore, the little increase in bulkiness easily can be overcompensated by the shorter effective chain length. This interpretation is supported by another study, where hexadecanol isomers were investigated at the air-water interface.30 In that work the small increase of interfacial area per molecule from 1-hexadecanol to 2-hexadecanol was interpreted with the same conformational model as described here. A reduced effective length of the hydrocarbon tail of 2-C10E5 would in addition explain the low efficiency of that surfactant. Compared to 1-C10E5 the structural changes are minimal, which should lead to almost similar phase behavior. The huge efficiency difference (29) Gompper, G.; Richter, D.; Strey, R. J. Phys.: Condens. Matter 2001, 13, 9055. (30) Can, S. Z.; Mago, D. D.; Walker, R. A. Langmuir 2006, 22, 8043. (31) Jang, S. S.; Goddard, W. A. J. Phys Chem B 2006, 110, 7992. Bresme, F.; Faraudo, J. Langmuir 2004, 20, 5127.

Langmuir, Vol. 23, No. 12, 2007 6535

can be explained if one assumes that the effective chain length of 2-C10E5 is nine C atoms. According to ref 16, the reduction of the hydrocarbon chain length by one C atom increases γ˜ by 5%-8%. This is in reasonable agreement with the values found for 1-C10E5 and 2-C10E5 (see Table 4).

Conclusions Our results clearly show that the fully linear alcohol ethoxylate 1-C10E5 is the surfactant having the highest emulsification capacity. This structure allows the best parallel alignment of the surfactant molecules and leads to the most rigid interfacial film, which is confirmed by our SANS study. DM-C10E5 and 4-C10E5 contain considerably more bulky hydrophobic tails and therefore are less perfectly aligned at the interface, which in turn leads to less rigid films and lower efficiencies. The increased bulkiness of the DM-C10E5 and 4-C10E5 hydrophobes in addition causes the surfactant films to be more curved toward the water domains, leading to significantly smaller T˜ values compared to that of 1-C10E5. 2-C10E5 does not fit into this scheme. Its behavior can be explained under the assumption that the methyl group in the C-1 position is oriented toward the water domains. This would reduce the effective hydrocarbon chain length to C9 and explain the very low efficiency as well as the exceptionally high T˜ . In addition, our study reveals that the phase behaviors of 1-C10E5 and monodisperse C10E5 are strongly different. The lower T˜ of monodisperse C10E5 can be explained quantitatively knowing the varying composition of the interfacially active fraction of 1-C10E5 as a result of the higher oil solubility of the lower ethoxylated molecules. However, the drastically reduced efficiency of monodisperse C10E5 is in disagreement with the depletion of lower ethoxylated surfactant molecules at the interface. It is also in disagreement with SANS results indicating almost the same bending rigidity for the monodisperse and the polydisperse surfactants. At the present we can only draw a rough picture with our experimental results, but we hope that this study encourages computer simulations (such as in ref 31) on branched CiEj surfactants and will complete or even exceed the interpretation of our experiments. Acknowledgment. We thank S. Willbold for the NMR measurements. C.F. and J.A. thank the Deutsche Bundesstiftung Umwelt for financial support. LA0637115