13376 DOI: 10.1021/la903260r Langmuir 2009, 25(23), 13376–13383Published on Web 10/23/2009

pubs.acs.org/Langmuir

© 2009 American Chemical Society

Shear-Induced Phase Separation in Polyelectrolyte/Mixed MicelleCoacervates

Matthew W. Liberatore,* Nicholas B. Wyatt, and MiKayla Henry

Department of Chemical Engineering, Colorado School of Mines, Golden, Colorado 80004

Paul L. Dubin and Elaine Foun

Department of Chemistry, University of Massachusetts, Amherst, Massachusetts 01003

Received June 1, 2009. Revised Manuscript Received September 25, 2009

A quantitative study of the shear-induced phase separation of a polycation/anionic-nonionic micelle coacervate ispresented. Simultaneous rheology and small-angle light scattering (SALS) measurements allow the elucidation ofmicrometer-scale phase separation under flow in three coacervate solutions. Below 18 �C, all three of the coacervatesolutions are optically clear Newtonian fluids across the entire shear rate range investigated. Once a critical temperaturerange and/or shear rate is achieved, phase separation is observed in the small-angle light scattering images and the fluidexhibits shear thinning. Two definitive SALS patterns demonstrate the appearance of circular droplets at low shear ratesnear the critical temperature and ellipsoidal droplets at higher temperatures and shear rates. The shear-induced dropletsrange in size from ∼1 to 4 μm. The ellipsoidal droplets have aspect ratios as high as 4. A conceptual picture in whichshear flow extends the polyelectrolyte chains of the clear coacervate liquid phase is proposed. The extended chains createinterpolyelectrolyte-micelle interactions and promote expulsion of small ions from the complex, resulting in theformation of micrometer-scale phase-separated droplets.

Introduction

When two oppositely charged macromolecules [such as poly-electrolytes (PEs)] are mixed, spontaneous liquid-liquid phaseseparation can occur with the formation of a densemacroion-richphase in equilibrium with a dilute macroion-poor phase. Thisprocess, called complex coacervation, converts a metastablesuspension of coacervate droplets to separate coacervate anddilute supernatant phases upon standing or centrifugation.1

Coacervates composedof polyelectrolytes andoppositely chargedcolloidal particles can be found in shampoos where the nanopar-ticles are micelles or in food formulations where the nanoparticlesare proteins. Such coacervates form a locally segregated environ-ment that can allow selective absorption of apolarmolecules fromthe surrounding medium, e.g., extracting aliphatic compoundsfrom solutions.2,3 Coacervates can also be used to deliver drugs,nutraceuticals, or topically active ingredients.4-6 Complex coa-cervation can be viewed as an entropically favorable ion-exchangeprocess, which occurs when the oppositely charged macroionsrelease some of their initially bound counterions upon forming acomplex.When the concomitant loss of hydration is smaller thanthat leading to precipitation, liquid-liquid phase separation willoccur. Since soluble complexes, or aggregates thereof, are theprecursors of coacervation, their mutual repulsion must be

weakened for this transition to take place. Liquid-liquid phaseseparation occurs only when the charge fraction f (the ratio ofthe number of macroion charges of a given sign to the totalmacroion charge) is close to 0.5. It is important to identify f asthe charge ratio in the complex (microstoichiometry) anddistinguish it from that of the entire system, fbulk (bulk ormacrostoichiometry, also known as the mixing ratio) and,further, to recognize that f itself could present some polydis-persity. Phase separation in the region of bulk charge stoichi-ometry has been reported formixtures of cationic chitosan withsynthetic polyanions,7 histones with DNA or with poly(styrenesulfonate),8 and lysozyme with poly(styrene sulfonate).9 Coa-cervation can take place in the vicinity of an fbulk of 0.5 becauseof system polydispersity or because of disproportionation orpolarization.10 However, in these cases, the residual charge ofthe coacervating species will result in structure formation atsubmicrometer length scales, the morphologies of which arecurrently being determined.9,11-16

*To whom correspondence should be addressed. E-mail: mliberat@mines.edu.(1) Stuart, M. A. C. Colloid Polym. Sci. 2008, 286(8-9), 855–864.(2) Sudbeck, E. A.; Dubin, P. L.; Curran, M. E.; Skelton, J. J. Colloid Interface

Sci. 1991, 142(2), 512–517.(3) Luque, N.; Rubio, S.; Perez-Bendito, D. Anal. Chim. Acta 2007, 584(1),

181–188.(4) Thimma, R. T.; Tammishetti, S. J. Microencapsulation 2003, 20(2), 203–210.(5) Zhang, L.; Liu, Y. Z.; Wu, Z. C.; Chen, H. X. Drug Dev. Ind. Pharm. 2009

35(3), 369–378.(6) Gander, B.; Blanco-Prieto, M. J.; Thomasin, C.; Wandrey, C.; Hunkeler, D.

Encycl. Pharm. Technol. 2006, 1–5.

(7) Mincheva, R.; Manolova, N.; Paneva, D.; Rashkov, I. Eur. Polym. J. 2006,42(4), 858–868.

(8) Raspaud, E.; Chaperon, I.; Leforestier, A.; Livolant, F. Biophys. J. 1999, 77(3), 1547–1555.

(9) Gummel, J.; Boue, F.; Clemens, D.; Cousin, F. SoftMatter 2008, 4(8), 1653–1664.

(10) Zhang, R.; Shklovskii, B. T. Physica A 2005, 352(1), 216–238.(11) Wang, X. Y.; Lee, J. Y.; Wang, Y. W.; Huang, Q. R. Biomacromolecules

2007, 8(3), 992–997.(12) Chodankar, S.; Aswal, V. K.; Kohlbrecher, J.; Vavrin, R.; Wagh, A. G.

Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2008, 78, 3.(13) Singh, S. S.; Aswal, V. K.; Bohidar, H. B. Int. J. Biol. Macromol. 2007, 41

(3), 301–307.(14) Kayitmazer, A. B.; Strand, S. P.; Tribet, C.; Jaeger, W.; Dubin, P. L.

Biomacromolecules 2007, 8(11), 3568–3577.(15) Kayitmazer, A. B.; Bohidar, H. B.; Mattison, K. W.; Bose, A.; Sarkar, J.;

Hashidzume, A.; Russo, P. S.; Jaeger, W.; Dubin, P. L. Soft Matter 2007, 3(8),1064–1076.

(16) Menjoge, A. R.; Kayitmazer, A. B.; Dubin, P. L.; Jaeger, W.; Vasenkov, S.J. Phys. Chem. B 2008, 112(16), 4961–4966.

DOI: 10.1021/la903260r 13377Langmuir 2009, 25(23), 13376–13383

Liberatore et al. Article

Coacervates formed from polyelectrolytes and proteins ofopposite charge have been studied by static and dynamic lightscattering (DLS),17 small-angle neutron scattering,12-14 fluores-cence recovery after photobleaching,15 rheology,17,18 total inter-nal reflectance microscopy, cryo-TEM,15 and pulsed-fieldgradient NMR.16 The results of these studies show that theseoptically clear, viscous fluids have complex internal structuresthat are heterogeneous on many length scales. While biologicaland biotechnological objectives motivate studies with proteins,their idiosyncratic charge anisotropies complicate elucidationof electrostatic effects. In this regard, PE/micelle coacervatesrepresent a simplification, even though micelle lability is aconsideration. In particular, mixtures of polyelectrolytes withionic-nonionic mixed micelles are good model systems becausethe micelle surface charge density can be modulated by the molefraction of ionic surfactant, especially when the surfactant con-centrations are much higher than the mixed surfactant criticalmicelle concentration.Dubin et al. performed extensive studies ona system comprised of poly(dimethyldiallylammonium chloride)(PDADMAC), together with mixed micelles of the anionicsurfactant sodium dodecyl sulfate (SDS) and the nonionic sur-factant Triton X-100 (TX100), using narrow molecular weightdistribution samples of the nonhydrophobic polycation. Theauthors showed that the micelle surface charge density (σ) andsurface potential (j) varied directly with the mole fraction of theanionic surfactant (“Y”).19 Consequently, gradual addition ofSDS to a mixture of polycation and nonionic micelles resulted inprogressive changes in σ and j, leading to transitions fromnoninteracting solutions to soluble complexes at “Yc”, and thento liquid-liquid phase separation (coacervation) at “Yφ”.

While Yφ appears to be a true liquid-liquid phase transition,becoming infinitely sharp when system polydispersity is elimi-nated,20 Yc may be more appropriately described as a second-order phase transition.21 These transitions can be identified bydynamic light scattering, electrophoretic mobility, or preciseturbidimetry. Turbidimetry was used to obtain phase behavioras a function of micelle surface charge density, PE molecularweight, PE:surfactant stoichiometry, and ionic strength. Thesemeasurements led to maps that identify the conditions corre-sponding to soluble complex formation, coacervation, and pre-cipitation.22 In contrast to PE/protein systems, the PE/micellesolutions undergo temperature-induced coacervation, which alsoappears to be a true liquid-liquid phase separation when systempolydispersity is minimized.23 The critical temperature of theliquid-liquid phase transition (Tφ) decreases with an increasingPE molecular weight, and Tφ is highly sensitive to the other keyvariables, including ionic strength, micelle surface charge density,and PE:surfactant stoichiometry.23 The effects of the micellesurface charge density and PE:surfactant stoichiometry areclosely correlated with conditions at which the effective

(electrophoretic) charge of soluble complexes is close to zero,20

which has also been observed for protein/PE systems.24,25

Low-speed centrifugation of the metastable droplet suspen-sions formed when T> Tφ yields optically clear, viscous fluids(typically 5-10 wt % surfactant and 1-2 wt % PE), whichdisplay some interesting phenomena. DLS shows abundantspecies with a diffusivity that is only 7 times smaller than that ofmicelles in dilute solution (despite a coacervate viscosity that is100 times larger) and are therefore described as “free mi-celles”.26 The equilibrium nature of the dense “coacervate”fluids thus formed is well-established bymultiple studies, whichdemonstrate that their properties are reversible and indepen-dent of both the time and route of preparation, as long asirreversible liquid-solid separation is avoided.27 When thiscoacervate is heated, a second phase separation temperature(denoted as Tφ

0) is observed turbidimetrically as shown inFigure 1 from ref 26. As pointed out in ref 26, the correspond-ing transition is essentially reversible, only showing a weakhysteresis. Of particular interest is the appearance of a secondphase with shear or elongational flows at temperatures slightlybelow Tφ

0.23 This flow-induced phase separation was observedwhen samples were loaded into confined geometries as shownin Figure 2 of ref 26.

A substantial body of literature describes shear-induced phaseseparation (SIPS)28 for wormlike micelles29-34 and selected high-molecular weight (MW) polymers,28,35,36 but this is the firstobservation of SIPS for a polymer-micelle complex. Sinceneither the micelle nor the PE alone exhibits such behavior, theimplication is that complexation with PE can cause ellipsoidalmicelles to behave like other macromolecular systems exhibitingSIPS. One interesting consequence of coacervation is the possi-bility of achieving very high viscosities at relatively low surfactantconcentrations, replacing surfactants with lower concentrationsof less expensive polymers.

Figure 1. Turbidity as a function of temperatures for a coacervatesolution. The transition from clear to turbid allows the determina-tion of critical temperature Tj

0. Reproduced from ref 26. Copy-right 2008. American Chemical Society.

(17) Bohidar, H.; Dubin, P. L.; Majhi, P. R.; Tribet, C.; Jaeger, W. Biomacro-molecules 2005, 6(3), 1573–1585.(18) Mohanty, B.; Bohidar, H. B. Int. J. Biol. Macromol. 2005, 36(1-2), 39–46.(19) Dubin, P. L.; The, S. S.; McQuigg, D. W.; Chew, C. H.; Gan, L. M.

Langmuir 1989, 5(1), 89–95.(20) Wang, Y. L.; Kimura, K.; Huang, Q. R.; Dubin, P. L.; Jaeger, W.

Macromolecules 1999, 32(21), 7128–7134.(21) McQuigg, D. W.; Kaplan, J. I.; Dubin, P. L. J. Phys. Chem. 1992, 96(4),

1973–1978.(22) Wang, Y. L.; Kimura, K.; Dubin, P. L.; Jaeger, W. Macromolecules 2000,

33(9), 3324–3331.(23) Kumar, A.; Dubin, P. L.; Hernon,M. J.; Li, Y. J.; Jaeger,W. J. Phys. Chem.

B 2007, 111(29), 8468–8476.(24) Xia, J. L.; Dubin, P. L.; Kim, Y.; Muhoberac, B. B.; Klimkowski, V. J.

J. Phys. Chem. 1993, 97(17), 4528–4534.(25) Gupta, A.; Reena; Bohidar, H. B. J. Chem. Phys. 2006, 125, 5.

(26) Dubin, P. L.; Li, Y. J.; Jaeger, W. Langmuir 2008, 24(9), 4544–4549.(27) Kaibara, K.; Okazaki, T.; Bohidar, H. B.; Dubin, P. L. Biomacromolecules

2000, 1(1), 100–107.(28) Larson, R. G. Rheol. Acta 1992, 31(6), 497–520.(29) Fischer, P.; Wheeler, E. K.; Fuller, G. G.Rheol. Acta 2002, 41(1-2), 35–44.(30) Oda, R.; Panizza, P.; Schmutz, M.; Lequeux, F. Langmuir 1997, 13(24),

6407–6412.(31) Narayanan, J.; Manohar, C.; Kern, F.; Lequeux, F.; Candau, S. J.

Langmuir 1997, 13(20), 5235–5243.(32) Liu, C. H.; Pine, D. J. Phys. Rev. Lett. 1996, 77(10), 2121–2124.(33) Rehage, H.; Hoffmann, H.; Wunderlich, I. Phys. Chem. Chem. Phys. 1986,

90(11), 1071–1075.(34) Hu, Y. T.; Boltenhagen, P.; Pine, D. J. J. Rheol. 1998, 42(5), 1185–1208.(35) Migler, K.; Liu, C. H.; Pine, D. J. Macromolecules 1996, 29(5), 1422–1432.(36) Rangelnafaile, C.; Metzner, A. B.; Wissbrun, K. F. Macromolecules 1984,

17(6), 1187–1195.

13378 DOI: 10.1021/la903260r Langmuir 2009, 25(23), 13376–13383

Article Liberatore et al.

In this work, we characterize the thermal and shear-inducedphase transitions in this polyelectrolyte/micelle system by small-angle light scattering (SALS) to determine the onset of phaseseparation as a function of the macromolecular solute concentra-tion, shear rate, and temperature. While previous SANS andcryo-TEM indicated the formation of dense phases (i.e., micelle-rich phases) at length scales of several hundred nanometers underquiescent conditions, this work probes the formation of the densephase under shear at larger length scales. In addition, the size andshape of the phase-separated droplets are measured and a con-ceptual picture of the transient structures is proposed. This workis representative of ongoing efforts to provide a molecular basisfor understanding the macroscopic properties of intermacroioniccoacervates.

Experimental Methods

Materials. Poly(diallyldimethylammonium chloride) (PDA-DMAC) was prepared by free radical aqueous polymerization ofdiallylmethylammonium chloride.37 The weight- and number-average molecular masses of the purified lyophilized polymerwere 2.19�105 and 1.41�105 Da, determined by light scatteringandmembraneosmometry, respectively.37TritonX-100 (TX100),a nonionic surfactant, was purchased from Fluka, and sodiumdodecyl sulfate (SDS), an anionic surfactant, with a purityof >99% and NaCl were purchased from Fisher. All wereused without further purification. Milli-Q water was used in allsamples.

Coacervate Preparation. PDADMAC/TX100 solutions(containing either 3 g/L PDADMAC with 20 mM TX100 or2 g/L PDADMAC with 10 mM TX100) and 60 mM SDSsolutions were prepared separately in 0.4 M NaCl. The sampleswere brought to coacervation via addition of SDS to the mixedPDADMAC/TX100 solution to produce the desired mole frac-tion of SDS defined as Y=[SDS]/([SDS] + [TX100]). Coacer-vates can also be prepared by the addition, at fixed Y and ionicstrength, of mixed micelles to PDADMAC, or PDADMAC tomicelles,38 although systematic comparisons of the coacervates soformed are incomplete. The equilibrium nature of this system isdemonstrated by the reversibility under conditions correspondingto soluble complexation or coacervation, although precipitation(e.g., induced by addition of SDS to PDADMAC in the absence

of nonionic surfactant) is effectively irreversible. As notedabove, coacervation could also be attained by increasing thetemperature above a critical value Tj which depends onY, polymer concentration (CP) andmolecular weight, and ionicstrength.23 The value of Tj for the formation of coacervatefrom a homogeneous polycation/mixed micelle solution istypically 5-10 �C below the value of Tj0, the temperature atwhich the coacervate itself undergoes additional phase transi-tion (see Figure 1). Thus, values of Y and CP were chosento produce different values of Tj and hence Tj

0. The turbidsample was then centrifuged for 1 h at 3500 rpm to produce anoptically clear dilute (upper) phase and a dense (lower) phase(“coacervate”). Preparation conditions and values of Tj andTj

0 are given in Table 1. The progressive increase in Tj0 fromsample A to sample B to sample C indicates that these coacer-vate samples (at, e.g., 20 �C) correspond to further progresstoward coacervate phase separation, which is confirmed byresults reported below. We note that values of Tj

0 (determinedby turbidimetry) in Table 1 are typically ∼2 �C higher than theonsets of phase separation as determined by SALS as discussedbelow. The differences in the values of Tj

0 are probably a resultof the heterogeneity of the coacervate, arising from the hetero-geneity of TX100 and the heterogeneity among the solublecomplexes from which coacervate forms. Because of the het-erogeneity of several components of these samples, relativecontributions of components may differ for turbidity versussmall-angle scattering. The effect of system heterogeneity,arising to a large extent from the chemical heterogeneity ofTX100 (vis-�a-vis, for example, C12E8

23), is to broaden allobservable transitions, making it difficult to establish theirorder.

Rheology and Small-Angle Light Scattering. Rheologyand SALS data were collected simultaneously using anAR-G2 rheometer (TA Instruments, New Castle, DE) withthe commercially available SALS attachment. A transparentparallel plate configuration (50 mm diameter) with a 1 mmgap was used for all tests. SALS images were recorded forat least four different shear rates (from 0.1 to 32 s-1) alongthe flow curves for each of the three different samples (A, B,and C) over a range of temperatures. All images were takenat steady state; i.e., the viscosity and SALS image werenot changing with time at a given shear rate. The sizescale of scattering objects captured by this SALS setup is0.94-5.0 μm (q = 1.3-6.7 μm-1). To examine the shear-induced phase separations, the SALS images were analyzedusing ImageJ with standard protocols for subtracting thebackground and removing the beam stop from the rawimages.39,40

The locations of the phase boundaries under shear weredetermined by the normalizedmean intensity of the SALS images.The normalizedmean intensity is a representation of the turbidityof the solutions from the perspective of small-angle scattering.Similar analysis has been used to determine critical shear rates

Figure 2. Photo of coacervate exhibiting flow-induced phaseseparation in a confined sample geometry. Reproduced fromref 26. Copyright 2008. American Chemical Society.

Table 1. Preparation Conditions and Properties of the CoacervateSolutionsa

sample Yb CPb (g/L) f-

c MWnb (kDa) Tj (�C)d Tj0 (�C)d

A 0.37 2 0.49 141 12 22B 0.35 3 0.35 141 19 24C 0.44 3 0.46 141 22 29( 2

aAll coacervates prepared in 0.40 M NaCl. b Solution from whichcoacervate is obtained. Y is the mole fraction of anionic surfactant,and CP and MW are the polymer concentration and number-averagemolecular weight, respectively. cThe number of SDS charges divided bythe sumof SDS and polycation charges. dProperty of coacervate (see thetext). Tj and Tj

0 values are all (1 �C, except as shown.

(37) Dautzenberg, H.; Gornitz, E.; Jaeger,W.Macromol. Chem. Phys. 1998, 199(8), 1561–1571.(38) Dubin, P. L.; Rigsbee, D. R.; Gan, L. M.; Fallon, M. A. Macromolecules

1988, 21(8), 2555–2559.

(39) Kline, S. R. J. Appl. Crystallogr. 2006, 39, 895–900.(40) TA Instruments.ARSeries Small Angle Light-Scattering (SALS)Accessory

Manual; TA Instruments: Newark, DE, 2008.

DOI: 10.1021/la903260r 13379Langmuir 2009, 25(23), 13376–13383

Liberatore et al. Article

fromSALS images.41 The normalizedmean intensity is defined asthe ratio of the average intensity of the image (after removal ofthe beam stop) to the intensity if all pixels in the image weresaturated (i.e., a value of 255 for the 8-bit camera used). Apositive value of the normalized mean intensity indicates thepresence of micrometer size structures in the flow; however,

the magnitude of the normalized mean intensity does notcorrelate with the size of the micrometer-scale scatteringobjects.

Following the method employed by Walker et al.,42 thecorrelation length (ac) and aspect ratio (ar) of the droplets weredetermined using Debye-Bueche43 plots (I-0.5 vs q2). A linearfit to the radially averaged data plotted in the Debye-Buecheformat was calculated in each case using the method ofleast squares (example plot included as Figure S1 of theSupporting Information). The slope and intercept of thelinear fit were used to calculate a characteristic length [ac=(slope/intercept)0.5] for the droplets at various shear rates.

Figure 3. Viscosity as a function of shear rate at several tempera-tures for three coacervate solutions: (a) sampleA, (b) sampleB, and(c) sample C.

Figure 4. Viscosity (either Newtonian μ or ηo from the Crossmodel fit) as a function of temperature for three coacervatesolutions: (a) sample A, (b) sample B, and (c) sample C.

(41) Saito, S.; Hashimoto, T.; Morfin, I.; Lindner, P.; Boue, F.Macromolecules2002, 35(2), 445–459.

(42) Walker, L. M.; Kernick, W. A.; Wagner, N. J.Macromolecules 1997, 30(3),508–514.

(43) Debye, P.; Bueche, A. M. J. Appl. Phys. 1949, 20, 518–525.

13380 DOI: 10.1021/la903260r Langmuir 2009, 25(23), 13376–13383

Article Liberatore et al.

A calibration of the Debye-Bueche characteristic length wasperformed using a Microbead NIST Traceable Particle SizeStandard (Polysciences, Inc., Warrington, PA) of polystyrenemicrospheres (3 μm in diameter). A correction factor wasdetermined by comparing the calculated ac with the reporteddiameter of the standard. This correction factor was thenapplied to our calculations of ac for the coacervate solutions.In addition, the distortion of the droplet shape with increasingshear rate was characterized by the aspect ratio, whichwas determined by taking the ratio of the q values in thevorticity and flow directions for a given value of the measuredintensity.

Results and Discussion

The rheology of the coacervate solutions shows dramaticchanges as a function of temperature and shear rate. Thethermodynamically stable coacervates behave as Newtonianfluids at low temperatures (typically below Tφ

0) across the entireshear rate range studied (Figure 3). Once the samples begin toexhibit shear-induced phase separation (as observed by SALS),their viscosity becomes shear thinning. The Cross rheologicalmodel (eq 1) was used to fit the non-Newtonian response ofthe coacervates.44 The model quantifies the zero shear viscosity,degree of shear thinning, and relaxation time of the fluid. The fits(included as Figure S2 and Table S1 of the SupportingInfromation) exhibit similarities in all three coacervates. First,the shear thinning index is 1 for almost all of the temperature-shear rate combinations exhibiting shear-induced phase separa-tion (as quantified by SALS). Since the shear thinning index is 1,the samples exhibit a stress plateau (i.e., shear stress is indepen-dent of shear rate), which is one indicator of possible shearbanding.34,45 In addition, the relaxation time increases with anincrease in temperature. Therefore, the onset of shear thinning,which corresponds to the appearance of SIPS, appears at lowershear rates at higher temperatures.

η¼ η0 -η¥1 þ ð _γτRÞm þ η¥ ð1Þ

The coacervate samples exhibit similarites in the temperaturedependence of the viscosity [either μ when Newtonian or ηo fromthe Cross model (Figure 4)]. Before entering the phase separatingregime, samples A and C exhibit a simple monotonic decrease inthe viscosity as the temperature is increased; however, theviscosity of sample B is nearly independent of temperature belowTj. As all of the coacervate solutions approach the phaseboundary (by temperature, shear, or the combination of the two),the solutions begin to exhibit shear thinning. All of the samplesexhibit an increase in the viscosity (for 5-8 �C) once in the shearthinning regime. The increased viscosity is an indication of thephase separation occurring; i.e., the transition from the nanoscaledomain to micrometer-scale structures would increase the visc-osity. The shear thinning regime begins 1-5 �C below the criticaltemperature, Tj

0. Finally, the viscosity decreases for the highesttemperature measured for all samples.

The turbidity (and thus the phase boundaries) of the solutionsfrom the perspective of small-angle scattering is quantified usingthe normalizedmean intensity (Figure 5). The beam stop does notpermit collection of the light exiting the sample at a scatteringangle of 0�; thus, direct comparison with the previous turbiditymeasurements40 used to locate the phase boundaries is notpossible. The lowest temperature with a non-zero normalized

mean intensity indicates the location of the phase boundary atthat representative shear rate (indicated by the gray regions inFigure 5). For example, sample A exhibits shear-induced phaseseparation at 20 �C for all four shear rates, which is 2 �CbelowTj0(Figure 5a). Sample B also exhibits SIPS below Tj

0 (at 22 �C and0.1 s-1 in Figure 5b). The first appearance of SIPS for sample C isat 27 �C for all measured shear rates (Figure 5c), confirming thatSIPS in all three samples is measured below Tj

0. The anomalousappearance of Figure 5c might suggest heterogeneity of transi-tions, and temperature-induced transitions in quiescent samplesdo become markedly sharper when system polydispersity isreduced via replacement of the chemically polydisperse nonionicsurfactant TX-100 with C12E8, a monodisperse analogue.

23 Overall,the nonmonotonic nature of the normalized mean intensity as afunction of temperature does not directly correlate with the size of

Figure 5. Normalized mean intensity of the SALS images as a func-tionof temperatureat several shear rates for samplesA(a),B (b),andC(c).Dashed lines are to guide the eye.Gray shading indicates the regionin which the samples exhibit SIPS. The vertical line indicates thelocation of Tj

0.

(44) Cross, M. M. J. Colloid Sci. 1965, 20, 417–437.(45) Hu, Y. T.; Lips, A. J. Rheol. 2005, 49(5), 1001–1027.

DOI: 10.1021/la903260r 13381Langmuir 2009, 25(23), 13376–13383

Liberatore et al. Article

the phase-separated domains. The definitive information fromdetermination of the normalized mean intensity is the location ofthe transition from a single-phase fluid to a phase-separatedsolution.

The shear-induced phase boundary may also be characterizedrheologically using a temperature sweep. For example, sample Awas heated at a rate of 1 �C/min while being held at a constantshear rate of 10 s-1 (Figure 6). The evolution of the viscosity of thecoacervate is continuous with three distinct regions (i.e., onephase at low temperatures, transition to the second phase nearTj

0, and a two-phase region at higher temperatures). At lowtemperatures, the viscosity decreases slowly with an increase intemperature until ∼22 �C (an average change in viscosity of-0.065 Pa s �C-1). Between 22 and 25 �C, the viscosity dropsmuch more dramatically (-0.12 Pa s �C-1). The clear coacervatesolution at temperatures below 20 �C undergoes a transition to acompletely phase-separated state at 25 �C and 10 s-1. Above25 �C, the viscosity changes very slowly with temperature(-0.0045 Pa s �C-1). The appearance of phase-separated struc-tures via SALS (see the SALS image at 22 �C inFigure 6) precedesthe period of strongly decreasing viscosity. Samples B and C alsoexhibit shear-induced phase separation before the large decreasein the viscosity of the solution. Therefore, SALS is a moresensitive measure of SIPS than viscosity measurements alone.

The native aggregate structure of these two samples can bedirectly compared using cryo-TEM (Figure 7). Sample B exhibitsa distribution of ca. 50 nm aggregates interconnected to formextended clusters. For sample A, these clusters appear to bedisconnected and collapsed intomore dense objects similar in size.These sizes are identical to those seen for soluble interpolymeraggregates prior to coacervation; the aggregates appear to arisefrom association of smaller intrapolymer complexes. For sampleA, the cryo-TEM “snapshot” is taken at a vitrification tempera-ture well above Tj

0. However, the Tj0 of sample B is close to thecryo-TEM vitrification temperature of 24 �C. The native 50 nmaggregates in the coacervates transform into themicrometer-scaleobjects under shear, which are characterized by SALS.

It is of interest to compare samples A and B vis-�a-vis the valuesof f- in Table 1, inasmuch as sample A, formed from a mixturecloser to charge stoichiometry than sample B (f- values of 0.49and 0.39, respectively), shows more collapsed aggregates(Figure 7) and shows a transition at a lower temperature(Figure 5). The importance of charge stoichiometry has beenunderlined by several experimental9 and theoretical10 works.

However, the correlation between the bulk or mixing chargestoichiometry, e.g., f-, and the microscopic stoichiometry(the value of f- in the coacervate or the complex aggregatesthat precede them) is less clear for phase separation in the saltysolutions studied here than in salt-free systems that undergophase separation stoichiometrically.46,47 For example, a coa-cervate formed at Y=0.50, at an ionic strength of 0.80 M, wasfound to contain only 23% of the total polymer and evenless (8%) of the total surfactant, preferentially the anionic, i.e.,aY value of 0.61 in the coacervate;48 with a similar result from asolution in 0.40 M NaCl and a Y of 0.35, yielding a coacervatewith aY of 0.51, and with excess polycation on a charge basis.49

The tendency toward charge neutralization and retentionof counterions are both factors that influence coacervatecomposition. In addition, comparisons of samples A and Cwith nearly identical values of f- show that total macro-molecular concentrations influence coacervate properties. Thepronounced effects of polycation MW at fixed f- noted else-where23 also suggested caution in interpretation of the effectsof bulk stoichiometry here.

Two-dimensional SALS images provide information about themicrostructure of a fluid by inspection. Overall, the coacervatesundergo a transition from a homogeneous isotropic one-phasesolution to a heterogeneous anisotropic phase-separated sys-tem. A set of scattering images for sample A are representativeof the three main types of scattering measured for the coacer-vate system as a function of temperature and shear rate(Figure 8). An almost completely black image is observed atlower temperatures and all shear rates. The coacervate phaseseparation is a nanoscale phenomenon under low-temperatureconditions and thus cannot be detected by SALS. Next, acircular region of high scattering intensity is observed (e.g., T=22 �C and shear rate = 0.1 s-1 in Figure 8). The circularscattering pattern indicates a nearly circular droplet on themicrometer scale. The larger areas of intense scattering in theimages at 22 �C versus 20 �C (at a shear rate of 0.1 s-1) indicatea smaller droplet size (i.e., images represent length scalesin inverse space). The third type of scattering image asseen at 24 �C and 1 s-1 (Figure 8, right column) is ellipsoidalwith the long dimension of high-intensity scattering in thevorticity direction. SALS is a more sensitive measurement ofshear-induced phase separation than flow rheology as indi-cated by the appearance of micrometer-scale structures viaSALS before the sample begins to shear thin (e.g., sample A at22 �C and 1 s-1). Measured SALS images for samples B and Care included as Supporting Information (Figures S3 and S4 ofthe Supporting Information).

Similar transitions fromno scattering to circular and ellipsoidalscattering are observed with increasing temperatures for all threesamples. Micron-sized scattering objects are observed at g20 �Cfor sample A (at all of the measured shear rates). The phase-separated droplets are very close to circular (assumed to bespherical if three-dimensional data were available) at a shear rateof 0.1 s-1 and ellipsoidal at all higher shear rates. Sample Bexhibits phase-separated droplets beginning at 24 �C (nearlycircular 2D scattering image). Ellipsoidal droplets are observedfor sampleB from24 �Cand 1 s-1 to 30 �Cand 32 s-1. SALS from

Figure 6. Viscosity as a function of temperature for sample A at ashear rateof 10 s-1 showing the transition fromaone-phase state toa phase-separated state. Arrows correspond to the locations ofreported SALS images.

(46) Ahmed, L. S.; Xia, J. L.; Dubin, P. L.; Kokufuta, E. J.Macromol. Sci., PureAppl. Chem. 1994, A31(1), 17–29.

(47) Tsuboi, A.; Izumi, T.; Hirata, M.; Xia, J. L.; Dubin, P. L.; Kokufuta, E.Langmuir 1996, 12(26), 6295–6303.

(48) Dubin, P. L.; Oteri, R. J. Colloid Interface Sci. 1983, 95(2), 453–461.(49) Davis, D. D. Intermacromolecular association of polycations with oppo-

sitely charged micelles. M.S. Thesis, Purdue University, West Lafayette, IN, 1984.

13382 DOI: 10.1021/la903260r Langmuir 2009, 25(23), 13376–13383

Article Liberatore et al.

sample C first appears at a higher temperature, 27 �C, than thatfrom samples A and B and exhibits the same progression throughcircular and ellipsoidal scattering as temperature and shear rateincrease.

The characteristic size and aspect ratio of the phase-separateddroplets were derived from the radially averaged intensity (e.g.,using Debye-Bueche plots like Figure S1 of the SupportingInformation). The droplet size ranges from ∼1 to 4 μm, and theaspect ratio of the droplet varies from ∼1 to 4. The continuous

transition from no small-angle scattering to circular and finallyelongated scattering patterns is observed with an increasing shearrate at a constant temperature. For example, sample A shows thecontinuous progression of shapes in the shear-induced structuresat a temperature above Tφ

0 (Figure 9). At 0.1 s-1 and 26 �C, anearly circular SALS pattern is observed, which corresponds to adroplet with an ac of 1.0 μm and an ar of 1.0. At 0.1 s

-1,approximately 1 μm nearly circular droplets appear as the first

Figure 7. Cryo-TEM images of (a) sample A and (b) sample B showing a native aggregate size of∼50 nm.

Figure 8. Small-angle light scattering images as a function oftemperature at two shear rates for sample A. SIPS occurs at highertemperatures and shear rates.

Figure 9. (a) Viscosity as a function of shear rate at 26 �C forsample A (top) with inset SALS patterns representing the smoothtransition from circular to ellipsoidal droplets with an increase inaspect ratio. (b) Characteristic length (ac) and aspect ratio of thephase-separated droplets for sampleA corresponding to the SALSpatterns in panel a.

DOI: 10.1021/la903260r 13383Langmuir 2009, 25(23), 13376–13383

Liberatore et al. Article

observation of shear-induced phase separation in all three coa-cervate solutions. The nearly circular droplet becomes elongatedat higher shear rates until the aspect ratio of the droplet reaches 3.0at 25 s-1 for sample A (Figure 9). The increasing aspect ratio withshear rate indicates the growth of the droplets is almost exclusivelyin the flow direction (and is observed for all three samples).

A conceptual picture of SIPS in coacervates can be derivedfrom the experimental observations presented earlier. Ingeneral, shear flow transforms the clear coacervate fluid withdomain sizes of 50-100 nm by transforming the polyelec-trolyte/micelle complexes into extended chains or “necklacesof polyelectrolyte decorated with micelle beads”50 as por-trayed in Figure 10. In a manner entirely analogous to shear-induced phase separation of simple polymers,28 these ex-tended chains create efficient inter(polyelectrolyte/micelle)interactions, presumably at the expense of intra(poly-electrolyte/micelle) interactions, with complementary spacingof bound micelles that can then interact electrostatically withmicelle-poor domains of adjacent complexes, a form of“polarization”.10 These more efficient interactions promotethe expulsion of small ions from the complex, resulting in theformation of micrometer-scale phase-separated droplets. Thephase transition temperature Tj

0 coincides with the transitionfrom shear-independent (Newtonian) to shear-dependentviscosity. The initial appearance of small-angle scatteringprecedes the transition in rheology from Newtonian to shearthinning. Further studies are needed to correlate transitiontemperatures for quiescent phase transitions reported fromturbidimetry and small-angle scattering with observationsmade under shear, a task that may be facilitated by thereduction of system polydispersity.

Conclusions

The phase behavior under flow of a polycation/mixed micellecoacervate was investigated by rheology and rheo-SALS.

Although shear-induced phase separation has been extensivelyreported for solutions of certain polymers and wormlike micelles,this is the first observation of SIPS for a polymer/micelle systemand likely the first quantitative study of SIPS in a complexcoacervate. Under shear, the coacervate solutions convert fromhomogeneous, isotropic, one-phase systems to heterogeneous,anisotropic, two-phase systems. Thus, behavior typicallyobserved for wormlike micelles is attained for small micellesbound to a polyelectrolyte. Figure 11 provides a summary ofthe shear- and temperature-induced phase transitions observed inthis work. Below 18 �C, all three of the coacervate solutions areoptically clear Newtonian fluids across the entire shear rate rangeinvestigated. Once a critical temperature and/or shear rate isachieved, phase separation occurs. Two definitive SALS patternsdemonstrate the appearance of circular droplets at low shear ratesnear the critical temperature and ellipsoidal droplets at highertemperatures and shear rates. The shear-induced droplets range insize from∼1 to 4 μm.The ellipsoidal droplets have aspect ratios ashigh as∼4. Overall, the shear-induced phase separation has beenexplored as a function of shear rate and temperature at the steadystate. Additional insights will be gained via exploration of thekinetics of the phase separation under flow and the possibility ofshear banding.

Acknowledgment. Partial support of this work was receivedfrom the donors of the Petroleum Research Fund (M.W.L.).Portions of this work were supported by grants from ShiseidoCorp. (P.L.D.) and from the donors of the Petroleum ResearchFund to A. Dinsmore and P.L.D. We acknowledge assistancefrom Dr. JoAn Hudson (AMRL, Clemson University, Clemson,SC) with cryo-TEM.

Supporting Information Available: Description of thematerial, an example fit of the small-angle light scatteringdata to the Debye-Bueche model, viscosity curves fit to theCross model, model fit parameters, and 2D SALS images asa function of shear rate and temperature for samples B andC. This material is available free of charge via the Internet athttp://pubs.acs.org.

Figure 10. Schematic representation of shear- and temperature-induced phase separation for PDADMAC/TX100-SDS coacer-vate. Both processes involve loss of counterions arising from anincreased number of polyelectrolyte-micelle interactions, but byan intercomplex vs intracomplex mechanism for the former.Reproduced from ref 26. Copyright 2008. American ChemicalSociety.

Figure 11. Schematic of the relevant domain size of the coacervatesolutions with changing temperature and shear rate.

(50) Lee, L. T.; Cabane, B. Macromolecules 1997, 30(21), 6559–6566.