Neutron Reflection Studies of Copolymers at the Hexane - American

Chemicals & Polymers, Runeorn Heath, Cheshire, U.K.. Received January 25, 1993. In Final Form: August 19, 199P. The conformations of three copolymers ...
0 downloads 0 Views 953KB Size
3530

Langmuir 1993,9, 3530-3537

Neutron Reflection Studies of Copolymers at the Hexane/ Water Interface J. S. Phipps,t R. M. Richardson,'*t T. Cosgrove,+and A. Eagleshamt School of Chemistry, University of Bristol, Cantock's Close, Bristol BS8 1TS, U.K., and I.C.I. Chemicals & Polymers, Runeorn Heath, Cheshire, U.K. Received January 25, 1993. In Final Form: August 19, 199P

The conformations of three copolymers at the hexane/water interface have been studied using neutron reflection. The experiment is made possible by using a specially-designedapparatus in which a thin ( 10 Fm) layer of hexane is spread on a water surface, thus allowing acceptable transmission of a neutron beam incident at a grazing angle to the surface. Since none of the polymers used was deuterated, sensitivity to the polymer structure was achieved using contrast variation of the two solvents. A triblock copolymer of poly(ethy1eneoxide)-poly(propy1eneoxide)-poly(ethy1ene oxide) (F127of the Pluronic series) is shown to adopt a conformationthat extends beyond the micellar radius in water and, thus, appears quite stretched. A random copolymer of poly(viny1 alcohol-co-acetate)is shown to adopt a very flat conformation at the interface, forming a dense layer no more than 20 8, thick. While both of these polymers have hydrophobic moieties,hexane is not a good solvent for either. In contrast, the hydrophobic block of a diblock copolymer of poly(dimethylsiloxane)-poly(2-vinylpyridine N-oxide) is very soluble in hexane, and in this case the hexane appears to have a much greater influence over the polymer structure. A substantial proportion of the polymer is found on the hexane side of the interface. N

Introduction The structure of block copolymers at a liquid/liquid interface is of fundamental interest as a result of the extensive application of these materials as surfactants, emulsifiers, and foaming and antifoaming agents. However, direct measurements of conformations at a liquid/ liquid interface have until recently not been possible. Simple parameters such as the average layer thickness, adsorbed amount, and area per molecule can be obtained from a number of techniques such as ellipsometry,' radiotracers,2and surface pressure/area measurement^.^ However, owingto the heterogeneous nature of emulsions, techniques such as depletion isotherms and photon correlation spectroscopy are rather unreliable when applied to the liquid/liquid interface in such systems. Direct measurements of polymer conformation at a liquid/liquid interface have been made on relatively monodisperse emulsions using small-angle neutron scattering (SANS)? a technique closely related to neutron reflection. Many neutron reflection measurements have now been reported from polymers at the aidliquid interface.%' In order to achieve sufficient constrast it has normally been necessary to prepare deuterated versions of the polymers under study. We have developed an apparatus enabling neutron reflection measurements from a hexane/water inferface and used this to study the conformation of three different copolymers: a triblock copolymer of poly(ethylene oxide)-poly(propy1ene oxide)-poly(ethy1ene ox+ University of Bristol. I.C.I. Chemicals and Polymers. *Abstract published in Advance ACS Abstracts, November 1, 1993. (1)Azzam,R. M. A.;Bashara, N. H.EllipsometryandPolarisedLight; North Holland Amsterdam, 1977. (2)Graham, D.E.;Phillips, M. C. J.Colloid Interface Sci. 1979, 70, 403. (3)Brooks, J. H.;Pethica, B. A. Trans. Faraday SOC.1964, 60, 208. (4)Cosgf.ove, T.;Mallagh, L. M.; Ryan, K.; Scheutjens, J. M. H. M. J. Surf. SCL Technol. 1988, 4,81. (5) Guiselin, O.;Lee, L. T.; Farnoux,B.;Lapp, A. J.Chem.Phys.1991, 95,4632. (6)Rennie, A. R.; Crawford, R. J.; Lee, E. M.; Thomas, R. K.; Crowley, T. L.; Roberta, S.; Qureshi,M. S.;Richards, R. W. Macromolecules 1989, 22, 3466. (7) Richards, R. W.; Henderson, J. A. Polymer, in press.

ide) (PEO-PPO-PEO or F127), a random copolymer of poly(viny1 alcohol-eo-acetate) (PVA-Ac), and a diblock copolymer of poly(dimethylsiloxane)-poly(2-vinylpyridine N-oxide) (PDMS-PVPO). It was not possible to produce deuterated versions of any of these polymers, and so the necessary contrast variation has been achieved by alternately deuterating the water and hexane phases.

Experimental Section Neutron reflectivity measurements were carried out on the CRISPinstrumentat the Rutherford Appleton Laboratory,near Didcot, England.* This is a dedicated reflectometer,which uses a fixed angle of incidence and a pulsed, polychromaticneutron beam. Specularly reflected neutrons are received by a single detector, and their wavelengths are analyzed by time-of-flight. In order to studythe liquid/liquidinterface,a special apparatus has been designed, which has been fully described elsewhere.g The large incoherent scattering cross section of hydrogen (and, to a lesser extent, deuterium) prevents the transmission of a neutron beam through more than a few millimetresof oil or water. The problem of penetration of the neutron beam through a bulk liquid phase is overcome by using a thin layer of oil, typically 10-20 Nm thick, spread on a water surface. For a 10 cm long sample at the grazing angles of incidence used (0 I1.5O),this reducesthe path lengththrough the oil phase to a few millimeters and gives acceptabletransmission. In all the experimentsthe oil used was n-hexane, for which the attenuation coefficient is approximatelyconstant over the wavelength range used. A pool of water is placed in a PTFE trough sealed in an enclosure and presents a proud meniscus to the incoming neutron beam. The film is maintained in contact with a small reservoir of hexane around the water. Its thickness is maintained by cooling the water (using an electrical Peltier device), such that the rate of condensation of hexane onto the surfacejust balances its rate of drainage. Since the drainage rate is very slow, a temperature differenceof 1 K between the water surfaceand the surroundings is usually sufficient. The layer thickness can be controlled by adjustment of this temperature differenceand can be estimated from the observed attenuation of the reflected intensity below the critical edge (when the reflectivity of the hexane/water (8)Penfold, J.;Ward, R. C.; Williams, W. G. J. Phys.E Sci.Znstrum. 1988,20, 1411.

(9) Cosgrove, T.; Phipps, J. S.; Richardson, R. M. Colloids Surf. 1992,

62, 199.

0743-7463/93/2409-3530$04.00/00 1993 American Chemical Society

Copolymers at the Hexanelwater Interface way;eaIed,,,Enclosure

Langmuir, Vol. 9, No. 12,1993 3531

Hexane Film

/

The reflectivity data shown in this paper have been analyzed by model-fitting, using the matrix method of AbBlb" to calculate the reflectivity from the perpendicular refractive index profile. The interface is divided into a number of uniform layers. A smoothly-changingrefractive index profile may be modeled by introducing an error function "roughness" parameter between the layers, as shown by NBvot and Croc6.12 The thickness, scattering length density, and interfacial diffuseness of the layers are adjusted to achieve an optimum fit to the data, using a nonlinear least-squares fitting routine. For a polymer in solution, the scattering length density of each layer is calculated according to the volume fraction, 4, of the polymer in that layer:

v n \

Hexane

Collimating Slits

Peltier Coolers Thermostatted Plate Figure 1. Schematic diagram of liquid/liquidinterface reflection experiment. interface is known to be unity). A schematic diagram of the experiment is given in Figure 1.

The Neutron Reflection Technique In a reflection experiment, reflectivity is measured as a function of wavevector transfer, Q, perpendicular to the surface or interface. The value of Q is related to the grazing incident angle, 8, and the wavelength, A, by Q = -4a sin 8

A

One can thus scan either 8 or A. CRISP scans A a t constant 8, with a useful wavelength range of 0.5-6.5 A. In order to measure the maximum possible range of Q, several experiments must be performed, using different angles of incidence, and the data then combined into a single reflectivity profile. For a typical sample as described in this paper, angles of 0.35', 0 . 6 O , and 1.5' would be used. The minimum measurable reflectivity is normally limited to about 10-6 by the isotropic, incoherent background scattering from the sample. In a typical experiment on CRISP, reflectivity can be measured over a Q range of 0.01-0.25 A-l. The reflectivity of an interface can be calculated from its refractive index profile (perpendicular to the surface) using the same equations as for perpendicularly polarized light.'O The real part of the refractive index, n,is simply related to the scattering length density, p , by

p

is related to the local density of scatterers by p =znibi

(3)

(4) P = 4Pp+ (1 - 4 ) P B where p p is the scattering length density of the pure polymer and p s is that of the solvent. A single volume fraction profile is sought that is consistent with all the different contrasts measured. However,with only a limited number of contrasts available, it is unfortunately not possible to guarantee that such a model is unique. Errors are assessed by fixing each parameter at several values around the optimum and allowing the other parameters to adjust accordingly. The error is then the region around that optimum for which the fit remains acceptable. The quality of each fit is assessed visually, and also by the calculated x2 parameter from the least-squares fitting routine. The latter approach is not entirely satisfactory, sincethe error bars are calculated only for neutron counting statistics, where systematic errors may outweigh purely statistical errors. The relative values of x2 for different fits to the same data set should, however, remain a good measure of their relative quality. Data are usually presented as either the log of reflectivity against wavevector, Q, or, as in this paper, as RIRF, where RF is the calculated reflectivity of a clean, sharp interface between decay that the two bulk media. This eliminates the is characteristic of all reflectivity profiles and highlights the sensitivity of the data to interfacial structure. Since the reflectivity profile very close to the critical edge is very sensitive to the incident angle and divergence of the beam, the plots of RIRF often show spurious peaks and troughs in this region. These do not affect the information concerning the adsorbed layer, which is essentially derived from the reflectivity profile at higher Q. The matrix method provides an exact calculation of the reflectivity but gives little insight into the relationship between reflectivity and structure. For reflectivities lower the reflectivity is given to an extremely good than 1k2, approximation by the formula13

I

where ni is the number density of the ith nucleus and bi is its scattering length. Typically, refractive indices differ from unity only by a few parts per million, and hence grazing angles of incidence must be used. Much of the versality of the technique arises from the fact that the scattering lengths of hydrogen and deuterium have opposite signs. This enables the scattering length density of compounds containing hydrogen to be adjusted over a substantial range using isotopic substitution. In some of the experiments described in this paper we use isotopic substitution in water or in hexane to obtain a scattering length of approximately zero, giving the hexane the same scattering length as air, and thus rendering the liquid effectively"invisible" to the incomingneutron beam. This is known as "contrast-matching". (10) Goldberger, M.; Seitz, F. Phys. Rev. 1947, 71, 294.

(5)

We make use of this formula to give a simple Guinier-type analysis of the reflectivity data, using a method first suggested by Cr0w1ey.l~The interfacial region is represented by a "film" on top of a homogeneous subphase. Representing the subphase by a Heavyside step function, H(-z), which is unity for z negative and zero for z positive, we have P ( Z ) = ApH(-z) + Pf(z) (6) where Ap is the difference in scattering length density between the incident medium and the subphase and p f is (11) Abklh, F. Ann. Phys. (Paris) 1948,3,504. (12) Nkvot, R.; Crock, P. Phys. Appl. 1980,15,761. (13) Als-Nieleen, J. Z . Phys. B Condens. Matter 1986, 61, 414. (14) Crowley, T. L. D. Phil Thesis, University of Oxford, 1984.

3532 Langmuir, Vol. 9, No. 12, 1993 the scattering length density contributed by the polymer film. By use of the known Fourier transform of the step function, this gives R(Q) = Ro(Q)Ap2- Rl(Q)Ap + &(Q)

(7)

**

Air (Hexane)

Water

Adsorbed Polymer P

Phipps et al.

(

a

(i) (ii)

tI I

where

!-

I P(z)



I

I ‘

P P L ’ A RoAp2is the reflectivity of a clean, sharp interface, which we call RF, the Fresnel reflectivity. By expanding the R1 and R2 terms as Taylor series and approximating the leading terms as exponentials, we obtain 16r2 R2 = -r2exp(-Q2u2)

(9)

Q2

R, = 32r2 r(z)exp( Q2

-)6(z)

-Q2 (2’)

(10)

u2 is the mean square deviation of the adsorbed layer thickness, (2”) is the nth moment of the distribution about z = 0, and

For the “Guinier” analysis we make the further approximation that

(12) This holds exactly for a uniform layer and is a good approximation for most other types of distribution. Hence the final approximation: ln[Q2(R- RF)] = h[16r2(r2- 2I’(z)Ap)] - Q2u2

(13) The approximation will break down with the Taylor expansion when Q u > 1and, of course,the kinematic theory is not valid at low Q where R > le2.Despite these limitations, we have found it a very useful for initial interpretation of reflectivity data which provides a starting point for futher analysis by model fitting using the exact matrix calculation. In practice we have found reasonable straight line regions at Q < l / a in most of the Guinier plots (of ln(Q2R) vs Q2). When the subphase is contrastmatched with air (Ap = 01,the R1 and Ro terms disappear, and we may use the intercept of the Guinier plot to obtain the scattering density integrated over the absorbed layer, r. This may then be converted directly in the adsorbed amount. In the more general case, R - RFmay be obtained by subtracting the reflectivity measured from a clean surface from that measured from the surface with adsorbed polymer and an estimate of the polymer layer thickness u derived from the Guinier plot. When the subphase is D20 and the adsorbing compound has a scattering length density close to zero, most of the scattering length density in the absorbed layer comes from the D2O. However, it is still possible to use the Guiniertype analysis to estimate the polymer layer thickness, u. This is not intuitively obvious, so we given an outline to the argument in the paragraph below. Consider the situation shown in Figure 2. Provided that the water/hexane interface remains reasonably sharp, with a D2O subphase the scattering length density profile will be as shown in Figure 2(ii). In the kinematic approximation, the reflectivity depends upon the modulus squared

(iii)

(iv)

Figure 2. Scattering length density profile of a hydrogenous polymer in D20.

of the Fourier transform of the scattering length density so we can apply some of the basic properties of Fourier transforms. For instance, the reflectivity must be unchanged by lateral inversion of p(z) to p(-2). Vertical inversion and addition of a constant also make no change to the reflectivity (except for a delta function at Q = 0 which is not observable). These results mean that Figure 2(iii) gives the same reflectivity as Figure 2(ii). Figure 2(iii) can be conveniently separated into the step function from the substrate and the (laterally inverted) polymer profile (Figure 2(iv)), and its reflectivity can then be divided into three components as in eq 8 above. The value of u that could be determined from a Guinier plot would then relate to the polymer profile inFigure 2(iv). However, (iv)was obtained by simple geometric operations on (ii) so the value of u must also describe the polymer profile within the D20 as in (ii).

Sample Materials D2O and perdeuterated hexane (hexane-du) were obtained from MSD Isotopes, Ltd. DzO was used as supplied, and hexane with approximately half the scattering length density of D2O was prepared by mixing hexane-& with hydrogenous hexane (Aldrich Gold Label) in the ratio 6 0 40. Air-contrast-matched water was prepared from the D2O (8%) and ultrapure water obtained from a Millipore Milli-Q system. N-Hexyl deuteride (hexane-&) was prepared from 1-bromohexanevia a Grignard reagent and D2O and is approximately contrast-matched with air. Three copolymers were selected for this study. They were chosen to be examples of different types of copolymer and because they were known to have some solubility in water. This is important because it helps maintain equilibrium in the polymer layer adsorbed between the bulk aqueous phase and the hexane film. The triblock copolymer of PEO-PPO-PEO was F127 of the Pluronic series, supplied by BASF Wyandotte Corp., MI. The random copolymer of PVA-AC was obtained by fractionation of a commercial Alcatex sample. NMR analysis showed it to be 88% hydrolyzed, with the acetate groups typically arranged in groups of three. The diblock copolymer of PDMS-PVPO was synthesized in the Bristol laboratory by Dr. Simon Biggs by sequential anionic polymerization of 2-vinylpyridine and hexamethylcyclotrisiloxane, followed by oxidation with peroxoethanoic acid.’S Details of the polymers are given in Table I. Results PEO-PPO-PEO Triblock Copolymer. Figure 3a shows the reflectivity ratio RIRF of a solution of 0.5 w t % F127 in D20 in contact with air-contrast-matched hexane. (15)Bigge, S.R. Ph.D. Thesis, University of Bristol, 1990.

Copolymers at the Hexanelwater Interface

Langmuir, Vol. 9, No. 12,1993 3633 Table I. Copolymer Detail8

block sequence a-b-a (a, PEO; b, PPO) random a-b (a, PDMS; b, PVPO)

polymer F127

PVA-AC PDMS-PVPO

1 .o o? \ E

0.5

0.0I 0.00

0.05

0.10

Q

0.15

a-1

Figure 3. Reflectivity ratio R/RFof PEO-PPO-PEO polymer at the hexane/water interface: (a) hexane-ddDz0; (b) 6040 hexane-dl.:hexane/DzO. -lo[*

I

I

r;

!I

: t/ I

V

0

V

-

-12 -14

I

-16)

I

0.000

\

\1

0.005

'

*

I

b

, 0.010 Q2

/

0.015

0.020

0.025

Figure 4. Guinier plot of reflectivityof PEO-PPO-PEO polymer at the hexane-dl/DzO interface.

The concentration was chosen to ensure a high adsorbed amount, using the surface tension data obtained by Wanka and co-workers.18 This concentrationis rather higher than the critical micelle concentration (cmc). Since the bulk neutron scattering length density of both PEO and PPO is very small (0.06 X 106 A-2 for PEO and 0.03 X 106A-z for PPO), the measurement is sensitive only to the depletion of D2O at the surface where it is displaced by the adsorbed polymer and is insensitiveto the polymer in the hexane. Figure 4 shows the Guinier plot of these data, which gives a u value (for the polymer on the water side of the interface) of 24 A. The low Q data were ignored in determining the slope because it is believed to be unreliable. It ie the result of subtractingtwo reflectivityprofiles (Le. from the interface with and without adsorbed polymer) very close to the critical angle where experimental factors such as a small difference in the setting of the incident beam angle can give a large difference in the data. The reflectivity data from this polymer adsorbed at the DzO/ air-contrast-matchedhexane interface can be successfully fitted with a two-layer model, with the polymer extending some 90A into the solution. The thickness is well-defined by the width of the interference fringe. The solid line in Figure 3 is the model-fit,the parameters of which are given in Table II. Although the thickness is quite well deter(16)Wanka, G.;Hoffmann,H.; Ulbricht,W.Colloid Polym. Sei. 1990, 268,101.

MnA block

MnB block

MJMn

4400 37000 (88% PVA) 3700

3700

1.3

1.1 6Ooo

1.2

mined, there is quite a strong correlation between the volume fraction and the diffuseness of the polymer layer. Fits can be achieved with a range of volume fractions and roughness parameters, since the effect of an increase in volume fraction is to increase the amplitude of the fringe, whereas increased roughness tends to reduce it. Figure 5 shows the reflectivity ratio of a similar solution at the air/water interface. As can be seen, the reflectivity is almost identical to that at the hexane/water interface, suggesting that the presence of hexane has little effect on either the adsorbed amount or the conformation of the polymer on the water side of the interface. The similarity of the two data sets suggests that perhaps the hexane might have evaporated in the first experiment; however, both monitoring of the absolute reflectivityand visual inspection of the sample, in which interference colors could be seen, confirmed that the layer remained throughout the experiment. PPO does dissolvein hexane, although the value of its Flory-Huggins exchange parameter, x , is quite high (-0.51, so it seems that changing the upper phase from air to hexane has no discernible influence on the distribution of the polymer in the aqueous phase. With a polymer of such a low scattering length, measurement of the reflectivity in air-contrast-matched water is not possible, since even at low Q the signal is not significantly higher than the background. In order to obtain more information either the polymer or the hexane phase must be deuterated. Deuterated Pluronics are not available, and so the reflectivity was measured with a partially deuterated hexane layer. A composition of 60% hexane-du 40% hexane was chosen in order to be able to measure the hexane layer thickness at the lowest angle by observation of two critical edges (one from the &/hexane interface and one from the hexane/D20 interface). For the purposes of model-fittingit was assumedthat the phase coherence of the neutron beam across the relatively thick hexane layer is lost, and so the total reflectivity is obtained by adding together all the contributions from the air/ hexane (R1) and hexane/water (Rz) interfaces so that R, + R2e-28t/&1 - 2 R l R 2 e - W ~ 1 RTotd

=

1- R R e-28t/aw1

(14)

1 2

where /3 is the attenuation coefficient of the hexane layer and 61 is the angle of the refracted beam in the hexane layer. Since /3 has been determined in a separate transmission experiment, the hexane thickness, t , may be determined by analysis of the drop in reflectivity between the two critical edges. Away from the critical edge region, where both R1 and R2 are small, this simplifies to RTotd

- R, + R2e-28t/sine,

(16)

The reflectivity ratio for this contrast is shown in Figure 3b. The polymer volume fraction profile corresponding to the solid line fita in Figure 3 is shown in Figure 6a. Introduction of the partially deuterated upper phase does not shift the position of the fringe significantly. It was found necessary to add a third layer to the model in order to fit both data seta. The necessity of adding this layer in the hexane phase is demonstrated by the dashed line in Figure 3 which is the model calculation for a partly deuterated upper phase with only the two polymer layers in the aqueous phase. The third layer, which was in the

3534 Langmuir, Vol. 9, No. 12, 1993 polymer F127

Table 11. Fit Parameters for Copolymers at the He.ane/Water Interface DdA 91 DdA 92 DdA 98 40 f 5 0.35 i 0.05 21 i 6 0.3 f 0.05 70 f 6 0.13 i 0.05

PVA-AC PDMS-PVPO 20 i 10 0.8i 0.1 0 Layer 1 in hexane and layers 2 and 3 in water.

1.0

Phipps et al.

14i2 25f5

-

0.8-

& \

0.6 -

E

0.40.2

1

0.00

0.05

0.10

0.15

Q / A-’ Figure 6. Reflectivity ratio R/RFof PEO-PPO-PEO polymer at the air/DzO interface.

hexane phase, had a polymer volume fraction of 0.3 and thickness 40 A. Unfortunately neither of these parameters is very accurately determined (only 30%) because the reflectivity also depends on the thickness of the hexane film that is only known approximately. However the form of the polymer profile must be close to that shown in Figure 6a. A potential hazard in the measurement of the specular reflectivity of a micellar solution is the contribution of small-angle scattering from the micelles in the bulk, which can lead to the measured intensity being rather larger than the true specular reflectivity. This small-angle scattering is not confined to the specular direction and so may be measured in an off-specular direction and subtracted from the apparent specular signal. This is best done with a multidetector by measuring the intensity around the specular peak and extrapolating to obtain the “off-specular” intensity in the specular direction. CRISP is now equipped with a multidetector, but a t the time of the experiment this was not available. In order to assess the contribution of off-specular scattering to the signal, the detector was misaligned from the specular beam by a fewmillimeters. The range in Q measured was thus almost identical to that in the specular experiment, but with the specular component eliminated. The intensity measured was 2 full orders of magnitude lower than the specular signal, and showed a virtually flat Q dependence. The contribution of small-angle scattering to the observed specular signal can thus be neglected in this case. PVA/Ac Random Copolymer. Figure 7 shows the reflectivity ratio of a solution of 2 X w t 7% PVA/Ac in DzO in contact with air-contrast-matched hexane, and Figure 8 is the Guinier plot of the same data. The data can be quite satisfactorily fitted with a single layer in the water 14 A thick, with a volume fraction of 0.5. The measurement with a partially deuterated upper layer was attempted, but unfortunately failure of the instrument meant that only the reflectivity a t an incident angle of 0.35O could be measured. This shows no appreciable deviation from the clean interface, which suggests that the adsorbed polymer layer is thin and provides no additional contribution to the reflectivity. This is consistent with the model fitted to the first set of data. The volume fraction profile fitted to these data is shown in

0.55 i 0.05 0.2 f 0.05

r/mg m-2 2.9 & 0.5 0.75 i 0.1 2.0 f 0.5

Figure 6b and is clearly much narrower than that of the triblock copolymer. PDMS/PVPO Diblock Copolymer. In the case of the PDMS/PVPO copolymer it was possible to measure the reflectivity at three different solvent contrasts, because PVPO has a substantial scattering length density and hence scatters sufficiently strongly for the measurement to be made with both the water and the hexane contrastmatched with air. All experiments were done with a concentration of 0.02 wt 7% PDMS/PVPO dissolved in the water. This is above the critical micellar concentration. The first measurement was done using DzO and aircontrast-matched hexane, in order to examinethe structure of the polymer on the water side of the interface. The reflectivity ratio at this contrast is shown in Figure 9. As can be seen, the effect on the DzO reflectivity is weak in comparison with the other polymers studied. It can be fitted with a single layer 25 A thick of volume fraction (assuming that it is primarily PVPO) between 0.1 and 0.2 in the DzO. There is quite a strong correlation between the volume fraction and the layer/subphase roughness which leads to the uncertainty in the volume fraction for this single measurement. Unfortunately, it was not possible to extract any information from a Guinier analysis, because the deviation from the clean interface reflectivity was too small. The error bars on the Guinier plot were sufficiently large to make it meaningless. The second measurement was done with both solvents contrast-matched with air, making the polymer the only reflecting constituent of the sample. The reflectivity of this sample (Figure loa) was rather stronger than might be expected from the DzO measurement and indicates that the adsorbed amount is substantial. As the hexane layer was formed the intensity of the reflection increased, despite the attenuation factor, indicating that the presence of the hexane increases the adsorbed amount. Unfortunately, it is not possible to obtain a clear value of r from this because of the attenuation of the hexane film. This is because, in the absence of critical edge, it is not possible to measure the hexane layer thickness, and so the absolute reflectivity of the interface is not known. However, in order for the hexane attenuation to have a positive value, it must be at least 1.5 mg m-2. If the polymer blocks were well separated so that all the PVPO was in the water and all the PDMS in the hexane, we would obtain an adsorbed amount of only -0.5 mg m-2 from the first measurement, so this is a clear indication that at least some of the PVPO protrudes out of the water. The overall thickness from the all-contrast-matched measurement is approximately 40 A, but on its own the measurement is not resolvable into sublayers of different density. The third contrast measured used a DzO subphase and a partially-deuterated hexane layer, of composition 60/40 (d/h) as before (Figure lob). The volume fraction profile used to fit all three contrasts simultaneously is shown in Figure 11. It consists of two polymer-containing layers, one in each solvent. In this model the diffuseness of each of the layers with ita bulk solvent is very broad, whereas the interface between the two layers is much sharper. In the absence of clear interference effects, the thickness of the layer is not very well defined, so the uncertainties quoted for layer thicknesses in Table I1 are much larger

Langmuir, Vol. 9, No. 12,1993 3536

Copolymers at the Hexanelwater Interface 0.6

0.4

0*41

0.2

0.2 0.0 -100

-50

50

0

100

150

0.0I 0.00

0.05

0.10

Q / A-'

z / A

Figure 6. Fitted volume fraction profiles for PEO-PPO-PEO

and PVA-Ac at the hexane/water interface: (a)PEO-PPO-PEO (b) PVA-Ac.

0.20

0.15

Figure 9. ReflectivityratioR/RpofPDMS-PVPO at the hexanedl/DzO interface.

,

4

I

I

2 \ E

0.4

1

0.00

1 0.05

0.10

0.15

Q / A-'

Figure 7. Reflectivity ratio R/Rp of PVA-Ac at the hexanedl/DzO interface. 1.0'

0.8

-

'

I

Hexane

-40 Q2

/ A-'

Figure 8. Guinier plot of reflectivity of PVA-Ac at the hexanedl/DzO interface. than those found for F127. Indeed, for the third contrast the data at different angles fit best to different thicknesses. This may be an effect of a slow change of conformation of the polymer. The volume fractions given in Table I1 assume that the aqueous polymer layer contains only PVPO segments, whereas the layer in hexane contains both PVPO and PDMS segments. Unfortunately, the values of are not particularly sensitive to this assumption. It does seem reasonable, however,since most of the polymer is contained within a dense layer in the hexane, leaving only a small amount (presumably water-soluble PVPO segments) in the water. Table I1 summarizes the fit parameters obtained.

I

-20

,

I

-

Water

0 z / A

20

40

Figure 11. Fitted volume fraction profile for PDMS-PVPO at the hexane/water interface.

Discussion Three different copolymers have been shown to adopt very different conformations at the hexane/water interface. In the case of F127 and PDMS/PVPO, solvent contrast variation has allowed the volume fraction profile to be obtained with some confidence. The choice of a 6040 mixture of hexane and hexane-du was necessary in order to monitor the hexane layer thickness by the observation of an intensity decrease between the two critical edges. As a result, however, the sensitivity to polymer conformation on the hexane side of the interface is not as good as that on the water side. It is of interest to compare the volume fraction profiles obtained with the measurements of other workers and current theories of polymer adsorption. Although no

Phipps et al.

3536 Langmuir, Vol. 9, No. 12, 1993 '

.

O

~ 1

hexane

water

C

.-c e LL 0)

-5 0

>

0.2

" 0

20

10

30

50

40

Layer number

Figure 12. Scheutjens-Fleer simulation of block copolymer at a liquialiquidinterfaceshowing the volume fraction of the polymer.

1

A II

PVA random copolymer

10

20

30

40

so

Layer number

Figure 13. Scheutjens-Fleer simulation of PVA/Ac random copolymer at a liquialiquidinterfaceshowingthe volume fraction of all three components.

equivalent measurements of this kind exist for F127, the conformations of several other Pluronics were measured at the toluene/water and cyclohexane/waterinterfaces by Cosgrove and co-workers4using SANS from oil-in-water emulsions. Wanka et al.l6 measured the micellar radius in water over the range of the concentration used and found to to be 60 A by both light scattering and SANS. This assumes a spherical micelle but is clearly smaller than the dimension of the F127 polymer shown in Figure 6. We can calculate a value of u = 33 A from the measured profile of F127 and compare it with the results obtained from SANS.14 The SANS studies at the toluene/water interface gave u values of 54 A (F108, Mn = 15 600),35 A (P104, Mn = 4800), and 37 A (P85, Mn = 4500), and a t the cyclohexane/water interface 37 A (F108) and 27 A (P104). The molecular weight of F127 lies between that of F108 and P104, so the u value of 33 A for the reflection measurement would appear to be consistent with the cyclohexane values. Toluene is a much better solvent for PPO than either hexane or cyclohexane,and sothe polymer distribution is rather wider at the toluene/water interface. The neutron reflection results are therefore consistent with the other available measurements on Pluronics. We now consider the conformation of the polymer in more detail. The estimated (un erturbed) radius of gyration of a PEO block alone is 20 and that of the PPO block 16 A. If we compare these dimensions with the 90 A (assumed to be PEO) in water and 40 A (assumed to be PPO) in hexane, it is apparent that the PEO is quite stretched. The PEO tails are stretched to 4.5R, while the PPO loop has adimension of only 2.5RP These qualitative considerations are borne out by a Scheutjens-Fleer (SF) simulation,17shown in Figure 12, for an A&,& polymer a t the liquid/liquid interface. The x parameters were chosen to reflect the solubilities of the EO and PO blocks in hexane and water and are given in Table 111. (A value of 0.45 for PEO in water might be a better choice, but in practice it makes vitually no difference to the simulation.) The resemblance of this profile to the experimentally-

determined structure of F127 is marked, although the apparent adsorbed amount in the simulation is unusually high. The approximate size of one lattice layer is 5 A, and so the overall width of the polymer distribution predicted by the SF model is some 30% less than the experimental value. The distribution of segments within the profile shows that the polymer is attached to the interface by the hydrophobic PPO block, and the PEO blocks behave in a similar manner to a grafted chain a t a solid surface, with a "parabolic" segment density profile. However, the proportion of the polymer in the hexane phase in the experimentally-determined profile is rather higher than in the simulation, so it appears that the SF model here overestimates the degree of segregation of the different blocks. This may be because polydispersity of the polymer has not been adequately modeled. The similarity of the polymer structure at the air/water and hexane/water interfaces is also worthy of comment. Hexane is not a particularly good solvent for PPO, and so the driving force for adsorption at both the air/water and hexane/water (the hydrophobic nature of the PPO block) is similar. Of course, there is no information on the conformation of the polymer on the air side of the air/ water interface, since it has such a low scattering length density. If we assume that the adsorbed amount is the same at both interfaces, which seems probable since the polymer structure in the water is the same, then the same amount of polymer that is found in the hexane must protrude out of the water. The volume fraction of the polymer layer in hexane is only 0.3, and so it would appear to be swelled by the hexane. The data obtained for PVA/Ac at the hexane/water interface suggest immediately that while the polymer is reasonably strongly adsorbed, ita conformation consists almost entirely of trains. The conformation is in stark contrast with that of the Pluronic triblock. While a narrow distribution was expected a t this interface and is consistent with existing measurements, the apparent total absence of loops is a little surprising. However, the observed conformation is consistent with the suggestion of Scholtens,'* who studied similar polymers at the water/butanol interface, that the polymer is pinned to the interface by the acetate groups, since the average distance between acetate groups is about 20 monomers (40 carbon atoms). This is perhaps not sufficient to form large loops. However the "predominantlytrains" conformation is also suggested by SF modeling. Assuming an adsorbed amount of 0.75 mg.m-2 and the interaction parameters given in Table 111 gives the result shown in Figure 13. Comparison with

(17) The Goliad program is deacribedin Mallagh,L. Thesis,University of Briatol, 1989.

(18) Scholtana, B. J. R.; Bijsterboch, B. H. J. Colloid Interface Sci. 1980, 77,162.

X

Table 111. x Values Used in SF Simulations water hexane PEO PPO Val

water hexane PEO PPO VA1 VAc

0.0

8.0 0.5 1.0 0.4 0.8

8.0 0.0 0.8

0.5

0.8 0.0

1.0 0.5 0.45

0.5

0.45

0.0

3.0 3.0

0.5

3.0 0.0 0.4

Vac

3.0 3.0 0.4 0.0

1

Copolymers at the Hexanelwater Interface

Figure 12 shows that the difference in the observed profiles can be rationalized by SF theory. The conformation of the PDMS/PVPO copolymer, where most of the polymer is found on the hexane side of the interface, is rather surprising, given that this copolymer is soluble in water but not in hexane. However its solubility in water is rather low, and the cmc is too small to detect. This is almost certainly a result of the unusual solution conformation of PVPO in water. It has been suggestedl9 that PVPO exhibits extensive intramolecular hydrogen bonding in aqueous solution, leading to a very compact conformation. In contrast with the two other copolymers, the hexane does appear to have a large influence on the segment distribution, as it is a good solvent for one of the polymer blocks. In order to set up the liquid/liquid experiment it was necessary a t first to measure the aidwater interface reflectivity at the lowest angle, and from this measurement it is clear that the adsorbed amount is higher at the hexane/ water interface. From the measurement using D2O and air-contrast-matched hexane we can conclude that the polymer penetrates only weakly into the D20. From the measurement with both solvents air-contrast-matched, however, it appears that the adsorbed amount is substantial. There must, therefore, be a significant amount of PVPO in the hexane phase. It appears that at the interface the insolubility of PDMS in water (and its high solubility in hexane) causes some of the PVPO to be drawn out of the water, where it forms a compact (and probably intramolecularly hydrogen-bonded) structure.

Conclusions Neutron reflection measurements have provided unique information on the conformation of copolymers a t the hexane/water interface. Three different copolymers are shown to adopt very different conformations. A triblock copolymer of PEO/PPO/PEO forms a diffuse layer that is rather thicker than the micellar radius. The shape of the distribution is in qualitative agreement with the meanfield theory of Scheutjens and Fleer. In contrast, a random (19)Tanami, B.;Boshehri, R. Polymer 1978,19,542.

Langmuir, Vol. 9, No. 12, 1993 3537

copolymer of PVA/Ac is shown to adsorb in an all-train conformation. Hexane is a poor solvent for the hydrophobic moieties of both of these polymers, and in the case of the PEO-PPO-PEO does not appear significantly to alter the conformation from the one adopted at the air/ water interface. Hexane does influence the conformation of a diblock copolymer of PDMS-PVPO, for whose hydrophobic block it is a good solvent. The polymer is almost completely drawn out of the water and forms a dense layer between the two phases. Clearly, it is now of great interest to make further neutron reflection studies of these polymers as a function of molecular weight and relative block size. The requirements of the experimental setup do, however, place some serious restrictions on the types of systems that can be studied. The presence of the oil as a thin film condensed from the vapor means that, in order to maintain “equilibrium” conditions, the polymer should be preferentially soluble in water, in order to guarantee rapid diffusion to the interface and a constant concentration in the oil phase. This in turn means that only systems with high HLB numbers can be studied. In addition to this, the need for a liquid that spreads spontaneously on and is less dense than water restricts the oil phase to the lighter alkanes, which are poor solvents for many hydrophobic chemical groupings. This is emphasized by the resulta for the Pluronic copolymer, where it appears that the hexane has little influence on the adsorbed layer structure. The sensitivity of the measurement to hydrogenous polymers is surprisingly high, as it is possible to deuterate both phases. It is not absolutely necessary, therefore, to prepare deuterated versions of molecules of interest. This allowsthe structures of many molecules to be determined that would otherwise not be possible. However, more information could be deduced, in particular on the distributions of different blocks in copolymer systems, if deuterated samples were available.

Acknowledgment. We gratefully acknowledge the help given in the neutron scattering experiments by staff at the Rutherford Appleton Laboratory. We also thank Shell Research and the SERC for financial support.