Neutron reflection from polystyrene adsorbed on silica from cyclohexane at temperatures at and below the theta temperature

Colloids and Surfaces A: Physicochemical and Engineering Aspects, 86 (1994) 185-192 0927-1751/94/$01.00 0 1994 ~ Elsevier Science B.V. All rights reserved.

185

Neutron reflection from polystyrene adsorbed on silica from cyclohexane at temperatures at and below the theta temperature J. McCarneya,

J.R. Lua, R.K. ThornaP*,

A.R. Rennieb

“Physical Chemistry Laboratory, South Parks Road, Oxford OX1 3QZ, UK bCavendish Laboratory, Madingley Road, Cambridge CB3 OHE, UK

(Received

10 October

1993; accepted

23 November

1993)

Abstract Using neutron reflection, we have studied the adsorption of polystyrene onto amorphous silica from a cyclohexane solution with changing temperature, thus varying the solvent conditions, from better than theta conditions to below the cloud point, 7&,, where phase separation occurs. For this system, TEloudwas determined by light scattering to be 21.9”C. We observe a dramatic increase in adsorption near the cloud point. The adsorption is reversible and the rate of establishment of equilibrium is fast. At 22”C, equilibrium is established in 25 min. Our measurements are very sensitive to adsorption at temperatures below 22.5”C where the layer is sufficiently thick and dense to give total reflection. Our reflectivity profiles are not sensitive to the shape of the tail of the decay into solution, but the thickness of the main part of the adsorbed layer is much larger than R,, of the order of 1000 A in comparison with an R, of 160 A. The surface excess at 22.5”C is 2.2kO.8 mg mm’. As the temperature is reduced, further adsorption occurs and at 15°C the surface excess is 6.5 +0.8 mg m-‘. Key words: Adsorption;

Neutron

reflection;

Polystyrene;

Silica; Theta temperature

Introduction The adsorption of polymers at the solid/liquid interface is known to play an important role in phenomena such as the steric stabilization of colloidal dispersions and in technological applications such as lubrication and adhesion [ 11. The adsorption of polymers at interfaces is determined by the substrate-polymer and polymer-solvent interaction parameters (xS and x respectively). If these are comparable, physisorption will result and the adsorption will be very sensitive to changes in the polymer-solvent interaction parameter, which is easily varied by altering the temperature. As the temperature is reduced below the theta temper-

*Corresponding

SSDI

author.

0927-7757(93)02714-P

ature, depending will increasingly and eventually, Tcloud, phase

on molecular weight, the polymer tend to separate from the solution at a well-defined temperature,

separation

occurs

(see Ref. 2 for a

discussion of the nature of this phase separation). Here we describe how neutron reflection was used to study polystyrene adsorbed onto amorphous silica from a cyclohexane solution, as the temperature was lowered from conditions

of good solvent

through the theta point to below Tcloud. In neutron reflection experiments the simplest way of investigating adsorbed species is to measure reflectivity profiles from solutions in which the solvent has the same scattering length density as the solid substrate. This ensures that the reflected signal arises only from the adsorbed material and gives a direct measure of the surface excess. Such

contrast

matching

is readily

achieved

by mixing

isotopically labelled species. The cyclohexane in this experiment was a mixture of fully deuterated and

fully protonated

species

prepared

in such a

way as to match the scattering length density of the amorphous silica, i.e. 3.42 x 10mh A-‘. The thermal expansion of cyclohexane on heating. which has a slight effect on its scattering length density (SLD), has been allowed for in the calculations described below (see Table 1). (Density values for cyclohexane at the specified temperature were calculated using a density-temperature equation

131.) One of the most useful tools in neutron reflectivity is that of isotopic substitution. For samples which differ only in their isotopic content the reflectivity profiles will differ but the same physical model should apply to both [4]. However, the method cannot be employed here because protonated and deuterated polystyrene solutions in cyclohexane have different phase properties [S], thus making direct comparison of models impossible. For example the difference in the theta temperature for the two isotopic species is 10 C [6]. Experimental The solid substrate used in this study was amorphous silica, purchased from Hellma (UK) Ltd. It had dimensions 100 x 50 x 10 mm3 with one of the large faces polished. This polished face was plasma etched for 15 min in oxygen and then for 45 min Table 1 Parameters Temperature

( ‘C)

15 21 22 22.5 40

of model fits to the reflectivity

in argon using a Bio-Rad etcher. The quartz a

3:l

mixture

PT7lOORF

plasma barrel

was then treated by soaking in of concentrated by volume

H,SO,/concentrated HNO, for 18 h, followed by repeated rinsing with pure water (Elgastat, Elga, UK). The sample was a 0.2% solution of deuterated polystyrene in cyclohexane contrast matched to the amorphous silica. The deuterated polystyrene (number-average molecular weight iz1, = 250 000; M,/M,= 1.08 where 121, is the weight-average molecular weight) was purchased from Polymer Laboratories and was used as received. In its anionic polymerization ri-butyllithium was used as an initiator and the terminator was methanol. Deuterated cyclohexane was supplied by MSD Isotopes and protonated cyclohexane (HPLC grade) was purchased from Aldrich. Both were used as received. The reflectivity measurements were made with use of the small-angle scattering spectrometer D17 at the Institute Laue-Langevin. Grenoble [7], adapted for reflection studies [Xl. In reflection experiments the reflectivity is determined as a function of momentum transfer Q, where Q = (471 sin O),‘i.. 0 being the angle of incidence and i the neutron wavelength. On D17 both 0 and i may be varied to effect a change in Q. With an incident angle range of 0.555” and a wavelength range of 8.7730 A, this gives a Q range of 0.00330.13 A-‘. Measurements at low Q are limited by the need to avoid over-illumination and at

profiles

Solvent SLD. corrected (A *1x 10h

Interfacial layer SLD (A ZlX 106

Interfacial laver thickness (A,

Polymer I;l>er SLD (A~*)x 10’

Polymer layer thickness (A,

Experimental decay length (A,

Surface excess I(mg IT-‘)

3.42 3.41 3.4 3.39 3.325

2.3 2.3 1.3 2.3 2.3

35 35 35 35 35

4.4 4.1 4 3.95

250” 250 250 200

900 900 700 500

6.5 4.6 3.3 2.2

* * f + _

0.8 0.8 0.8 0.8

“The errors in the individual thickness of the two layer5 are about i 50%. but the error in the total thickness is such that it cannot be less than about 1000 A, although the exponential decay length could be significantly larger than the tabulated values.

high Q by the background chiefly to the incoherent

radiation scattering

which is due by the solid

the full length of the block without subtraction

of the background

reflection.

the reflectivity

After was

substrate.

obtained

The light scattering experiments used to determine the cloud point were carried out using a

the reflectivity profiles are obtained point by point, the momentum transfer being varied as mentioned

Malvern light scattering goniometer. The cloud point for our system was found to be 21.9 + 0.1 ‘C.

previously. reflectivity

The sample cell used for the reflection measurements has been described in detail elsewhere [S] and consisted of a PTFE trough clamped against the amorphous silica block with two metal securing plates. The trough contained the solution under study and was sealed to the quartz with a nitrile rubber O-ring. It was fully enclosed in an acrylic plastic box fitted with mica windows. The O-ring was cleaned in ethanol while the PTFE trough was soaked for 24 h in a Hellmanex solution and then rinsed thoroughly with pure water. The solution was transferred by pipette into the cell through two filling ports sealed with PTFE stoppers. With the cell in the vertical orientation the solution was in contact with the substrate at all times. Any air bubbles created during transfer rose to the filling ports and did not interfere with the interface. Water from a thermostat was circulated through the body of the securing plates, maintaining the solution at the chosen temperature to within 0.1 K. The sample cell was mounted with the large face vertical in the neutron beam. The collimated beam, defined by a 15 mm circular diaphragm and a 0.2 mm vertical slit 2.4 m apart, entered the substrate block through the 10 mm thick end face and was reflected on striking the solid/liquid interface. The reflected beam emerged through the opposite end face. The intensity of the reflected beam was measured on a position-sensitive detector giving an integrated peak intensity I,. A value for the background scattering was determined for each measured point of reflection using an area of the multidetector around the reflected peak. Neutron beams are attenuated on passing through the amorphous silica, primarily by wide-angle scattering, but there is also some absorption. To allow for this, the intensity of the attenuated beam was measured at each wavelength after passing through

but the most common

by taking

the ratio of I, and I,. On D17

Information can be extracted profiles in a number of different method

from ways

is that of comparing

calculated model profiles, obtained using the optical matrix method [9,10], with the experimental data 141. Results and discussion Before studying adsorption it is important to characterize the roughness of the solid/liquid interface and to ensure that it is clean. In neutron reflectivity this is done by measuring the reflectivity from the interface in a number of different solvent contrasts. Ideally for a clean surface and for two homogeneous media of the same scattering length density, i.e. when the cyclohexane contrast matched to amorphous silica, no reflection is expected. However, in practice some reflection is usually observed [S]. For our system we observed such a reflected signal indicating that there is some type of interfacial layer other than a sharp interface. Since we have measured only three reflectivity profiles with pure solvents we cannot determine unambiguously the nature of this interfacial layer but it is likely that it is due primarily to surface roughness with some water adsorption. The best model for the interfacial layer, i.e. one that is consistent with the three solvent profiles, is a layer 35 A thick composed of 30% amorphous silica, 30% water and 40% cyclohexane. Other related, but slightly different, models of the interfacial region also fit the data, and the interface could be more complex than that represented by this threecomponent layer. Figure 1 shows the fit of this model to the three reflectivity profiles. The fit is adequate for two of the three profiles at Q values above about 0.005 A-‘. Below this the reflectivity from deuterated cyclohexane falls significantly below unity, almost certainly because of slight

J. McCarnev

et al./Colloids

Surfaces A: Ph~sicochrm

the interface

Eng. Asprcts 86 (I 994) 185-192

and some sort of decay in polymer

concentration on moving away from the surface into the solution, Our profiles are, however, insensitive to the shape of this decay, i.e. Gaussian, linear or exponential

decays fit equally

well.

Figure 2 shows reflectivity profiles at a number of temperatures and the parameters of the model

Q/A’ from amorphous profiles the 1. Reflectivity silica/cyclohexane interface obtained for three solvent contrasts with model fits. The model fits (continuous lines) assume a layer 35 A thick at the interface composed of approximately 30% water, approximately 30% amorphous silica and approximately 40% cyclohexane. The three contrasts are (0) cyclohexand (C) cyclohexane contrast ane-d,,, (0) cyclohexane-h,, matched to amorphous silica. Error bars are largest when the signal is low and sample error bars are drawn where they are significant.

over-illumination of the surface. The poor fit below about 0.01 A-’ for the matched cyclohexane suggests that the interface may be more complex than the model assumed; for example there may be a “gel” layer [S]. Attempts to fit this profile more accurately are not worthwhile, partly because any large-scale structure included would be ambiguous without independent data and partly because the inclusion of such a layer has few consequences for the fitting of the adsorbed polymer layer, which is the primary interest. Models of the adsorbed polymer profiles must take into account the structure of the bare interface. To be able to draw useful conclusions about the adsorbed polymer it is important that the fits of the model polymer profile are independent of the range of structures that will account for the reflectivity from the bare interface. At temperatures below 22.5”C there is indeed total reflection and the model fit is independent of this slight variation in the modelling of the bare interface. Those profiles which show total reflection (as shown in Fig. 2) require a model with a polymer-rich block next to

fits (with an exponential decay) are shown in Table 1. We see from the table that, at 22.5”C, just above the cloud point, there is significant adsorption and as the temperature is lowered more and more, adsorption occurs. At higher temperatures the fit is dependent on the model we choose for the interface, so this prevents us making any firm above 22.5” C. conclusions at temperatures However, we can say that within the accuracy of our experiment there is little adsorption at 4O”C, i.e. in the vicinity of the theta temperature. This is demonstrated clearly by comparing the “bare” surface reflectivity profile (Fig. 1) with the observed reflectivity from the polymer solution which has been heated to 40°C (Fig. 3). The continuous line in each figure is calculated for the same structural model. The accuracy of the measured coverages is good for those temperatures for which we have quoted a value and is little affected by changes in the model of the bare surface used to fit the data. However, the uncertainty about the structure of the bare surface leads to considerable uncertainty when the coverage is approximately zero. The adsorption

was found to be reversible

on a

time scale of about 30 min. Figure 4 shows the change in reflectivity at a Q value of 0.0044 A-’ after a change in temperature from 25”C, where there appears to be little adsorption, to 22°C. Note that the reflectivity varies as the square of the scattering length density difference between the layer and the substrate. The time was taken to be zero when the temperature of the cell had reached 22’C, so the equilibration time may be a little longer than the 20 min shown in Fig. 4. The existence of a significant critical angle in profiles below 22S”C means that we cannot be very precise about the exact size and shape of the concentration profile because the value of the

J. McCarney

et al.lColloids

Surfaces

A: Physicochem.

Eng. Aspects 86 (1994) 185-192

(a)

189

-9

(b)

-?

(d)

Fig. 2. Reflectivity profiles of polystyreneecyclohexanesilica at different temperatures: (a) T= 15°C; (b) T=21”C; (c) T=22”C; (d) T=22,5”C. The fits of the model described in the text with the parameters of Table 1 are drawn as continuous lines. Error bars caused by statistical inaccuracy have only been drawn at the lowest reflectivity; elsewhere they are insignificant.

011 , *, 0

Y)

20

30

40

Time/minutes 0.02

0.04

0.06

a/A-’ Fig. 3. Reflectivity profile at 40°C compared ity calculated for the bare interface.

with the reflectiv-

Fig. 4. The kinetics of adsorption of thepolymer as shown by the change in reflectivity at Q=O.O044 A-’ after a change in temperature from 25 to 22°C. Time t=O is the time at which the temperature had reached 22°C.

critical angle only depends on composition. not on dimensions. If the boundaries of the layer are

in pure solvent.

sharp, interference mine the thickness,

fringes may be used to deterbut a slow decay on one side

above, the rate at which the adsorption process came to equilibrium was rapid, of the order of a

of the interface will damp out any fringes. What is certain is that the thickness of the adsorbed polymer layer is large, of the order of 1000 A (several

few minutes at 34.5 C. Cosgrove et al. [ 141 and Satija et al. [ 151 have used neutron reflectivity to

times the unperturbed

onto solid substrates. Cosgrove et al. found that the polystyrene physically adsorbed to mica (M, =

radius

of gyration

(approxi-

mately 160 A) of the polymer), even at 22.5 C, only 0.6’C above the cloud point, although the layer rapidly becomes more tenuous as the temperature increases.

also

study

found

that.

adsorbed

is approximately in slight

polystyrenes

2 mg m ‘. They

contradiction

from

to the

cyclohexane

In comparing our results with those of other workers we note that a wide diversity of behaviour has been observed for the system polystyreneecyclohexaneesolid and this undoubtedly stems from variations in the nature of the solid surface. We have found in reflectivity experiments on surfactants adsorbed on solid surfaces that there may be a wide variation of the amount adsorbed even on surfaces that are nominally identical 181. The surfaces of quartz, silica and the oxide layer on silicon are very sensitive to the way they are prepared [ 1l] and this may be particularly important when hydrophobic species are adsorbed from hydrocarbon solution onto the hydrophilic silica surface. Granick and Johnson [ 121 have used Fourier transform infrared spectroscopy in the attenuated

49500) had a monotonically decaying profile at room temperature (well above Tcloud). They found the r.m.s. thickness to be of the order of 100 A. which is similar to the unperturbed radius of gyration of 70 A. Satija et al. [ 151 studied the adsorption of carboxylic-acid-terminated polystyrene at the cyclohexane,‘SiO, interface at 21 ‘C. They found an adsorbed amount similar to ours, 5.6 mg nr~’ for deuterated polystyrene in protonated cyclohexane and 4.1 mg ln-’ for ordinary polystyrene in deuterated cyclohexane. Some other features observed by Satija et al. are also in qualitative agreement with our results. For example they observed a layer sufficiently dense to give an interference fringe with a thickness of about 100 A. However, the temperature of the experiment was only approximate and Satija et al. give no details of whether the cloud point was determined fat their system. Their polymer molecular weight was much smaller than that used in our study (M,= 14000) so no quantitative comparison is possible. It should be noted that for a solution of such a small molecular weight polymer. even at 2 1 C they

total reflection mode to study polystyrene (M,= 355000) adsorption on oxidized silicon from a cyclohexane solution. They found significant adsorption at temperatures above the cloud point and even in the vicinity of the theta point the surface excess was significant, approximately 3 mg m -‘. Terashima et al. [ 131 have used a microbalante technique to study the adsorption of polystyrene (M, = 107 000) from cyclohexane to mica and they also found significant adsorption at the theta temperature. They found that the amount of strongly adhering polymer, i.e. that which is not easily removed by rinsing and prolonged soaking

are working well above the cloud point of their system. A further observation was that the adsorption was sensitive to the nature of the terminating group and this might also invalidate any comparison between their work and ours. The presence of the carboxylic acid group will mean that the end binding of the polymer will be strong and ahnost equivalent to grafting the polymer on to the surface. The only result that is in clear disagreement with our results is that no change in the film thickness of the polymer was observed on increasing the temperature to 42 C. However, it is not clear how the “bare” interface was characterized

Discussion

and conclusions

J. McCarney

et al./Colloids

Surfaces A: Physicochem.

for their system. We also obtained but that could

be attributed

from the bare interface, agreement

ments performed rene system

to reflection

as was shown in Fig. 3. and

between

sometimes

even

the different experi-

on the solid-cyclohexane-polysty-

illustrates

the experimental larly

the condition

clear

from

our

a signal at 40°C

entirely

The lack of quantitative, qualitative,

Eng. Aspects 86 (1994)

the difficulties

conditions

of defining

accurately,

of the surface experiments

is

we have

a

surface that has a lower affinity for the polystyrene than any of the others so far studied, with the result that we have only weak physisorption polymer.

of the polymer near its cloud point. Wetting theories predict that the solid surface may be able to ble in the bulk, persist above

layer

Johner

layer should blobs

close-packed

may

and Joanny

point,

have the form of a

on the surface

solution

temperature

the dimensions

theoretical

8 9 10

with

a

predictions.

range

above

12

even

13 14

the cloud

are quite different Thus,

11

of blobs next to it.

Although our experiments seem to suggest the persistence of a layer of the polymer-rich phase over a short

6 I

that, in the case of a polymer

of Gaussian

uniform

case multilayers

the cloud point.

layer, the wetting

4

phase that would be unsta-

in which

[2] have predicted

1 2 3

5

This is the type of system that is interes-

a microscopic

cannot be very precise about the exact size and shape of the concentration profile, what is certain is that the thickness of the layer is of an order of magnitude larger (1000 A) than expected for Gaussian blobs, which will be of the order of the radius of gyration (160 A).

of the

ting from the point of view of the wetting behaviour

stabilize

from the

though

191

References

particu-

itself. What

is that

185-192

we

15

M. Rosoff, Phys. Methods Macromol. Chem., 1 (1967) 1. A. Johner and J.F. Joanny, J. Phys. II, 1 (1991) 181. International Critical Tables, Vol. 3, McGraw-Hill, 1928, p. 27-29. E.A. Simister, E.M. Lee, R.K. Thomas and J. Penfold, J. Phys. Chem., 96 (1992) 1373. T. Engels, L. Belkoura and D. Woermann. Ber. Bunsenges. Phys. Chem., 88 (1984) 635. C. Strazielle and H. Benoit, Macromolecules, 8 (1975) 203. B. Maier (Ed.), Neutron Facilities at the High Flux Reactor, lnstitut Laue-Langevin, Grenoble, 1990. D.C. McDermott, J.R. Lu, E.M. Lee, R.K. Thomas and A.R. Rennie, Langmuir, 8 (1992) 1204. J. Lekner, Theory of Reflection, Martinus Nijhoff, Dordrecht, 1987. OS. Heavens, Optical Properties of Thin Films, Butterworth. London, 1955. R.K. Iler, The Chemistry of Silica, Wiley-Interscience, New York, 1979. S. Granick and H.E. Johnson, Macromolecules, 24 (1991) 3023. H. Terashima, J. Klein and P.F. Luckham, Symposium on Adsorption from Solution, Bristol, 1982. T. Cosgrove. T.G. Heath, J.S. Phipps and R.M. Richardson, Macromolecules, 24 (1991) 94. SK. Sati_ja, J.F. Ankner, C.F. Majkrzak, T. Mansfield, G. Beaucage, R.S. Stain, D.R. Iyengar, T.J. McCarthy and R.J. Composto, Abst. Pap. Am. Chem. Sot., 204 (1992) 88.

192

Discussion Speaker: A.R. Rennie Questioner: P. Luckham Q. At the end of your talk, you said that the polymer

layer thickness

you had a thickness of around 200 8. Could you explain? A. The model used to fit the data consists of one layer of uniform a further

region modelled

as an exponential

decay to the bulk solution

was about density

1000 A. On your table

of polymer

and solvent

and

density.

Speaker: A.R. Rennie Questioner: J.S. Phipps Q. When you lower the temperature below the cloud point of the solution, does the small-angle scattering from the solution interfere with the neutron reflectivity signal and thus make interpretation more difficult? A. In general it is necessary to account for any small-angle scattering that occurs in the forward direction close to the specular beam. These experiments were performed on D17 at the ILL, Grenoble, which is equipped with a two-dimensional area detector. Background scattering was subtracted by considering the intensity observed in areas adjacent to the specular beam. No particular increase in small-angle scattering was observed in these experiments - after the cloud point was reached the aggregates appear to be fairly large and to settle rapidly to the bottom of the cell there is not a high concentration close to the surface under investigation which is held vertically. Speaker: A.R. Rennie Questioner: R.A.L. Jones Q. Another group which has attempted to use neutron reflection to probe the growth of a wetting layer as a binary mixture is quenched below coexistence is L. Norton and E. Kramer. They found that while the experiment was sensitive to the growth of the wetting layer while the width of the layer was less than the correlation length which governs the width of the interface between the wetting phase and the bulb, as soon as the wetting phase becomes depressed from the wall there is no sensitivity to the wetting layer thickness. Can you be confident of your value of 250 A for the thickness of your layers‘? A. I hope that I was careful in presenting these results not to be over confident about the accuracy. The presence of a thick layer is evident from the critical angle that is observed. The model used to fit the data was of a relatively concentrated layer followed by an exponential decay in segment density. While the exact form of the profile is not accurately determined, we cannot find models that fit the slope of these reflection profiles without a long decay. It is probably difficult to distinguish between the region of uniform concentration and the decaying profile. There are limits imposed by the range of data from the critical momentum to the maximum momentum transfer that limit the range of distances that can be probed.