Langmuir 1987, 3, 1131-1135
scanning ~alorimetry,'~ and radi01abeling.l~ We showed that the terminal D nodules undergo a t the interface a denaturation similar to the one identified in solution around 60 OC.lO-l2 In the present work, we analyzed the structural alterations of a fibrinogen layer adsorbed at the surface of pure hydrophilic silica, using the technique of angular scanning reflectometry. It has been observed that the occurrence of a molecular thermal denaturation of the adsorbed molecules induces opposite effects a t low and high surface concentrations. Fibrinogen molecules dispersed a t the surface show important desorption upon denaturation, while those remaining adsorbed adopt a conformation leading to an increased layer thickness. For a dense adsorbed layer, which presents a more complex structure, denaturation leads to
1131
a kind of superficial polymerization, which strongly stabilizes the layer. This description correlates well with a thermal desorption study performed with radiolabeled m01ecules.l~It appears therefore that SAR is a technique well adapted to study structural changes within an adsorbed layer. It would be of great potential interest to follow continuously, as a function of time, these structural changes, but this is not yet possible with our present system.
Acknowledgment. The authors are grateful for financial support from the CNRS under the joint projects "GRECO Polymbres HBmocompatibles" and "ATP MaSriaux". The technical assistance of G. Maennel and F. Woehl is also acknowledged.
Reflectometry as a Technique To Study the Adsorption of Human Fibrinogen at the Silica/Solution Interface? P. Schaaf, P. Dejardin, and A. Schmitte Znstitut Charles Sadron (CRM-EAHP), 67083 Strasbourg Cedex, France Received March 3,1987. I n Final Form: June 1, 1987 In this paper, we show that the technique of reflectometry,applied to the reflection of a "p" wave around the Brewster angle, provides information similar to ellipsometry to characterize a layer of adsorbed macromolecules. Application of this method to study adsorption from solutions of human fibrinogen at the surface of pure silica leads to the following observations: (1) When the external equilibrium concentration is raised, but remains low, a layer of almost constant optical thickness builds up, while the concentration (w/w), this first layer within the layer increases. (2) At an external crossover concentration close to appears to be saturated. (3) Above this concentration threshold, an increase in the equilibrium concentration induces structural changes within the adsorbed layer, which remain to be completely explained. These observations correlate with former results obtained by using a hydrodynamic flow technique.
Introduction When dissolved synthetic or biological macromolecules are trapped in the interfacial force field of solid/liquid interfaces, they may undergo conformational transitions affecting primarily their overall size and shape. In the case of biopolymers, properties related to local interactions may in turn be altered, leading to enhancement or inhibition of various biological responses. It is, therefore, of importance to characterize the structure of a macromolecular adsorbed layer, that is, to know at least its molecular lateral (mean area occupied by one molecule) and orthogonal (mean layer thickness) dimensions a t the surface. Because of its high sensitivity, the most popular method used to determine the surface concentration r, and thus the mean area per adsarbed molecule, is /3 or y radiolabeling. Various hydrodynamic techniques have been applied to measure the thickness of interfacial mono- or multilayers.l4 Ellipsometry is also known to be a powerful and sensitive tool in surface science, because it provides information on both adsorbed amounts and layer thicknesses or, in other words, on the zeroth and first moments of the monomer distribution function, in a direction normal to the i n t e r f a ~ e . ~ ~We ' recently developed a simplified version of this optical technique, namely, reflectometry? 'This paper is based in part on a presentation to the Division of Colloid and Surface Chemistry, 192nd National Meeting of the American Chemical Society, Anaheim, CA, Sept 7-12, 1986.
0743-7463/87/2403-ll31$01.50/0
and showed that the same physical parameters, the mean layer thickness and the mean refractive index within the layer, may be obtained; therefore, the adsorbed amount may also be evaluated. In the present work, we used this new approach to investigate variations in the structure and composition of an adsorbed fibrinogen layer, when the interfacial concentration was progressively increased, a t the surface of optically polished silica. Previous work related to the adsorption behavior of fibrinogen at various model interfaces led us to identify a t least two adsorption r e g i m e ~ . ~ J ~ We postulated that a t low interfacial concentration the (1) Rowland, F. W.; Eirich, F. R. J. Polym. Sci., Polym. Chem. Ed. 1966,4, 2401. (2) Silberberg, A. In Polymer Adsorption and Dispersion Stability; Goddard,E. D., Vincent, B., Eds.; ACS Symposium Series 240; American Chemical Society: Washington DC, 1984; p 161. (3) Silberberg, A,; Klein, J. Biorheology 1981, 18, 589. (4) Pefferkorn, E.; Schmitt, A.; Varoqui, R. Biopolymers 1982, 21, 1451. ( 5 ) de Baillou, N.; DBjardin, P.; Schmitt, A.; Brash, J. L. J. Colloid Interface Sci. 1984, 100, 167. (6) Ellipsometry and Polarized Light; Azzam, R. M. A., Bashara, N., Eds.; North-Holland Amsterdam, 1977. (7) Charmet, J. C.; de Gennes, P. G. J. Opt. SOC.Am. 1983,73,1777. (8)Schaaf, P.; D€jardin,P.; Schmitt, A. Rev. Phys. Appl. 1986,21,741. (9) Schmitt, A.; Varoqui, R.; Uniyal, S.; Brash, J. L.; Pusineri, C. J. Colloid Interface Sci. 1983, 92, 25. (10) de Baillou, N.;Voegel, J. C.; Schmitt, A. Colloids Surfaces 1985, 16, 271.
0 1987 American Chemical Society
1132 Langmuir, Vol. 3, No. 6,1987
Schaaf et al.
Cylinder
Y
e
-n = l
C5J
'ra ns
0'
I ,
20
I
I
I
LO
60
80
time
5
Imn)
Figure 2. Variation of the reflectivity coefficient of a p wave at the Brewster angle OB, as a function of the adsorption time, for silica/solution interfaces prepared in different ways. Treatments subsequent to the one described in the text are as follows: (a) surface dried at 50 "C for 3 h (LO= 22 nm, n = 1.358), (b) surface rinsed with acetone and dried at 110 "C for 4 h, (c) experiment performed without any drying (Lo= 23 nm, n = 1.342), (d) surface rinsed with acetone and dried for 1min at room temperature (Lo = 13 nm, n = 1.348).
4 2 20
4230
4240
4250
42 6 0
Figure 1. (a, top) Schematic experimental system. The lower hemicylinder is made of silica. F represents the thermostated liquid phase, while L is a focusing lens. Other parts are described in the text. (b, bottom) Variation of the reflectivity coefficient of a p wave, It,(@), with the incidence angle 8, for a Fresnel interface (silica/buffer solution) ( 0 )and for an interface containing an adsorbed fibrinogen layer (equilibrium concentration of 5 X w t % ) (+).
elongated fibrinogen molecules adsorb flat on the surface. Once the layer of molecules attached "sideon" is saturated, the structure of the monolayer changes with additional adsorption: the hydrodynamic thickness increases sign i f i ~ a n t l y while ,~ the affinity for the surface appears to decrease. Hence, our investigation was first aimed to obtain complementary information with respect to these farmer observations. The first part of the paper briefly describes the technique of reflectometry and the experimental procedures used. Whether the way of preparing a clean silicajsolution interface plays a role in the adsorption results is then investigated. Finally, the main body of the paper is devoted to a detailed study of what may be called a "reflectometry isotherm".
Experimental Section Protein and Buffer Solutions. The buffer solution used throughout this work contained 0.05 M Tris (hydroxyaminomethane) and 0.15 M NaCl, with the pH adjusted to 7.35 (physiological conditions) with 0.5 M HC1. Human fibrinogen, grade L, was purchased from KAI3I and handled as indicated by the manufacturer. Each solution was centrifuged at 17300g for 45 min before introduction into the adsorption cell. Description of the Technique. A schematic representation of our experimental system is shown in Figure la. The light source is a 5-mW He-Ne laser (A = 6328 A), producing almost linearly polarized light. Polarizers P1 and P2 are aimed to select the appropriate polarization plane for the incident and reflected
light beams. In the present work, polarization is parallel to the incidence plane, so that we measure the reflectivity of a p wave. The experimental cell is made of a hemicylindrical silica block, optically polished at the plane interface S. A microscopic pinhole PH of 100-pm diameter is fixed in front of the photomultiplier PM and selects a well-defined reflected angle, with an angular definition better than 0.01". Small rotations around the Brewster angle BB are obtained with microcontrolled rotation (Mrot) and translation (Mtrans) devices. Details may be found in a previous publication? Temperature control appears to be important: the system is located within a thermostated room. At the beginning of an experiment, the cell is filed with buffer solution. The refractive index within the cell then varies abruptly if one moves from the buffer solution phase to the solid silica phase. Thus, the reflectivity coefficient R, of this interface (ratio between reflected and incident intensities for a p wave) is given by the well-known Fresnel formula.'l In Figure lb, the measured reflection intensities are plotted against the incidence angle, around the Brewster angle BB = 42.43" (refractive indexes are nl = 1.332 for the buffer solution, and no = 1.457 for silica, at 25 "C). It is seen that the best fit with a Fresnel function (dotted line) is very satisfactory. Once the buffer solution is replaced by a fibrinogen solution, protein adsorption proceeds at the solid/liquid interface and the reflectivity around 8B increases, as observed in Figure lb, where the data were collected after 2 h, that is, once equilibrium was almost reached. We notice that the increase in reflectivity depends on the incidence angle: the experimental curve after adsorption is not simply a vertical shift of the Fresnel curve; since our system is not yet automated, it is not possible to determine the function RP(8,t)at any adsorption time. We thus proceed in two steps. Variation of the reflectivity with time is first measured at a fied angle, very close to the Brewster angle, and Figure 2 presents typical curves showing these variations of R,(OB,t). The conditions under which the data were obtained will be discussed in the next section. What should be pointed out is that angular variations of the reflected intensity can only be recorded once the reflectivity displays a quasi-plateau in its time variation, since to perform a set of measurements takes about 50 min. To determine the parameters characterizing the adsorbed protein layer, one has to postulate a given refractive index profile in a direction normal to the interface. The simplest approach is to consider a homogeneous adsorbed layer, with constant "optical" thickness LO and mean refractive index n. For such a simple situation, an analytical expression of the function R,(O) (11)Landau, L.;Lifchitz, E. Electrodymmique des milieux continw; Editions Mir: Moscow, 1969;p 363.
Langmuir, Vol. 3, No. 6, 1987 1133
Adsorption of Human Fibrinogen Studied by Reflectometry
6oo
coo
tc
1.345 1.364
/L“
1.3L3
1.342
A loq-2conc l % w / w-1)
-5 -L -3 Figure 3. Variation of the mean optical thickness Lo (+) and
of the hydrodynamic thickness LH (X) of an adsorbed fibrmogen layer, as a function of the equilibrium solution concentration,in decimal logarithmic scale. is available.” With a nonlinear least-squares regression procedure, a best fit of the experimental R&B) curve is achieved, and thus
the parameters LOand n are determined. Once these parameters are known, it is possible to compute the product (n-nl)Lo, which is, if the refractive index increment dn/dc remains constant, proportional to the amount of protein adsorbed.12 We show in the Appendix that Rp($) is, to a f i t approximation, proportional to [(n-nl)Lo12,a result which is of special interest in the study of adsorption kinetics. As shown in the Appendix, this approximation is only accurate at the beginning of the fibrinogen adsorption process. Preparation of Clean Interfaces. The experimental cell was designed to avoid both tedious angular corrections (incident and reflected beams are normal to the silica/air interface) and significant diffusion by the solution. However, the cell has to be cleaned after each experiment with a concentrated sulfochromic acid solution remaining in contact with the interface for 15 h at room temperature. Afterwards, we proceeded by thoroughly rinsing the cell with deionized (super Q Millipore) and filtered (0.2-pm pore diameter) water. We experimented with several subsequent treatments. Details are given in the caption of Figure 2. What appears clearly is that the recorded optical signal depends significantly on the method of preparation of the surface. We observed that the highest interfacial concentrations are obtained with the surfaces that have been dried and heated (curves a and b, Figure 2) and which display, therefore, some hydrophobic properties. Surfaces that have been quickly dried at room temperature (curve d) or not dried (curve c) adsorb lower amounts, probably because of their dominant hydrophilic character.13 Finally, we also noticed that surfaces of type c lead to the most reproducible results. It is the reason why this latter cleaning procedure was used to obtain the following results.
Results and Discussion We performed a set of experiments in which a fibrinogen solution was put into contact for approximately 2 h with a clean silica surface, preequilibrated with buffer solution. Protein concentrations varied between 5 X lo4% and 5 X (w/w). Once adsorption equilibrium was almost attained, the reflected beam intensities around the Brewster angle were quickly measured and the parameters Lo and n calculated. Figures 3 and 4 illustrate the variation of these parameters as a function of the bulk solution concentration (logarithmic scale). It appears that a bulk concentration close to separates, as already noticed with several (12)De Feijter, J. A.; Benjamim, J.; Veer, F.A. Biopolymers 1978,17, 1759. (13) J6neaon, U.;Malmqvist, M.;Mnnberg, I. J. Colloid Interface Sci. 1985,103, 360.
t
-2.
--- --+ I
I o q conc 1 %
1.339 I
-3 0
I
-2.5
I
-2.0
I
wiwl c
-1 5
Figure 4. Variation of the mean refractive index n within the adsorbed fibrinogen layer as a function of the equilibrium solution concentration, in decimal logarithmic scale.
other i n t e r f a c e ~ , ~ two J ~ adsorption domains. At low concentrations, the thickness of the layer remains constant, within experimental error, and close to 12 nm (Figure 3). This probably corresponds to the rod-shaped molecules lying flat on the adsorbing surface: their lateral dimension, in the dry state, is close to 9 nm, but the excess thickness may be due to a contribution of the a chains, which protrude partly into the solution. As the external concentration is raised, the layer of constant thickness fills up progressively with adsorbed molecules, and this shows up through the gradual increase of the mean refractive index within the layer (Figure 4). The highest excess volume concentration within the adsorbed layer may be estimated through eq 1: where dn/dc represents the refractive index increment of fibrinogen dissolved in the buffer solution. It has been shown by De Feijter et all2 that this increment remains constant up to the high concentrations existing within interfacial layers. In addition, it does not vary greatly with the kind of protein. Therefore, we adopt for fibrinogen at 25 “C the mean value dn/dc = 0.18 g/cm3. This leads, according to eq 1,to a concentration within the layer (the bulk concentration being negligible) of approximately 7 X g/cm3. It is possible to estimate the concentration within a monolayer of fibrinogen molecules attached side-on at a surface and packed in parallel arrays: one finds 0.20 g/cm3. Hence, at the bulk crossover concentration of some “vacancies” apparently remain a t the surface, but they do not allow additional lateral adsorption of the elongated fibrinogen molecules, because of the surface exclusion effect. Above this concentration threshold, the thickness of the “homogeneous” layer increases steadily with the external concentration, while the mean refractive index, and thus the mean concentration within the layer, decreases. These data suggest that the structure of the adsorbed layer becomes more complex, and the validity of our simplified model should therefore be questioned. We might conjecture the building up of a second layer, adsorbing on top of the “saturated” layer which has just been described; it is also possible that additional adsorption proceeds with molecules having their long axis tilted in a direction which is no longer parallel to the interface. Anyway, a t higher bulk concentrations, the model of a homogeneous layer provides only a crude first-order approximation to describe what happens in the interfacial layer. It is in principle possible to introduce some interfacial concentration profile
1134 Langmuir, Vol. 3, No. 6,1987
Schaaf et al. l O ’ x R,(B,i
I
,150A
, /
/
ciO 00
*o
2c
30
i c
1O’X
c I%w/w I
50
Figure 5. Variation of the parameter (n- nl)Lo,which is proportional to the amount of protein adsorbed, assuming a constant refractive index increment, as a function of the equilibrium solution concentration.
c(z ) ( z being the distance to the interface) involving more than two structural parameters and to calculate the corresponding refle~tivities.’~J~ Nevertheless, to assess the validity of such a model, it is then necessary to explore with accuracy a wide angular domain in the vicinity of the Brewster angle. This is not yet possible with our experimental system. I t is instructive to compare the layer thickness Lo measured by the present optical technique with the socalled hydrodynamic thickness LW5Let us just recall that LH was calculated by measuring viscous flow, before and after adsorption, in a porous filter made of fritted Pyrex glass beads. Cleaning and preparation of the adsorbing surface were not exactly the same as in the present work.5 Comparison of the data appearing in Figure 3 suggests the following comments: (1) LH is systematically higher than LO, a result which is qualitatively well understood. It has been shown theoreticallylaJ7that the hydrodynamic technique measures more or less the cutoff distance of the interfacial concentration profile c(z), while optical methods measure the zeroth and first-order moments of this distribution function and are therefore mainly sensitive to the dense part of the concentration profile. The presence of the dangling a chains has accordingly a more pronounced effect on the hydrodynamic thickness. (2) Below the concentration crossover, LH also reaches a value, characteristic of B layer which filt4 up progressively, with molecules adsorbed in a direction parallel to the surface.18 (3) A t the concentration threshold, we observe a sharp increase in the LH value, which shows once more the high sensitivity of the technique to detect structural changes a t the interface, without reference to any model. It is possible to compute, from our results, the interfacial concentration r, using the relation r = Lo(n - nl)/(dn/dc) (2)
00
5.00
10.00 15.00 1 0i ~ ~ ~~ i *
20.00
Figure 6. Variation of the reflectivity coefficient at the Brewster angle, R,(BB),versus the square of the refractive index difference An between the layer and the solvent, for different layer thicknesses LO: (-) values calculated without any approximation; (- - -) values calculated using eq A14 for thickness Lo = 40 8, (bottom) and Lo = 150 A (top).
with macromolecular species. Scattering in the experimental values is apparent. The pseudoplateau value is much lower than the one measured on glass (I’ = 0.8 pg/cm2) by using a radiolabeling t e ~ h n i q u e . ~This is, however, not surprising in view of the results obtained with different methods of surface preparation. It has already been shown by Jonsson et al.13 that the chemical modification of silica surfaces induces pronounced modulation of amounts of adsorbed proteins, measured by ellipsometry and radiolabeling.
Conclusion To describe structural or kinetic properties related to the adsorption of macromolecules, the technique of ellipsometry has been widely used during the last few years.6J2J3J9,21We wished to show that the analysis of the reflectivity of a p wave around the Brewster angle provides the same information as ellipsometry. We investigated how the two parameters characterizing an adsorbed fibrinogen layer at the surface of silica (namely, the mean refractive index within the layer and the layer thickness) vary when the adsorbed amount increases steadily along the adsorption isotherm. The results obtained show a “bimodal” adsorption mechanism, which is in good agreement with former observations in our laboratory5i9J0concerning the adsorption of fibrinogen on various types of interfaces. It should be mentioned that a “kink” at about half saturation has indeed been observed by several authors in protein adsorption isotherms.22 Adsorbances calculated from our data are, however, not precise enough to assess the existence of such kinks. Acknowledgment. We are grateful for financial support by the CNRS under the joint projects “GRECO PolymGres HBmocompatibles” and “ATP Mat6riaux”. The help of Professor Benoit is also acknowledged.
It may be shown that this equation remains valid for inhomogeneous surface layers, Lo and n being then the mean thickness and refractive index within the layer.12 In Figure 5, the isotherm presents the shape usually observed
Appendix When two media containing respectively the incident and refracted light beams are separated by an interface
(14) AbelBs, F.Ann. Phys. 1950,5, 596. (15) Schaaf, P.; DBjardin, P.; Schmitt, A. Rev. Phys. Appl. 1985,20, 631. (16) Varoqui, R.;DBjardin, P. J. Chem. Phys. 1977, 66, 4395. (17) De Gennes, P. G. Macromolecules 1981, i4, 1637. (18) Priel, Z.;Silberberg, A. J. Polym. Sci. 1978, 16, 1917.
(19) Corsel, J. W.; Willems, G. M.; Kop, J. M. M.; Cuypers, P. A.; Hermens, W. T. J. Colloid Interface Sci. 1986, 111, 544. (20) Sundgren, J. E.; Bodo, P.; Ivarsson, B.; Lundstrom, L. J. Colloid Interface Sci. 1986, 113, 530. (21) Lee, J. J.; Fuller, G. G. J. Colloid Interface Sci. 1985, 103, 569. (22) Andrade, J. D.; Hlady, V. In Aduances in Polymer Science; Springer-Verlag: Berlin, Heidelberg, 1986; Vol. 79, p 1.
Langmuir, Vol. 3, No. 6, 1987 1135
Adsorption of Human Fibrinogen Studied by Reflectometry characterized by a one-dimensional index profile n(z),it is possible to calculate the reflectivity coefficient of this interface, by using the so-called “method of integrals“ developed by Abe16s.14 Let us consider a p wave a t the Brewster angle OB To relate the amplitudes of the electric and magnetic fields, in the incidence and refraction media, respectively, Abelgs introduces a transfer matrix Mp(z):
phase), it follows that, for z
>0
+
e(z) = n t 2n,An(z) (A7) Let us call 2, the cutoff to the index profile, that is
An(%)= 0 for z L 2, (A8) The first-order moment q of the function An(%)is written 4 = r z m a n ( z )dz
(A9
d 0
With the foregoing definitions and hypotheses, it is easily shown that
F’(Z,)
= -Z,(nl2 - a2)- 2 ( a 2 q / n l )
-koeocnl(nlZ, + 2q) (A10) Whence, the matrix M,(Z,) is expressed simply by
4(zm)
ko, c, and eo are, in vacuo, the modulus of the wave vector, the velocity of the electromagnetic wave, and the dielectric permittivity, respectively, &) is the relative dielectric permittivity a t a point of abscissa z within the interface: it is related to the refractive index n ( z ) by the relation c,(z) = n2(z)
(A2)
F’(z) and @(z) are the derivatives, with respect to z, of the functions F ( z ) and &z): m
F(z) = Cko2jFj(z) j=O
;1
E OC
1
1
i k ~ ~ ~ c n l ( n+l 2Zqm) r m ( n 1n2l z- a’)
22) n13 1
(All)
Let Uo+,Uo-,and Ul+be the amplitudes of the magnetic induction, far from the interface, for the incident, reflected, and refracted beams, respectively. They are related by the matricial expression
(D
These zeroth, first-order, etc., components are respectively defined by the following: F ~ ( Z=) 1
A t the Brewster angle, we have the particular relations nl no nl tan Bo = -. -= (A131 no’ cos 8, cos 6, The reflectivity coefficient for a p wave is therefore expressed by
If we assume that the refractive index increment remains constant up to the concentrations within the layer, we may for k 3 0
(A4)
The constant a is related to the incidence angle Bo and to the refractive index no within the incidence medium (silica in our experiments) by a = no sin 60 (A5) In the following, we suppose that the thickness of the interface is small compared to the wavelength X in vacuo, It is therefore possible to approximate the functions F ( z ) and 4(z) by their expansions to first order. The refractive index n ( z ) may be written, in our adsorption studies, as n ( z ) = no z I O n ( z ) = n , + An(%) z > 0 646) If we suppose An(%)
