3635
Langmuir 1994,lO,3635-3641
Preparation of Tailorable Interfaces: Adsorption of Reactive Copolymers onto Silica Particles M. P. Vivarat Perrin, C. Amiel,* and B. Sebille Laboratoire de Physico-Chimie des Biopolymkres, CNRS, Universitd Paris Val de Marne, 2-8 rue H . Dunant 94320 Thiais, France Received February 2, 1994. In Final Form: July 27, 1994@ The adsorption of two reactive copolymerspoly(vinylpyrrolidonelviny1chloroformate),poly(VPlCF),and poly(vinylpyrrolidonelviny1chloroformatemodified with N-hydroxysuccinimide),poly(VP/NHS),onto silica particles (porous and nonporous) has been studied. FTIR spectroscopy was used for the study of the adsorption onto nonporous silica particles, allowing the determination of the fractions of links between the adsorbed copolymers and the surface. A preferential adsorption of the vinylpyrrolidone units over the other ones has been observed. Moreover, a mechanism of polymer-polymer association has been demonstrated in the adsorbed layers of poly(VP/NHS). The influence of the surface structure on the adsorption has been studied with porous particles generally used for chromatography. It has been shown that the adsorption was enhanced with increased pore size for poly(VP1CF) copolymers. The porosity influence on the adsorption of the poly(VP/NHS) copolymer is complicated because of the additional interaction mechanism of polymer-polymer association.
Introduction Interfaces play an important role in many problems such as colloid stabilization,l adhesion,2coating, painting, In and separation processes such as ~hromatography.~ all these applications, one needs to control the interactions involved at the interfaces. This can be done by adsorption of a copolymer, sequential or random, with one of the monomers interacting directly with the surface and playing the role of an anchor and the other one extending in solution and giving the desired properties to the interface. Adsorption of block copolymershas been widely studied both experimentally and theoretically during the last few year^.^-^ It has been shown that the nonadsorbing blocks can be in an extended conformation, giving rise to ’ * ~the other hand, the so called “ b r u s h s t r u ~ t u r e . ~ ,On the available body of knowledge for random copolymer adsorption is more limited,1°-14 despite the wide practical use of these materials related to the easiest copolymer synthesis. In the present work, random copolymers where one of the monomer bears a reactive function are adsorbed on
*
Abstract published in Advance ACS Abstracts, September 15, 1994. (1)Napper, D. H.Polymeric Stabilization of Colloid Dispersions; Academic Press: London, 1975. (2)Lee, L. H. Adhesion and Adsorption of Polymers; Plenum Press: New York, 1980. (3)Snyder, L. R.; Stadalius, M. A. High Performance Liquid Chromatography; Advances and Perspectives; Horvath Editor 1986, Vol4. Jaulmes, A.; Vidal-Madjar, C. In Advances in Chromatography; Giddings, J. C., Grushka, E., Brown, P. R., Eds.; Marcel Dekker: New York, 1989;Vol. 28,p 1. (4)Kawaguchi, M.; Takahashi, A. Adu. Colloid Interface Sci. 1992, 37, 219. (5) Halperin, A,; Tirrell, M.; Lodge, T. P. Adu. Polym. Sci. 1992,100, 31.
(6)Marques, C. M.; Leibler, L.; Joanny, J. F. Macromolecules 1988, 21, 1051.
(7)Hair, M. L.; Aguzonas, D.; Boils, D. Macromolecules 1992, 24, 341. (8)DOrgan, J. R.; S t a ” , M.; Toprakcioglu, C.; Jerome, R.; Fetters, L. J. Macromolecules 1993,26, 5321. (9)Auroy, P.; Mir, Y.; Auvray, L. Phys. Rev. Lett. 1992, 69, 1, 93. (10)Marques, C. M.; Joanny, J. F. Macromolecules 1990,23, 268. (11)Cosgrove, T.; Finch, N. A.; Webster, J. P. R. Macromolecules 1990,23, 3353. (12)Van Lent, B. A.; Scheutjens, J. M. H. M. J . Phys. Chem. 1990, 31, 5033. (13)Kawaguchi, M.; Aoki, M.; Takahashi, A. Macromolecules 1989, 22, 2199. (14)Yamagiwa, S.;Kawaguchi, M.; &to, T.; Takahashi, A. Macromolecules 1990,23, 268.
silica particles. The use of a reactive copolymer allows the synthesis of interfaces which can be tailored with the desired properties in the last stage of preparation: after adsorption of the copolymer on the solid surface, the reactive functions can be further modified to create the desired chemical functions. Potential applications of the present work are materials for chromatography. Thus the adsorbed layers of copolymer on silica particles have a double function: to block the silica surface sites (silanol) by making hydrogen bonds and to control the interactions with the external medium by bearing specific chemical functions. Two kinds of copolymers have been synthesized for this study. They both contain a hydrophilic monomer, vinylpyrrolydone (VP),which is known to make hydrogen bonds with silanol functions. The reactive monomers are vinyl chloroformate (CF), which is highly reactive in organic solvents, and vinyl chloroformate modified with N-hydroxysuccinimide(MIS),which is reactive in aqueous mediums. Nonporous silica particles (Aerosil) have been chosen as a model system for the first part of this study. Adsorption isotherms and measurements of the fractions of saturated surface sites as well as measurements of the fractions of monomers directly interacting with the surface as a function of the adsorbed amounts have been performed with poly(VP1CF)and poly(VP/NHS)copolymers in order to obtain better insight into the adsorption mechanisms and the conformation of the chains in the adsorbed layers. These parameters were deduced from infrared spectroscopy measurements. The second part of the paper is devoted to the adsorption of poly(VP/CF) and poly(VP1 NHS) onto highly porous silica particles used in chromatography. The influences of the porosity and of the chemical composition of the V P M S copolymers have been analyzed.
Materials and Methods 1. Copolymers. Copolymer Poly(VP/CF). The radicalar
copolymerization of N-vinylpyrrolidone (VP) (Fluka, Buchs, Switzerland)and vinyl chloroformate (CF)(Aldrich, Milwaukee, WI)has been described in detail.15J6The molar masses of the two monomers are 106.5g for VP and 99g for CF. The structural formulas are given in Figure 1. The monomers, previously (15)Sebille, B.; Vivarat-Penin, M. P.; Thuaud, N. French Patent 9 112 872,1991. (16)Vivarat-Perrin, M. P.; Sebille, B.; Vidal-Madjar, C. J . Chromatogr. Biomed. Appl. 1992, 584, 3.
0743-746319412410-3635$04.50/00 1994 American Chemical Society
Vivarat Perrin et al.
3636 Langmuir, Vol. 10, No. 10, 1994
Table 1. Adsorbent Characteristics
f CH, -i Hj /
\
j
+*-y VP
P
c=o
I
CI A
0
NHS
CF Figure 1. Structural formulas ofthe monomer units. VP,CF,
and NHS represent the units of, respectively, poly(l\r-vinylpyrrolidone), poly(viny1chloroformate), and poly(viny1chloroformate) modified with N-hydroxysuccinimide. distilled, were diluted in dichloromethane (SDS, Peypin, France) and dried with a 4 A molecular sieve. A small amount of AIBN (Eastman Kodak, Rochester, NY) was added, the molar ratio of initiator/monomers being about 1.2%. The reaction bath was kept under a nitrogen atmosphere a t 35 "C for 48 h. The copolymersamples were purified by precipitationin diethyl ether. The precipitates were collected by filtration under a nitrogen atmosphere and vacuum dried for 12 h. The copolymer samples were then stored a t -30 "C under vacuum. The chemical composition (molar fractions) of the studied sample was determined by elemental analysis of carbon, chlorine, and nitrogen. This composition (representing 50% of each comonomer) was in good agreement with the values calculated from the reactivity ratios rl(CF) = 0.26 and rZ(VP)= 2.25. The rl and rz values were previously determined by the Alfred Price method." The number average molecular weight, ca. 30 000, was measured in dichloroethane at 20 "C using a high-speed membrane osmometer (Mechrolab 501). No data are available in the literature about the relative solvent qualities of the two co-units in dichloroethane or dichloromethane. However, a better solvent quality is expected for the VP sequences because of their higher acid-base properties. Copolymer Poly(VP/NHS). This Copolymer is obtained by modification of the poly(VP/CF) copolymer with N-hydroxysuccinimide by following the method described by Meunier.l* Solvents are dried on a molecular sieve of 4 A pore size, and the glasswares are carefully dried a t 150 "C before the reaction. NHS is dissolved in dimethylformamide and added drop by drop into a dichloromethane copolymer solution which is maintained a t low temperature in an ice bath. A pyridine solution in dichloromethane is added simultaneously. The molar ratios NHS/CF and NHS/pyridine are respectively 3 and 1.3. The reaction bath is kept under a saturated nitrogen atmosphere with magnetic stirring for 3 h. The modified copolymeris purified by precipitation in diethyl ether, and the NHS molecules in excess are eliminated by three successive dissolutions and precipitations, the solvents being successively DMF, a mixture containing 50% DMF and 50%dichloromethane, and pure dichloromethane. The precipitate is then filtered out on regenerated cellulose membrane under nitrogen pressure and dried under vacuum. The copolymer is stored at room temperature under vacuum. Infrared spectroscopy of this copolymer in dichloroethane solution shows that the carbonyl vibration band a t 1774 cm-l of the chloroformate units has disappeared. From elemental analysis of carbon, chlorine, and nitrogen, the yield of the reaction is estimated to be better than 70%. (17)Alfred,T.; Price, J. R. J. Polym. Sci. 1947,2,1 , 101. (18)Meunier, G.;Boivin, S.; Boileau, S.; Senet, J. P. Polymer 1982, 23, 861.
adsorbent Aerosil200 Nucleosil Polygosil IBF Lichrospher
particle size (Pm) .01 5 20 15-25 10
pore size (A) nonporous 300 300 1200 4000
surface area (m2/g) 200 100 100 25-35 10
2. Adsorbents and Solvents. Adsorbents are silica particles whose characteristics are listed in Table 1. The nonporous adsorbents are colloidal pyrogenic silica Aerosil200 (Degussa). The porous adsorbents are spherical silica particles from different origins used in chromatographic applications: Nucleosil and Polygosil have been purchased from Macherey-Nagel (Germany) and Lichrospher from Merck (Germany). The silica IBF was a gift from IBF (France). In order to remove the physisorbed water and to obtain a reproducible chemical state of the surfaces, the silica particles were heated at 120 "C for 48 h before use. The adsorption solvents are l,2-dichloroethane and dichloromethane. They are of spectroscopic uality (SDS, Fr), and they are stored over a molecular sieve of4 pore size to eliminate traces of water. 3. Adsorption. With the nonporous silica particles, a homogeneous suspension was first prepared by ultrasonication (35 khz,Bioblock) of a flask containing 0.2 g of silica and 5 mL of solvent for 15 min. The adsorption solvent was 1,a-dichloroethane. Adsorption experiments were carried out by adding 5 mL of copolymer solution t o each silica dispersion flask. The totalpolymer concentration was Ct (expressed in weightholume). The samples were gently mixed for 24 h. At this point, equilibrium was assumed t o have been reached among the three components of each sample. The suspensions were then centrifuged for 20 min at 1300g, and the polymer concentration of the supernatant was measured by monitoring the absorbance of the carbonylband a t 1685cm-1 of the VP segments. The adsorbed amount A (mg/m2)was derived from
1
A = (Ct - C,)V/S
(1)
where Vis the volume of the solution and S the surface area of the silica. The medium was washed three times with pure solvent by repeating the same process: suspension, centrifugation, and elimination of the supernatant. The solvent was thereafter evaporated, and the powder was dried a t 40 "C under reduced pressure for 24 h. The process was slightly different for the adsorption onto the porous substrates because the standard method used in our laboratory for the preparation of the chromatographic columns was applied. Moreover, the silica particles are bigger than in the previous case and cannot form a stable suspension. The adsorption solvent was dichloromethane, whose quality is similar to 1,2-dichloroethane but which is more easily evaporated after adsorption because of its lower boiling point. A 10 mL portion of polymer solution, concentration C,, was added to 1g of silica. The medium was sonicated for 3 min in an ice bath in order to degas the particles and then gently stirred for 4 h at room temperature. The polymer solution was then removed by filtration of the medium on a regenerated cellulose membrane (0.45pm)under a nitrogen atmosphere. As before, the medium was washed three times in pure solvent, and the particles were dried under vacuum for 24 h. The adsorbed amounts were deduced from elemental analysis of carbon,nitrogen, and chlorine. 4. Infrared Measurements. The infrared measurements were all done on a Perkin Elmer 1760Fourier Transform infrared spectrometer. For the analysis of the dry powders, a Perkin Elmer Diffuse Reflectance FTIR accessory was used. Two kinds of measurements were performed with the nonporous particles: transmission spectra of the suspensions, to deduce the fraction of monomers directly interacting with the surface by hydrogen bonds, and diffise reflectance of the dry substrates, to measure the fraction of saturated surface sites. Measurements of the Fractions of Bound Segments. The carbonyl functions present on each unit of the copolymers can interact with the silanol surface sites by forming hydrogen bonds. When such an interaction exists, the carbonyl vibration band is
Preparation of Tailorable Interfaces
Langmuir, Vol. 10, No. 10, 1994 3637
shifted to lower values, and this can be seen by infrared spectroscopy.19 The relative intensity of the shifted and nonshifted bands is directly correlated to the fraction of bound segments. After the period of adsorption and before centrifugation of the samples, small volumes of the dispersions were directly analyzed m. The adsorption in an infrared NaCl cell of path length 2 x spectrum obtained is the sum of the contributions of the solvent, the silica particles, the adsorbed copolymer, and the free copolymer in solution. The spectra of the solvent and of the silica are first substracted. Details of the method used for the data analysis are described elsewhere.20 In summary, the carbonyl peak of interest is composed of three different populations: (a) bound monomer, (b) free monomer belonging to adsorbed copolymer, and (c) free momomer belonging t o free copolymer in solution. c a b , c.f, and c, are their respective concentrations. The calculation of the bound fraction is based on the assumption that populations b and c have the same extinction coefficient. The carbonyl peak can then be resolved into two bands: a bound band and a free band, the last one being modeled by the solution spectrum. The total concentration in free units CG is obtained by substracting their contribution in the peak:
4 C
o " " ' " " " " ' ~
-,
0.1
CS.V/S
0.15
0.2
(mg/m2)
Figure 2. Isotherm for adsorption of poly(VP/CF)copolymer onto Aerosil 200 silica particles from dichloroethane. The adsorbed amounts are plotted versus a reduced quantity of the supernatant concentration C,. Vis the volume of the solution and S the total area of the silica particles. 4
The concentration C,, in free carbonyl units belonging to free polymers is derived from the supernatant concentration C. and the copolymer content in the studied units. The partial bound fraction Pais then deduced:
0 05
0
7
3.2 . -
1
(3)
Difuse ReflectanceFTIR Measurements. This method is very sensitive to the surface properties ofthe studied medium. It has been chosen for the analysis of the silica surface sites (silanols) in the study with nonporous silica. The samples were prepared in the following way: after the adsorption, washing, and drying processes described in section 3, the coated silica powders were diluted to 5%in dry KBr powder and sampled in a well-defined way in a 10-mm-diameter cup, without packing the top of the mixed powder. The reflectance spectra were stored against a reference, which was the spectrum of the diluent and nonabsorbing medium (KBr). The method used for a quantitative analysis of the spectra is described in detail elsewhere.20 The Kubelka Munk equation, which is the diffise reflectance analogue to Beer's law for transmission measurements was applied?l F(R)= (1 - R)'/UZ = k/s
(4)
where s is a scattering coefficient and k is the molar adsorption coefficient of the medium. Providing that s remains constant (s is dependent on particle size), a linear relationship is expected between F(R)and the absorbing species in the sample. For the measurements of the degree of saturation of the surface sites, we have looked at the attenuation of the free silanol stretching band (3747cm-l). The absorbance ofthis band (in KubelkaMunk units) was proportional t o the number of unsaturated surface sites nOH. The Si0 combination band (1872 cm-l) was used as an internal reference.22 The relative change in absorbance for coated silica compared to that of bare silica prepared under the same conditions allowed the determination of the degree of saturation of the surface sites 8:
e = 1 - noH/(noH)o
(5)
where (nodo is the number of free silanol surface sites for the bare silica particles.
Results and Discussion A. Adsorption onto Nonporous Silica Particles. (A1)Adsorption Isotherms. The isotherms for absorption (19)Fontana, B. J.; Thomas, J. R. J. Phys. Chem. 1961,65,480. (20)Amiel, C.; Sebille, B. J.Colloid Interface Sci. 1992,149,2,481. (21)Brimmer, P.J.; Griffith, P. R. Anal. Chem. 1986,58,2179. (22)Murthy, R. S. S.; Leyden, D. E. Anal. Chem. 1986,58, 1228.
I
Figure 3. Isotherm for adsorption of poly(Vp/NHS)copolymer onto Aerosil 200 silica particles from dichloroethane. The adsorbed amounts are plotted versus a reduced quantity of the supernatant concentration C.. Vis the volume of the solution and S the total area of the silica particles.
of poly(VP/CF) and poly(VP/NHS) copolymers onto silica Aerosil from dichloroethane are shown in Figures 2 and 3. In order to take into account the different adsorption conditions between the nonporous and porous silica particles, the adsorbed amounts A (mg/m2)are plotted versus a reduced quantity of the final supernatant concentration Cs:23
B = C,(V/S) where V is the volume of the adsorption solution and S the total area of the particles in the dispersion. B is in mg/m2. The isotherms are of the high-affinity type for both Copolymers with a very steep rise at low values of C, (B). Both isotherms cannot be easily fitted because they seem to be of the "two plateau" type. Comparable variations have often been observed in the adsorption of competitive homopolymer^.^^ The polydispersitiesin size and composition of the two copolymers could also induce a similar behavior. For instance, if the VP sequences are preferentially adsorbed on the surface, the chains having the highest VP content would displace the other ones. (23)Cohen Stuart, M. A,; Fleer, G. J.;Bijsterbosch, B. J. J.Colloid Interface Sci. 1982,90,2, 310.
Vivarat Perrin et al.
3638 Langmuir, Vol. 10, No. 10,1994
v
' L 1
LL 0 6 ' 0
"
0.625
'
, 1.25
i - ' o I
0
05
1.875
Adsorbed amounts (mum2) Figure 4. Plots of the saturation degree of the silanol surface groups for the poly(VP/CF) (0)and poly(VP/NHS)(*) copolymers as a function of the adsorbed amounts.
Nevertheless, the transition to the saturation plateau is sharper for poly(VP/CF) than for poly(VP/NHS). This difference should reflect different adsorption mechanisms. The saturation values are of the order of 2.5 mg/m2, identical for both copolymers. This value should be compared to the ones obtained with PVP homopolymers. Several authors have studied the adsorption of PVP on silica aerosil from different In chloroform, whose quality is comparable to dichloroethane,they found a plateau value of 0.75mg/m2 almost independent of the molecular weight. Thus the plateau adsorbed amounts are much higher for the copolymers than for the PVP homopolymers. We will see below that this behavior cannot be attributed to a higher affinity of CF and NHS comonomers for the silica surface. The density of silanol surface sites being of the order m ~ l / mthe ~ ,number ~~ of monomer units in the of 3 x adsorbed layer at saturation is approximately 7 times higher than the number of available surface sites. We can deduce that, at saturation, only 13% of the monomer units are linked to the surface with the silanols. (A2)Surface Coverage of the Silanol Groups. In Figure 4 are presented the variations of the degree of saturation of the surface groups, 8, as a function of the amount adsorbed for the two copolymers. The important variations of 8 (from 0 to 1)and the good correlation with the adsorbed amount make it obvious that the silanol sites constitute the greater part of the adsorption sites. 0 increases more sharply for poly(VP/CF)than for poly(VP/NHS),but for both copolymers the number of free silanols becomes negligible, in the sensitivity domain of the infrared method, for adsorbed amounts higher than 1.5-1.75 mg/m2. (A3)Fractions of the Bound Segments. The fractions of the monomer units bound to the solid surface have been determined from IR measurements by the method detailed in the previous section. For the poly(VP/CF)copolymer, each comonomer has a carbonyl function that can make hydrogen bonds with the silanols. For the poly(VP/NHS) copolymer, three carbonyl functions can interact with the silica surface: carbonyl belonging to the pyrrolidone units, carbonyl belonging to the carbonate function of the NHS comonomer, and (24)Kawaguchi, M.Adv. Colloid Interface Sci. 1990,32,1. ( 2 5 ) Killman, E.; Bergmann, M. Colloid Polymer. Sci. 1986,263,372. (26)Day, J. C.;Robb, I. D. Polymer 1980,21,408. (27)Korn,M.;Killmann, E. J. J.Colloid Interface Sci. 1980,76, 19.
1
-
-I--
IS
J
-
I
2
25
Adsorbed amounts (ms/m2)
Figure 5. Plots of the fractions Pvp (e)and PCF(0)of VP and CF bound segments of the poly(VP/CF)copolymer as a function of the adsorbed amounts. Table 2. Vibration Frequencies of the Carbonyl Groups"
function vinylpyrrolidone (VP)
chloroformate (CF) carbonate (NHS) succinimide (NHS)
v ( C 0 ) free v ( C 0 ) bound v(free-bound) (cm-') (cm-l) (cm-l) 1685 1774 1746 1790 1812
1660 1762 1738 1785 1808
25 12 8 5
4
carbonyls belonging to the succinimide function of the
NHS comonomer. The last ones have two valence vibrations, one is assymetrical at 1790 cm-' and the other one is symetrical at 1812 cm-'. The carbonyl vibration bands are shifted to lower values when hydrogen bonds are created. The vibration frequencies of the free and bound carbonyl bands are reported in Table 2. The infrared frequency s a have been related to the adsorption enthalpies:26the higher the adsorption enthalpy, the larger is the frequency shift. The shift is 25 cm-' for the VP units, much larger than for the other carbonyl functions. Thus the VP units have the higher affinity for the surface. PolyW/CF) Copolymer. The fractions of VP and CF units making links with the surface (Pvpand PCF) have been measured, and their values are plotted in Figure 5 as a function of the adsorbed amounts. On the whole range of adsorbed amount values, the fraction of bound VP segments Pvp is higher than the fraction of bound CF segments PCF.This is an additional proof that the VP carbonyl groups have a higher affinity for the silica surface than the CF carbonyl groups. Only 10% of the CF units are involved in hydrogen bonds with the silanol surface sites. This value is independent of the adsorbed amount. On the other hand, Pvp decreases markedly from 0.52 at low adsorbed amount to 0.26at the plateau of adsorption. This variation is in qualitative agreement with polymer adsorption models. The theoretical model of Scheutjens and initially developed for monodisperse homopolymer samples, predicts a behavior of this kind. At low surface coverage the polymer adopts relatively flat conformationscorrespondingto large P values (closeto 11,in order to maximize the interactions with the surface, whereas at high surface coverage additional interactions between polymers in the layer have to be considered, which result in a coiling of the chain conformation,leading to low P values. The average bound fraction PTOThas a smaller variation range (0.31-0.2)) (28) Hair, M. L. J. colloid Interface Sci. 1977,59, 532. (29)Scheutjens, J. M.H. M.; Fleer, J. G. J.Phys. Chem. 1970,83, 1619.
Preparation of Tailorable Interfaces
Langmuir, Vol. 10, No. 10, 1994 3639 I .o
1
I
I
I
I
I
0 C
c
0
00
Adsorbed amounts (mg/m2) Figure 6. Fractions PW (e),Pms(carbonate) (A), and PNHsfeucc-de) (0,O) ofw,NHs (carbonatefunction),and NHs (succinimide functions)bound segments of the poly(VP/NHS) copolymer plotted as a function of the adsorbed amounts.
showing that the adsorbed chains do not reach a flat conformation at very low surface coverage. This behavior has been observed with homopolymers (PVP,23PMMA26) and random c o p o l y m e r ~ and l ~ ~ has ~ ~ been attributed to the local stiffness of the adsorbed chains, due to sterical hindrance of the carbonyl groups of PVP which are not located on the backbone but are side groups of the chain. At the plateau of adsorption, the average bound fraction is 0.20. This value is in good agreement with the previous determination of 0.13 based on an estimation of the silanol density on the silica surface. This constitutes an additional proof that the adsorption of poly(VP/CF) copolymers is controlled by the hydrogen bonding between carbonyl groups of the copolymer and silanol sites of the silica. Poly(vp/NHS)Copolymers. Four carbonyl bands are present on this copolymer, one belonging to the VP units and the others to the NHS units. Their fractions of links have been measured, and the variations are plotted in Figure 6 as a function of the adsorbed amounts. These variations present special features. As for poly(VP/CF), the fraction of bound VP segments Pvp is always higher than the other ones. For adsorbed amounts lower than 1 mg/m2, Pvp and PNHS(carbonate) decrease with the adsorbed amount, as can be expected from polymer adsorptionmodels.29In this domain, the succinimide NHS bands have negligible fractions of links. This very weak interaction with the silica surface is corroborated by the observed low frequency shifts of 4-5 cm-'. For adsorbed amounts higher than 1-1.5 mg/m2, the four fractions of links show a marked increase with the adsorbed amounts. These variations are surprising because they are opposite of what is generally expected. Moreover, in this domain of the adsorption isotherm, the silanol surface sites are saturated (see Figure 4). To our knowledge, such a behavior has never been described in the literature. Our interpretation of this phenomenon is that at high surface coverage there are polymer-polymer interactions between the chains in the adsorbed layer of poly(VP/NHS) copolymers. Such an interaction cannot be seen in a diluted solution of polymer. In the same way, there is no interaction of this kind in the adsorbed layer at low surface coverage because the chains are diluted on the surface. On the other hand, polymer-polymer interactions can be favored in a concentrated polymer layer such as an
cm-I
Figure 7. Comparison of the infrared transmission spectra of the poly(VP/NHS) copolymer in a dilute solution of dichloro-
ethane (bold line) and in the molten state (sharp line).
adsorbed layer with high surface coverage. The interactions are favored by the high density of segments and their reduced mobility. To demonstrate the associating properties of poly(VP/ NHS) copolymers in concentrated media, we have studied the extreme situation of the copolymer in the molten state: a thick layer (ca. 20 pm) of copolymer has been realized on a CaF2 lamellae. The transmission infrared spectrum of the thick film is compared in Figure 7 to the spectrum of the copolymer in a dilute solution of dichloroethane. It can be seen that the four carbonyl bands of interest are shifted t o lower frequencies. This shows clearly the high ability of the chains to make polymerpolymer associations in a concentrated medium, through interactions between the NHS and VP segments. As is shown in Figure 7, the four carbonyl groups are involved in polymer-polymer associations. S h i h in these bands are thus representative of two kinds of interactions: hydrogen bonding with the silanol surface sites and polymer-polymer associations. The measurements of the bound fractions are based on the band shifts. The bound fractions cannot be taken as the fractions of units linked with the silica but as the fractions of units involved in links either with silica or with polymer. (A41Discussion. The adsorption study of poly(VP/CF) and poly(VP/NHS) copolymers onto nonporous silica particles has shown strong interactions between the macromolecules and the silica. Isotherms are ofthe highaffinity type, and the adsorbed amounts at saturation, 2.5 mg/m2, are higher than the ones obtained with the homopolymers. Moreover, we have shown that the silanol surface sites play a driving role in the adsorption mechanism and the amount of residual silanol is negligible at the plateau of adsorption. The infrared study has demonstrated a preferential adsorption of the VP units over the other ones. The reactive functions are thus less involved in hydrogen bonds with the surface but are likely to be in the mobile parts of the adsorbed chains: loops and tails. The reactive fimctions should then provide good accessibility for further modifications for preparing tailorable interfaces. The infrared study has also demonstrated a mechanism of polymer-polymer association for poly(VP/NHS) copolymers. This interaction, sensitive to the local concentration of the chains, is seen at high surface coverage of the particles. This additional interaction mechanism present in poly(VP/NHS) copolymers but not in poly(VP/ CF) copolymers explains the differencesobserved between the two adsorption isotherms: the transition to the
Vivarat Perrin et al.
3640 Langmuir, Vol. 10,No. 10, 1994
Table 3. Adsorbed Amounts of Poly(VP/CF)Copolymers as a Function of the Pore Size of the Silica Particles
v
silica Nucleosil
r
Polygosil IBF
Lichrospher D
i2!
0,5r w
0
25
7 5
5
10
cs . v / s
(mg/m2) Figure 8. Isotherm for the adsorption of poly(VP/CF)polymer on Nucleosil silica particles (porous) from dichloromethane.
3.2 -
1.6
0.8
0 -
-
0
"
"
'
2
4
6
8
cs. v / s
(ms/m2) Figure 9. Isotherm for the adsorption of poly(VP/NHS) copolymer on Nucleosil silica particles (porous) from dichlo-
romethane.
saturation plateau is much sharper for poly(VP/CF)than for poly(VP/NHS). It should be noted that adsorption isotherms with very slow saturation have been observed with proteins;30this has been attributed to multilayer adsorption, i.e. protein-protein interactions. B. Adsorption onto Porous Silica Particles. (Bl) Adsorption Isotherms. Adsorption isotherms have been realized with Nucleosil silica particles whose characteristics are listed in Table 1. As in the previous section, adsorption isotherms are represented with adsorbed amount A versus a reduced parameter B of the supernatant concentration (6). The variations are shown in Figure 8 and 9 for poly(VP/CF) and poly(VP/NHS), respectively. The isotherms present similar features as in the nonporous case, showing a high affinity of the polymers for the surfaces. A comparison of Figures 8 and 9 and Figures 2 and 3 shows that the graphs of the porous and nonporous experiments do not superimpose on a single master curve for each copolymer. The presence of pores of small size appears to induce different conformations of the adsorbed molecules and thus different adsorption mechanisms. For the poly(VP/CF) copolymer, the adsorbed amount at saturation, 1.54mg/m2,is lower than the one obtained with Aerosil silica (2.5 mg/m2). The adsorbed amounts are greatlyinfluenced by the structure ofthe surface. The influence of pore size is discussed in the next section. (30)Schmitt, A.;Varoqui, R.; Uniyal, S.; Brash, J. L.; hsinieri, S.
J.Colloid Interface Sei. 1983,92, 25.
mre size (A) 300 300 1200 4000
adsorbed amounts (mdm2) 1.45 1.35 1.68 2.65
It seems that the isotherm of the poly(VP/NHS) copolymer does not reach a saturation value for the range of concentrations studied. It should be underlined that concentrations of the polymer solution have been increased to 100g/L for the last point of the isotherm. The maximum adsorbed amount, 3 mg/mz,is higher than in the nonporous case, contrary to the trend observed with poly(VP/CF).It seems that the presence of small pores on the surface gives opposite effects for poly(VP/CF) and poly(vp/NHS). We have seen in sectionAthat NHS groups are responsible for polymer-polymer interactions in the adsorbed layers of poly(VP/NHS) copolymers. These attractive interactions are favored in a porous medium where the macromolecules are confined inside a restricted volume. The adsorbed layers tend to thicken (multilayer adsorption), and this explains why no saturation is observed on the isotherm. (B2)Influence of the Pore Size on the Adsorption of Poly(VP/CF). Four kinds of chromatographic silica particles with different pore sizes (300,1200,4000 8)have been studied. Adsorptions have been performed in the same conditions, from polymer solutions of 50 g/L. For Nucleosil silica, this corresponds to the plateau of adsorption (see Figure 8). As the other supports have an average pore size of the same order or higher, and the surface areas are of the same order or lower, we can assume that the plateau of adsorption is reached for all the supports. The adsorbed amounts have been measured from elemental analysis of carbon, nitrogen, and chlorine performed on the coated silica particles. The results are shown in Table 3. Before the results are discussed, we should underline that we compare the performances of silica particles with different structures. The pore sizes given in Table 3 are average values, and they are not representative of the broadness of the pore size distributions. Moreover, the adsorbed amounts are calculated using the surface areas given by the manufacturers. Very often, small molecules are used for the measurements of the surface areas. Micropores which are not accessible to macromolecules are taken into account. Thus, uncertainties which are linked to the experimental methods of surface area measurement and to the microporous structures of the material have to be taken into account for the adsorbed amounts values. We estimate that this gives an uncertainty of 10%. The adsorbed amounts are an increasing function of the pore size. For the particles having the highest pore sizes (4000A), the adsorbed amount is equivalent to the one obtained with nonporous particles. This behavior has already been observed with homopo1yme1-s.~~ This had been attributed to the steric hindrance of the chains in small pores. Chains are constrained to flatten on the surface instead of adopting an extended conformation toward the solution that maximizes the number of adsorbed chains. (B3) Influence of Copolymer Composition on the Adsorption ofPoly(W/NHS). Four copolymer samples have (31)Letot, L.; Lesec, J.;Quivoron, C.J.Liq.Chromutogr. 1981,4(8), 1311.
Preparation of Tailorable Interfaces
Langmuir, Vol. 10, No. 10, 1994 3641
Table 4. Influence of the NHS Composition of Poly(vp/ NHS) Copolymers on the Adsorbed Amounts
NHS
adsorbed amount
mole fraction
(mdm2) 1.77 1.70
.06 .OB .25 .35
2.20 2.62
been synthesized. The NHS contents were varied from 6 to 35%. The copolymers were adsorbed on Nucleosil silica from polymer solutions of concentration 50 g/L. The adsorbed amounts have been measured from elemental analysis of carbon and nitrogen of the coated silica particles and are reported in Table 4. There is an evident correlation between the adsorbed amounts and the NHS content ofthe copolymers: adsorbed amounts are an increasing function of the NHS mole fraction. The copolymer samples having the lowest NHS mole fraction (6%)give an adsorbed amount (1.7 mg/m2) that tends to be closer to the one obtained with poly(VP1 CF) copolymer on the same silica particles (1.45 mg/m2). This behavior corroborates our interpretation that the NHS segments, by linking with VP segments, are responsible for polymer-polymer interactions giving high adsorbed amounts on porous silica.
Discussion and Conclusion The adsorption mechanisms of the poly(VP/CF) and poly(VP/NHS) copolymers have been first studied using a model surface: nonporous Aerosil silica. The adsorption studies of the two copolymers have shown high-affinity isotherms with plateau values of the adsorbed amounts (2.5 mg/m2)higher than the one obtained with homopolymer PVP. The infrared study has evidenced the driving force for the formation of the hydrogen bonds between the silanol surface sites and the carbonyl groups of the different monomer units in the adsorption process. There is a preferential adsorption of the V P units over the other comonomers for both copolymers, resulting from a higher affinity for the silanol surface sites. As a consequence, the CF and NHS segments have a higher probability to be included in loops or tails rather than in trains. These are favorable conformations of the adsorbed chains for further chemical modifications of the surface because the reactive functions borne by CF and NHS segments are more accessible. The fractions of bound segments of the poly(VP/NHS) copolymer have shown nonmonotonic variations as a function of the surface coverage. This particular behavior has been attributed to polymer-polymer interactions, due to interactions between the VP and NHS segments.
The particles which are generally used in chromatographic applications are highly porous materials. The influence of porosity on adsorption has been studied in the second section. The adsorption isotherms of the two copolymers are still of the high aEnity kind, but they do not superimpose on the same master curves with isotherms of nonporous particles. Much higher excesses of polymer material are needed to saturate the surfaces in the porous case than in the nonporous case. This is probably due to an accessibility of the surface sites for the macromolecules, which is poorer when the sites are inside small pores. We have shown, by studying several kinds of chromatographic particles, that the presence of small pores greatly influences the adsorption by two antagonistic effects: The adsorbed amounts at the plateau are an increasing function of the average pore size. This has been observed with the poly(VP/CF) copolymers; it has been attributed to the constraints of adsorption to the restricted volume of the pores. The chains are forced to flatten on the surface instead of extending toward the solution. The adsorbed amount is thus decreasing with a decrase of the pore size. The presence of small pores leads to an increase of the adsorbed amounts in the case of poly(vp/NHS) copolymers. In this case, there is an attractive polymer-polymer interaction. These interactions are favored when the chains are confined to pores of restricted volumes. There is then a tendency to thicken the adsorbed layers. In terms of tailorable interfaces, the poly(VP/CF) and poly(VP/NHS) copolymers can be used as adsorbants on porous silica particles. In view of chromatographic applications 16,32-36 silica particles with larger average pore size (1200 will be preferred as they accomodate the adsorbed chains in a more extended conformation. To ensure a good blocking of the silanol surface sites, the adsorption should be performed at the saturation plateau. For poly(W/NHS) copolymers, macromolecules having a low content of NHS functions should be chosen, to reduce the thickness of the adsorbed layers.
A)
Acknowledgment. We thank J. Perichon for her technical assistance. (32)Alpert, A. J.; Regnier, F. E. J. Chromatogr. 1979, 185, 375. (33) Kopaciewicz, W.; Rounds, M. A.; Regnier, F. E. J.Chromatogr. 1985,318, 157. (34) Shomburg, G. LC-GC 1986, 6, 36. (35) Lemque, R.; Vidal-Madjar, C.; Sebille, B. J. Chromatogr. 1991, 553, 165. (36) Renard, J.; Vidal-Madjar, C.; Sebille, E.J. Liq. Chromatogr. 1992, 15, 71.