Adsorption of Diblock Copolymers of Poly(ethylene oxide) and Polylactide at Hydrophobized Silica from Aqueous Solution

Journal of Colloid and Interface Science 228, 326–334 (2000) doi:10.1006/jcis.2000.6930, available online at http://www.idealibrary.com on

Adsorption of Diblock Copolymers of Poly(ethylene oxide) and Polylactide at Hydrophobized Silica from Aqueous Solution Dries Muller,∗,1 Martin Malmsten,∗, † Siriporn Tanodekaew,‡ and Colin Booth‡ ∗ Institute for Surface Chemistry, P.O. Box 5607, S-114 86 Stockholm, Sweden; †Department of Chemistry, Surface Chemistry, Royal Institute of Technology, SE-100 44 Stockholm, Sweden; and ‡Department of Chemistry, University of Manchester, Manchester M13 9PL, United Kingdom Received December 8, 1999; accepted April 24, 2000

The adsorption of a series of amphiphilic diblock copolymers of poly(ethylene oxide) (PEO) and poly(DL-lactide) (PL) at hydrophobized silica from aqueous solution was studied using time-resolved ellipsometry and reflectometry. The adsorbed amounts only display a weak dependence on the copolymer composition in both water and phosphate-buffered solution. For the short copolymers, the layer thickness decreases slightly with increasing length of the hydrophobic block. Furthermore, in comparison with the short copolymers, the layer thickness of the long copolymers is substantially higher. Upon degradation of the PL block, the adsorbed amount is found to decrease and approach that of the corresponding PEO homopolymer. Protein rejection studies indicate that the adsorption of fibrinogen is inhibited by copolymer preadsorption. The protein rejection is enhanced with increasing surface coverage of the preadsorbed copolymer, but largely independent of the length of the PL block and the PEO block. For all polymers investigated, essentially complete protein rejection is obtained above a critical surface coverage that is significantly lower than the saturation coverage of the copolymers. Removing the copolymer from bulk solution after preadsorption causes a partial desorption, resulting in reduced protein rejection. However, the protein rejection capacity with and without copolymer in the bulk solution is found to be similar at a given surface coverage. Contrary to the behavior of the intact copolymers, fibrinogen adsorption is found to be significant at surfaces pretreated with an extensively degraded copolymer and, in fact, quantitatively comparable to that at the hydrophobic surface in the absence of preadsorption. This finding, together with that of the effect of the copolymer composition on protein rejection, suggests that an efficient protein rejection is maintained until only a few L units remain in the copolymer, i.e., until nearly completed degradation. °C 2000 Academic Press Key Words: adsorption; copolymers; degradation; desorption; ellipsometry; fibrinogen; poly(ethylene oxide); polylactide; protein.

INTRODUCTION

The interfacial behavior of copolymers containing poly(ethylene oxide) (PEO) is of interest in a range of industrial applica1 To

whom correspondence should be addressed.

0021-9797/00 $35.00

C 2000 by Academic Press Copyright ° All rights of reproduction in any form reserved.

tions (1–4). Although PEO-containing copolymers, and PEO– PPO–PEO block copolymers in particular, are used extensively for the purpose of stabilization of colloidal systems in a range of pharmaceutical applications, a particularly interesting application area is parenteral administration of colloidal drug carriers (5–8). The reason for this is that PEO-containing copolymers can be used to control, reduce, or eliminate the unspecific adsorption of proteins, cells, bacteria, etc., at a range of interfaces, which is of critical importance to improve, e.g., the biological response to exogenous material. Examples of pharmaceutical colloidal systems stabilized with PEO-containing amphiphiles include emulsion droplets, liposomes, polymer particles, and solid drug particles (9–12). Much work regarding the interfacial behavior of PEOcontaining copolymers is related to block copolymers of PEO and PPO. This is not in the least due to the early commercial availability of a large series of these copolymers as well as the relatively good toxicity aspects displayed by at least some of these polymers (4). More fundamental aspects of the interfacial behavior of, e.g., diblock copolymers in aqueous solution have also been the subject of many theoretical (13–16) and experimental studies (17–24). However, from both the fundamental point of view and the perspective of biomedical and other technical applications, there is a need to broaden the scope and include also other types of copolymers. In addition, there is an interest in finding systems that perform better than the ones previously investigated or display other features desired, e.g., in a particular technical application. Recently, diblock copolymers of polylactide (PL) and PEO were successfully synthesized and characterized (25). These constitute a good example of new block copolymers of interest for, e.g., parenteral drug delivery of colloidal drug carriers. In addition, the hydrolytic degradation of the PL block allows for a range of interesting features. Thus, on intravenous administration of colloidal drug carriers, these are rapidly cleared from circulation in the bloodstream by the reticuloendothelial system (RES) and accumulated in RES-related tissue, e.g., liver and spleen (7). In most cases this is disadvantageous due to a limited resulting bioavailability and dose-limiting acculumation-related

326

327

DIBLOCK COPOLYMER ADSORPTION

side effects. Surface treatment with PEO-containing copolymers or lipid conjugates has significantly increased circulation time and a more even tissue distribution has been amply demonstrated (8). Despite this, however, the circulation time is still relatively short (≈days at the longest) and, therefore, protein rejection for a longer period of time than this is essentially unneccessary. If, instead, the polymer were to degrade after this period, this would reduce the limitations in the use of these polymers, at the same time as the resorbability of the polymer and of the drug, could be expected to be improved. To facilitate a controlled use of degradable copolymers in these applications, it is important to learn more about the adsorption properties of the copolymers. With respect to the currently investigated diblock copolymers of PL and PEO, the poorly soluble PL blocks (25, 26) could be expected to anchor at the surface, possibly forming a melt layer, whereas the soluble blocks are expected to extend into the solution (27, 28). Depending on the packing density of the molecules at the surface, and hence the overlap, the PEO chains will adapt a mushroom or brush configuration (27). This, in turn, is expected to have implications on the protein rejection and, hence, in the performance of the various applications discussed above. In the present investigation, we have studied the behavior of PEO–PL diblock copolymers, their effects on protein adsorption, and the effects of hydrolytic degradation on both the polymer adsorption and the protein rejective capacity of the surface coatings. In doing so we used the previously employed method of in situ ellipsometry and reflectometry. EXPERIMENTAL

Materials The water used was purified by a Milli-RQ 10PLUS unit (Millipore Corp.), including filtration, carbon adsorption, and decalcification followed by reverse osmosis. The water was then led through a Milli-Q PLUS185 unit (UV light, 185 and 254 nm) and a Q-PAK unit consisting of an active carbon unit, a mixed ion bed exchanger, an Organex cartridge, and a final 0.22-µm Millipak 40 filter. PEO-diol homopolymers with molecular weights of 2000 and 3400 (Mw /Mn = 1.03) were purchased from Shearwater, United States, and were used without further purification. Monodisperse pentaethylene glycol monododecyl ether, C12 E5 was purchased from Nikko Chemicals and was used without further purification. Diblock copolymers comprising poly(ethylene oxide) and poly(DL-lactide), abbreviated here as En Lm , with varying compositions were investigated in this study. In aqueous solution at 25◦ C, the solvent is selective for PEO, whereas the PL block is essentially insoluble (25). The detailed procedure of the preparation and synthesis of the copolymers was described elsewhere (25); the composition and most important characteristics of the diblock copolymers are listed in Table 1. Throughout the paper, a distinction is made between longer and shorter copolymers, based on blocks of EO≈80 and EO≈40 , respectively. Fibrinogen

TABLE 1 Number-Average Molecular Weight of the Polymer (Mn ), Polydispersity as Given by the Ratio between the Weight-Average Molecular Weight (Mw ) and Mn , Refractive Index Increment (dn/dc), and the Critical Micelle Concentration (cmc) of the Polymers Investigated in the Present Study (25) Polymer

Mn

Mw /Mn

dn/dc (cm3 g−1 ) (30◦ C)

(cmc) (g dm−3 ) (25◦ C)

E39 L5 E42 L12 E38 L16 E39 L20 E41 L26 E78 L14 E77 L26

2080 2710 2820 3160 3680 4440 5260

1.07 1.10 1.16 1.15 1.14 1.13 1.25

0.128 0.124 0.120 0.119 0.118 0.122 0.126

N.I 0.35 0.04 0.003 N.I. N.I. N.I.

(Fraction I, 92% clottable protein) was obtained from Sigma Chemical Co., St. Laus, MO. Chemicals used for the preparation of PBS buffer were of analytical grade and used without purification. Surfaces Silica surfaces were prepared from polished silicon slides (Okmetric, Finland). These were oxidized thermally and then annealed and cooled in argon flow to generate an oxide layer thickness of about 30 nm. The slides were then cleaned in a mixture of 25% NH4 OH, 30% H2 O2 , and H2 O (1 : 1 : 5, by volume) at 80◦ C for at least 5 min and then cleaned in a mixture of 32% HCl, 30% H2 O, and H2 O2 (1 : 1 : 5, by volume) at 80◦ C for at least 5 min. Thereafter, the slides were rinsed in distilled water and ethanol. Hydrophobic surfaces were prepared from the hydrophilic surfaces by a double rinsing subsequently with water, ethanol, and trichloroethylene (pro Analysi, Merck followed by treatment with a 0.2 wt% solution of Cl2 (CH3 )2 Si (Merck and Co., Rahway, NJ) in trichloroethylene for 45 min. Alternatively, the surfaces were hydrophobized by treatment with (CH3 )3 CCH2 CH2 Si(CH3 )2 Cl (Kebo) in the gas phase; no differences in adsorption behavior were observed in applying either of the two methods. After treatment, the slides were rinsed extensively with trichlorethylene and ethanol. Advancing and receding contact angles of 95◦ and 88◦ , respectively, were found for the hydrophobized slides. The surfaces were stored in ethanol. A treatment with boiling distilled water just prior to use was found to increase the stability of the surfaces hydrophobized using the dichlorosilane. This was probably due to the removal of residual silane accumulated at the surfaces. The contact angles after treatment with boiling water were identical to those for untreated hydrophobic surfaces. Methods The ellipsometry measurements were all performed by means of null ellipsometry. A thorough description of the experimental

328

MULLER ET AL.

setup and methodology involved can be found elsewhere (29–32). The instrument used in this study was an automated Rudolf thin-film ellipsometer, type 436, with the polarizer, analyzer, compensator, and sample cuvet in a horizontal arrange˚ was used as the light ment. A xenon lamp, filtered to 4000 A, source. Before the actual adsorption measurement was initiated, the complex refractive index (N2 = n 2 − ik 2 ) of the bulk silicon as well as the thickness (δ1 ), and refractive index (n 1 ) of the oxidized silica layer were determined. This was done by measuring the ellipsometric angles ψ and 1 in subsequently air and buffer, according to a four-zone procedure (31). Hereafter, adsorption was initiated by injection of a concentrated solution of the copolymer into the cuvet followed by a continuous measurement of the ellipsometric angles ψ and 1. The mean refractive index (n f ) and the average thickness (δf ) of the adsorbed polymer layer were calculated using an optical four-layer model (30, 33). n f and δf were then used to calculate the adsorbed mass (0), according to de Feijter et al. (34), using dn/dc values for the diblock copolymers as listed in Table 1 (25), 0.188 cm3 /g for fibrinogen and 0.133 cm3 /g for the PEO homopolymer (35). For the substrates used in this study, the perpendicular surface ˚ Simulations of roughness has previously been estimated at 5 A. surface roughness between the Si/SiO2 and the SiO2 /ambient media interface show that this results in very small effects on the determination of n f , δf , and 0 (31). Reflectometry was used for the titration experiments and has been extensively described by Dijt et al. (35, 36). Reflectometry measurements are characterized by a linear dependence between the signal and the adsorbed mass, and in this case the ˚ were calibrated ussubstrates (oxide layer thickness of 300 A) ing the adsorption of pentaethylene glycol monododecyl ether (C12 E5 ) as a reference (31). The determination of the adsorbed amounts included correction for the differences in the refractive index increment of pentaethylene glycol monododecyl ether (dn/dc = 0.131) (31) and the adsorbing copolymer. All experiments were performed at 25◦ C in pure water or in PBS buffer containing 5 mM phosphate buffer and 145 mM NaCl at pH 7.2. RESULTS AND DISCUSSION

Adsorption from Pure Water Time-resolved adsorption. In Fig. 1, the time-resolved adsorption of E41 L26 and E77 L26 is shown. The bulk concentration is 3000 ppm in both cases, corresponding to plateau conditions in the respective isotherms of both copolymers. A fast initial adsorption, notably the effect of a high bulk concentration, results in a surface coverage of over 75% within 30 s followed by a sharp drop in the adsorption rate. Near-saturation conditions are reached after approximately 1800 s, but a stable plateau is not reached within 10,000 s, which may indicate structural rearrangements and/or exchange effect occurring within the surfacebound layer over time. Maximum adsorbed amounts of 3.0 and 3.3 mg/m2 were found for E41 L26 and E77 L26 , respectively,

FIG. 1. Time evolution of the adsorbed amount (0, (a)), mean layer thickness (δf , (b)), and mean refractive index (n f , (c)) at the interface between water and hydrophobized silica at T = 25◦ C, for E 41 L 26 (e) and E 77 L 26 (r). In part (a), the adsorption of the homopolymers PEO-2000 (d) and PEO-3400 (n) are also shown. The bulk concentration was 3000 ppm in all cases.

329

DIBLOCK COPOLYMER ADSORPTION

showing that the difference in adsorption between the shorter and the longer copolymers is rather insignificant. However, the structure of the adsorbed layer shows considerable differences between the two polymers. In Fig. 1b, it can be seen that δf is substantially higher for E77 L26 than for E41 L26 , indicating that the former forms a more extended layer. Since the surface excess of the two polymers is rather similar, this also implies that E41 L26 forms a more compact layer. It is no surprise therefore that the layer refractive index (Fig. 1c) is higher for E41 L26 compared to that for E77 L26 . Using the refractive index increment for the polymers (Table 1) yields an average adsorbed layer polymer concentration of about 0.39 g/ml for E77 L26 and 0.55 g/ml for E41 L26 . The copolymer adsorption at hydrophobic silica is much higher than that displayed by the corresponding PEO homopolymer (Fig. 1a), which suggests that the PL block promotes the adsorption, most likely through a direct attractive interaction between the PL blocks and the surface. For PEO–PPO–PEO block copolymers and other PEO-containing amphiphilic di- or triblock copolymers adsorbing at hydrophobic surfaces, it is expected that the more hydrophobic block is located in the vicinity of the surface, whereas the PEO chains should protrude into the solution phase (27, 28). It is not unreasonable to expect that for our diblock copolymers the adsorbed layer is formed in a similar way, i.e., with the PL block largely responsible for anchoring the PEO chains at the surface. Saturation adsorption and layer thickness. For the adsorption of the copolymers from pure water at bulk concentrations of 3000 ppm (corresponding to plateau adsorption), Fig. 2 shows the saturation adsorbed amount and layer thickness as a function of the number of L units. As mentioned above, the adsorption of the copolymers is quite extensive compared to that of the corresponding PEO homopolymer, probably due to the PL

FIG. 2. The saturation adsorbed amount (e) and the layer thickness (n) for adsorption at hydrophobic silica from aqueous solution at T = 25◦ C, as a function of the length of the hydrophobic block, for E ≈40 copolymers (closed symbols) and the E ≈80 copolymers (open symbols). The bulk concentration was 3000 ppm in all cases.

TABLE 2 Saturation Adsorbed Amount, Γmax , and Layer Thickness, δf , at Hydrophobic Silica for Adsorption from Pure Water at T = 25◦ C (Given Also Are the Chain Density, σp , the Radius of Gyration, Rg , in Bulk Solution,37 and the Asymmetry Ratio of the Copolymers βs ) Polymer

0max (mg/m2 )

δf ˚ (A)

βs

σp (nm−2 )

Rg (nm)

E39 L5 E42 L12 E38 L16 E39 L20 E41 L26 E78 L14 E77 L26

3.03 2.75 2.63 2.74 3.05 3.36 3.27

68 65 62 59 55 85 78

16.2 7.4 4.9 4.1 3.3 13.3 7.1

0.88 0.61 0.56 0.52 0.50 0.46 0.37

1.7 1.7 1.7 1.7 1.7 2.5 2.5

block anchoring the adsorbed copolymer layer. However, despite the considerable variation in hydrophobic and hydrophilic block length of the copolymers the adsorbed amounts (at a bulk concentration of 3000 ppm) are rather similar, and the minimum and maximum values (2.6 and 3.4 mg/m2 , respectively) are found to be a mere 25% apart. However, although the saturation adsorbed amount appears to be rather independent of the length of the hydrophobic block, a significant variation is found in the interfacial chain density, σp , as shown in Table 2. It can be seen that when the PEO block length of the copolymer is constant, σp increases with decreasing PL block length. It is interesting to note that, for E39 L5 , containing the shortest PL block within the investigated polymer series, by far the highest interfacial chain density is found. The results thus hint at a strong attractive interaction between the PL block and the surface, allowing strong anchorage of the adsorbed layer, even when the length of the anchor is short (27). When the PL block length is constant, it can be seen that the presence of a longer hydrophilic block results in a lower value of σp , probably due to increased repulsive interaction between the longer PEO chains in the adsorbed layer (27). Comparing the observed chain density with the interfacial overlap concentration [σol = (π Rg2 )−1 = 0.113 and 0.050 chains/ nm2 for PEO homopolymers corresponding to E40 and E80 , respectively (37, 38)] shows that all copolymers experience a high degree of overlap as chain densities in the adsorbed layer are 3–9 times higher than σol (Table 2). The observed overlap indicates that the PEO chains must be deformed from their random coil conformation and it is not surprising therefore that the observed layer thickness (Fig. 2, Table 2) is significantly higher than the dimensions of the unperturbered coil as given by 2Rg [≈3.4 nm for corresponding E40 homopolymers and ≈5.0 nm for E80 homopolymers (37, 38)]. Furthermore, the layer thickness δf decreases somewhat with increasing PL block length, i.e., from ˚ ˚ for the shorter copolymers whereas δf is 82 and 78 A 68 to 55 A for the longer copolymers. The more extended layers formed by the latter are consistent with the presence of longer PEO blocks in the brush layer. Furthermore, the layer thickness is found

330

MULLER ET AL.

Protein Rejection

FIG. 3. Plateau values of the copolymer surface density σp in Fig. 2 vs −12/23 −6/23 NB , for the short copolymers (e) and the long copolymers (r). NA The solid line represents the best linear fit to the experimental data.

to decrease with decreasing interfacial chain density, which agrees well with a lower degree of chain stretching in the brush layer (27). In the scaling picture developed by Marques, Joanny, and Leibner (15) for the adsorption of diblock copolymers from a selective solvent, different regimes are found depending on the size of the anchoring and the buoy block. In the van der Waals brush regime the anchor layer is assumed to form a melt at the surface that fully wets the latter; σp is then found from a balance between the van der Waals force acting on the melt film and the osmotic pressure inside the brush. σp increases with the 6/5 asymmetry ratio βs (=NA−1 NB , where NA and NB represent the number of units in the anchor block and the buoy block, respec−12/23 −6/23 NB in the van der Waals tively) and scales as σp ∝ NA −12/23 −6/23 NB brush regime (15). In Fig. 3, we plotted σp vs NA using the data from Fig. 2, resulting in a good agreement with scaling predictions for the van der Waals brush regime. Thus, the observed chain densities follow the scaling relation both in dependence of the hydrophobic and hydrophilic block length. To enable the PL blocks to form a fully saturated anchoring layer, the PEO blocks would have to be highly stretched to fit into the small lateral space dictated by the relatively small anchoring blocks. The high interfacial density and high layer thickness displayed by the adsorbed copolymers are consistent with this notion. Despite the asymmetry of the copolymers [βs varies from 3.3 to 17 (Table 2)], this finding suggests that the copolymer adsorption is fully dominated by the adsorption of the PL blocks in the anchoring layer. While this is consistent with the effect of the L block length on the adsorbed amount (Table 2, Fig. 2), it is a somewhat unexpected result requiring further investigation.

Before discussing the protein rejection capacity of the surfaces precoated with copolymers, it is useful to look into the mechanisms of protein rejection by PEO-containing coatings. The considerations made here will form the basis for discussion of our results concerning the protein rejection behavior. The ability of neutral hydrophilic polymer coatings, particularly those containing PEO, to reduce the adsorption of a range of proteins is widely used in a range of medical applications (8, 39). The protein rejection properties of PEO-containing coatings are well known, and by now there seems to be an understanding that they are related to a combination of different effects and that steric interaction between proteins and the polymer layer is particularly important. In the presence of a PEO layer, the interaction between the surface and the protein is modified in a number of ways. Under conditions where PEO experiences good or fair solvency, i.e., in most medical applications, the osmotic equilibrium between chain entropy and the polymer–solvent interaction favors a high degree of hydration (27). In adsorbed layers containing nonadsorbing PEG chains attached to the surface through, e.g., an adsorbing block or a covalent coupling, analogous osmotic pressure is experienced with a driving force to maximize the hydration. Naturally, this leads to a state of saturation at a higher packing density at the surface and steric repulsion in the case of interaction between PEO-coated particles (27). Similarly, in the absence of protein-induced polymer desorption, there will be a repulsive effect when a protein approaches a PEO-coated surface since the interfacial polymer concentration must increase if proteins are to penetrate the PEO layer so that adsorption can take place (40–42). Furthermore, since PEO is uncharged and can be expected to reduce the effects of an underlying surface charge density, electrostatic interactions between the PEO-coated surface and the protein are diminished (43). Also, the water content in the PEO layers is generally rather high (44), which leads to a small van der Waals interaction between the PEO layer and the protein (40, 43). It has been found that the protein rejection of PEO coatings correlates with the PEO chain length. However, once a certain chain length has been reached, the PEO layers effectively eliminate protein adsorption, i.e., they maximize their protein rejection, which is then essentially constant when the chain length is increased further (8). As mentioned above, the protein rejection capacity is expected to be enhanced with increasing interfacial chain density, which has indeed been found for both adsorbed and covalently attached PEO-containing interfacial layers (44). It has furthermore been shown that the protein rejection capacity at a fixed chain length becomes essentially independent of the nature of the underlying surface, thus indicating that the chain density at the surface of the adsorbed layer is the dominant factor (8). In applications such as intravenous administration of colloidal drug carriers, the polymer concentration surrounding the drug

DIBLOCK COPOLYMER ADSORPTION

formulation is extensively diluted once administered. Thus, the protein rejection in relation to desorption effects is an important consideration in relation to the effiency of a formulation. If the saturation of the adsorbed PEO layer occurs at high bulk concentrations, extensive desorption may be the result upon dilution of the bulk phase (27). Depending on the magnitude of the desorption effect, it would be expected that part of the protein rejection properties would be lost. Polymers often display some degree of practically irreversible adsorption, the origin of which can be found in a number of factors, e.g., polymer polydispersity, preferential adsorption, or exchange effects at the surface, all leading to the presence of longer and more hydrophobic polymer fractions at the surface (24, 27, 45). In addition, polymer systems typically display exceedingly slow rates for interfacial conformational changes, exchange, and desorption upon dilution. In summary, for applications of PEO coatings it is essential to gain insight into the shape of the adsorption isotherm as well as the degree of desorption upon dilution. It is important to bear in mind that although complete desorption is not frequently observed, the initial desorption may be quite significant. Copolymer adsorption from PBS. Protein rejection of fibrinogen by hydrophobized silica after preadsorption with a copolymer was investigated. The experiments were performed using physiological buffer (PBS, pH 7.2), which is widely used when model systems are investigated for the adsorption of blood proteins. Parallel to the protein rejection experiments, we investigated the adsorption behavior of the copolymers from PBS. In Fig. 4, the adsorption isotherms from PBS buffer are given for E39 L5 , E41 L26 , E78 L14 , and E77 L26 . As can be seen in all cases, the isotherms show a gradual increase until plateau adsorption is reached. Pseudo-plateau conditions are reached at concentrations that are approximately 20 ppm for E41 L26 and 1000 ppm for E39 L5 , and thus the adsorption affinity in the initial part of the isotherm is sharply enhanced for the copolymer with the longer PL block length. The same trend, although far less pronounced, can be observed for the longer copolymers, where plateau conditions are reached at bulk concentrations around 50 ppm. On one hand, it thus appears that an increase in hydrophobicity can promote adsorption at the lower bulk concentrations. On the other hand, there is little difference in the shape of the isotherms displayed by E41 L26 and E77 L26 despite the longer hydrophilic block of the latter. From a practical point of view, the use of a more hydrophobic copolymer would thus be advantageous when coating a surface since saturation of the adsorbed layer may be achieved at quite low bulk concentrations. When one focuses on the plateau adsorbed amounts, it can be seen that the adsorption of both the longer and the shorter copolymers is lower from PBS buffer (Fig. 4) than from pure water (Fig. 2). In particular, the adsorption of the longer copolymers is significantly decreased. Thus, adsorption from PBS shows a more straightforward dependence on the composition of the copolymer than that found in the case of pure water. From Fig. 4 it can be seen that the adsorbed amount increases both with in-

331

FIG. 4. Adsorption isotherms at hydrophobized silica for adsorption from PBS at T = 25◦ C for the copolymers E 39 L 5 (h), E 41 L 26 (d), E 78 L 14 (e), and E 77 L 26 (m).

creasing length of the PL block and decreasing length of the hydrophilic block. From a mechanistic point of view, consideration of the interfacial chain density of the adsorbed copolymer layers when adsorbed from PBS yields similar results to those discussed above; i.e., the results indicate that the copolymer adsorption from PBS is qualitatively similar to that from pure water. The lower adsorption of the copolymers from PBS compared to the adsorption from pure water might be explained by a somewhat weaker effective interaction between the anchoring PL block and the surface in the presence of PBS. Interaction between counterions and the surface may affect the interaction of the copolymer layer with the surface and may indeed cause uncharged polymers to desorb from charged surfaces (27). In addition, the copolymers solutions in pure water have a pH value of approximately 4, presumably due to residual lactic acid (25). The native silica siloxy groups at the surface are more extensively protonated in this situation than in PBS at pH 7.2, thus resulting in a higher surface charge density in the latter case (46). It should be noted here that hydrophobization of silica surfaces has previously been shown to leave the surface charge density largely unaffected (47) and simultaneous hydrophobic and negatively charged surfaces have also been found for hydrophobized glass (24). All of these effects are expected to result in a decreased adsorbed amount. For the copolymers E39 L5 and E78 L14 , the time-dependent desorption is shown in Fig. 5. Desorption was initiated by rinsing the cuvet, after adsorption for 1 hour at a bulk concentration of 1000 ppm. The initial desorption is rapid, followed by decreasing desorption rates until, after approximately 1.5 h, the adsorption has more or less stabilized. In the case of E39 L5 , the adsorption decreases from 2.3 to 1.3 mg/m2 , corresponding to a decrease of roughly 40%. For E78 L14 , the adsorbed amount decreases from 1.8 to 0.8 mg/m2 , corresponding to a decrease of over 50%. Clearly, in the case of the longer copolymer, the

332

MULLER ET AL.

FIG. 5. Desorption of the copolymers E 39 L 5 (r) and E 78 L 14 (e) upon rinsing, after preadsorption at a bulk concentration of 1000 ppm for 1.5 h.

remaining adsorption is lower than that for the short copolymer. In both cases, however, desorption is incomplete, which may be related to a slow detachment rate. However, polydispersity can also affect the adsorption behavior. This can be manifested by a highly preferential adsorption of polymer fractions notably present in very low concentrations, which can successfully compete for adsorption with the monodisperse copolymer fraction or cause fractionation and/or interfacial exchange at the surface (24, 45). The adsorption may then be considered practically irreversible. Fibrinogen adsorption. The increase in adsorption at hydrophobized silica after copolymer preadsorption when fibrinogen is added to the system is shown in Fig. 6. Although ellipsometrically detected adsorption only provides information about the total adsorbed amount and therefore cannot as such be used to exclude the possibility of exchange adsorption at a fixed total adsorbed amount (44), the latter is less likely, as discussed previously. The increase in adsorption is therefore attributed to fibrinogen. As can be seen, the adsorption of fibrinogen onto hydrophobized silica is strongly affected by copolymer preadsorption. More precisely, the adsorption of the protein is increasingly suppressed at higher amounts of preadsorbed copolymer. In good agreement with previously reported results (48–50), the adsorption of fibrinogen at bare hydrophobized silica is approximately 5 mg/m2 (Fig. 8). Thus, it can be seen from Fig. 6 that the onset of protein rejection occurs at a copolymer coverage of approximately 0.5 mg/m2 . With an increasing amount of preadsorbed copolymer, the adsorption of fibrinogen decreases in a straightforward manner and leads to complete rejection of the latter at copolymer adsorbed amounts of approximately 1.9– 2.3 mg/m2 . It is important to note that though the adsorption of fibrinogen is dependent on the preadsorbed amount of copolymer, it appears to be largely unaffected by the composition of the latter. As discussed above, increasing the interfacial chain

density, leading to an increased overlap of PEO chains and thus an increased effective steric repulsion, is expected to be an important factor in the rejection of protein adsorption. The results thus imply that at a fixed surface coverage the protein rejection exerted by the copolymer layer is largely unaffected by the minor changes in the layer structure, occurring between the different copolymers. Instead, these findings indicate that the protein rejection capacity of these copolymer layers is largely dominated by the surface coverage. At their maximum surface coverage, all polymers displayed in Fig. 6 show a complete protein rejection. In the case of E39 L5 and E78 L14 , the maximum surface coverage of approximately 2 mg/m2 coincides with the surface coverage at which full protein rejection is displayed. In the case of E41 L26 , complete protein rejection occurs at a slightly higher surface coverage of 2.3 mg/m2 , which is, however, well below the maximum surface coverage of this polymer. As was seen in Fig. 5, the copolymer layer is partially desorbed upon dilution of the bulk phase. The practical importance of this effect in relation to protein rejection is the manner in which the remaining copolymer layer is able to resist protein adsorption. In Fig. 6, it can be seen that, with equal amounts of preadsorbed copolymer, there is essentially no difference between the protein rejection capacity in the absence and in the presence of the block copolymers in solution; i.e., the functionality of the copolymer layer is similar before and after rinsing. However, due to desorption, the maximum surface coverage that can be obtained after rinsing has decreased below the surface coverage where complete protein rejection is observed. Nevertheless, the effective reduction in protein adsorption compared to that of the bare hydrophobized silica surface remains high at 80%.

FIG. 6. The adsorption increase upon addition of fibrinogen after different preadsorbed amounts of E 39 L 5 (r), E 41 L 26 (d), and E 78 L 14 (j). The closed symbols indicate the presence of bulk copolymer during the adsorption of fibrinogen, whereas the open symbols indicate that the adsorbed layer has been rinsed prior to the adsorption of fibrinogen.

333

DIBLOCK COPOLYMER ADSORPTION

It should be noted that the observed results are subject to some degree of uncertainty due to possible exchange effects that might take place. More specifically, the copolymers are not covalently attached to the surface and might thus be exchanged to some extent by fibrinogen during protein adsorption. Nevertheless, the observed effectiveness of the preadorbed copolymer layers in resisting protein adsorption and its straightforward dependence on the preadsorbed amount are a strong indication that the copolymer layers remain essentially intact when exposed to fibrinogen. Furthermore, as has been discussed previously, complementary results indicate that little exchange occurs for similar systems (44). Degradation. Degradation of the copolymers is achieved by hydrolysis of the ester bonds linking the units in the PL chain (51–55). Disintegration of the PL chain can be expected ultimately to lead to the decrease in functionality of the anchoring block in binding the copolymer to the surface. Thus, it is anticipated that the adsorption behavior of the degraded copolymer will resemble that of the PEO homopolymer rather than that of the intact copolymer. In PBS at pH 7.2, hydrolysis is found to be a very slow process, which is in agreement with the results previously found for the degradation of PL. Hydrolysis is greatly enhanced at either high or low pH as well as at elevated temperature, where extensive degradation occurs (52, 55). For the current experiments, copolymers were degraded in aqueous solution at 80◦ C, without adjustment of the pH. Using HPLC, the lactic acid concentration in solution was followed until all PL was degraded, i.e., until the lactic acid concentration reached a maximum and constant value (56). In Fig. 7, the effect of complete degradation of E39 L5 on the adsorption onto hydrophobized silica is displayed. As can be seen, the adsorption of the degraded copolymer is greatly reduced compared to that of the intact copolymer. In fact, the adsorption is similar to that of the homopolymer PEO-2000.

FIG. 8. The adsorption of fibrinogen at hydrophobized silica from PBS at 25◦ C without copolymer preadsorption (e), after preadsorption of 4000 ppm E 39 L 5 (n), and after preadsorption of extensively degraded E 39 L 5 at a bulk concentration corresponding to 4000 ppm of the undegraded polymer (r).

The effect of degradation on the effective protein rejection of preadsorbed E39 L5 is shown in Fig. 8. As discussed above, preadsorption of the copolymer E39 L5 leads to a complete protein rejection. However, when preadsorption is performed after extensive degradation of the copolymer, the resulting copolymer layer, which is notably consisting mainly of PEO homopolymer, is not able to reduce protein adsorption. In that case, the adsorption of fibrinogen is essentially indistinguishable from that on bare hydrophobized silica. The adsorption and canceled protein rejection of the degraded copolymer E39 L5 (Figs. 7 and 8, respectively) indicate that degradation prevents the formation of a dense brush layer, thus leading to a reduced anchoring and hence adsorption of the copolymer. However, considering the interfacial behavior of the block copolymers as discussed above, it is obvious that the PL block needs to be extensively degraded before loosing its ability to anchor the copolymer at the surface. In turn, this implies that the functionality of the copolymer in rejecting protein adsorption may be maintained throughout degradation until a few L units remain in the anchoring block. SUMMARY

FIG. 7. Adsorption at hydrophobized silica from PBS at T = 25◦ C and a bulk concentration of 4000 ppm of undegraded E 39 L 5 (e) and extensively degraded E 39 L 5 (r).

The interfacial behavior of a series of PEO–PL diblock copolymers from an aqueous solution on hydrophobized silica was investigated. It was found that the hydrophobic PL block anchored the copolymers at the surface, resulting in a high interfacial density and layer thickness, which increased with a decreasing length of the anchoring block. The adsorption results were in agreement with those of the scaling relation for the van der Waals brush regime. The copolymers displayed a basic capacity to reject protein adsorption, regardless of their composition, when preadsorbed onto hydrophobized silica. Protein rejection

334

MULLER ET AL.

increased with the preadsorbed amount of copolymer, and complete protein rejection was observed at a sufficiently high surface coverage. After rinsing, the surface coverage decreased but was still sufficient to achieve roughly 80% protein rejection. Upon degradation of E39 L5 , the adsorption was greatly reduced compared to that of the intact copolymer and resembled instead the adsorption of the PEO homopolymer. Furthermore, the preadsorbed copolymer lost all of its protein rejection capacity after degradation. ACKNOWLEDGMENT This work was a part of the center for Amphiphilic Polymers from Renewable Sources, a jointly funded research program subsidized by the Swedish National Board for Industrial and Technical Development (NUTEK), and an industrial consortium.

REFERENCES 1. Alexandridis, P., and Hatton, T. A., Colloids Surf. A 96, 1 (1995). 2. Edens, M. W., in “Nonionic Surfactants: Polyalkoxyalkylene Block Copolymers” (V. M. Nace, Ed.), Chap. 5, p. 185, Surfactant Science Series 60. Dekker, New York, 1996. 3. Harris, J. M., and Zalipsky, S., Eds., “Poly(ethylene glycol) Chemistry and Biological Applications.” American Chemical Society, Washington, DC, 1997. 4. Schmolka, I. R., in “Polymers for Controlled Drug Delivery.” (P. J. Tarcha, Ed.), Chap. 10, p. 189, CRC Press, Boca Raton, FL, 1991. 5. Semple, S. C., and Chonn, A., J. Liposome Res. 6, 33 (1996). 6. Chonn, A., Semple, S. C., and Cullis, P. R., J. Biol. Chem. 267, 18759 (1992). 7. Patel, H. M., Crit. Rev. Ther. Drug Carrier Syst. 9, 39 (1992). 8. Malmsten, M., Ed., “Biopolymers at Interfaces,” Surfactant Science Series 75. Dekker, New York, 1997. 9. Jones, M. N., Adv. Colloid Interface Sci. 54, 93 (1995). 10. Lieberman, H. A., Rieger, M. M., and Banker, G. S., Eds., “Pharmaceutical Dosage Forms: Disperse Systems,” Vol. 2. Dekker, New York, 1989. 11. M¨uller, R. H., Mehnert, W., Lucks, J.-S., Schwartz, C., zur M¨uhlen, A., Weyhers, H., Freitas, C., and R¨uhl, D., Eur. J. Pharm. Biopharm. 41, 62 (1995). 12. Tarcha, P. J., Ed., “Polymers for Controlled Drug Delivery.” CRC Press, Boca Raton, FL, 1991. 13. Evers, O. A., Scheutjens, J. M. H. M., and Fleer, G. J., Macromolecules 23, 5221 (1990). 14. Wijmans, C. M., Leermakers, F. A. M., and Fleer, G. J., Langmuir 10, 1331 (1994). 15. Marques, C., Joanny, J. F., and Leibler, L., Macromolecules 21, 1051 (1988). 16. Marques, C., and Joanny, J. F., Macromolecules 22, 1454 (1989). 17. Halperin, A., Tirrell, M., and Lodge, T. P., Adv. Polym. Sci. 100, 31 (1992). 18. Griffiths, P. C., Cosgrove, T., Shar, J., King, S. M., Yu, G.-E., Booth, C., and Malmsten, M., Langmuir 14, 1779 (1998). 19. Bijsterbosch, H. D., Cohen Stuart, M. A., Fleer, J., van Caeter, P., and Goethals, E. J., Macromolecules 31, 7436 (1998). 20. Wu, D. T., Yokoyama, A., and Setterquist, R. L., Polym. J. 23, 709 (1991). 21. Pagac, E. S., Prieve, D. C., Solomentsev, Y., and and Tilton, R. D., Langmuir 13, 2993 (1997).

22. Pelletier, E., Stamouli, A., Belder, G. F., and Hadziioannou, G., Langmuir 13, 1884 (1997). 23. Belder, G. F., ten Brinke, G., and Hadziioannou, G., Langmuir 13, 4102 (1997). 24. Schill´en, K., Claesson, P. M., Malmsten, M., Linse, P., and Booth, C., J. Phys. Chem. B 101, 4238 (1997). 25. Tanodekaew, S., Pannu, R., Heatley, F., Attwood, D., and Booth, C., Macromol. Chem. Phys. 198, 927 (1997). 26. Ivanova, T. Z., Panaiotov, I., Boury, F., Proust, J. E., and Verger, R., Colloid Polym. Sci. 275, 449 (1997). 27. Fleer, G. J., Cohen Stuart, M. A., Scheutjens, J. M. H. M., Cosgrove, T., and Vincent, B., “Polymers at Interfaces.” Chapman & Hall, London, 1993. 28. Tiberg, F., Malmsten, M., Linse, P., and Lindman, B., Langmuir 7, 2723 (1991). 29. Azzam, R. M. A., and Bashara, N. M., “Ellipsometry and Polarized Light.” North-Holland, Amsterdam, 1989. 30. Landgren, M., and J¨onsson, B., J. Phys. Chem. 97, 1656 (1993). 31. Tiberg, F., and Landgren, M., Langmuir 9, 927 (1993). 32. Tiberg, F., J¨onsson, B., Tang, J., and Lindman, B., Langmuir 10, 2294 (1994). 33. Hansen, W. N., J. Optical Soc. Am. 58, 380 (1968). 34. de Feijter, J. A., Benjamins, J., and Veer, F. A., Biopolymers 17, 1759 (1978). 35. Dijt, J. C., Cohen Stuart, M. A., Hofman, J. E., and Fleer, G. J., Colloids Surf. 51, 141 (1990). 36. Dijt, J. C., Cohen Stuart, M. A., and Fleer, G. J., Adv. Colloid Interface Sci. 50, 79 (1994). 37. Bhat, R., and Timasheff, S. N., Protein Sci. 1, 1133 (1992). 38. Devanand, K., and Selser, J. C., Macromolecules 24, 5943 (1991). 39. Malmsten, M., in “Amphiphilic Block Copolymers: Self-Assembly and Applications” (B. Lindman and P. Alexandridis, Eds.), Elsevier, Amsterdam, in press. 40. Jeon, S. I., Lee, J. H., Andrade, J. D., and de Gennes, P. G., J. Colloid Interface Sci. 142, 149 (1991). 41. Jeon, S. I., and Andrade, J. D., J. Colloid Interface Sci. 142, 159 (1991). 42. Halperin, A., Langmuir 15, 2525 (1999). 43. Israelachvili, J. N., “Intermolecular and Surface Forces.” Academic Press, London, 1992. 44. Malmsten, M., Emoto, K., and Van Alstine, J. M., J. Colloid Interface Sci. 202, 507 (1998). 45. Eskilsson, K., Grant, L. M., Hansson, P., and Tiberg, F., Langmuir 15, 5150 (1999). 46. Burns, N. L., Van Alstine, J. M., and Harris, J. M., Langmuir 11, 2768 (1995). 47. Malmsten, M., Burns, N. L., and Veide, A., J. Colloid Interface Sci. 204, 104 (1998). 48. Malmsten, M., J. Colloid Interface Sci. 168, 247 (1994). 49. Wahlgren, M. C., and Arnebrant, T., J. Colloid Interface Sci. 142, 503 (1991). 50. Wojciechowski, P. W., and Brash, J. L., Colloids Surf. B 1, 107 (1993). 51. Mason, N. S., Miles, C. S., and Sparks, R. E., Polym. Sci. Technol. Biomed. Dent. Appl. Polym. 14, 279 (1981). 52. de Simone, V., Maglio, G., Palumbo, R., and Scardi, V., J. Appl. Polym. Sci. 46, 1813 (1992). 53. Makino, K., Ohsima, H., and Kondo, T., J. Microencapsulation 3, 203 (1986). 54. Belbella, A., Vauthier, C., Fessi, H., Devissaguet, J.-P., and Puisieux, F., Int. J. Pharm. 129, 95 (1996). 55. Wallis, K. H., and M¨uller, R. H., Pharm. Ind. 55, 168 (1993). 56. Muller, D., Malmsten, M., Tanodekaew, S., and Booth, C., J. Colloid Interface Sci., in press.