Sedimentation field flow fractionation in the determination of surface

Jan 2, 1991 - Jenq-Thun Li and Karin D. Caldwell*. Department of Chemical Engineering, Center for Biopolymers at Interfaces, University of. Utah, Salt...
0 downloads 0 Views 1MB Size
2034

Langmuir 1991, 7, 2034-2039

Sedimentation Field-Flow Fractionation in the Determination of Surface Concentration of Adsorbed Materials Jenq-Thun Li and Karin D. Caldwell' Department of Chemical Engineering, Center for Biopolymers at Interfaces, University of Utah, Salt Lake City, Utah 841 12 Received January 2, 1991. I n Final Form: April 12, 1991 Due to their large surface-to-volumeratios, colloidal materials are often used as substrates in adsorption studies, where evaluationof the surface concentrationof an adsorbed component requires exact knowledge of both the surface area of the substrate and the mass of the adsorbed species. Sedimentation field-flow fractionation (FFF) is frequently used to determine the size of colloidal materials of known density from their degree of retention in the fractionator. However, if the colloid is allowed to interact with an adsorbing species, the density of the adduct becomes a function of the level of adsorption and is, as such, not directlyapplicableto evaluationsof adduct size. Here, we show that it is possible to extract information from sedimentation FFF retention which permits characterization of the adlayer in terms of the amount adsorbed per particle or related quantities such as surface concentration and the area occupied per adsorbed molecule. The method is appliedto evaluate the adsorptionof a PEO-PPO-PEO triblock polymeric surfactant to standard PS latex particles of different sizes, and the results are shown to compare well with quantification based on isotope labeled surfactant.

Introduction Studies of adsorption phenomena require quantification of the often minute amounts of analyte associated with a given surface area. This problem is generally solved in either of two ways. One approach is to use flat substrates for the adsorption, in which case the surface area is small but easily quantified, while quantification of the minute amounts of adsorbate requires sophisticated surface spectroscopic techniques. The alternative approach is to use a substrate which presents a large surface area and therefore takes up relatively large quantities of solute; this simplifies quantification of the adsorbed component, which can now be performed with standard analytical techniques, but requires close attention to the determination of the surface area accessible to the adsorbing species. The favorable surface-to-volume ratios characteristic of colloidal materials have frequently led to their use as substrates in the determination of adsorption isotherms. If the colloid is of uniform size, the surface area available for adsorption is then easily determined from its concentration, provided the particle size is accurately known. Colloids of broad distribution require estimates, rather than precise determinations, of the surface area available for adsorption and are therefore less desirable for quantitative work. Once the substrate area is determined, the isotherm is established by means of a series of adsorption experiments in which given amounts of particles are incubated with different concentrations of the adsorbing species. After removal of the equilibrated adsorption complex, the amount of adsorbed solute is determined either indirectly from the levels of solute depletion or from direct assessments of the amounts of solute adsorbed under the different solution concentrations. The determination of colloid size frequently involves the use of photon correlation spectroscopy (PCS)1.2or

* To whom correspondence should be addressed. (1) Weiner,B. B. InModernMethodsofPurticleSireAnalysis;Barth, H. G., Ed.; Wiley: New York, 1984, p 93. (2) Cummins, P. G.; Staples, E. J. Langmuir 1987,3, 1109.

sedimentation methods such as sedimentation field-flow fractionation ( s ~ ~ F F Fand ) ~analytical -~ ultracentrifugation (AUC).6J Of these, the PCS is an in situ method which provides an average size of the particles in a suspension. The ultracentrifugation analysis is generally based on measurements of sedimentation rates; for simple systems of uniform density, this technique gives an accurate representation not only of the average size but of the entire size distribution in a given colloid. At the end of the AUC analysis the various components, which were temporarily separated during the size measurement, are again mixed as the sedimentation field is removed. Sedimentation FFF is an elution method that has been used extensively for the purification and characterization of colloidal materials.* It is particularly the high inherent mass resolution and the well-established relationship between retention and buoyant particle mass which are responsible for this use. If the goal of a particular experiment is separation alone, as opposed to size or mass analysis, sedFFF is often the technique of choice because of its ability to generate fractions of well-resolved components. If however, the method is to be used for the purpose of determining either size or mass of the eluted particles, one needs to have exact knowledge of the densities of both the particles and their suspension medium in order to convert observed retention volumes into size/ mass assignments for the colloid. By means of simple pychnometry one can generally obtain accurate densities for the liquid eluant; densities for synthetic sample particles of uniform chemical composition are also obtainable by this method, provided the materials are (3) Giddings, J. C.; Yang, F. J.; Myers, M. N. Anal. Chem. 1974, 46,

1917.

(4) Kirkland, J. J.; Yau, W. W.; Doerner, W. A. Aml. Chem. 1980,52, 1944. ( 5 ) Giddings, J. C.;Caldwell, K. D. In Method8 inPhy8icd Chemietry; Rossiter, B. W., Hamilton, J. F., Eds.; Wiley: New York, 1989; Vol3B, p 867. (6) Mlchtle, W. Makromol. Chem. 1984,185, 1025. (7) Li, J. M.; Caldwell, K. D.; Mlchtle, W. J. Chromatogr. 1990,517, 361. (8) Caldwell, K. D. In Modern Methods of Particle Size Anolyek; Barth, H. G.,Ed.; Wiley: New York, 1984; p 211.

0143-7463f 91f 2407-2034$02.50/0 0 1991 American Chemical Society

Determination of Surface Concentration by FFF available in sufficient amounts to permit accurate dry weight determinations or other assessments of concentration. Samples of nonuniform composition, such as those resulting from the adsorption of one sample-component onto another, present a problem in terms of density assessment, since the densities of such complexes are a direct reflection of the unknown relative amounts of two interacting species. This is particularly true if the adsorbing species is a loosely coiled macromolecule whose conformation on the substrate will vary with surface concentration. Due to their complexity, such systems have so far eluded examination by sedimentation-based separation techniques such as sedFFF and AUC. In fact, from measurements of sedimentation rates in the AUC, one cannot expect to determine an adsorption-induced increase in particle mass. This is due to the fact that the observed sedimentation velocity is directly proportional to the particle's buoyant mass while being indirectly proportional to its friction against the suspension medium, i.e. its size. When flexible macromolecules adsorb to colloidal particles, there is no clear-cut relationship between the composite's mass and size, and consequently there is no direct correlation between mass or size, on the one hand, and the sedimentation velocity on the other. In the present study we wish to show that the mass of a particle coating, or an ad-layer, can be determined by using the sedimentation FFF technique?i.e. by the use of a sedimentation equilibrium method as opposed to the measurement of sedimentation velocities. Specifically, we will examine a number of three-component systems consisting of latex core particles (A), a flexible macromolecular substance with affinity toward these particles (B), and the liquid medium (C)in which the particles are suspended and the macromolecules dissolved. For these systems we will show that it is possible to determine the level of adsorption of B on A, provided accurate densities of each of the components A, B, and C are at hand. A similar approach has been developed by Giddings and Beckett and is described in a companion article.9 In order to seek independent verification of the validity of the sedimentation FFF based measurements of the surface concentration of B on A, we have developed a protocol for radiolabeling the macromolecule, B and determining the amount of B which is irreversibly attached to each bead.1° It is our purpose here to evaluate the polymer surface densities determined from such labeling experiments and to compare the results with those determined by sedimentation FFF.

Theory The mechanism behind retention and selectivity in FFF has been detailed in a number of arti~les;~~~~8for the purpose of the present discussion we will only briefly summarize a few key features. The fractionation takes place in a thin channel (thickness dimension w ) whose rectangular cross section has a breadth-to-thickness ratio in excess of 40. In practice, flow through such channels approximates that between infinite parallel plates. The channel is filled with the intended carrier fluid which in general is selected for its compatibility with the sample. After injection of a sample at the head of the channel, the flow of fluid is interrupted and a field is applied in a direction perpendicular to the major walls of the channel. (9) Giddin 8, J. C.;Beckett, R. Langmuir, following paper in this issue. (10) Li, J. Caldwell, K. D.; Tan, J. S. In Particle Size Assessment and Characterization;Provder,T., Ed.;ACS SymposiumSeries;American Chemical Society: Washington, DC, in press.

5.;

Langmuir, Vol. 7, No. 10,1991 2035 In the present case, this field is a sedimentation field generated by spinning a rotor basket in which the channel is contained. Under the influence of the field, the injected particles will migrate and concentrate at either the outer or the inner wall, depending on whether they are denser or less dense than the carrier. This concentration is opposed by the particles' Brownian motion, so that at equilibrium the sample is confined to a thin layer whose concentration c decays exponentially with distance x from the wall

Parameter A, which represents the dimensionless thickness of the layer, is an explicit function of the strength of the applied field G, as well as of the buoyant mass m' of each particle"

kT m'Gw Here, k and T have the usual meaning of Boltzmann constant and temperature. Replacing m' in eq 2 by (mA - Vue) where mA is the actual particle mass, V Athe particle's volume, i.e. its mass divided by its density, and pc the density of the medium, gives A=-

(3) If one assumes that the particles are spherical with diameter d , and replaces mA with the product of volume and density, one obtains has an explicit function of particle diameter A=

6kT

(4)

d37rApGw In this expression, Ap symbolizes the difference in density between particle and fluid. Once the particles are at equilibrium in the channel, one initiates a flow of liquid tangential to the accumulation wall, i.e. perpendicular to the direction of the field. This flow, which is characterized by a parabolic velocity distribution, will transport the equilibrated particle zones downstream with mean velocities determined by the compactness of each zone. Thus, the more compact distributions will on the average encounter slower moving flow lines and will trail those zones which are less compressed. Their relative migration rates, expressed by the retention ratio R, will therefore uniquely reflect the reduced layer thickness of each component (R is experimentally determined as the ratio of the column void volume V o to the observed retention volume Vr)3 VO

vr = R = 6h[ coth(') 2h - 2 x 1 By a combination of eqs 4 and 5 one can directly convert retention volumes into size data for known values of the density difference Ap between the liquid medium and the particle. If the particle is allowed to adsorb a surface coating of unknown thickness, its new density will be a reflection of the unknown amount of adsorbed material, and the direct application of eq 4 is therefore unfeasible. Under such conditions one must consider the particle moving through the fluid as a composite, consisting of the core, whose mass (11) Giddings, J. C.; Yang, F. J.; Myers, M. N. Science 1976,193,1244.

Li and Caldwell

2036 Langmuir, Vol. 7, No. 10, 1991

Figure 1. Schematic representation of the adsorption complex, indicating the core particle, with its mass m ~density , p ~ and , volume VA, and the adsorption layer consisting of unsolvated polymer of mass m~ and density p~ plus the solvation shell whose mass is mc and density pc.

and volume are mA and VA, respectively, the adsorbed species,which we assume to be a polymer whose unsolvated mass and volume are m g and Vg, and the solvation layer with mass mC and volume Vc, as illustrated in Figure 1. The densities of the three components are assumed to be PA, pg, pc, respectively. Furthermore, it is assumed that the ad-layer is irreversiblyadsorbed, so that no lossesoccur when the particle is transferred from the adsorption medium (consisting of B and C) to the pure liquid (C) in which the fractionation takes place. Consideringthe mass of the composite, m,to equal the sum of its components, i.e. m = mA + m g + mc, and the volume of the composite, V, to equal ( ~ A / P A+ ~ B / P B+ m c / ~ c )or m/p, where P is the compositedensity, one can express the buoyant mass m' in eq 2 as

or m' = mA( 1 - p.) + m,( 1 - p.) PA

PI3

By combining eqs 2 and 6 one now obtains the following expression for the retention parameter X L W

A=

K1

[m,(

1 - p.) + m,( 1 --

(7)

PA

It is thus clear that an observed elution volume Vr,which according to eq 5 corresponds to a unique value of parameter A, can give information about either of the four sample specific properties mA, mg,PA,or pg, provided the remaining three, and the carrier density pc, are known quantities. In using colloidal particles as substrates for adsorbing macromolecules, one frequently has independent knowledge of the size (mass) of the core particle obtained from sedFFF, photon correlation spectroscopy (PCS), electron microscopy (EM), or a combination of these technique^.^ The densities, PA, pg, and pc, in turn, are generally determined by pychonometry on the pure liquid (in case of pc) or on suspensions (for PA) or solutions (for pg) of known concentrations. Therefore, with the aid of these input parameters it is possible to convert an observed sedFFF retention into a value for the mass mg of a given particle coating. Experimental Section Materials. Polystyrene latex standards with nominal diameters of 69,130,212,272,394, and 482 nm (Seradyn) were used as substrates in the adsorption experiments, while the adsorbing

species was the polymeric surfactant Pluronic F108 from the BASF Corporation. This triblock polymer with a molecular weight of 14 600 consists of PEO/PPO/PEO (poly(ethy1ene oxide),PEO;poly(propy1ene oxide),PPO) in the monomeric ratio of 129/56/129. The density of this material in its unsolvated state was earlier determined by a PAAR Model DMA 60 digital precision density meter to be 1.186 g/mL.l0 During the adsorption experiments, the latex particles (2.5% (w/w)) were incubated with F108 (4.0% (w/w)) in phosphate buffered saline (PBS, I = 0.15 M, pH 7.4) for periods of 24 h under constant end-over-end shaking a t room temperature. The sizes of the coated latex particles were then measured by FFF. For comparison, the coated particles were also sized by PCS. In these measurements it became necessary to include a filtration step before the actual sizing, since a certain amount of aggregates appeared to form in the adsorption process and since these aggregatesaffected the average particle size determined by PCS. In the filtration steep, Millipore filters with a pore size of 1.0 pm (type FA) were used for particles larger than 200 nm, while for the smaller particles, a pore size of 220 nm (Type Millex GV) was used. Methods. In this work, both core particles and particle-surfactant composites were sized by means of photon correlation spectroscopy using a Brookhaven Model BI90 fixed angle instrument. The sizing of the core particles was performed in 0.1% aqueous FL 70 surfactant (Fisher Scientific), while the compositeswere sized in PBS. All liquids used for sample dilution were previously filtered through a 220-nm Millipore Millex GV filter. Each reported diameter is the mean of a t least ten observations. In addition, the bare particles were sized by using either sedimentation or flow FFF,l* where the inherently more highly resolving sedFFF technique was used for all particles except the smallest one of the set, which lies outside the resolution range for our sedFFF system. The sedFFF system was built in-house, essentially according to the specifications used to construct the Model 100 particle fractionator from FFFractionation, Inc.; its flow channel has the following dimensions: length 96.0 cm, breadth 2.0 cm, and thickness 0.0254 cm, for a measured void volume (V o )of 4.7 mL. The field strengths used to accomplish sizing with this instrument were 2000 rpm (694g) for the 130-nm particle, 1500rpm (390g)for the 212-nm particle, 1OOOrpm (173g) for the 272-nm particle, 550 rpm (5%) for the 394-nm particle, and 400 rpm (28g) for the 482-nm particle, and the carrier flow was maintained a t 2.1 mL/min throughout. Injections of 5 pL of unfiltered sample were followed by relaxation (equilibration in the absence of flow) for periods of less than 30 min in all cases; retention times were likewise less than 30 min. The same conditions were employed for the subsequent analyses of the composites,except that the carrier in these experiments was 0.1 % F108 with the addition of 12 mM NaC1. All reported sizes are the mean of at least three determinations made under identical conditions. The smallest particle of the lot had a nominal diameter of 69 nm. This particle, with and without the F108 coating, was sized by flowFFF,ll using an instrument built in-house whose flow channel had the dimensions of 45.0 cm X 2.0 cm X 0.0254 CM for an experimentally determined void volume of 1.17 mL. Here, the sizing was performed under a cross-flow of 15.6 mL/h and acarrierflowof6.5mL/h. ThecarrierwasO.l% FL'lOforanalyses of the core particles and PBS for the composites. Reported sizes represent the mean of a t least three determinations. The mass of Pluronic FlO8 adsorbed on the latex particles was quantified both by sedFFF, according to the above procedure, and by measurements of radioisotope labeled surfactant. In the labeling step, to be described in detail elsewhere,a Bolton-Hunter reagent (Pierce) was conjugated with the end-hydroxyl groups of the surfactant's PEO blocks, and lWIwas then introduced into the reagent's phenolic moiety. Following a 24-h adsorption of labeled surfactant (mixed 150 with unlabeled F108) from a 4% (w/w) F108solution in PBS, which is well into the plateau region of the adsorption isotherm,12the coated polystyrene spheres were pelleted in a Fisher Micro-centrifuge, Model 235A, and the supernatant was removed. After resuspension in PBS the (12) Kayes, J.

B.;Rawline, D.A. Colloid Polym. Sei. 1979,257,622.

Determination of Surface Concentration by FFF particles were again centrifuged and the supernatantcollected for radioactivity measurement with a Beckman 170M radiocounter. These washing cycles were repeated until the supernatant was free from measurable radioactivity. When coated with l9-labeled F108, the smaller PS particles (69 and 130 nm) cannot be forced to pellet using the above method. Instead, a centrifugal concentrator containing a 30 OOO MW cutoff filter (Amicon,Centricon-30)was used to separate solids from supernatant. During careful washing of the solids collected on the filter, the filtrate was repeatedly counted for radioactivity; the washing continued until no further activity could be detected. At this point, the radioactivity of the solid was measured and related to the known amount of particles present to give the surface densities of F108 on the different particles.

Results The PEO-PPO-PEO triblock with the trade name of Pluronic F108 is a macromolecular surfactant with an average molecular weight of 14600. Triblocks of this general composition are known to adsorb strongly to polymeric colloids, such as polystyrene latex particles, via their relatively hydrophobic PPO center-block. This mode of adsorption leaves the PEO side-arms in a mobile state as they extend outward from the particle surface and provide stability to the suspension by suppressing aggregation.12J3 The adsorption of FlO8 to PSis for all practical purposes irreversible. Thus, in a previous studylo we have shown that radiolabeled F108 remains completely associated with the PS substrate even three days after removal of the particles from the coating solution and resuspension in phosphate buffered saline (PBS). Although the carrier in the present set of sedFFF experiments is of slightly lower ionic strength (12 mM NaC1) and contains a small amount (0.1%) of F108, we assume that the composition of the adducts, which form when the various latex particles of this study adsorb F108 from a 4% solution, remains unaffected by the less than hour-long FFF retention process. We are likewise assuming that the removal of nonadsorbed 12%labeled F108 from the coated particles by PBS washing leaves the adsorbed surfactant layer intact. The thickness of the ad-layers which form when polymeric surfactants adsorb to colloidal substrates has been the subject of numerous investigations.12-16 Most frequently the PCS technique is used to measure the sizes of both core particles and adducts and the thickness of the coating is found from the difference between the two. In our earlier studylo we demonstrated how the presence of minor amounts of aggregates, which sometimes form during the adsorption of F108, affect the PCS-derived size. By introducing a sedFFF separation step prior to the PCS sizing of the adsorbate or by performing the sizing in conjunction with a separation by flowFFF, it was possible to improve the accuracy in these size measurements. While a filtration procedure is adopted to remove such aggregates prior to PCS sizing of the adsorbate (see Experimental Section),the core particles are sized directly by using either PCS or FFF in the sedimentation or flow modification; these sizes are listed in Table I. Such filtration is not necessary for samples to be examined by FFF, where existing aggregates are either eluting separately under the applied field, as shown previously,lOorwould emerge upon removal of the field. We note that each of the FFF analyses requires less than an hour for completion (seeExperimental Section). By contrast, quantification of the irreversibly (13) Baker, J. A,; Berg, J. C. Langmuir 1988, 4, 1055. (14) Killmann, E.; Maier, H.; Baker, J. A. Colloids Surf. 1988,31,51. (15) Lee, J.;Matric,P.A.;Tan,J.S.J.ColloidInterfaceSci. 1989,131,

252.

Langmuir, Vol. 7, No.10, 1991 2037 Table I. Determination of Coating Thickneer, F-108on PS diameter (bare), nm nominal determined

69 130 212 272 394 482

69 f 2 145 f 1 214 f 0 275 2 390f 2 458 f 20

diameter (coated),nm FFF 82 f 2

coating

thickness, nm 6.5 f 2.0

*

PCS 69 68f3 84f2 8.0 & 2.5 130 141 f 0 157 f 0 8.0 f 0.0 212 206f 1 230 f 1 12.0 f 1.0 272 272 3 320 f 2 24.0 f 2.5 394 404 f 4 446f5 21.0 4.5 482 484f6 505*8 10.5 f 7.0 a Uncorrected for wall effects.' For FFF, the quoted standard deviations are the result of a minimum of three determinations for each sample;for PCS, the standard deviationsresult from a minimum of 10 measurements on the same sample at the same time.

*

*

Table 11. Determination of Surface Concentration, F-108on PS adsorbed surface diameter technique amt X 1018, density, (bare), nm usedb mg/particle nm*/molecule 69 I 0.18 0.03 20.1 0.3 130 I 1.04 f 0.04 12.4 0.5 130 S 1.42 f 0.00 9.1 0.0 212 I 3.75 f 0.02 9.2 f 0.1 212 S 3.79 0.17 9.1 f 0.4 272 I 7.16 0.09 7.9 0.1 272 S 8.07 f 0.35 7.0 0.3 394 I 13.88 f 0.15 8.5 f 0.1 394 S a a 482 I 19.86 f 0.04 8.9 0.1 482 S a a Out of range for SedFFF. b I = Isotope labeling, S = SedFFF. For both techniques,the standard deviationsresult from a minimum of three measurements.

*

*

** * **

adsorbed 1261-labeled surfactant is a relatively timeconsuming, labor-intensive, and error-prone procedure which includes both the initial modification of the surfactant as well as the extensive washing of the composite to remove all traces of unadsorbed radiolabel. If losses of particles during the wash cycles can be treated as negligible, a given initial amount of particles will be associated with a measured amount of radiolabel taken up from a surfactant solution which contains a small, well-known fraction of labeled F108. Assuming no preferential adsorption of the labeled molecules, the amount of radiolabel, together with the size and concentration of the particles, is used to calculate the total amount of surfactant per particle, listed in Table 11,and the related surface densities (see also Figure 2). From this data set it appears that the F108 surfactant adsorbs to the smaller particles in a less crowded arrangement than to particles of larger diameter and lesser curvature. Given the relative freedom of motion experienced by the PEO side chains a t the lower surface concentration, one might expect a thinner ad-layer at these concentrations.le A detailed study of the relationship between the surface density, layer thickness, and chain mobility of F108 on the one hand, and PS substrate curvature on the other, will be the subject of a separate communication. The difference in sedimentation FFF behavior between bare and coated particles is exemplified by Figure 3, which (16)de Gennee, P.G. Adu. Colloid Interface Sci. 1987,27, 189.

Li and Caldwell

2038 Langmuir, Vol. 7, No. 10, 1991

. h

a