Langmuir 2004, 20, 101-108
101
Monitoring Solute Interactions with Poly(ethylene oxide)-Modified Colloidal Silica Nanoparticles via Fluorescence Anisotropy Decay Dina Tleugabulova, Andy M. Duft, Michael A. Brook, and John D. Brennan* Department of Chemistry, McMaster University, Hamilton, Ontario, L8S 4M1, Canada Received July 22, 2003. In Final Form: October 28, 2003
The fluorescence-based nanosize metrology approach, proposed recently by Geddes and Birch (Geddes, C. D.; Birch, D. J. S. J. Non-Cryst. Solids 2000, 270, 191), was used to characterize the extent of binding of a fluorescent cationic solute, rhodamine 6G (R6G), to the surface of silica particles after modification of the surface with the hydrophilic polymer poly(ethylene oxide) (PEO) of various molecular weights. The measurement of the rotational dynamics of R6G in PEO solutions showed the absence of strong interactions between R6G and PEO chains in water and the ability of the dye to sense the presence of polymer clusters in 30 wt % solutions. Time-resolved anisotropy decays of polymer-modified Ludox provided direct evidence for distribution of the dye between bound and free states, with the bound dye showing two decay components: a nanosecond decay component that is consistent with local motions of bound probes and a residual anisotropy component due to slow rotation of large silica particles. The data showed that the dye was strongly adsorbed to unmodified silica nanoparticles, to the extent that less than 1% of the dye was present in the surrounding aqueous solution. Addition of PEO blocked the adsorption of the dye to a significant degree, with up to 50% of the probe being present in the aqueous solution for Ludox samples containing 30 wt % of low molecular weight PEO. The addition of such agents also decreased the value and increased the fractional contribution of the nanosecond rotational correlation time, suggesting that polymer adsorption altered the degree of local motion of the bound probe. Atomic force microscopy imaging studies provided no evidence for a change in the particle size upon surface modification but did suggest interparticle aggregation after polymer adsorption. Thus, this redistribution of the probe is interpreted as being due to coverage of particles with the polymer, resulting in lower adsorption of R6G to the silica. The data clearly show the power of time-resolved fluorescence anisotropy decay measurements for probing the modification of silica surfaces and suggest that this method should prove useful in characterization of new chromatographic stationary phases and nanocomposite materials.
Introduction Birch1-5
introduced In the past few years, Geddes and a new theoretical framework that describes the relationship between the rotational characteristics of entrapped fluorescent probes and the evolution of particle growth in silica-based sols. According to this theory, short (picosecond time scale) rotational correlation times correspond to probe that is free in solution; long (nanosecond scale) rotational correlation times, which are often observed in the anisotropy decays of cationic probes such as rhodamine 6G (R6G) in silica sols, correspond to dye molecules that are electrostatically bound to anionic primary silica nanoparticles; and residual anisotropy (r∞) values correspond to dye molecules that are bound to large silica structures existing in the medium that rotate too slowly to cause fluorescence depolarization during the 1-10 ns emission lifetime of a typical fluorescent probe. Provided that the viscosity of the surrounding solution is known, the nanosecond anisotropy decay component can be related to the hydrodynamic radius (R) of the * To whom correspondence should be addressed. Tel: (905) 525-9140 (ext 27033). Fax: (905) 527-9950. E-mail: brennanj@ mcmaster.ca. (1) Geddes, C. D.; Birch, D. J. S. J. Non-Cryst. Solids 2000, 270, 191. (2) Birch, D. J. S.; Geddes, C. D. Phys. Rev. E 2000, 62, 2977. (3) Geddes, C. D.; Karolin, J.; Birch, D. J. S. J. Fluoresc. 2002, 12, 113. (4) Geddes, C. D.; Karolin, J.; Birch, D. J. S. J. Fluoresc. 2002, 12, 135. (5) Geddes, C. D. J. Fluoresc. 2002, 12, 343.
primary particles by applying the Stokes-EinsteinDebye (SED) equation,6
η)
3φkT 4πR3
(1)
where φ is the rotational correlation time on the nanosecond scale, η is the microviscosity of the sol solution, k is the Boltzmann constant, and T is the temperature. Recently, we applied the nanoparticle metrology approach to follow the evolution of aqueous and glyceroldoped sodium silicate sols through the sol-to-gel transition.7 The possibility of differentiating between the free molecular rotation of the probe and the slow motion of silica-probe complexes led us to hypothesize that the nanosize metrology approach may be appropriate for testing the efficiency of silica surface modification using covalently coupled silanes such as (aminopropyl)triethoxysilane.8 In this work, we are extending our fluorescencebased studies of silica surface modification to include physisorption of water-soluble polymers, such as poly(ethylene oxide) (PEO), onto Ludox particles. The silica particles in Ludox are discrete, uniform, negatively charged spheres of silica that have no internal surface area or detectable crystallinity. R6G, being a cationic (6) Debye, P. Polar Molecule; Chemical Catalog Co.: New York, 1929. (7) Tleugabulova, D.; Zhang, Z.; Brennan, J. D. J. Phys. Chem. B 2003, 107, 10127. (8) Tleugabulova, D.; Chen, Y.; Zheng, Z.; Brook, M.; Brennan, J. D. Langmuir 2004, in press.
10.1021/la035333z CCC: $27.50 © 2004 American Chemical Society Published on Web 12/09/2003
102
Langmuir, Vol. 20, No. 1, 2004
probe, adsorbs strongly to the surface of Ludox particles and under such conditions will rotate with the correlation time of the silica particle.5 The interest in this specific polymer comes from the similarity of specific interactions of PEO (hydrogen bonding and hydrophobic interactions) to those found in proteins and other biological materials. Additionally, PEO is known to stabilize colloidal dispersions,9 act as an antifouling agent on membranes,10 and resist the adsorption of proteins and prevent the growth of cells on surfaces.11 Furthermore, sol-gel-derived silica doped with PEO and other polymers has been shown to provide enhanced activity for entrapped enzymes12 and to alter the dynamics of entrapped biomolecules.13 All these qualities make the incorporation of PEO chains into silicabased nanocomposites the basis of a general methodology for the incorporation of bioactive molecules into ceramics and glasses14 and make such materials useful for biomedical and environmental immunochromatography and immunosensing.15 As shown here, the anisotropy decays of R6G in both PEO/water and PEO/Ludox systems are sensitive to the concentration and molecular size of the polymer, and in particular the data show that increasing the total moles of polymer in solution, either by increasing total polymer weight percent or by reducing polymer molecular weight, leads to increased surface coverage of the silica with the polymer and a concomitant decrease in the degree of adsorption of R6G to the silica surface. The fluorescence results are considered in light of atomic force microscopy (AFM) images of unmodified and PEO-modified silica particles, which suggest that polymer-modified silica forms aggregates, resulting in large structures that appear to adsorb less R6G than unmodified particles, based on the anisotropy data. The implications for preparing surfacemodified silica nanocomposites are discussed. Experimental Section Chemicals. PEO of average molecular weight 550, 10 000, and 100 000 (polydispersity index of 1.06, 1.14, and 1.18, respectively, as stated by the manufacturer) and Ludox AM-30, 30 wt % SiO2 (average particle radius of 6 nm, specific surface area of 220 m2 g-1, Du Pont) were purchased from Aldrich (Milwaukee, WI). Rodamine 6G was obtained from Sigma (St. Louis, MO). All water was distilled and deionized using a Milli-Q Synthesis A10 water purification system. All reagents were used without further purification. Procedures. Polymer-Modified Colloidal Silica. Homogeneous dispersions of silica particles in the polymer matrix were prepared by diluting Ludox (200 µL) with distilled water containing 1-30 wt % of PEO to produce a 3 wt % SiO2 colloidal dispersion at pH 8.5. The dispersions were homogenized by continuous stirring for 24 h. The final concentration of organic modifiers in the silica sol ranged from 0.01 to 30 wt %, as specified in each case. The pH of the solution was checked to ensure that it remained constant during the modification process and (9) Raghavan, S. R.; Walls, H. J.; Khan, S. A. Langmuir 2000, 16, 7920. (10) Hester, J.; Banerjee, P.; Mayes, A. Macromolecules 1999, 32, 1643. (11) Golander, C.-G.; Herron, J. N.; Lim, K.; Claesson, P.; Stenius, P.; Andrade, J. D. In Poly(ethyleneglycol) Chemistry. Biotechnical and Biomedical Applications; Harris, J. M., Ed.; Plenum Press: New York, 1992; pp 221-245. (12) (a) Chen, Q.; Kenausis, G. L.; Heller, A. J. Am. Chem. Soc. 1998, 120, 4582. (b) Heller, J.; Heller, A. J. Am. Chem. Soc. 1998, 120, 4586. (c) Keeling-Tucker, T.; Rakic, M.; Spong, C.; Brennan, J. D. Chem. Mater. 2000, 12, 3695. (13) (a) Baker, G. A.; Jordan, J. A.; Bright, F. V. J. Sol.-Gel Sci. Technol. 1998, 11, 43. (b) Baker, G. A.; Pandey, S.; Maziarz, E. P., III.; Bright, F. V. J. Sol.-Gel Sci. Technol. 1999, 15, 37. (14) Bronshtein, A.; Aharonson, N.; Avnir, D.; Turniansky, A.; Alstein, M. Chem. Mater. 1997, 9, 2632. (15) Jin, W.; Brennan, J. D. Anal. Chim. Acta 2002, 461, 1.
Tleugabulova et al. subsequent fluorescence measurements. R6G (1 µM) was added to the dispersions prior to fluorescent measurements and incubated for at least 12 h before fluorescence measurements commenced. Steady-State Fluorescence Measurements. Fluorescence anisotropy measurements were performed using a SLM 8100 spectrofluorimeter (Spectronic Instruments, Rochester, NY) as described elsewhere.16 Single-point fluorescence anisotropy measurements were made at λex ) 495 nm and λem ) 551 nm. Appropriate blanks were subtracted from each of the intensity values (IVV, IVH, IHV, IHH) used to calculate the anisotropy values,17 and all fluorescence anisotropy values were corrected for the instrumental G factor to account for any polarization bias in the monochromators. The values reported represent the average of 5 measurements each on three different samples. Time-Resolved Fluorescence. Time-resolved fluorescence intensity and anisotropy decay data were acquired in the timedomain using an IBH 5000U time-correlated single photon counting fluorimeter, as described elsewhere.7 The intensity decay of R6G could be fit to a single decay time for all samples tested. For anisotropy decay analysis, the experimentally obtained parallel and perpendicular fluorescence decays were used to generate the sum, S(t), difference, D(t), and time-resolved anisotropy, r(t), functions as follows:18-20
r(t) )
IVV(t) - IVH(t) IVV(t) + 2IVH(t)
)
D(t) S(t)
(2)
The anisotropy decay was fit to a two-component hindered rotor model according to the following equation:21
r(t) ) fr0 exp(-t/φ1) + (1 - f - g)r0 exp(-t/φ2) + gr0 (3) where φ1 reflects rapid rotational motions associated with rotation of free probes in solution, φ2 reflects slower reorientation of probes bound to silica particles, f is the fraction of fluorescence originating from free probe in solution, (1 - f - g) is the fraction of fluorescence owing to the local motion of adsorbed probe, g is the fraction of fluorescence due to probe that is rigidly bound to larger particles that rotate more slowly than can be measured with the R6G probe (φ > 60 ns), and r0 is the limiting anisotropy. In some cases, the value of gr0 is denoted as r∞, the residual anisotropy. Fits were considered acceptable if the reduced chi-squared (χR2) was close to 1.0 and the residuals showed no clearly nonrandom pattern. Viscosity Measurements. The kinematic viscosity of aqueous polymer solutions was determined from the transit time of the liquid meniscus through a capillary measured with a precision of (0.1 s in a Schott-Gera¨te AVS 350 automatic Ubbelohde viscosimeter. The viscosimeter was immersed in a bath, and the precision of the temperature control in all the measurements was (0.05 °C. Each measurement was repeated five times with a maximum deviation of less than 0.4%. The dynamic viscosity was calculated by multiplying the kinematic viscosity by the corresponding density. Atomic Force Microscopy. AFM imaging was performed using a Digital Instruments NanoScope IIIa Multimode instrument. The Ludox samples (3 wt % SiO2) were modified with 1 wt % of 0.55 or 10 kDa PEO and then diluted (1/30 v/v) with distilled water. One drop of the diluted sample (0.1 wt % SiO2) was placed onto an electronics grade silicon substrate (mean roughness < 0.150 nm over a 5 µm × 5 µm area) and allowed to dry. Height data were collected from a 250 nm scan area in tapping mode (16) Zheng, L.; Reid, W. R.; Brennan, J. D. Anal. Chem. 1997, 69, 3940. (17) Lakowicz, J. R. Principles of Fluorescence Spectroscopy, 2nd ed.; Plenum Press: New York, 1999. (18) Demas, J. N. Excited-State Lifetime Measurements; Academic Press: New York, 1983. (19) Lakowicz, J. R.; Gryczynski, I. In Topics in Fluorescence Spectroscopy; Lakowicz, J. R., Ed.; Plenum: New York, 1991; Vol. 1, Chapter 5. (20) (a) Bright, F. V.; Betts, T. A.; Litwiler, K. S. CRC Crit. Rev. Anal. Chem. 1990, 21, 389. (b) Bright, F. V. Appl. Spectrosc. 1995, 49, 14A. (21) Geddes, C. D.; Karolin, J.; Birch, D. J. S. J. Phys. Chem. B 2002, 106, 3835.
Monitoring Solute Interactions with Nanoparticles
Langmuir, Vol. 20, No. 1, 2004 103
Table 1. Anisotropy Decay Data for R6G in Polymer Solutions solvent 0.5 kDa PEG τ, nsa χR2 φ1, ns φ2, ns f (1 - f - g) r0 r∞ g η1 η2 η3, Pa s × 103 η0, Pa s × 103 χR2
10 kDa PEG
100 kDa PEO
0.1 wt %
1 wt %
30 wt %
0.1 wt %
1 wt %
30 wt %
0.1 wt %
1 wt %
3.88 ( 0.01 1.25 0.16 ( 0.01
3.88 ( 0.01 1.25 0.24 ( 0.01
3.86 ( 0.01 1.23 0.15 ( 0.01
3.89 ( 0.01 1.17 0.20 ( 0.01
3.88 ( 0.01 1.23 0.26 ( 0.01
0.992
0.990
0.973
0.980
0.980
0.376 0.001 0.003 0.9
0.368 0.003 0.008 1.3
0.385 0.004 0.010 0.9
0.330 0.009 0.027 1.1
0.300 0.006 0.020 1.1
0.356 0.007 0.020 1.5
0.9 0.9 1.01
1.3 1.0 0.95
0.9 0.9 0.90
1.1 1.1 1.07
3.94 ( 0.01 1.08 0.28 ( 0.03 0.95 ( 0.07 0.208 0.739 0.359 0.019 0.053 1.6 5.3 4.5 6.1 1.00
3.84 ( 0.01 1.22 0.21 ( 0.01
0.997
3.87 ( 0.01 1.31 0.21 ( 0.02 0.86 ( 0.02 0.091 0.887 0.400 0.009 0.022 1.2 4.7 4.2 5.1 1.02
1.1 1.0 0.90
1.5 3.0 0.99
a τ, fluorescence lifetime; φ and φ , rotational correlation times; f, fractional contribution to anisotropy decay owing to φ ; (1 - f - g), 1 2 1 fractional contribution to anisotropy decay owing to φ2; r0, limiting anisotropy; r∞, residual anisotropy; g, fraction of fluorescence arising from slow motion of probes corresponding to limiting anisotropy (g ) r∞/r0); η1 and η2, viscosity calculated from φ1 and φ2, respectively; η3, total viscosity (η3 ) η1(1 - f - g) + η2f); η0, experimental bulk viscosity.
using an Olympus silicon tapping probe (tip radius < 10 nm) operated at a 2.44 Hz scan rate. Particle sizes were determined from regions where single particles were present and are based on height differences between the top of the particle and the substrate.
Results and Discussion Aqueous Polymer Systems. The interpretation of time-resolved anisotropy results depends strongly on the nature of the probe and all possible interactions it can undergo in a chosen system. The occurrence of strong interactions between the probe and any component present in the system can drastically alter the fluorescence properties, particularly the rotational dynamics. For example, in silica sols, R6G is bound to the silica particles and its rotational dynamics primarily reflects the rotation of silica spheres.5 Similarly, the rotational dynamics of rhodamine B covalently attached to a cationic polyelectrolyte is much slower than that of the free probe in water because it reflects the rotation of a probe-polymer segment and of a portion of the polymer coil in solution, rather than the mobility of free dye.22 To assess such interactions in our system, the fluorescence properties of R6G were examined in the presence of the polymer alone prior to examination of polymer/Ludox systems. There were no obvious changes in the absorbance maximum, emission intensity, or quantum yield of the probe in the presence of any of the polymers, regardless of molecular weight or weight percent. However, the steady-state and timeresolved anisotropy of the probe was quite dependent on the presence of the polymer. As shown in Table 1, the fluorescence anisotropy of R6G in pure water and in 0.1-1 wt % PEO solutions fit well to a single-exponential function with a lifetime of ∼3.8 ns and a rotational correlation time of 0.16 ( 0.01 ns in water and ∼0.20 ns or higher in polymer solutions, with the rotational correlation time generally increasing with polymer concentration, as expected. However, upon increasing the polymer concentration up to 30 wt % for either the 0.55 or 10 kDa PEO, the anisotropy decay data could no longer be adequately fit by a single-exponential component (note: higher concentrations of 100 kDa PEO were not tested owing to the limited solubility of 100 kDa PEO in water). In such cases, monoexponential fits led to (22) Smith, T. A.; Irwanto, M.; Haines, D. J.; Ghiggino, K. P.; Millar, D. P. Colloid Polym. Sci. 1998, 276, 1032.
nonrandom residual plots and unrealistically low r0 values, as shown in Figure 1. On the other hand, fitting the anisotropy decay data to two subnanosecond rotational components led to random residual plots and reasonable r0 values (Figure 1). In 30 wt % solution, the faster rotational component was in the range of ∼0.25 ns, consistent with the rotational correlation times of R6G in dilute polymer solutions. However, this component only accounted for 10-20% of the total fluorescence. The majority of the fluorescence arose from a longer rotational component of ca. 0.90 ns (see Table 1), consistent with probe that was located in higher viscosity domains that were likely present within solvent-filled pores in the polymer network. For PEO with Mw ) 10 and 100 kDa, the radius of gyration of the polymer coil is 3 and 10 nm, respectively,23 which should lead to the rotation of whole coils on the order of a few or tens of nanoseconds, respectively. Hence, the two picosecond components in the anisotropy decays of R6G in PEO solutions cannot be due to the rotation of dye that is rigidly bound to the polymer chains. The absence of slow rotational motion on the scale of a few nanoseconds or slower and the low r∞ values observed in the decays of R6G in PEO solutions indicate that there is no direct binding of the probe to the polymer network. Based on this assumption, the observed anisotropy decays likely reflect the rotation of R6G within solvent-filled spaces encompassed by the polymer network. The rotational mobility of R6G in dilute PEO solutions is slowed by a factor of approximately 2 owing to collisions with single polymer coils and also by cooperative motions of the polymer molecules due to the creation and annihilation of the network by the fluctuations in the polymer backbone. In concentrated PEO solutions, the rotation of R6G is further slowed by ∼6-fold (relative to pure water) due to the formation of polymer clusters above a critical selfassociation concentration.24,25 Such aggregates of PEO have been detected even in diluted solutions, as shown by light scattering, electron microscopy, viscometry, spectroscopic techniques, sedimentation velocities, and gel (23) Singh, S.; Khulbe, K. C.; Matsuura, T.; Ramamurthy, P. J. Membr. Sci. 1998, 142, 111. (24) Polverari, M.; van de Ven, T. G. M. J. Phys. Chem. 1996, 100, 13687. (25) Atkins, D. T.; Ninham, B. W. Colloids Surf. 1997, 129, 23.
104
Langmuir, Vol. 20, No. 1, 2004
Tleugabulova et al.
Figure 1. Residual plots and recovered fitting parameters for monoexponential and biexponential fits to anisotropy decay data for a solution containing 30 wt % of 10 kDa PEO.
permeation chromatography.26-28 Thus, the R6G molecule can be assumed to be distributed between a less dense region similar to that found in dilute polymer solutions (faster component) and a more dense region associated with polymer clusters (slower component). The presence of two rotational components in polymer solutions containing a fluorescent probe has been previously reported.29,30 We cannot exclude the possibility of many other intermediate viscosity regions between these extreme cases, which would lead to a distribution of anisotropy decay times. However, such distributions would require fitting of anisotropy decays to a distribution of rotational correlation times, which is beyond the capabilities of our fitting programs.31 Polymer clusters coexist in a thermodynamic equilibrium with free polymer coils and can be removed from aqueous solutions, at least temporarily, by filtration through a 0.22 µm membrane filter.24 If our assumption that the slower rotational component of R6G is related to larger and more densely packed polymer clusters is true, filtration of the polymer solution should result in a reduction in the fractional contribution arising from such clusters. When solutions containing 1-8 wt % of 10 kDa PEG were filtered and analyzed by time-resolved anisotropy, there was no detectable change in the rotational characteristics of R6G (all decays were monoexponential). However, when the 30 wt % PEG solution was filtered through the 0.45 and 0.22 µm membrane filters, both the value and fractional contribution of the slower rotational correlation time were markedly decreased, as shown in Table 2, consistent with removal of polymer clusters. This experiment demonstrates that the rotational dynamics of R6G is sensitive to the presence of polymer clusters. The absence of strong interactions between R6G and PEG chains in water, as indicated by low r∞ values (Table 1), makes this probe a good choice for measurement of local microviscosity. However, in our particular case, the viscosity is calculated from the rotational correlation time (26) Nonionic Surfactants; Schick, M. J., Ed.; Marcel Dekker: New York, 1967. (27) Cox, J. E.; Dunlop, E. H.; North, A. M. Nature 1974, 249, 243. (28) Dunlop, E. H.; Cox, L. R. Phys. Fluids 1977, 20, 10 (II) S203. (29) Sen, S.; Sukul, D.; Dutta, P.; Bhattacharyya, K. J. Phys. Chem. B 2002, 106, 3763. (30) Shirota, H.; Segawa, H. J. Phys. Chem. A 2003, 107, 3719. (31) van der Sijs, D. A.; van Faassen, E. E.; Levine, Y. K. Chem. Phys. Lett. 1993, 216, 559.
Table 2. Anisotropy Decay Data for R6G in 10 kDa PEG (30 wt %) before and after Filtration
τ, nsa χR2 φ1, ns φ2, ns f (1 - f - g) r0 r∞ g χR2 a
before filtration
filtered through 0.45 µm
filtered through 0.22 µm
3.94 ( 0.01 1.08 0.29 ( 0.03 0.95 ( 0.07 0.208 0.739 0.359 0.019 0.053 1.00
3.95 ( 0.01 1.25 0.26 ( 0.07 0.72 ( 0.15 0.271 0.697 0.312 0.010 0.032 0.95
3.96 ( 0.01 1.19 0.50 ( 0.03 0.990 0.381 0.004 0.010 0.98
Abbreviations are as shown in Table 1.
of a small fluorescent dye in polymer solutions with substantially more spatial conformations than are found in aqueous mixtures containing a low molecular weight solute. According to our calculations using eq 1, there are two major viscosity domains (1-1.7 Pa s × 103 and 4-5 Pa s × 103, respectively) in concentrated PEO solutions. Upon calculation of the average viscosity based on the rotational characteristics of the dye in each of these two viscosity domains, the average viscosity is observed to increase with the polymer concentration, as expected. However, the values are slightly different than those measured for the respective polymer solutions with a bulk viscosimeter. In the latter case, the fluidity of polymer solution is measured, whereas the viscosity calculated from the rotational mobility of R6G reflects friction caused by the polymer network on the free rotation of a small molecule located in the solvent-filled voids in the polymer solution. This explains the observed difference and highlights the importance of the nanoparticle metrology approach for the determination of microviscosity in polymer solutions. Given that the R6G probe does not interact directly with PEO in solution, it is likely that the introduction of the polymer to Ludox solutions will not directly affect the rotational properties of the R6G probe, with the exception of viscosity-based effects. Thus, it is expected that changes in the rotational properties of R6G upon addition of the polymer will reflect polymer-Ludox interactions, which block the adsorption of R6G to the silica surface. Time-Resolved Anisotropy of R6G in PolymerLudox Solutions. Ludox particles are stable and scarcely
Monitoring Solute Interactions with Nanoparticles
Langmuir, Vol. 20, No. 1, 2004 105
Figure 2. Decays of fluorescence anisotropy for R6G in surface-modified and unmodified Ludox suspensions. Panel a: unmodified Ludox. Panel b: 30 wt % of 0.55 kDa PEO. Panel c: 30 wt % of 10 kDa PEO. Panel d: 1 wt % 100 kDa PEO.
precipitate when in suspension despite a high specific gravity of 2.2.32 The particle buoyancy is a result of the strong negative charge and the thick hydration layer on the silica surface. However, if the modification of the particle surface leads to neutralization of the surface charge, flocculation is expected, making the sample highly scattering and thus unsuitable for time-resolved fluorescence measurements. Thus, the effect of different polymers on the flocculation of Ludox was first tested and timeresolved fluorescence measurements were performed on samples containing the highest level of each given modifier that could be used before flocculation occurred. In all cases, the polymer-containing Ludox formed nonflocculated sols that were transparent and showed no visual evidence of phase separation, precipitation, or increased turbidity over a long period of time. R6G exhibited essentially the same fluorescence lifetime, extinction coefficient, and quantum yield in water and PEO solutions, in the absence or presence of Ludox, indicating that the fluorescent properties of the silicabound and free probe were the same. Thus, the fractional fluorescence values corresponding to bound and free probe are likely to reflect the true distribution of the probe in the Ludox solution. However, there were slight changes in the spectral characteristics of R6G after the addition of polymers to the Ludox sol. The absorption maximum for R6G was found at 527 ( 1 nm in diluted polymer solutions and unmodified Ludox but shifted to 530 nm in the Ludox containing 30 wt % PEO. Accordingly, the position of the emission maximum (551 nm) was essentially unchanged in diluted polymer solutions and unmodified Ludox and shifted to 555 nm in Ludox containing 30 wt % PEG indicating specific hydrogen (32) Iler, R. K. The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties, and Biochemistry; Wiley-Interscience: New York, 1979; p 348.
bonding interactions between the dye and the polymersilica composite. Ludox at moderately high concentrations can be used as an efficient light scattering solution to obtain instrument response functions for fluorescence lifetime measurements. Such scattering can lead to the presence of a very short lifetime component (a few picoseconds) and may also distort the data obtained in the VV channel during anisotropy decays, producing inaccurate data if not accounted for properly. To minimize problems related to light scattering, we limited the time-resolved measurements exclusively to optically transparent, homogeneous samples. Furthermore, in addition to monochromators, a short-pass filter was placed in the excitation path and a long-pass filter in the emission path to remove any directly scattered light from the light-emitting diode (LED) source. Using these conditions, our Ludox samples did not show any significant scattering signal above background using excitation at 495 nm and emission at 551 nm, even when VV polarization conditions were utilized. Figure 2 shows typical decays of fluorescence anisotropy for R6G in unmodified and modified Ludox. In all cases, it is clear that the anisotropy does not decay to a value of zero, indicative of a significant fraction of dye that is rigidly bound to silica structures that do not rotate on a time scale of at least 15τ (ca. 60 ns).5 Given that a significant residual anisotropy was evident in all anisotropy decays, the data were fit to a two-component hindered rotor model to account for the fraction of rigidly bound probe (fitting to a model that did not account for a residual anisotropy resulted in unrealistically long rotational correlation times on the order of several microseconds or more). Tables 3 and 4 show the fits to the R6G intensity and anisotropy decays in polymer-Ludox solutions. Several features merit special attention. First, the intensity decays
106
Langmuir, Vol. 20, No. 1, 2004
Tleugabulova et al.
Table 3. Anisotropy Decay Data for R6G in Ludox Sol Containing 10 kDa PEG solvent: Ludox/10 kDa PEG τ, nsa χR2 φ1, ns φ2, ns f (1 - f - g) r0 r∞ g η1, Pa s × 103 χR2
0 wt %
1 wt %
2 wt %
5 wt %
15 wt %
30 wt %
3.99 ( 0.01 1.18 0.16 ( 0.09 5.63 ( 0.61 0.006 0.194 0.319 0.251 0.80 0.9 0.92
4.04 ( 0.01 1.05 0.20 ( 0.06 4.14 ( 0.35 0.025 0.276 0.333 0.226 0.679 1.1 0.90
4.04 ( 0.01 1.12 0.43 ( 0.11 3.89 ( 0.36 0.020 0.217 0.325 0.248 0.763 2.4 0.87
4.02 ( 0.01 1.05 0.34 ( 0.07 3.41 ( 0.30 0.043 0.266 0.363 0.252 0.694 1.9 0.93
3.97 ( 0.01 1.11 0.36 ( 0.02 3.17 ( 0.12 0.100 0.327 0.351 0.201 0.573 2.0 0.95
3.91 ( 0.01 1.22 0.37 ( 0.04 2.47 ( 0.29 0.191 0.432 0.321 0.121 0.377 2.1 0.96
a Abbreviations are as shown in Table 1. η , viscosity calculated according to eq 1, based on φ , molecular radius of 0.56 nm for R6G, 1 1 and a temperature value of 298 K.
Table 4. Anisotropy Decay Data for R6G in Ludox Sol Containing Different Polymers solvent Ludox/100 kDa PEO τ, nsa χR2 φ1, ns φ2, ns f (1 - f - g) r0 r∞ g χR2 a
Ludox/0.55 kDa PEG
0.1 wt %
1 wt %
1 wt %
30 wt %
3.96 ( 0.01 1.16 0.12 ( 0.03 4.36 ( 0.60 0.040 0.178 0.348 0.272 0.782 1.83
3.99 ( 0.01 1.26 0.24 ( 0.16 3.52 ( 0.56 0.050 0.413 0.354 0.190 0.537 1.00
3.96 ( 0.01 1.27 0.16 ( 0.03 2.74 ( 0.30 0.090 0.395 0.344 0.177 0.515 0.90
3.85 ( 0.01 1.28 0.55 ( 0.02 2.76 ( 0.40 0.470 0.350 0.301 0.054 0.180 0.88
Abbreviations are as shown in Table 1.
of R6G were best fit to a single fluorescence lifetime of ∼4 ns in all samples, in agreement with previous reports.7 Second, all samples show two correlation times and a residual anisotropy value, consistent with the probe being distributed between multiple environments (free, partially bound, and rigidly bound states or bound to particles of varying size).5 Third, the ratio of residual anisotropy values to limiting anisotropy values (g ) r∞/r0) is dependent on sample modification, generally decreasing after modification with polymer species but decreasing most when high amounts of low molecular weight PEO were used. Fourth, the fractional fluorescence associated with the nanosecond rotational component increases upon polymer modification. Finally, the value of the nanosecond decay component decreases upon polymer modification. Each of these points is discussed in more detail below. Considering first the unmodified Ludox samples, the R6G probe shows fractional fluorescence values of 80% for rigidly bound probes that rotate with large particles (g ) 0.8), 19.4% for probes that rotate with a 5 ns correlation time (discussed below), and only 0.6% for free probe (0.16 ns correlation time). The 5 ns correlation time is far too long to be associated with free probe, even if such probe is present in polymer-rich solution; hence, the data are consistent with greater than 99% of the probe being adsorbed to silica. Previous studies of rotational anisotropy decays within Ludox have demonstrated that as much as 95% of the fluorescence from Ludox-associated probes corresponds to an ∼400 ns rotational correlation time (associated with particles having diameters in the range of 6-7 nm), with only a small fraction of a 5 ns component being present.5 However, these studies were conducted with the longer lifetime probe CG437, which has solvent-dependent biexponential fluorescence decay times of 8.5-14 and 25-34 ns in a 30% aqueous Ludox
sol. In our system, the probe lifetime is on the order of 4 ns and thus is more sensitive to faster rotational correlation times, which may explain the higher fractional fluorescence originating from the 5 ns decay component. The polydispersity of Ludox is only about 10-15% in water, which is not likely to lead to a clearly detectable multiexponential decay. Furthermore, as described below, the fractional contribution from the nanosecond decay component can be greater than 40% in some polymer-Ludox samples. Hence, we do not attribute the 5.6 ns component to a fitting artifact. The origin of the 5 ns component in the anisotropy decays of R6G associated with Ludox is at present unknown. However, possible explanations may include the presence of small particles (1-3 nm radius), a contribution from local rotational motion (i.e., wobble motions21) of probes that are bound to larger particles, or an average rotational correlation time corresponding to a fraction of dye that rapidly associates and dissociates with the silica surface on the time scale of the fluorescence decay process. Based on the results of AFM studies (described below) and the effects of polymer adsorption on the value and fractional contribution of the nanosecond rotational correlation time, we tentatively assign this component to local motion of dye which may be adsorbed through either electrostatic or hydrogen-bonding interactions with the silica. As noted above, the addition of polymer to the Ludox samples causes the g values to decrease, the value of the nanosecond decay component to decrease, and the fraction of this component to increase in a concentration-dependent manner. For example, considering the 10 kDa PEO samples (Table 3), it is apparent that the g value gradually decreased from 0.8 (plain Ludox) to 0.38 (30 wt % PEO), with the fractional fluorescence of the nanosecond component increasing from 20% to 43% and that of the picosecond component increasing from 0.6% to 19%. Since the emission properties of R6G do not appear to change upon adsorption, we assume that the total fluorescence from the nanosecond and residual anisotropy components (1 - f - g + g ) 1 - f) corresponds to the fraction of bound dye. Thus, the fraction of rigidly bound dye is reduced from >99% to 81% upon adsorption of the highest concentration of 10 kDa PEO to the Ludox suspension. As shown in Figure 3, the extent of blockage of dye adsorption, shown as the fraction of free probe, was dependent on both the molecular weight and the concentration of added polymer. More significant changes in the fraction of free dye were obtained after introduction of low molecular weight PEO to the Ludox samples. In this case, the g value decreased from 0.8 (plain Ludox) to 0.18 (30 wt % PEO), with the fractional fluorescence of the nanosecond component increasing from 20% to 35% and
Monitoring Solute Interactions with Nanoparticles
Langmuir, Vol. 20, No. 1, 2004 107
Figure 3. Fractional contributions from free probe to the total anisotropy decay of R6G in polymer-containing Ludox sols (3 wt % SiO2) determined from time-resolved measurements: (b) 10 kDa PEO modified Ludox; (9) 0.55 kD PEO modified Ludox.
that of the picosecond component increasing from 0.6% to 47% (see Table 4). More importantly, the majority of the changes occurred at low polymer concentrations in the case of 0.55 kDa PEO. These data clearly show that the lower molecular weight polymer is better able to block the adsorption of the dye to the silica surface and thus suggest that the smaller polymer chains are better able to coat the silica surface. This is not unexpected, as the larger polymer would lose more degrees of freedom relative to a smaller polymer if all segments adsorbed to the silica surface. An unexpected finding from the anisotropy measurements was that the fractional fluorescence originating from dye with a nanosecond rotational correlation time increased at the expense of fluorescence from rigidly bound dye. In addition, the value of the nanosecond component decreased significantly with increasing PEO concentration. For example, the rotational correlation time decreased from 5.6 ( 0.6 ns (19.4% contribution) in unmodified Ludox to 2.5 ( 0.3 ns (43% contribution) in Ludox containing 30 wt % of 10 kDa PEO. Similar changes in the rotational correlation times with increasing polymer concentration were observed in solutions of PEO of different molecular weight. One explanation for this behavior is that the two rotational correlation times that are extracted from the fit are actually average values that emerge from the combination of three correlation times (low picosecond component for free probe, long picosecond component for probe in polymer clusters, nanosecond component for dye adsorbed to particles). However, such a situation should exist only for high polymer concentrations where the two picosecond components were observed in solution. In our systems, significant decreases in the value of the nanosecond decay component are observed even at 0.1-1 wt % of polymer, where the anisotropy decay of the polymer solution is monoexponential. Furthermore, all fits to the polymer/Ludox anisotropy decays produced reasonable r0 values and random residual plots, which would not likely be the case if there was an underlying three-component decay. Thus, it is unlikely that the observed decreases in the nanosecond anisotropy decay component are a simple fitting artifact. Overall, the changes in the value and fractional contribution of the nanosecond decay component suggest that the bound dye adsorbs more weakly or perhaps in a different orientation when polymer is present and thus displays a higher degree of local motion.
Figure 4. Atomic force microscopy images of unmodified and PEO-modified Ludox suspensions. Panel A: unmodified Ludox. Panel B: 1 wt % of 10 kDa PEO. Panel C: 1 wt % of 0.55 kDa PEO.
Atomic Force Microscopy. Atomic force microscopy experiments were performed to further assess the origin of the nanosecond rotational decay component (i.e., adsorption to small particles vs local motion of dye bound to larger particles). Figure 4 shows AFM images obtained for unmodified, 0.5 kDa PEO modified, and 10 kDa PEO modified Ludox suspensions that were cast on silicon surfaces. As shown in panel A, the unmodified Ludox is composed predominantly of particles with radii between 6 and 7 nm, consistent with the expected presence of secondary particles.5 Topological analysis of the images provided no evidence for the presence of smaller primary
108
Langmuir, Vol. 20, No. 1, 2004
Tleugabulova et al.
Figure 5. Height and phase contrast AFM images of Ludox modified with 1 wt % of 0.55 kDa PEO.
particles (1-3 nm radius), which supports our interpretation of the ∼5 ns rotational correlation being due to local motion of the dye bound to large (>6 nm radius) particles. Addition of 10 kDa PEO to Ludox (panel B) results in what appear to be relatively minor changes in the particle distribution. In this case, there is evidence for particle aggregation (based primarily on height data). However, there are also a significant number of discrete particles that appear not to be modified to a significant degree. Such particles would be expected to be able to adsorb R6G onto their surface, resulting in the presence of significant amounts of bound probe, as was observed. Upon addition of 0.55 kDa PEO (panel C), the Ludox particles appear to change from spherical to irregular shapes, and a halo appears to surround the particles which is likely due to adsorbed polymer. Indeed, the image appears to be consistent with Ludox particles embedded in a polymer phase. There is evidence for significant aggregation of particles, and there is an almost complete absence of discrete particles. Thus, the image is consistent with extensive coverage of the silica with the polymer, which would reduce the overall exposed surface area and result in poor adsorption of R6G, consistent with our fluorescence results. To further examine the morphology of the 0.55 kDa PEO-Ludox sample, both height and phase contrast AFM images were obtained, as shown in Figure 5. The contrast mechanism in the phase contrast image is due to variations in the mechanical properties of the sample. In this case, light areas are harder than dark areas. Thus, the phase image clearly shows the presence of a PEO film (dark regions) between the silica particles (light regions). Given that the molecular weight of the PEO is low, it is expected that the polymer should be able to form a layer through which the underlying structure can easily be sensed by the vibrating probe used to obtain the phase contrast images. The presence of a darker ring around each spherical particle in the height images suggests that the PEO makes some specific contact angle with the beads such that a wetting layer is present between the silica particles. This would result in the dark bands around the
particles (as seen in the height images), and it is also consistent with the phase images. Taken together with the fluorescence data, it is clear that the 0.55 kDa PEO adsorbs as a continuous film over the surface of the silica beads, thus reducing the ability of the silica to interact with R6G that is present in solution. Conclusions The extent of binding of a fluorescent cationic solute, rhodamine 6G, to the surface of silica particles can be examined using time-resolved fluorescence anisotropy decay measurements and can be used to probe the adsorption of nonionic polymers to the silica surface. Our data show that in all cases, the dye is distributed between bound and free states, with the bound dye displaying both local motion and global motion of the silica nanoparticles. The degree of R6G adsorption generally decreased upon addition of polymer to the silica, consistent with polymer adsorption onto the silica surface. The most extensive blockage of R6G adsorption was obtained by adding low molecular weight PEO, which had a pronounced effect on R6G adsorption even at low loading levels. In optimal cases, up to 50% of the probe could be blocked from adsorbing on the surface. The addition of such agents also altered the degree of local motion of the dye, consistent with an alteration in the strength of the dye-silica interaction. The data clearly show the power of timeresolved fluorescence anisotropy decay measurements for probing the modification of silica surfaces and should prove useful in characterization of new chromatographic stationary phases. Acknowledgment. The authors thank MDS-Sciex, the Ontario Ministry of Energy, Science and Technology, the Canada Foundation for Innovation, and the Ontario Innovation Trust for support of this work. We also thank Dr. Alicja M. Mika and Dr. Zhang Aheng from McMaster University for helpful discussions. J.D.B. holds the Canada Research Chair in Bioanalytical Chemistry. LA035333Z
