Journal of Colloid and Interface Science 452 (2015) 215–223
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Surface activity and flocculation behavior of polyethylene glycol-functionalized silica nanoparticles Sanna Maria Sofi Björkegren a,b,⇑, Lars Nordstierna a, Anders Törncrona b, Michael E. Persson a,b, Anders E.C. Palmqvist a a b
Chemistry and Chemical Engineering, Chalmers University of Technology, SE-412 96 Göteborg, Sweden AkzoNobel Pulp and Performance Chemicals AB, SE-445 80 Bohus, Sweden
g r a p h i c a l a b s t r a c t
a r t i c l e
i n f o
Article history: Received 23 February 2015 Accepted 20 April 2015 Available online 27 April 2015 Keywords: PEGylated colloidal silica NMR diffusometry Surface activity Dynamic light scattering (DLS)
a b s t r a c t Colloidal silica nanoparticles have been functionalized with methyl polyethylene glycol silane (mPEG silane) and the PEGylated particles have been characterized with focus on exploring their surface chemical properties. The degree of surface functionalization was quantified using NMR diffusometry, and the measurements showed that the silane binds covalently to the silica surface. Samples with surface coverages ranging from 0.068 to 0.315 lmol silane/m2 have been analyzed. The functionalized particles proved to be surface active and showed a significant reduction in surface charge and zeta potential with increasing degree of PEG functionalization. All samples showed colloidal stability at neutral pH and above within the range studied. At lower pH, the samples with low surface coverage displayed a reversible flocculation behavior, while samples with a high surface coverage and samples without functionalization remained stable. This suggests that steric stabilization is effective at low pH when the surface coverage is high enough; electrostatic stabilization is effective for samples without functionalization; and that inter-particle PEG–silica interactions cause flocculation of particles with too low degrees of PEG functionalization. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction ⇑ Corresponding author at: Chemistry and Chemical Engineering, Chalmers University of Technology, SE-412 96 Göteborg, Sweden. E-mail address: [email protected] (S.M.S. Björkegren). http://dx.doi.org/10.1016/j.jcis.2015.04.043 0021-9797/Ó 2015 Elsevier Inc. All rights reserved.
Dispersions of colloidal silica, also referred to as silica sols, are utilized in numerous industrial processes and commercial products, such as retention aid in paper making, as additives in paints,
216
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
as binder in foundry applications, for beverage clarification, and in polishing applications [1]. In all these applications the nature of the silica surface and the ability to control the aggregation of the colloidal particles are of great importance. Studies concerning the surface chemistry of amorphous silica have previously been undertaken [2–4]. The surface chemistry is complex and further characteristics remains to be understood. To further expand the uses of silica, and to improve the use in existing applications, development of new composite silica materials has been extensively explored. Surface modification is a common approach to customizing the silica material properties, which requires further investigations of the material in terms of characterization and evaluation methods. Publications addressing the grafting of poly (ethylene glycol) (PEG) or poly(ethylene oxide) (PEO) onto silica surfaces are frequently appearing, often with uses in biomedical applications due to the biocompatibility of the material [5–9]. PEG-containing coatings is another field of interest, where for example Malmsten et al. [10] found that a high enough interfacial density of PEG on flat silica surfaces results in efficient protein rejection. Leckband et al. studied intermolecular forces and show that grafted PEG chains can exist in both protein-repulsive and protein-attractive states, depending on factors such as compression and polymer chain length [11]. The grafting of PEO onto the surface of silica particles was reported already in the 1980s by Bridger and Vincent [12]. They investigated two methods for terminal grafting of PEO chains of which one was suitable for aqueous conditions, using in situ grafting of isocyanate-capped PEO during particle formation (by Stöber method [13]). However, the modified particles obtained had a low colloidal stability in water [12]. Other examples are Zhang et al. [14] who prepared PEGylated silica particles in methanol addressing both chemical and colloidal stability issues in water and Xu et al. [15] who obtained PEG-coated silica through synthesis in a methanol–ammonia mixture. Joubert et al. [16] prepared PEO-grafted silica through graft polymerization; the polymerization of EO was initiated from the silica surface, onto which alcohol groups had been attached. Still, the grafting of PEG onto silica particles where the synthesis is carried out by a simple process in purely aqueous conditions is not as recurrent. In this paper we report on the characterization of colloidal silica particles functionalized with methyl end-capped tri-methoxy poly(ethylene glycol) silane (mPEG silane) prepared via a simple water-based route. A direct measurement of the mPEG silane attached to the silica particles has been employed using NMR diffusometry, providing quantitative information of the grafting efficiency. The aggregation behavior of the PEGylated particles has been studied with dynamic light scattering (DLS) and UV–Vis spectroscopy. Further, the modified silica surface has been characterized through polyelectrolyte adsorption and zeta potential measurements. In addition, surface activity of the particles has been assessed, since the use of particles as stabilizers for emulsions has recently gained much interest [17,18]. This full study and characterization provides both quantitative information concerning the degree of surface functionalization and surface activity as well as qualitative information concerning flocculation behavior of surface modified colloidal silica.
2. Experimental section
a native pH of 9.1 and a concentration of 40 wt% silica. Silquest A-1230 (from Momentive), a tri-methoxy poly(ethylene glycol) silane end-capped with a methyl group and denoted mPEG silane, was used for the functionalization. The chemical structure of the mPEG silane is shown in Fig. 1 and was derived from standard high-resolution NMR studies (COSY, HMQC, HMBC), giving an average molecular weight of 679 g/mol. The polydispersity index of the mPEG is less than 1.2, as verified with HPLC measurements. For salt and pH adjustments reagent grade sodium hydroxide, sodium chloride and hydrochloric acid from Sigma–Aldrich were used. In addition, Amber-Jet 1500H from Dow was used for pH adjustments, a strong cation-exchange resin with sulfonic acid groups in the hydrogen form. 2.2. Methods 2.2.1. Functionalization with mPEG silane The mPEG silane was added to the silica suspension during agitation and at constant temperature (70 °C or room temperature) using a syringe pump. The synthesis was carried out at pH 9–10, since at this pH interval the desired condensation reaction as well as hydrolysis of the silane are fast [20]. In addition, this pH provides an electrostatically stabilized colloidal system where aggregation of the silica particles is avoided [4]. When the mPEG silane is added to alkaline water, the methoxy groups are hydrolyzed. The formed Si-OH groups of the mPEG silane become available for reaction with the silanol groups of the silica particle surface and through condensation reactions, siloxane bonds form. This results in covalent attachment of the mPEG silane to the silica surface. The mPEG silane has three functional groups, however, only up to two siloxane bonds to the surface can be formed [21], and adjacent silanes may well bind to each other. Hydrolyzed mPEG silane can also react in solution with one another. The samples were purified either by dialysis or by using an ultrafiltration cell from Millipore with an applied N2 pressure of 1.6 bar. The dialysis membranes (from Sigma–Aldrich) were of cellulose with a cutoff of 14 kDa. The ultrafiltration membranes of regenerated cellulose (from Millipore) had a cut-off of 100 kDa, removing free unreacted silanes from the suspensions. 2.3. Characterization and evaluation 2.3.1. NMR diffusometry NMR self-diffusion measurements were conducted on a Bruker Avance 600 spectrometer, equipped with Diff30 diffusion probe with a maximum gradient strength of 1200 G/cm and with a 5 mm RF insert. Using the conventional stimulated spin-echo sequence [22], the diffusion experiments were performed at 25 °C with diffusion time D = 100 ms, gradient pulse length d = 1.5 ms, gradient ringing delay sr = 1 ms and the gradient strength, g, ramped in at least 16 steps from 5 to 1200 G/cm. At each gradient step, the 1H spectrum was collected from 8 to 16 acquisitions dependent on sample concentration. The relaxation delay was set to 5 s. It was however noted that full longitudinal relaxation was not achieved during this time, why this was
H3 C
2.1. Materials
H3C Sodium ion-stabilized colloidal silica particles with the trade name Bindzil 40/130 were provided by AkzoNobel Pulp and Performance Chemicals. These anionic particles have a surface area of 130 m2/g, as measured by Sears titration [19], corresponding to an equivalent spherical diameter of 21 nm [4]. The suspension has
O O
Si
O O
11
CH3
O H 3C Fig. 1. Chemical structure of tri-methoxy poly(ethylene glycol) silane (mPEG silane) as derived from high resolution NMR spectroscopy.
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
compensated for during data processing (see Eq. (1) where sum of relaxation delay and acquisition time is given). Correct calibration of the gradient amplifier was controlled by obtaining the self-diffusion coefficient of HDO trace in a standard sample of pure D2O [23]. For all samples, and also from experiments with varying diffusion time, a strict two-component attenuation of the logarithmic NMR signal intensity against k = (cgd)2(D d/3) was noted, where c is the 1H gyromagnetic ratio. The data pattern allowed for a twocomponent fitting by standard non-linear regression using Eq. (1).
IðkÞ ¼ p1 expfkDf g þ p2 expfkDb g pf ¼
ð1Þ
p1 C b p1 C b þ p2 C f
pb ¼ 1 pf
Cf ¼
relax:delay þ aq:time 1 exp T 1;f ðD sr Þ 2ðd sr Þ exp exp T 1;f T 2;f
Instruments. The suspensions of functionalized silica particles were diluted in pH adjusted water solutions to 0.5 wt% silica, containing HCl or NaOH and compensated with the necessary amount of NaCl required to keep the electrolyte concentration constant throughout the variations in pH. Dynamic light scattering (DLS) particle size measurements were conducted on the same instrument. The colloidal stability of diluted suspensions, at 5 wt% silica, was investigated by adjusting the pH using hydrochloric acid, sodium hydroxide or cation exchange resin. The turbidity of the suspensions was measured with a UV–Vis spectrometer (HP 8453) in quartz cuvettes. The absorbance of the suspension was recorded at 700 nm, since the molecular absorbance of the system is low at this wavelength, and is assumed to be unchanged with the variation in pH [25]. An increase in absorbance, i.e. increased scattering of light is therefore due to an increase in turbidity. The samples were analyzed within 20 min from the pH adjustment, with some additional measurements of longer time intervals. 2.3.3. Surface activity Surface tension measurements of the colloidal systems were made using a Sigma 70 tensiometer employing the DuNoüy ring method. The platinum ring (radius 9.545 mm and wire-thickness 0.185 mm) was rinsed with distilled water and ethanol, and heated with a Bunsen burner in between measurements.
Cb ¼
217
relax:delay þ aq:time 1 exp T 1;b ðD sr Þ 2ðd sr Þ exp exp T 1;b T 2;b
3. Results and discussion As can be seen in Eq. (1) the spin–lattice (longitudinal) and spin– spin (transverse) relaxation times, T1 and T2 respectively, are required for both free (f) and bound (b) silane in order to quantify the relative amount of the two species. The relaxation times were obtained by standard inversion recovery and CPMG pulse sequences where both revealed strict bi-exponential behavior. Likewise to the self-diffusion coefficients, the relaxation times also provide information about local molecular mobility, e.g. a decreased value of T2 indicates reduced mobility [24]. It was noted that the diffusion coefficients were found to be of significantly higher relative difference compared to the relaxation times, comparing free and bound silane (two orders of magnitude for diffusion and one order of magnitude for relaxation). The superior sensitivity of diffusometry, for this particular system, thus argued for quantification by determination of the self-diffusion coefficients. 2.3.2. Particle charge, zeta potential and colloidal stability The apparent surface charge density of purified samples was assessed by particle charge density measurements with a Mütek Particle Charge Detector (PCD), equipped with an oscillating piston that detects the current between two electrodes. The charge was measured during titration with a Mettler-Toledo Autotitrator of the cationic polyelectrolyte polybrene (C13H30Br2N2). Polybrene has two equivalent charges per mol, and the titration is terminated once the suspension reaches the point of zero charge (PZC). The equivalent charge per mass (eq/g) was calculated according to Eqs. (2) and (3) and compared to the charge of the unmodified silica sol of the same size. The titrations were conducted on diluted suspensions (2 wt% silica), and the polybrene concentration was 3.748 g/L thus obtaining Npolybrene of 0.02 eq/L.
C polybrene concentration eq 2 mol MW polybrene
ð2Þ
V volume polybrene Npolybrene V volume silica suspension C silica concentration
ð3Þ
Npolybrene ¼
eq=g ¼
Zeta potential measurements of dialyzed samples, as a function of pH, were conducted on Zetasizer Nano-ZS from Malvern
3.1. Functionalization procedure Various syntheses were conducted, in order to investigate the effect of the reaction conditions on the yield in terms of bound mPEG silane in relation to the added amount of silane, and also to acquire particles with varying degree of surface coverage. The synthesis could be conducted at room-temperature. A slow addition rate, preferably below 0.25 lmol/(m2 h), was required to minimize self-condensation of the mPEG silanes. The maximum surface coverage obtained at the reaction conditions in this work was 0.315 lmol mPEG silane/m2 of silica particle surface, which corresponds to approximately 0.19 molecules per nm2. The surface coverage obtained is relatively low compared to the theoretically available 8 lmol silanol groups/m2 [26]. Steric hindrance of the PEG-chain from already attached silanes, the high solubility of the mPEG silane in water and self-condensation of the silane species reduce the reaction yield. 3.2. Determination of amount of PEG surface coverage The mPEG silane is water soluble and the PEG-chains protruding from the particle surface have low affinity to silica at alkaline conditions, where the silanol groups are dissociated and the surface is charged [27]. This makes NMR an excellent tool for evaluating the degree of functionalization, since a signal, of high sensitivity, from the repeating ethylene oxide units of the PEG chain could be obtained for both free and bound silane, see spectrum in Fig. 2A. Although the signal is an overlap of free and bound signals, it could be analyzed by two-component behavior using diffusometry. Fig. 2B shows experimental data points and the fitted lines of Eq. (1) profile of non-purified and purified PEGylated silica particles in aqueous suspension, and free aqueous mPEG silane. These particles were found to be grafted with 0.155 lmol mPEG silane/m2. The latter value was obtained from the fraction of bound silane, which was calculated, using Eq. (1), via the diffusion data together with the relaxation times T1 and T2 for free and grafted mPEG silane, respectively. The diffusion decay is clearly
218
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
Fig. 2. (A) 1H NMR spectrum of functionalized silica sol. (B) Stimulated spin echo profile from NMR diffusometry measurements of silica particles grafted with 0.155 lmol mPEG silane/m2 and of free aqueous mPEG silane (blue ). Before purification by ultrafiltration (black j), 17% of the silanes are attached to the particles, and after purification 75% of the silanes are attached to the particles (red ). The steeper slope of the filtered suspension is due to a more diluted sample. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
bi-exponential, suggesting a strong association or, most probably, covalent attachment of a fraction of the mPEG silane species. The covalent linkage was further substantiated by variation of the diffusion time D that provided identical results verifying a surface residence time significantly larger than several hundreds of milliseconds. Non-covalent adsorption would most probably result in shorter residence times. The strict mono-exponential appearance of the two joint but distinct diffusion patterns, respectively, combined with the reproduced results when varying the diffusion time, implies that the degree of adsorbed or motionally restricted molecules with long residence time is negligible in this work. Any influence on calculated diffusion coefficients, by such populations, is thus insignificant. The transverse relaxation time T2 is significantly reduced for bound mPEG silane, having a value of 16 ms compared to 797 ms for free aqueous silane and 259 ms for free silane in suspension. The longitudinal relaxation time T1 on the other hand is increased from 336 ms for free silane to 1.96 s for bound silane. This demonstrates that the bound fraction is partly immobilized and associated to the surface of the silica particle, the latter with a long rotational correlation time. The self-diffusion coefficients obtained are in the order of 3 1010 m2/s for free silane, and 5 1012 m2/s for bound, with variations due to differences in concentrations, i.e. viscosity and obstruction. Table 1 provides an overview of the results collected for the samples discussed in this work. Also the purified samples contain small amounts of free mPEG silane in aqueous solution. Both ultrafiltration and dialysis remove 70–80% of the unbound silanes, and dialysis is slightly more efficient than ultrafiltration. The condensation reaction that attaches the mPEG silane to the silica surface is in principle reversible, which could result in detachment of the Table 1 Overview of surface functionalized silica particles discussed in this work and the particle sizes as measured with DLS at alkaline conditions. The yield was calculated from the NMR diffusion data together with the relaxation times, from which the surface coverage, and subsequently the amount of free silane, is obtained. The synthesis temperature was 70 °C for all samples except 0.21mPEG, for which the synthesis was conducted at room temperature. Sample name
Unmodified 0.07mPEG 0.15mPEG 0.16mPEG 0.21mPEG 0.31mPEG
Surface coverage; bound mPEG silane (lmol/m2 SiO2)
(molecules/nm2)
– 0.068 0.155 0.162 0.215 0.315
– 0.04 0.09 0.10 0.13 0.19
Yield
– 25% 17% 18% 22% 17%
Diameter (nm) Z-average
Volume average
32 34 35 35 – 36
30.8 29.7 31.0 31.2 – 31.7
silane at conditions of high pH [20]. The NMR integral intensity of the signal from the bound mPEG silane in the purified samples was compared to the corresponding signal in the non-purified samples, to ensure that no loss of bound mPEG silane occurred during the purification steps. The intensity did not change, showing that the amount of bound silane is maintained intact during the purification process and that the reversibility of the condensation reaction has no significant influence. It also substantiates that the bound silane is covalently attached, since a loss of bound silane would be expected if the association is only physical adsorption. In addition, through analyzing purified samples that had been kept at high pH over a time period of months we found that further detachment did not yet occur over time. 3.3. Surface charge of functionalized silica Single, vicinal and geminal silanol groups as well as surface siloxanes are found at the silica surface, and physically adsorbed water molecules are hydrogen-bonded to all types of silanol groups [1]. The 8 lmol of silanol groups/m2 surface of fully hydroxylated amorphous silica can in water be ionized bearing a charge density that increases with pH; pKa of the SiOH is around 9.2 [4]. At the isoelectric point, reported to be found around pH 2, a silica sol is metastable, and above pH 7 the concentration of surface charge is sufficient to cause mutual repulsion between the particles, which provides stability toward gelling of the sol [1,4]. Functionalization of the silica surface, such as the PEGylation presented here, would yield a reduction of the surface charge density, and may induce steric stabilization thus allowing applications in a broader pH range. Partial coverage of the silica surface allows studies of how the pH influences particle–PEG interactions. 3.3.1. Measurements of the apparent surface charge Particle charge density (PCD) measurements show that the apparent surface charge density is indeed reduced and can be correlated to the surface coverage of the particles, see Fig. 3A. It indicates that the number of available silanol groups on the particle surface has decreased significantly; a reduction of surface charge above 60% is observed for the highest surface coverage. For comparison, silica functionalized with a shorter epoxy silane, having a surface coverage of about 2.3 lmol/m2, reduces the apparent surface charge with 80–85% [28]. This further implies that the mPEG chains from the attached silanes shield the surface silanol groups to such an extent that (i) it prevents further condensation of hydrolyzed mPEG silanes and (ii) it has a major effect on the surface properties even though the surface coverage is low. It should be
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
219
Fig. 3. (A) The apparent surface charge in meq/g as measured with PCD polyelectrolyte adsorption as a function of the surface coverage of the particles. The surface charge correlates well with the surface coverage of mPEG silane; the higher coverage yields a larger reduction of charge. (B) Apparent surface charge as measured with PCD polyelectrolyte adsorption of unmodified silica (black j), silica grafted with 0.155 lmol mPEG/m2 (green ) and 0.315 lmol mPEG/m2 (red ) monitored as a function of pH of the suspensions. The error bars represent the standard deviation obtained from three replicates. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
noted that the reduction of surface charge is indeed apparent, since the access to the silanol groups also depends on the type and size of the polymer used for the titration, in this case polybrene. However, the relative difference between the samples is still of interest. The PCD was also monitored as a function of pH, see Fig. 3B. The suspensions were pH adjusted prior to the measurements using hydrochloric acid. As for unmodified particles, the functionalized particles lose most of their charge between pH 9 and 7 and below pH 2 virtually all charge is gone. However, the decrease in surface charge for modified particles is not as steep as for the unmodified particles, again pointing to the effect on the surface properties of the bound mPEG groups. 3.4. Aggregation behavior of the mPEG modified silica particles That attractive forces exist between poly(ethylene oxide) (PEO) and silica are well-known [29]. This interaction depends on both pH and temperature [25], where an increasing pH decreases the adsorption [30]. In this work, strong pH dependence in terms of aggregation is observed for the mPEG functionalized silica particles having a surface coverage of 0.215 lmol mPEG silane/m2 and below, where the turbidity of the suspension increases with decreasing pH. If a turbid sample is left undisturbed at a low pH it settles and after a month when some of the water is allowed to evaporate it forms a gel. Samples were also analyzed with DLS to obtain the particle size distribution, as a complementary method to both the zeta potential and the turbidity measurements. With DLS, the hydrodynamic radius is calculated from the diffusion coefficient of the particles. At low pH values and high electrolyte concentrations particle–particle interactions may influence the diffusion. When aggregates become large (>1000 nm) and start to sediment, it is no longer possible to derive reliable results, which occurs for a surface coverage of 0.215 lmol mPEG silane/m2 and below, at low pH and/or high electrolyte concentration. It is also important to bear in mind that the radii obtained are estimates, with the purpose to demonstrate the difference between the samples at different conditions. In general the DLS measurements show that the increase in turbidity as measured with UV–Vis corresponds well to an increase in mean particle size. 3.4.1. Colloidal stability and PEG–silica interactions Fig. 4 shows the apparent absorbance (i.e. turbidity) as a function of pH, for A) purified and B) non-purified samples. It can be seen that for low (0.068 lmol/m2) and medium (0.155 lmol/m2)
surface coverages, the turbidity increases rapidly when the pH is below 6. Flocculation of silica, resulting in phase separation, due to adsorption of PEO polymers on the silica surface is known to be dependent on ionic strength, pH and also the amount of polymer adsorbed [31]. The non-purified samples containing unreacted mPEG silane have a similar flocculation behavior as the purified sample, why it can be concluded that adsorption of unreacted mPEG silane has no significant influence in terms of stabilization. Fig. 4C shows turbidity measurements performed while adjusting the pH using ion-exchange resin, maintaining the same ionic strength throughout the variations in pH. Flocculation occurred also when the sample was ion-exchanged, although the increase in turbidity was smaller and the process was slower. It is shown in Fig. 4D that the flocculation is reversible; an increase of pH from an acidic solution returned a flocculated system almost to its original turbidity. This means that the particles are stabilized also when the surface coverage is low in a way that allows the control of aggregation by adjusting the pH. This further indicates that the particles most likely form larger aggregates through weak bridge flocculation that, upon storage, rigidify and form a gel network. For the highest surface coverage obtained (0.315 lmol/m2) no increase in turbidity was seen neither by pH adjustment through ion-exchange nor by addition of acid. The size as measured with DLS was constant at a Z-average of 37 ± 1 nm in diameter. For samples with lower surface coverage, the size started to increase at pH 6–7 to around 100 nm in diameter. At lower pH values the aggregates become >1000 nm, but by adding NaOH to the system, the original size ±2 nm was regained. An important parameter to examine when studying particle interactions and colloidal stability is the zeta potential. It may also be used to study responsiveness toward changes in pH, and to compare the modified and the unmodified silica particles. Fig. 5 shows the zeta potential of dialyzed samples as a function of pH, at (i) low salt concentration (7 mM salt) and (ii) higher salt concentrations (100 and 200 mM). Overall, the zeta potential for all PEGylated samples is negative and the absolute values are small compared to unmodified silica and other types of unmodified and modified silica samples [32]. Low salt concentration: At alkaline conditions the absolute value of the zeta potential of the functionalized samples is smaller and less affected by a change in pH compared to unmodified sol. A distinct difference between the samples was also observed, where an increase in surface coverage results in a reduced absolute zeta
220
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
Fig. 4. Absorbance at 700 nm measured with UV–Vis spectroscopy, showing the variations in turbidity as a function of pH. (A) Purified samples and (B) non-purified samples: Comparing particles without functionalization (black j) and with low (blue ), medium (green ) and high (red ) degree of functionalization; 0.068, 0.155, and 0.315 lmol mPEG silane/m2, respectively. It is evident that the particles with a lower surface coverage display aggregation at low pH, while for the higher degree of surface functionalization aggregation is absent. (C) Comparing samples with medium surface coverage ion-exchanged and measured shortly after pH adjustment ( ); ion-exchanged and stored for two days ( ); pH adjusted using HCl ( ). (D) Showing the reversibility of the aggregation of 0.068 (blue ), 0.155 (green ) and 0.215 lmol mPEG silane/m2 (green ), respectively. Upon increase in pH after the samples have been aggregated, the samples almost regain their original turbidity (filled symbols). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 5. Zeta potential of dialyzed samples as a function of pH, with varying salt concentrations. 0.068 (blue ), 0.155 (green ) and 0.315 lmol mPEG silane/m2 (red ) are at 7 mM salt. The zeta potential of particles without functionalization (black j) is shown for comparison, having a slightly higher salt concentration (10 mM). For the sample with high surface coverage it was possible to measure zeta potential at higher salt concentrations; 100 mM (red ) and 200 mM (red ). The error bars represent the standard deviation obtained from six replicates. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
potential. Due to the severe aggregation resulting in sedimentation of the low surface coverage sample, measurements were not possible below pH 6.
Higher salt concentrations: Only the sample with the highest surface coverage was possible to measure at these high salt concentrations since the samples with lower surface coverage displayed aggregation too severe for measurements to be possible to perform. The high surface coverage sample revealed a zeta potential close to zero in the solutions with high salt concentrations. Despite this, the sample showed no aggregation at high pH values and only moderate aggregation at low pH values; the particle size is increased to 40 nm at 100 mM and 90 nm for 200 mM, at pH < 3. Since for the unmodified silica sol, the higher concentration of salt is close to the critical coagulation point of silica, the corresponding measurements of this sample were excluded. It should be noted that PEGylated silica sols may offer an advantage over unmodified silica for situations where colloidal stability in high salt concentrations are required. To further understand the silica–PEG interactions causing the severe aggregation observed, additional NMR experiments were conducted on a Varian 400 MHz spectrometer. Ultrafiltrated samples were diluted in D2O and ion-exchanged using cation-exchange resin, thus maintaining the same concentrations in all samples analyzed. 1H spectra were collected as a function of pH, seen in Fig. 6, of (A) 0.155 lmol mPEG silane/m2 and (B) 0.315 lmol mPEG silane/m2. For (A), the intensity decreases with pH, presumably due to severe broadening of the peaks from mPEG silane that adsorbs to the silica surface as the affinity increases. It supports the hypothesis that most probably inter-particle bridge flocculation
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
221
Fig. 6. 1H NMR spectra showing the peak arising from the poly(oxy ethylene) units of the mPEG silane (compare Fig. 2) of (A) 0.155 lmol mPEG silane/m2 and (B) 0.315 lmol mPEG silane/m2, as the pH is decreased.
Fig. 7. Surface tension as a function of time for mPEG functionalized silica particles at a concentration of 5 wt% SiO2 and increasing surface coverage, from unmodified (black), 0.068 (blue), 0.155 (green) and 0.315 (red) lmol mPEG silane/m2. (A) Samples purified through ultrafiltration (UF) and dialysis (Dial.). (B) Non-purified samples. The purified samples contain less than 0.3–0.7 mM free mPEG silane, while the non-purified samples have a larger amount of free mPEG silane present, and a faster and more pronounced surface activity. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
occurs via adsorption of mPEG silanes causing the aggregation. Further, it also shows that this process is pH dependent, although addition of electrolytes facilitates the process. For the sample with high coverage, no difference in intensity is observed as the pH is decreased, in agreement with the aggregation studies. This finding is somewhat surprising since the surface is not fully covered with mPEG silane. Nevertheless, the results again support the idea that despite a low surface coverage compared to the theoretical maximum coverage, the covalently attached mPEG chains still have a major influence on the surface properties of the silica particles. 3.5. Surface activity The particles have a molecular weight in the order of 106 g/mol which means that they diffuse slowly in solution, in contrast to a surfactant with a molecular weight in the order of 102 g/mol. To quantify the effect of the functionalized particles on the surface tension, measurements were conducted during a long time period (around 2–3 h) to allow a sufficient amount of particles to reach the air/water interface and the surface tension values to level out. The unmodified silica particles have a very small effect (2 mN/m) on the surface tension, consistent with findings in other studies addressing the surface tension of colloidal silica [33,34]. However, Dong and Johnson [35] found a minimum in surface tension at around 5–7 wt% of charged SiO2 particles, with a reduction of 4 mN/m. The PEGylated particles on the other hand have a significant effect, as can be seen in Fig. 7. The measurements were conducted at a fixed silica concentration of 5 wt%, for both purified and non-purified samples, diluted in distilled water.
When free silane was present in the samples in concentrations exceeding 1 mM, as for the non-purified samples, the surface tension was lower and leveled out faster. This indicates either that a composite system of both free silane and surface modified silica particles facilitates the diffusion of the particles to the surface, or that the reduction of surface tension is simply due to the presence of the free silane. The concentrations of free silane in the samples can be seen in Table 2. Fig. 8 shows the surface tension of hydrolyzed mPEG silane in water. A comparison of Figs. 7 and 8 shows that the reduction of surface tension observed can be explained by the presence of free mPEG silane species only for the non-purified solutions and not for the purified samples. For the low surface coverage, there is no significant effect on the surface activity that can be ascribed to the particles. For the medium and high surface coverage samples a decrease in surface tension of 2 mN/m is observed, after the effect from the free silane has been subtracted. This is comparable to the surface activity of unmodified silica particles and particles with hydrophilic modification and higher surface coverage [34]. It is most probable that during the first 10–20 min of the measurements the air–water interface becomes packed with free silane, giving rise to the main reduction of surface tension, and then the slow-moving particles reduce the surface tension further. For the purified samples, the concentration of free silane is below 0.3–0.7 mM in all samples, rendering a surface tension (ST) of 60–62 mN/m ascribed to the free silane, as seen in Fig. 8. In this case the interface is not as densely packed with free silane. A reduction in surface tension well exceeding that caused by the free silane was observed for the high and medium surface coverage
222
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
Table 2 Free silane concentrations, based on NMR diffusometry data, of the samples analyzed for surface tension. Sample
0.07mPEG 0.15mPEG 0.31mPEG
Concentration free silane (mM) Non-purified
Purified
1.3 5.2 10
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Surface activity and flocculation behavior of polyethylene glycol-functionalized silica nanoparticles Sanna Maria Sofi Björkegren a,b,⇑, Lars Nordstierna a, Anders Törncrona b, Michael E. Persson a,b, Anders E.C. Palmqvist a a b
Chemistry and Chemical Engineering, Chalmers University of Technology, SE-412 96 Göteborg, Sweden AkzoNobel Pulp and Performance Chemicals AB, SE-445 80 Bohus, Sweden
g r a p h i c a l a b s t r a c t
a r t i c l e
i n f o
Article history: Received 23 February 2015 Accepted 20 April 2015 Available online 27 April 2015 Keywords: PEGylated colloidal silica NMR diffusometry Surface activity Dynamic light scattering (DLS)
a b s t r a c t Colloidal silica nanoparticles have been functionalized with methyl polyethylene glycol silane (mPEG silane) and the PEGylated particles have been characterized with focus on exploring their surface chemical properties. The degree of surface functionalization was quantified using NMR diffusometry, and the measurements showed that the silane binds covalently to the silica surface. Samples with surface coverages ranging from 0.068 to 0.315 lmol silane/m2 have been analyzed. The functionalized particles proved to be surface active and showed a significant reduction in surface charge and zeta potential with increasing degree of PEG functionalization. All samples showed colloidal stability at neutral pH and above within the range studied. At lower pH, the samples with low surface coverage displayed a reversible flocculation behavior, while samples with a high surface coverage and samples without functionalization remained stable. This suggests that steric stabilization is effective at low pH when the surface coverage is high enough; electrostatic stabilization is effective for samples without functionalization; and that inter-particle PEG–silica interactions cause flocculation of particles with too low degrees of PEG functionalization. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction ⇑ Corresponding author at: Chemistry and Chemical Engineering, Chalmers University of Technology, SE-412 96 Göteborg, Sweden. E-mail address: [email protected] (S.M.S. Björkegren). http://dx.doi.org/10.1016/j.jcis.2015.04.043 0021-9797/Ó 2015 Elsevier Inc. All rights reserved.
Dispersions of colloidal silica, also referred to as silica sols, are utilized in numerous industrial processes and commercial products, such as retention aid in paper making, as additives in paints,
216
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
as binder in foundry applications, for beverage clarification, and in polishing applications [1]. In all these applications the nature of the silica surface and the ability to control the aggregation of the colloidal particles are of great importance. Studies concerning the surface chemistry of amorphous silica have previously been undertaken [2–4]. The surface chemistry is complex and further characteristics remains to be understood. To further expand the uses of silica, and to improve the use in existing applications, development of new composite silica materials has been extensively explored. Surface modification is a common approach to customizing the silica material properties, which requires further investigations of the material in terms of characterization and evaluation methods. Publications addressing the grafting of poly (ethylene glycol) (PEG) or poly(ethylene oxide) (PEO) onto silica surfaces are frequently appearing, often with uses in biomedical applications due to the biocompatibility of the material [5–9]. PEG-containing coatings is another field of interest, where for example Malmsten et al. [10] found that a high enough interfacial density of PEG on flat silica surfaces results in efficient protein rejection. Leckband et al. studied intermolecular forces and show that grafted PEG chains can exist in both protein-repulsive and protein-attractive states, depending on factors such as compression and polymer chain length [11]. The grafting of PEO onto the surface of silica particles was reported already in the 1980s by Bridger and Vincent [12]. They investigated two methods for terminal grafting of PEO chains of which one was suitable for aqueous conditions, using in situ grafting of isocyanate-capped PEO during particle formation (by Stöber method [13]). However, the modified particles obtained had a low colloidal stability in water [12]. Other examples are Zhang et al. [14] who prepared PEGylated silica particles in methanol addressing both chemical and colloidal stability issues in water and Xu et al. [15] who obtained PEG-coated silica through synthesis in a methanol–ammonia mixture. Joubert et al. [16] prepared PEO-grafted silica through graft polymerization; the polymerization of EO was initiated from the silica surface, onto which alcohol groups had been attached. Still, the grafting of PEG onto silica particles where the synthesis is carried out by a simple process in purely aqueous conditions is not as recurrent. In this paper we report on the characterization of colloidal silica particles functionalized with methyl end-capped tri-methoxy poly(ethylene glycol) silane (mPEG silane) prepared via a simple water-based route. A direct measurement of the mPEG silane attached to the silica particles has been employed using NMR diffusometry, providing quantitative information of the grafting efficiency. The aggregation behavior of the PEGylated particles has been studied with dynamic light scattering (DLS) and UV–Vis spectroscopy. Further, the modified silica surface has been characterized through polyelectrolyte adsorption and zeta potential measurements. In addition, surface activity of the particles has been assessed, since the use of particles as stabilizers for emulsions has recently gained much interest [17,18]. This full study and characterization provides both quantitative information concerning the degree of surface functionalization and surface activity as well as qualitative information concerning flocculation behavior of surface modified colloidal silica.
2. Experimental section
a native pH of 9.1 and a concentration of 40 wt% silica. Silquest A-1230 (from Momentive), a tri-methoxy poly(ethylene glycol) silane end-capped with a methyl group and denoted mPEG silane, was used for the functionalization. The chemical structure of the mPEG silane is shown in Fig. 1 and was derived from standard high-resolution NMR studies (COSY, HMQC, HMBC), giving an average molecular weight of 679 g/mol. The polydispersity index of the mPEG is less than 1.2, as verified with HPLC measurements. For salt and pH adjustments reagent grade sodium hydroxide, sodium chloride and hydrochloric acid from Sigma–Aldrich were used. In addition, Amber-Jet 1500H from Dow was used for pH adjustments, a strong cation-exchange resin with sulfonic acid groups in the hydrogen form. 2.2. Methods 2.2.1. Functionalization with mPEG silane The mPEG silane was added to the silica suspension during agitation and at constant temperature (70 °C or room temperature) using a syringe pump. The synthesis was carried out at pH 9–10, since at this pH interval the desired condensation reaction as well as hydrolysis of the silane are fast [20]. In addition, this pH provides an electrostatically stabilized colloidal system where aggregation of the silica particles is avoided [4]. When the mPEG silane is added to alkaline water, the methoxy groups are hydrolyzed. The formed Si-OH groups of the mPEG silane become available for reaction with the silanol groups of the silica particle surface and through condensation reactions, siloxane bonds form. This results in covalent attachment of the mPEG silane to the silica surface. The mPEG silane has three functional groups, however, only up to two siloxane bonds to the surface can be formed [21], and adjacent silanes may well bind to each other. Hydrolyzed mPEG silane can also react in solution with one another. The samples were purified either by dialysis or by using an ultrafiltration cell from Millipore with an applied N2 pressure of 1.6 bar. The dialysis membranes (from Sigma–Aldrich) were of cellulose with a cutoff of 14 kDa. The ultrafiltration membranes of regenerated cellulose (from Millipore) had a cut-off of 100 kDa, removing free unreacted silanes from the suspensions. 2.3. Characterization and evaluation 2.3.1. NMR diffusometry NMR self-diffusion measurements were conducted on a Bruker Avance 600 spectrometer, equipped with Diff30 diffusion probe with a maximum gradient strength of 1200 G/cm and with a 5 mm RF insert. Using the conventional stimulated spin-echo sequence [22], the diffusion experiments were performed at 25 °C with diffusion time D = 100 ms, gradient pulse length d = 1.5 ms, gradient ringing delay sr = 1 ms and the gradient strength, g, ramped in at least 16 steps from 5 to 1200 G/cm. At each gradient step, the 1H spectrum was collected from 8 to 16 acquisitions dependent on sample concentration. The relaxation delay was set to 5 s. It was however noted that full longitudinal relaxation was not achieved during this time, why this was
H3 C
2.1. Materials
H3C Sodium ion-stabilized colloidal silica particles with the trade name Bindzil 40/130 were provided by AkzoNobel Pulp and Performance Chemicals. These anionic particles have a surface area of 130 m2/g, as measured by Sears titration [19], corresponding to an equivalent spherical diameter of 21 nm [4]. The suspension has
O O
Si
O O
11
CH3
O H 3C Fig. 1. Chemical structure of tri-methoxy poly(ethylene glycol) silane (mPEG silane) as derived from high resolution NMR spectroscopy.
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
compensated for during data processing (see Eq. (1) where sum of relaxation delay and acquisition time is given). Correct calibration of the gradient amplifier was controlled by obtaining the self-diffusion coefficient of HDO trace in a standard sample of pure D2O [23]. For all samples, and also from experiments with varying diffusion time, a strict two-component attenuation of the logarithmic NMR signal intensity against k = (cgd)2(D d/3) was noted, where c is the 1H gyromagnetic ratio. The data pattern allowed for a twocomponent fitting by standard non-linear regression using Eq. (1).
IðkÞ ¼ p1 expfkDf g þ p2 expfkDb g pf ¼
ð1Þ
p1 C b p1 C b þ p2 C f
pb ¼ 1 pf
Cf ¼
relax:delay þ aq:time 1 exp T 1;f ðD sr Þ 2ðd sr Þ exp exp T 1;f T 2;f
Instruments. The suspensions of functionalized silica particles were diluted in pH adjusted water solutions to 0.5 wt% silica, containing HCl or NaOH and compensated with the necessary amount of NaCl required to keep the electrolyte concentration constant throughout the variations in pH. Dynamic light scattering (DLS) particle size measurements were conducted on the same instrument. The colloidal stability of diluted suspensions, at 5 wt% silica, was investigated by adjusting the pH using hydrochloric acid, sodium hydroxide or cation exchange resin. The turbidity of the suspensions was measured with a UV–Vis spectrometer (HP 8453) in quartz cuvettes. The absorbance of the suspension was recorded at 700 nm, since the molecular absorbance of the system is low at this wavelength, and is assumed to be unchanged with the variation in pH [25]. An increase in absorbance, i.e. increased scattering of light is therefore due to an increase in turbidity. The samples were analyzed within 20 min from the pH adjustment, with some additional measurements of longer time intervals. 2.3.3. Surface activity Surface tension measurements of the colloidal systems were made using a Sigma 70 tensiometer employing the DuNoüy ring method. The platinum ring (radius 9.545 mm and wire-thickness 0.185 mm) was rinsed with distilled water and ethanol, and heated with a Bunsen burner in between measurements.
Cb ¼
217
relax:delay þ aq:time 1 exp T 1;b ðD sr Þ 2ðd sr Þ exp exp T 1;b T 2;b
3. Results and discussion As can be seen in Eq. (1) the spin–lattice (longitudinal) and spin– spin (transverse) relaxation times, T1 and T2 respectively, are required for both free (f) and bound (b) silane in order to quantify the relative amount of the two species. The relaxation times were obtained by standard inversion recovery and CPMG pulse sequences where both revealed strict bi-exponential behavior. Likewise to the self-diffusion coefficients, the relaxation times also provide information about local molecular mobility, e.g. a decreased value of T2 indicates reduced mobility [24]. It was noted that the diffusion coefficients were found to be of significantly higher relative difference compared to the relaxation times, comparing free and bound silane (two orders of magnitude for diffusion and one order of magnitude for relaxation). The superior sensitivity of diffusometry, for this particular system, thus argued for quantification by determination of the self-diffusion coefficients. 2.3.2. Particle charge, zeta potential and colloidal stability The apparent surface charge density of purified samples was assessed by particle charge density measurements with a Mütek Particle Charge Detector (PCD), equipped with an oscillating piston that detects the current between two electrodes. The charge was measured during titration with a Mettler-Toledo Autotitrator of the cationic polyelectrolyte polybrene (C13H30Br2N2). Polybrene has two equivalent charges per mol, and the titration is terminated once the suspension reaches the point of zero charge (PZC). The equivalent charge per mass (eq/g) was calculated according to Eqs. (2) and (3) and compared to the charge of the unmodified silica sol of the same size. The titrations were conducted on diluted suspensions (2 wt% silica), and the polybrene concentration was 3.748 g/L thus obtaining Npolybrene of 0.02 eq/L.
C polybrene concentration eq 2 mol MW polybrene
ð2Þ
V volume polybrene Npolybrene V volume silica suspension C silica concentration
ð3Þ
Npolybrene ¼
eq=g ¼
Zeta potential measurements of dialyzed samples, as a function of pH, were conducted on Zetasizer Nano-ZS from Malvern
3.1. Functionalization procedure Various syntheses were conducted, in order to investigate the effect of the reaction conditions on the yield in terms of bound mPEG silane in relation to the added amount of silane, and also to acquire particles with varying degree of surface coverage. The synthesis could be conducted at room-temperature. A slow addition rate, preferably below 0.25 lmol/(m2 h), was required to minimize self-condensation of the mPEG silanes. The maximum surface coverage obtained at the reaction conditions in this work was 0.315 lmol mPEG silane/m2 of silica particle surface, which corresponds to approximately 0.19 molecules per nm2. The surface coverage obtained is relatively low compared to the theoretically available 8 lmol silanol groups/m2 [26]. Steric hindrance of the PEG-chain from already attached silanes, the high solubility of the mPEG silane in water and self-condensation of the silane species reduce the reaction yield. 3.2. Determination of amount of PEG surface coverage The mPEG silane is water soluble and the PEG-chains protruding from the particle surface have low affinity to silica at alkaline conditions, where the silanol groups are dissociated and the surface is charged [27]. This makes NMR an excellent tool for evaluating the degree of functionalization, since a signal, of high sensitivity, from the repeating ethylene oxide units of the PEG chain could be obtained for both free and bound silane, see spectrum in Fig. 2A. Although the signal is an overlap of free and bound signals, it could be analyzed by two-component behavior using diffusometry. Fig. 2B shows experimental data points and the fitted lines of Eq. (1) profile of non-purified and purified PEGylated silica particles in aqueous suspension, and free aqueous mPEG silane. These particles were found to be grafted with 0.155 lmol mPEG silane/m2. The latter value was obtained from the fraction of bound silane, which was calculated, using Eq. (1), via the diffusion data together with the relaxation times T1 and T2 for free and grafted mPEG silane, respectively. The diffusion decay is clearly
218
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
Fig. 2. (A) 1H NMR spectrum of functionalized silica sol. (B) Stimulated spin echo profile from NMR diffusometry measurements of silica particles grafted with 0.155 lmol mPEG silane/m2 and of free aqueous mPEG silane (blue ). Before purification by ultrafiltration (black j), 17% of the silanes are attached to the particles, and after purification 75% of the silanes are attached to the particles (red ). The steeper slope of the filtered suspension is due to a more diluted sample. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
bi-exponential, suggesting a strong association or, most probably, covalent attachment of a fraction of the mPEG silane species. The covalent linkage was further substantiated by variation of the diffusion time D that provided identical results verifying a surface residence time significantly larger than several hundreds of milliseconds. Non-covalent adsorption would most probably result in shorter residence times. The strict mono-exponential appearance of the two joint but distinct diffusion patterns, respectively, combined with the reproduced results when varying the diffusion time, implies that the degree of adsorbed or motionally restricted molecules with long residence time is negligible in this work. Any influence on calculated diffusion coefficients, by such populations, is thus insignificant. The transverse relaxation time T2 is significantly reduced for bound mPEG silane, having a value of 16 ms compared to 797 ms for free aqueous silane and 259 ms for free silane in suspension. The longitudinal relaxation time T1 on the other hand is increased from 336 ms for free silane to 1.96 s for bound silane. This demonstrates that the bound fraction is partly immobilized and associated to the surface of the silica particle, the latter with a long rotational correlation time. The self-diffusion coefficients obtained are in the order of 3 1010 m2/s for free silane, and 5 1012 m2/s for bound, with variations due to differences in concentrations, i.e. viscosity and obstruction. Table 1 provides an overview of the results collected for the samples discussed in this work. Also the purified samples contain small amounts of free mPEG silane in aqueous solution. Both ultrafiltration and dialysis remove 70–80% of the unbound silanes, and dialysis is slightly more efficient than ultrafiltration. The condensation reaction that attaches the mPEG silane to the silica surface is in principle reversible, which could result in detachment of the Table 1 Overview of surface functionalized silica particles discussed in this work and the particle sizes as measured with DLS at alkaline conditions. The yield was calculated from the NMR diffusion data together with the relaxation times, from which the surface coverage, and subsequently the amount of free silane, is obtained. The synthesis temperature was 70 °C for all samples except 0.21mPEG, for which the synthesis was conducted at room temperature. Sample name
Unmodified 0.07mPEG 0.15mPEG 0.16mPEG 0.21mPEG 0.31mPEG
Surface coverage; bound mPEG silane (lmol/m2 SiO2)
(molecules/nm2)
– 0.068 0.155 0.162 0.215 0.315
– 0.04 0.09 0.10 0.13 0.19
Yield
– 25% 17% 18% 22% 17%
Diameter (nm) Z-average
Volume average
32 34 35 35 – 36
30.8 29.7 31.0 31.2 – 31.7
silane at conditions of high pH [20]. The NMR integral intensity of the signal from the bound mPEG silane in the purified samples was compared to the corresponding signal in the non-purified samples, to ensure that no loss of bound mPEG silane occurred during the purification steps. The intensity did not change, showing that the amount of bound silane is maintained intact during the purification process and that the reversibility of the condensation reaction has no significant influence. It also substantiates that the bound silane is covalently attached, since a loss of bound silane would be expected if the association is only physical adsorption. In addition, through analyzing purified samples that had been kept at high pH over a time period of months we found that further detachment did not yet occur over time. 3.3. Surface charge of functionalized silica Single, vicinal and geminal silanol groups as well as surface siloxanes are found at the silica surface, and physically adsorbed water molecules are hydrogen-bonded to all types of silanol groups [1]. The 8 lmol of silanol groups/m2 surface of fully hydroxylated amorphous silica can in water be ionized bearing a charge density that increases with pH; pKa of the SiOH is around 9.2 [4]. At the isoelectric point, reported to be found around pH 2, a silica sol is metastable, and above pH 7 the concentration of surface charge is sufficient to cause mutual repulsion between the particles, which provides stability toward gelling of the sol [1,4]. Functionalization of the silica surface, such as the PEGylation presented here, would yield a reduction of the surface charge density, and may induce steric stabilization thus allowing applications in a broader pH range. Partial coverage of the silica surface allows studies of how the pH influences particle–PEG interactions. 3.3.1. Measurements of the apparent surface charge Particle charge density (PCD) measurements show that the apparent surface charge density is indeed reduced and can be correlated to the surface coverage of the particles, see Fig. 3A. It indicates that the number of available silanol groups on the particle surface has decreased significantly; a reduction of surface charge above 60% is observed for the highest surface coverage. For comparison, silica functionalized with a shorter epoxy silane, having a surface coverage of about 2.3 lmol/m2, reduces the apparent surface charge with 80–85% [28]. This further implies that the mPEG chains from the attached silanes shield the surface silanol groups to such an extent that (i) it prevents further condensation of hydrolyzed mPEG silanes and (ii) it has a major effect on the surface properties even though the surface coverage is low. It should be
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
219
Fig. 3. (A) The apparent surface charge in meq/g as measured with PCD polyelectrolyte adsorption as a function of the surface coverage of the particles. The surface charge correlates well with the surface coverage of mPEG silane; the higher coverage yields a larger reduction of charge. (B) Apparent surface charge as measured with PCD polyelectrolyte adsorption of unmodified silica (black j), silica grafted with 0.155 lmol mPEG/m2 (green ) and 0.315 lmol mPEG/m2 (red ) monitored as a function of pH of the suspensions. The error bars represent the standard deviation obtained from three replicates. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
noted that the reduction of surface charge is indeed apparent, since the access to the silanol groups also depends on the type and size of the polymer used for the titration, in this case polybrene. However, the relative difference between the samples is still of interest. The PCD was also monitored as a function of pH, see Fig. 3B. The suspensions were pH adjusted prior to the measurements using hydrochloric acid. As for unmodified particles, the functionalized particles lose most of their charge between pH 9 and 7 and below pH 2 virtually all charge is gone. However, the decrease in surface charge for modified particles is not as steep as for the unmodified particles, again pointing to the effect on the surface properties of the bound mPEG groups. 3.4. Aggregation behavior of the mPEG modified silica particles That attractive forces exist between poly(ethylene oxide) (PEO) and silica are well-known [29]. This interaction depends on both pH and temperature [25], where an increasing pH decreases the adsorption [30]. In this work, strong pH dependence in terms of aggregation is observed for the mPEG functionalized silica particles having a surface coverage of 0.215 lmol mPEG silane/m2 and below, where the turbidity of the suspension increases with decreasing pH. If a turbid sample is left undisturbed at a low pH it settles and after a month when some of the water is allowed to evaporate it forms a gel. Samples were also analyzed with DLS to obtain the particle size distribution, as a complementary method to both the zeta potential and the turbidity measurements. With DLS, the hydrodynamic radius is calculated from the diffusion coefficient of the particles. At low pH values and high electrolyte concentrations particle–particle interactions may influence the diffusion. When aggregates become large (>1000 nm) and start to sediment, it is no longer possible to derive reliable results, which occurs for a surface coverage of 0.215 lmol mPEG silane/m2 and below, at low pH and/or high electrolyte concentration. It is also important to bear in mind that the radii obtained are estimates, with the purpose to demonstrate the difference between the samples at different conditions. In general the DLS measurements show that the increase in turbidity as measured with UV–Vis corresponds well to an increase in mean particle size. 3.4.1. Colloidal stability and PEG–silica interactions Fig. 4 shows the apparent absorbance (i.e. turbidity) as a function of pH, for A) purified and B) non-purified samples. It can be seen that for low (0.068 lmol/m2) and medium (0.155 lmol/m2)
surface coverages, the turbidity increases rapidly when the pH is below 6. Flocculation of silica, resulting in phase separation, due to adsorption of PEO polymers on the silica surface is known to be dependent on ionic strength, pH and also the amount of polymer adsorbed [31]. The non-purified samples containing unreacted mPEG silane have a similar flocculation behavior as the purified sample, why it can be concluded that adsorption of unreacted mPEG silane has no significant influence in terms of stabilization. Fig. 4C shows turbidity measurements performed while adjusting the pH using ion-exchange resin, maintaining the same ionic strength throughout the variations in pH. Flocculation occurred also when the sample was ion-exchanged, although the increase in turbidity was smaller and the process was slower. It is shown in Fig. 4D that the flocculation is reversible; an increase of pH from an acidic solution returned a flocculated system almost to its original turbidity. This means that the particles are stabilized also when the surface coverage is low in a way that allows the control of aggregation by adjusting the pH. This further indicates that the particles most likely form larger aggregates through weak bridge flocculation that, upon storage, rigidify and form a gel network. For the highest surface coverage obtained (0.315 lmol/m2) no increase in turbidity was seen neither by pH adjustment through ion-exchange nor by addition of acid. The size as measured with DLS was constant at a Z-average of 37 ± 1 nm in diameter. For samples with lower surface coverage, the size started to increase at pH 6–7 to around 100 nm in diameter. At lower pH values the aggregates become >1000 nm, but by adding NaOH to the system, the original size ±2 nm was regained. An important parameter to examine when studying particle interactions and colloidal stability is the zeta potential. It may also be used to study responsiveness toward changes in pH, and to compare the modified and the unmodified silica particles. Fig. 5 shows the zeta potential of dialyzed samples as a function of pH, at (i) low salt concentration (7 mM salt) and (ii) higher salt concentrations (100 and 200 mM). Overall, the zeta potential for all PEGylated samples is negative and the absolute values are small compared to unmodified silica and other types of unmodified and modified silica samples [32]. Low salt concentration: At alkaline conditions the absolute value of the zeta potential of the functionalized samples is smaller and less affected by a change in pH compared to unmodified sol. A distinct difference between the samples was also observed, where an increase in surface coverage results in a reduced absolute zeta
220
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
Fig. 4. Absorbance at 700 nm measured with UV–Vis spectroscopy, showing the variations in turbidity as a function of pH. (A) Purified samples and (B) non-purified samples: Comparing particles without functionalization (black j) and with low (blue ), medium (green ) and high (red ) degree of functionalization; 0.068, 0.155, and 0.315 lmol mPEG silane/m2, respectively. It is evident that the particles with a lower surface coverage display aggregation at low pH, while for the higher degree of surface functionalization aggregation is absent. (C) Comparing samples with medium surface coverage ion-exchanged and measured shortly after pH adjustment ( ); ion-exchanged and stored for two days ( ); pH adjusted using HCl ( ). (D) Showing the reversibility of the aggregation of 0.068 (blue ), 0.155 (green ) and 0.215 lmol mPEG silane/m2 (green ), respectively. Upon increase in pH after the samples have been aggregated, the samples almost regain their original turbidity (filled symbols). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 5. Zeta potential of dialyzed samples as a function of pH, with varying salt concentrations. 0.068 (blue ), 0.155 (green ) and 0.315 lmol mPEG silane/m2 (red ) are at 7 mM salt. The zeta potential of particles without functionalization (black j) is shown for comparison, having a slightly higher salt concentration (10 mM). For the sample with high surface coverage it was possible to measure zeta potential at higher salt concentrations; 100 mM (red ) and 200 mM (red ). The error bars represent the standard deviation obtained from six replicates. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
potential. Due to the severe aggregation resulting in sedimentation of the low surface coverage sample, measurements were not possible below pH 6.
Higher salt concentrations: Only the sample with the highest surface coverage was possible to measure at these high salt concentrations since the samples with lower surface coverage displayed aggregation too severe for measurements to be possible to perform. The high surface coverage sample revealed a zeta potential close to zero in the solutions with high salt concentrations. Despite this, the sample showed no aggregation at high pH values and only moderate aggregation at low pH values; the particle size is increased to 40 nm at 100 mM and 90 nm for 200 mM, at pH < 3. Since for the unmodified silica sol, the higher concentration of salt is close to the critical coagulation point of silica, the corresponding measurements of this sample were excluded. It should be noted that PEGylated silica sols may offer an advantage over unmodified silica for situations where colloidal stability in high salt concentrations are required. To further understand the silica–PEG interactions causing the severe aggregation observed, additional NMR experiments were conducted on a Varian 400 MHz spectrometer. Ultrafiltrated samples were diluted in D2O and ion-exchanged using cation-exchange resin, thus maintaining the same concentrations in all samples analyzed. 1H spectra were collected as a function of pH, seen in Fig. 6, of (A) 0.155 lmol mPEG silane/m2 and (B) 0.315 lmol mPEG silane/m2. For (A), the intensity decreases with pH, presumably due to severe broadening of the peaks from mPEG silane that adsorbs to the silica surface as the affinity increases. It supports the hypothesis that most probably inter-particle bridge flocculation
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
221
Fig. 6. 1H NMR spectra showing the peak arising from the poly(oxy ethylene) units of the mPEG silane (compare Fig. 2) of (A) 0.155 lmol mPEG silane/m2 and (B) 0.315 lmol mPEG silane/m2, as the pH is decreased.
Fig. 7. Surface tension as a function of time for mPEG functionalized silica particles at a concentration of 5 wt% SiO2 and increasing surface coverage, from unmodified (black), 0.068 (blue), 0.155 (green) and 0.315 (red) lmol mPEG silane/m2. (A) Samples purified through ultrafiltration (UF) and dialysis (Dial.). (B) Non-purified samples. The purified samples contain less than 0.3–0.7 mM free mPEG silane, while the non-purified samples have a larger amount of free mPEG silane present, and a faster and more pronounced surface activity. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
occurs via adsorption of mPEG silanes causing the aggregation. Further, it also shows that this process is pH dependent, although addition of electrolytes facilitates the process. For the sample with high coverage, no difference in intensity is observed as the pH is decreased, in agreement with the aggregation studies. This finding is somewhat surprising since the surface is not fully covered with mPEG silane. Nevertheless, the results again support the idea that despite a low surface coverage compared to the theoretical maximum coverage, the covalently attached mPEG chains still have a major influence on the surface properties of the silica particles. 3.5. Surface activity The particles have a molecular weight in the order of 106 g/mol which means that they diffuse slowly in solution, in contrast to a surfactant with a molecular weight in the order of 102 g/mol. To quantify the effect of the functionalized particles on the surface tension, measurements were conducted during a long time period (around 2–3 h) to allow a sufficient amount of particles to reach the air/water interface and the surface tension values to level out. The unmodified silica particles have a very small effect (2 mN/m) on the surface tension, consistent with findings in other studies addressing the surface tension of colloidal silica [33,34]. However, Dong and Johnson [35] found a minimum in surface tension at around 5–7 wt% of charged SiO2 particles, with a reduction of 4 mN/m. The PEGylated particles on the other hand have a significant effect, as can be seen in Fig. 7. The measurements were conducted at a fixed silica concentration of 5 wt%, for both purified and non-purified samples, diluted in distilled water.
When free silane was present in the samples in concentrations exceeding 1 mM, as for the non-purified samples, the surface tension was lower and leveled out faster. This indicates either that a composite system of both free silane and surface modified silica particles facilitates the diffusion of the particles to the surface, or that the reduction of surface tension is simply due to the presence of the free silane. The concentrations of free silane in the samples can be seen in Table 2. Fig. 8 shows the surface tension of hydrolyzed mPEG silane in water. A comparison of Figs. 7 and 8 shows that the reduction of surface tension observed can be explained by the presence of free mPEG silane species only for the non-purified solutions and not for the purified samples. For the low surface coverage, there is no significant effect on the surface activity that can be ascribed to the particles. For the medium and high surface coverage samples a decrease in surface tension of 2 mN/m is observed, after the effect from the free silane has been subtracted. This is comparable to the surface activity of unmodified silica particles and particles with hydrophilic modification and higher surface coverage [34]. It is most probable that during the first 10–20 min of the measurements the air–water interface becomes packed with free silane, giving rise to the main reduction of surface tension, and then the slow-moving particles reduce the surface tension further. For the purified samples, the concentration of free silane is below 0.3–0.7 mM in all samples, rendering a surface tension (ST) of 60–62 mN/m ascribed to the free silane, as seen in Fig. 8. In this case the interface is not as densely packed with free silane. A reduction in surface tension well exceeding that caused by the free silane was observed for the high and medium surface coverage
222
S.M.S. Björkegren et al. / Journal of Colloid and Interface Science 452 (2015) 215–223
Table 2 Free silane concentrations, based on NMR diffusometry data, of the samples analyzed for surface tension. Sample
0.07mPEG 0.15mPEG 0.31mPEG
Concentration free silane (mM) Non-purified
Purified
1.3 5.2 10
