Journal of Colloid and Interface Science

Transcription

1 Journal of Colloid and Interface Science 394 (2013) Contents lists available at SciVerse ScienceDirect Journal of Colloid and Interface Science Electrostatically driven adsorption of silica nanoparticles on functionalized surfaces Xue Li, Olivia Niitsoo, Alexander Couzis Department of Chemical Engineering, The City College of New York, NY 10031, United States article info abstract Article history: Received 11 June 2012 Accepted 19 November 2012 Available online 13 December 2012 Keywords: Electrostatic adsorption Silica Nanoparticles Coatings Solid liquid interface Amino-functionalized substrate Adsorption of nanoparticles on solid supports is a scientifically interesting and technologically important phenomenon that has been attracting ever-increasing attention. Formation of particle-based films onto surfaces from stable suspensions is at the center of the development of new devices that utilize the plethora of newly synthesized nanoparticles with exciting properties. In this study we exploit the attractive electrostatic interactions between silica (SiO 2 ) nanoparticles and functionalized substrates that display an amine termination in order to devise a simple method for the fabrication of SiO 2 nanoparticle films. Electrostatically controlled adsorption allows for uniform coverage of nanoparticles over large areas. The Stöber method (a sol gel approach) was employed to prepare uniformly sized SiO 2 nanoparticles with a diameter of nm. Native oxide-covered silicon wafer substrates were amino-functionalized utilizing the self-assembled monolayer of 3-aminopropyltrimethoxysilane (APS). The adsorption of SiO 2 nanoparticle film onto the silicon wafer substrate was controlled by modulation of the electrostatic interaction between nanoparticles and the substrate. Modification of surface charge of either the SiO 2 NP or the substrate is a crucial step in the process. Thus the effect of APS adsorption time on the surface energy of the substrate was investigated. Also, process parameters such as NP concentration and solvent composition were varied in order to investigate the extent of NP adsorption. Moreover, NaCl was introduced to the SiO 2 suspension as a charge-screening agent to reduce the inter-particle repulsion in the suspension as well as interaction of the particles with the surface. This resulted in denser/thicker films. Published by Elsevier Inc. 1. Introduction The continuing expansion of synthetic methods for nanoparticles has brought to the forefront the need to form structures using these nanoparticles as building blocks. Many material properties related to nanoscale that once were only accessible through expensive semiconductor type processing are now accessible because of the nanoparticles. However, as complex structures are required to exploit these properties and build devices, assembly of these materials becomes crucial. Technological applications of such approaches include catalysis [1,2], sensors [3], photonics [4], antireflection [5], antifogging [6], self-cleaning [7,8] and so on. For applications in electronic and optical devices, most often films (coatings) of well-ordered closely packed nanoparticles over large areas are needed. A number of methods have been developed to achieve that. In this manuscript we briefly discuss the methods where nanoparticles are prepared prior to film deposition ( bottom-up approach), leaving aside the approaches where the NP are formed in situ during the deposition process (some sol gel techniques, spray pyrolysis, (electro) chemical deposition and physical vapor deposition methods with bulk precursor). Corresponding author. Fax: address: couzis@ccny.cuny.edu (A. Couzis). Spin-coating is well established in the electronics industry for depositing thin films of polymers for lithography [9]. It has also been used for depositing nanoparticle films [10]. Ogi et al. [11] have coated films of submicron-sized SiO 2 particles from aqueous suspensions on sapphire substrate with high surface coverage. While spin-coating enables good control of film thickness by adjusting the spin rate, the morphology and uniformity of the film is highly sensitive to ambient conditions and the solvent used to disperse the nanoparticles in. Controlled atmosphere (clean-room) fabrication conditions are likely to upset the costs especially when NP-based materials are pursued as a more economical alternative to their bulk counterpart. Dip-coating is relatively simple and easily automated [12]. It can be adjusted for coating almost any substrate size [13]. Widely used for 3D-assembly of photonic crystals, and referred to as vertical evaporation, is a process related to dip-coating. In vertical evaporation [14], instead of withdrawing the substrate from the NP suspension at controlled velocity, the substrate (placed vertically or at a fixed angle in the deposition vessel) is stationary and the solvent is allowed to evaporate, leaving behind a NP film. Just as dip-coating, also vertical evaporation can be performed repeatedly to increase film thickness. In electrophoretic deposition (EPD) charged particles are attracted onto a conducting substrate by applying electric field to /$ - see front matter Published by Elsevier Inc.

2 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) the suspension [15,16]. EPD cannot be used for depositing NP films on insulating substrates (with the exception of porous membranes when placed in front of the electrode). Magnetic field has also been used for NP film deposition, but this approach is limited to magnetic NP [17]. Spray deposition has been successfully applied for many commercial coatings for decades [18]. However, it is difficult to obtain uniform thin (<2 lm) continuous films by spraying. Screen-printing and doctor blading are both simple and cheap, but do not offer very precise control on film thickness [19]. Langmuir Blodgett (LB) technique entails self-assembly of NP at the gas liquid interface, followed by compression by mechanical barriers and transfer of the NP film onto a solid substrate. The method is best suited for NP monolayer preparation, and can be applied on large surfaces [20]. Similarly to LB, NP can also be assembled at liquid liquid interface and transferred to a solid substrate. Different methods can be combined to obtain the desired NP film, e.g. by Tsai et al. [21] who successfully coupled LB and electrostatic assembly by introducing a low dielectric constant solvent (ethanol) as a driving force for Au NPs to assemble from a suspension onto a monolayer of octadecylamine at air water interface. Unfortunately, these approaches are not always useful for practical applications and all have some drawbacks such as the requirement for specific equipment; not applicable to large area substrate; limited to a small set of surfaces (e.g. conducting or flat substrates); the morphology and uniformity of the film is highly sensitive to environmental conditions. Compared to these traditional strategies, the electrostatic adsorption technique offers an easy and inexpensive process for layer formation and allows a variety of materials to be incorporated within the film structures. Therefore, the assembly process based on adsorption can be considered as a versatile bottom-up assembly technique. Adsorption (either by physisorption or chemisorption) of nanoparticles onto a substrate is the simplest method of fabricating self-organized 2D and 3D networks or arrays. One of the most prominent advantages of the electrostatic adsorption is its simplicity [22]. Electrostatic adsorption is mainly directed by electrostatic interactions between the particles and the substrate [23 25]. In the adsorption process, the substrate is immersed into nanoparticle dispersion for a time optimized for adsorption. It is then withdrawn, rinsed, and dried. Nanoparticles can be adsorbed and arrayed to the substrate to form ordered two-dimensional networks or three-dimensional arrays. For some applications a monolayer of NP is desired, and often the lengthy adsorption time is not of concern. For multilayer coatings repeated adsorption steps are usually implemented. This approach is referred to as the layerby-layer (LbL) self-assembly technique [26 30]. Polyelectrolytes are often introduced into LbL self-assembly method in order to obtain opposite charged substrates and particles. Ahn [31] and coworkers made use of the electrostatic attraction and formed random close-packed polystyrene latex particle monolayers, which find uses as anti-reflection coatings. Qiu et al. [32] studied systematically the key parameters for controlling the quality of the selfassembled ceramic particle film. Tettey et al. [33] successfully showed that LbL assembly of oppositely charged particles could be performed in a nonpolar solvent, toluene, as a general film fabrication technique. Not only electrostatic attraction but also secondary bonding interactions, e.g. hydrogen bonding, are often employed in the LbL assembly [34]. Li et al. [35] successfully prepared asymmetrically functionalized Janus particles as well as hollow microcapsules by coating the particles with cross-linked hydrogen-bonded multilayers followed by polymer-on-polymer stamping. In the absence of polyelectrolytes, LbL assembly can still be performed simply by taking advantage of the chemical nature of particles and substrates under certain process conditions. Lee et al. [7] reported all-nanoparticle thin-film coatings composed of only negatively charged SiO 2 and positively charge TiO 2 NP. Otherwise, when particles and substrates have the same surface charge, surface charge modification is necessary. Usually there are two different approaches to surface charge modification. The first one involves the self-assembly of a bifunctional monolayer onto a substrate (formation of a self-assembled monolayer, SAM) and then subsequent attachment of nanoparticles to the monolayers [36,37]. The other approach involves the surface charge modification of nanoparticles themselves by bifunctional molecules [38 42]. 3-aminopropyltrimethoxisilane (APS) is one of this kind of bifunctional molecule which is widely used and is also adopted in this research [43]. Lee and co-workers presented one-component all-nanoparticle multilayers assemblies comprising oppositely charged SiO 2 NP by sequential adsorption of negatively charged SiO 2 NP and amino-functionalized SiO 2 NP [44]. Applications related to the electrostatic adsorption have developed rapidly and there are a number of theoretical studies [45 47]. In this manuscript we report on the development of a framework of electrostatically driven nanoparticle adsorption utilizing a carefully selected set of experiments to test the impact of process variables and material properties. In our work we have used SiO 2 nanoparticles for the formation of electrostatically adsorbed films onto silicon wafer substrates. The unique component of this study is that by using APS monolayers to control the charge density on either surface or particle we can investigate the balance of the forces that direct the adsorption of particles onto a surface. We are not uniquely locked into one state of charge for surface or particle, and we can independently control that parameter from the ionic strength of the solution. 2. Materials and methods 2.1. Materials 3-aminopropyltrimethoxysilane (APS, 99%) was obtained from Gelest Inc., stored in a desiccator and used as received. Tetraethoxysilane (TEOS, 98%, from Gelest Inc.), ammonium hydroxide (Acros Organics), chloroform (HPLC grade, Spectrum), toluene (anhydrous, Alfa Aesar), acetone (ACS grade, Alfa Aesar), methanol (laboratory reagent, Sigma Aldrich), ethanol (200 proof, Decon Laboratories Inc.), diiodomethane (CH 2 I 2, 99%, stabilized with copper, Alfa Aesar), sodium chloride (NaCl, Spectrum) were used without further purification. All water used was deionized water. Silicon wafers (p-type, h100i, from Silicon Quest) were polished on one side Preparation of silica nanoparticles Silica particles were prepared by following the Stöber method M TEOS was added to a mixture of 0.51 M ammonium hydroxide, 15 M ethanol and 2.5 M deionized water. The resulting solution was kept stirring for 24 h. Monodispersed colloidal silica dispersion with narrow size distribution was formed. In order to purify the silica particles, the colloid was centrifuged and redispersed in ethanol for three times Preparation of APS coated silica nanoparticles (SiO 2 /APS) APS was used to functionalize SiO 2 NP surface with positively charged amino groups. The as-prepared SiO 2 NP were treated with APS by dispersing the vacuum-dried SiO 2 NP in 5 mm APS solution in toluene. The mixture was stirred from 0.5 h, 1 h to 1.5 h. The functionalized particles were then centrifuged and transferred back to ethanol.

3 28 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) Substrate preparation First, the silicon wafers were cut into 3 1 cm pieces using a diamond pen, cleaned with chloroform at 60 C, then with acetone also at 60 C and then with methanol at 75 C. Between solvents, the Si substrates were dried in nitrogen flow. The substrates were then treated with the Samco UV Ozone Stripper to remove organic residue. Then the cleaned substrates were immediately placed into 1 mm APS solution in toluene for 1.5 h. The substrates were taken out from the toluene solution of APS, rinsed rigorously in toluene with ultrasonic treatment for 20 min. Finally, the substrates were blown dry with nitrogen and used immediately Self-assembly of silica NP film on APS modified silicon wafer A dispersion of silica NP in an ethanol water mixture (100%, 90%, 70% and 30% of ethanol by volume) was ultra-sonicated for 10 min. Then above described freshly APS-functionalized Si wafer substrates were immersed vertically at a slight incline with the polished face down (to prevent exclude sedimentation effects, as shown in Scheme 1.) at room temperature. After elapsed immersion time, the substrates were rinsed with mixture of ethanol and water at the same composition as the SiO 2 NP suspension and finally dried in nitrogen flow. For charge screening experiments, NaCl was added to the SiO 2 NP suspensions in concentrations of 0.01 mm, 0.1 mm and 1 mm Characterization f-potential of SiO 2 NP was measured by laser Doppler electrophoresis on a Malvern Zetasizer Nano. Dynamic Light Scattering (DLS) was performed on the same instrument to measure hydrodynamic size of the NP. Scanning Electron Microscopy was performed on Zeiss Supra 55 for SiO 2 NP size measurement and for NP film morphology and coverage evaluation. NP coverage on substrates was quantified by analyzing the SEM images using ImageJ software. Surface energy of substrates was evaluated by contact angle measurements with DI water and diiodomethane on a contact angle goniometer (Ramé Harte Goniometer). These measurements are conducted on the APS modified silicon substrate surfaces with different APS treatment time (varying from 0 h to 1.5 h). For each sample, five different points on the sample were taken for measurements and the contact angles were determined by averaging the five values measured at the five different position. The surface energy of each sample was calculated using the following two equations [48,49] The geometric mean [28] equation: h ð1 þ cos h i Þc i ¼ 2 c d i c d 1=2 i S þðc p i cp S Þ1=2 The harmonic mean (HM) equation: ð1 þ cos h i Þc i ¼ 4 " # c d i c d S þ cp i cp S c d i þ c d S c p i þ c p S HM is valid between low-energy material, predicts well compared to GM while GM prefers high-energy surfaces. Here, h i is the contact angle of the probe liquid on the sample surface; c is the surface energy, the superscripts d and p refer to nonpolar (dispersive) and polar components, respectively. As the surface tension values of c i, c d i and c p i of the two probe liquids are known (c H2 O ¼ 72:8 dyne/cm, c CH2 I 2 ¼ 50:8 dyne/cm, for c d H 2 O ¼ 21:8 dyne/cm, cd CH 2 I 2 ¼ 49:5 dyne/ cm) we could get the surface energy values for c p S and cd S of our samples by solving the equations above, and estimate the surface energy (c S ) by taking the sum of the calculated c p S and cd S values. 3. Results and discussion 3.1. Synthesis of SiO 2 nanoparticles The silica nanoparticles were prepared directly by following the well-known Stöber method [50]. The SEM study shows that the synthesized silica nanoparticles are monodisperse and spherical (see Fig. 1.). The silica particle size from SEM image is about nm while DLS shows the dynamic particle size of about 67 nm. This is reasonable since the underlying principles of the two methods are different: from SEM image we measure the solid particle size, while DLS measures the hydrodynamic radius of the particles. The surface charge properties of the silica NP in suspension were investigated in terms of zeta potential. The zeta potential of SiO 2 in pure ethanol is negative, a result consistent with the fact that the surface of the particles are occupied by SiO functionality (as shown in Table 4) which means that the SiO 2 NP are good candidates for an adsorbent onto a positively charged surface. It was also observed that the SiO 2 suspension was very stable without agglomeration for periods of up to at least a month. ð1þ ð2þ Scheme 1. Illustration of adsorption between negatively charged SiO 2 nanoparticles and amino-terminated positively charged Si substrate. Fig. 1. FE-SEM image of SiO 2 nanoparticles used for adsorption.

4 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) Table 1 Surface energy (c S ) and its dispersive (c d S ), polar (cp S ) components for APS treated silicon substrate surface. APS treatment time (h) h H2O ( ) h CH2I 2 ( ) 3.2. Formation of amino-terminated SAM on silicon substrates In general, immobilization of NP by means of self-assembly is accomplished via surface modification of a substrate with functional groups. Organosilane self-assembled monolayers (SAMs) have been frequently employed to introduce amino group onto the substrates [51 53]. Here, 3-aminoproryltrimethoxysilane (APS) was used to introduce positive primary amino groups onto the Si-wafer substrates by silylation reaction. The silicon substrate is natively coated with an oxide (SiO x ) layer with the thickness of about Å [54]. Thus the Si substrate surface is occupied by hydroxyl groups and is ready for silylation with APS molecules. After immersion of the pre-cleaned silicon substrate into APS toluene solution, APS molecules are chemisorbed onto the negative charged silicon substrate surface. At this point, the negative surface charge of the Si substrate is reversed to positive because of the exposed positively charged amino groups on the surface. The aminoterminated SAM of APS is applied as an anchor surface for adsorbing SiO 2 nanoparticles. SiO 2 NP adsorption also serves as verification to the presence of the amino groups on the substrate surface. The APS modified Si-wafer substrates should be used directly after the preparation or stored in vacuum since the APS SAM is found to adsorb impurities, which might be hydrophilic, from the ambient atmosphere or even inside a desiccator under certain pressure [55]. In our experiments, we use the amino-terminated Si-wafer substrates immediately after the modification Surface energy calculation Geometric mean method c S (dyn/ cm) x p x d c S (dyn/ cm) Harmonic mean method Surface Energy (dyn/cm) GM HM APS Deposition Time (h) Fig. 2. Plot of surface energy of APS treated Si substrate as a function of the APS deposition time. Both the surface energy calculated by harmonic mean method (HM) and geometric mean method (GM) gave out the same trend. The surface energy of the amino-terminated silicon substrate was investigated using both the harmonic mean method and x p x d Table 2 Surface coverage quantification for SiO 2 NP at the silicon substrates with different APS treatment time and f-potential for APS-functional SiO 2 suspension (a mixture of ethanol/ water with the ratio of 9:1). APS deposition time (h) f-potential (mv) SiO 2 NP coverage (%) 6.54 ± ± ± ± 3.48 geometric mean method. In order to evaluate the surface energy of the APS treated silicon substrates, contact angle measurements were performed on samples after different APS treatment time. The results of contact angles and calculated surface energies using the above two equations are summarized in Table 1. Both harmonic mean method and geometric mean method show the same trend with increasing APS treatment time: the dispersive component of the surface energy increased while the polar component decreased. With time, more and more hydroxyl groups on the silicon substrate surface reacted with the silane moiety of APS molecules, leading to increase of dispersive component of surface energy. This suggests that the degree of surface coverage by APS molecules increased with time since the dispersive component was increasing. Also, the total surface energy of the APS modified substrate was decreased with the reduction of initial hydroxyl groups on the silicon substrate surface (as shown in Fig. 2.). Interestingly, the total surface energies of 1 h and 1.5 h surface modification of APS were close in values (For GM, the c S value was 57.5 dyn/cm and 57.4 dyn/cm respectively; for HM, the c S value was 65.2 dyn/cm and 65.1 dyn/cm respectively). This indicates that the surface modification of 1 h and 1.5 h with APS left almost the same amount of APS in the SAM on Si wafer substrate, suggesting an equilibrium was reached. In order to make sure that maximal APS surface coverage was achieved, 1.5 h treatment was employed as our standard surface charge modification step. In order to verify that the surface charge has been modified from negative to positive, f-potential measurement was performed on amino-functionalized SiO 2 NP suspension by employing the SiO 2 /APS as a model for the amino-terminated substrates. The results were summed in Table 2, showing that already after 0.5 h modification of APS the surface charge has been successfully changed from negative to positive. From these results, we can conclude that the substrate surface also acquired a positive charge. Moreover, the f-potential values for the 1 h and 1.5 h APS modified SiO 2 NP were almost identical suggesting that 1 h treatment is enough to achieve sufficient APS coverage. Once the surface charge modification of Si substrates is complete, they are ready for electrostatic adsorption of the negatively charged SiO 2 NP. During immersion in negatively charged SiO 2 NP suspension, the electrostatic attraction between the positively charge Si substrate and the negatively charged particles serves as the driving force for the adsorption of SiO 2 NP. SEM images of self-assembled SiO 2 NP on APS modified silicon substrate are shown in Fig. 3. Using Image J we evaluated the surface coverage of SiO 2 NP on the silicon substrate surface. The quantification results are listed in Table 2 and plotted in Fig. 4. Itis evident that the surface coverage of SiO 2 NP increases with the charge density on the substrate by varying the modification time of APS. Very few particles adsorbed onto unmodified silicon substrate (the coverage is as low as 6.54%) due to electrostatic repulsion between similarly charged particle and substrate. The driving force here for the particle to attach onto the substrate could be the random Brownian motion or the capillary force exerted when the Si wafer was pulled out of the suspension. The SiO 2 assemblies formed on 1 h APS modified substrate were as dense as that of 1.5 h treatment of APS (see Figs. 3C and D) with the sur-

5 30 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) Fig. 3. FE-SEM images of SiO 2 NP assemblies on Si-wafer substrate with different APS treatment time (A) 0 h, (B) 0.5 h, (D) 1 h, (D) 1.5 h. Adsorption has been carried out with 2.04 wt.% SiO 2 suspension in the solvent mixture of ethanol and water with ratio of 9:1 v/v. SiO 2 Surface Coverage (%) face coverage of 61.0% and 66.0%, respectively, which substantiates that the electrostatic interaction between silicon substrate and SiO 2 NP is almost the same, on the other hands indicates that same amount of APS was chemisorbed onto the silicon substrate. It is clear that 1 h and 1.5 h APS modification on the substrate resulted in the same surface charge density that translates into the same surface energy and the ability to adsorb same amount of SiO 2 NP onto the substrate Effect of SiO 2 concentration APS Deposition Time (h) Fig. 4. Plot of surface coverage of SiO 2 nanoparticles as a function of the deposition time of APS (0 h, 0.5 h, 1 h, 1.5 h). Adsorption has been performed on 1.5 h APS modified Si-wafer substrate by using SiO 2 suspension of 2.04 wt.%. The interaction force for SiO 2 NP adsorption is the electrostatic attraction between particles and the substrate. Higher the concentration of SiO 2 in the suspension, higher the probability of attraction between the absorbent and the substrate. The influence of the SiO 2 NP concentration in the suspension on the surface coverage of SiO 2 NP assemblies on the amino-functionalized silicon substrate surface was investigated by varying the concentration of SiO 2 NP from 0.04 to 2.04 wt.% (see Figs. 5 and 6). The surface coverage quantification results are listed in Table 3. The SiO 2 NP density increased rapidly with the particle concentration of 0.62 wt.% and then continued to increase, reaching the maximal value (of 70.1% SiO 2 surface coverage) at 2.04 wt.% (see Fig. 5A). For a low concentration, there are fewer nanoparticles per unit volume in the suspension compared to high concentration case leading to loose coverage. With increasing suspension concentration, the number density of nanoparticles also increased (from nr/ml for 0.04 wt.% to nr/ml for 2.04 wt.%), resulting in high frequency of attachment between the SiO 2 NP and substrates and thus, dense coverage. In observing the morphology of SiO 2 assemblies (light-colored areas on the SEM images represent the SiO 2 NP), it is notable that the particles form clusters in isolated regions on the surface. Similar observation of film morphology for LbL assembly between charged particles and oppositely charged polyelectrolyte has been reported [56]. This study concluded that the phenomenon is a result of particles adhering to existing islands of particles rather than the bare substrate surface. As we increased the suspension concentration, there are more SiO 2 particles per unit volume and thus the probability of interaction between particle and functionalized substrate increases and particles adhere not only on the existing particle clusters but also to the amino-functionalized surface, eventually resulting in uniform dense coverage as shown in Fig. 5A Effect of solvent medium In this portion of the paper we discuss the influence of the solvent composition on the SiO 2 adsorption behavior. This effect was investigated systematically by employing f-potential measurements to quantitatively assess the extent of the electrostatic repulsion between SiO 2 nanoparticles in a given medium. We used a mixture of DI water and ethanol as the solvent medium. It is

6 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) Fig. 5. FE-SEM images of SiO 2 NP assembly on 1.5 h APS modified Si-wafer substrate with different SiO 2 NP concentration (A) 2.04 wt.%, (B) 0.62 wt.%, (C) 0.2 wt.%, (D) 0.04 wt.%, the suspension is in a mixture of ethanol and water with the ratio of 9:1 v/v. SiO 2 Surface Coverage (%) Silica Particle Concentration (wt %) Fig. 6. Plot of surface coverage of SiO 2 nanoparticles as a function of the SiO 2 NP concentration (0.04 wt.%, 0.2 wt.%, 0.62 wt.%, 2.04 wt.%). The silicon substrate has been treated 1.5 h with APS and the solvent is mixture of ethanol and water with ratio of 9:1 v/v. Table 3 Surface coverage quantification for SiO 2 NP at the silicon substrates with different concentration of SiO 2 suspension all for 19 h adsorption. SiO 2 concentration (wt.%) Coverage (%) important to note that the f-potential does not directly reflect the net charge on the surface but rather indicates the electrostatic potential at the interface of double layer between the dispersion medium and the stationary layer of fluid attached to the dispersed particle. The influence of suspension parameters on f-potential may be predicted by the Debye Hückel approximation recognizing that double layer is thin compare to the particle radius (the Debye length j 1 for all the suspension used in our experiment j has a value greater than 0.03 nm 1, and the particle radius, a, has a value greater than 50 nm, thus ja > 1)[57]. f ¼ r ej Here, r is the nanoparticle surface charge, and the suspending medium property, dielectric constant, e, was varied by changing the extent of ethanol in the mixture. According to the relationship above, f is inversely proportional to j and e and directly proportional to r. Silica suspensions were prepared as described in the experimental section with a concentration of 0.04 wt.%, which is sufficiently dilute for the f-potential measurement and particle size analysis. f-potential measurement results for different ethanol content in both SiO 2 and APS coated SiO 2 suspension medium were shown in Table 4. The significance of f-potential is that its value can be related to the stability of colloidal dispersions. In our results, the absolute value of f-potential for suspensions with ethanol content higher than 90% were greater than 30 mv indicating that they were very stable. It is also notable that an increase in the ethanol content generally leads to an increase in the absolute value of f-potential. And this increase in f-potential is in agreement with the decrease of dielectric constants for different suspending medium (as shown in Table 4) [58,59] according to the Debye Hückel approximation. The morphology of SiO 2 particles adsorbed onto the silicon substrate from suspending medium of different ethanol content was shown in Fig. 7. From the images of Fig. 7 together with the surface coverage results shown in Fig. 8 and Table 4, the surface coverage of SiO 2 assembly increased from 36.1% to 63.2% as the ethanol content increased from 50% to 90% and then experienced a decrease to 48.8% surface coverage when no water was present, suggesting that the surface coverage is not simply determined by the interparticle interaction representing by f-potential of SiO 2 NP but also the particle substrate interaction. An important point is that our SiO 2 NP are consolidated, hence the presence of water cannot affect ð3þ

7 32 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) Table 4 Effects of ethanol ratio in EtOH H 2 O mixture on f-potential of SiO 2 NP suspension, surface coverage of SiO 2 assemblies and also dielectric constant (at 25 C). EtOH volume concentration 100% 90% 70% 50% f-potential (mv) SiO SiO 2 /APS Coverage (%) Dielectric constant (from Ref. [58]) Dielectric constant (calculated) Fig. 7. SEM images of SiO 2 NP assembly on 1.5 h APS modified Si-wafer substrate by using 2.04 wt.% silica suspension with different ethanol content (A) 100%, (B) 90%, (C) 70%, (D) 50% in the solvent medium. SiO 2 Surface Coverage (%) EtOH fraction in solvent medium (%) Fig. 8. Effects of the solvent medium composition (volume concentration of EtOH in EtOH/H 2 O mixture: 50%, 70%, 90%, 100%) on the SiO 2 surface coverage by using 2.04 wt.% silica suspension. The silicon substrate has been treated 1.5 h with APS. substrate. From Table 4, for water content of 30% and 50%, the f- potential of SiO 2 /APS was very low, practically zero, which suggests that the substrates exert almost no attraction to the SiO 2 particle in the suspension. As the alcohol content is increased to 90% and in pure ethanol, the substrates demonstrate strong attraction by the f-potential of about 20 mv. However, the densest surface coverage (see Fig. 7B.) occurs in 90% ethanol suspension while the surface coverage for pure ethanol case was lower. Since the f-potential indicates the degree of repulsion between adjacent, similarly charged particles in the dispersion, the relatively high f-potential of 42.9 mv in pure ethanol means that the repulsion force between adjacent particles is stronger than between particles with f-potential of 31.4 mv. Hence the two particle films did not achieve the same coverage even though the attraction to the substrate is equally strong. We conclude that the optimal suspending medium for achieving densely packed films is ethanol/water mixture with 90% ethanol. the system by promoting chemical bridging between NP via unhydrolyzed alkoxy groups. The amino-functionalized SiO 2 (SiO 2 /APS) particles were used as a model for the APS modified silicon substrate. The f-potential of SiO 2 /APS suspension indicated the interaction between particle and substrate: the bigger the absolute value of f-potential for SiO 2 /APS, the stronger the SiO 2 particles were attracted onto the 3.6. Charge-screening effect of NaCl The negatively charged SiO 2 NP are subject to repulsion force between particles and also from the particles already adsorbed on the substrate. This inter-particle repulsion is the main obstacle for achieving dense packing. An efficient way to minimize the inter-particle electrostatic repulsion is increasing ionic strength in the solution. This is commonly done by introducing salt ions into

8 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) Table 5 f-potential for SiO 2 nanoparticles dissolved in solutions of different ionic strength, the solution is ethanol and water with a ratio of 9:1 v/v. NaCl concentration (mm) f-potential (mv) Coverage (%) Zeta-potential (mv) EtOH fraction in solvent medium (%) NaCl EtOH fraction NaCl (mm) Fig. 9. Plot of f-potential as a function of both volume fraction of EtOH in the solvent mixture of EtOH/H 2 O (50%, 70%, 90%, 100%) and the NaCl concentration (0 mm, 0.01 mm, 0.1 mm, 1 mm) with ethanol volume fraction of 90%. the solution that screen the Coulombic interactions between SiO 2 NP with negative surface charge [60]. We increased ionic strength by adding NaCl to the SiO 2 suspension. The effective hard-sphere model [61] is useful to describe the range of the electrostatic interaction. In this model, the SiO 2 particles are regarded as hard spheres with an effective radius that is defined as: r eff ¼ r þ h where r is the radius of the particles and h is the enlargement of r due to the electrostatic repulsion between particles. The range of this repulsion is characterized by the Debye length (j 1 ) assuming that h is proportional to the Debye length: h j 1 ¼ p 0:304 ffiffiffiffiffiffiffiffiffiffiffiffiffi nm ð5þ ½NaClŠ at 25 C the unit for [NaCl] is moles per liter. Eq. (5) indicates that h is a function of NaCl concentration in SiO 2 suspension only and thus the effective particle size depends on the salt concentration. In the case of no salt in SiO 2 suspension, we assume that the salt concentration is very low and on the order of 10 4 M, which means that under this assumption, the enlargement h is about 30 nm. Similar assumptions have been previously reported [32].f-potential measurements were also performed to verify the charge screening effects of NaCl, the results are summed in Table 5 and Fig. 9 shows the plot of f-potential as a function of both NaCl concentration and H 2 O/EtOH ratio. The silica nanoparticle suspension shows a relatively high f-potential of 31.4 mv in the absence of salt. It is evident that the absolute value of f-potential for the silica suspension was decreased from 31.4 mv to 25.7 mv by increasing the salt concentration from 0 mm to 1 mm indicating that the addition of NaCl has reduced the inter-particle repulsion. On the other hand, for [NaCl] is 1 mm the enlargement of radius h is about 9.6 nm comparing with that of 0.01 mm [NaCl] (h = 96 nm). The distance between particles has been greatly reduced according to the effective hard sphere model, which provides theoretical support for our results. Also, the f-potential for the silica suspension with 0.01 mm NaCl was consistent with that of no salt, which matches our assumption for the no salt case that the salt concentration is on the order of 0.1 mm. The influence of the NaCl concentration in SiO 2 suspension on the surface coverage of SiO 2 NP assemblies on the amino-functionalized silicon substrate surface was investigated by varying the concentration of NaCl from 0 to 1 mm (see Figs. 10 and 11). The ð4þ Fig. 10. Effects of the ionic strength in the suspension with NaCl concentrations of: (A) 0 mm (B) 0.01 mm (C) 0.1 mm and (D) 1 mm on the SiO 2 surface coverage by the 2.04 wt.% silica suspension. The silicon substrate has been treated 1.5 h with APS. The solvent medium is a mixture of ethanol and water with the ratio of 9:1 v/v.

9 34 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) SiO 2 Surface Coverage (%) surface coverage results are listed in Table 5. For NaCl concentrations of 0.01 mm and even 0.1 mm, the surface coverage of SiO 2 on silicon substrate stays almost the same. On the other hand, the surface coverage did not change from that of no salt case because nearly no salt was introduced to the system. However, when the salt concentration was increased to 1 mm, the surface coverage increased dramatically up to 84.2%. The surface morphology of SiO 2 NP film was uniform and dense due to the reduction of inter-particle repulsion (Fig. 10D.). In addition, even multilayers were observed, which could be the result of the relevant instability of the particle suspension caused by screening of surface charge on particles by NaCl ions. One can conclude when analyzing the above data that ionic strength of the suspension from which particle adsorption takes place exerts a profound effect on the surface coverage of SiO 2 assemblies onto oppositely charged substrate. If in 2D we model the SiO 2 NP on silicon substrate as circles on a planar surface, we could estimate the packing density of SiO 2 by the circle-packing model. In two-dimensional ideal packing of identically sized circles, there are two close periodic circle packing regimes: hexagonal packing and square packing with a maximum possible surface coverage of SiO 2 NP of 90.7% and 79%, respectively [62]. Also, if we assume that the electrostatic adsorption of particles is a random process for non-interacting hard spheres, the random sequential adsorption (RSA) model could be used to estimate the maximum possible surface coverage [63]. In the RSA model [45], particles arriving at the substrate surface are adsorbed sequentially. Adsorbed particles are anchored on random sites irreversibly and the newly arrived particles are not allowed to occupy sites overlapping the previously anchored particles. There are no interparticle interactions, and the jamming limit results in the maximum surface coverage of 54.7%. Our highest observed coverage of NP (surface coverage of 84.2% for addition of 1 mm NaCl to the SiO 2 suspension) is between the RSA limit and the ideal hexagonal packing limit. This is reasonable because real SiO 2 particle are not perfect circles and have some polydispersity that may increase the packing density [64]. Moreover, the RSA model accounts only for steric effects and omits the particle particle, particle substrate electrostatic interactions, particle diffusion and multilayer formation possibilities. All of these may lead to a higher coverage than the jamming limit. However, the 84.2% still does not exceed the surface coverage of close packed (90.7%) for identical circles in ideal case. 4. Conclusions NaCl Concentration (mm) Fig. 11. Plot of SiO 2 surface coverage as a function of NaCl concentrations (0 mm, 0.01 mm, 0.1 mm, 1 mm). Amino-functionalized silicon substrate with 1.5 h treatment of APS has been immersed in the 2.04 wt.% silica suspension with solvent composition of 9:1 v/v for ethanol/water. In conclusion, we have demonstrated a general method for creating nanoparticle coatings by electrostatically controlled adsorption. The NP film formation by electrostatic assembly occurs at solid liquid interface. The key point is the modification and control of surface charge of both the particles and the substrate. We show that with simple chemisorption of APS SAM, the surface charge of the native oxide coated Si substrate can be adjusted from negative to positive. We found that the deposition time of APS has effect on the NP film coverage. Interestingly, more than a monolayer of SiO 2 NP can be deposited during single adsorption cycle in the liquid environment. Charge screening by introducing NaCl to the suspension reduces interparticle repulsion, thus enabling deposition of thicker films. Up to 84.2% surface coverage was obtained when negatively charged SiO 2 NP were adsorbed onto positively charged substrate when introducing salt into the silica suspension. Electrostatically controlled adsorption can be applied over large areas with uniform surface coverage, and by simple adjustment of surface charge, could be a method of choice for depositing films of a large variety of different nanoparticles. Optimal adsorption parameters can be devised for fabricating NP films of desired surface coverage. References [1] S. Kidambi, J. Dai, J. Li, M.L. Bruening, J. Am. Chem. Soc. 126 (2004) [2] J.F. Ciebien, R.E. Cohen, A. Duran, Supramol. Sci. 5 (1998) 31. [3] T.A. Duchesne, J. Brown, K. Guice, Y. Lvov, M. McShane, Sensors Mater. 14 (2002) 293. [4] L.L. Beecroft, C.K. Ober, Chem. Mater. 9 (1997) [5] Y. Du, L.E. Luna, W.S. Tan, M.F. Rubner, R.E. Cohen, ACS Nano 4 (2010) [6] F.Ç. Cebeci, Z. Wu, L. Zhai, R.E. Cohen, M.F. Rubner, Langmuir 22 (2006) [7] D. Lee, M.F. Rubner, R.E. Cohen, Nano Lett. 6 (2006) [8] L. Zhai, F.Ç. Cebeci, R.E. Cohen, M.F. Rubner, Nano Lett. 4 (2004) [9] A. Friedl, P. Adam, S. Leshkov, J. Homola, J. Nanosci. Nanotechnol. 11 (2011) [10] P. Jiang, M.J. McFarland, J. Am. Chem. Soc. 126 (2004) [11] T. Ogi, L.B. Modesto-Lopez, F. Iskandar, K. Okuyama, Colloids Surf. A: Physiochem. Eng. Aspects 297 (2007) 71. [12] Z.L. Hua, J.L. Shi, L. Wang, W.H. Zhang, J. Non-Cryst. Solids 292 (2001) 177. [13] C.J. Brinker, G.C. Frye, A.J. Hurd, C.S. Ashley, Thin Solid Films 201 (1991) 97. [14] S. Anachkov, P. Vasileva, C. Dushkin, J. Optoelectron. Adv. Mater. 11 (2009) [15] S. Fraden, A.J. Hurd, R.B. Meyer, Phys. Rev. Lett. 63 (1989) [16] T.Y. Gong, D.T. Wu, D.W.M. Marr, Langmuir 19 (2003) [17] T. Ding, K. Song, K. Clays, C.H. Tung, Adv. Mater. 21 (2009) [18] K. Somroop, R. Pornprasertsuk, S. Jinawath, Thin Solid Films 519 (2011) [19] H.T. Yang, P. Jiang, Langmuir 26 (2010) [20] G. Yu, A. Cao, C.M. Lieber, Nat. Nanotechnol. 2 (2007) 372. [21] H. Tsai, Y.L. Lee, Soft Matter 5 (2009) [22] K. Ariga, J.P. Hill, Q.M. Ji, Phys. Chem. Chem. Phys. 9 (2007) [23] M. Sastry, M. Rao, K.N. Ganesh, Acc. Chem. Res. 35 (2002) 847. [24] M. Sastry, Pure Appl. Chem. 74 (2002) [25] J.H. Fendler, Korean J. Chem. Eng. 18 (2001) 1. [26] C.Y. Jiang, V.V. Tsukruk, Adv. Mater. 18 (2006) 829. [27] M.M. de Villiers, D.P. Otto, S.J. Strydom, Y.M. Lvov, Adv. Drug Deliv. Rev. 63 (2011) 701. [28] K. Ariga, Q. Ji, J.P. Hill, Y. Bando, M. Aono, NPG Asia Mater. 4 (2012) e17. [29] P.R.V. Tassel, Curr. Opin. Colloid Interface Sci. 17 (2012) 106. [30] B.M. Wohl, J.F.J. Engbersen, J. Control. Release 158 (2012) 2. [31] J.S. Ahn, P.T. Hammondb, M.F. Rubnerc, I. Lee, Colloids Surf. A 259 (2005) 45. [32] H. Qiu, W. Lee, S. Sukhishvili, J. Am. Ceram. Soc. 89 (2006) [33] K.E. Tettey, M.Q. Yee, D. Lee, Langmuir 26 (2010) [34] D. Lee, M.F. Rubner, R.E. Cohen, Chem. Mater. 17 (2005) [35] Z. Li, D. Lee, M.F. Rubner, R.E. Cohen, Macromolecules 38 (2005) [36] A. Gole, S.R. Sainkar, M. Sastry, Chem. Mater. 12 (2000) [37] G.K. Toworfe, R.J. Composto, I.M. Shapiro, P. Ducheyne, Biomaterials 27 (2006) 631. [38] A. Kumar, A. Mandale, M. Sastry, Langmuir 16 (2000) [39] M. Sastry, Curr. Sci. 78 (2000) [40] G. Deng, M.A. Markowitz, P.R. Kust, B.P. Gaber, Mater. Sci. Eng. C 11 (2000) 165. [41] M.A. Markowitz, G. Deng, B.P. Gaber, Langmuir 16 (2000) [42] A. Vanblaaderen, A. Vrij, J. Colloid Interface Sci. 156 (1993) 1. [43] F.X. Zhang, M.P. Srinivasan, Langmuir 20 (2004) [44] D. Lee, Z. Gemici, M.F. Rubner, R.E. Cohen, Langmuir 23 (2007) [45] M.R. Oberholzer, J.M. Stankovich, S.L. Carnie, D.Y.C. Chan, A.M. Lenhoff, J. Colloid Interface Sci. 194 (1997) 138. [46] A.P. Angelopoulos, J. Colloid Interface Sci. 243 (2001) 292. [47] K.V. Tretiakov, K.J.M. Bishop, B. Kowalczyk, A. Jaiswal, M.A. Poggi, B.A. Grzybowski, J. Phys. Chem. A 113 (2009) [48] S. Wu, Polymer Interface and Adhesion, Marcel Dekker, New York, [49] E.N. Dalal, Langmuir 3 (1987) 1009.

10 X. Li et al. / Journal of Colloid and Interface Science 394 (2013) [50] W. Stöber, A. Fink, E. Bohn, J. Colloid Interface Sci. 26 (1968) 62. [51] N. Kumar, C. Steiner, C. Maldarelli, A. Couzis, Langmuir 17 (2001) [52] F. Fan, C. Maldarelli, A. Couzis, Langmuir 19 (2003) [53] N. Kalyankar, M. Sharma, S. Vaidya, D. Calhoun, C. Maldarelli, A. Couzis, L. Gilchrist, Langmuir 22 (2006) [54] I. Haller, J. Am. Chem. Soc. 100 (1978) [55] D.F.S. Petri, G. Wenz, P. Schunk, T. Schimmel, Langmuir 15 (1999) [56] J.W. Ostrander, A.A. Mamedov, N.A. Kotov, J. Am. Chem. Soc. 123 (2001) [57] C. Hiemenz, R. Rajagopalan, Principles of Colloid and Surface Chemistry, third ed., Marcel Dekker, New York, [58] G. Åkerlöf, J. Am. Chem. Soc. 54 (1932) [59] S.J. Suresh, V.M. Naik, J. Chem. Phys. 116 (2002) [60] A.R. Ferhan, L. Guo, D.H. Kim, Langmuir 26 (2010) [61] P. Hanarp, D.S. Sutherland, J. Gold, B. Kasemo, J. Colloid Interface Sci. 241 (2001) 26. [62] H. Steinhaus, Mathematical Snapshots, third ed., Dover, New York, [63] C.A. Johnson, A.M. Lenhoff, J. Colloid Interface Sci. 179 (1996) 587. [64] Z. Adamczyk, B. Siwek, M. Zembala, P. Weroński, J. Colloid Interface Sci. 185 (1997) 236.