Journal of Sol-Gel Science and Technology

, Volume 50, Issue 3, pp 328–336

Synthesis of silica nanoparticles by modified sol–gel process: the effect of mixing modes of the reactants and drying techniques


  • M. Jafarzadeh
    • School of Chemical SciencesUniversiti Sains Malaysia
    • School of Dental Sciences, Health CampusUniversiti Sains Malaysia
  • C. S. Sipaut
    • School of Chemical SciencesUniversiti Sains Malaysia
Original Paper

DOI: 10.1007/s10971-009-1958-6

Cite this article as:
Jafarzadeh, M., Rahman, I.A. & Sipaut, C.S. J Sol-Gel Sci Technol (2009) 50: 328. doi:10.1007/s10971-009-1958-6


A modified preparation of silica nanoparticles via sol–gel process was described. The ability to control the particle size and distribution was found highly dependent on mixing modes of the reactants and drying techniques. The mixture of tetraethoxysilane and ethanol followed by addition of water (Mode-A) produced monodispersed powder with an average particle size of 10.6 ± 1.40 nm with a narrow size distribution. The freeze drying technique (FD) further improved the quality of powder. In addition, the freeze dried samples have shown unique TGA decomposition steps which might be related to the well-defined structure of silica nanoparticles as compared to the heat dried samples. DSC analysis showed that FD preserved the silica surface with low shrinkage and generated remarkably well-order, narrow and bigger pore size and pore volume and also large endothermic enthalpies (ΔHFD = −688 J g−1 vs. ΔHHD = −617 J g−1) that lead to easy escape of physically adsorbed water from the pore at lower temperature.


Silica nanoparticlesSol–gelFreeze drying

1 Introduction

Synthesis of silica nanoparticles have drawn great interest of research owing to their potential application in industries (electronic devices, insulators, catalysis, etc.) and pharmaceuticals (enzyme encapsulation, drug delivery and cell markers) [1, 2]. Sol–gel process has become an attractive research area, in which extensive studies have been made on the synthesis of silica nanoparticles [35].

In general, sol–gel process is very sensitive towards the experimental conditions that affect the rate of hydrolysis and condensation reactions, e.g., the type and concentration of starting materials (alkoxides), H2O/alkoxide ratio, catalyst’s nature and concentration/pH, nature of solvent, temperature, time of reaction, aging and drying method [1, 2, 69]. The Stöber method was firstly introduced using ammonia catalyzed hydrolysis and condensation of ethoxysilanes in low molecular-weight alcohols as solvent to produce uniform silica particles [10]. Various methods have been reported for the preparation of silica nanoparticles [5, 1120]. Bogush and Zukoski [5] successfully prepared monodispersed silica particles in the range of 40 nm to few micrometers using modified Stöber method. The authors believed that concentration of TEOS, concentration of ammonia, concentration of water, as well as types of solvent (alcohol) and reaction temperature were the main parameters, which governed the particle size and its distribution. By optimizing these parameters, Park et al. [17] conveniently prepared ultrafine silica particles within the range of 13.7 ± 4.5 nm. Kim et al. [21] were able to reduce the particle size up to 17.5 nm with 3.33 mL min−1 feed rate through addition of small amount of NaI as an electrolyte additive during the synthesis. Rahman et al. [22] reported that monodispersed silica of average size 20.5 ± 3.5 nm under 0.1 mL min−1 feed rate can be synthesized in the presence of small quantity of NH4Br. Most recently, Rahman et al. [23], have reported a sol–gel route for the synthesis of silica nanoparticles in the primary size range by optimizing various parameters such as concentration of the reactants, ammonia feed rate, temperature under ultrasonic mixing.

There are several drawbacks in sol–gel process such as the difficulties in controlling the particle size, aggregation and agglomeration, and longer reaction time. In this work, we have synthesized silica nanoparticles via modified sol–gel process by emphasizing the reactants mixing mode and drying techniques that are not reported in earlier studies [17, 2224]. The effect of mixing mode of reactant and drying techniques on particle sizes and distribution of the powders is highlighted.

2 Experimental

2.1 Reagents

Tetraethoxysilane (TEOS, 99%, Fluka), absolute ethanol (EtOH, 99.5%, Systerm), ammonia (NH3 25%, A & M Chemicals) and distilled water have been used in this work. The chemicals were employed without any further purification.

2.2 Standard procedure

Silica nanoparticles are synthesized by using a standard procedure with experimental conditions provided in Table 1. A quantity of 5 mL of TEOS was first dissolved in 30 mL of absolute ethanol under low frequency ultrasound (Bransonic, Model 5510, 42 kHz) at room temperature for 10 min. Then, 1 mL of distilled water was dropped into the reaction media with the feed rate of 0.2 mL min−1, to facilitate hydrolysis of TEOS in the ultrasonic bath. After 1.5 h, 2 mL of ammonia (catalyst) was fed into the reaction mixture at the feed rate of 0.01 mL min−1. Sonication was continued for 3 h. Gelation was allowed for 1 h. The gel was centrifuged and washed with ethanol and distilled water (3 × 7 min, 6,000 rpm). Drying was carried out using either a conventional oven at 70 °C for 24 h or freeze drying (FD) under vacuum for overnight in a freeze dryer (Labconco, Freezon 12). The samples were calcined at 600 °C for 2 h.
Table 1

Optimized experimental parameters for the preparation of silica nanoparticles


Optimal value

TEOS (mol L−1)


NH3 (mol L−1)




Feed rate (mL min−1)


Temperature (°C)


Reaction time (h)


Sonication (kHz)


2.3 Synthesis of nanosilica at different mixing modes of the reactants

Silica nanoparticles were prepared under similar conditions as described in Sect. 2.2 with different mixing modes of the reactants. Mode-A: TEOS + EtOH (TEOS was first dissolved in ethanol for 10 min, and then water was added into the reaction medium). Mode-B: one-pot (TEOS, EtOH and H2O were mixed simultaneously). Mode-C: H2O + EtOH (mixed for 10 min, then TEOS was added to the mixture). The flow chart of the different mixing modes is given in Fig. 1.
Fig. 1

Flow chart for nanosilica preparation by different mixing modes

2.4 Characterizations

Morphology of the samples was studied by using a transmission electron microscopy (TEM, Philips CM 12) operated at an acceleration voltage of 80 kV. The particle size and size distributions (PSD) were determined using analysis Docu Version 3.2 image analysis software. Dilute dispersion of powder of samples in ethanol were prepared under ultrasonication for 10 min to obtain a homogeneous suspension. One drop of suspension was evaporated on a carbon-coated copper grid. Determination of the particle size and statistical parameters were based on the measurement of more than 300 particles from the TEM micrograph.

Surface area and porosity of samples was measured by adsorption–desorption of nitrogen isotherm at 77 K on an automatic physisorption analyzer (Micromeritics ASAP 2000). The samples were degassed overnight under vacuum at 105 °C before measurement (10−3 mmHg). Thermogravimetric analysis (TGA) was carried out on TGA 7 instrument (Perkin–Elmer) with heating rate of 20 °C min−1 under flowing nitrogen. TG-MS (ThermoStar™, equipped with V8.10 Star software) was used to recognize the species being released during the thermal treatment at a heating rate of 10 °C min−1 under flow of N2. Differential scanning calorimetry (DSC) was used to study the heat capacity using Pyris 1 DSC instrument (Perkin–Elmer) at a heating rate of 10 °C min−1 in the range of −50 to 400 °C.

3 Results and discussion

3.1 Effect of experimental parameters on the particle size and morphology

Silica particles were prepared at the same conditions with different feed rates of ammonia (0.01, 0.02, 0.03 mL min−1). Table 2 shows the effect of feed rates on particle size of silica obtained via FD. The average particle size was determined to be 10.6 ± 1.4, 11.4 ± 1.7 and 15.0 ± 1.6 nm for feeding rates of 0.01, 0.02, and 0.03 mL min−1, respectively. Figure 2 shows the particle size distribution (PSD) and TEM of nanoparticles prepared in different ammonia feed rates. Results show that particle size decreased with the decrease in the feed rate. At 0.01 mL min−1 feed rate, the smallest particle size and narrowest particle distribution was obtained. TEM analyses illustrate uniform and relatively aggregated particles prepared for 0.01 mL min−1 feed rate (Fig. 2b). Increase of the feed rate enhanced the rate of hydrolysis, which increased the concentration of hydrolyzed monomer that lead to higher nucleation rate. Since the rate of nucleation is equal to the rate of growth of particles [24], hence, associated of hydrolyzed monomer under “monomer addition model” [3] produced larger particles for the higher feed rates. Lower feed rate can control the nucleation and growth of the primary particles and leads to smaller nanoparticles.
Table 2

Effect of feed rate on particle size of silica obtained via freeze drying

Feed rate (mL min−1)

Particle size range (nm)

Average particle size (nm)



10.6 ± 1.4



11.4 ± 1.7



15.0 ± 1.6
Fig. 2

PSD and TEM of silica nanoparticles obtained via freeze drying in different ammonia feed a PSD, b 0.01 mL min−1, c 0.02 mL min−1 and d 0.03 mL min−1

3.2 Effect of mixing mode

Silica nanoparticles were prepared under similar condition at different mixing mode of the reactants as illustrated in Fig. 1. Table 3 shows the effect of mixing mode on the particle size of silica obtained via FD. The average particle size was found to be 10.6 ± 1.4, 13.8 ± 1.7 and 14.9 ± 1.6 nm for Mode-A, B, and C, respectively. PSD and TEM of the silica are given in Fig. 3. TEM images reveal a narrow size and shape distribution in the morphology of the particles with roughly low aggregation and agglomeration by using Mode-A (Fig. 3b).
Table 3

Effect of mixing mode on particle size of silica obtained via freeze drying

Mixing mode

Particle size range (nm)

Average particle size (nm)



10.6 ± 1.4



13.8 ± 1.7



14.9 ± 1.6
Fig. 3

PSD and TEM of silica nanoparticles obtained from freeze drying in different mixing modes a PSD, b Mode-A, c Mode-B and d Mode-C

TEOS is non-polar, immiscible in water, but dissolves easily in ethanol. Thus the mixing mode greatly affects the homogeneity of TEOS molecules in the reaction medium (TEOS-ethanol) and interactions between TEOS and water molecules. In Mode-C the presence of strong hydrogen bonds between ethanol and water restricts the accessibility (it means availability and proximity of reactants that lead to higher probability of collision among them) of TEOS to water molecules for the hydrolysis reactions. According to natural structure of liquid water in which the molecules are linked to some of neighbors by hydrogen bonding interaction [25], that can forms the different arrangement [(H2O)n, n = 3–60] and shape such as dimmer, trimer (cyclic), tetramer (cyclic), pentamer (cyclic), hexamer (cyclic, cage or prism) and so on [26]. The same arrangement or cluster between water and ethanol might be expected. It is clear that the interaction among polar molecules (water–ethanol cluster) is stronger than polar-nonpolar molecules (TEOS-ethanol). On the other hand, low homogeneity of TEOS molecules in the mixture of water and ethanol (water–ethanol cluster) lead to microphase separation among them. In other words, hydrolysis reaction performs when TEOS penetrate into the pore of water–ethanol cluster through homogeneous TEOS-ethanol interface. Due to microphase separation phenomena, accessibility between TEOS and water was restricted, consequently, the growth of nanoparticles were increased in competition with the nucleation in the cluster, leading to relatively larger particles. For Mode-B due to moderate homogeneity among the starting materials, the PSD shows better results compared to Mode-C. The mixing of the starting materials simultaneously provides the proximity between the hydrolyzing agent (water) and target molecule (TEOS), thus enhancing the hydrolysis reaction. On the other hand, for Mode-A, there has been high homogeneity among TEOS and ethanol compared to that in Mode-B and C. This leads to easy accessibility between TEOS and water. Moreover, ethanol as an interface helps to establish an effective contact between TEOS and water. In conclusion, the effective homogeneity of precursor in the solvent at the initial stage ought to be very critical factor in sol–gel process. Thus, Mode-A is a recommended procedure for the preparation of monodispersed silica nanoparticles.

3.3 Effect of drying techniques

The freshly synthesized silica nanoparticles (in gel form) were dried under heat drying (HD) in oven for 24 h or FD in deep vacuum for overnight. HD process cause the decrease in gel volume due to the loss of liquid by evaporation through the silica pores. It is well-known that the rapid removal of the water from the pores causes blistering, shrinkage and shriveling of silica gel [27]. However, FD (lyophilization) preserves the structure of materials without shrinkage of the gel structure [28]. It appears to be one of the most attractive methods to stabilize the colloidal nanoparticles [29].

Particle size distribution of these two drying techniques is illustrated in Fig. 4 showing no significant different. However, FD technique revealed an interesting surface characteristic of the particles. The specific surface area, micropore area, average pore diameter and micropore volume of nanoparticles produced from FD are higher than of HD drying techniques (Table 4). This may due to the slow removal of water or other solvents through sublimation process under vacuum, reduce the capillary forces (decrease the thermal tension), resulting in a minimal shrinkage and preserve a high surface area and porosity of the particles.
Fig. 4

PSD of silica nanoparticles in different drying process: freeze drying (FD-C) and heat drying (HD-C)

Table 4

The effect of drying process on specific surface area, pore diameter, and particle size of the samples




BET surface area (m2 g−1)



Micropore area (m2 g−1)



Micropore volume (mL g−1)



Average pore diameter (nm)



Particle size range (nm)



Average particle size (nm)

10.6 ± 1.4

11.0 ± 1.4

The adsorption–desorption isotherm plot and pore size distribution are shown in Fig. 5. The hysteresis loop shows good agreement with typical pattern for mesoporous materials based on Brunauer’s classification [30]. In addition, pore size diameters in Table 4 obviously support the mesoporosity behavior in the silica according to the IUPAC definition (2–50 nm in pore size) for mesoporosity [31]. There are distinct differences in the shape and amount of adsorbate that adsorbed on adsorbent in isotherm plot. High volume of adsorbed gases reveals the larger volume capacity in FD-C that shows a good consistent with resulting pore volume in Table 4. The threshold capillary condensation leading to the sharp increase in adsorbed volume commence at lower relative pressure (P/P0 = 0.57) in HD-C than that in FD-C (P/P0 = 0.8). Also, the narrower plot in FD-C may indicate the more similarity between the adsorption and desorption behaviour compared to that in HD-C. It might be proposed a wider pore size distribution in HD-C.
Fig. 5

Nitrogen sorption isotherm (a), pore area (b) and pore volume (c) distribution (BJH method) of silica nanoparticles obtained via freeze (FD-C) and heat (HD-C) drying

Pore size distribution based on desorption isotherm was calculated using the capillary condensation model assuming cylindrical pores according to BJH (Barett–Joyner–Halenda) method [32] (Fig. 5b, c). The results reveal uniform and narrower pore area as discussed earlier.

Concentration of hydroxyl groups (δOH) on the surface of silica is the key parameter that influences the physical interaction between coupling agents and matrices in the preparation of nanocomposites. It is expected that high number of hydroxyl group on the silica surface will enhance the adhesion between inorganic filler and functional organic matrix. Nevertheless, determination of silanol concentration of colloidal silica particles by adapting the reported procedure (titration method) [33] reveals different silanol concentration as 1.33 and 1.27 mmol g−1 for FD and HD, respectively. The finding was very consistent with the result from BET; therefore, it can be another evidence to highlight the difference between freeze and HD process.

3.4 Thermal analysis

The TGA analysis showed two stage mass losses for HD samples (Fig. 6a). The DTG reveals the initial loss at below 100 °C which is related to the physically adsorbed water on the silica surface [31]. The second stage of mass loss between 150 and 900 °C represent the release of trapped water within the pores, degradation of residual starting materials and dehydroxylation of silanol groups [31].
Fig. 6

TGA and DTG thermograms of silica particles obtained a via heat drying, before calcination (HD) and after calcination (HD-C) b via freeze drying, before calcination (FD) and after calcination (FD-C)

Significantly, there were four distinctive stages of weight loss for FD sample (Fig. 6b). The m/e data was extracted from TG equipped with mass spectrometer as detector. At below 130 °C, the weight loss related to the physically adsorbed water (m/e: 18 = H2O+, 17 = HO+) and trapped residual ammonia (m/e: 17 = +NH3). The second stage (130–320 °C) indicated the escape of trapped water (m/e: 18 = H2O+, 17 = HO+), ammonia (m/e: 17 = +NH3) and small portion of residual solvent (m/e: 46 = EtO+H) in the pores. Stage 3 (320–540 °C) was due to the dehydroxylation of the silanol group (m/e: 18 = H2O+, 17 = HO+) and the degradation of residual reactant [m/e: 43 = C2H3O+ (TEOS)]. Finally, stage 4 (540–900 °C) was suggested to be due to the degradation of residual TEOS (m/e: 43 = C2H3O+) and dehydroxylation of silanol groups (m/e: 18 = H2O+, 17 = HO+).

Uniformity, cylindrically shaped pores (assumed cylindrical pores according to the BJH model), larger pore sizes in the FD samples help to release the pore liquid at distinctive temperature ranges. However, in HD samples the shrinkage caused deformity of cylindrical-shape pores. The comparison between lost of physically adsorbed water via freeze (0.88%) and heat (1.41%) drying obtained by TG results had indicated the efficiently dehydration process for FD. DSC results (Fig. 7) showed a large endothermic enthalpy for FD (ΔHFD = −688 J g−1) as compared to HD (ΔHHD = −617 J g−1). These results indicate that the ease of evaporation of physically adsorbed water for FD that occurred at lower temperature (Tonset = 10.82 °C, Tmax = 96.19 °C) as compared to HD (Tonset = 28.78 °C, Tmax = 116.19 °C). Also, it can be due to the uniformed and well-ordered porous structure which makes the escape of water molecules easy.
Fig. 7

DSC thermograms of silica particles obtained via heat drying (HD) and freeze drying (FD)

For the calcined silica only two steps of decomposition were observed. The initial loss [30–150 °C (HD-C) or 30–165 °C (FD-C)] was due to the loss of adsorbed water from the silica surface. Less volatile were released from HD-C due to the increase in siloxane bonds and reduced surface hydroxyl groups. The second decay [150–900 °C (HD-C) or 165–900 °C (FD-C)] related to the dehydroxylation process (release of water molecules due to condensation of silanol groups and formation of siloxane linkage). Ek et al. [34] elucidated an interesting method for determination of the number of hydroxyl groups by thermogarvimetry (TG). The hydroxyl content was calculated by using the following Eq. 1:
$$ n{\text{OH}}_{{{\text{SiO}}_{2} }} = \frac{{2[\% {\text{ wt}} (T_{0} ) - \% {\text{ wt}} (T_{f} )]}}{{100\;M_{{{\text{H}}_{ 2} {\text{O}}}}}} $$
In which \( n{\text{OH}}_{{{\text{SiO}}_{2} }} \) is the number of hydroxyl groups per gram of silica. The weight loss is calculated in the interval between initial (T0) and final (Tf) temperatures. \( M_{{{\text{H}}_{ 2} {\text{O}}}} \) is the molar mass of water.
The content of hydroxyl groups in FD-C samples was found more than in HD-C samples due to higher surface area of FD-C samples (Table 5). The large quantity of physisorbed water in the FD-C samples was due to high amount of hydroxyl groups present on its surface which could bond more water molecules through hydrogen bonding. The results (obtained from TG data) were found consistent with the results obtained from titration method.
Table 5

The effect of drying process on number of hydroxyl content


Temperature interval (°C)

Mass loss (wt%)

Absorbed water (mmol g−1 SiO2)

OH groupsa (mmol g−1 SiO2)

OH groupsb (mmol g−1 SiO2)



















aTGA method

bTitration method

4 Conclusions

The ultrafine silica particles were synthesized by modified sol–gel process under influence of different mixing mode of the reactants and drying techniques. By decreasing the feed rate of ammonia, smaller silica nanoparticles were obtained. Mixing mode showed significant effect on the average particle size and PSD. Dehydration process exerted efficiently via FD besides preservation of particle structure. FD provided high surface area and mesoporosity compared to conventional drying process. The TG thermogram results revealed four distinctive and defined weight loss steps for the samples obtained via FD. A high amount of silanol groups in silica nanoparticles can enhances interaction between surface hydroxyl groups and functional groups of organic moieties in an inorganic–organic hybrid. This unique property has made the silica nanoparticles suitable for preparation of silica-based polymeric nanocomposites.


The authors appreciate Ministry of Higher Education for financial support of this research under Fundamental Research Grant Scheme (FRGS, Grant No: 203/PKIMIA/671174). M. Jafarzadeh would like to express his gratitude to Universiti Sains Malaysia (USM) for USM-Fellowship.

Copyright information

© Springer Science+Business Media, LLC 2009