- Email: [email protected]
Tailoring the properties of sub-3 μm silica core–shell particles prepared by a multilayer-by-multilayer process
Journal of Colloid and Interface Science 437 (2015) 50–57
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Tailoring the properties of sub-3 lm silica core–shell particles prepared by a multilayer-by-multilayer process Hanjiang Dong, John D. Brennan ⇑ Biointerfaces Institute and Department of Chemistry & Chemical Biology, McMaster University, 1280 Main St. West, Hamilton, ON L8S 4L8, Canada
a r t i c l e
i n f o
Article history: Received 10 May 2014 Accepted 15 September 2014 Available online 28 September 2014 Keywords: Silica core–shell particles Porous shells Post-treatment procedures Thermal stability Mechanical stability
a b s t r a c t Sub-3 lm silica core–shell particles (CSPs) were fabricated by a multilayer-by-multilayer method recently developed in our group. In this work, we report on methods to prepare and modify the properties of these CSPs by high temperature calcination, pore size enlargement under basic conditions, and rehydrolyzation in boiling water to make them more suitable as starting materials for preparation of HPLC columns. The chemical, physical and mechanical properties of these modified CSPs were characterized by scanning electron microscopy (SEM), infrared spectroscopy (IR), thermogravimetric analysis (TGA), and nitrogen sorption porosimetry. CSPs obtained after these treatments were observed to have the following properties: particle diameter 2.7 lm, shell thickness 0.5 lm, surface area 200 m2/g, pore diameter 10 nm (and almost no mesopores), pore volume 0.5 cc/g, and Si–OH group surface concentration 4 OH/nm2. These properties are in line with those of commercially available sub-3 lm CSP products. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Silica core–shell particles (CSPs) consisting of a nonporous silica core and a porous silica shell are a special type of partially porous particles that have found wide use in the development of chromatographic stationary phases [1–4]. CSPs are also known as pellicular, superficially porous, fused core–shell, Poroshell, or shell particles. Interest in these materials dates back to the 1960s, with the goal of designing chromatographic stationary phases with less resistance to solute mass transfer compared to totally porous materials [5]. However, various early versions of CSPs with a small shell thickness/core diameter ratio (<0.06), large core size (>4.5 lm), and large pore size (>20 nm) had limited commercial success as packing materials due to the rapid development of small, totally porous particles that provided better separation efficiency and higher sample loading capacity [6]. Another key drawback of these early CSPs was the loose connection between not only core and nanoparticle coating but also between nanoparticles due to the relatively large nanoparticles used to prepare the porous shells, which led to poor mechanical strength. Nevertheless, persistent efforts in industry finally led to breakthrough proprietary technologies with the introduction of Halo™ particles by Advanced Materials Technology
⇑ Corresponding author. E-mail address: [email protected] (J.D. Brennan). URL: http://www.brennanlab.ca (J.D. Brennan). http://dx.doi.org/10.1016/j.jcis.2014.09.033 0021-9797/Ó 2014 Elsevier Inc. All rights reserved.
(AMT, Wilmington, DE) in 2006, followed by the introduction of similar materials by several other companies. Such particles have the combined characteristics of narrow particle size distribution (standard deviation 5%), large pore size (9–30 nm), and high surface area (30–200 m2/g) [7]. The sub-3 lm particles deliver an unparallelled minimum reduced HETP (height equivalent to a theoretical plate) of ca. 1.2 for small molecular weight compounds, which is well below the reduced HETP of columns packed with totally porous particles (usually 2.0) [7]. However, the methods used to synthesize such particles remain proprietary and post-synthesis treatments of these particles have not been disclosed. Inspired by Kirkland’s original work [8] while using a poly(diallyldimethylammonium chloride) (PDADMA)/silica nanoparticle (6 nm)/silica core particle (500 nm) model system, we recently reported a modifed layer-by-layer process we termed a multilayerby-multilayer (ML-b-ML) deposition method [9]. In that work, we found that shell growth could be adjusted by changing the solution pH and ionic strength. Up to 6–7 layers of nanoparticles could be deposited onto silica core particles and subsequent core–shell particles in 0.1 M NH4NO3 at pH 2.7 in a single coating cycle, thus significantly improving manufacturing efficiency. However, these particles after simply removing the PDADMA template at 550 °C had a significant fraction of small pores (<5 nm diameter) and poor mechanical stability, making them unsuitable for HPLC applications. In this paper, we utilized our original ML-b-ML process to rapidly grow sub-3 lm silica CSPs, but then undertook several
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
additional modification steps to make the particles more suitable for HPLC column development. These treatments including high temperature calcination to remove the PDADMA template and increase mechanical strength, high pH incubation to enlarge pore size, and rehydrolyzation to maximize surface Si–OH group concentration. The focus of this work was to investigate how these treatments affected the properties of CSPs (e.g., particle size, shell thickness, pore volume, surface area, average pore diameter, pore size distribution, mechanical strength, and surface Si–OH concentration), as these parameters have a dramatic influence on the performance of such particles as chromatographic stationary phases. The physical, chemical, and mechanical properties were characterized using a number of techniques such as SEM, IR, TGA, and nitrogen sorption porosimetry, and compared to the properties of commercially available core–shell products. 2. Experimental section 2.1. Materials Poly(diallyldimethylammonium chloride) (PDADMA, Average MW 100 kDa to 200 kDa, 35%) and tetraethoxysilane (TEOS, 98%) were purchased from Sigma–Aldrich. All other chemicals were of analytical grade and were used as received. All water was twice distilled and deionized to a specific resistance of at least 18 MO cm using a Milli-Q Synthesis A10 water purification system. 2.2. Preparation of nanoparticles, core particles and primary CSPs Both silica core particles (1.65 lm diameter) and silica nanoparticles (6 nm diameter) were prepared by the well-known Stöber process [10] and a detailed description of their preparation can be found elsewhere [11,12]. The preparation of CSPs using the ML-b-ML process was previously reported by our group [9]. Briefly, 1 g of silica cores (1.65 lm) were suspended in 50 ml H2O and the pH was adjusted to 2.7 by adding 1.0 M HNO3. Ionic strength was adjusted to 0.1 M by adding 5.0 M NH4NO3. To these silica cores was added 5 ml of a 5% (w/w) aqueous solution of positively charged PDADMA polyelectrolyte and the suspension was again adjusted to pH 2.7 if necessary. In the silica nanoparticle coating step, to 45 ml of the PDADMA coated particles (pH 2.7) was added 4 ml of 8% (w/w) silica nanoparticles (pH 2.5). The coating time for both PDADMA and silica nanoparticles was 20 min and the suspension was stirred with a magnetic stirrer at 500 rpm. The excess polymer and silica nanoparticles were removed by three repeated centrifugation (2000 rpm, 5 min)/water wash/redispersion cycles. The desired numbers of PDADMA/silica multilayers were deposited by repeating the procedure under the same conditions outlined above for the first layer. The as-prepared CSPs were dried at 80 °C for 10 h, 120 °C for 2 h, 200 °C and then 300 °C for 12 h each, and then calcinated at 550 °C for 1 day to remove PDADMA prior to further treatments. Under these experimental conditions, the increase of particle diameter per coating cycle is 60 ± 10 nm (i.e., 30 ± 5 nm shell thickness). As such, the targeted 650 nm shell thicknesses of as prepared particles, which is decreased to about 600 nm after calcination at 550 °C, requires 10–12 deposition cycles. 2.3. Post treatment of primary CSPs Thermal treatment of CSPs was carried out in air for 20 h in the range of 650–1050 °C. In order to increase the pore size, 0.1 g of CSPs obtained after the above thermal treatment were suspended in 10 ml H2O, which was then sonicated for 1 h to fully disperse the particles (Branson 50T Sonicator). To these particles was added
51
28% (v/v) NH4OH to adjust the pH to 11, which was then reacted at room temperature for 6 h with gentle stirring. Rehydroxylation of 0.1 g of particles obtained after the above thermal treatment was carried out in 10 ml H2O refluxed at 100 °C for 1 day. The resultant materials were collected by three repeated centrifugation (4000 rpm, 0.5 min)/water wash/redispersion cycles. The samples were then dried at 80 °C for 10 h, 120 °C for 2 h, 200 °C for 2 h, and 300 °C for 1 day prior to characterization. The samples for mechanical strength studies were prepared by a method identical to that used for the preparation of KBr pellets that were used for IR spectra. After testing, the KBr pellets were dissolved in water and the samples were collected by the centrifugation method.
2.4. Characterization of modified CSPs SEM images were obtained on a JEOL JSM-7000F scanning electron microscope with a 15 kV accelerating voltage and a probe distance of 10 nm. Porosity measurements were performed by nitrogen sorption porosimetry using a QuantaChrome Nova 2000 surface area/pore size analyzer. All samples were degassed at 300 °C for 3 h before measurement. The specific surface area (7 points, 0.025 < p/p0 < 0.30) and pore size distribution were calculated using the multi-point BET equation and BJH (Barrett, Joyner, and Halenda) model, respectively [13]. The total pore volume was found at the last point of the adsorption branch at a relative pressure of 0.99. Fourier transform infrared (FTIR) spectra of samples in KBr pellets were measured with a Bio-Rad FTS-40 spectrometer. Each spectrum represents an average of 32 scans with a 2 cm1 spectral resolution from 4000 to 400 cm1. Thermogravimetric analysis (TGA) was performed on a NETZSCH STA 409 instrument in the range of 30–1000 °C, with a heating rate of 5 °C min1, under 50 ml min1 argon flux.
3. Results and discussion 3.1. Effect of high temperature calcination Our originally reported CSPs were calcined at 550 °C to remove the PDADMA template. In the current study, we were interested in determining how different thermal treatments would affect the morphology and size of the final particles. Fig. 1 shows SEM images that show typical silica CSPs treated at different temperatures. The average particle size and the associated standard deviation of CSPs at each temperature were obtained by calculating values over approximately 100 particles (doublets (2%) are not included). The average particle sizes at 550 °C, 650 °C, 750 °C, 850 °C, 950 °C, and 1050 °C were 2.88, 2.79, 2.75, 2.72, 2.65 and 2.37 lm, respectively. As expected, the particle size deceases steadily with increasing temperature as a result of several factors, which will be discussed later. While the particle size distribution of CSPs was slightly broader than that of the core particles, their coefficients of variation were still very small at approximately 4.5%. The core diameter of particles that were calcined at 550 °C was 1.65 ± 0.05 lm and this value showed almost no change over the temperature range studied in this work. Therefore, the shell thicknesses of CSPs were 0.62 lm at 550 °C, 0.55 lm at 750 °C, and 0.50 lm at 950 °C, respectively. These values are in line with Halo particles and much smaller than the values of totally porous particles in the micron-size range, which have typical CV values of 15% [14]. The narrow particle size distribution is very likely to be the major reason that silica CSPs have exceptional separation efficiency as chromatographic packing materials [15]. The notable difference between Halo products and our materials is the particle surface morphology. As shown in Fig. 1C, our CSPs display a much
52
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Fig. 1. SEM images of CSPs treated at different temperatures.
smoother surface, which can be attributed to the very small nanoparticles (6 nm) used to synthesize the porous shells in this work. Fig. 2 compares the TGA/DTA curves of CSPs over the range of 200–1000 °C and TGA curves of samples that were calcined at 750 °C and 950 °C, respectively. Also included in Fig. 2 are two calcined samples after rehydroxylation, referred as 750 °C-RH and 950 °C-RH, which will be discussed later. The TGA/DTA curves are normalized at 200 °C in order to remove the effects of adsorbed water on the porous silica surfaces, which are very different for the five samples. The TGA curve of the initial, non-calcined sample (dried at 120 °C for 1 day) can be divided into four distinct regions based on the slopes of weight loss (200–260 °C, 260–330 °C, 330–650 °C, and 650–1000 °C). The weight loss at low temperature from 200 to 260 °C was 0.4%, mainly due to dehydration (the removal of physically adsorbed water) and the total weight loss of water from 30 to 260 °C was 7.8%, accompanied by an endothermic DTA peak. There was a drastic weight loss (5%) in a relatively narrow range of temperature from 260 to 330 °C, accompanied by a large and narrow exothermic DTA peak. We also observed
Fig. 2. TGA/DTA analysis of CSPs and their rehydroxylated counterparts (750 °C-RH and 950 °C-RH).
the change of the sample color from white to light brown to deep black over this temperature range, which can mainly be attributed to the decomposition of the PDADMA template. Further increasing temperature caused the sample to change color again to light brown (400 °C) and then finally white (>475 °C), indicating that decomposition was complete at temperatures >475 °C. In addition, dehydroxylation (the removal of silanol groups from the silica materials) occured as low as 200 °C and this reaction continued well above 1000 °C due to the different types of Si–OH groups (vicinal, geminal, and single isolated) that required different activation energies to be removed [16]. Therefore, the 6% weight loss from 330 to 650 °C, accompanied by a small but very broad exothermic DTA peak, was due to the combined effect of PDADMA decomposition and dehydroxylation, with the latter likely being the main contributor. The weight loss from 650 to 1000 °C was about 1% and was due to the dehydroxylation of more stable geminal and single Si–OH groups. Fig. 3 shows the FTIR spectra of CSPs as a function of treatment temperature. The interpretation of the IR spectra of porous silica gels has been extensively reported in the literature [17,18]. From Fig. 3, one can observe three characteristic bands for all silica materials located approximately at 1090 (with a shoulder at 1200), 810, and 465 cm1 that result from the silica framework vibrations and are associated with vibrational modes for the Si–O–Si antisymmetric stretch, Si–O–Si bend (or symmetric stretch), and Si–O–Si (or O–Si–O) rock, respectively. The weak bands at 1970 and 1865 cm1 are due to higher-order vibrations (overtones and combinations) of the SiO2 network. There are several bands resulting from the residual Si–OH groups and adsorbed H2O. The broad absorption ranging from 3000 to 3700 cm1 arises from stretching vibrations of hydrogen bonded H2O and various Si–OH groups, while the sharp but weak absorption band at 3745 cm1 is due to the stretching vibration of isolated Si– OH groups. The presence of H2O and Si–OH groups was further confirmed by the bands at 1630 cm1 (H–O–H deformation) and 965 cm1 (Si–OH stretching), respectively. However, the Si–OH stretching band at 956 cm1 can be observed only at low temperatures (<650 °C). There was a large decrease of H2O associated bands at 3450 cm1 and 1630 cm1 at elevated temperatures but these
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
53
Fig. 3. IR spectra of CSPs treated at different temperatures. (A) Hydroxyl stretching region. (B) Silica stretching region.
bands never completely disappeared, likely due to the strong adsorption of H2O by the residual Si–OH groups and KBr tablet during the IR experiments. The overall decrease in the H2O associated bands can be attributed to a decreased concentration of Si–OH groups with increased temperature (as indicated in Fig. 2). Fig. 4 shows nitrogen adsorption and desorption isotherms and desorption pore size distributions of CSPs as a function of temperature. As shown in Fig. 4A, our materials had typical type IV isotherms with a type H2 hysteresis loop, characteristic of mesoporous materials. The amounts adsorbed were progressively reduced with the increase of temperature. Similarly, the shapes of pore size distributions were very similar while the positions of the curves gradually decreased with increasing temperature (Fig. 4B). The effect of temperature on specific surface area (SSA), total pore volume (TPV), and average pore size (APS) values is listed in Table 1. Both SSA and TPV values decreased while APS
values remained similar (8.0 nm) as temperature increased up to 950 °C. Continued condensation reactions along with structural relaxation with increasing temperature probably lead to the fusing of nanoparticles that were in contact, growth of necks between these particles and skeleton densification. These changes do not alter pore sizes but do decrease the particle sizes (about 8% from 550 °C to 950 °C) and the associated pore volumes from 0.63 cc/g at 550 °C to 0.42 cc/g at 950 °C and surface areas from 314.2 m2/ g at 550 °C to 198.5 m2/g at 950 °C. For comparison, the surface area, pore volume, and pore size of Halo 2.7 lm particles are 124 m2/g, 0.288 cc/g, and 9.5 nm, respectively [14]. At 1050 °C, viscous flow of the porous nanoparticles may become possible, which leads to the pore collapse of the porous shell, accompanied by a significant shrinkage of core–shell particles from 2.65 lm at 950 °C to 2.37 lm at 1050 °C. As such, the TPV value drops substantially to 0.059 cc/g, which is the main reason that the
Fig. 4. (A) Nitrogen sorption isotherms and (B) corresponding pore size distribution of CSPs after heat treatment at different temperature. (C) is the normalized pore size distribution data from panel (B).
54
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Table 1 Nitrogen sorption data for CSPs calcined at different temperatures. Temperature (°C)
SSA (m2/g)
TPV (cc/g)
APS (nm)
550 650 750 850 950 1050
314.2 ± 12.5 295.5 ± 6.4 273.2 ± 6.5 258.7 ± 5.2 198.5 ± 4.9 32.0 ± 2.7
0.63 ± 0.08 0.56 ± 0.07 0.53 ± 0.06 0.51 ± 0.06 0.42 ± 0.05 0.06 ± 0.02
8.15 ± 0.10 7.80 ± 0.09 7.94 ± 0.09 8.02 ± 0.10 8.12 ± 0.10 7.08 ± 0.09
SSA – specific surface area; TPV – total pore volume; APS – average pore size.
surface area at 1050 °C (31.95 m2/g) is more than 10 times less than that at 550 °C (314.2 m2/g). The upper limit temperature for this particular sample without pore collapse was 1000 °C, which is similar to colloidal gels prepared by other methods [19,20]. Finally, we point out that not only did the average pore sizes at different temperatures up to 950 °C remain similar but so to did their pore size distributions, as shown in Fig. 4C after normalizing their data to those obtained at 550 °C. These results indicate that pores of different sizes contract at a similar rate, even though the smaller pores have higher surface energy. This surprising finding seems to be in contrast to the intuition that small pores disappear before large ones due to the higher surface energy of the former. The reasons for our finding remain unknown. It is likely that pore disappearance and/or collapse requires the long migration of silica materials such as viscous flow or Ostwald ripening, which may not occur during high temperature treatment up to 1000 °C. In summary, high temperature calcination leads to CSPs with strong mechanical strength (See Section 3.4) and very similar particle geometry as compared to commercial products. However, further treatments are required to increase their pore size (10 nm) and the surface Si–OH concentration in order for them to be used as HPLC columns [21]. In the following, we discuss the effects of different treatments on pore size and Si-OH content of CSPs after calcination at 750 °C and 950 °C. 3.2. Rehydroxylation CSPs after high temperature calcination lose a significant amount of Si-OH groups due to condensation reactions. The purpose of rehydroxylation for chromatographic applications is to restore these lost Si–OH groups, allowing them to be silanized at a high density [21]. Fig. 2 shows the TGA analysis of four CSP samples: two samples calcinated at 750 °C and 950 °C, respectively, and their counterparts after rehydroxylation. The weight loss and Si–OH group concentration aOH of those samples are listed in Table 2. aOH, expressed as the number of Si–OH group per nm2, was estimated from the weight loss (WL) in Fig. 2 between 200 and 1000 °C using the following formula:
aOH ¼ ððWL 100Þ=18ÞÞNA 1018 S1
ð1Þ
which is modified from the literature [22]. In the above formula, the first term (WL 100)/18 is the concentration of OH groups on the CSP surface in the unit of mol of Si–OH group/1 g of CSPs, NA is the Avogadro number, and S is the specific surface area of CSPs. Table 2 Comparison of the weight loss and Si–OH surface concentration before and after rehydroxylation. Sample
Surface area (m2/g)
Weight loss (%)
Si–OH concentration (nm2)
750 °C 750 °C-RH 950 °C 950 °C-RH
273.2 ± 6.5 266.5 ± 5.4 198.5 ± 4.9 195.7 ± 4.7
1.4 3.5 0.9 2.1
1.72 4.40 1.52 3.60
From Table 2, one can find that the weight loss of CSP samples from 200 to 1000 °C was in the order of 750 °C-RH (3.5%) > 950 °CRH (2.1%) > 750 °C (1.4%) > 950 °C (0.9%). Their corresponding aOH values were 4.40, 3.60, 1.72, and 1.52. While the aOH values are reasonable compared to the literature information summarized by Zhuravelev [16], our data may not be accurate owing to the method by which we obtained the weight loss data from TGA measurements. In the low temperature region (e.g., <250 °C), the weight loss of Si–OH concentration could be overestimated due to the desorption of physically adsorbed water. On the other hand, Si–OH groups are still clearly present at 1050 °C (see Fig. 3) and condensation reactions continue up to 1200 °C, which will lead to the underestimation of the Si–OH concentration on the silica surface, partially compensating for this issue. Future work will use more accurate tools such as the deuterium-exchange method combined with mass spectrometric analysis to determine the aOH values [22]. Nevertheless, our data show clear and expected trends in terms of the effects of calcination and rehydroxylation. Calcination at high temperature led to a decrease in aOH due to continuous condensation reactions. The subsequent rehydroxylation significantly increased the surface Si–OH coverage even if the sample was calcined at 950 °C. However, higher calcination temperatures resulted in a lower degree of rehydroxylation due to the high hydrophobicity of the surface and high activation energy for the breakdown of Si-O-Si bonds. These trends are further supported by the IR spectra shown in Fig. 5, in which the same calcined samples before and after rehydroxylation are compared. While it is difficult to analyze the data quantitatively, the IR results clearly indicate that rehydroxylation led to an increase of surface Si–OH concentrations (750 °C-RH > 750 °C and 950 °C-RH > 950 °C) and was less complete after high temperature calcination (750 °C-RH > 950 °C-RH). 3.3. Pore size enlargement under basic conditions As stated previously, the pore size of CSPs is 8 nm, which remains stable up to 1000 °C. This value is at the low end for chromatographic column applications. There are a number of chemicals (NH4F, HF, NaOH, KOH, etc.) and processes (hydrothermal treatment, aging in the mother liquor) that can be used to tailor the pore size of silica materials. The basic principle of these post synthesis treatment methods is the well-known Ostwald ripening process: dissolution of silica regions with small positive curvature and redeposition of the dissolved silicate species on the surfaces of silica regions with small negative curvature [23]. We attempted to avoid the use of corrosive chemicals and autoclaving to alter the mesopore sizes of CSPs in order to make eventual manufacturing more facile and environmentally friendly. Our method to enlarge the pore size of CSPs was, therefore, carried out at room temperature in aqueous solution by adding NH4OH to adjust the pH to approximately 11 for 6 h. We refer these samples as 750 °C-pH and 950 °C-pH based on their calcination temperature. Fig. 6 shows the isotherms of CSP samples 750 °C-pH and 950 °C-pH and their corresponding desorption pore size distributions (PSDs). Fig. 6A shows that CSPs after aging in diluted NH4OH solution had typical type IV isotherms with a H2 hysterisis loop associated with capillary condensation in the mesopore structures, similar to those shown in Fig. 4A before high pH treatment. However, the whole PSD curves were shifted to the large pore size region compared to those obtained before high pH treatment, while the small mesopores (<5 nm) were almost completely absent. The PSD curves of these two samples showed a monomodal distribution with peak positions centered at 17 nm (750 °C-pH) and 12 nm (950 °C-pH). From Fig. 6B, one can observe that our materials displayed narrower PSDs (much higher peak position > 1.2 cc/g) compared to commercial products [14] and
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Fig. 5. IR spectra of CSPs before and after rehydroxylation after calcining at 750 °C and 950 °C, respectively.
had essentially no pores with diameters less than 4 nm. These properties may be important to reduce the solute mass transfer resistance and increase column separation efficiency in eventual HPLC applications. The specific surface area, total pore volume, and average pore size of these samples were 199.6 m2/g, 0.529 cc/g, and 11.79 nm for the 750 °C-pH sample and 185.0 m2/g, 0.453 cc/g, and 9.80 nm, for the 950 °C-pH sample, respectively. These results indicate that the post treatment (pH 11, room temperature, and reaction time of 6 h) was very effective for producing CSPs with a 10 nm pore size with a high surface area (190 m2/g). As stated previously, the thermodynamic basis for these changes is Ostwald ripening. In our particular case, small silica nanoparticles and aggregated clusters, and especially those with small positive curvature that form the small intraparticle pores that have high free energies, were dissolved first at high pH. The dissolved silicate species then redeposited onto large particles (or clusters) of lower solubility, thus eliminating the small pores and increasing the average pore sizes but reducing the surface areas of CSPs. Similar to rehydroxylation, high pH treatment also involves the breakdown of Si–O–Si bonds, which become more stable with increasing temperature. As such, the porosity of the 750 °C-pH sample showed less change compared to the 950 °C-pH sample.
3.4. Mechanical strength of CSPs For chromatographic applications, the mechanical strength of the particles is critically important because HPLC column packing is carried out at very high pressure (e.g., 10,000 psi) and with the
55
advent of ultrahigh pressure LC, experiments may also be done at pressures approaching 10,000 psi. However, the porous shells prepared by the layer-by-layer method are assembled by loosely packed silica nanoparticles held together by weak electrostatic attractions and hydrogen bonding. These noncovalent forces are unlikely to impart sufficient mechanical stability to produce robust shells, as demonstrated by others [24,25]. To enhance the mechanical strength of porous silica films, low temperature methods such as hydrothermal treatment [24] and atomic layer deposition [26] have recently been developed. The advantage of these methods is that they are compatible with organic substrates, however it is highly unlikely that silica shells formed by these methods, which simply increase the neck thickness of connected nanoparticles, are robust enough to endure the packing pressure of HPLC columns. In this work, we explored high temperature treatment as a method to enhance the mechanical strength of CSPs. This is the most widely used method to increase the robustness of porous silica materials prepared by numerous approaches [20,27]. In order to test the mechanical strength of our materials, we prepared KBr pellets similar to those used for recording IR spectra, using different final pressures to form the pellets. Fig. 7 shows SEM images of the 5 lm totally porous particles and CSPs after being subjected to different compressive pressures. The 5 lm totally porous silica particles are commercial products and are used as a reference material to evaluate the mechanical properties of our materials. CSPs are named based on how these materials are post treated and tested. For example, 750 °C-pH-2500 psi represents CSPs that were calcinated at 750 °C followed by pore size enlargement at high pH and compression at 2500 psi to produce the KBr pellet. From Fig. 7, one can produce semi-quantitative data on mechanical stability by determining the percentage of particles that remained intact after compression. We note that 90% of 5 lm particles remained intact at 500 psi, while only 10% were intact after treatment at 2500 psi. CSPs treated at 750 °C remained largely intact (>98%) at 500 or 1000 psi, but only 40% of these particles were intact after compression at 2500 psi. Finally, >99% of particles treated at 950 °C remained intact at 2500 psi. These results indicate that CSPs after high temperature treatment are much more robust than 5 lm porous particles, which can withstand at least 6000 psi pressure during slurry packing. The main reason that porous particles show a much higher compressive modulus during slurry packing is likely due to the fact that the liquid that fills the pores provides extra resistance under compressive loading. These results also indicate that the mechanical strength of CSPs increases with increasing calcination temperature as explained below. First, it is well-known that the mechanical properties of
Fig. 6. (A) Nitrogen sorption isotherms and (B) corresponding pore size distribution of CSPs treated at different temperatures and then aged under NH4OH solutions at pH = 11.
56
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Fig. 7. SEM images of 5 lm totally porous particles and CSPs treated at different temperatures and after high pH pore enlargement that are subjected to high compressive pressure.
porous silica materials (e.g., elastic and compressive modulus) are dependent on their density (q) and have a power-law relationship with the density (E / qa), where a (ranging from 3 to 4) is the scaling exponent [28,29]. Increasing the density of CSPs with increasing temperature is obviously due to shrinkage (Fig. 1) and reduced pore volume of the particles (Table 1). Second, we expect the presence of string-of-bead (individual nanoparticle) like structures in the as-prepared particles due to the weak interactions involved in the mL-b-mL method. At elevated temperature, especially at 950 °C, the individual nanoparticles are expected to lose their identity by fusing together and also onto the core particles. This change is the result of continuing condensation of Si–OH groups (see Fig. 2), structural relaxation, and partial sintering, evidenced by an accompanying decrease in particle surface area and pore volume. The thus formed thickened necks stabilize the solid network and increase network connectivity, therefore, enhancing the mechanical strength of CSPs. Finally, condensation reactions themselves, similar to aging in the sol–gel process, which increase the degree of condensation, are also expected to play a positive role in strengthening porous silica materials. Therefore, the enhanced mechanical strength of CSPs with increased calcination temperature can be attributed to the reduced porosity, the high degree of condensation, the formation of thick necks between nanoparticles and the fusion of nanoparticles onto the core particles.
a very narrow pore size distribution and no pores less than 4 nm in diameter. High temperature calcination also drastically increased the mechanical strength of CSPs. CSPs calcinated at 950 °C and after high pH treatment showed much higher mechanical strength than 5 lm totally porous commercial silica particles. In addition, the above CSPs possessed the following properties: particle diameter 2.65 lm, shell thickness 0.5 lm, surface area 185 m2/g, pore diameter 9.80 nm, pore volume 0.453 cc/g, and Si–OH group surface concentration 3.6 OH/nm2, which are in line with the values of commercial products. The performance of these materials as HPLC stationary phases is currently being evaluated and the results will be reported in due course.
4. Conclusions
References
In this work, the previously developed ML-b-ML process was used to prepare sub-3 lm silica core–shell particles with a focus on altering their properties to make them comparable to commercial products. To this end, we performed various post-synthesis treatments including high temperature calcination, high pH pore enlargement, and rehydroxylation. We found that surface areas and pore volumes decreased steadily while pore sizes and their distributions remained similar with increases in temperature up to 1000 °C. These results indicate that the porous shells prepared by the ML-b-ML process have similar thermal stability to colloidal gels and the small pores are as stable as the large pores. Rehydroxylation (100 °C in water for 24 h) and pore size enlargement (in NH4OH, pH = 11 for 6 h) involve the breakdown of Si–O–Si bonds, which becomes increasingly difficult with increases in calcining temperature. Nevertheless, the high pH treatment effectively led to increases in pore size from 8 nm to 10 nm with
Acknowledgments The authors thank the Natural Sciences and Engineering Research Council of Canada for funding this work through a network grant-SENTINEL Canadian Network for the Development and Use of Bioactive Paper. We also thank Aglient Corporation for providing funding through an Agilent University Research Fund grant. The authors also thank the Canada Foundation for Innovation and the Ontario Innovation Trust for providing funding to the Biointerfaces Institute. JDB holds the Canada Research Chair in Bioanalytical Chemistry and Biointerfaces.
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
H.J. Dong, J.D. Brennan, J. Mater. Chem. 22 (2012) 13197. S. Fekete, E. Olah, J. Fekete, J. Chromatogr. A 1228 (2012) 57. H. Nishi, K. Nagamatsu, Anal. Sci. 30 (2014) 205. K.K. Unger, R. Skudas, M.M. Schulte, J. Chromatogr. A 1184 (2008) 393. J.J. Kirkland, J. Chromatgr. Sci. (1969) 361. F. Gritti, I. Leonardis, D. Shock, P. Stevenson, A. Shalliker, G. Guiochon, J. Chromatogr. A 1217 (2010) 1589. G. Guiochon, F. Gritti, J. Chromatgr. A 1218 (2011) 1915. J.J.K.a.T.J. Langlois, US. Patent 0189944 A1 (2007). H.J. Dong, J.D. Brennan, Chem. Commun. 47 (2011) 1207. W. Stöber, A. Fink, E. Bohn, J. Colloid Interface Sci. 26 (1968) 62. H.G. Klaus Unger, Joachim Kinkel, US. Patent 4777520. W.C.T.-C. Wei, US. Patent 7846337 B2. E.P. Barrett, L. Joyner, P.P.J. Halenda, J. Am. Chem. Soc. 73 (1951) 373. F. Gritti, I. Leonardis, J. Abia, G. Guiochon, J. Chromatogr. A 1217 (2010) 3819. J.J. Kirkland, T.J. Langlois, J.J. DeStefano, Am. Lab. 39 (2007) 18. L.T. Zhuravlev, Colloids Surfaces A – Physicochem. Eng. Asp. 173 (2000) 1. D.L. Wood, E.M. Rabinovich, D.W. Johnson, J.B. Macchesney, E.M. Vogel, J. Am. Ceram. Soc. 66 (1983) 693. T.A. Guiton, C.G. Pantano, Colloids Surfaces A – Physicochem. Eng. Asp. 74 (1993) 33.
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57 [19] T. Yamada, M. Nakajima, T. Suemoto, T. Uchino, J. Phys. Chem. C 111 (2007) 12973. [20] D.W. Johnson, E.M. Rabinovich, J.B. Macchesney, E.M. Vogel, J. Am. Ceram. Soc. 66 (1983) 688. [21] J. Kohler, J.J. Kirkland, J. Chromatogr. 385 (1987) 125. [22] L.T. Zhuravlev, Langmuir 3 (1987) 316. [23] R.K. Iler, The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties and Biochemistry, Wiley, New York, 1979.
57
[24] Z. Gemici, H. Shimomura, R.E. Cohen, M.F. Rubner, Langmuir 24 (2008) 2168. [25] L.E. Blue, J.W. Jorgenson, J. Chromatogr. A 1218 (2011) 7989. [26] M.I. Dafinone, G. Feng, T. Brugarolas, K.E. Tettey, D. Lee, ACS Nano 5 (2011) 5078. [27] S. Satoh, K. Susa, I. Matsuyama, T. Suganuma, J. Am. Ceram. Soc. 68 (1985) 399. [28] T. Woignier, J. Phalippou, R. Vacher, J. Mater. Res. 4 (1989) 688. [29] A. Jain, S. Rogojevic, S. Ponoth, N. Agarwal, I. Matthew, W.N. Gill, P. Persans, M. Tomozawa, J.L. Plawsky, E. Simonyi, Thin Solid Films 398 (2001) 513.
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Tailoring the properties of sub-3 lm silica core–shell particles prepared by a multilayer-by-multilayer process Hanjiang Dong, John D. Brennan ⇑ Biointerfaces Institute and Department of Chemistry & Chemical Biology, McMaster University, 1280 Main St. West, Hamilton, ON L8S 4L8, Canada
a r t i c l e
i n f o
Article history: Received 10 May 2014 Accepted 15 September 2014 Available online 28 September 2014 Keywords: Silica core–shell particles Porous shells Post-treatment procedures Thermal stability Mechanical stability
a b s t r a c t Sub-3 lm silica core–shell particles (CSPs) were fabricated by a multilayer-by-multilayer method recently developed in our group. In this work, we report on methods to prepare and modify the properties of these CSPs by high temperature calcination, pore size enlargement under basic conditions, and rehydrolyzation in boiling water to make them more suitable as starting materials for preparation of HPLC columns. The chemical, physical and mechanical properties of these modified CSPs were characterized by scanning electron microscopy (SEM), infrared spectroscopy (IR), thermogravimetric analysis (TGA), and nitrogen sorption porosimetry. CSPs obtained after these treatments were observed to have the following properties: particle diameter 2.7 lm, shell thickness 0.5 lm, surface area 200 m2/g, pore diameter 10 nm (and almost no mesopores), pore volume 0.5 cc/g, and Si–OH group surface concentration 4 OH/nm2. These properties are in line with those of commercially available sub-3 lm CSP products. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Silica core–shell particles (CSPs) consisting of a nonporous silica core and a porous silica shell are a special type of partially porous particles that have found wide use in the development of chromatographic stationary phases [1–4]. CSPs are also known as pellicular, superficially porous, fused core–shell, Poroshell, or shell particles. Interest in these materials dates back to the 1960s, with the goal of designing chromatographic stationary phases with less resistance to solute mass transfer compared to totally porous materials [5]. However, various early versions of CSPs with a small shell thickness/core diameter ratio (<0.06), large core size (>4.5 lm), and large pore size (>20 nm) had limited commercial success as packing materials due to the rapid development of small, totally porous particles that provided better separation efficiency and higher sample loading capacity [6]. Another key drawback of these early CSPs was the loose connection between not only core and nanoparticle coating but also between nanoparticles due to the relatively large nanoparticles used to prepare the porous shells, which led to poor mechanical strength. Nevertheless, persistent efforts in industry finally led to breakthrough proprietary technologies with the introduction of Halo™ particles by Advanced Materials Technology
⇑ Corresponding author. E-mail address: [email protected] (J.D. Brennan). URL: http://www.brennanlab.ca (J.D. Brennan). http://dx.doi.org/10.1016/j.jcis.2014.09.033 0021-9797/Ó 2014 Elsevier Inc. All rights reserved.
(AMT, Wilmington, DE) in 2006, followed by the introduction of similar materials by several other companies. Such particles have the combined characteristics of narrow particle size distribution (standard deviation 5%), large pore size (9–30 nm), and high surface area (30–200 m2/g) [7]. The sub-3 lm particles deliver an unparallelled minimum reduced HETP (height equivalent to a theoretical plate) of ca. 1.2 for small molecular weight compounds, which is well below the reduced HETP of columns packed with totally porous particles (usually 2.0) [7]. However, the methods used to synthesize such particles remain proprietary and post-synthesis treatments of these particles have not been disclosed. Inspired by Kirkland’s original work [8] while using a poly(diallyldimethylammonium chloride) (PDADMA)/silica nanoparticle (6 nm)/silica core particle (500 nm) model system, we recently reported a modifed layer-by-layer process we termed a multilayerby-multilayer (ML-b-ML) deposition method [9]. In that work, we found that shell growth could be adjusted by changing the solution pH and ionic strength. Up to 6–7 layers of nanoparticles could be deposited onto silica core particles and subsequent core–shell particles in 0.1 M NH4NO3 at pH 2.7 in a single coating cycle, thus significantly improving manufacturing efficiency. However, these particles after simply removing the PDADMA template at 550 °C had a significant fraction of small pores (<5 nm diameter) and poor mechanical stability, making them unsuitable for HPLC applications. In this paper, we utilized our original ML-b-ML process to rapidly grow sub-3 lm silica CSPs, but then undertook several
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
additional modification steps to make the particles more suitable for HPLC column development. These treatments including high temperature calcination to remove the PDADMA template and increase mechanical strength, high pH incubation to enlarge pore size, and rehydrolyzation to maximize surface Si–OH group concentration. The focus of this work was to investigate how these treatments affected the properties of CSPs (e.g., particle size, shell thickness, pore volume, surface area, average pore diameter, pore size distribution, mechanical strength, and surface Si–OH concentration), as these parameters have a dramatic influence on the performance of such particles as chromatographic stationary phases. The physical, chemical, and mechanical properties were characterized using a number of techniques such as SEM, IR, TGA, and nitrogen sorption porosimetry, and compared to the properties of commercially available core–shell products. 2. Experimental section 2.1. Materials Poly(diallyldimethylammonium chloride) (PDADMA, Average MW 100 kDa to 200 kDa, 35%) and tetraethoxysilane (TEOS, 98%) were purchased from Sigma–Aldrich. All other chemicals were of analytical grade and were used as received. All water was twice distilled and deionized to a specific resistance of at least 18 MO cm using a Milli-Q Synthesis A10 water purification system. 2.2. Preparation of nanoparticles, core particles and primary CSPs Both silica core particles (1.65 lm diameter) and silica nanoparticles (6 nm diameter) were prepared by the well-known Stöber process [10] and a detailed description of their preparation can be found elsewhere [11,12]. The preparation of CSPs using the ML-b-ML process was previously reported by our group [9]. Briefly, 1 g of silica cores (1.65 lm) were suspended in 50 ml H2O and the pH was adjusted to 2.7 by adding 1.0 M HNO3. Ionic strength was adjusted to 0.1 M by adding 5.0 M NH4NO3. To these silica cores was added 5 ml of a 5% (w/w) aqueous solution of positively charged PDADMA polyelectrolyte and the suspension was again adjusted to pH 2.7 if necessary. In the silica nanoparticle coating step, to 45 ml of the PDADMA coated particles (pH 2.7) was added 4 ml of 8% (w/w) silica nanoparticles (pH 2.5). The coating time for both PDADMA and silica nanoparticles was 20 min and the suspension was stirred with a magnetic stirrer at 500 rpm. The excess polymer and silica nanoparticles were removed by three repeated centrifugation (2000 rpm, 5 min)/water wash/redispersion cycles. The desired numbers of PDADMA/silica multilayers were deposited by repeating the procedure under the same conditions outlined above for the first layer. The as-prepared CSPs were dried at 80 °C for 10 h, 120 °C for 2 h, 200 °C and then 300 °C for 12 h each, and then calcinated at 550 °C for 1 day to remove PDADMA prior to further treatments. Under these experimental conditions, the increase of particle diameter per coating cycle is 60 ± 10 nm (i.e., 30 ± 5 nm shell thickness). As such, the targeted 650 nm shell thicknesses of as prepared particles, which is decreased to about 600 nm after calcination at 550 °C, requires 10–12 deposition cycles. 2.3. Post treatment of primary CSPs Thermal treatment of CSPs was carried out in air for 20 h in the range of 650–1050 °C. In order to increase the pore size, 0.1 g of CSPs obtained after the above thermal treatment were suspended in 10 ml H2O, which was then sonicated for 1 h to fully disperse the particles (Branson 50T Sonicator). To these particles was added
51
28% (v/v) NH4OH to adjust the pH to 11, which was then reacted at room temperature for 6 h with gentle stirring. Rehydroxylation of 0.1 g of particles obtained after the above thermal treatment was carried out in 10 ml H2O refluxed at 100 °C for 1 day. The resultant materials were collected by three repeated centrifugation (4000 rpm, 0.5 min)/water wash/redispersion cycles. The samples were then dried at 80 °C for 10 h, 120 °C for 2 h, 200 °C for 2 h, and 300 °C for 1 day prior to characterization. The samples for mechanical strength studies were prepared by a method identical to that used for the preparation of KBr pellets that were used for IR spectra. After testing, the KBr pellets were dissolved in water and the samples were collected by the centrifugation method.
2.4. Characterization of modified CSPs SEM images were obtained on a JEOL JSM-7000F scanning electron microscope with a 15 kV accelerating voltage and a probe distance of 10 nm. Porosity measurements were performed by nitrogen sorption porosimetry using a QuantaChrome Nova 2000 surface area/pore size analyzer. All samples were degassed at 300 °C for 3 h before measurement. The specific surface area (7 points, 0.025 < p/p0 < 0.30) and pore size distribution were calculated using the multi-point BET equation and BJH (Barrett, Joyner, and Halenda) model, respectively [13]. The total pore volume was found at the last point of the adsorption branch at a relative pressure of 0.99. Fourier transform infrared (FTIR) spectra of samples in KBr pellets were measured with a Bio-Rad FTS-40 spectrometer. Each spectrum represents an average of 32 scans with a 2 cm1 spectral resolution from 4000 to 400 cm1. Thermogravimetric analysis (TGA) was performed on a NETZSCH STA 409 instrument in the range of 30–1000 °C, with a heating rate of 5 °C min1, under 50 ml min1 argon flux.
3. Results and discussion 3.1. Effect of high temperature calcination Our originally reported CSPs were calcined at 550 °C to remove the PDADMA template. In the current study, we were interested in determining how different thermal treatments would affect the morphology and size of the final particles. Fig. 1 shows SEM images that show typical silica CSPs treated at different temperatures. The average particle size and the associated standard deviation of CSPs at each temperature were obtained by calculating values over approximately 100 particles (doublets (2%) are not included). The average particle sizes at 550 °C, 650 °C, 750 °C, 850 °C, 950 °C, and 1050 °C were 2.88, 2.79, 2.75, 2.72, 2.65 and 2.37 lm, respectively. As expected, the particle size deceases steadily with increasing temperature as a result of several factors, which will be discussed later. While the particle size distribution of CSPs was slightly broader than that of the core particles, their coefficients of variation were still very small at approximately 4.5%. The core diameter of particles that were calcined at 550 °C was 1.65 ± 0.05 lm and this value showed almost no change over the temperature range studied in this work. Therefore, the shell thicknesses of CSPs were 0.62 lm at 550 °C, 0.55 lm at 750 °C, and 0.50 lm at 950 °C, respectively. These values are in line with Halo particles and much smaller than the values of totally porous particles in the micron-size range, which have typical CV values of 15% [14]. The narrow particle size distribution is very likely to be the major reason that silica CSPs have exceptional separation efficiency as chromatographic packing materials [15]. The notable difference between Halo products and our materials is the particle surface morphology. As shown in Fig. 1C, our CSPs display a much
52
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Fig. 1. SEM images of CSPs treated at different temperatures.
smoother surface, which can be attributed to the very small nanoparticles (6 nm) used to synthesize the porous shells in this work. Fig. 2 compares the TGA/DTA curves of CSPs over the range of 200–1000 °C and TGA curves of samples that were calcined at 750 °C and 950 °C, respectively. Also included in Fig. 2 are two calcined samples after rehydroxylation, referred as 750 °C-RH and 950 °C-RH, which will be discussed later. The TGA/DTA curves are normalized at 200 °C in order to remove the effects of adsorbed water on the porous silica surfaces, which are very different for the five samples. The TGA curve of the initial, non-calcined sample (dried at 120 °C for 1 day) can be divided into four distinct regions based on the slopes of weight loss (200–260 °C, 260–330 °C, 330–650 °C, and 650–1000 °C). The weight loss at low temperature from 200 to 260 °C was 0.4%, mainly due to dehydration (the removal of physically adsorbed water) and the total weight loss of water from 30 to 260 °C was 7.8%, accompanied by an endothermic DTA peak. There was a drastic weight loss (5%) in a relatively narrow range of temperature from 260 to 330 °C, accompanied by a large and narrow exothermic DTA peak. We also observed
Fig. 2. TGA/DTA analysis of CSPs and their rehydroxylated counterparts (750 °C-RH and 950 °C-RH).
the change of the sample color from white to light brown to deep black over this temperature range, which can mainly be attributed to the decomposition of the PDADMA template. Further increasing temperature caused the sample to change color again to light brown (400 °C) and then finally white (>475 °C), indicating that decomposition was complete at temperatures >475 °C. In addition, dehydroxylation (the removal of silanol groups from the silica materials) occured as low as 200 °C and this reaction continued well above 1000 °C due to the different types of Si–OH groups (vicinal, geminal, and single isolated) that required different activation energies to be removed [16]. Therefore, the 6% weight loss from 330 to 650 °C, accompanied by a small but very broad exothermic DTA peak, was due to the combined effect of PDADMA decomposition and dehydroxylation, with the latter likely being the main contributor. The weight loss from 650 to 1000 °C was about 1% and was due to the dehydroxylation of more stable geminal and single Si–OH groups. Fig. 3 shows the FTIR spectra of CSPs as a function of treatment temperature. The interpretation of the IR spectra of porous silica gels has been extensively reported in the literature [17,18]. From Fig. 3, one can observe three characteristic bands for all silica materials located approximately at 1090 (with a shoulder at 1200), 810, and 465 cm1 that result from the silica framework vibrations and are associated with vibrational modes for the Si–O–Si antisymmetric stretch, Si–O–Si bend (or symmetric stretch), and Si–O–Si (or O–Si–O) rock, respectively. The weak bands at 1970 and 1865 cm1 are due to higher-order vibrations (overtones and combinations) of the SiO2 network. There are several bands resulting from the residual Si–OH groups and adsorbed H2O. The broad absorption ranging from 3000 to 3700 cm1 arises from stretching vibrations of hydrogen bonded H2O and various Si–OH groups, while the sharp but weak absorption band at 3745 cm1 is due to the stretching vibration of isolated Si– OH groups. The presence of H2O and Si–OH groups was further confirmed by the bands at 1630 cm1 (H–O–H deformation) and 965 cm1 (Si–OH stretching), respectively. However, the Si–OH stretching band at 956 cm1 can be observed only at low temperatures (<650 °C). There was a large decrease of H2O associated bands at 3450 cm1 and 1630 cm1 at elevated temperatures but these
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
53
Fig. 3. IR spectra of CSPs treated at different temperatures. (A) Hydroxyl stretching region. (B) Silica stretching region.
bands never completely disappeared, likely due to the strong adsorption of H2O by the residual Si–OH groups and KBr tablet during the IR experiments. The overall decrease in the H2O associated bands can be attributed to a decreased concentration of Si–OH groups with increased temperature (as indicated in Fig. 2). Fig. 4 shows nitrogen adsorption and desorption isotherms and desorption pore size distributions of CSPs as a function of temperature. As shown in Fig. 4A, our materials had typical type IV isotherms with a type H2 hysteresis loop, characteristic of mesoporous materials. The amounts adsorbed were progressively reduced with the increase of temperature. Similarly, the shapes of pore size distributions were very similar while the positions of the curves gradually decreased with increasing temperature (Fig. 4B). The effect of temperature on specific surface area (SSA), total pore volume (TPV), and average pore size (APS) values is listed in Table 1. Both SSA and TPV values decreased while APS
values remained similar (8.0 nm) as temperature increased up to 950 °C. Continued condensation reactions along with structural relaxation with increasing temperature probably lead to the fusing of nanoparticles that were in contact, growth of necks between these particles and skeleton densification. These changes do not alter pore sizes but do decrease the particle sizes (about 8% from 550 °C to 950 °C) and the associated pore volumes from 0.63 cc/g at 550 °C to 0.42 cc/g at 950 °C and surface areas from 314.2 m2/ g at 550 °C to 198.5 m2/g at 950 °C. For comparison, the surface area, pore volume, and pore size of Halo 2.7 lm particles are 124 m2/g, 0.288 cc/g, and 9.5 nm, respectively [14]. At 1050 °C, viscous flow of the porous nanoparticles may become possible, which leads to the pore collapse of the porous shell, accompanied by a significant shrinkage of core–shell particles from 2.65 lm at 950 °C to 2.37 lm at 1050 °C. As such, the TPV value drops substantially to 0.059 cc/g, which is the main reason that the
Fig. 4. (A) Nitrogen sorption isotherms and (B) corresponding pore size distribution of CSPs after heat treatment at different temperature. (C) is the normalized pore size distribution data from panel (B).
54
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Table 1 Nitrogen sorption data for CSPs calcined at different temperatures. Temperature (°C)
SSA (m2/g)
TPV (cc/g)
APS (nm)
550 650 750 850 950 1050
314.2 ± 12.5 295.5 ± 6.4 273.2 ± 6.5 258.7 ± 5.2 198.5 ± 4.9 32.0 ± 2.7
0.63 ± 0.08 0.56 ± 0.07 0.53 ± 0.06 0.51 ± 0.06 0.42 ± 0.05 0.06 ± 0.02
8.15 ± 0.10 7.80 ± 0.09 7.94 ± 0.09 8.02 ± 0.10 8.12 ± 0.10 7.08 ± 0.09
SSA – specific surface area; TPV – total pore volume; APS – average pore size.
surface area at 1050 °C (31.95 m2/g) is more than 10 times less than that at 550 °C (314.2 m2/g). The upper limit temperature for this particular sample without pore collapse was 1000 °C, which is similar to colloidal gels prepared by other methods [19,20]. Finally, we point out that not only did the average pore sizes at different temperatures up to 950 °C remain similar but so to did their pore size distributions, as shown in Fig. 4C after normalizing their data to those obtained at 550 °C. These results indicate that pores of different sizes contract at a similar rate, even though the smaller pores have higher surface energy. This surprising finding seems to be in contrast to the intuition that small pores disappear before large ones due to the higher surface energy of the former. The reasons for our finding remain unknown. It is likely that pore disappearance and/or collapse requires the long migration of silica materials such as viscous flow or Ostwald ripening, which may not occur during high temperature treatment up to 1000 °C. In summary, high temperature calcination leads to CSPs with strong mechanical strength (See Section 3.4) and very similar particle geometry as compared to commercial products. However, further treatments are required to increase their pore size (10 nm) and the surface Si–OH concentration in order for them to be used as HPLC columns [21]. In the following, we discuss the effects of different treatments on pore size and Si-OH content of CSPs after calcination at 750 °C and 950 °C. 3.2. Rehydroxylation CSPs after high temperature calcination lose a significant amount of Si-OH groups due to condensation reactions. The purpose of rehydroxylation for chromatographic applications is to restore these lost Si–OH groups, allowing them to be silanized at a high density [21]. Fig. 2 shows the TGA analysis of four CSP samples: two samples calcinated at 750 °C and 950 °C, respectively, and their counterparts after rehydroxylation. The weight loss and Si–OH group concentration aOH of those samples are listed in Table 2. aOH, expressed as the number of Si–OH group per nm2, was estimated from the weight loss (WL) in Fig. 2 between 200 and 1000 °C using the following formula:
aOH ¼ ððWL 100Þ=18ÞÞNA 1018 S1
ð1Þ
which is modified from the literature [22]. In the above formula, the first term (WL 100)/18 is the concentration of OH groups on the CSP surface in the unit of mol of Si–OH group/1 g of CSPs, NA is the Avogadro number, and S is the specific surface area of CSPs. Table 2 Comparison of the weight loss and Si–OH surface concentration before and after rehydroxylation. Sample
Surface area (m2/g)
Weight loss (%)
Si–OH concentration (nm2)
750 °C 750 °C-RH 950 °C 950 °C-RH
273.2 ± 6.5 266.5 ± 5.4 198.5 ± 4.9 195.7 ± 4.7
1.4 3.5 0.9 2.1
1.72 4.40 1.52 3.60
From Table 2, one can find that the weight loss of CSP samples from 200 to 1000 °C was in the order of 750 °C-RH (3.5%) > 950 °CRH (2.1%) > 750 °C (1.4%) > 950 °C (0.9%). Their corresponding aOH values were 4.40, 3.60, 1.72, and 1.52. While the aOH values are reasonable compared to the literature information summarized by Zhuravelev [16], our data may not be accurate owing to the method by which we obtained the weight loss data from TGA measurements. In the low temperature region (e.g., <250 °C), the weight loss of Si–OH concentration could be overestimated due to the desorption of physically adsorbed water. On the other hand, Si–OH groups are still clearly present at 1050 °C (see Fig. 3) and condensation reactions continue up to 1200 °C, which will lead to the underestimation of the Si–OH concentration on the silica surface, partially compensating for this issue. Future work will use more accurate tools such as the deuterium-exchange method combined with mass spectrometric analysis to determine the aOH values [22]. Nevertheless, our data show clear and expected trends in terms of the effects of calcination and rehydroxylation. Calcination at high temperature led to a decrease in aOH due to continuous condensation reactions. The subsequent rehydroxylation significantly increased the surface Si–OH coverage even if the sample was calcined at 950 °C. However, higher calcination temperatures resulted in a lower degree of rehydroxylation due to the high hydrophobicity of the surface and high activation energy for the breakdown of Si-O-Si bonds. These trends are further supported by the IR spectra shown in Fig. 5, in which the same calcined samples before and after rehydroxylation are compared. While it is difficult to analyze the data quantitatively, the IR results clearly indicate that rehydroxylation led to an increase of surface Si–OH concentrations (750 °C-RH > 750 °C and 950 °C-RH > 950 °C) and was less complete after high temperature calcination (750 °C-RH > 950 °C-RH). 3.3. Pore size enlargement under basic conditions As stated previously, the pore size of CSPs is 8 nm, which remains stable up to 1000 °C. This value is at the low end for chromatographic column applications. There are a number of chemicals (NH4F, HF, NaOH, KOH, etc.) and processes (hydrothermal treatment, aging in the mother liquor) that can be used to tailor the pore size of silica materials. The basic principle of these post synthesis treatment methods is the well-known Ostwald ripening process: dissolution of silica regions with small positive curvature and redeposition of the dissolved silicate species on the surfaces of silica regions with small negative curvature [23]. We attempted to avoid the use of corrosive chemicals and autoclaving to alter the mesopore sizes of CSPs in order to make eventual manufacturing more facile and environmentally friendly. Our method to enlarge the pore size of CSPs was, therefore, carried out at room temperature in aqueous solution by adding NH4OH to adjust the pH to approximately 11 for 6 h. We refer these samples as 750 °C-pH and 950 °C-pH based on their calcination temperature. Fig. 6 shows the isotherms of CSP samples 750 °C-pH and 950 °C-pH and their corresponding desorption pore size distributions (PSDs). Fig. 6A shows that CSPs after aging in diluted NH4OH solution had typical type IV isotherms with a H2 hysterisis loop associated with capillary condensation in the mesopore structures, similar to those shown in Fig. 4A before high pH treatment. However, the whole PSD curves were shifted to the large pore size region compared to those obtained before high pH treatment, while the small mesopores (<5 nm) were almost completely absent. The PSD curves of these two samples showed a monomodal distribution with peak positions centered at 17 nm (750 °C-pH) and 12 nm (950 °C-pH). From Fig. 6B, one can observe that our materials displayed narrower PSDs (much higher peak position > 1.2 cc/g) compared to commercial products [14] and
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Fig. 5. IR spectra of CSPs before and after rehydroxylation after calcining at 750 °C and 950 °C, respectively.
had essentially no pores with diameters less than 4 nm. These properties may be important to reduce the solute mass transfer resistance and increase column separation efficiency in eventual HPLC applications. The specific surface area, total pore volume, and average pore size of these samples were 199.6 m2/g, 0.529 cc/g, and 11.79 nm for the 750 °C-pH sample and 185.0 m2/g, 0.453 cc/g, and 9.80 nm, for the 950 °C-pH sample, respectively. These results indicate that the post treatment (pH 11, room temperature, and reaction time of 6 h) was very effective for producing CSPs with a 10 nm pore size with a high surface area (190 m2/g). As stated previously, the thermodynamic basis for these changes is Ostwald ripening. In our particular case, small silica nanoparticles and aggregated clusters, and especially those with small positive curvature that form the small intraparticle pores that have high free energies, were dissolved first at high pH. The dissolved silicate species then redeposited onto large particles (or clusters) of lower solubility, thus eliminating the small pores and increasing the average pore sizes but reducing the surface areas of CSPs. Similar to rehydroxylation, high pH treatment also involves the breakdown of Si–O–Si bonds, which become more stable with increasing temperature. As such, the porosity of the 750 °C-pH sample showed less change compared to the 950 °C-pH sample.
3.4. Mechanical strength of CSPs For chromatographic applications, the mechanical strength of the particles is critically important because HPLC column packing is carried out at very high pressure (e.g., 10,000 psi) and with the
55
advent of ultrahigh pressure LC, experiments may also be done at pressures approaching 10,000 psi. However, the porous shells prepared by the layer-by-layer method are assembled by loosely packed silica nanoparticles held together by weak electrostatic attractions and hydrogen bonding. These noncovalent forces are unlikely to impart sufficient mechanical stability to produce robust shells, as demonstrated by others [24,25]. To enhance the mechanical strength of porous silica films, low temperature methods such as hydrothermal treatment [24] and atomic layer deposition [26] have recently been developed. The advantage of these methods is that they are compatible with organic substrates, however it is highly unlikely that silica shells formed by these methods, which simply increase the neck thickness of connected nanoparticles, are robust enough to endure the packing pressure of HPLC columns. In this work, we explored high temperature treatment as a method to enhance the mechanical strength of CSPs. This is the most widely used method to increase the robustness of porous silica materials prepared by numerous approaches [20,27]. In order to test the mechanical strength of our materials, we prepared KBr pellets similar to those used for recording IR spectra, using different final pressures to form the pellets. Fig. 7 shows SEM images of the 5 lm totally porous particles and CSPs after being subjected to different compressive pressures. The 5 lm totally porous silica particles are commercial products and are used as a reference material to evaluate the mechanical properties of our materials. CSPs are named based on how these materials are post treated and tested. For example, 750 °C-pH-2500 psi represents CSPs that were calcinated at 750 °C followed by pore size enlargement at high pH and compression at 2500 psi to produce the KBr pellet. From Fig. 7, one can produce semi-quantitative data on mechanical stability by determining the percentage of particles that remained intact after compression. We note that 90% of 5 lm particles remained intact at 500 psi, while only 10% were intact after treatment at 2500 psi. CSPs treated at 750 °C remained largely intact (>98%) at 500 or 1000 psi, but only 40% of these particles were intact after compression at 2500 psi. Finally, >99% of particles treated at 950 °C remained intact at 2500 psi. These results indicate that CSPs after high temperature treatment are much more robust than 5 lm porous particles, which can withstand at least 6000 psi pressure during slurry packing. The main reason that porous particles show a much higher compressive modulus during slurry packing is likely due to the fact that the liquid that fills the pores provides extra resistance under compressive loading. These results also indicate that the mechanical strength of CSPs increases with increasing calcination temperature as explained below. First, it is well-known that the mechanical properties of
Fig. 6. (A) Nitrogen sorption isotherms and (B) corresponding pore size distribution of CSPs treated at different temperatures and then aged under NH4OH solutions at pH = 11.
56
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57
Fig. 7. SEM images of 5 lm totally porous particles and CSPs treated at different temperatures and after high pH pore enlargement that are subjected to high compressive pressure.
porous silica materials (e.g., elastic and compressive modulus) are dependent on their density (q) and have a power-law relationship with the density (E / qa), where a (ranging from 3 to 4) is the scaling exponent [28,29]. Increasing the density of CSPs with increasing temperature is obviously due to shrinkage (Fig. 1) and reduced pore volume of the particles (Table 1). Second, we expect the presence of string-of-bead (individual nanoparticle) like structures in the as-prepared particles due to the weak interactions involved in the mL-b-mL method. At elevated temperature, especially at 950 °C, the individual nanoparticles are expected to lose their identity by fusing together and also onto the core particles. This change is the result of continuing condensation of Si–OH groups (see Fig. 2), structural relaxation, and partial sintering, evidenced by an accompanying decrease in particle surface area and pore volume. The thus formed thickened necks stabilize the solid network and increase network connectivity, therefore, enhancing the mechanical strength of CSPs. Finally, condensation reactions themselves, similar to aging in the sol–gel process, which increase the degree of condensation, are also expected to play a positive role in strengthening porous silica materials. Therefore, the enhanced mechanical strength of CSPs with increased calcination temperature can be attributed to the reduced porosity, the high degree of condensation, the formation of thick necks between nanoparticles and the fusion of nanoparticles onto the core particles.
a very narrow pore size distribution and no pores less than 4 nm in diameter. High temperature calcination also drastically increased the mechanical strength of CSPs. CSPs calcinated at 950 °C and after high pH treatment showed much higher mechanical strength than 5 lm totally porous commercial silica particles. In addition, the above CSPs possessed the following properties: particle diameter 2.65 lm, shell thickness 0.5 lm, surface area 185 m2/g, pore diameter 9.80 nm, pore volume 0.453 cc/g, and Si–OH group surface concentration 3.6 OH/nm2, which are in line with the values of commercial products. The performance of these materials as HPLC stationary phases is currently being evaluated and the results will be reported in due course.
4. Conclusions
References
In this work, the previously developed ML-b-ML process was used to prepare sub-3 lm silica core–shell particles with a focus on altering their properties to make them comparable to commercial products. To this end, we performed various post-synthesis treatments including high temperature calcination, high pH pore enlargement, and rehydroxylation. We found that surface areas and pore volumes decreased steadily while pore sizes and their distributions remained similar with increases in temperature up to 1000 °C. These results indicate that the porous shells prepared by the ML-b-ML process have similar thermal stability to colloidal gels and the small pores are as stable as the large pores. Rehydroxylation (100 °C in water for 24 h) and pore size enlargement (in NH4OH, pH = 11 for 6 h) involve the breakdown of Si–O–Si bonds, which becomes increasingly difficult with increases in calcining temperature. Nevertheless, the high pH treatment effectively led to increases in pore size from 8 nm to 10 nm with
Acknowledgments The authors thank the Natural Sciences and Engineering Research Council of Canada for funding this work through a network grant-SENTINEL Canadian Network for the Development and Use of Bioactive Paper. We also thank Aglient Corporation for providing funding through an Agilent University Research Fund grant. The authors also thank the Canada Foundation for Innovation and the Ontario Innovation Trust for providing funding to the Biointerfaces Institute. JDB holds the Canada Research Chair in Bioanalytical Chemistry and Biointerfaces.
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
H.J. Dong, J.D. Brennan, J. Mater. Chem. 22 (2012) 13197. S. Fekete, E. Olah, J. Fekete, J. Chromatogr. A 1228 (2012) 57. H. Nishi, K. Nagamatsu, Anal. Sci. 30 (2014) 205. K.K. Unger, R. Skudas, M.M. Schulte, J. Chromatogr. A 1184 (2008) 393. J.J. Kirkland, J. Chromatgr. Sci. (1969) 361. F. Gritti, I. Leonardis, D. Shock, P. Stevenson, A. Shalliker, G. Guiochon, J. Chromatogr. A 1217 (2010) 1589. G. Guiochon, F. Gritti, J. Chromatgr. A 1218 (2011) 1915. J.J.K.a.T.J. Langlois, US. Patent 0189944 A1 (2007). H.J. Dong, J.D. Brennan, Chem. Commun. 47 (2011) 1207. W. Stöber, A. Fink, E. Bohn, J. Colloid Interface Sci. 26 (1968) 62. H.G. Klaus Unger, Joachim Kinkel, US. Patent 4777520. W.C.T.-C. Wei, US. Patent 7846337 B2. E.P. Barrett, L. Joyner, P.P.J. Halenda, J. Am. Chem. Soc. 73 (1951) 373. F. Gritti, I. Leonardis, J. Abia, G. Guiochon, J. Chromatogr. A 1217 (2010) 3819. J.J. Kirkland, T.J. Langlois, J.J. DeStefano, Am. Lab. 39 (2007) 18. L.T. Zhuravlev, Colloids Surfaces A – Physicochem. Eng. Asp. 173 (2000) 1. D.L. Wood, E.M. Rabinovich, D.W. Johnson, J.B. Macchesney, E.M. Vogel, J. Am. Ceram. Soc. 66 (1983) 693. T.A. Guiton, C.G. Pantano, Colloids Surfaces A – Physicochem. Eng. Asp. 74 (1993) 33.
H. Dong, J.D. Brennan / Journal of Colloid and Interface Science 437 (2015) 50–57 [19] T. Yamada, M. Nakajima, T. Suemoto, T. Uchino, J. Phys. Chem. C 111 (2007) 12973. [20] D.W. Johnson, E.M. Rabinovich, J.B. Macchesney, E.M. Vogel, J. Am. Ceram. Soc. 66 (1983) 688. [21] J. Kohler, J.J. Kirkland, J. Chromatogr. 385 (1987) 125. [22] L.T. Zhuravlev, Langmuir 3 (1987) 316. [23] R.K. Iler, The Chemistry of Silica: Solubility, Polymerization, Colloid and Surface Properties and Biochemistry, Wiley, New York, 1979.
57
[24] Z. Gemici, H. Shimomura, R.E. Cohen, M.F. Rubner, Langmuir 24 (2008) 2168. [25] L.E. Blue, J.W. Jorgenson, J. Chromatogr. A 1218 (2011) 7989. [26] M.I. Dafinone, G. Feng, T. Brugarolas, K.E. Tettey, D. Lee, ACS Nano 5 (2011) 5078. [27] S. Satoh, K. Susa, I. Matsuyama, T. Suganuma, J. Am. Ceram. Soc. 68 (1985) 399. [28] T. Woignier, J. Phalippou, R. Vacher, J. Mater. Res. 4 (1989) 688. [29] A. Jain, S. Rogojevic, S. Ponoth, N. Agarwal, I. Matthew, W.N. Gill, P. Persans, M. Tomozawa, J.L. Plawsky, E. Simonyi, Thin Solid Films 398 (2001) 513.