Microporous and Mesoporous Materials 2004 in press
Well-Ordered Mesoporous Silica Prepared by Cationic Fluorinated
Stephen E. Rankin*§, Bing Tan§, Hans-Joachim Lehmler†, Kimberly P. Hindman§ and Barbara L.
§ Chemical and Materials Engineering Department, University of Kentucky,
Lexington, KY 40506-0046, U.S.A.
† Department of Occupational and Environmental Health, University of Iowa, College of Public
Health, Iowa City, IA 52242-5000 U.S.A.
* To whom correspondence should be addressed. Phone: +1-859-257-9799. E-mail:
We describe the synthesis and characterization of ordered mesoporous silica using the
template 1H,1H,2H,2H-perfluorooctylpyridinium chloride. This surfactant forms several
lyotropic mesophases at room temperature, including hexagonal close packed cylinders, an
isotropic liquid crystal, and a lamellar phase. Using this surfactant, mesoporous silica is
synthesized by room temperature precipitation and surfactant extraction. Both the surfactant and
product material are thoroughly characterized. From nitrogen adsorption, the product has a
specific surface area of 982 m2/g and a pore diameter of 2.6 nm. X-ray diffraction and
transmission electron microscopy confirm 2-dimensional close-packed hexagonal long-range
ordering. This is the first example of cationic fluorinated surfactant templating of an ordered
porous ceramic material.
Keywords: Surfactant templating, silica, mesoporous, fluorinated
S. Rankin et al. 2 of 17
Since their discovery, researchers have tried to control the size, structure and orientation
of the pores in surfactant-templated silica[2,3]. These materials are formed by co-assembly of
surfactants and silica precursors, followed by silica network formation and surfactant removal by
calcination or extraction. All major classes of hydrocarbon surfactants (cationic, nonionic and
anionic) including single- and double- tailed surfactants have been used as templates. It
has been shown that the pore size decreases as the length of the alkyl chain decreases for
alkyltrimethylammonium halide templates, but that the ability of these surfactants to form well-
ordered assemblies with silica diminishes for chains shorter than decyl. Octyl chains give
uniform pores without long-range order, unless double-tailed surfactants are added to promote
micelle ordering. This limitation is significant, because it leaves unspanned the gap in pore
size between zeolites (maximum pore size ~1.4 nm for aluminophosphates) and ordered
surfactant-templated silica (pore sizes > 2.7 nm unless double-tailed surfactants are used).
However, fluorinated surfactants are a qualitatively different class of surfactants that have
not yet been explored extensively. There has to date been one report of using a fluorinated
surfactant in combination with a nonionic block copolymer as a template for mesoporous
silica, but the pore size does not seem to be related to the fluorosurfactant used. In that report,
the precise role of the fluorinated surfactant in the pore templating process is not entirely clear.
In fact, Han et al. suggest that “…fluorocarbon surfactants are not suitable as templates for the
preparation of well-ordered materials.” However, fluorinated surfactants form aggregates
more easily and with lower curvature than their hydrocarbon analogues[10,11]. This suggests
that they may be suitable as porogens, if the right surfactant and synthesis conditions were
chosen. The strong carbon-fluorine bond also causes them to have strikingly different molecular
S. Rankin et al. 3 of 17
interactions than hydrocarbon surfactants. While this paper was under review, Blin et al.
published the first example of mesoporous silica synthesized with only a fluorinated surfactant
template, in their case a nonionic surfactant. In this brief article, we will show that cationic
fluorinated surfactants also can be used as templates for ordered porous ceramics and that their
ease of assembly allows small-pore ordered structures to be formed from a single-tailed
The surfactant, 1H,1H,2H,2H-perfluorooctylpyridinium chloride (FSPCl-1) was synthesized
as described previously, by alkylation of anhydrous pyridine with 1H,1H,2H,2H-perfluorooctyl
iodide. The resulting pyridinium iodide was carefully washed with anhydrous diethyl ether
and recrystallized from acetone. The corresponding chloride was obtained by ion exchange
chromatography on Amberlite IRA-410 and used without further purification. Melting points
were determined using a Thermal Analysis 2920 Differential Scanning Instrument.
Thermograms were recorded using a heating rate of 20°/minute from 20 to 230°C and the onset
temperatures of the main transitions were determined. Mass spectra were recorded at the
University of Kentucky Mass Spectrometry Facility using a Kratos Kompact SEQ or Bruker
Autoflex time-of-flight mass Spectrometer equipped with MALDI. The formula weight of the
pyridinium cation of both the iodide and chloride surfactants was 426 g/mol.
Materials were prepared by first adding 0.750 g aqueous ammonium hydroxide (~28 wt%)
and 0.400 g FSPCl-1 to 13.9 g of deionized ultrafiltered water. After mixing for one hour, 1.08 g
of tetraethoxysilane (TEOS) was slowly added with continuous stirring. Within two minutes
after the addition of TEOS to the surfactant solution, a white powder precipitated. The solution
was aged at room temperature for 24 hr and then filtered. The as-synthesized material was dried
S. Rankin et al. 4 of 17
in air for 48 hr. The surfactant was extracted from the dried material by twice stirring with a
mixture of 150 ml ethanol (95 wt% with water) and 5 ml aqueous HCl (~36 wt%) for 24 hr,
followed by filtration. The shorthand notation for this sample will be UK-1.
Samples for FTIR analysis were ground finely, mixed with KBr at a concentration of 1 wt%
of the sample, and pressed into translucent pellets. Spectra were collected with a ThermoNicolet
Nexus 470 spectrometer. Nitrogen sorption isotherms of the samples after surfactant extraction
and drying were collected with a Micromeritics Tristar 3000 automated adsorption instrument.
The samples were outgassed for 4 hours at 150 ºC. X-ray powder diffraction (XRD) patterns
were recorded with a Siemens 5000 diffractometer using 1.54098 nm CuKα radiation. Powders
were mixed with a small amount of acetone and pressed into a holder. For transmission electron
microscopy (TEM), samples were ground, suspended in acetone and deposited on lacey carbon
TEM grids for analysis with a JEOL 2010F instrument.
Results and Discussion
The surfactants were thoroughly characterized for purity and composition. The results of the
DSC and NMR investigations of the surfactants are as follows:
1H,1H,2H,2H-perfluorooctyl pyridinium iodide:
m.p. = 110, 130, 170 and 186°C. 1H-NMR (DMSO, 400 MHz) δ 3.14-3.32 (m, -CH2CF2-,
2H), 5.02 (t, -CH2CH2CF2, J = 7.2 Hz), 8.23 (“t”, J ~ 7.0 Hz, 2H), 8.68 (t, J = 8.0 Hz, t, J = 1.2
Hz, 1H), 9.23 (“d”, J ~ 5.2 Hz, 2H) ppm. 13C-NMR (DMSO, 100 MHz) δ 30.55 (t, -CH2CF2-,
JCF = 20 Hz), 52.83(t, -CH2CH2CF2), 128.07, 145.52, 146.34 ppm. 19F-NMR (D3COD, 376
MHz) δ -81.63 (-CF3), -113.67, -122.10, -123.12, -123.58, 126.57 ppm.
1H,1H,2H,2H-perfluorooctyl pyridinium chloride (FSPCl-1)
S. Rankin et al. 5 of 17
m.p. = 178°C. 1H-NMR (DMSO, 400 MHz) δ 3.14-3.32 (m, -CH2CF2-, 2H), 5.02 (t, -
CH2CH2CF2, J = 7.2 Hz), 8.20 (“t”, J ~ 7.0 Hz, 2H), 8.69 (t, J = 8.0 Hz, t, J = 1.2 Hz, 1H), 9.18
(“d”, J ~ 5.2 Hz, 2H) ppm. 13C-NMR (DMSO, 100 MHz) δ 30.55 (t, -CH2CF2-, JCF = 20 Hz),
52.83(t, -CH2CH2CF2), 128.07, 145.52, 146.34 ppm. 19F-NMR (D3COD, 376 MHz) δ -81.65 (-
CF3), -113.73, -122.13, -123.14, -123.64, 126.60 ppm.
Before attempting to synthesize materials, we used polarization contrast optical microscopy
to rapidly scan the phase behavior of FSPCl-1. The phase diagram of 1H,1H,2H,2H-
perfluorodecylpyridinium chloride was recently reported, and hexagonal close packed cylindrical
phases were found to be less favorable near room temperature than “intermediate” phases such as
rectangular and mesh phases at room temperature. Therefore, we could not be sure a priori
whether our surfactant would be able to form conventional pore structures such as hexagonal
close packed cylinders. Figure 1a shows the results of a flooding experiment, in which a solid
crystal of surfactant was surrounded with deionized ultrafiltered water. The solid material was
initially near the top center of the figure, so the concentration decreases moving away from the
point. The micrograph was collected very shortly after the water was introduced. Moving
outward, transitions are seen from a lamellar structure (near the center) to an isotropic liquid
crystal phase (the black band) to a hexagonal close packed phase, to an isotropic micellar phase.
Samples were also prepared of well defined composition and allowed to equilibrate at room
temperature (22 ± 1 °C) for 72 hours. Figure 1b shows that a 50 wt% solution in water has a
hexagonal close packed (HCP) structure, as indicated by the fanlike texture. The 60 wt%
solution also has a HCP structure (not shown). Figure 1c shows that a 70 wt% solution of
FSPCl-1 in water has a lamellar structure, indicated by the “streaky” appearance of the
birefringence texture. This rapid screening study of the phase behavior of the surfactant suggests
S. Rankin et al. 6 of 17
that it may be possible to prepare a material with HCP, isotropic liquid crystal, or lamellar
With the confirmation that our surfactant is able to form ordered mesophases at room
temperature, we discuss the synthesis. Infrared spectroscopy (Figure 2) confirms that we
incorporate FSPCl-1 into the silica particles, and that our extraction procedure completely
removes the surfactant. The solid surfactant spectrum (Fig. 2a) shows several bands in the range
(1400 cm-1 to 700 cm-1) associated with fluorinated alkyl groups.[16,17] Of these bands, the
most prominent ones have been assigned to asymmetric CF2 stretching (1247 cm-1), CCC
deformation (1207 cm-1), and symmetric CF2 stretching (1147 cm-1).[16,17] In the as-
synthesized material (Fig. 2b) all of these bands may be present, but the 1147 cm-1 band is most
clearly defined (indicated with an arrow). In the extracted product (Fig. 2c), the 1147 cm-1 band
cannot be observed, but there still is a shoulder at about 1200 cm-1. This shoulder is typical of
high-surface area silica[18,19] and, like the more prominent band at 1093 cm-1, is assigned to
asymmetric Si-O-Si stretching. The spectrum of fumed silica (>99%, Aldrich) shown in Fig.
2d has the same features as the extracted material, and confirms the assignment of the shoulder at
1200 cm-1 to silica. A clearer indication of the presence of the surfactant in the as-synthesized
product (Fig. 2b) is the narrow band at 1495 cm-1 that does not coincide with any other bands.
This band has not been assigned to fluorocarbons and is more likely to be associated with
pyridinium motion. After 2 extraction steps, this band is completely absent (Fig. 2c).
The nitrogen adsorption isotherm after surfactant extraction is shown in Figure 3. This type
IV isotherm has an inflection at p/p0 = 0.116, suggesting that capillary condensation in uniform
pores is occurring. To learn more about the pore structure, we prepared and analyzed an αs-plot
S. Rankin et al. 7 of 17
according to the procedure of Sayari et al.. Data for the macroporous reference material,
LiChrospher Si-1000, are taken from Jaroniec et al..
Analysis of the αs plot determines features of the structure of these materials without
reference to a particular adsorption model (other than the BET method used to determine the
surface area of the reference material, 22.1 m2/g). We observe the same features as have
been observed for other mesoporous silicates. A line fit through the data at low αs passes
through the origin, indicating that the product does not contain micropores. The slope of this line
gives the total specific surface area of the sample, St. After capillary condensation ends,
adsorption continues outside of the mesopores, and the slope of a second line at high αs gives the
external specific surface area, Sext. These results, and the BET surface area, are reported in
Table 1. Extrapolating the external adsorption (high-αs) line to αs =0 gives the primary pore
Availability of well-defined porous silica has recently led to improved methods for
determining the size of cylindrical silica pores by nitrogen adsorption. A modified Kelvin
equation approach (the KJS method) has been shown to give accurate estimates of pore size for
MCM-41 materials. In the inset of Figure 3, we show the pore size distribution calculated
using the KJS method. The peak pore diameter, dKJS, is given in Table 1. Consistent results are
found using non-local density functional theory values, based on the position of the
inflection in the isotherm (dNLDFT), or from a simple geometric calculation based on the
assumption of cylindrical pores.
Values calculated by the KJS method, NLDFT, and geometric methods are all consistent
with a pore diameter of 2.6 nm. This disagrees with the prediction of the often-used BJH
method, 1.9 nm. The BJH method is known to underpredict pore diameters near the
S. Rankin et al. 8 of 17
micropore-mesopore boundary so other estimates can be considered to be more reliable. A
similar pore diameter was recently reported when a large amount of the cosurfactant butylamine
was added to an MCM-41 synthesis solution, but the long-range ordering of that material was
limited (as indicated by only two diffraction peaks).
To demonstrate long-range order in sample UK-1, XRD results are presented in Figure 4 for
the as-prepared and extracted samples. The as-prepared sample (dashed line) shows poorly
resolved peaks, probably because of poor contrast between silicon and the fluorocarbon
surfactant. The most clearly resolved peak has a d-spacing of 2.8 nm. The extracted sample
(solid line) clearly shows evidence for three diffraction peaks which are indexed according to a
2D hexagonal close packed (HCP) structure. Contrasting this, MCM-41 prepared using
octyltrimethylammonium bromide has a uniform pore size, but no evidence from higher-order
diffraction peaks of well-defined order .
Extraction of the surfactant from UK-1 does not change the d-spacing of the (100) peak
significantly (it is 2.83 nm). The unit cell size for the 2D HCP structure is 3.28 ± 0.01 nm.
Based on the pore size estimated above, this gives an approximate wall thickness of 0.7 nm. The
diffraction peaks are relatively broad, and from the Scherrer equation, we estimate a scattering
domain size of approximately 13 nm.
TEM confirms the presence of cylindrical pores with 2D hexagonal symmetry. Figure 5
shows two representative images of UK-1 after surfactant extraction. Figure 5a shows a direct
image of the 2D HCP structure. A Fourier transform of this region (inset) shows six spots
characteristic of a 2D HCP structure. Figure 5b shows a region with layers of light and dark
stripes. The Fourier transform of this region (inset) shows two spots characteristic of a lamellar
structure. However, the image is also consistent with a 2D HCP cylindrical structure viewed
S. Rankin et al. 9 of 17
from the edge of the cylinders. The spacing between layers is consistent with the 2D HCP (100)
d-spacing measured by XRD. There is no indication from XRD or from TEM sampling that
lamellar regions are actually present in our sample. Because there is no orientation preference in
these particles, other parts of the image appear to be disordered, but that could be due to
superposition of ordered regions with different orientations. The observation of small oriented
domains in the TEM images is consistent with the broad XRD peaks observed in Figure 4.
Using a partially fluorinated octylpyridinium chloride surfactant, we synthesized silica
particles with 2.6 nm, 2D hexagonal close packed cylindrical pores. The pore size was
confirmed by several theoretical interpretations of the nitrogen adsorption data. The long-range
2D hexagonal ordering was confirmed by XRD and TEM. This is the first time that a cationic
fluorinated surfactant alone has been used as the template for ordered mesoporous silica. This
also is the first time that a single-tailed surfactant has been used to generate a porous material
with long-range order from a cationic octyl surfactant. This finding suggests that fluorinated
surfactants may have potential to further reduce the available pore sizes in surfactant templated
porous ceramic materials. We are currently investigating in detail the effects of fluorinated
surfactant architecture and synthesis conditions on the pore structure of these materials.
We thank Prof. Elizabeth Dickey and Ms. Colleen Frazer of Pennsylvania State University
for helpful discussions of the characterization. We thank Dr. Alan Dozier of the University of
Kentucky Materials Characterization Facility for his help with the TEM. We thank Prof.
Kimberly Anderson of the University of Kentucky and her group for helping with the use of the
optical microscope. This material in based upon work supported by the National Science
S. Rankin et al. 10 of 17
Foundation under Grant No. DMR-0210517. BT was supported by a Kentucky Research
Challenge Trust Fund graduate fellowship.
C.T. Kresge et al., Nature 359 (1992) 710.
U. Ciesla, F. Schüth, Microporous and Mesoporous Mater. 27 (1999) 131.
J.Y. Ying et al., Angew. Chem. Int. Ed. 38 (1999) 56.
Q.S. Huo et al., Nature 368 (1994) 317.
Q. Huo et al., Chem. Mater. 8 (1996) 1147.
R. Ryoo et al., J. Am. Chem. Soc. 123 (2001) 1650.
M.E. Davis, Chem. Eur. J. 3 (1997) 1745.
Y. Han et al., Angew. Chem. Int. Ed. 42 (2003) 3633.
ibid, p. 3633
M.P. Krafft, J.G. Riess, Biochim. 80 (1998) 489.
M. Monduzzi, Current Opinion in Colloid & Interface Science 3 (1998) 467.
J.L. Blin et al., Langmuir 20 (2004) 491.
T. Asakawa et al., Langmuir 11 (1995) 478.
H.J. Lehmler et al., Journal of Fluorine Chemistry 107 (2001) 141.
K. Wang et al., Langmuir 16 (2000) 1042.
S.L. Hsu et al., Macromolecules 23 (1990) 4565.
Y. Ren et al., J. Chem. Phys. 113 (2000) 1162.
S.M. Holmes et al., J. Chem. Soc., Faraday Trans. 94 (1998) 2025.
D.C. Calabro et al., Microporous Materials 7 (1996) 243.
E.I. Kamitsos et al., Phys. Rev. B 48 (1993) 12499.
A. Sayari et al., Chemistry of Materials 9 (1997) 2499.
M. Jaroniec et al., Langmuir 15 (1999) 5410.
S. Brunauer et al., J. Am. Chem. Soc. 60 (1938) 309.
M. Kruk et al., Langmuir 13 (1997) 6267.
P.I. Ravikovitch et al., Langmuir 11 (1995) 4765.
E.P. Barrett et al., J. Am. Chem. Soc. 73 (1951) 373.
A.V. Neimark et al., Journal of Colloid and Interface Science 207 (1998) 159.
F. Kleitz et al., Langmuir 18 (2002) 4963.
J. Choma et al., Coll. Surf. A: Phys. Eng. Aspects 203 (2002) 97.
S. Rankin et al. 11 of 17
Table 1. Nitrogen adsorption analysis results for sample UK-1
St Sext SBETa Vp dKJS dNLDFT 4Vp/(St-Sex) dBJH
(m2/g) (m2/g) (m2/g) (cm3/g) (nm) (nm) (nm) (nm)
982. 39.4 1134 0.608 2.57 2.60 2.58 1.90
a Calculated using data in range of p/p0 from 0.04 to 0.1.
S. Rankin et al. 12 of 17
Figure 1. Polarization contrast micrographs of FSPCl-1 solution in water at room temperature.
(a) Low-magnification view immediately after flooding a single particle of solid surfactant with
deionized water, (b) 200x magnification view of equilibrated sample of 50 wt% surfactant, and
(c) 200x view of equilibrated sample of 70 wt% surfactant.
Figure 2. FTIR spectra of KBr pellets pressed with 1 wt% of (a) the pure surfactant, FSPCl-1,
(b) sample UK-1 as synthesized, (c) sample UK-1 after extraction, and (d) fumed silica. The
arrows indicate surfactant bands in the as-synthezied product that are absent after extraction.
Figure 3. Adsorption isotherm for sample UK-1. Filled circles are adsorption points and open
circles are desorption points. The inset shows the pore size distribution of sample UK-1 after
extraction calculated by the KJS method .
Figure 4. XRD results for sample UK-1 as-prepared (lower) or after extraction (upper).
Figure 5. Representative transmission electron micrographs of UK-1 after surfactant extraction.
The views illustrate (a) a 2D HCP region and (b) a striped region consistent with the 2D HCP
structure viewed from the edge of the cylindrical pores. The insets show Fourier transforms of
S. Rankin et al. 13 of 17
H He ex xa ag go on na al l
L La am me el ll la ar r
S. Rankin et al. 14 of 17
80010001200 1400 1600
S. Rankin et al. 15 of 17 Download full-text
0 0.20.4 0.60.81
V (cm3/g STP)