Formation of silica precipitates on membrane surfaces in two and three dimensions

Article abstract of DOI:10.1021/la1021627

Ether lipids with alkyl chains of uniform length and varying amine headgroups were synthesized and assembled into bilayer structures in aqueous solution, which served as templates for the formation of silica in two and three dimensions produced under ambient conditions. Dynamic light scattering revealed that unilamellar vesicles of the aminolipids are formed by the extrusion method. The alkylation of the polar amine headgroup was systematically increased from a primary, secondary, and tertiary amine to a quaternary ammonium salt, and the amount of silica was quantified by the -silicomolybdate method as a function of the headgroup. A lysinol-connected ether lipid harboring two primary amine groups was also investigated. This variation enabled us to compare the influence of the headgroup on the properties of the precipitated silica in detail. By spreading of unilamellar aminolipid vesicles onto planar silicon substrates, two-dimensional planar bilayers can be produced. By means of ellipsometry in conjunction with atomic force microscopy, we were able to demonstrate that very thin silica layers with a thickness of a few nanometers are formed within minutes on the surface of the aminolipid bilayers. All layers are composed of silica nanospheres, and the thickness turned out to be independent of the amine headgroup.

Full text of DOI:10.1021/la1021627

pubs.acs.org/Langmuir
© 2010 American Chemical Society
Formation of Silica Precipitates on Membrane Surfaces in
Two and Three Dimensions
†
†
†
‡
€
Anja Bernecker, Joanna Ziomkowska, Svenja Heitmuller, Ralph Wieneke,
Armin Geyer,*,‡ and Claudia Steinem*,†
†Institute of Organic and Biomolecular Chemistry, University of Go€ttingen, Tammannstr. 2, 37077 Go€ttingen,
‡
Germany, and Philipps-Universitat Marburg, Hans-Meerwein-Str., 35032 Marburg, Germany
€
Received May 27, 2010. Revised Manuscript Received June 30, 2010
Ether lipids with alkyl chains of uniform length and varying amine headgroups were synthesized and assembled into
bilayer structures in aqueous solution, which served as templates for the formation of silica in two and three dimensions
produced under ambient conditions. Dynamic light scattering revealed that unilamellar vesicles of the aminolipids are
formed by the extrusion method. The alkylation of the polar amine headgroup was systematically increased from a
primary, secondary, and tertiary amine to a quaternary ammonium salt, and the amount of silica was quantified by the
β-silicomolybdate method as a function of the headgroup. A lysinol-connected ether lipid harboring two primary amine
groups was also investigated. This variation enabled us to compare the influence of the headgroup on the properties
of the precipitated silica in detail. By spreading of unilamellar aminolipid vesicles onto planar silicon substrates,
two-dimensional planar bilayers can be produced. By means of ellipsometry in conjunction with atomic force micro-
scopy, we were able to demonstrate that very thin silica layers with a thickness of a few nanometers are formed within
minutes on the surface of the aminolipid bilayers. All layers are composed of silica nanospheres, and the thickness turned
out to be independent of the amine headgroup.
Introduction
identified in these organic molecules.8-10 However, as also other
parameters such as the chain length and the distance between and
the position of the amine groups dictate the precipitation proper-
ties, the individual factors contributing to the process of silica
formation still remain unclear.
Most studies have focused on the precipitation of silica nano-
spheres from solution,2,11-15 and only very little is known about
the silica precipitation process if it occurs at a membrane interface
like it might be the case in the SDV. It has been shown that silica
nanoparticles are formed at the outer surface of catanionic lipid
vesicles.16,17 Only very recently, a few reports on the silica precipita-
tion properties of membrane confined catalytic centers arranged in
a planar geometry have been published.18-20 However, the local
environment of catalytic sites, mainly provided by amine function-
alities, can influence the precipitation properties, and thus, more
information needs to be gathered on this aspect.
Diatoms are unicellular photosynthetic organisms, which are a
major contributor to global carbon fixation with up to 25%.1
There are more than 10000 species known to science, and species
identification is mainly based on the shape and structure of their
silica-based shells. Silica polymerization occurs in a specialized
vesicle, called silica deposition vesicle (SDV). As yet, this orga-
nelle could not be isolated and characterized. However, it is
known that the SDV interior is slightly acidic. Moreover, it is
established that at least two main types of organic compounds are
associated with the silica formation process in diatoms, namely
(poly)peptides with propylamine-functionalized lysine side chains
and phosphorylated serine side chains, termed silaffins,2-5 and
long-chain polyamines.6-8 In a species-specific manner, silaffins
and long-chain polyamines are found in a complex mixture,
suggesting their important functions in structure formation
of the silica shell. A common feature is that both classes of
compounds contain a variety of amine functionalities with
different methylation patterns. Primary, secondary, and tertiary
amine groups as well as quaternary ammonium cations have been
Ether lipids with alkyl chains of uniform length and an amine
containing headgroup are expected to assemble in a bilayer struc-
ture in aqueous solution. In this study, we were able to demon-
strate that such molecules can indeed form unilamellar vesicles
(10) Sumper, M.; Brunner, E. Adv. Funct. Mater. 2006, 16, 17–26.
(11) Bernecker, A.; Wieneke, R.; Riedel, R.; Seibt, M.; Geyer, A.; Steinem, C.
J. Am. Chem. Soc. 2010, 132, 1023–1031.
(12) Kroger, N.; Lorenz, S.; Brunner, E.; Sumper, M. Science 2002, 298,
584–586.
(13) Knecht, M. R.; Sewell, S. L.; Wright, D. W. Langmuir 2005, 21, 2058–2061.
(14) Knecht, M. R.; Wright, D. W. Chem. Mater. 2004, 16, 4890–4895.
(15) Knecht, M. R.; Wright, D. W. Langmuir 2004, 20, 4728–4732.
(16) Hentze, H. P.; Raghavan, S. R.; McKelvey, C. A.; Kaler, E. W. Langmuir
2003, 19, 1069–1074.
*Corresponding authors: Tel þ49 6421 28 22030, e-mail geyer@staff.
uni-marburg.de (A.G.); Tel þ49 551 39 3294, Fax þ49 551 39 3228, e-mail
€
csteine@gwdg.de (C.S.).
(1) Johnston, A. M.; Raven, J. A.; Beardall, J.; Leegood, R. C. Nature 2001, 412,
40–41.
€
(2) Kroger, N.; Deutzmann, R.; Sumper, M. Science 1999, 286, 1129–1132.
€
(3) Kroger, N.; Deutzmann, R.; Sumper, M. J. Biol. Chem. 2001, 276, 26066–
26070.
€
(4) Poulsen, N.; Sumper, M.; Kroger, N. Proc. Natl. Acad. Sci. U.S.A. 2003, 100,
12075–12080.
(17) Lootens, D.; Vautrin, C.; Van Damme, H.; Zemb, T. J. Mater. Chem. 2003,
13, 2072–2074.
(5) Wenzl, S.; Deutzmann, R.; Hett, R.; Hochmuth, E.; Sumper, M. Angew.
Chem., Int. Ed. 2004, 43, 5933–5936.
(18) Kent, M. S.; Murton, J. K.; Zendejas, F. J.; Tran, H.; Simmons, B. A.;
Satija, S.; Kuzmenko, I. Langmuir 2009, 25, 305–310.
(19) Li, H.; Pfefferkorn, D.; Binder, W. H.; Kressler, J. Langmuir 2009, 25,
13328–13331.
(20) Wallace, A. F.; DeYoreo, J. J.; Dove, P. M. J. Am. Chem. Soc. 2009, 131,
5244–5250.
€
(6) Kroger, N.; Poulsen, N. Annu. Rev. Genet. 2008, 42, 83–107.
€
(7) Kroger, N.; Sumper, M. The biochemistry of silica formation in diatoms.
In Biomineralization; Baeuerlein, E., Ed.; Wiley-VCH: Weinheim, 2000; pp 151-170.
(8) Sumper, M.; Brunner, E.; Lehmann, G. FEBS Lett. 2005, 579, 3765–3769.
(9) Sumper, M.; Lehmann, G. ChemBioChem 2006, 7, 1419–1427.
13422 DOI: 10.1021/la1021627
Published on Web 07/19/2010
Langmuir 2010, 26(16), 13422–13428

Bernecker et al.
Article
Scheme 1. Syntheses of the Compounds pAL, sAL, tAL, qAL, and LysinolL
and planar lipid bilayers attached to a solid substrate. Formation
of these aggregates is driven by the hydrophobic alkyl chains. The
alkylation of the polar headgroup was systematically increased
from a primary, secondary, and tertiary amine to a quaternary
ammonium salt. In addition, a lysinol-connected ether lipid harbo-
ring two primary amine groups was investigated. This variation
enabled us to compare the influence of the headgroup on the
properties of the precipitated silica in detail not only in vesicular
three-dimensional structures but also in two dimensions in a planar
geometry.
Vesicle Preparation. Large unilamellar vesicles (LUVs) were
prepared according to the extrusion method.26 First, the lipid was
dissolved in chloroform (2.5-5.0 mg/mL) and placed in a glass
test tube. Lipid films were obtained by evaporation of the solvent
under a stream of nitrogen followed by vacuum for 3 h to remove
remaining solvent. The lipid films were stored at 4 °C. A lipid
suspension was obtained by incubating the lipid films in the
respective buffer solution for 15 min and vortexing three times
(30 s every 5 min). LUVs were produced by pressing the lipid
suspensionthrough polycarbonate membraneswith nominal pore
sizes of 200 nm using a miniextruder (LiposoFast, Avestin,
Canada). In the case of pAL, all steps were carried out at T=
63°C. In order to producefluorescentlylabeledvesicles, 0.5 mol%
sulforhodamine 101 DHPE was added to the lipid films.
Experimental Section
Materials. Sulforhodamine 101 DHPE, tetramethyl ortho-
silicate, and silicon atomic absorption standard solution were
from Sigma-Aldrich (Taufkirchen, Germany). Ammonium hep-
tamolybdate was obtained from Merck (Darmstadt, Germany).
Silicon substrates were purchased from Si-Mat (Landsberg/Lech,
Germany) and CrysTec (Berlin, Germany). Water was purified
first through a Millipore water purification system Milli-RO
3 plus and finally with a Millipore ultrapure water system Milli-Q
plus 185.
Synthesis of the Ether Lipids. 3-O-Benzyl-sn-glycerol (1)21
served as the common starting compound for anchoring five
different polar headgroups on 1,2-di-O-dodecyl-sn-glycerol (2)22,23
(Scheme 1). The different amines required an activation of com-
pound 2 as tosylate (3), bromide (4), and triflate (5). Tosylate 3 was
converted into the azide24 to obtain the primary aminolipid pAL in
91% yield after hydrogenation. The methylated aminolipids were
achieved by direct alkylation of the corresponding amine with
bromide 4 in dry ethanol in a pressure tube at 90 °C,25 yielding
93% of sAL, 89% of tAL, and 82% of qAL. Reduction of di-tert-
Preparation of Solid Supported Bilayers by Vesicle
Spreading. Silicon wafer pieces, either with a thin native oxide
layer or with a 20 nm thick SiO2 layer, served as substrates for the
preparation of solid supported membranes. They were cleaned in
NH3 (25%)/H2O2 (30%)/H2O (1:1:5, v/v/v) at 70 °C for 15 min
followed by rinsing with water. For fluorescence and atomic force
microscopic experiments, the silicon substrates (with a 20nmSiO2
layer) were incubated with the respective LUV suspension
(0.2 mg/mL) for 1 h at room temperature (63 °C for pAL) and sub-
sequently rinsed with buffer solution (50 mM MES, pH 6.0). For
ellipsometry measurements, bilayer formation was directly fol-
lowed in the ellipsometry cell. The LUV suspension was added to
1.5 mL of buffer solution (lipid concentration: 0.08 mg/mL)
covering the silicon wafer (with a thin native SiO2 layer). During
incubation (1 h), the solution was stirred. Subsequently, the
substrate was rinsed with buffer solution. In the case of pAL, the
spreading process was not carried out in the measuring cell due
to the higher temperature (63 °C) but under the same conditions.
Here, the thickness of the membrane was determined by ellipso-
metry after bilayer formation.
butyloxycarbonyl-L-lysine (6) with sodium boron hydride leads to
the double Boc-protected
triflate 5. Final deprotection of 8 was carried out with HCl in cold
L-lysinol (7), which was alkylated with
Quantification of SiO2 Precipitation on LUVs. For all
experiments, orthosilicic acid (Si(OH)4) was freshly prepared from
tetramethyl orthosilicate by the addition of 1 mM HCl followed by
incubation for 20 min. Silica precipitation on LUVs was quantified
according to a modified β-silicomolybdate method.27,28 Si(OH)4
(100 mM) was added to the LUV suspension (1 mM) and incu-
bated for 10 min after vortexing. The samples were centrifuged
at 14000g for 5 min, and the pellet was washed with water.
diethyl ether to obtain the hydrochloride of LysinolL.
(21) Yamauchi, K.; Une, F.; Tabata, S.; Kinoshita, M. J. Chem. Soc., Perkin.
Trans. 1 1986, 765–770.
(22) Abdelmageed, O. H.; Duclos, R. I.; Griffin, R. G.; Siminovitch, D. J.;
Ruocco, M. J.; Makriyannis, A. Chem. Phys. Lipids 1989, 50, 163–169.
(23) Kretschmar, G.; Toepfer, A.; Sonnentag, M. Tetrahedron 1998, 54, 15189–
15198.
(24) Bhattacharya, S.; Ghanashyam, S. N. Langmuir 2000, 16, 87–97.
(25) Bhattacharya, S.; De, S. Chem.;Eur. J. 1999, 5, 2335–2347.
(26) Hope, M. J.; Bally, M. B.; Webb, G.; Cullis, P. R. Biochim. Biophys. Acta
1985, 812, 55–65.
(27) Iler, R. K. The Chemistry of Silica; Wiley: New York, 1979.
(28) De Smedt, F.; Stevens, G.; De Gendt, S.; Cornelissen, I.; Arnauts, S.;
Meuris, M.; Heyns, M. M.; Vinckier, C. J. Electrochem. Soc. 1999, 146, 1873–1878.
Langmuir 2010, 26(16), 13422–13428
DOI: 10.1021/la1021627 13423

Article
Bernecker et al.
Table 1. Distribution of the Hydrodynamic Diameter dH of the Aggregates (c = 1 mM) Obtained after Extrusion through Polycarbonate
Membranes with a Nominal Diameter of 200 nm As Obtained by Dynamic Light Scattering Together with the Relative Change in the Hydrodynamic
Radius after Addition of Si(OH)4
a
compound
dH/nmb
rH-end/rH-start/%
A/molecule-1/nm2
mSi/μgc
mSi/μgd
pAL
sAL
tAL
qAL
180 ( 30
160 ( 40
210 ( 80
100 ( 70
350 ( 170
800
600
100
4000
1300
0.15
0.23
0.24
0.31
0.51e
22 ( 4
28 ( 5
9 ( 4
59 ( 5
76 ( 34
4 (5.5)
6 (4.7)
6 (1.5)
8 (7.4)
13 (5.8)
LysinolL
a Taking the area per molecule into account, the mass of precipitated silica, given as mSi, obtained in 50 mM MES, pH 6.0 can be compared to the
theoretically expected mSi. The number in parentheses gives the factor by which the measured mSi exceeds the calculated one. b The values are given as
mean values ( standard deviation. c The values are given as mean values ( standard deviation (n = 5). d Calculated values. e Determined by film balance
measurements.
The angle of incidence was set to 70°. All measurements were
performed in a fluid ellipsometer cell. From the ellipsometric
angles δ and ψ the thicknesses of the layers were calculated
~
using the following refractive indices: NSi = 4.0000 - i0.0538,
nSiO =1.4610, nmembrane = 1.4884, nbuffer = 1.3328.
2
Atomic Force Microscopy (AFM). AFM experiments were
carried out with a NanoWizard II atomic force microscope (JPK
Instruments, Berlin, Germany). All measurements were perfor-
med in ultrapure water using intermittent contact mode. Silicon
CSC37 (spring constant k = 0.3 N/m, resonance frequency f0 =
21 kHz, Mikromasch, Tallinn, Estonia) and gold-coated silicon
nitride MSCT (k = 0.5 N/m, f0 = 120 kHz, Veeco, Camarillo)
cantilevers were used.
Results and Discussion
Figure 1. Size distribution of aggregates formed from LysinolL as
determined by dynamic light scattering in the presence of different
salt concentrations.
Synthesis. 1,2-Di-O-dodecyl-sn-glycerol (1) serves as a com-
mon precursor of ether lipids with various amine headgroups
(Scheme 1). Two dodecyl chains are necessary to form lipid
bilayers. Activation of the primary hydroxyl group of compound
3 required three different leaving groups, namely tosylate (3),
bromide (4), or triflate (5), in order to maximize the yield of the
ether lipids on the gram scale. The size of the headgroup increases
linearly from pAL to tAL. The headgroup of LysinoL is con-
siderably larger and harbors two amine groups.
This procedure was repeated three times to remove free silicic
acid. Silica was then dissolved in 2 M NaOH for 60 min. For
the modified β-silicomolybdate method, 1.35 mL of HCl
(37%) was dissolved in 40.3 mL of H2O, and 774.2 mg of
[(NH4)6Mo7O24 4H2O] was dissolved in 9.7 mL of H2O. Both
3
solutions were mixed, and the pH was adjusted to 1.12 with
2 M NaOH. Cuvettes were filled with 160 μL of H2O, 800 μL of
molybdate solution, and a defined volume of silica/NaOH solution
(0, 20, or 40 μL) and a corresponding volume of 2 M NaOH (0, 20,
or 40 μL). The absorbances of the solutions were measured at a
wavelength of 380 nm. A silicon atomic absorption standard solu-
tion was used to generate calibration curves.
Determination of the Aggregates in Solution. Prerequisite
for the formation of planar bilayers on substrates by vesicle
spreading is the formation of unilamellar vesicles in solution.
Driven by the hydrophobic tail of each aminolipid in conjunction
with the geometry of the molecules, it is expected that unilamellar
vesicles can be formed by the extrusion method.29,30 Aminolipid
suspensions were extruded through polycarbonate membranes
with a nominal diameter of 200 nm, and the resulting aggregates
immediately analyzed by dynamic light scattering (DLS) as it is
known that positively charged vesicles tend to aggregate and fuse
into larger structures.31 The results are summarized in Table 1. All
compounds form aggregates in 50 mM MES, pH 6.0 with
hydrodynamic diameters larger than 100 nm indicative of the
formation of unilamellar vesicles. However, the size varies with
the methylation pattern of the headgroup. While pAL, sAL, and
tAL form aggregates with a mean diameter of around 200 nm as
expected from the nominal pore diameter of the polycarbonate
membranes, the average size of the aggregates for qAL is only half
the size. In case of LysinolL, the situation is completely different
(Figure 1). Here, the average size of the aggregates is 350 nm,
twice as large as expected, which might be a result of the larger
headgroup. Besides these large aggregates, also small ones with
Dynamic Light Scattering (DLS). For the determination of
aggregate sizes in solution, DLS measurements were carried out.
A laser witha wavelength ofλ = 532.1 nmwas used, and scattered
light was detected at an angle of 90°. The temperature of the
sample solutions in the measuring cell was kept constant at 25 °C.
The concentration ofall lipid suspensions was 1 mM, and Si(OH)4
was added in a concentration of 100 mM.
Fluorescence Microscopy. The quality of the supported bi-
layers was analyzed by fluorescence microscopy on silicon substrates
coated with a 20 nm thick SiO2 layer to minimize fluorescence
quenching. An AxioTech Vario microscope (Zeiss, Jena, Germany)
equipped with an Achroplan 40ꢀ/0.8 W water immersion objective
and the filter set 45 was used for fluorescence microscopy inspection.
Scanning Electron Microscopy (SEM). An electron micro-
scope Leo Supra 55VP (Zeiss, Oberkochen, Germany) operating
at 5.00 kV was used to image silica structures resulting from pre-
cipitationreactions onLUVs. Sampleswerepreparedasdescribed
above but were not dissolved in NaOH after centrifugation and
washing. The pellet was resuspended, and 3 μL of the solution was
dropped onto a SEM sample holder and dried overnight. The
as-prepared samples were investigated without further treatment.
Ellipsometry. A commercially available ellipsometer (EL
X-02C, DRE GmbH, Ratzeburg, Germany) was used, which
was equipped with a laser with a wavelength of λ = 632.8 nm.
(29) Israelachvili, J. N.; Mitchell, D. J.; Ninham, B. W. Biochim. Biophys. Acta
1977, 470, 185–201.
(30) Macdonald, R. C.; Macdonald, R. I.; Menco, B. P. M.; Takeshita, K.;
Subbarao, N. K.; Hu, L. R. Biochim. Biophys. Acta 1991, 1061, 297–303.
(31) Carmona-Ribeiro, A. M. Chem. Soc. Rev. 1992, 21, 209–214.
13424 DOI: 10.1021/la1021627
Langmuir 2010, 26(16), 13422–13428

Bernecker et al.
Article
Figure 2. Dynamic light scattering experiment. Time-resolved
change of the hydrodynamic radius of a vesicle suspension com-
posed of 1 mM LysinolL in 50 mM MES, pH 6.0 after the addition
of 100 mM Si(OH)4 at t = 0 min.
Figure 3. Scanning electron micrograph of silica structures result-
ing from the addition of 25 mM Si(OH)4 to a LysinolL LUV
suspension (prepared withpolycarbonate membranes with a nomi-
nal pore diameter of 400 nm). Scale bar: 200 nm.
a size of about 5 nm and below are observed, which are attributed
to micelles formed in solution. By reducing the Debye length by
adding 500 mM KCl, a single and rather narrow size distribution
is observed with a mean diameter of 130 ( 50 nm. One broad size
distribution with a mean diameter of 120 ( 120 nm is found if an
intermediate KCl concentration of 200 mM was chosen.
samples were taken for SEM images. A characteristic SEM image
of LysinolL LUVs after silica precipitation is shown in Figure 3.
Hollow and rather spherical structures are found supporting the
idea that the vesicle acts as a template for the formation of a
surface bound thin silica layer.
Silica Formation in the Presence of LUVs. We asked the
question whether the formed aggregates are capable of precipitat-
ing silica and how the methylation pattern influences the silica
precipitation process. First, the change in the hydrodynamic
radius was followed by DLS upon addition of silicic acid. Figure 2
shows a representative time course of the change in hydro-
dynamic radius for LysinolL. At t = 0, silicic acid was added, and
within 1-3 min, the hydrodynamic radius increased considerably
by several hundred percent (Table 1). This observation is attri-
buted to the formation of silica around the vesicles and aggre-
gation processes of the silicified vesicles. After 10 min, the process
is already completed. For all aminolipids, almost the same time
course was observed with a significant increase in the hydro-
dynamic radius (Table 1). However, the hydrodynamic radius
does not represent the actual size of the vesicles. Hence, to be able
to quantify the amount of precipitated silica, the β-silicomoly-
bdate method was used. The results of this assay, given as mSi
obtained after a reaction time of 10 min, are presented in Table 1.
The smallest silicon quantity was found for tAL, which was
considerably less than that found for pAL and sAL. In the
presence of qAL, double as much silicon was precipitated as for
sAL. In the presence of LysinolL containing two primary amine
groups, a silicon mass of 76 μg was obtained, which is more than
double the mass obtained in presence of pAL. To roughly
estimate the silicon mass mSi that can be theoretically expected,
we assumed that all lipids (1 mM) in the suspension are arranged
in LUVs with diameters as determined by DLS prior to the
addition of Si(OH)4 (Table 1). Taking the area per molecule of
each aminolipid into account, the total surface area can be
calculated. For LysinolL, the area per molecule was determined
by film balance measurements, while for the other aminolipids,
the areas per molecule were estimated using three-dimensional
models. For the calculation of the silica precipitates, it was further
assumed that only one layer of edge-linked tetrahedrons is formed
on the vesicle surface and that each SiO2 unit exhibits an area of
0.13 nm2. The calculated masses are summarized in Table 1. From
this rough approximation in comparison to the measured silicon
masses, it becomes obvious that only few layers of silica are
formed on the surface of the LUVs. To corroborate this finding,
For pAL and sAL, as well as for LysinolL, a factor of roughly 5
between the calculated and measured silicon mass is found. This is
not the case for tAL, where only a factor of 1.5 is calculated, while
for qAL, a larger factor of 7.7 was determined. This finding might
be discussed in terms of the surface charge density. It is expected
that positively charged ammonium cations are decisive for silica
precipitation. Thus, the pKa value of each compound except for
qAL, which is permanently positively charged, in an aqueous
environment was calculated using the program SPARC.32 While
the pKa values of pAL (pKa = 8.7), sAL (pKa = 9.1), and
LysinolL (pKa = 9.1, 10.3) are around 9-10 as expected, the
pKa value of tAL (pKa = 5.4) is considerably smaller as a result of
the differential solvation effect in water; i.e., the ammonium
cation gets less well solvated with an increased number and length
of alkyl chains attached to the nitrogen atom. This smaller pKa
value implies that the surface charge density for tAL at pH 6.0 is
significantly decreased compared to that of pAL and sAL.
However, another aspect needs to be considered. The pKa values
were calculated assuming amine monomers, and it is controver-
sially discussed how the accumulation of charged molecules at an
interface influences the pKa value. Several different techniques
have been developed to measure the pKa values of self-assembled
monolayers (see ref 33 and references therein). Fears et al.33
reported on a measured pKa value of 6.5 for a self-assembled
monolayer (SAM) composed of HS-(CH2)12-NH2 based on the
bulk pH, while they obtained 8.9 based on the theoretical analysis
of the local pH at the SAM surface. This demonstrates that it is
still difficult to calculate the exact ionization state based on the
determined pKa value. The tendency is that the pKa value, if
altered, changes to smaller values, which would then lead to a
reduced surface charge density. This is in line with the observation
that the permanently charged ammonium cation of qAL is the
(32) Hilal, S. H.; Karickhoff, S. W.; Carreira, L. A. Quant. Struct.-Act. Relat.
1995, 14, 348–355.
(33) Fears, K. P.; Creager, S. E.; Latour, R. A. Langmuir 2008, 24, 837–843.
Langmuir 2010, 26(16), 13422–13428
DOI: 10.1021/la1021627 13425

Article
Bernecker et al.
Figure 4. Fluorescence micrographs of membranes of (a) pAL, (b) sAL, (c) tAL, (d) qAL, and (e) LysinolL doped with 0.5 mol %
sulforhodamine 101 DHPE. Membranes were prepared by spreading of LUVs on silica wafers under the conditions described in Table 2 and
subsequent rinsing with 50 mM MES buffer, pH 6.0. Scale bars: 50 μm.
Table 2. Ellipsometrically Determined (d) and Calculated (dcal)
Thicknesses of the Planar Bilayers after Spreading of Aminolipid
LUVs (c = 0.08 mg/mL), n = 3
compound
d/nm
dcal/nm
spreading conditions
pAL
sAL
tAL
qAL
2.9 ( 1.1
3.9 ( 0.5
3.1 ( 0.1
2.9 ( 0.1
3.4 ( 0.3
2.9
3.1
3.1
3.1
4.0
50 mM MES, pH 2.4, 63 °C
50 mM MES, pH 3.7
50 mM MES, pH 4.0
50 mM MES, pH 6.0
25 mM MES/0.5 M KCl, pH 6.0
LysinolL
Table 3. Thickness of the Precipitated Silica Layer on the
Aminolipid Bilayera
compound d(model 1)/nm d(model 2)/nm k1 ꢀ 10-2/s-1 k2 ꢀ 10-3/s-1
pAL
sAL
tAL
qAL
0.9 ( 0.2
1.3 ( 0.2
1.6 ( 0.2
1.3 ( 0.1
1.8 ( 0.3
1.5 ( 0.3
2.1 ( 0.4
2.6 ( 0.3
2.2 ( 0.1
3.0 ( 0.6
4.4 ( 2.6
4.2 ( 1.5
3.9 ( 0.5
1.7 ( 0.4
2.2 ( 0.5
4.6 ( 1.6
3.1 ( 1.1
2.4 ( 0.1
2.9 ( 1.0
2.3 ( 0.3
LysinolL
a For model 1, the thickness was calculated assuming the deposition
of a homogeneous bulk silica layer with a dielectric constant of 2.135.
For model 2, an effective medium approximation according to
Bruggeman36 was applied assuming a volume fraction for silica of 0.6.
k1 and k2 are the kinetic constants as obtained from a biexponential fit to
the time-resolved ellipsometric data.
results were found for catanionic surfactant vesicles. Catanionic
surfactant vesicles composed of a mixture of cationic cetylmethyl-
ammonium hydroxide and anionic myristic acid were used, which
produced polyhedral vesicles. Silica coating at the surface of these
structures was observed by SEM, indicating a thickness of about
10 nm.17 Similarly, Hentze et al.16 found hollow spheres molded by
surfactant vesicles with a thickness of 1-2 nm. Both groups made
use of positively and negatively charged surfactants, while in this
study only one component, a positively charged lipid, was used; i.e.,
no negative charges or even polyvalent ions were required for the
precipitation process.
Figure 5. (a) Time-resolved ellipsometric measurement of the
thickness change during the spreading and fusion process of qAL
vesicles (c = 0.08 mg/mL) in 50 mM MES, pH 6.0. The vesicles
were added at t = 0. (b) Time-resolved ellipsometric measurement
ofthe thickness changeafter the additionofSi(OH)4 (c = 100 mM)
at t = 0 to a qAL bilayer in 50 mM MES, pH 6.0.
Formation of Planar Bilayers on Silicon Substrates. To
form planar bilayers composed of the different aminolipids,
unilamellar vesicles were used, which were expected to spread
and fuse on a solid substrate like it was demonstrated for other
positively charged lipid vesicles.34,35 The process was followed by
one that precipitates the largest amount of silica as it is not
influenced by the pH and the geometry.
The results clearly demonstrate that the assembly of positively
charged lipids in spherical structures allows the very rapid forma-
tion of silica under ambient conditions. The positively charged
headgroups are the catalytic centers for silica formation. Only in
the headgroup region, silica is formed leading to very thin and
hollow structures, which are molded by the unilamellar vesicles.
Even though not performed under biomimetic conditions, similar
(34) Richter, R.; Mukhopadhyay, A.; Brisson, A. Biophys. J. 2003, 85, 3035–
3047.
(35) Richter, R. P.; Brisson, A. Biophys. J. 2005, 88, 3422–3433.
(36) Bruggeman, D. A. G. Ann. Phys. (Leipzig, Ger.) 1935, 24, 636–664.
13426 DOI: 10.1021/la1021627
Langmuir 2010, 26(16), 13422–13428

Bernecker et al.
Article
Figure 6. Atomic force microscopy images of (a) a qAL bilayer after vesicle spreading and (b) the same sample after silica formation.
(c) Height profile along the white line in (b). (d, e) Silica structures on LysinolL bilayers. The bilayers in (b, d, e) were incubated with 100 mM
Si(OH)4 in 50 mM MES, pH 6.0 for 1 h. Scale bars: 2 μm.
means of ellipsometry (see Figure 5a). Within 10 min after the
addition of vesicles composed of qAL, a lipid layer was formed
with a mean thickness of 2.9 ( 0.1 nm. This value is in good
agreement with that of a lipid bilayer with a calculated thickness
of 3.1 nm taking the geometry of the molecules into account. For
all five compounds, thicknesses were obtained, which indicate the
formation of lipid bilayers on the silicon substrate (Table 2).
A slightly larger thickness, found for sAL, can be attributed to
residual vesicles attached to the bilayer, while a slightly smaller
thickness is indicative of some defects in the bilayer structure.
To further corroborate the formation of planar bilayers on the
surface, fluorescently labeled vesicles were spread and the surface
was inspected by means of fluorescence microscopy after the
process had been completed (Figure 4). All images show a rather
homogeneous fluorescence intensity demonstratingthe formation
of lipid layers on the substrate. In addition, there are some bright
spots discernible, which can be attributed to vesicles attached to
the bilayer. While the spreading processes were performed under
optimized buffer conditions (Table 2), the fluorescence images
were all taken in 50 mM MES, pH 6.0.
Silica Precipitation at Planar Bilayers. Time-resolved
ellipsometry measurements allow an investigation of the silica pre-
cipitating properties of the planar bilayers composed of the five
compounds differing in their headgroups. Figure 5b shows the
result of a characteristic time-resolved ellipsometric measurement.
After a qAL bilayer had been formed on the silicon substrate,
Si(OH)4 was added and the change in δ and ψ was monitored in a
time-resolved manner. As a first approach, we calculated the
thickness of the precipitated silica layer by assuming a continuous
and homogeneous SiO2 layer deposited on the lipid bilayer. This
results in a thickness change of 1.3 ( 0.1 nm on top of the qAL
bilayer. The same experiment was performed with all five com-
pounds resulting in thicknesses, which are summarized in Table 3.
From the overall thickness changes, it can be concluded that a thin
layer of silica is precipitated on the lipid bilayer similar to what has
been found on the surface of unilamellar vesicles. However, a
significant difference in thickness change as a function of the head-
group could not be observed. This might be a result of the different
geometries of the two systems (vesicles and planar bilayers). The
planar arrangement of the molecules on the solid support influences
their packing geometry, their flexibility as well as their pKa values,
and thus the surface charge density. As the charge of the ammonium
cation as well as the distance between the positively charged groups
appear to be decisive for silica precipitation, the given geometry
might influence the amount of precipitated silica and does not need
to solely follow the pKa values of the amine groups.
To analyze the structure of the precipitated silica inmore detail,
we performed AFM imaging. Figure 6a shows an AFM image of
a qAL bilayer before the addition of Si(OH)4. A smooth surface is
observed with no defects. After the addition of Si(OH)4 for 1 h
and rinsing with buffer, the topography of the surface has consi-
derably changed. Silica spheres with diameters in the nanometer
regime are found on the surface, which are difficult to image. The
tip of the cantilever gets contaminated very quickly, resulting in
imaging artifacts. To be able to obtain height information from
the sample, a window was scratched in the SiO2 decorated bilayer
by applying high load forces. A height difference between the
silicon substrate and the silica precipitate of about 8 nm is ob-
tained (Figure 6c). Taking the bilayer thickness of about 3 nm into
account, the silica thickness can be estimated to be roughly 5 nm,
which suggests that only a few layers of silica spheres are formed
on top of the bilayer surface. However, the thickness obtained
from the AFM images is larger than that found by ellipsometry.
The ellipsometric thickness was achieved by assuming a solid
silica layer, which is obviously not the case. To account for a
heterogeneous silica layer composed of silica spheres in an
aqueous environment, an effective medium approximation was
applied.36 From the AFM images, it is estimated that 60% of the
layer is composed of silica spheres, while the remaining 40% void
is filled with water. Then, a thickness of 2.2 nm instead of 1.4 nm is
calculated for the silica layer atop the qAL bilayer (Table 3). This
is still almost by a factor of 2 smaller than the thickness observed
in the AFM images. Assuming that only 40% of the layer is
composed of silica spheres would result in an additional increase
in thickness of about 1 nm, which is still slightly lower than the
observed thickness and is not fully supported by the AFM images.
Langmuir 2010, 26(16), 13422–13428
DOI: 10.1021/la1021627 13427

Article
Bernecker et al.
It is more likely to assume that the silica spheres are not
completely made from bulk SiO2 with a dielectric constant of
2.135 but are rather porous and hence have a reduced dielectric
constant.
Conclusions
The results demonstrate that synthetic ether lipids with amine
headgroups are stable in aqueous solution and form bilayer
structures, which can serve as templates for the formation of
silica structures. They not only form three-dimensional vesicles in
solution, which can be silicified, but they also enable one to
produce planar bilayers on solid substrates. These planar ar-
rangements of positively charged aminolipids serve as catalytic
sites for silica formation on a fast time scale and under ambient
conditions, resulting in a very thin layer of silica nanospheres
attached to the bilayer without need of polyvalent or monovalent
anions. The size of the spheres as well as the overall thickness of the
layer is almost independent of the amine headgroup. By inducing
phase separation between silica precipitating aminolipids and non-
silica inducing lipids, it is well conceivable to be able to obtain two-
dimensional patternings of silica on planar surfaces.
Another aspect of silica precipitation is the time scale, on which
this process occurs. We asked the questionwhether the headgroup
of the aminolipids influences the kinetics. A biexponential func-
tion was fit to the ellipsometric data. A monoexponential fit did
not sufficiently describe the data. The results are summarized in
Table 3. The parameter k1 accounts for the first fast process, while
k2 describes the slower process, which is ∼10 times slower. We
ascribe the first process to the fast headgroup-induced silica
precipitation process, while the second one can be attributed to
the sol-gel process occurring while the Si(OH)4 solution is in
contact with the surface. Hence, the kinetic constantk2 is expected
to be the same for all five systems under investigation, which is the
case within the error of the experiment. The headgroup-assisted
kinetics of the precipitation process is, however, also only very
slightly influenced by the structure of the molecule with a
tendency of LysinolL and qAL having a 2-fold lower k1 value
compared to pAl, sAL, and tAL.
€
Acknowledgment. We thank Hans-Jorg Langer for experi-
mental assistance. Financial support by the VW-Stiftung (I/82 042)
is gratefully acknowledged.
13428 DOI: 10.1021/la1021627
Langmuir 2010, 26(16), 13422–13428