DEVELOPMENTS IN BIO-NANOTUBE SCIENCE AND TECHNOLOGY
DAVID TANNER MITCHELL
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
This dissertation is dedicated to my parents, Dr. James P. and Terri Tanner Mitchell.
I would like to thank Dr. Charles Martin and the Martin group members I have had the
opportunity to work with over the course of my tenure. Dr. Martin never stopped
thinking of applications for my research and taught me much about effective scientific
communication. The Martin group members have been unanimously supportive and
terrific examples of ingenuity and perseverance. Kshama Jirage, Veronica Cepak, John
Hulteen, and Silvia di Vito Luebben showed great patience in training me in various
analytical techniques and instrumentation. Michelle Steen and Shawn Sapp gave
insightful advice on experiment design. In particular I would like to thank Charlie
Patrissi for his help in proofreading this dissertation.
Lacramioara Trofin and Naichao Lee spent many hours growing aluminum oxide for
my experiments, for which I am in their debt. Marc Wirtz and Erich Steinle taught me
the fundamentals of impedance spectroscopy and both were helpful in running
experiments on ion channel mimetic membranes. The enantioselective antibody
experiments would have been impossible without the aid of Sang Bok Lee who helped
both procure the antibody and analyze the extraction experiment results.
Scott Miller, Elizabeth Medeiros, and Shufang Yu discussed the experiments in this
dissertation with me at length and provided many helpful insights.
Professor Vaneica Young was very kind to run and help interpret the XPS
experiments. Drs. Bob Lee, John Chandler and Ameila Dempere gave me excellent
training in electron microscopy. Professor Dave Grainger was instrumental in my
decision where to attend graduate school and has been a continual source of sage
scientific and career advice. Drs. Donn Dennis and Tim Morey were very helpful in
discussing experimental results and planning future experiments. Drs. Hans Soderlund
and Tarja Nevanen kindly provided the enantioselective antibody ENAI lHis.
My life during graduate school would have been much less interesting without the
leavening influences ofLiisa Kauri, Cathy Zelenski, Dean Medeiros, and Nathalie Fosse.
Special thanks for keeping me climbing to my friends Craig Luebben, Lane Hale, Corey
Salzer, Andrea Koenig, Jason McDannold, Joie and Jann Contado, Mitzy Carlo, and
Finally, my family provided unflagging moral support throughout my graduate career.
For this support I thank Heather, John, Benjamin, and James Howell; Jonathan,
Rebekkah, Anthony, Rachel, Adam, and Tanner Mitchell; Heidi Mitchell, Adam and
Shannon Mitchell, and Jenni Mae, Matthew, and Ashley Mitchell. Above all, thanks to
my parents Jim and Terri Mitchell for instilling in me a love of learning.
TABLE OF CONTENTS
A CK N O W LED G M EN TS ................................................................................... iii
ABSTRACT .... ..... .................................... viii
1 INTRODUCTION AND TEMPLATE SYNTHESIS........................................ 1
Introduction................... ................................. ......... .. ................. ....... 1
B ack g ro u n d ............................................................................. ... ................. ...... 3
Template Synthesis .......................... ........................................... .................3
Three-dim ensional templates ............................................ ................... 3
Two-dim ensional tem plates ............................. .... ......... ................. 5
One-dimensional templates................................... 5
A lum ina G row th ............................................ ....... ..................................... 11
P olishing ................................................................... ... ...... 11
Perfect pores ......................................................................... .............. 15
Separation of the alumina film ................ .............................. .......... ..... 16
Sol-G el C hem istry ................ ...................... ... 19
S ila n iz atio n ..................................................................... ................. 2 0
C chapter Sum m aries...................................... ................................ ......... ... 24
2 CHEMICAL MODIFICATION OF SILICA NANOTUBES.................................... 26
In tro d u ctio n ......................................................................................................... 2 6
B background .......................................................... ......... ........ ............ .. 26
D rug D delivery .................... .. .. .......... .... ................... 27
Types of nanoparticles for drug delivery ..................................................... 27
Advantages of using nanoparticles for drug delivery ..................................31
Engineered Sorbants and Extractants .....................................................32
E xperim mental ............................................ ............. ................. ............... .... 33
T em plate Synthesis ............................................................................................. 33
Silanization ...................... ...... ....... .................. ....... 34
Protein Modification ............................................................ .......... 39
Biotinylation .................................... ............................... ........ 39
C om posite N anotubes ............................................... ................ ................ 41
R esu lts and D discussion ............................................................................................. 42
D ifferential Inside/Outside M odification ............................. ........................... 42
Hydrophilic/lipophilic nanotubes ............................................ ............ 42
Quinine/Au nanoparticle nanotubes.................. .. ............ ..... 45
Au nanoparticle/fluorescent protein nanotubes .............................................47
N anotubes as Extractants .............................................. .............................. 49
Cls nanotubes for drug detoxification.............................. ............. 49
EDTA nanotubes for copper(II) extraction .................................... 56
Composite Silica-Polymer Nanotubes .................... .................... 57
Insulating polymers................................................ 58
P o ly (p y rro le)........................................................................ .................. 59
Self-A ssem bly .............. ............... ................................ .... ......... .......63
Biotin/avidin self-assem bly......................... ........................................ 64
D ithiol self-assem bly ............... .......................................................... 69
C conclusions .................. ................................ ......... .. ............... 70
3 PROTEIN MODIFICATION OF SILICA NANOTUBES ....................................... 72
In tro d u ctio n ......................................................................................................... 7 2
Background .................................. ................. ............ .............. ..... 72
S m art N ano particles ............................................................................................ 73
Biomolecule-Modified Particles............................................... 76
E x p erim mental ................................................................ ...... .... ........ .... .... 7 8
R esu lts and D iscu ssion ..................................................................................... 8 1
P rotein C oupling ............................... ................................ .... ....... .............. 82
Extractions Using Protein-Modified Nanotubes ................ .................. 85
A vidin/biotin ............................ .. ... ............... .... ............... 85
Immunoprotein extractions ............... .. ......................... ................ 87
Enantioselective extractions............................ .............. ................... 88
Directed Nanotube Binding............................ .. ................... ...................... 95
Enzyme Attachm ent and Assays ................................................................... 100
G lucose oxidase ................................... ......... .... ........... ........ ... 100
Cytocrome P450 enzymes for drug detoxification ................................... 102
Conclusions ............... ......... ....................................... ....... ............ 106
4 ION CHANNEL MEMETIC SENSORS............. ................. 108
Introduction ...................... .. ............................ 108
B ackg ro u nd .................. ....... .......... .. .... .. .... .... .. ................... .......... ...... 10 8
Experimental .................................... ................ 10
M aterials.................................................. ............ .. .... ... .. ................ 110
M em brane Preparation........................................................ .................. 110
AC Impedance Measurements......................................... 112
X-Ray Photoelectron Spectroscopy (XPS) ............................................ 112
Transport Experim ents.......................................... 113
Ion-Current Measurements.......................................... .................... 114
R results and D isscussion......................... ....................................................... 114
XPS Studies................................................................... ................ 114
AC Impedance Experiments with DBS Analyte ........................................ 115
Transport Experim ents......................................................... ................. 118
M easurem ents of Ion Current....................................................................... 120
Effect of Alkyl Chain Length and Nature of the Surfactant Head Group ........... 121
D election of D rug M molecules .......................... ....................... .............. 123
The Effect of Pore Density and Pore Diameter on Analyte Detection................ 124
C o n clu sio n s..................................................... ................ 12 6
5 C O N C LU SIO N S ........................................................................ ................. 127
LIST OF REFERENCES ......................... ................ 130
BIOGRAPHICAL SKETCH ..................................... ............................................ 140
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DEVELOPMENTS IN BIO-NANOTUBE SCIENCE AND TECHNOLOGY
David Tanner Mitchell
Chair: Dr. Charles R. Martin
Chapter 1 provides background information on the template synthesis of
nanomaterials. The template synthesis method is examined with special attention to the
use of membranes containing monodisperse cylindrical pores as templates. Several
examples of the utility of template-synthesized nanomaterials are given. The production
of one type of template membrane, nanopore alumina, is reviewed. Reviews of sol-gel
and silane chemistry are also provided.
In Chaper 2, a sol-gel template synthesis process is used to produce silica nanotubes
within the pores of alumina templates. The nanotubes can be modified using a variety of
chemistries, typically via a silanization process. Because the nanotubes are formed in a
template, the interior and exterior surface can be modified independently. Modified
nanotubes can be used for drug detoxification or as extractants for the removal of metal
ions. The nanotube surface can also be biotinylated, which causes binding to avidinated
surfaces. Composite microtubes of silica and various polymers are also prepared.
Additionally, Au nanowires are shown to assemble with colloidal Au particles using
dithiols as linkers.
Chapter 3 describes the attachment of proteins onto template-synthesized silica
nanotubes. The proteins are covalently linked via an aldehyde silane bridge that binds to
pendant primary amino moieties on the protein. Protein-modified nanotubes function as
highly specific extractants. Avidin-modified nanotubes extract biotin-coated Au
nanoparticles from solution with high extraction efficiency. Immunoprotein-modified
nanotubes extract the corresponding antibody from solution with high specificity.
Antibody-modified nanotubes extract one enantiomer from a racemic mix. Enzymes,
including drug detoxification enzymes, were also attached to the nanotubes and were
shown to retain their catalytic activity. Immunoproteins on the outside of nanotubes can
be used to direct nanotube binding, creating specific labeling agents.
Chapter 4 describes the production of synthetic microporous membranes that mimic
ligand-gated ion channels. The membranes are prepared from microporous alumina that
is silanized with a silane containing a hydrophobic C18 chain. The hydrophobic C18
chains exclude aqueous solutions from entering the pores. Ionic surfactant molecules act
as a chemical stimulus to allow water to flood the pores, which in turn permits molecular
flux across the membrane. The effect of concentration and hydrophobicity of the surfact
analyte species are examined, as is the effectiveness of several drug compounds as
analytes. The results and conclusions of this dissertation are summarized in Chapter 5.
INTRODUCTION AND TEMPLATE SYNTHESIS
Nanomaterials constitute an emerging subdiscipline at the intersection of the chemical
and materials sciences (1, 2). This field poses an important fundamental question-how do
the electronic, optical, and magnetic properties of a nanoscopic particle differ from those
of a bulk sample of the same material? This issue is of particular importance because all
properties of a material change as the particle size approaches molecular dimensions and
because it is often the unique properties of the nanomaterial that make it useful for a
One example of this concerns the optical properties of gold. The yellow metallic
luster of bulk gold is well known. In contrast, solutions ofnanoparticles of gold are
colored from ruby red to purple, depending on the nanoparticle size. This is due to a
characteristic visible absorption band called the plasmon resonance absorption (3).
Extinction coefficients for these nanoparticles are extremely high, ranging from 108 to
1011 MTcm1', depending on particle size. The high extinction coeffients of gold
nanoparticles make them useful as labeling agents in optical microscopy or as indicators
for colorimetric tests, such as pregnancy tests (4).
There is no universally agreed-upon definition of what constitutes a nanomaterial. It
is generally considered that at least one dimension (d) must be less than or equal to 100
nm. Just larger than this are microscale materials, with dimensions in the range of 100
nm < d< -10 im. Materials in both these regimes will be discussed in this dissertation.
In addition to size, control of shape is of primary importance in nanomaterials science.
Because (at the nanometer scale) surface properties are dominant, particle shape often
determines material properties. Again using gold nanoparticles as an example, a large
shift in particle color is observed when gold nanoparticles are transformed from an
elliptical to a spherical shape (5).
There are two general methods for controlling the shape of a material. These are
usually termed top-down or bottom-up. An example of top-down shape control is the
production of a two-by-four. A round tree of dimensions larger than 2" x 4" is sawn
down to those dimensions. Bottom-up shape control is exemplified by the pouring of
concrete. Because the concrete is fluid before setting, it will take the shape of the
recepticle or form into which it is poured. After setting, the shape of the concrete is
locked and the form can be removed.
In recent years, our laboratory has pioneered a bottom-up method for the production of
nanomaterials. This method involves synthesizing the nanomaterials inside of a porous
template (and therefore is called template synthesis). This chapter provides background
information on the template synthesis of nanomaterials. The template synthesis method
is examined with special attention to the use of membranes containing monodisperse
cylindrical pores as templates. Several examples of the utility of template-synthesized
nanomaterials are given. The production of one type of template membrane, alumina, is
then reviewed. Reviews of sol-gel and silane chemistry that are extensively used in later
chapters are then provided.
One of the easiest ways to control the shape and size of a material is to synthesize it
inside a form or template. This process is ubiquitous in the manufacturing world. Car
tires, concrete walls, and gelatin desserts are all examples of products produced by
casting the precursor material into a mold or form. The size and shape of the resulting
object is determined by the size and shape of the template. This remains true down to
nanoscopic dimensions. Several natural and synthetic materials have pores or voids in
the micro- or nanoscale and have been used as templates. Typically these templates are
classified by the dimensionality of their void spaces and therefore the dimensionality of
the resulting templated structures: three-dimensional, two-dimensional, or
Three-dimensional template materials include micelles, glasses, zeolytes, and block
copolymers. In the case of glasses or block copolymers, the pores are the result of
dissolution of a heterogenous phase. For example Vycor 7930 glass, manufactured by
Corning, is produced by annealing borosilicate glass to produce intercalating phases that
are correspondingly boron rich or silica rich. The boron-rich phase can then be dissolved
with acid to leave a three-dimensional network of pores. The average size of the pores
can be controlled by the length of the annealing time. Commercially available sizes
range from 4 to 20 nm. Indium has been templated into Vycor glass, resulting in an
electronically conducting composite material (6). Vycor has also been used to template
polymers such as poly(acrylic acid) (7). The composites were found to have restricted
polymer chain dynamics compared to bulk polymer (7).
Zeolites, also known as molecular sieves, are crystalline solids that contain highly
ordered pores. The framework of zeolites are usually made of oxides of Si and Al. The
pores may be empty or may contain water, ions, or molecules. As the moniker
'molecular sieves' would indicate, the pores are generally molecule sized, from
sub-Angstrom to a few nanometers in diameter. The pores are the result of the structure
of the highly crystalline oxides that make up the framework. As such they are almost
perfectly monodisperse. There are over a hundred types of zeolites, including natural and
synthetic varieties. Zeolites have been widely used as templates, including for such
substances as carbon (8, 9), polymers (10), semiconductors (11, 12) and metals (13). Due
to the sub-nanometer size of the template, the resulting materials can have extremely high
surface areas. The surface area of carbon templated in zeolites has been reported as high
as 2000 m2/g (8).
Pores in block copolymers are formed in a similar manner to those in glasses. The
block copolymer is made of two chemically distinct polymer types that phase separate.
One phase is then dissolved with an appropriate solvent. The size and shape of the
resulting pores are dependant on the molecular weight and concentrations of the polymer
types. In contrast to glasses, which have a random pore structure, the pores formed by
block copolymers can be highly ordered. This also is a function of the phase separation.
The three-dimensional network of pores in block copolymers has allowed them to be
used as templates for synthetic molecular sieves, allowing control over pore dimensions
(14, 15). They have also been used to template nanoparticles of magnetic oxides (16) and
nanowires of gold (17).
An interesting recent three-dimensional template is synthetic colloidal crystal or
synthetic opal. The crystals are made by physical close packing ofmicrospheres or
nanospheres, usually either latex or silica, from a solution. After or during packing, the
liquid is removed and a monomer solution is added. The monomer is then polymerized
and the original colloidal particles dissolved away (18). The distance between layers can
be controlled by the diameter of the colloids used. Materials templated in this way
include photopolymers (19) Ge (20) and titania (21). The resulting structures have been
shown to be photonic band gap materials (20, 21). This 'lost wax' replica of the crystal
can then be further used as a template in its own right. For example, the laser dye
rhodamine 6G was templated using a reverse opal and found to exhibit morphology
dependant resonances due to the confinement (19).
Two-dimensional templates are layered materials. Some examples include minerals,
Langmuir-Blodgett (L-B) films, and ionic polymer multilayers. Most of the layered
minerals used as templates, such as muscovite and montmorillonite, are oxides of Al and
Si, similar to zeolites. As with zeolites, the layers are the result of the molecular
structure. Polymer multilayers and L-B films, in contrast, are built up manually, one
layer at a time, usually by repeated dipping of a substrate into a solution containing or
supporting the layer. Examples of two-dimensional template synthesis include
organization of colloids on polyelectrolyte templates (22), and graphite synthesis in
montmorillonite clay (23).
The primary concern of this dissertation is template materials of one dimension.
These are typically membranes with pores running through the thickness of the
membrane. The pores may be naturally occurring or may be the result of extensive
preparation and engineering.
Carbon and lipid nanotubes are naturally occurring one-dimensional templates.
Nanomaterials that have been deposited inside carbon nanotubes include metals (24, 25),
metal oxides (26), and semiconductors (27). Lipid nanotubes have been used to template
nanoscopic structures ofpolypyrrole (28).
Another natural one-dimensional template material is aluminum oxide or alumina
(A1203). When grown under appropriate conditions, cylindrical pores form normal to the
surface of the membrane (Figure 1-1). The pores result from the electrochemical
oxidation of aluminum. Pore size is dependant on the voltage used for oxidation and can
be varied from -10 nm to -350 nm. The pores are highly monodisperse. Pore densities
range from 109 to 1012 pores/cm2. Alumina membranes with pores of nominally 200 nm
diameter are available from Whatman International, Maidstone, England. Nominal pore
sizes of 100 nm and 20 nm are commercially available as well. For sizes other than 200
nm, however, the stated pore size reflects less than 1% of the pore length. The other
99+% of the pore is -200 nm in diameter. Many groups, including ours, have used
porous alumina as a template material. A sampling of the materials that have been
prepared by template synthesis includes metals (29), polymers (30), lithium ion battery
materials (31), thermoelectric materials (32), carbon (33), and semiconductors (34, 35).
A recent interesting application is the use of alumina to template nanometer scale
metallic barcodes (36). To form the bar codes, the researchers electrochemically
deposited layers of metal in the alumina membrane pores. By varying the metal type and
layer thickness, a unique set of nanowires could be composed each time. The nanowires
Figure 1-1. Scanning electron micrographs of aluminum oxide. A) Side view of
homegrown 60 nm. B) Top view of commercial 200 nm (Whatman).
Figure 1-2. Scanning electron micrograph of track-etched polycarbonate membrane
with 5 pm pores.
are then freed by dissolving the template. Nanobarcodes can be mixed in with another
substance as a type of tracer or marker or used in bioassays (36).
In contrast to the above one-dimensional template materials, polymeric and mica
membranes have pores created by irradiation with radioactive decay particles (Figure
1-2). The high-energy decay particles cause damage tracks as they pass through the
polymer or mica. These damage tracks are etched much faster than nondamaged
material. Etchants are usually either acid or base. The resulting porous membranes are
said to be 'track-etched'. Pores produced by this method tend to be highly monodisperse.
In mica the pores have a diamond-shaped cross-section while in polymers the
cross-section is circular. In both cases they are predominantly normal to the membrane
surface although there is some deviation in angle. The distribution of pores is random for
the same reason. Pore size is dependant on etching time and ranges from -10 nanometers
to tens of microns. Porosity depends on the pore density and size of the pores but is
typically from 15% to 0.01%.
Track-etched mica was one of the earliest materials used as a one-dimensional
template. Possin (37) deposited nanowires of several types of metals including Zn, Sn,
and In as early as 1970. Other researchers later modified the technique to produce
nanowires of gold as small as 8 nm in diameter (38).
Track-etched polymer membranes have been widely used as templates by the Martin
lab (30, 39, 40) and others (41). Metal nanostructures have been synthesized by
electrochemical reduction (42) or an electroless process (39). Solution casting (30), melt
casting, and polymerization from a monomer solution (43) have all been used to template
polymeric materials in track-etched membranes. Examples of the polymers templated
include many types of insulating polymers (30) and conducting polymers (44, 45).
Template-synthesized nanofibers of conducting polymers have been shown to be more
conductive than the bulk material (44, 45). Polymers and metals have been templated
sequentially to form concentric composites consisting of an outer metal shell, a
passivating layer of insulating polymer, and an inner wire of conducting polymer (46).
An early application of template-synthesized nanomaterials entails preparing
nanoscopic electrodes. Research in our lab has shown that an electroless plating method
can be used to prepare ensembles of gold nanodisk electrodes with a disk diameter as
small as 10 nm. Such nanoelectrode ensembles (NEEs) have potential applications in
electroanalytical chemistry because the signal-to-background ratio (S/B) observed at the
NEE can be orders of magnitude higher than at a conventional gold macroscopic
electrode disk. Thus the detection limit at the NEE that can be orders of magnitude lower
than that for conventional macroscopic electrodes (39).
The NEEs were obtained by electroless plating of gold within the pores of the
template membrane for sufficiently long times so that solid nanowires were produced. If
plating is done for shorter times, ensembles ofnanotubes that span the complete thickness
of the template membrane can be obtained. These nanotubes can have inside diameters
of molecular dimensions, (<1 nm) and can be used to separate molecules on the basis of
size (47). By applying a bias to the gold nanotubes, the membranes can be rendered
cation permselective, anion permselective, or nonpermselective (48). Furthermore, the
inside of the nanotubes can be modified using thiol chemistry to create chemically
selective filters (40, 49).
An additional type of engineered one-dimensional template is nanochannel glass (50,
51). This differs from the Vycor glass discussed earlier in that the pores are all parallel to
each other, normal to the membrane surface, and highly monodisperse. Nanochannel
glass is created by bundling fibers of glass together, heating to soften and bind the fibers,
then pulling the softened collection of fibers to decrease their cross-section. The process
is repeated until there are as many as 101 fibers/cm2. Each fiber consists of a sheath of
high-silica-content glass and a core of high-boron-content glass. The high-boron-content
glass is then etched away with acid to leave an equivalent density of pores. Nanochannel
glass has been used to template ferromagnetic wires that show enhanced magnetic
coercivity (50), and to pattern mesoscopic ring structures of Au (51).
As cited earlier, aluminum oxide or alumina can be produced with highly
monodisperse cylindrical pores (Figures 1-1 and 1-3). This is a result of the
electrochemical oxidation process in acid electrolytes. (If a basic electrolyte is used, a
nonporous barrier oxide is formed) (52). In the growth process, high-purity aluminum
metal (99.99%) is used as the anode. A stainless steel cathode is used as the other half of
the cell. Both cathode and anode are immersed in aqueous acid electrolye and voltage is
supplied by a variable power supply. The geometry of cathode and anode is important
for uniform alumina growth. Our system uses a cylindrical cathode that surrounds the
aluminum anode to assure homogenous current density on both sides of the aluminum
plate (Figure 1-4).
Because the alumina film is converted directly from the aluminum metal, a rough
metal surface will produce a rough alumina film, albeit with the incorporation of pores.
I \ I
Scanning electron micrographs of homegrown aluminum oxide.
A) Aluminum metal. B) Barrier layer.
ikfe Jeb.'Jl. jb -0^
Electrochemical growth cell for anodic oxidation of aluminum.
To produce an even and flat template, highly polished aluminum metal is needed. The
aluminum may be mechanically polished if the surface is severely scratched or pitted.
Mechanical polishing is done with a slurry of either silicon carbide or alumina particles.
Larger particles (>10 urn) are used to remove material quickly, and polishing is continued
with slurries of progressively smaller particles until the particles are submicron in size.
The final polishing stage is done electrochemically. This process is analogous to that
of alumina film formation but it is done in concentrated acid electrolyte solutions and at
high temperatures to favor immediate dissolution of the alumina. The electropolishing
solution used in our lab is 95% concentrated phosphoric acid, 5% concentrated sulfuric
acid with 20 g/L Cr03 added to prevent pitting. The aluminum is again the anode and a
lead plate is used as the cathode. The temperature of the electrolyte is kept around 70C
and between 10 and 15 V is applied for periods of roughly 5 minutes. If the aluminum
has been previously polished to a submicron surface roughness, two or three
electropolishing steps will generally result in a mirror finish.
The rate of the oxidation reaction is also important to final film characteristics. If this
becomes too high during the oxidation process, the resulting heat will cause a runaway
reaction that will preferentially etch the aluminum at the air/electrolyte interface. This
will eventually result in loss of electrical connection to the aluminum. To keep the
reaction rate low, the electrolyte is chilled to between 0 and 4 degrees C.
Polyprotic acids are used as the electrolyte (usually sulfuric acid, phosphoric acid, or
oxalic acid). The type of acid used is dependant on the size of the pores to be grown.
Smaller pore sizes require low applied voltages and therefore require more highly
conductive electrolytes (e.g., sulfuric acid) (53). Larger pores need larger voltages that
would cause a runaway reaction in a highly conductive electrolyte. Thus large-pore
formation uses lower conductivity electrolytes (such as oxalic acid). The concentration
of the electrolyte is also important. Higher concentrations increase the current density for
any given voltage. This increase in current density results in an increased rate of film
formation but also enhances the possibility of a runaway reaction.
The most important parameter in determining the final pore size for the alumina film
is the applied voltage. When proper electrolyte choice and temperature control constrain
the current density between 1 and 6 mA/cm2, the pore size is observed to be -1.4 times
the applied voltage (54).
During the initial stages of film growth, the pores are somewhat random in shape and
packing, and there is some heterogeneity in size. This is a result of the pores nucleating
from random pits and valleys on the aluminum surface (55). As film growth continues,
the pores self organize into a hexagonal array and the pores become highly monodisperse
(56). A good review of the theory of this process can be found in the dissertation of G. L.
Homyak (54). To produce membranes with a "perfect" hexagonal packing and uniform
pore size, a two-step process was developed by Matsuda and coworkers (56) in 1995.
The first step in the process is growth of alumina film for sufficient time to allow
self-organization and homogenization of pore size. That film is then removed with acid.
This leaves an indentation or pit in the underlying aluminum corresponding to each pore.
The aluminum is then re-anodized at the same voltage and in the same electrolyte as the
original anodization. The pores nucleate in the pits and thus are already highly ordered
and monodisperse, and continue to grow as patterned (55, 56).
"Perfect" pores can also be formed by indenting the aluminum on a nanoscopic scale
preceding anodization. Matsuda and coworkers created a SiC stamp by electron beam
lithography and used this to pattern the aluminum surface. Pore formation again initiated
in the preformed indentations and gave rise to highly ordered, monodisperse pores (57).
Separation of the alumina film
There are two methods of removing the alumina film from the underlying aluminum
after growth is complete. The most conceptually simple process is to dissolve the
aluminum metal with HgCl2. This does not damage the alumina. This method works
best if thin foils of aluminum are being used to grow the alumina films. Thin films have
lower quantities of aluminum to dissolve, especially after anodizatoin. It should be noted
that the pores in alumina do not extend all the way through to the aluminum metal. Each
pore is separated from the metal by a thin barrier layer that is continuously forming and
dissolving as the film grows (Figure 1-3) (58). When the aluminum is dissolved, the
pores on the resulting alumina are closed on one end by this barrier layer. It can be
removed by acid or base etching, but as all of the alumina is exposed to the etchant this
may slightly widen the pores.
The other method of freeing the alumina from the aluminum involves a process called
voltage reduction (59). After the film reaches the desired thickness, the voltage is
reduced to about 70% of its original value. Because the pore size is determined by the
applied voltage, the pores at the barrier layer branch to smaller sizes. The current also
decreases but subsequently proportionally to the increasing voltage. The thickness of the
barrier layer, like the pore size, is proportional to anodization voltage. The voltage
reduction process is continued through many branching cycles until the barrier layer is
very thin and there are many small pores branching into it. Anodization is stopped, and
the alumina/aluminum composite is placed in an acid or base etchant solution. This
dissolves the alumina everywhere. However, the thin barrier layer and the small pores
are most quickly dissolved and the alumina is detached from the underlying aluminum
metal. The alumina film separates en masse if thick (above 20 mun) or flakes off in pieces
if thin. The resulting membrane has a branching pore structure that resembles a tree. A
silica replica templated from alumina demonstrating the structure of the barrier layer is
shown in Figure 1-5. The process used to make this replica is discussed in detail in
A membrane produced by voltage reduction has two distinct faces: the solution side
and the barrier side. Pores on the barrier side are always smaller than the solution side
due to the voltage reduction and pore branching. Voltage reduction can be done rapidly,
such that the branching constitutes only a fraction of the total pore length. For example,
in commercially available 20 nm alumina (Whatman), the pores branch from 200 nm to
20 nm over just 250 nm distance. Since each pore is 60 pm long, this represents only
0.42% of the overall pore length. If a membrane with uniform pores on both sides is
desired, the barrier side can be removed with acid, base, or reactive ion etching.
For several types of template synthesis, it is not necessary to separate the alumina film
from the underlying aluminum metal. The metal is often used to give mechanical support
to the membrane, as will be shown in Chapter 2. Although there is a barrier layer of
alumina preventing direct electrical contact, various materials have been electrodeposited
in alumina templates using AC deposition in which the attached aluminum metal acts as
the cathode. Examples include Ag (60)and CdS (61).
Transmission electron micrograph of silica templated in branched pore
Sol-gel technology originated in the 1970s as researchers attempted to find low
temperature routes to glass synthesis (62, 63). Glass is an amorphous material, however
it is generally made from crystalline oxide precursors. The high temperatures usually
needed to form glass (1300 to 2000* C) are a result of the need to destroy the crystallinity
of the precursors. A method using noncrystalline precursors was sought. As glass is
mostly SiO2, liquid alkoxysilanes were used. It was found that alkoxysilanes hydrolyze
readily in the presence of water to form silanols:
R'3Si-O-R + H20 -- R'3Si-O-H + R-OH (1-1)
The silanols can then undergo polymerization reactions with other silanols or with
R'3Si-O-H + HO-SiR'3 -4 R'3Si-O- SiR'3 + H20 (1-2)
R'3Si-O-H + RO-SiR'3 -> R'3Si-O- SiR'3 + ROH (1-3)
In either case, for alkoxysilanes containing two or more alkoxy groups, the result is
formation of a three-dimensional siloxane network. At the start of polymerization, many
small particles of siloxane are formed. These are well dispersed in the liquid phase to
form a colloid. When the particles are well isolated from each other, the density of the
suspension resembles that of the solvent. At this stage the colloid is called a sol. As
polymerization continues, the particles increase in size and the liquid increases in
viscosity. When polymerization has progressed to the point where the particles develop a
three-dimensional network throughout the liquid, the colloid becomes very viscous and is
termed a gel (64).
Several parameters influence the rate of the hydrolysis and polymerization reactions.
These include temperature, pH, water concentration, and type of alkyl group (65). An
increase in temperature increases the rate of gelation. The reactions can be either acid or
base catalyzed; extremes of either high or low pH will speed the reaction, as will
increasing the amount of water (66). A larger, more sterically bulky alkyl group slows
the reaction rate (66).
Gels can be converted to silica through two routes. In the first method, the gel is
heated or placed under vacuum to remove the solvent phase, typically alcohol, as well as
water and other alcohols produced during polymerization. The resulting surface tension
collapses the open three-dimensional network of the silica phase of the colloid,
condensing it to a dense phase called a xerogel (65). If the solvent and water are
extracted instead through a critical-point drying procedure, the surface tension is
eliminated and the open porous nature of the gel is maintained (67). This type of
monolith is termed an aerogel. The maximum temperature for either process can be kept
under 100'C (66).
Since its inception, this process has been extended from silica to many types of metal
oxides (68). Because at the sol stage the colloid has liquid properties, it can be molded or
dip coated into nearly any size or shape, including nanomaterials. On dip coating, a gel is
formed at the interface of the material immersed in the sol. This gel layer is compacted
by drying to form a dense layer that replicates the surface topology of the immersed
template (65). Sol-gel chemistry has been used to fabricate template-synthesized
nanostructures of TiO2 for photocatalysis (34), V20s as a lithium ion battery material
(69), and WO3 (70).
As mentioned above, the surface properties of a material become increasingly
important as the size decreases. It is therefore useful to have a method for controlling or
modifying the surface chemistry of materials. One relatively simple method for
modifying surface chemistry is the use of organosilanes. These are compounds
containing an Si-C bond. Because Si is in the same chemical group as carbon, these
compounds share much of their chemistry with organic compounds. Organosilanes form
stable covalent bonding with siliceous materials (e.g., silicates and aluminates) and many
metal oxides (71). The predominant types of organosilanes used for surface
modifications are chlorosilanes and alkoxysilanes. Both are usually deposited from
alcohol solutions. The attachment chemistry of both types is equivalent because when
chorosilanes are dissolved in alcohol, they react with the alcohol to form alkoxysilanes
and HCl (71).
R'3Si-Cl + ROH R'3Si-OR + HCI (1-4)
The extent of this reaction can be monitored by measuring the pH.
Alkoxysilane chemistry is analogous to the chemistry of sol-gel production. Silica
sol-gel precursors are organosilanes, generally either tetramethoxysilane (TMOS) or
tetraethoxysilane (TEOS). Due to its high vapor pressure and high reactivity,
tetrachlorosilane is used for vapor phase reactions to produce fumed silica. All of these
compounds have four reactive sites and thus can form three-dimensional networks. The
reactions are analogous to Equations 1-1 to 1-3, although in the chlorosilane case, the
chlorine is the leaving group.
If only a monolayer of surface modification is desired, an organosilane is used which
has only one reactive chloro- or alkoxy- substituent. Because there is only one reactive
site, these molecules can either dimerize or bind to the surface. Dimers cannot bind
further and are rinsed away. Binding to the surface takes place via surface hydroxyl
groups through the reaction
R'3Si-OH + R-O-SiR'3 -> Si-O-SiR'3 + ROH (1-5)
Often, a higher degree of surface modification is desired. To improve the density of
coverage, silanes with two or, more commonly, three reactive groups are used. These are
first allowed to oligimerize in a slightly aqueous alcohol solution (71). The presence of
water, typically around 5% vol/vol, initiates the formation of silanols (Equation 1-1)
increasing the reaction rate. The pH is adjusted to between 4.5 and 5.5 with acetic acid to
further facilitate the substitution reaction of the alkoxysilanes (71). The material to be
surface modified is then added to this solution and the oligimers bind through surface
hydroxyl sites (Figure 1-6). Surface modification by this route generally requires only a
few minutes of immersion. If a 2% solution of trialkyoxy- or trichlorosilane is used and
five minutes is allotted for oligimer formation, the resulting surface modification is
normally from 3-8 monolayers thick (72). After solution deposition, the silanized surface
must be cured. This increases cross-linking and removes residual solvent. It can be done
by heating to 120C for 20-30 min, or for longer times at lower temperatures. At room
temperature, 24 hours is generally sufficient for full curing.
Chlorosilanes can also be deposited from aprotic solvents, such as toluene and
tetrahydrofuran (71). If these solvents are anhydrous (and the substrate is free of bound
water), there will be no chance for alkoxysilanes to form. There can therefore be no
polymerization of the silane. The reaction must proceed by nucleophilic attack from
surface hydroxyl sites. As a result, only a monolayer can form (71). Surface
modification by this route requires longer times, from 12-24 hours. Higher temperatures
ROi-OR RO- OR
R0 jifOR + RO- O
RO-- i---- iO--i-OR
Trialkoxysilane oligier formation and attack ent to surface.
Trialkoxysilane oligimer formation and attachment to surface.
or refluxing are also usually required, however the process does not need further curing
In Chaper 2, a sol-gel template synthesis process is used to produce silica micro- and
nanotubes from alumina templates. The nanotubes can be modified with a variety of
chemistries, typically via a silanization process. Because the nanotubes are formed in a
template, the interior and exterior surface can be modified independently. For example,
the interior of the nanotubes can be modified with fluorophores while the exterior is
modified independently to make the nanotubes hydrophilic or hydrophobic. Modified
nanotubes can be used for drug detoxification or as extractants for the removal of metal
ions. The nanotube surface can also be biotinylated, which causes binding to avidinated
surfaces. Composite microtubes of silica and various polymers are also prepared.
Additionally, Au nanowires are shown to assemble with colloidal Au particles using
dithiols as linkers. The position of linkage can be controlled by using the template as a
Chapter 3 describes the attachment of proteins onto template synthesized micro- and
nanotubes. The proteins are covalently linked via an aldehyde silane bridge that binds to
pendant primary amino moieties. Because most proteins have pendant primary amines, a
wide range of proteins can be used to modify the nanotubes. Protein-modified nanotubes
function as highly specific extractants. Avidin-modified nanotubes extract biotin coated
Au nanoparticles from solution with high extraction efficiency. Immunoprotein-modified
nanotubes extract the corresponding antibody from solution with high specificity.
Antibody-modified nanotubes extract one enantiomer from a racemic mixture. Enzymes,
including drug detoxification enzymes, are also attached to the nanotubes and are shown
to retain their catalytic activity. Nanotubes modified in this way function as removable
bioreactors. Immunoproteins on the outside of nanotubes can be used to direct nanotube
binding, creating specific labeling agents.
Chapter 4 describes the production of synthetic microporous membranes that mimic
ligand-gated ion channels. The membranes are initially 'off, that is, they do not allow
molecules or ions to pass. On addition of analyte, the membranes switch to an 'on' state
and become permeable to molecules and ions. The membranes are manufactured from
microporous alumina that is silanized with a C18 silane. The hydrophobic C18 chains
prevent aqueous solutions from entering the pores. Ionic surfactant molecules act as a
chemical stimulus to turn the membranes 'on'. The hydrophobic portion of the surfactant
partitions into the C18 layer, while the ionic portion of the molecule becomes the
prevalent chemistry at the pore wall. Water can then flood the now ionic pores and
molecules and ions can cross. The effect of concentration and hydrophobicity of the
surfact analyte species are examined, as is the effectiveness of several drug compounds as
analytes. The results and conclusions of this dissertation are summarized in Chapter 5.
CHEMICAL MODIFICATION OF SILICA NANOTUBES
In this chapter, a sol-gel template synthesis process is used to produce silica micro-
and nanotubes within the pores of alumina templates. These silica nanotubes can be
modified via silanization with a variety of chemical functional groups. The nature of the
template synthesis process allows independent modification of the inner and outer
nanotube surfaces. Several examples of independent modification are given. Nanotubes
modified with C18 on the inside and unmodified on the outside are effective at sorbing
drugs and drug analogs from aqueous solution. Modifying the nanotubes with chelating
agents makes them useful as extractants of metal ions. Template synthesized structures
can also be used as building blocks for the assembly of larger structures. For example,
silica microtubes can be biotinylated, which causes binding to avidinated microspheres.
Additionally, Au nanowires are shown to assemble with colloidal Au particles using
dithiols as linkers. Composite microtubes of silica and various polymers are also
A recurrent theme in the recent literature involving nanoparticles concerns their use as
drug delivery agents (73-76). This process is the inverse of using nanoparticles as
extractants, which will be a primary subject of this chapter. Because of the similarities, a
review of the basic tenants of nanoparticle drug delivery is instructive. The use of
macroscopic particles as extractants is also reviewed.
There is increasing interest in using nanoparticles for drug delivery (73-76). Spherical
nanoparticles are almost always used because that shape is easy to make. The term
'nanoparticle' as used in the drug delivery literature encompasses everything from 5 nm
to 5 pm diameter. There are a few biologically important size parameters. Particles
below 200 nm in diameter are less susceptible to clearance by the reticulo-endothelial
system (RES), which increases their dwell time in the bloodstream and thus their efficacy
(77). Particles larger than 5 pm diameter will not be able to pass through capillaries and
will therefore not be well distributed by the circulatory system (77). Particles less than
100 nm in diameter are said to be 'ideal' in that they avoid RES uptake and facilitate
access to cells and tissues (77).
Types of nanoparticles for drug delivery
Most nanoparticles used for drug delivery are self-assembled from block copolymers
(27, 77, 78). Polymers are formed by linking together small monomer molecules. A
homopolymer is formed if only one type of monomer is used. When two different types
of monomers are joined in the same polymer chain, the result is called a copolymer (79).
Two types of monomer molecules can be made into a copolymer in many different
ways. If the two types of monomer are labeled A and B, the copolymer could be formed
by alternating A and B units:
This type of copolymer is called an alternating copolymer. If the monomer units are
arranged in a random fashion, the copolymer is called a random copolymer. An example
would be as follows:
In a block copolymer, the monomers of each type are grouped together. A block
copolymer can be thought of as two homopolymers joined together (80):
Block copolymers used for drug delivery usually consist of either two or three
components (77). For a two-component system, one block is hydrophilic and the other
hydrophobic. In a three-block system, the scheme is
hydrophilic-hydrophobic-hydrophilic. These systems form micelles in a manner
analogous to that of surfacts. In aqueous systems, the hydrophobic regions cluster
together to form the interior or core of the micelle, while the hydrophilic regions create
an outside shell or corona.
Poly(ethylene oxide) (PEO), or the related polymer poly(ethylene glycol) (PEG), are
often used as the hydrophilic component (77). Poly(ethylene oxide) and PEG have nearly
unlimited water solubility, and therefore a high degree of hydration and large excluded
volume (81). These properties help to stabilize the micelles and prevent protein
adsorption (77). The type of hydrophobic polymer used varies widely. For a partial list
of biocompatible hydrophobic polymers used for micelle formation see Table 1 of
There are two typical routes to micelle formation, dissolving the block copolymer
directly in water or other aqueous solvent (e.g., buffer), and dialysis (77). The easiest
method is direct dissolution, in which the block copolymer is simply dissolved in the
aqueous system at a concentration high enough to cause spontaneous formation of
micelles. This concentration is known as the critical micelle concentration (CMC).
However, in many cases the block copolymers are not soluble in water. If they are not
soluble, the polymer is dissolved first in a water miscible organic solvent or
organic/aqueous mix, and then dialyzed against water or heated until the organic solvent
is removed. Again the micelles form spontaneously. For drug delivery applications, the
core of the micelle is hydrophobic and the exterior shell or corona is hydrophilic (77).
It is generally important that the working concentration of the block copolymer (that
is, the concentration in the blood after injection) be kept above the critical micelle
concentration (CMC) or the micelles may degrade (77). This can be avoided by cross-
linking either the core or the corona polymer. However, if the ratio of the hydrophobic
core material to the hydrophilic corona is very high, micelles can be kinetically stable
over long times even below their CMC (77). This is also true if the core material is well
below the temperature at which there is easy movement of the polymer chains, known as
the glass transition temperature (Tg), or if it is highly crystalline (77).
The size of the block copolymer micelle is influenced by the lengths of the core and
corona blocks (77). The size of the core generally determines the amount of drug the
micelle can carry. A typical core volume is only about 0.5% of the overall micelle
volume (77). The amount of drug a micelle can hold is also dependant on the
hydrophobicity of the drug, the drug size, and the physical state of the core (crystalline
vs. liquid-like) (77). These factors also influence how quickly the drug is released from
the micelle, as do other factors such as the density of the corona and biodegradation rates
of the polymers (77).
Non-micellar nanoparticles can be prepared by polymerization of monomer in the
presence of a stabilizing agent. Polymers preferred for this are usually hydrophobic and
biodegradable. Examples include poly(esters) (82), poly(alkylcyanoacrylates) and
analogs of poly(methylidenemalonate) (83). Stabilizers are usually surfactants or
carbohydrate polymers (e.g., dextrans). Nanoparticle size is often controlled by
templating the polymer inside the stabilizer micelle or in the oil phase of an oil-in-water
Dr. Karen Wooley and coworkers have modified block copolymer micelle formation
by cross-linking the corona (shell) after assembly. The nanoparticles can have liquid-like
or solid-like interiors, depending on the Tg of the hydrophobic polymer. For example,
poly(styrene) (Tg = 105"C), yields solid nanoparticles, while poly(isoprene) (Tg = -75C)
cores are liquid (78). The cores can also be comprised of a degradable material, which
can be subsequently removed to leave a nanocage structure (84). The outer hydrophilic
shells are usually poly(acrylic acid) with multiple functional sites to allow cross-linking
The other major synthetic route to nanoparticles for drug delivery is dendrimer
chemistry. Compared to polymer micelles, dendrimers are very highly monodisperse and
have precisely defined chemistries. Research in this area is typified by the work of
Frechet, who has synthesized amphiphilic dendrimers hydrophilicc PEG exterior,
hydrophobic phenolic core), and demonstrated their ability to incorporate and release
model drugs, notably phenol (85).
Micelles formed from biological macromolecules such as proteins, lipids, and nucleic
acids are called coacervates. Like the micelles mentioned above, these form
spontaneously in aqueous systems as a result of aggregation of hydrophobic regions of
the molecules. Like block copolymers, coacervates are often used as a delivery vehicle.
Because the macromolecules from which they are made are ionic, the coacervates are
also ionic in character. They typically have a densely charged ionomer core and a more
sparsely charged ionomer corona of the opposite sign (86). The inability of close ion
paring caused by the different charge densities causes the sparsely charged ionomer to
extend into solution, which stabilizes the particles (86). Because of their strongly ionic
nature, they are not often used as drug carries, as most drugs are hydrophobic. Their
niche is rather the transport of charged species which can be part of the coacervate, for
example DNA (87). One 'smart' example of this is discussed in Chapter 3.
Advantages of using nanoparticles for drug delivery
Paul et al. found that using nanoparticles to deliver drugs increased the drug efficacy
versus the free drug (88). They used pentamidine, an antileishmanial drug to treat
Leishmania-infected mice. (Leishmania are protozoan parasites, usually spread by sand
flea bites. The disease is most common in India and Bangladesh). The efficacy was
measured by the amount of drug needed to inhibit the disease. The micelles were 130 to
150 nm in diameter. When they used poly(D,L-lactide), which is biodegradable, as the
core polymer they observed a three fold increase in drug efficacy. When they used
poly(methacrylate) as the core, the efficacy increase was even greater- 6.5 times the free
drug, but this polymer is not biodegradable (88).
Nanoparticles can reduce drug toxicity. Perkins et al. made nanoparticles of 50 to 100
nm diameter using drug molecules as the core and PEG-conjugated lipids as the corona.
They found that administering Taxol (an anticancer drug) in this way to treat ovarian
cancer in mice resulted in far less toxicity than intraperitoneal (shot) or intravenous (IV)
administration of the free drug (89).
Encapsulation in nanoparticles can also protect fragile drugs from severe
gastrointestinal conditions and allow alternate means of treatment. Carino et al.
encapsulated insulin in nanoparticles of poly(lactide-co-glycolide) which were then taken
orally. The nanoparticles protect the insulin from enzymatic degradation in the intestinal
tract, and then cross the intestinal epithelium and release the insulin into the bloodstream
(90). The particles produced in this study were widely variable in size, from 100 nm to 5
um. Since only particles smaller than about 4 um can pass the intestinal epithelium, and
since the percentage of passage increases with decreasing size, most of the effective
insulin was probably encapsulated in the smaller particles.
Engineered Sorbants and Extractants
Macroscale engineered sorbants have proven useful in a diversity of applications
(91-93). For example, the addition of hygroscopic promoters such as CaCl2 and LiBr to
silica solutions improves water adsorption of the resulting material (91). Composites of
SiO2-CaCI2 and SiO2-LiBr prepared by a sol-gel process have been shown to more than
double the amount of water absorbed versus conventional zeolytes and silica gels (91).
Other examples of engineered bulk silica sorbant materials include solgel prepared
optical chemical sensors (93) and metal ion selective sorbants for drinking water
purification (92). There are limited examples ofnano-sized particles specifically
engineered as sorbants and extractants. Radioactive Cs can be removed from waste
materials using nanoporous silica silanized with Cs-specific binders (94). Polyurethane
nanoparticles have been used to remove a model pollutant from soil (95).
In this chapter, we build on these prior systems to synthesize silica based nanotube
systems that can be engineered to selectively extract a variety of species from solution.
Commercial Anopore alumina membranes used as templates were obtained from
Whatman (Clifton, NJ). These were disks 47 mm in diameter with cylindrical pores of
nominally 200 nm diameter. Alumina filter membranes of nominally 20 nm pore
diameter were purchased from the same source. All other sizes of templates were grown
in house by anodic oxidation of high purity aluminum (see Chapter 1). Unless otherwise
noted, chemicals were reagent grade, obtained from commercial sources and were used as
received. Water was 18 MOhm pure from a Barnstead Epure model D4641 system.
UV-VIS spectra were obtained using a Hitachi U-3501 spectrophotometer.
Inductively-coupled plasma (ICP) analysis was done on a Varian Vista RL ICP-AES.
Scanning electron microscopy (SEM) was performed using a Phillips 505 scanning
electron microscope. SEM samples were made conductive by sputtering a thin layer of
Au (-15 nm), using a Hummer model 6 sputter coater. Transmission electron
microscopy (TEM) was performed on a JEOL model 200CX TEM.
A sol-gel template process was used to prepare the nanotubes (70). See Chapter 1 for
a review of the chemistry involved. First, a sol-gel silica precursor was prepared by
mixing absolute ethanol, tetraethylorthosilicate (TEOS, Figure 2-1A), and 1 M HCI
(50:5:1 vol/vol). This solution was allowed to hydrolyze for 30 minutes. Alumina
template membranes were then immersed into the sol-gel with sonication for 1 minute,
after which they were air dried for 10 minutes at room temperature and oven cured
overnight at 1500C. To produce plain silica nanotubes, the sol-gel treated membranes
were polished on both surfaces with an aqueous aluminum oxide slurry of-lum diameter
particles. This removed the surface layers of SiO2 that would otherwise bind the
nanotubes together. The membranes were then immersed overnight in a 25% (wt/wt)
solution ofH3P04 to dissolve the aluminum oxide template. The nanotubes were filtered
from the acid solution and repeatedly rinsed with pH 7.0 buffer and water, then dried. An
SEM micrograph of silica nanotubes templated in commercial 200 nm alumina can be
seen in Figure 2-2. In Figure 2-3, an alumina membrane produced in house with a pore
diameter of 30 nm was used as the template.
Structures of all silanes used for surface modification are shown in Figure 2-1. Cls
functionalized surfaces were prepared by immersing either the SiO2-coated alumina
membrane or freed nanotubes in a 5% aqueous, 5% octadecyltrimethoxysilane (Aldrich,
Figure 2-1B) solution (vol/vol) in ethanol, pH adjusted to 5.0 with acetate (72). The
deposition solution was stirred for 20 minutes before addition of the membrane to allow
formation of siloxane oligimers. Silane deposition was usually done overnight to insure
maximum coverage of the C18. Following deposition, the membranes were rinsed with
ethanol and dried 20 minutes at 150C to cure the silane layer. The silicon dioxide
surface layers of the membrane were removed by mechanical polishing with a slurry of
-1 um alumina particles and the template alumina was dissolved with a 25% wt/wt
solution of phosphoric acid. The nanotubes were filtered and rinsed, as above.
Modifications with N-triethoxysilylpropylquinineurethan (Gelest, Morrisville, PA,
Figure 2-1C), N-triethoxysilylpropyldansylamide, (Gelest, Figure 2-1D),
2-[methoxy(polyethyleneoxy)propyl]trimethoxysilane (Shearwater, Huntsville, AL,
Figure 2-1E), or 3-mercaptopropyltrimethoxysilane (Sigma, Figure 2-1F), were
Structures of silanes used for modifications.
G) N-(trimethoxysilylpropyl)ethylenediamine triacetic acid, trisodium salt
Na OOC COO Na
Na OOC bH2
Figure 2-1. Continued
analogous. Each was deposited from an ethanolic solution containing 5% (wt./vol.) of
the silane. The deposition solution was also 5% in distilled water by volume and was
adjusted to pH 5.0 with sodium acetate. In all cases the silane solution was allowed to
polymerize 20 minutes before addition of the membrane. Silane deposition was
accomplished by immersing the SiO2 coated template into these solutions, followed by an
ethanol rinse and heat fixing at 140C for 20 minutes. Depositions were usually
performed overnight to maximize coverage. The modified nanotubes were released from
the template by dissolving with 25% H3PO4, filtered from the acid solution, and rinsed
repeatedly with water.
A L w-
Figure 2-2. Scanning electron micrographs of silica nanotubes templated in
commercial 200 nm diameter alumina membranes. A) Nanotubes held
together by a film of silica. B) Dispersed nanotubes.
Figure 2-3. Scanning electron micrographs of silica nanotubes templated in 30 nm
diameter alumina membranes produced in house. A) Nanotubes held
together by a film of silica. B) Dispersed nanotube.
EDTA-functionalized nanotubes were prepared in a similar manner but from an
aqueous silane solution. The EDTA silane, N-(trimethoxysilylpropyl)ethylenediamine
triacetic acid, trisodium salt (Figure 2-1G), is not soluble in alcohol and is supplied as a
50% solution in water (Gelest). This solution is added to water 10:90 vol/vol, which
makes the final silane concentration of the deposition solution ~5%. The pH is adjusted
to 5.0 with sodium acetate/acetic acid. Silica nanotubes produced by template synthesis
were added to this solution and stirred 3-4 hours. The nanotubes were then filtered by
vacuum filtration, rinsed with water and dried. Curing of the silane was accomplished by
heating in a 120' oven for 25 minutes.
Extraction experiments were generally performed using either 2.5 or 5.0 mL of
solution. Unless otherwise noted, the concentration of nanotubes added was ~1.0 mg
nanotubes per mL solution.
Protein functionalized SiO2 nanotubes were prepared from the unmodified nanotubes
via an aldehyde terminated siloxane linker (96). (See Chapter 3 for a detailed discussion
of protein attachment). Unmodified nanotubes (as prepared above) were stirred in a 5%
aqueous, 10% aldehyde methoxysilane (PSX1050, United Chemical Technologies,
Figure 2-1H) in ethanol, pH adjusted to 5.0 with acetate. The aldehyde-modified
nanotubes were filtered, rinsed with ethanol, and cured 24 hours in an oxygen-free glove
box at room temperature. After drying, the nanotubes were incubated overnight at 4C
with a solution of protein in buffer (-1 mg protein/ml buffer). Aldehydes have been
shown to react with pendant primary amines on proteins to covalently link the proteins to
the substrate (97-99). Protein-modified nanotubes were filtered from the protein solution
and extensively rinsed with buffer before use. Unless otherwise noted, proteins were
obtained from Sigma in the highest purity possible and used as received.
The nanotubes were biotinylated using a multistep modification process (see Figure
2-4). An amino silane was first deposited as a coupling agent. A 2% (wt./vol.) aqueous
solution of 3-aminopropyltriethoxysilane (United Chemical Technologies, Figure 2-11)
was stirred 20 minutes to allow oligimer formation. Silica nanotubes from commercial
templates were added to this solution and sonicated briefly to suspend. The nanotubes
were stirred in solution for a further 60 minutes, removed by filtration, and rinsed with
water. Heating to 120 for 20 minutes cured the aminosilane layer. The amino-modified
nanotubes were then added to 5 mL of 10 mM sulfosuccinimidyl
6-(biotinamido)hexanoate (Sigma) in N,N-dimethylformamide in water and again
sonicated to suspend. Sonication was continued for 1 hour. The nanotubes were
-OH + H3CH2CO-i^/"\ NH2
0 Na O0 0 H
i/ NH2 + 03 N-
00 H 0
Figure 2-4. Schematic ofbiotinylation process.
removed from the reaction solution by vacuum filtration. They were rinsed with water
and then with pH 8.0 phosphate buffer.
Avidinated silica microspheres in buffer were purchased from Bangs Labs and used as
received. Two sizes were purchased, 5.0 gm and 0.5 lim diameter. Both sizes were
stored at 4'C until used.
Composite nanotubes of two types were prepared: polymer exterior with a silica
interior, or a silica exterior with a polymer interior. The ordering of the silica versus the
polymer layer was determined by order of deposition, with the outermost material being
the first deposited into the template. If the silica layer was to make up the outside of the
nanotube, it was deposited as detailed above. Inner layers of silica were deposited after
polymer deposition from the same sol-gel precursor solution. The cure, however, was
modified to 70C for 24 hours to prevent oxidation of previously deposited polymer.
Insulating polymers were deposited by dip coating the template membrane into a
polymer solution and allowing it to air dry. Poly(ethylene-vinyl acetate) (PEVA), was
dissolved in toluene to make a 0.5% (wt./vol.) solution. The PEVA was 40% by weight
vinyl acetate and had an average molecular weight of-100,000 Da. Poly(methyl
methacrylate) (PMMA) was prepared as a 4% (wt./vol.) solution in chloroform. The
molecular weight of this polymer was ~35,000 Da. Both polymers were supplied by
Scientific Polymer Products (Onterio, NY). Residual solvent was driven off by heating
the membrane to 70C for 15 minutes. The conducting polymer poly(pyrrole) was
deposited chemically from an aqueous 0.2 M monomer solution. The pyrrole monomer
(99%, TCI, Portland, OR) was twice distilled under argon and stored at -20C prior to
use. The template was immersed in 20 mL of the aqueous 0.2 M monomer solution and
an equivalent volume of a solution containing 0.5 M FeC3 and 0.5 Mp-toluenesulfonic
acid was added, which oxidatively polymerized the pyrrole monomer (100). This
reaction was done at 4' C. The alumina templates were removed with 25% phosphoric
acid and the nanotubes were collected by filtration.
Biotinylated Au nanoparticles of several sizes were supplied by Sigma. Other Au
nanoparticle colloid solutions were produced in house by reduction of hydrogen
tetracholoraurate (101, 102). To make the Au colloid, 300 mL of 0.5 mM HAuC14'3H20
was heated to boiling. 15 mL of boiling 1% (wt./wt.) sodium citrate dihydrate was
added. The combined solution is boiled a further 10 minutes, then allowed to cool to
room temperature. The resulting citrate-stabilized Au nanoparticles have an absorbance
maximum at 519 nm and average 15 nm in diameter, as measured by TEM.
Results and Discussion
Differential Inside/Outside Modification
Perhaps the most important attribute of the nanotubes is their distinct inner and outer
surfaces, which can be independently chemically and biochemically functionalized.
Furthermore, the template method provides a very simple way to differentially modify
the inner vs. outer surfaces. This concept is illustrated in Figure 2-5. While still
embedded within the pores of the template membrane, the inner surfaces are reacted with
one type of silane to attach this chemistry to the inner surfaces. The outer surfaces are
masked because they are in contact with the pore walls of the template membrane. The
amphoteric alumina template is then dissolved with acid or base to liberate the nanotubes.
These are collected by filtration, resuspended and reacted with the second silane. This
second silane attaches to the now accessible outer surfaces of the nanotubes. Several
experiments were conducted to prove this concept.
As conclusive proof of independent inside/outside modification, we modified the
exterior nanotube surface chemistry in a way that would control nanotube partitioning in
an organic/aqueous two-phase system. One batch of nanotubes was labeled on their inner
A) Cross-section ofAl203 template membrane showing two pores.
B) Sol-gel synthesis of SiO2 nanotubes.
J E 0
C) Attach first silane to inner tube surfaces.
D) Remove surface films on faces of membrane
and dissolve template membrane.
E) Attach second silane to outer surfaces.
Figure 2-5. Schematic of inside/outside modification of nanotubes.
surfaces with the fluorescent silane N-triethoxysilylpropylquinineurethan while still in
the alumina template. The outer surfaces of these nanotubes were protected by the
template and were therefore not silanized. After silanization, the template was dissolved
and the bare silica outer surface of the nanotubes was exposed. These nanotubes were
added to a vial containing 10 mL of water and 10 mL of the immiscible organic solvent
cyclohexane. The solvents were mixed by shaking the vial and then allowed to separate.
Because silica is naturally hydrophilic, when these nanotubes are added to the
cyclohexane/water two phase system, the pale blue fluorescence of the quinineurethan
(445 nm) is only seen from the lower aqueous phase (Figure 2-6A). The fluorescence at
the top of the cyclohexane phase is due to reflection of light from the meniscus. This
result confirms that exterior chemistry can be modified independent of interior chemistry.
If the initial quinineurathane silanazation were not confined to the inner nanotube
surface, the hydrophobic nature of the fluorophore would cause partitioning into the
A separate batch ofnanotubes was prepared with the fluorescent silane
N-(triethoxysilylpropyl)dansylamide attached to their inner surfaces. This fluorophore
emits in the green region (525 nm) when excited by UV light. The template membrane
was then dissolved and the outer surfaces were derivatized with the very hydrophobic
octadecyl silane (Cls). The nanotubes were added to an identical cyclohexane/water
system which was shaken and allowed to phase separate. The fluorescence image in
Figure 2-6B shows that because these nanotubes are hydrophobic on their outer surfaces,
they partition into the upper, hydrophobic cyclohexane phase. Hence, the pale green
fluorescence from the dansylamide is only seen from this phase.
A B C
Figure 2-6. Fluorescently modified nanotubes in biphasic cyclohexane/water mix
(cyclohexane on top). A) Quinineurathane inside/bare silica outside
nanotubes partition into water phase. B) Dansylamine inside/C18 outside
nanotubes partition into cyclohexane phase. C) Mix of both types
independently phase separate.
When both sets of nanotubes are added to the solvent mixture in the same vial, the
tubes with the Ci8 outer surface chemistry partition into the cyclohexane and the tubes
with the silica outer surfaces go to the aqueous phase (Figure 2-6C).
Quinine/Au nanoparticle nanotubes
We first created nanotubes with a quinineurethane silanized interior and an
avidinated/biotin-Au nanoparticle exterior. Details of the modifications are given in the
experimental section. The interior silanization with quinineurethane was done while the
nanotubes were still in the alumina template. For this reason, the quinineurathane must
attach on the interior surface only. The outer nanotube surface is masked from
silanization by contact with the alumina template. In this experiment, 200 nm
commercial alumina templates were used to make the nanotubes. After quinine
silanization, the alumina template was dissolved in acid. This did not affect the attached
quinine, which showed both absorption and fluorescence following the acid treatment
0 I I- I
250 300 350 400 450 500 550 600 650
Figure 2-7. UV-Vis spectra of modified silica nanotubes. A) Nanotubes modified
inside with quinineurathane and outside with avidin. B) The same
nanotubes after incubation with biotinylated Au nanoparticles.
Freed nanotubes were then silanized with an aldehyde silane (96). Because the
interior had already been functionalized, this attachment is posited to be preferential to
the nanotube exterior. Avidin was incubated with the aldehyde modified nanotubes
overnight. Avidin is a tetrameric 68 kDalton protein found in egg whites (103-105).
Each monomer unit is essentially identical and has a molecular weight of 17 kDaltons
(103). Incubation resulted in Shiffbase formation between the aldehyde and pendant
primary amino groups on the avidin, covalently binding it to the nanotube exterior (96).
Details of the protein attachment process are given in Chapter 3.
Avidin is well known to bind the small molecule biotin (104, 105). The binding
constant for the reaction is among the highest known for protein systems, -1 x 1015 NM"
(105). Because avidin is a tetramer, it can bind up to four molecules of biotin (103). To
be able to monitor the biotin binding to the nanotubes, 20 nm diameter biotinylated Au
nanoparticles (Sigma) were used. As mentioned in Chapter 1, nanoparticles of gold have
extremely high molar extinction coefficients due to their surface plasmon resonance
(106). The absorbtion maximum for the Au nanoparticles is 540 nm, well separated from
the 340 nm maximum absorbance of the quinineurathane.
Biotinylated Au nanoparticles were incubated with the nanotubes and bound to the
attached avidin. A UV-Vis spectrum of the resulting dual-functionalized nanotubes
shows two distinct absorption peaks, one at 340 nm due to the interior bound
quinineurathane and another at 540 nm caused by the exterior bound Au nanoparticles
(see Figure 2-7). The above results are encouraging, although not conclusive proof that
the Au nanoparticles are attached only to the exterior surface. To further prove this
concept, a control experiment was attempted in which both the inside and outside
surfaces of the nanotubes were modified with quinineurathane. This was done by
silanizing after template dissolution. It was then attempted to silanize these
quinineurathane-modified nanotubes with aldehyde silane and incubate them with
protein, but this proved impossible as the nanotubes would not suspend in aqueous
solutions. This result increases the probability that avidinization, and the subsequent
binding of biotinylated Au nanoparticles, occurs on the outer surface of the nanotubes.
Au nanoparticle/fluorescent protein nanotubes
Another example of independent inside/outside modification is the preparation of
nanotubes with Au nanoparticles on the inside and FITC-labeled antibodies on the outer
surface. As previously, we began by silanizing the nanotubes while still in the template.
In this case, the nanotubes were functionalized with a mercaptosilane. See experimental
section for details.
400 500 600 700 800 900
Figure 2-8. UV-Vis spectra of modified silica nanotubes. A) Nanotubes modified
inside with mercaptosilane to which Au nanoparticles have been bound.
B) The same nanotubes after outside modification with FITC-labeled
Citrate stabilized Au nanoparticles (diameter = 15 nm) prepared in our lab were bound
by the mercapto group to the inside of the nanotubes while still in the template. The
template was then removed with acid. Acid dissolution did not remove the Au
nanoparticles, as evidenced by UV-Vis spectrophotometry (Figure 2-8). Note that the
absorption band for the mercapto-bound Au nanoparticles is much broader than that for
the avidin-bound biotinylated Au nanoparticles (Figure 2-8). This is a function of the
agglomeration of the mercapto bound nanoparticles that changes their surface plasmon.
This shift in Au nanoparticle plasmon resonance caused by agglomeration has been used
as a transducer for the detection of analyte species, for example by Mirkin and coworkers
As before, subsequent silanization with aldehyde silane is thought to occur primarily
on the silacious nanotube exterior, as the interior has already been modified. A protein
labeled with a fluorescent tag was then attached via the aldehyde-amino Shiffbase
reaction (96). The protein attached was immunoglobulin G (IgG), which had a
covalently-attached fluorescein isothiocyanate (FITC) group. The FITC absorbs in a
narrow region centered at 495 nm and thus serves as a label for protein attachment. The
dual-modified nanotubes again show two distinct absorbance peaks in the UV-Vis. The
broad 550-650 nm peak is due to the agglomerated Au nanoparticle absorbance (inside)
while the sharp 495 peak is a result of the FITC labeled protein (outside).
Nanotubes as Extractants
C1i nanotubes for drug detoxification
Differentially modified nanotubes have a practical application as drug detoxification
agents. Drug overdoses affect hundreds of thousands of people per year in the United
States alone (108). If the overdosed drugs are ingested, the patient's stomach can be
pumped and/or filled with a slurry of carbon black to sorb the remainder. However, if the
drugs are administered intraveinously or have already been absorbed into the
bloodstream, health professionals have little recourse.
Most drugs are soluble in aqueous solution but generally prefer a more lipophilic
environment due to the prevalence of hydrophobic moieties. For this reason, drugs tend
to deposit in fatty tissue in the body. It was posited that addition of lipophilic
(hydrophobic) particles to the blood would sorb the drug and therefore decrease toxicity.
Purely lipophilic particles would function as sorbants but may not disperse in aqueous
solution. Agglomeration of the particles in the bloodstream could lead to blockage,
compromising the health and even life of the patient. As demonstrated above, silica
nanotubes can be differentially modified to have two different surface chemistries, inside
and outside. Our goal was to create nanotubes with a lipophilic interior to sorb drugs and
a hydrophilic exterior to ensure suspension in aqueous solution.
Silica nanotubes embedded in an alumina template were silanized with Cis silane as
detailed in the experimental section. The template masked the outer nanotube surface
and consequently only the nanotube interior was modified. The template was dissolved
in acid and the nanotubes collected by filtration. The bare silica outer surface remains
hydrophilic, which causes the nanotubes to suspend in aqueous solution. Aqueous
suspendibility of nanotubes with bare silica outer surfaces is illustrated by Figure 2-6. In
this figure, nanotubes with bare silica exteriors and fluorescently silanized interiors are
shown to suspend in the water phase of a cyclohexane/water mixture. Nanotubes
modified on the exterior surface with C18 do not suspend in aqueous solution but rather
float on top.
Figure 2-9. 7,8-benzoquinoline.
A model molecule was chosen to examine the efficacy of the nanotubes as sorbents.
The model needed to be of a moderate molecular weight, highly lipophilic, yet at least
somewhat soluble in water. It should also be easy to monitor, and stable over time. The
compound 7,8-benzoquinoline (BQ, Figure 2-9) satisfies all these conditions. It has a
molecular weight of 179 g/mol and is water soluble to about 2 x 10.4 M. Additionally the
tricylic structure of BQ results in strong absorbance in the UV. Concentrations down to
1.0 x 10"7 M can be easily monitored with UV spectroscopy using the absorbance peaks
at 233 or 266 nm. It is also stable under ambient conditions.
For the BQ extraction experiments, an initial concentration of 1.0 x 10-5 M BQ was
used. A UV spectrum of this concentration was taken to serve as a reference (see the
diamond top trace in Figure 2-10). To 2.5 mL of this solution was added 2.5 mg of
nanotubes modified with Cs1 on the interior surface. This amount of 1.0 mg
nanotubes/mL solution was arbitrarily set as a benchmark. A higher concentration of
nanotubes would obviously remove more drug.
o Control Solution (10^-5 M benzoquinoline)
0.3 1 A After Extraction with SiO2 Tubules
*.* ~ After Extraction with C-18 SiO2 Tubules
S 7 After 2nd Extraction with C-18 SiO2 Tubules
220 240 260 280 300
Figure 2-10. UV-Vis spectra of 7,8-benzoquinoline extracted with C18 modified silica
The solution was briefly sonicated to suspend the nanotubes and then filtered through
a 20 nm alumina filter. The lipophilic BQ should partition into the lipophilic nanotube
interior and be removed from solution when the nanotubes are filtered. After filtration,
the concentration of the remaining BQ was again determined by UV-Vis spectroscopy.
The results are shown in the circle trace of Figure 2-10. Results are tabulated in Table
Table 2-1. Absorbance of 7,8-Benzoquinoline Before and After Extractions.
1.0 x 10^-5 M Extraction with Extraction with 2nd Extraction
Starting Sol. Silica Nanotubes C18 Nanotubes C18 Nanotubes
Abs. @ 265 nm 0.1889 0.1708 0.0339 0.015
% Extracted 10% 82% 92%
It is evident that most of the BQ has been removed. The absorbance of BQ remaining
in solution was 0.0339, meaning that 82% of the BQ was extracted by the C18 modified
nanotubes. If the BQ solution is extracted twice, each time with a concentration of 1 mg
modified nanotubes per mL of solution, the amount of BQ removed increases to 92%.
The bottom trace of Figure 2-10 shows this result.
To insure that the lipophilic C18 was responsible for the extractions, a control
experiment was performed with unmodified silica nanotubes. Again 1.0 mg
nanotubes/mL solution was added, sonicated to disperse, and filtered. As shown in the
upper triangle trace of Figure 2-10, only a small amount of BQ was removed from the
solution by the unmodified nanotubes, -10%. It should be noted that for all the lipophilic
extraction experiment glass transfer pipettes, syringes, and vials were used. The
lipophilic nature of polymer (generally poly(ethylene)) storage and transfer devices
would otherwise cause sorbtion of the BQ and skew the results.
To more closely examine the effect of compound lipophilicity on relative extraction, a
series of increasingly lipophilic compounds was chosen. These included the moderately
hydrophilicp-cresol, and, in order of increasing lipophilicity, 2,4-dimethylphenol and
2,4,6-trimethylphenol. The increasing lipophilicity of these compounds is confirmed by
their increasing octanol/water partition coefficients. The log of the octanol/water
partition coefficient for p-cresol is 1.97 (109), for 2,4-dimethylphenol it is 2.35 (109), and
for 2,4,6-trimethylphenol it is 2.73 (110). Structures of all compounds are shown in
Figure 2-11. Solutions of all compounds at a concentration of 1.0 x 10-5 M were made in
pH 9.0 phosphate buffer. The buffer was used to insure protonation of the phenol.
Deprotonation of the phenol would alter the lipophilicity of the compounds. Extractions
with interior modified Csi nanotubes were analogous to those for 7,8-benzoqinoline.
Each extraction used 1 mg nanotubes/mL solution. The absorbances of the solutions
were monitored before and after extraction using UV-Vis spectroscopy. Results are
shown in Figure 2-12 and tabulated in Table 2-2.
CH3 CH3 CH3
H3 H3C CH3
A. OH B. H C. H
Figure 2-11. Structures of Lipophilicity Study Compounds A)p-cresol.
B) 2,4-dimethylphenol. C) 2,4,6-trimethylphenol.
Table 2-2. Absorbance Before and After Extraction.
Compound Wavelength Starting Abs. Abs. after extraction % Extracted
p-cresol 277 0.01559 0.01547 0.10%
2,4-dimethyl- 278 0.02131 0.01484 30%
2,4,6-trimethyl- 280 0.01267 0.00735 42%
It would be expected that the most hydrophilic compound would be extracted the least.
As shown in Figure 2-12A, this is the case. The concentration of p-cresol before and
after extraction is essentially identical. The more lipophilic 2,4-dimethylphenol is
extracted by the Cs1 modified nanotubes, as seen if Figure 2-12B. The concentration
after extraction is 7.0 x 106 M, thus 30% was extracted. Figure 2-12C shows that
1 x 10"-5 M p-cresol
After C-8 1()02 tubule extraction
270 275 280 285 290
S1 x 10^-5 M 2,4-dimethylphenol
SAtter extraction with C-l18 i-)2 tubules
265 270 275 280 285 290 295
Figure 2-12. UV-Vis spectra showing increasing extraction for more lipophilic
compounds. A) p-cresol. B) 2,4-dimethylphenol.
C 1 x 10^-5 M 2,4,6-trimethylphenol
A After C-18/Si02 tubule extraction
S0.008 loss:" AA',
o A"A A
0.006 A "A
0.002 & .
265 270 275 280 285 290 295
Figure 2-12. Continued.
extraction efficiency increases for 2,4,6-trimethylphenol, the most lipophilic of the
compounds. Approximately 42% is sorbed by the Cls nanotubes.
The Cs1 modified nanotubes were next used to extract an actual drug. An aqueous
solution ofbupivacaine, a commonly used local anesthetic, was prepared at a
physiological dosage level of 5 x 10'6 M. The structure of bupivacaine is shown in
Figure 2-13. Due to the cationic nature of the bupivacaine, it is not as lipophilic as the
7,8-benzoquinoline and would not expect to be as readily extracted by the Cs1 nanotubes.
C 3 HCI
Figure 2-13. Bupivacaine
Extraction with 1.0 mg Cs1 nanotubes per mL solution removed 65% of the
benzoquinoline, or 1.1 ug bupivacaine per mg of nanotubes. More bupivacaine could be
extracted from a more concentrated starting solution. The same concentration of
nanotubes removed 15 ug bupivacaine per mg nanotubes from a 1 x 104 M solution.
However, the efficiency of the extraction was decreased, with only 45% of the total
EDTA nanotubes for copper(H) extraction
As a second type of chemical extraction, the nanotubes were modified with an EDTA
analog (Figure 2-1G) which we refer to as EDTriA. The details of this modification are
discussed in the experimental section. Note that the free silica nanotubes were silanized,
which results in modification of both interior and exterior surfaces. The EDTriA, like
EDTA, is hydrophilic and the tubes suspend easily in aqueous systems, even if modified
on the exterior.
The EDTriA is pentadentate, with one acetic acid group of EDTA replaced by an
organosilane through which it is bound to the nanotubes. EDTA and related multidentate
compounds are known to strongly chelate many types of metal ions (111). It was
therefore hypothesized that the EDTriA-modified nanotubes would be useful in removing
metal ions from solution. Copper (II) was chosen as a representative ion for this
extraction system. The binding constant for Cu2+ to hexadentate EDTA is 6.3 x 1018
(112). Although the binding constant for the pentadentate analog will be somewhat
smaller, is should still be considerable.
Extraction experiments were similar to those involving C18 nanotubes. First the
concentration of a buffered solution of Cu2+ was measured using Inductively Coupled
Plasma-Atomic Emission Spectroscopy (ICP-AES). The initial concentration of Cu2'
was 2.0 ppm. EDTriA-modified nanotubes were then added to the solution and sonicated
to suspend. As before, 1 mg nanotubes was used per mL solution. The nanotubes were
filtered and the Cu2+ concentration of the filtrate solution remeasured by ICP-AES. As
shown in Figure 2-14 and Table 2-3, extraction removed 52% of the Cu2+. The control
experiment, using unmodified SiO2 nanotubes, removed only 5% of the Cu2+ (Figure
2-14 and Table 2-3).
Table 2-3. Cu(II) Extraction with EDTriA Nanotubes.
2.0 ppm Extraction with Extraction with
Starting Sol. Silica Nanotubes C18 Nanotubes
ICP-AES Area Counts 503 481 247
Percent Extracted 5% 52%
The results of the C8/benzoquinoline and EDTriA/Cu2+ experiments demonstrate that
our modified nanotubes can function effectively as extractants. The interior and exterior
surfaces can be differentially modified to tailor nanotube chemistry for a particular use,
(e.g., suspension in aqueous solution).
Composite Silica-Polymer Nanotubes
During template synthesis, most materials preferentially deposit on the template
surface. This minimizes surface energy (113), of particular importance for a high surface
energy template such as alumina (114). Because of this phenomenon, nanotubes are the
most common structure produced by template synthesis. Solid rods or wires can be
templated under certain conditions, for example if electroless metal deposition is
performed at a slow rate for long times (48). However, it is typical for the pores in the
template membrane to remain open after deposition, although with a smaller diameters.
This fact is useful in the differential silanization of silica nanotubes as it allows exclusive
alteration of the interior surface. It also allows a second material to be template
synthesized inside the first.
E 2.0 ppm Cu(II)
[ After extraction with silica nanotubes
600 1- After extraction with EDTA-silcia nanotubes
Figure 2-14. ICP-AES data showing Cu2+ extraction with EDTriA-modified nanotubes.
The Martin group has used this approach to manufacture concentric metal-insulating
polymer-conducting polymer composites (46). Using 3 pm diameter polyester track
etched membranes as templates, Au nanotubes were deposited by an electroless process.
This yields an Au microtube in each pore of the template membrane. Using the Au
nanotubes as working electrodes, poly(phenylene oxide) was next deposited
electrochemically from a solution of 2,6-dimethylphenol monomer. This polymer is
insulating and passivates the gold nanotubes. After gold and polymer depositions,
electronmicrographs showed that the pores of the membrane were still open. Finally, a
conducting wire of poly(pyrrole) was grown electrochemically to fill the center of the
These results show that composite structures can be synthesized using polymer track
etched templates. The same process should be amenable to templating with alumina
membranes. To explore this area, nanotubes of silica and polymer were prepared as
described in the experimental section. Two types of insulating polymers were used,
poly(ethylene-vinyl acetate) (PEVA), and poly(methyl methacrylate) (PMMA). Both
were deposited from solution by dip coating. Polymers could be deposited either before
or after silica nanotube formation. The silica was deposited from a sol-gel precursor, as
previously discussed. Commercial 200 nm alumina membranes were used as the
templates. Electronmicrographs show that the membrane pores are still open after
polymer and silica deposition (see Figure 2-15). The template can be removed by acid or
base treatment to free the composite structures (Figure 2-16).
Although it is difficult to see the silica or polymer layers from Figure 2-16, proof of
the composite nature is given by the nanotube solubilities. As with the C18 modified
nanotubes, the dispersion of polymer-silica composite nanotubes is controlled by the
chemistry of the outer surface. Composites with silica on the outer surface and polymer
inside suspend well in aqueous systems but do not disperse in organic solvents.
Conversely, composites with polymer exteriors do not disperse in water but suspend in
many organic solvents, such as ethanol or toluene. Care must be taken to not use solvents
that will dissolve the polymer. The polymer type can also affect dispersion. For
example, nanotubes with PMMA on the outside disperse readily in ethanol, while
nanotubes coated with the more hydrophobic PEVA do not.
Conducting polymer-silica nanocomposites were also prepared. In this process the
silica was deposited first, followed by chemical polymerization of poly(pyrrole) (Ppy)
from an aqueous solution of the monomer. Details of the synthesis are available in the
experimental section. A homegrown alumina membrane with -60 nm pores was used as
the template. The resulting nanostructures are seen in Figure 2-17.
Figure 2-15. Scanning electron micrographs of alumina templates after polymer and
silica depostion showing open pores. A) Poly(methyl methacrylate) and
silica. B) Poly(ethylene-vinyl acetate) and silica.
Figure 2-16. Scanning electron micrographs of composite polymer outside/silica inside
nanotubes templated in 200 nm commerical alumina templates.
A) Poly(methyl methacrylate) and silica. B) Poly(ethylene-vinyl acetate)
Figure 2-17. Scanning electron micrographs ofpoly(pyrrole) inside/silica outside
composite nanotubes templated in homegrown 60 nm dia. alumina
membranes. A) Nanotubes held together by a film of silica. B) Dispersed
In Figure 2-17A the structures are held together by a surface layer of silica. Typically
this layer would be polished away to free the nanotubes, but here it was left intact and the
nanotubes protrude like bristles from a brush. After template dissolution, the nanotubes
are rinsed with water, then dried. The surface tension of the evaporating water draws
together collections of nanotubes to form the islands visible in Figure 2-17A. Figure
2-17B shows composite Ppy-silica nanotubes prepared in the same fashion but with the
surface silica layer removed. As the template is dissolved, these nanotubes are freed and
suspend in solution.
To prove that Ppy had been deposited into the silica nanotubes, an absorbance profile
of an aqueous dispersion of the composite nanotubes was taken over the visible region.
Poly(pyrrole) has a characteristic broad absorbance from 600 nm to -2 pm. Absorbance
300 400 500 600 700 800
Figure 2-18. UV-Vis spectra of poly(pyrrole)/silica composite nanotubes showing
in this region is not seen for the bare silica nanotubes. The UV-Vis spectrum is seen in
Figure 2-18. The broad peak starting at -600 nm indicates that Ppy has been deposited.
This type of composite nanotube may have analytical applications, as the Ppy being
oxidized or reduced could generate an analytical signal.
The small size of micro- and nanoscale materials often makes it difficult to assemble
them into organized assemblies. Physical manipulation in this size regime is challenging,
even with highly specialized equipment. A better solution is relying on interacting
particle chemistries to self-assemble. The Whitesides group has been very active in this
area, using a variety of techniques to self-assemble structures from macro to sub-micro
scale (115-118). Natan and coworkers have used various chemical and biochemical
strategies to self-assemble colloidal Au particles onto surfaces (119, 120). Mirkin's
group used thiol linkers to bind colloidal Au particles to single stranded oligonucleotides.
These nanoparticles were subsequently self-assembled into a polymeric network by
addition of the complementary oligonucleotide (121, 122). Thiol linkers have also been
used to attach Au nanoparticles to carbon nanotubes (123).
To date, template synthesized micro- and nanostructures have not been self-assembled
in this way. To examine the possibility of self-assembling silica nanotubes, the
nanotubes were modified with biotin. As noted above, biotin binds very tightly to the
protein avidin. Incubation ofbiotinylated nanotubes with avidinated microspheres should
result in binding of the two to form an agglomerated structure.
Biotinylation of silica nanotubes is described in detail in the experimental section.
Briefly, the nanotubes were first silanized with a primary amino group that then bound a
succinimide ester containing the biotin. The biotin was at the end of a long linker arm to
ensure that steric hindrance would not compromise binding (124). Biotinylation occurred
on both the inside and outside surfaces of the nanotubes.
The biotin-modified nanotubes were incubated with commercially available avidinated
silica microspheres (Bangs Labs). Two sizes of microspheres were used, 5.0 gm and 0.5
pm diameter. Control experiments were performed with poly(ethylene glycol) (PEG)
silanized silica nanotubes. Poly(ethylene glycol) is known to prevent adsorbtion to
proteins. If bare nanotubes were used as a control, the silica would likely bind to the
avidin on the microspheres non-specifically.
When biotinylated nanotubes were incubated with the 5.0 pm avidinated silica
particles, they adhered to the particle surface (Figure 2-19). If PEG coated nanotubes
were used, no adhesion was seen (Figure 2-20). Similarly, 0.5 pm silica particles
agglomerated with the biotinylated nanotubes (Figure 2-21), but not with PEG-ylated
nanotubes (Figure 2-22).
Figure 2-19. Scanning electron micrographs of biotinylated silica nanotubes on 5.0 pm
diameter avidinated silica spheres.
Figure 2-20. Scanning electron micrographs ofpoly(ethylene glycol) modified silica
nanotubes and 5.0 pun diameter avidinated silica spheres.
Scanning electron micrographs of biotinylated silica nanotubes and 0.5
pm diameter avidinated silica spheres. A) Ball/stick clusters.
B) Agglomerated ball/stick mass.
Figure 2-22. Scanning electron micrographs ofpoly(ethylene glycol) coated silica
nanotubes and 0.5 gm diameter avidinated silica spheres. A) Isolated balls
and sticks. B) Agglomerated stick mass and isolated balls.
While the self-assembled structures described thus far have dimensions in the
micrometer regime, there is no fundamental reason that this strategy cannot be applied to
much smaller nanowires and nanoparticles. An alternative self-assembly chemistry was
used to prove this point. This entailed the use of 1,9-nonanedithiol to assemble colloidal
Au particles (10 nm diameter) to the ends of 30 nm diameter template synthesized Au
nanowires (see experimental section for details).
A gold film was sputtered onto one face of a polycarbonate template membrane with
30 nm diameter pores. This film served as the working electrode for deposition of Au
nanowires within the pores of this membrane (125). The nanowire-containing membrane
was then immersed overnight into a 1 mM ethanolic solution of the dithiol. This resulted
in chemisorption of the dithiol onto the ends of the Au nanowires and also onto the
sputtered Au surface film. The Au film was removed by polishing with a laboratory
tissue. Because this polishing removed the surface film, the only remaining dithiolated
surface was the exposed ends of the Au nanowires. The membrane was then immersed
overnight into an aqueous colloid of Au nanoparticles, which resulted in binding of the
Au nanoparticles to the ends of the dithiolated nanowires. The membrane was then
rinsed with water to remove any unbound Au nanoparticles. The template membrane was
dissolved away by immersion in CHCl3 to yield a suspension (125) of the Au
nanowire/colloidal particle assemblies. A drop of this suspension was applied to a
transmission electron microscopy grid.
Figure 2-23 shows transmission electron micrographs of gold nanowire/colloidal
particle assemblies. Note that because the nanowires were embedded within the template
when they were exposed to the dithiol and colloidal nanoparticle suspension, the Au
particles are self-assembled to only the ends of the Au nanowires. About 10% of the
nanowires in such images were terminated with colloidal particles.
Figure 2-23. Transmission electron micrographs of template synthesized 30 nm
diameter Au nanowires to which 10 nm Au particles have been
self-assembled using 1,9-nonanedithiol.
Novel nanoparticulate silica-based extractants have been prepared with a variety of
chemistries. The initial step involves the template synthesis of silica nanotubes in a
porous alumina membrane. The nanotubes can then be modified with different chemical
moieties depending on the extraction desired, typically via a silanization process.
The inner and outer surface chemistries can be modified differentially. Mixed
nanotubes can be prepared in which the inside is silanized with a quinineurathane and the
outside coated with protein-bound Au nanoparticles. A further example is the
preparation of nanotubes with thiol-bound Au nanoparticles on the inside and
fluorescently labeled proteins on the outside. Outer surface chemistry can direct
nanotube phase segregation. Nanotubes with unmodified silica exteriors partition into
and suspend in aqueous solutions. In contrast, nanotubes modified with Cis on the outer
surface partition into and suspend in hydrophobic organic solvents.
Template synthesized nanotubes can be used as extractants. Using C18 modified
nanotubes, model and actual drugs can be extracted from aqueous solution. Extraction
efficiencies range from 45% to 85%, with higher efficiencies for lower initial drug
concentrations. Nanotubes modified with an EDTA analog extract copper ions from
Composite polymer-silica nanotubes have been prepared. The polymer can be
compose either the external or internal surface. As with the differentially modified
nanotubes, surface chemistry controls dispersion properties.
Template synthesized nanotubes and nanowires can be used as building blocks for
self-assembly. Biotinylated silica nanotubes bind to avidinated silica microspheres.
Dithiol chemistry can be used to link Au nanowires with Au nanoparticles.
PROTEIN MODIFICATION OF SILICA NANOTUBES
This chapter describes the attachment of proteins onto template synthesized nanotubes.
The proteins are covalently linked via an aldehyde silane bridge that binds primary amino
groups. Because most proteins have pendant primary amines, a wide range of proteins
can be used to modify the nanotubes. The molecular recognition capabilities of certain
proteins can be retained after binding to make the protein-modified nanotubes
exceptionally specific extractants. Avidin-modified nanotubes extract biotin coated Au
nanoparticles from solution with high extraction efficiency. Immunoprotein-modified
nanotubes extract the corresponding antibody from solution with high specificity.
Nanotubes modified with an enantioselective antibody extract one enantiomer from a
racemic mix. Enzymes, including drug detoxification enzymes, are also attached to the
nanotubes and shown to retain their catalytic activity. Nanotubes modified in this way
function as bioreactors. Immunoproteins on the outside of nanotubes can be used to
direct nanotube binding, creating specific labeling agents.
Nanoparticles have recently found use in a variety of biomedical applications. These
include enzyme encapsulation (126, 127), DNA transfection (128, 129), biosensors (36,
130-132), and, as detailed in Chapter 3, drug delivery (73-76). Linkage of the
biomolecule of interest to the nanoparticle is a critical step to increasing nanoparticle
utility. Several examples ofbiomolecule particle modification are reviewed, as well as
'smart' nanoparticles which have been engineered to deliver material to or bind to a
specific biological site.
Nanoparticles can be engineered to deliver to specific organs or sites in the body. The
majority of this work again comes from drug delivery research. For example, drugs that
do not normally cross the blood/brain barrier (BBB) can be carried across in designed
nanoparticles. Schroeder et al. incorporated two analgesic peptides, dalargin and
kyotorphin, which are known to not cross the BBB, into nanoparticles (75). The
nanoparticles were made of poly(butylcyanoacrylate) cores and non-ionic surfactant
coronas, either of dextran 70,000 or polysorbate 85. A coating of another non-ionic
surfactant, polysorbate 80, was added prior to injection into mice to enhance the BBB
crossing rate. Nanoparticle size was about 200 nm for the dextran 70,000 stabilized
particles and about 300 nm for those stabilized with polysorbate 85. Tests found that
neither drug caused analgesia in the mice when administered free, nor did the non-drug
carrying nanoparticles. When the nanoparticles were loaded with either drug, analgesia
was induced, indicating that the drugs passed the BBB (75).
In another interesting nanoparticle modification, Levy et al. increased the arterial
uptake of drug carrying nanoparticles by changing the nanoparticle surface charge (74).
The need for arterial drug delivery stems from the high recurrence of artery obstruction
after surgery (30-50% of patients develop reoclusion within 3-6 months) (74). In this
study, a model drug, U-86, was used to test the efficacy of the delivery system. The
nanoparticles were composed of polylactic-polyglycolic acid copolymer and were
10040 nm in diameter. Several modifying agents, (including heparin, cyanoacrylate,
and ferritin) were coated on the surface of the nanoparticles; a complete list is presented
in Table 1 of reference (74). The most effective modifier was the cationic surfactant
didodecylammonium bromide (DMBA). The free U-86 concentration in the artery walls
was 7-10 times higher after treatment with nanoparticles modified with DMBA compared
to non-modified (but still drug containing) nanoparticles (74). The authors postulate that
this is due to the change in the surface charge of the particles. The unmodified particles
have an electrical surface charge, or zeta potential, of-28 mV while the DMBA modified
particles have a zeta potential of+22 mV. This change in charge may increase the
permeability of the arterial walls (74).
Bae and coworkers modified nanoparticles to specifically interact with cancerous liver
cells. They made their nanoparticles from block copolymers ofcurdlan (corona) and
sulfonylurea (core). Curdlan is a naturally occurring polysaccharide produced by
microbes, and is known to have anti-tumor activity. Although the mechanism for this
activity is not understood, the authors propose that the interaction between curdlan and
cancer cells may make nanoparticles of this substance effective anti-tumor drug delivery
agents. Three different ratios ofcurdlan to sulfonylurea were tested, but all of the
nanoparticles were about 200 nm in their swollen state and 30-50 nm when dried (73).
Because Bae and coworkers were specifically interested in treatment of liver cancers,
they further modified the surface of the nanoparticles with glactose moieties (73). Liver
cells have glactose receptors and preferentially uptake glactosylated proteins and
macromolecules. A fluorescent probe, rhodamine isothiocyanate (RITC), was also
attached to the nanoparticles to quantify their uptake. Results show that the glactosylated
nanoparticles are well taken up by the cancerous liver cells (73). Control experiments
with other cell types or with non-glactosylated nanoparticles show only 20% of the
In another example of'smart' nanoparticles, Wooley et al. functionalized the surface
of shell cross-linked particles with a protein derived from HIV that facilitates particle
incorporation into cells (84). They first prepared this protein (known as the protein
transduction domain or PTD) by peptide synthesis on a solid support. Then, in a one step
process, they detached the PTD from the support, bound it to the nanoparticles, and
degraded the interior core of the particles. This yielded hollow nanocages coated with
PTD. The size of the nanoparticles was 55 20 nm (84). The PTD nanocages were then
labeled with the fluorophore fluorescein-5-thiosemicarbazide (FTSC), and incubated with
Chinese hamster ovary (CHO) cells and HeLa cells. (HeLa cells are cancer cells often
used for research that originated from a patient named Henrietta Lacks). The fluorescent
PTD-coated nanoparticles attached to and were incorporated by both the CHO and HeLa
cells (84). Similar nanoparticles without the PTD coating showed no evidence of binding
to either cell type. The nanoparticle labeled cells could be separated from unlabeled cells
via flow cytometry (84). Drugs have not yet been incorporated into the nanoparticles for
use in targeted drug delivery, but this is the goal.
Coacervate nanoparticles, which consist of self-assembled nanospheres of ionic
polymers, have also been engineered to be 'smart'. The composition of coacervates is
described in Chapter 2. August et al. synthesized DNA-gelatin coacervate nanoparticles
coated with the protein transferring and containing calcium and the drug chloroquine, all
of which are said to increase transfection (87). Transfection is the transfer of the
nanoparticle DNA into the cell nucleus. The extent of transfection is determined by
measuring which cells produce a protein coded by the nanoparticle DNA after incubation
with the nanoparticles. Gelatin and DNA were used as the ionic polymers to create these
coacervate nanoparticles (87). The resulting particles were about 400 nm in diameter.
The cells used for transfection were human tracheal epithelial cells. Transferrin, an iron-
binding protein present in blood plasma, was bound to the outside of the nanoparticles.
This was found to be necessary for transfection (87). Nanoparticles lacking transferring
showed no transfection ability. The researchers presume that this protein acts as a ligand
to a cell surface receptor, effectively docking the nanoparticle and allowing DNA transfer
into the cell. Calcium incorporated into the nanoparticles (in the form of CaPO4) was
also found to increase cell transfection about 10 fold versus nanoparticles without
calcium (87). This may be due to the calcium enhancing the permeability of the cell
membrane. The researchers also incorporated the drug chloroquine, postulating that it
would protect the DNA in the nanoparticles from degradation from DNAse enzymes
(87). There were, however no control experiments of nanoparticles without chloroquine
to prove or refute the supposed protection.
An interesting recent example of biological modification of nanoparticles concerns the
functionalization of carbon nanotubes. Dai and coworkers used a pyrene-based
succinimide ester to link proteins to the exterior surface of single-walled carbon
nanotubes (133). The pyrene structure irreversibly adsorbs to the nanotube via
n-stacking. The succinimide ester is then used to covalently bind to amine groups on the
protein. The group bound both ferritin and Au nanoparticle-labeled streptavidin, and
showed the presence of both using transmission electron microscopy (133).
The group of Dr. Weihong Tan at the University of Florida has also developed
bio/nano particles for labeling applications (134). Using a sol-gel microemulsion
process, silica nanoparticles were formed containing luminescent chromophores
immobilized in the silica matrix. This silica network protects the luminophores from
quenching by environmental oxygen (134). To make the luminescent nanoparticles into
labels, they were covalently coated with immunoproteins using a cyanogen bromide
coupling reaction. First, the hydroxylated silica surface was derivatized with cyanogen
bromide. Amino groups on the antibody then bound to the carbon of the cyanogen
bromide. The nanoscopic labels were used to identify leukemia cells, reportedly easily,
clearly, and with high efficiency (134).
Recently, nanoparticles of ZnS have been arranged into highly ordered crystals by
attaching a self-ordering virus (135). Belcher and coworkers first screened a collection of
viruses to find those that effectively bound to ZnS surfaces. A virus that bound
effectively was cloned and amplified. A solution of the amplified viruses was then
incubated with ZnS nanoparticles. The ZnS binding sites on the virus bound to the ZnS
nanoparticles such that each virus contained a nanoparticle at one end (135). Because the
virus is water soluble, this resulted in a viral/nanoparticle suspension. The
viral/nanoparticles formed ordered arrays in the suspension similar to liquid crystals due
to the van der Walls interactions of the long, rod-shaped viruses (135). Ordered packing
over several micrometers was observed when the viral/nanoparticle suspension was cast
onto smooth surfaces. The authors postulate that because the ZnS nanoparticles are
magnetic, the highly ordered arrays may find application in high-density information
Polymer particles have been modified with a variety of biomolecules and used
effectively as extractants (136, 137). Because the polymer particles in solution resemble
colloids of polyterpene produced by rubber plants, the polymer particles are termed
'latex' particles and this type of extraction process is termed 'affinity latex'. The
pioneers in this area are the Kawaguchi group at Keio University in Japan. Latex
polymer particles are usually on the order of 200 to 1000 nm, slightly larger than true
nanoparticles. Affinity latex particles have been modified with DNA to detect point
mutations in genes (136) and with proteins for identification of drug receptors (137).
Silica nanotubes were prepared as described in Chapter 3. Protein functionalized SiO2
nanotubes were typically prepared from the unmodified nanotubes via an aldehyde
terminated siloxane linker (96). Unmodified nanotubes were stirred in a 5% aqueous,
10% aldehyde methoxysilane (PSX1050, United Chemical Technologies) in ethanol, pH
adjusted to 5.0 with acetate. The aldehyde-modified nanotubes were filtered, rinsed with
ethanol, and dried 24 hours in an oxygen free glove box. After drying, the nanotubes
were incubated with a solution of protein in buffer (-1 mg protein/ml buffer). Incubation
was at 4 *C overnight. Aldehydes have been shown to react with pendant primary amines
on proteins (lysine residues) to covalently link the proteins to the substrate (97-99). This
resulted in proteins being bound to both the interior and exterior of the nanotubes. The
protein-modified nanotubes were filtered from the protein solution and extensively rinsed
with buffer before use.
Nanotubes were modified with proteins on the inside only using a three step
gluteraldehyde process. While still in the alumina template membrane, the nanotubes
were silanized with an amino silane. The silica-coated alumina membrane was immersed
for two hours in an ethanol solution containing 5% by volume
3-aminopropyltriethoxysilane and 5% (vol./vol.) water. After rinsing with ethanol, the
silane was cured by heating two hours at 80C. The alumina template was then dissolved
with aqueous 25% phosphoric acid and the nanotubes were filtered and extensively
rinsed. The amino-modified interior surface was further modified with aldehyde by
suspending the nanotubes in a 150 mM pH 7.4 phosphate buffer containing 2.5%
(vol./vol.) gluteraldehyde for five hours. The nanotubes were again filtered, rinsed with
buffer, and suspended in protein solutions overnight at 4C for protein attachment.
Poly(ethylene glycol) (PEG) was attached to the nanotubes using a 5000 MW PEG
silane obtained from Shearwater Inc, Huntsville, AL. An ethanol solution containing 5%
water by volume was stirred and the PEG silane was added until it reached a
concentration of 5% by volume. The pH of the solution was adjusted to 5.1 with sodium
acetate/acetic acid. To this mixure, silica nanotubes were added and stirred for 30
minutes. The nanotubes were then filtered, rinsed with ethanol to remove excess silane,
and cured at 60C overnight.
Avidin, bovine serum albumin, biotinylated gold nanoparticles, and immunoproteins
other than ENAl lHis were purchased from Sigma and used as received. Unless
otherwise noted, all instruments used were as listed in Chapter 2.
The a (R,S) and d(S,R) enantiomers of
4-[3-(4-fluorophenyl)-2-hydroxy- 1-[1,2,4]triazol- 1-yl-propyl]-benzonitrile and the Fab
antibody fragment ENA1 lHis were kindly provided Dr. Hans Soderlund and coworkers
at VTT Biotechnology, Finland. All extraction experiments were from pH 8.5 sodium
phosphate buffer containing 5% (vol./vol.) dimethylsulfoxide. Separation and
concentration determination of the enantiomers was done using a Shimadzu VP system
High Performance Liquid Chromatograph (HPLC) under the following conditions: An
Ultron ES-OVM column was used at 30 C with a running buffer of 20 mM ammonium
phosphate adjusted to pH 5.0 containing 20% methanol by volume. The flow rate was
1.0 mL/min. Detection was at 230 nm and the detection limit was at a signal to noise
level of 5. The injection loop had a volume of 20 pL. Chiral HPLC data were obtained
by Dr. Sang Bok Lee.
Extraction experiments were conducted with either 2.5 or 5.0 mL of solution. Unless
otherwise stated, the concentration of nanotubes added was -1.0 mg nanotubes per mL
The activity of glucose oxidase was assayed according to a modified literature
procedure (138). First a 50 mM, pH 5.1 acetate buffer was prepared containing 0.17 mM
o-dianisidine and 1.72% (wt./vol.) 3-D-glucose. To 2.90 mL of this solution was added
0.10 mL of aqueous peroxidase solution at a concentration of 60 purpurogallin units/mL.
Glucose oxidase-modified nanotubes were added to this mixture and shaken to disperse
the nanotubes. Production of the oxidized o-dianisidine was monitored using visible
spectroscopy at 500 nm.
Enzymatic assay of the human drug detoxification enzyme cytochrome P450 3A4 was
based on a standard method (139). An aqueous potassium phosphate buffer (0.10 M) was
prepared at a pH of 7.4. To 930.4 pL of this buffer was added 50 pL of a solution
containing 20mg/mL glucose-6-phosphatate, 20mg/mL nicotinamide adenine
dinucleotide phosphate (NADP), and 13.3 mg/mL magnesium chloride hexahydrate. Ten
microliters of a 5 mM tribasic sodium citrate solution containing 40 units/mL of
glucose-6-phosphate dehydrogenase were also added. To this mixture was added 9.6 pL
of 20.8 mM testosterone in acetonitrile. The testosterone serves as the substrate for the
enzymatic reaction. The final solution volume was 1 mL and thus the final testosterone
concentration was 200 pM. Before addition of the enzyme, the solution was warmed to
35"C. To the prewarmed mix was added 5 mg ofP450 3A4 modified silica nanotubes.
Incubation at 35C was continued for 60 minutes. At 20 minutes, half the solution (0.50
mL) was removed and 250 pL of acetonitrile was added to denature and precipitate the
enzyme and to stop product formation. As a further impediment to the reaction, the
denatured mix was transferred to an ice bath immediately after addition of acetonitrile.
This procedure was repeated for the remaining half of the assay solution at 60 minutes.
Precipitated enzyme was removed by centrifuging at 14,000 rpm for 5 minutes. The
supernatant was then filtered through 20 nm diameter alumina filters (Whatman) to
remove the enzyme-modified nanotubes.
If the enzyme is active, the testosterone substrate will be converted to
6-P-hydroxytestosterone. Analysis of product formation was performed on a Shimadzu
VP HPLC system. The separation conditions were as follows: the running buffer was
60:40 methanol:water at a flow rate of 1 mL/min. Temperature was constant at 45'C.
The column used was a Supelcosil C18 that was 15 cm x 4.6 mm with an average particle
size of 5 pum. Absorbance was monitored at 254 nm. Under these conditions the
6-1-hydroxytestosterone eluted at 2.75 minutes and the testosterone at 5.3 minutes.
Results and Discussion
Over the course of millennia, Mother Nature has designed highly specific molecular
recognition agents. Examples include enzymes, transport proteins, and immunoproteins.
These agents are termed 'smart' due to their ability to recognize one particular molecule
in an array of molecular species. The capacity to couple these agents to our nanotubes
would give the nanotubes the same molecular recognition ability, and thus render them
'smart' as well. As noted by the examples, many of the most selective of these molecular
recognition agents are proteins. A method is therefore needed to couple proteins to the
The primary difficulty in protein attachment is the preservation of protein structure.
The molecular recognition ability of proteins is based on three-dimensional structure and
this must be preserved for retention of selectivity. Protein attachment must therefore
occur without denaturation. If the protein is denatured during the binding process, the
resulting modified nanotubes would show little selectivity or molecular recognition.
A typical approach for protein modification is the gluteraldehyde route (97, 98)
(Figure 3-1). In this approach, the surface to which the protein is to be attached is first
modified with primary amino groups. For a glass or silica surface, this is usually
accomplished via silanization with an amino silane. The surface is then treated with a
solution containing gluteraldehyde, a five-carbon dialdehyde. The aldehyde moieties on
the gluteraldehyde can bind to primary amino groups to form a Schiffbase. In this
manner, the amino surface is converted to an aldehyde surface after gluteraldehyde
treatment. Proteins are then incubated with the aldehyde-modified surface. Primary
pendant amino groups in the protein, (e. g., lysine residues), can bond to the remaining
aldehyde functionality of the gluteraldehyde. Schiffbase formation for the
protein/gluteraldehyde linkage is identical to the reaction linking the gluteraldehyde to
the amino modified surface. The result of this procedure is that protein is covalently
linked to the surface.
A OH + H3CH2CO Si /' NH2
-0 0 0
B -Oi/^ NH2 + H H
-ON H HProtein
_0'i/"'N=C iH --N- Protein
Figure 3-1. Schematic of gluteraldehyde protein attachment process. A) Silanization
of silica surface with aminosilane. B) Attachment ofgluteraldehyde.
C) Attachment of protein.
There are several drawbacks to the gluteraldehyde attachment route. It is a three-step
process, involving first amino modification, then dialdehyde modification, and then
modification with the protein. The total amount of protein coverage will be the product
of the coverage for each step. A loss of coverage in any one step cannot be recovered in
subsequent steps. For example, if the amino modification of the surface is low, only a
low degree of protein coverage can be expected even if the subsequent steps in the
process have high yields.
-OH H3CH2CO H
A OH + H3CH2CO-ir V' H
T f + H2N Protein
S-0 i ----N
~-O K\^-( :Protein
Figure 3-2. Schematic of aldehyde silane protein attachment process. A) Silanization
of silica surface with aldehyde silane. B) Attachment of protein.
A particular problem is the gluteraldehyde step. Because gluteraldehyde is a
dialdehyde, it can bond both aldehyde ends to amino groups during exposure to the
amino modified surface. If this cross-linking occurs, there can be no further bonding of
In an attempt to simplify this process, an aldehyde silane route was chosen (Figure
3-2) (96). The surface is first derivitized with the aldehyde silane, followed by
incubation with protein and attachment. The three-step process is decreased to two steps,
obviating the need for amino derivitization of the surface. This process also eliminates
the possibility of cross-linked aldehyde groups and insures that the aldehyde moiety will
be available for binding of the protein (96). The exact procedure for aldehyde
silanization and protein attachment can be found in the experimental section.
Extractions Using Protein-Modified Nanotubes
To determine if protein binding to the nanotubes occurred successfully and without
denaturation, we attempted to bind the protein avidin. Avidin and its binding to the small
molecule biotin are discussed in Chapter 2. To be able to monitor the binding of biotin
by the avidin, biotinylated 20 nm Au nanoparticles (Sigma) were chosen as the analyte.
Au nanoparticles have very strong absorbance in the visible region due to their surface
plasmon resonance (106), which allows straightforward monitoring of their concentration
using UV-Vis spectroscopy.
The idea was to modify the surfaces of the nanotube with avidin, then use the
modified nanotubes to extract the biotinylated Au nanoparticles from solution. It has
been seen in Chapter 3 that chemically modified nanotubes are effective extractants.
Here we wanted to use specific binding between proteins and ligands to enhance the
specificity of the extraction. If the extraction was successful, it would also indicate that
the protein is attached to the nanotubes and not significantly denatured by the attachment
An extraction experiment was performed that was similar to the drug extraction
experiments in Chapter 2. The absorbance spectra of the Au nanoparticle solution was
measured prior to addition of the nanotubes to determine the initial Au nanoparticle
concentration. Avidin modified nanotubes were added to an aqueous dispersion of
biotinylated Au nanoparticles at a concentration of 1 mg modified nanotubes per mL
solution. The solution was briefly sonicated after addition of the nanotubes to suspend
them. Following sonication, the solution was filtered, removing the nanotubes and any
biotinylated Au nanoparticles bound to the avidinated surfaces. The absorbance of the
filtered solution was then measured again to determine the amount ofAu nanoparticles
removed (Figure 3-3).
450 500 550 600 650
Figure 3-3. UV-Vis spectra biotinylated Au nanoparticle solution. A) Starting
solution. B) After extraction with 1 mg/mL bovine serum albumin-
modified nanotubes. C) After extraction with 1 mg/mL avidin-modified
As a control experiment, the SiO2 nanotubes were modified with bovine serum
albumin, which should have little specific affinity for the biotinylated Au nanoparticles,
and the extraction efficiency was determined in the same manner. In this case, significant
nonspecific adsorption of the Au nanoparticles was observed, with 34% removed from
solution (Figure 3-3). This is likely due to the fact that the gold nanoparticles are
stabilized with albumin, which may bind non-specifically to the albumin-coated silica
nanotubes. For the extraction with the avidinated SiO2 nanotubes, the extraction
efficiency was much higher, with 100% of the Au nanoparticles removed from solution
within our ability to measure (Figure 3-3).
These results show that the protein binding via the aldehyde silane is successful. The
protein is not significantly denatured in that the three-dimensional binding pocket for
avidin is still intact and functional.
To mitigate the possibility of nonspecific adsorption, we next attempted to use a more
specific protein/protein interaction for our extraction. Immunoproteins have long been
known to exhibit high specificities, with little or no affinity for analytes other than those
against which they were raised (140, 141). For example, antihuman immunoglobulin
proteins bind strongly to human immunoglobulins, but not to immunoglobulins from
other species, (e.g., mouse or rat immunoglobulins). The most common type of
immunoglobulin is Immunoglobulin G (IgG) (140). This is a Y-shaped molecule with a
molecular weight of 146 kDaltons (142). It consists of two identical top fragments called
the Fab region, and one stem fragment, called the Fc region. The molecule can be
cleaved into its Fab and Fc fragments using enzymes (142). The overall dimensions of
the molecule are -10nm tall and 10 nm wide (140). The Fab fragments are 6.0 nm by 3.5
nm and the Fc fragments are 4.0 nm wide and 4.5 nm tall. The binding site for antigens
is on the end of each of the Fab fragments (142). The binding site area has dimensions of
3.0 nm by 1.0 nm by 0.6 nm (140).
Human IgG was attached via aldehyde silanes to one batch of SiO2 nanotubes (see
experimental). Rat IgG was attached in the same way to another batch of nanotubes to
serve as a control. The analyte for these experiments was a Fab fragment of antihuman
IgG antibody that had been labeled with a fluorescent fluorescein isothiocyanate (FITC)
tag (143) for absorbance monitoring. The structure of FITC is seen in Figure 3-4.
Figure 3-4. Structure of fluorescein isothiocyanate (FITC).
A solution of the FITC-labeled antihuman IgG was extracted with the modified
nanotubes in a manner analogous to the avidin/biotin extractions. The absorbance before
and after extraction was measured by UV-Vis spectroscopy. Extractions used I mg of
modified nanotubes per mL of solution. Results are seen in Figure 3-5. As expected, the
human IgG-modified nanotubes effectively extracted the FITC-labeled antihuman IgG.
Comparing the initial absorbance to the absorbance after extraction shows that 76% of
the FITC-labeled antihuman IgG has been removed from solution (Figure 3-5). In
contrast, Figure 3-5 also shows that there was no appreciable antihuman extraction using
the rat IgG-functionalized nanotubes. The concentration ofFITC-antihuman IgG is
indentical before and after extraction, as indicated by the overlapping absorbance spectra.
This result proves that protein modified nanotubes can retain the selectivity of the protein
used for modifications.
The most specific and therefore most difficult extraction is the selective removal of
one enantiomer from a racemic mixture. Enantiomeric extraction is contingent on having
0 O c
0 0.08 -
400 450 500 550
Figure 3-5. UV-Vis spectra FITC-labled antihuman IgG solution. A) Starting
solution. B) After extraction with 1 mg/mL rat IgG-modified nanotubes.
C) After extraction with 1 mg/mL human IgG-modified nanotubes.
an extractant species that shows selectivity for one of the enantiomers. The recombinant
antibody fragment ENA IHis, developed by Dr. Hans Soderlund and coworkers, is
known to be selective for one stereoisomer of the drug
4-[3-(4-fluorophenyl)-2-hydroxy-l-[1,2,4]triazol-1-yl-propyl]-benzonitrile (FTB) (144).
As there are two stereocenters for this molecule, there are four stereoisomers, (R,S),
(S,R), (R,R), and (S,S). The (R,R) and (S,S) isomers are diastereomers to the (R,S) and
(S,R) isomers and can be separated by differences in physical properties. Structure of the
a (R,S) and d (S,R) enantiomers are shown in Figure 3-6. FTB is used for postsurgical
treatment of breast cancer. It functions by inhibiting the conversion of testosterone to
To produce enantioselective antibody fragments, Soderlund and coworkers first
injected a mouse with FTB. They then screened antibodies produced by the mouse for
their ability to bind each enantiomer of FTB using an enzyme-linked immunosorbant
assay (ELISA) (144). Once an antibody that selectively bound the a enantiomer
selectively had been identified, the DNA which coded the production of the Fab fragment
of that antibody was spliced into an E. coli host. The E. coli then produced a monoclonal
version of the enantioselective Fab antibody fragments, which was termed ENA1 IHis
(144). These enantioselective fragments were then attached to chromatographic gel
material and placed in a chromatographic column. A column containing the
ENA1 1His-modified gel retained the a enantiomer but allowed the d enantiomer to pass
Figure 3-6. 3-D structure of the enantiomers of
A) The a or (R,S) enantiomer. B) The d or (S,R) enantiomer.
ENAI IHis obtained from Dr. Soderlund was attached to silica nanotubes via an
aldehyde siloxane linker (see experimental). The antibody fragment modified-nanotubes
were used to extract racemic solutions of the FTB enantiomers. A concentration of 1 mg
nanotubes per mL solution was used for these initial extraction experiments. After
extraction and filtering to remove the nanotubes, the amounts of a and d enantiomer
remaining in solution were determined by chiral HPLC. Chromatography data were
obtained by Dr. Sang Bok Lee. Results show that the a enantiomer was indeed
selectively extracted from solution (Figures 3-7, 3-8, and 3-9).
Retention Time (min)
Chromatograms from enantioselective extractions. Absorbance was at
230 nm. A) Racemic 5.0 LM solution after extraction with 1.0 mg/mL
ENA IHis-modified nanotubes. B) Racemic 3.0 piM solution after
extraction with 1 mg/mL ENA 1His-modified nanotubes. C) Racemic 2.0
pM solution after extraction with 1 mg/mL ENA 11His-modified
nanotubes. D) Racemic 1.0 VM solution after extraction with 1 mg/mL
ENA1 His-modified nanotubes. E) Racemic 1.0 pM solution
(unextracted) for comparison. HPLC data were obtained by Dr. Sang Bok
The extraction ratio of the a versus the d enantiomer is dependant on the initial
concentration of the racemic mixture. For a solution that was initially 1 gM in both
enantiomers, all of the a enantiomer was extracted by the nanotubes, within our ability to
measure (Figures 3-7 D and 3-8 B). Approximately half of the d enantiomer was
extracted. As the starting racemate concentration is increased to 2 pM, 3 pM, and 5 pM,
~cl~l~rrylCYhSAFk~' *C~C~' '