Nanotube membranes for chemical and biochemical sensing and separation


Material Information

Nanotube membranes for chemical and biochemical sensing and separation
Physical Description:
xv, 109 leaves : ill. ; 29 cm.
Trofin, Lacramioara
Publication Date:


Subjects / Keywords:
Chemistry thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF   ( lcsh )


Thesis (Ph. D.)--University of Florida, 2005.
Includes bibliographical references.
General Note:
General Note:
Statement of Responsibility:
by Lacramioara Trofin.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 003320253
oclc - 726715280
System ID:

Full Text






Copyright 2004


Lacramioara Trofin

This dissertation is dedicated to my parents, Lucia and Nicolae Darie.


I think if any of us honestly reflects on who we are, and how we got here, we

discover a debt to others that spans nearly our whole lives.

The work and support of some people made my life easier everyday.

Some people

have directly shaped my life and my work.

I would like to acknowledge the inspiration and guidance that I received from my

Ph.D. advisor, Dr. Charles R. Martin.

He encouraged and supported my work.

He also

taught me how to effectively communicate science and successfully write scientific


I am grateful for the many suggestions and comments that I received from the

members of the Martin group. I have been fortunate to be a member of such a diverse,

team-oriented research group. Many colleagues contributed directly to my research

projects with their ideas and energies. David T. Mitchell introduced me to the field of

He taught me how to make alumina membranes, silica nanotubes and he

gave me excellent training in electron microscopy.

advice on the microbattery project.

impedance spectroscopy. I am gra

SangBok Lee gave me insightful

Marc Wirtz taught me the fundamentals of

teful to Punit Kohli for his help in the protein

separation project and for his contagious enthusiasm for science.

chatting with Zuzanna Siwy have led to fruitful developments. S

with her knowledge and energy to my scientific development. I

The many hours I spent

The contributed directly

would like to thank


Elizabeth Heins, David Mitchell, Shufang Yu, Punit Kohli, Marc Wirtz gave me ideas

and helped me in preparing the talks that I presented during graduate school. I am very

grateful to Myungchan Kang, who helped me with the microarrays project. He taught me

about plasma etching technique and fluorescence spectroscopy.

I am indebted to Miguel

O. Mota, who did his best to improve on my best. Miguel spent many hours making

alumina membranes and solutions necessary for my projects.

I also want to thank

Heather Hillebrenner who has shared her needs and experiences with me.

I have had the

wonderful experience to truly believe that in some small way I helped her; that means so

very much to me. I am also grateful to Zuzanna Siwy and Lane Baker who proofread this

dissertation and, by their feedback, they helped me structure better the dissertation.

My life during the graduate school would have been much less interesting without

my friends Isabela and Calin Briciu, Zuzanna Siwy, Cristina Cosma, Nathalie Kohli, and

Violeta Bacanu.

I would like to specially thank my boyfriend, Hermann Schulte auf'm

Erley, for his love and continuous support during my last years of graduate school.

I would like to thank to my chemistry high school teacher, Ion Dumbrava, and my

mentor, Dr. Gheorghe Ivan from CERELAST-Bucharest, for their untiring support and

seemingly unlimited belief in me.

Special gratitude is reserved to my parents, Lucia and Nicolae Darie, who have

encouraged me in so many ways to pursue my life dreams.


ACKNOW LEDGM ENTS...... ... ............... ........ ...................................... v

LIST OF TABLES..... ............ ... ........ ..... ... a.. e..... .. ... ..... .. .......... ........... ix

LIST OF FIGURES... ................. ....... .... ...... .... .......... .. ......... ..x

AB STR A C T .................................................... ...... ... .. ......... .......... X V


INTRODUCTION AND BACKGROUND .............. ........................ ..........1


B background ..... ..... ...... .....................
Membrane-Based Template Synthesis..
Membrane templates .................
Electroless deposition................

. . .. m m
C tees tee c.

* m .. m m .

. . .. .. ...3
. . . ... 3
. . . .... ..3

Applications of gold nanotube membranes
Porous Alumina Membranes......................
Two-step anodization method..............
M mechanism ...................................

* me .... ... S .. .....
.. .. ..............aacc. *...c e. .c me.
a ac. ... C5 .. .. .. ..t *m c e .e S *. e a....


Applications of alumina membranes ....... ..... ........................... .. ....16
Sol-Gel Chemistry ................. ................................. ...... .. ..... 19
Silane Chemistry ......... ............... ........... ... ................................20
Carrier Facilitated Transport ............................................................23
Single Pore Polymeric Membranes ...................................................27
Irradiation with heavy ions.............. ....................... ....................29
Ion track etching ....... ........ ..... .... .... ............ ..... ................. 30
Dissertation Overview .................. ...... ......... ..a........... ...................... 32


MICROBATTERY ...........

* e e c cm C C J k ll C C 3 C Jft Ck cc .m e, ... at 5 m C C S tt /.35

Introduction .. ... ....... ..... ..e... ... .... ... .. ..... 35
FxYnerimnPntnl .


Microbattery Fabrication ...........SE . . .......... ....... ..... . . . ... .. 37
Scanning Electron Microscopy (SEM) ...................................... ........38
Cell Assembly and Battery Discharge Measurements............... .............38
Results and Discussions...................................... ................................. 39
Characterization of the Electrode Films .......................................... .. 39
Battery Iischarge Experiments......................................................... 41
Effect of Surfactant Concentration on the Response Time.........................46
Investigations of Alkyl Chain Length on the Response Time ......................47
Conclusions.. ........ ........ ......... ............ ... .. ........ ......... ... .... ... 48


M aterials ........................... .. ............................ *.....S. .. ........ .. *...... ... 51

Fabrication of the Nanoporous Alumina Membranes.................................51

Antibody Immobilization...
Transport Experiments......
Results and Discussions .........
Effect of Antibody Affinity
Effect of the Feed Solution
Coefficient ...................
Effect of the Pore Diameter
Conclusions... ................

* C** ttt C..1q1 ec sss, qq.uss
baatcs C 11S1u1t cCC mu ... C*5 Ca

* t q c a
eq. . ..


on Selectivity Coefficient...................
Concentration on the Flux and Selectivity

on the Flux and Selectivity Coefficient.

. . . 52
. . ... 53


... ..........57

... .. ..61


Introduction. ................. ................... ..... ........... .. . . .. .... ..... .. .. .. .63
Experimental. e....... ... ...... ..s.. ..... c...... .. ... ... ...... .... .64
M ateri als... .... ...................... ............................................ ... 64
Fabrication of Porous Alumina Microarrays.......................................67

M ethod 1.............
M ethod 2 .............
Membrane Modification
Microarray Modification
Results and Discussions.......

Effect of the

s . .. q s .e . c e c s s e e.. .. ...6 7
for Sensitivity Studies ......... ...........................71
for Selectivity Studies .... ... ........ ...... ........ .....72
t..., q gs s. *. ,..,. c. a .. *,*., ., ... 73

Fabricated by Method 1............. ..... ............ ......... .......73
Made by Method 2 .. ..... ............. ..... .... .................. .74
Silica on the Sensitivity Measurements.............................75
........................ .. ... ...... C .. .76
c....... .q.q. q ..s. ......... ..t.. .. .. ,. .. .... .... ..... ...... ,7 7

C conclusions ...... ................... ... .. ........ ........... ......... .. ............... .. ..80

Introduction........ .................. .... ...... ............ .. ...... .. .. .. ... .. ..... .. ..82

Experim mental ........ ........ ..... ....... .... ... ...... .. ...
M aterials..........................c. .......... ................ .
Electroless Plating of PET Membranes........... .......
Proteins ...... .. .... e. ... ....... ........

.m. .m.. ... .. ..... .. 84
S...................... 84
.. ...... ....... .. ... .85
........ .. ... .... ..... 85

ExpCerimental Set e...... .......... ... ................. .......... ...... ................85
Results and Discussions ................ ........................................................86

......... e e c .94

6 C(ONC1LUS(IONSJ.(S..............g ...........g.g.. ..........g.c.ccg.......

LIST OF REFERENCES.........g.......................Ceeg*eam .e e aec*..e. ec........

BIOGRAPHICAL SKETCH..................................... ............... .............109








Effect of DBS concentration on the response time................................ 42

Effect ofDTA concentration on the response time............ .................... .43

Effect of alkyl chain length in alkyl trimethylammonium surfactants on the
response time .................. ................. .................... ........... 44




Scanning electron micrographs of the surfaces of the (A) alumina and (B)

polycarbonate membranes......

S S 8 SS**

Electrochemical cell setup for alumina growth........................................10

Variation of the pore diameter with the voltage applied..............................11

Variation of the pore diameter with the concentration of the electrolyte


Scanning electron micrographs of the (A) solution side and (B) barrier side of an
alumina membrane formed at 50 V in 5% oxalic acid.................................1

Scanning electron micrographs of the (A) surface and (B) cross-section of an

alumina membrane obtained at 50 V in 5% oxalic acid...

S ** S ** 9.*..**11

. .... ..13










Scanning electron micrographs of the (A) surface and (B) cross-section of a
commercial alumina membrane (Whatman) with pore diameter of 200 nm. .....14

Schematic representation of the pore formation in the porous alumina film...... 15

Steps involved in the silane chemistry.. .......... ....... .................. ........23

Typical nonlinear flux pattern for carrier-mediated diffusion. .....................25

Schematic representation of how the facilitated transport works...................26

Chemical formula of the PET (A) and Kapton (B). .................................30

Schematic representation of a conductivity cell used for chemical etching.......30

Schematic representation of the microbattery fabrication.........................38

Schematic representation of a U-tube permeation cell...............................39

Cross-sectional (upper) and surface (lower) SEM images of the battery electrode

EDS spectra of the membrane surface after deposition of Ag (A) and after

conversion of the Ag surface to AgCl (B).

EDS spectrum of the opposite

membrane surface that had been coated with Zn (C). ................................41

Current-vs.-time response for the transmembrane microbattery applied to an
alumina membrane that was not rendered hydrophobic by silane functionalization
(A). Analogous current-vs.-time response for the hydrophobic C18-modified
m em brane (B ) .......................... ........... .... ..... .. .... .. .... .. ............. .... .... 43

Current-vs.-time response for a hydrophobic membrane before and after injection


Scanning electron micrograph of a porous alumina membrane with pores of 50
nm in diameter and pore density of- 1010


Modification steps involved in antibody

S... .. 53

Transport plots of GFP-Hevein and RFP through ENA 11 His (A), 1 C2 (B) and

1A4 (C) nanotube membranes ...................................
Effect of the antibody immobilized on the selectivity .......
Plot of fluxes of GFP-Hevein and RFP versus feed solution

concentration ..........................

C *C C. C.55

C p

. . . . . ..57

Transport plots of GFP-Hevein and RFP through a 1A4 antibody-modified
membrane when using 5 nM (A), 20 nM (B), 50 nM (C) and 100 nM (D) feed

solution concentration................

*SSSS SSCC 55S* *S.*S SW CS*S e*S* ** ** CS* Se m... C CC .5

Variation of the selectivity coefficient with feed solution

Transport plots obtained when used membranes with pores of 50 nm (A), 70 nm
(B) and 100 nm (C) in diameter ...................................... ................. .60
Selectivity coefficient variation with the pore


. ..... .. ... ...61

Schematic representation of the microarray fabrication by method 1...............66

Schematic representation of the microarray fabrication by method 2...............68

Electrochemical cell setup for silver electrodeposition: A, Ag wire counter and
reference electrode; B, Ag plating solution; C, Cu foil; D, Au-Pd modified
alumina membrane as working electrode; E, stainless steel plate; F, teflon tape; G,
O -ring seal .... ....... .................... .......... .......... ... .... ... ...... ... ...69

Modification steps for sensitivity studies .................................................71

Scanning electron micrographs of the porous alumina microarrays fabricated by
method 1 at a low (A) and higher (B) magnification.............. ...................72


of DBS surfactant solution ................... .........,......., .....







Fluorescence spectra for a rhodamine B-APTES-alumina sample with (green) and
w without (red) silica ....... ............................ ...................... ...............76

Relative fluorescence coefficient for rhodamine B-modified membranes of
different thicknesses...... ................. .................................................77

Optical (left) and fluorescence (right) image of a 350 pm x 350 pm area of the
microarrays after immobilization of the target proteins.............................78

Excitation with 495 nm light: A, 2D fluorescence image; B. 3D fluorescence
image; C, fluorescence intensity profile ........... ........... ......... .............79

Excitation with 590 nm light: A, 2D fluorescence image; B. 3D fluorescence
image; C, fluorescence intensity profile................. ......................


Sensing lysozyme with a single conical gold nanotube. (A) Current-voltage
characteristic of the Au nanotube before (red points) and after modification with
thiolated biotin (blue points). The diameters of the pore opening are 5 nm and 0.6
p-m, respectively. (B) Ion current versus time through the Au nanotube modified
with biotin recorded at 1 M KC1, pH 7. (C) Ion current versus time as in (B) at
presence of 100 nM lysozyme in contact with the small opening of the pore....87

Sensing streptavidin with a single conical gold nanotul
characteristics of a single conical Au tube modified wi
180 pM streptavidin added on the small side of the con
current in time through a single Au nanotube modified
M KCI, pH 4.5, recorded at -1000 mV. (C) Ion current
presence of 180 pM streptavidin...........................

Blockage time vs.

be. (A) Current-voltage
th SH-biotin at presence of
lical nanotube. (B) Ion
with biotin, recorded at 1
in time as in (B), at

. ....... 88

-log of the molar streptavidin concentration ...................90

Chemical modifications of a
dimensional nanoimmunoass
interactions with protein G.
modifications of a single Au
1 M KC1, pH 7, performed a
modifications has diameters

single Au tube leading to preparations of 3
say for detection of IgGs and probing their
(A) Schematic representation of the subsequent
tube. (B) Current-voltage characteristics recorded at
after each modification step. The gold tube after
of-I15 nm and 0.6 pm, respectively..................91


Sensing of cat IgG with a single conical Au nanotube modified with protein G as
shown in Fig. 4. (A) Current-voltage characteristic of a single Au tube recorded at
1 M KC1, pH 8.7. Ion current in time recorded at 500 mV transmembrane
potential before (B) and after (C) adding 100 nM cat IgG. The gold tube after
modifications has diameters of-15 nm and 0.6 pm, respectively..................92

Sensing of horse IgG with a single conical Au nanotube modified with protein G
4 a a^ t aj 4 4 P, fl4 a**r .*j 4 4* .

5-7. I-V curves for the ricin sensor in the presence of no protein (x), 100 nM BSA( ),
and -100 nM ricin (e). ............. 94.................. ...............94

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



Lacramioara Trofin

May 2005


Charles R. Martin

Major Department:


The discovery of novel materials, processes, and phenomena at the nanoscale, as

well as the development of new experimental and theoretical techniques for research,

provide fresh opportunities for the development of innovative nanodevices and

nanostructured materials.

Nanostructured materials can be made with unique

nanostructures and properties, and finding various and unique applications of these is a

continuous challenge for researchers in this field.

As part of this emerging research, the

work presented here is focused on development of new nanostructures based on

nanoporous membranes, and investigation of their applications as sensors and separation


A template synthesis method is used to produce nanotubes inside the pores of

both aluminum oxide and polymeric membranes.

After an introduction in the template

synthesis method and the processes of fabrication of the porous membranes, the

dissertation is centered on investigating new applications of these nanotube membranes.

There are three applications of the nanotube alumina membranes, and one application of

the single nanotube polymeric membranes that are explored.

Porous alumina is used to mimic the function of the ligand-gated ion channel by

applying a porous battery cathode film to one face of the hydrophobic membrane and a

porous battery anode film to the other face.

Hence, in analogy to the naturally occurring

channel case, we have a membrane with a built in electrochemical potential difference

across the membrane.

The application of silica nanotube membranes in selective separation of proteins is




The membranes were modified with antibodies that selectively bind one

These nanotube systems lead to the transport at much higher rates of the analyte

which binds to the membrane.

Another application studied is the fabrication of the protein microarrays which

features a three dimensional substrate based on porous aluminum oxide membranes.

These membranes represent distinct microfeatures on a robust platform, and they have

cylindrical pores with monodisperse nanoscopic diameters.

The advantages of using this

substrate and its application in antibody specificity screening are presented.

Finally, a new family of protein biosensors based on a single conical nanotube

membrane is described.

Three different protein systems were investigated: (i)

biotin/streptavidin, (ii) protein-G/immunoglobulins, and (iii) anti-ricin/ricin.



Nanotechnology refers to technologies in which matter is manipulated on the

atomic and molecular level to create new materials and observe new processes.

It is not

just the study of the very small; it is also the practical application of that knowledge.

Nanotechnology is a truly interdisciplinary field.

Materials scientists, electronic and

mechanical engineers, as well as medical researchers are working together with

biologists, physicists and chemists.

Research at the nanoscale is unified by the need to

share the knowledge and expertise required to work at the atomic and molecular level.

Powerful new concepts and capabilities, such as atomic-scale imaging and manipulation,

self-assembly and biological structure-function relationship, together with increasingly

powerful computing tools are rapidly converging from different research areas.

Nanotechnology is not a new area, though.

Mother Nature serves as a model for having

many materials and processes that functions at the nanoscale (1); small molecular

building blocks are joined together to produce nanostructures with defined geometries

and functions.

The top-down approach becomes increasingly difficult, as the final

products approach the nanometer levels.

It has become evident that Nature's bottom-up

approach can be emulated to produce new materials with nanosized dimensions and

engineered properties.

KT t a, .. a a4 n 1 n-^ n j (,j / ) \ n \ n 4 r. .' a:,, 4. .L I La a I a a a a c

silver particles.

More recently, a wider variety of nanoparticles have been synthesized;

for example, commercially available magnetic beads are used for cell preparation (6,7),

quantum dots are used for long-term fluorescence assay in cells (8), and colloidal gold

has been used for gene therapy (9).

Since the discovery of carbon nanotubes in 1991

(10), the synthesis and functionalization of the nanotubular materials has become one of

the most highly energized research areas (11).

Nanotubes have numerous potential

commercial and technological applications, including their use in nanoelectronics

(12,13,14,15), catalysis (16,17,18), hydrogen storage (19), scanning probe microscopy

(20), biosensors (21,22) and drug delivery systems (23).

Research in the field of nanotube membranes will have a great impact on

membrane technology.

Membranes are utilized to perform separations for a wide range

of applications such as water and wastewater treatment, electrodialysis, gas separation,

and fuel cell development (24).

The use of membranes and biological tools are important

in the development of biomedical and biotechnological applications (25).


membranes have been gaining attention as options for biological sensors (26).


modem biotechnology and separation science have presented new challenges to

membrane technology, including the requirement of pores with diameters similar to those

of molecules under study, therefore as small as several nanometers.

Nanometer scale

pores are necessary in achieving optimal control of the flow of biomolecules as well as in

developing sensors for their detection (27-30).

Another challenge is the development and

characterization of membranes possessing well-controlled, stable, and uniform nanometer

dimension pores capable of the separation and sensing of molecules in a restricted


In light of these challenges, Martin's group has pioneered a bottom-up method for

the production of nanotube membranes, called template synthesis (31). This method

involves synthesizing nanotubes inside of a porous membrane (template). This chapter

provides background information on the following: membrane-based template synthesis,

fabrication and applications of porous alumina (one type of membrane template), sol-gel

and silane chemistry, carrier facilitated transport, and single pore polymeric membranes.

An overview of the dissertation is also presented.

This information will be used in the

next chapters.


Membrane-Based Template Synthesis

In recent years, the Martin group has been investigating a versatile method to

produce nanomaterials, called template synthesis (31).

In this approach, a membrane

with uniform dispersed, micro or nanometer diameter pores acts as a template.


material is deposited into the cylindrical pores of the membranes, it adopts their shape.

the template is dissolved, the material can retain the high aspect ratio of the pores,

yielding wires or tubes with nanometer diameters.

the type of materials that can be prepared. For ins

The method is versatile with regard to

stance, track-etched polymeric

membranes have been used to prepared nanostructures composed of metals (32,33),

insulating polymers (34) or conductive polymers (35,36).

Membrane templates

Two types of template are most often used for this approach: polycarbonate and

alumina membranes.

Polycarbonate membranes are prepared by the "track-etch" method

pore densities approaching 109 pores/cm2

electrochemically from aluminum foils (38

International), or can be prepared in the lab

Alumina membranes are prepared

). They are commercially available (Whatman

oratory. The process of making alumina

membranes and their applications will be discussed in detail in the following section.

Figure 1-1. Scanning electro micrographs of the 3 pim pore diameter polycarbonate (A)


Figure 1-1 shows scanning electron micrographs of the surfaces of the

polycarbonate and alumina membranes, respectively.

Templates with diamond shapes

pore in mica have also been reported (39).

The sensing and transport properties of the gold nanotube membranes prepared by

the electroless deposition method (40) were investigated extensively in the Martin group


Electroless deposition

Electroless deposition involves a chemical reducing agent which is used to plate a

metal from a solution onto a surface.

be summarized as follows. The men

solution of SnCl2.

The method for electroless deposition of gold can

ibrane is first "sensitized" by exposing it to a

This results in deposition of Sn" onto the membrane surfaces and the

pore walls. After the sensitization, the membrane is immersed into an ammonia silver

nitrate solution. A surface redox reaction occurs (Equation 1-1) and Ag' is reduced by

, which results in absorbtion of Ag nanoscopic particles on the membrane surfaces.

Sn~ + 2Agf01

- SnV + 2Ag>

The subscripts "surf" and "sol" denote species absorbed to the membrane surfaces

and species in solution, respectively.

plating solution at 40C.

Then, the membrane is immersed into a gold

A second redox reaction occurs, and Auo displaces the Ag

particle, yielding the membrane surfaces to be coated with Au particles (Equation 1-2).

Au' + Ago,


-> Auo + Ag\o

These Au particles are excellent autocatalysts for the reduction of Au' to Au,


Applications of gold nanotube membranes

Gold nanotube membranes are a new class of molecular filters, capable of sensing

and transporting both small and large molecules.


Jiraje et al. took advantage of the

charge density present on the inner walls of the gold nanotubes and showed the

regulation of ion transport through the membranes (40).

They showed the fluxes of

anionic and cationic permeates changed with the potential applied to the gold nanotube

membranes. The tubes transport ions which have the opposite charge as the gold

nanotubes (40). Because the inner diameter of the gold tubes can be of molecular

dimensions (<1 nm), nanotube membranes have also been used to separate small

molecules on the basis of molecular size (41).

In these experiments, a large molecule, (a

tris-bipyridal complex of Ru(In Ru(bpy)32+), and a small molecule (methyl viologen

MV2+), were used.

A selectivity coefficient was defined as the ratio between the fluxes

of MV2+

Ru(bpy)32+ through the membranes.

They report a selectivity coefficient

of 50 when the inner diameter of the gold tubes was 5.5 nm.

They have also showed that

as the inner diameter decreases, the selectivity coefficient increases, reaching a value of


nm inner diameter (41).

The gold nanotube membranes were also used to

study the DNA transport, both by diffusion and electrophoretically.

The flux of the

single-stranded homooligonucleotides made of thymidine bases (poly(Tn) where n

represents the number of bases decreased as the size (base number) of the poly(Tn)

increased (42).

Another way of introducing chemical and biochemical transport selectivity is by

adsorbing thiols on the gold nanotubes (43.44.451.

Hvdronhobic thiols vield membranes

gold tubes was used to make pH-switchable ion transport membranes.

Depending on the

solution pH, the membrane can have excess positive charge (low pH), no net charge

(isoelectric point ), or excess negative charge (high pH ).

As a result, these membranes

can be switched between cation-, non-ion-permselective, and anion-transporting states


By controlling the inner diameter of the gold tubes, these membranes can also show

good selectivity for transport of proteins on the basis of molecular size (46).

In this work,

chemisorbtion of a PEG-thiol prevented the non-specific adsorbtion of the protein on the

gold tubes.

A transmembrane pressure was applied to the feed solution to force the

solution through the membrane.

The effect of nanotube diameter on the flux and

selectivity for lysozyme, bovine serum albumin and P-lactoglobulin A was investigated


Recently, through the immobilization of molecular recognition elements, gold

nanotube membranes were used to obtain

selectivity (47).

DNA single base mismatch transport

Single-stranded DNA molecules with a thiol at one end were

chemisorbed on the inner walls of the tubes.

These DNA functionalized tubes selectively

recognize and transport the DNA sequences which are complementary to the DNA on the

tubes, relative to the uncomplementary DNA sequences.

Porous Alumina Membranes

Electrochemical oxidation (anodization) of aluminum surfaces under controlled

conditions can produce aluminum oxide or alumina with a structure of essentially

cylindrical, parallel pores (48).

Anodic porous alumina membranes can be made with

pore diameters varying between few nanometers to 200 nm, with lengths up to 300 urm

I_.flX. -1L: r.-. ,-d2ln1

r(-1.. LI-.~- L~ l*f..L,,,: 3',

/ Af \


solutions (e.g. phosphoric acid (49), oxalic acid (50) and sulfuric acid) are used as

electrolyte (51).

In recent years, there has been a growing interest in preparing alumina membranes

with a perfect pore array architecture, having a high aspect ratio at the nanometer scale.

This interest was drawn by the possibility of applying these membranes as hosts or

templates for the fabrication of the nanodevices.

There are two reported methods for the fabrication of highly ordered anodic porous

alumina membranes.

One is the two-step anodization method which will be discussed in

detail in the next section; this is the method that is used in our laboratory to make the

alumina membranes (52,53).

In the other method, the layout of the initiation sites for

hole development in anodic alumina is achieved by a process based on nanoindentation

of the aluminum substrate.

In this process, an array of shallow depressions is formed on

aluminum by indentation, and these depressions serve as initiation sites for hole

generation at the initial stage of generation.

Masuda et al. (54) used a SiC mold to form

an array of concave features with the desired arrangement (square, triangular) on


In addition, Mikulskas et al. (55) showed that the nanoindentation twice with

commercially available optical grating rotated by an angle of 600 to each other, can create

pre-structures with rombohedral ridges on aluminum.

The advantage of this patterning is

that it eliminates the high cost of the mold, which requires electron beam lithography to

produce it.

Masuda et al. (56) reported another patterning method, by using a

nanoindentation apparatus attached to a scanning probe microscope.

Twn-sten anodization method

Polishing of aluminum foils. In order to obtain a high quality alumina film, the

starting material, aluminum, must be very smooth. High purity aluminum foils (99.99 %)

are first mechanically polished with a slurry of alumina particles.

Larger particles

(>10pam) are used to remove material fast, and polishing is continued with slurries of

progressively smaller particles of submicron size.

If the aluminum foils have severe

scratches, mechanically polishing with fine sand paper is applied until the scratches


This mechanical polishing is followed by an electrochemical one.

In this

process, a potential difference of 15 V is applied between the aluminum foil (which

serves as the anode) and a lead plate which serves as the cathode.

The polishing solution

(95% concentrated phosphoric acid, 5% concentrated sulfuric acid and 20 g/L chromic

oxide) is heated to 700C.

This process is analogous with that of alumina film formation,

but the electrolyte, being a very concentrated acidic solution at high temperature, favors

immediate dissolution of alumina.

The electropolishing step (usually 5 minutes) is

repeated as many times as it is necessary, until the aluminum foil has a mirror-like


Anodization steps.

Electropolished aluminum is electrochemically oxidized into a

first step at a constant voltage, using an electrochemical cell setup like the one presented

in figure 1-2.

In the process of forming the alumina film referred in the literature as the growth

process, the aluminum is the anode and a stainless steal plate is used as the cathode.

anode and cathode are immersed into an electrolyte solution and a voltage is applied

jicino a rn,-^r ciinnlrl


Th ftemnnratlire /(lrol/llx; hehl xin (Y sanl 1 0 \ i rnntmnllrl ilcinoa


strongly on experimental conditions, such as temperature, voltage applied, type of

electrolyte and concentration of the electrolyte (57-59).

Smaller pore sizes require lower

applied voltages and therefore more conductive electrolytes (such as sulfuric acid) (48).
Larger pores need larger voltages, which causes a high rate of dissolution in highly
conductive electrolytes (62).








Figure 1-2. Electrochemical cell setup for alumina growth.
Thus, the formation of large pores will require lower conductivity electrolytes, such as

oxalic acid.

The pore diameter in the grown alumina films varies in direct proportion to

the voltage applied and the concentration of the electrolyte.

Figure 1-3 shows the

dependence of the pore diameter on the voltage applied, when using 5% oxalic acid as


electrolyte solution, when 50 V was applied.


The thickness of the formed alumina film

depends on the anodization time; longer anodization times yield thicker membranes.

After the first anodization, alumina film is removed in a solution which is 0.4 M in

phosphoric acid and 0.2 M in chromic oxide at 600C.

The removal of alumina film leaves

behind aluminum with a hexagonal scalloped pattern, due to the self-organizing into a

hexagonal arrays of the pores during the anodization (50).

That is, the removal of the

alumina leaves indentations, or pits, in the aluminum that correspond to each pore.

1 2 3 4 5
Oxalic Acid Concentration, %

Figure 1-3.

Variation of pore diameter with applied voltage.

140 -
100 -


To obtain this pattern on aluminum, the duration of the anodization process must at

least 12 hours.

The pre-pattemed aluminum is then re-anodized in exactly the same conditions,

which were used in the first anodization step. The por

already highly ordered and monodisperse (52). The se

out for a length of time, depending on the thickness of

es nucleate in the pits which are

cond anodization step is carried

alumina membranes desired.

our laboratory, we obtained alumina membranes of thicknesses between 0.3 and 150 p.m.

Detaching. After the second anodization step, alumina can be detached from the

unoxidized aluminum by two methods.

a saturated solution of HgC12. One sid<

One method utilizes dissolving the aluminum in

e of the formed alumina membrane, which faced

during the growth the electrolyte solution has open pores while pores on the other side

are closed (60).

The two sides are called "solution" and "barrier" side, respectively.

Figure 1-5 shows the scanning electron micrographs of the solution side and barrier side

of an alumina membrane obtained in our laboratory.

4r- J J f 4 -r 6 t 4 f


This barrier layer can be removed by etching of alumina films for very short time in

dilute acid or basic solutions.

The second method for detaching the alumina films is

known as the "voltage reduction process" (61).

This process uses a progressive reduction

of the applied voltage until it reaches 4-5% from the initial value.

Because the pore

diameter is directly proportional to the voltage applied, the resulting pores branch down

to smaller sizes.

The anodization is then stopped and the aluminum/alumina system is

placed into an etching solution, which can be a dilute acidic or basic solution.

The thin

barrier layer and the small pores dissolve faster, resulting in detachment of the alumina

from the aluminum.

In our laboratory, using two-step anodization method, we obtained highly ordered,

porous alumina membranes. The pores are very uniform throughout the whole thickness

of the membrane, as can be seen in Figure 1-6.

.. .l... .

******** e* ...:. ---

Figure 1-6. Scanning electron micrographs of (A) surface and (B) cross-section of an
alumina membrane obtained at 50 V in 5% oxalic acid.

Alumina membranes with pore diameters of 200 nm are commercially available

frrnm Whntman Tnternmtinnnl

PnreA nf 1 Oi nm nnd 90 nrm in rliameter arp alAn nvnilahle


on one side the pores have these diameters on a length of 200 nm; on the other side the

membranes have pore with 200 nm in diameter. Figure 1-7 shows scanning electron

micrographs of the (A) surface and (B) cross-section of a commercially available alumina

membrane with a pore diameter of 200 nm. Compared to the home-grown membranes

(Figure 1-6), these membranes have pores which are not uniformly distributed throughout

the membrane and are characterized by much more heterogeneous diameter.


Figure 1-7. Scanning electron micrographs of (A) surface and (B) cross-section of a
commercial alumina membrane (Whatman) with pore diameter of 200 nm.


The formation of a highly ordered pore array in the alumina membranes is the

result of two competing processes: 1) the pore initiation process due to a geometric

effect, which is called "field assisted dissolution process" (52) and 2) the self-

organization of the pores which is thought to be driven by mechanical stress at the

alumina/aluminum interface (62).

In the pore initiation process, the alumina film developed at the metal-film-

electrolyte interface, at preferred sites (small pits or defects), undergoes dissolution,

outer film surface, which are the precursors of the pores.

effectively polarizes the Al-O bonds, allowing more A13

of the field.

Field-assisted dissolution

dissolution than in the absence

As a consequence of the pore development, the electric field and the ionic

current become concentrated in the barrier layer beneath the major pores.

This implies

continued migration of the 02/0H' ions from the electrolyte to form a solid film at the

metal-film interface and corresponding A13+

as well as field-assisted dissolution of A13 i

ejection at the pore base-electrolyte interface

ons (59). This mechanism of pore nucleation

is illustrated schematically in Figure 1-8.



Figure 1-8. Schematic representation of the pore formation in the porous alumina film.

The pore initiation process is followed by a steady-state film formation. In this

state, there is a dynamic equilibrium between film growth at the metal-film interface and

field-assisted dissolution at the pore base-electrolyte. The mechanical stress, a possible

origin of repulsive forces between neighboring pores, is associated with the expansion of

the aluminum during oxide formation (62). This leads to the self-organization of the


Applications of alumina membranes

Due to the packed array of columnar hexagonal cells with cylindrical, uniformly

sized pores, porous alumina membranes have been used to fabricate many types of

nanocomposites using the template synthesis method.

For instance, template pores were

filled with metals or semiconductors used for the preparation of magnetic recording

media (63,64), optical devices (65), functional electrodes (66,67), electrochromic (68),

and electroluminescence display devices (69,70).

The outside diameter of the

nanocomposites is determined by the pore diameter of the membranes, and the length of

the nanocomposites is controlled by the thickness of the membranes.

Natan and co-

workers synthesized submicrometer metallic barcodes by alternating Au and Ag

segments along the length of a nanowire (71).

For sensing and nanoelectrode

applications, the nanowires can remain in the template and function as an array.

single-nanowire applications, removing the template produces individual nanowires that

can be isolated.

Porous alumina has also been used as a template to make Au, Ni and Si

nanoring arrays by a sputtering redeposition method (72).

The channels of alumina membranes were used to produce a new kind of artificial

lipid membrane system.

In this system, lipid bilayers were immobilized on the surface

(73) and inside of the pores of the membranes (74), creating a platform with potential

applications for biosensing.

Porous alumina membranes were also used as a support to incorporate metal

clusters, or colloid particles (75).

This opens the way to new applications, such as

chemical complexation inside the membranes of radioactive organo-metallic compounds

Lahav et. al. reported a procedure to make metal "nanoparticles nanotubes" that

combines nanotube geometry with nanoparticle morphology and properties (76).

When the alumina membranes are used as templates to make nanotubes, an

important issue is controlling the inside diameter of the formed nanotubes.

This problem

has been approached

by the layer-by-layer film deposition process.

In this method, films

of materials are deposited layer-by-layer on the pore walls of the membranes to make


The resulting inner diameter of the nanotubes is dictated by the thickness and

the number of film layers deposited.

polyelectrolytes (77).

Using this method, Ai, et al. deposited layers of

Kovtyukhova et al. have also used a method based on alternate

SiC14/H20 deposition cycles to make silica nanotubes (78), and Hou, et al. used

Mallouk's alternating a,wo-diorganophosphonate/Zr chemistry (79) to prepare nanotubes

within the pores of alumina template membranes (80).

Highly selective silica nanotube membranes can be used as both sensors and as

molecular filters (81,82,83).

nanotube membranes (84).

A sol-gel template method was used to prepare the silica

In the next section, a detailed review of the chemistry

involved is presented.

Two applications of the silica nanotube membranes are their use for biological

extraction and for biocatalysis.

In this procedure, silica nanotubes were removed from

the membrane by dissolving the template, and collected by filtration.

They were

functionalized with octadecyl silane on the inside, resulting into a hydrophobic nanotube

interior, while the outside was left unfunctionalized, giving an hydrophilic nanotube


These hvdrophilic/hvdroohobic nanotubes were used successfully to extract


solution of 7,8-benzoquinoline, a lipophilic compound that preferentially entered the

hydrophobic interior of the tubes.

removed from the solution (81).

In this way, more than 90% of the compound was

In the same work, Mitchell et al. showed that

enantiomers of a drug can be separated using a suspension of nanotubes (81).

In this case

the nanotubes were functionalized with an antibody that binds the RS isomer of the drug

over the SR isomer.

These nanotubes successfully extracted 75% of the RS isomer from

a 20 ptM racemic mixture, and all of the RS isomer from a 10pM racemic mixture.

Another application of the silica nanotube membranes is their use in bioseparation.

Lee and co-workers also looked at the separation of RS and SR enantiomers of a drug


In this case, the membrane was not removed, and the silica nanotube membranes

were modified again with an antibody that selectively binds the RS isomer.

They found

that these membranes facilitate the transport of the RS enantiomer, as the RS flux was

twice the SR flux, for nanotubes with an inner diameter of 35 nm.

It was also shown that

the binding affinity of RS over SR could be tuned by addition of DMSO to the protein

buffer solution, leading to an optimal DMSO concentration that maximized the


The selectivity could be further enhanced by decreasing the silica nanotube

diameter, yielding a selectivity of 4.5 when the nanotube diameter was 20 nm.

Yamaguchi and co-workers have reported a method to form a hybrid membrane

composed of silica-surfactant nanocomposites inside a porous alumina membrane, which

functions as a nanometer-order size-exclusive separation (83).

In this work, they added a

precursor solution of TEOS (tetraethoxyortosilicate) and CTAB

(netvltrimethvlnmmnniuim hrnmide' tn the aluminn memhrann raevltino in the


were able to separate two small protein molecules, myoglobin and bovine serum albumin,

(diameter > 4 nm) from two other smaller molecules, rhodamine B and vitamin B12,


< 2.4 nm).

Nanorods made in alumina membranes were used as gene delivery systems (85).

Leong and co-workers fabricated nickel-gold nanorods by template electrodeposition in

alumina membranes. Transferrin, an iron transport protein, was bound to the Au

segments by a thiol linkage. They served to promote cellular uptake of the rods by a

receptor mediated pathway. The Ni portions were functionalized with DNA plasmid that

contained a fluorescent reporter gene. They showed that the nanorods were internalized

by the cell, but they did not enter to the nucleus. These nanorods have two functions, one

to target the cells and to deliver the DNA. In the nucleus, green fluorescence was

observed, indicating the delivery of the reporter gene into the nucleus.

Porous alumina membranes in a tubular shape were made when the aluminum to be

oxidized was purchased in the form of wires or cylinders.

were used in studies of catalysis (86),

These porous alumina tubes

and as drug delivery systems (87).

Sol-Gel Chemistry

Sol-gel chemistry is a powerful method to generate inorganic materials. It

originated in the 1970's, as scientists attempted to find low temperature routes to glass

synthesis (88,89).

The high temperatures (1300 to 20000C) needed to form glass are a

result of the need to destroy the crystallinity of the precursors; that is, even the glass is an

amorphous material, it is generally made from crystalline oxide precursors.

method to use noncrystalline precursors was searched for.

As a result, a

It was found that liquid

R3Si-O-R+H20 R'Si-O-H +R-OH


The silanols than can undergo further polymerization reactions with another

silanols or other alkoxysilanes:


- R3Si- O -SiR3




-- R3Si- O SiR3+R-OH


In both cases, the result is formation of a three-dimensional siloxane network.

the start of the polymerization, many small siloxanes particles are formed.

well dispersed in the liquid phase and form colloids.

They are very

When the particles are well isolated

from each other, the density of the suspension resembles that of the solvent.

stage, the colloidal form is called a "sol"

At this

. As polymerization continues, the particles

increase in size and the viscosity of the solution increases.

The particles develop a three-

dimensional network throughout the solution, which is named a "gel" (90).


solution pH, water concentration and the type of the alkyl group are parameters that

influence the rate of hydrolysis and polymerization (91).

with the temperature.

The rate of gelation increases

The hydrolysis reaction is very slow and entails the replacement of

alkoxy groups with hydroxyl groups.

or base catalyzed (92).

It is much faster when the reactions are either acid

A larger, more sterically bulky alkyl group slows down the

reaction rate (92).

There are two ways of converting gels to silica. In the first method, the gel is

heated or placed under vacuum to remove the solvent phase. The open three-dimensional

structure collapses, condensing it into a dense phase called a xerogel (91).

In the second

material, called an aerogel (93).

The maximum temperature for both processes can be

kept below 1000C.

Silane Chemistry

Siliceous surfaces (e.g. silicates and aluminates) can be derivatized with a large

variety of different functional groups using silane chemistry.

covalent bonding with these surfaces (94).

Organosilanes form

The general formula of an organosilane is

RnSiX(4-n), where X is a hydrolyzable group capable of forming strong covalent bonds

with the hydroxyl groups on silica surfaces such as halogen, alkoxy or aciloxy.

group is a nonhydrolyzable group that may posses a desired functionality (95).

most used types of organosilanes for surface modification are chlorosilanes and


The R

The two

The attachment chemistry of both types is equivalent because when the

chlorosilanes are dissolved in alcohol, they react with the alcohol to form alkoxysilanes:




The extent of this reaction can be monitored by measuring the pH. Alkoxysilane

chemistry is analogous to the sol-gel formation chemistry. Silanes with one hydrolyzable

group can be utilized to produce monolayers on the surfaces. Because there is only one

reactive group, the silanes can either bind to the surface or dimerize. Dimers cannot bind

further and can be washed away.

hydrophobic surfaces (95). Whe

The silanes with one hydrolyzable group yield to

;n surfaces with a higher degree of coverage are desired,

silanes with two or three hydrolyzable groups are used.

These are first allowed to

oligomerize in a slightly aqueous alcohol solution (water content is typically 5% vol/vol)

alkoxysilanes (94).

The surface to be modified is than added to this solution and the

oligomers bind to the surface through the surface hydroxyl groups (coupling step).

Figure 1-9 shows the reactions involved in the hydrolysis and coupling steps.

modification by this route requires only a few minutes of immersion.


If a 2% of

trialcoxy- or trichlorosilane solution is used, the resulting modified surface is normally 3-

8 monolayers thick (95).

The silanized surfaces must be cured usually at 1200C for 30 minutes, or for 24

hours at room temperature.

Chlorosilanes can also be deposited from aprotic solvents, such as toluene and

tetrahydrofuran (94).

If these solvents are anhydrous (and the surface is free of water),

then no alkoxysilane can be formed, and no polymerization of the silanes can take place.

The reaction must proceed by the nucleophilic attack from surface hydroxyl sites, and as

a result, only one monolayer can be formed (94).

Surface modification in these cases

takes longer time, usually 12-24 hours.

Step 1. Hydrolysis

y/\V\S -OR

+ 3H20


+ 3ROH

Step 2. Coupling

Si Si O Si Glass surface

Si -O- Si-O-Si

0 H OH





Figure 1-9 Steps involved in the silane chemistry.

Carrier Facilitated Transport

In biological systems, the simplest mechanism by which molecules pass through a

plasma membrane is passive diffusion (116).

During the passive diffusion, a molecule

simply dissolves in the lipid bilayer, diffuses across it and passes to the other side of the


Larger, uncharged, polar molecules, such as glucose, and charged molecules

of any size, like small ions K Na

passive diffusion.

, Cr, are unable to cross the plasma membrane by

Their passage through the membrane is mediated by proteins that

enable the transport of molecules and ions through the plasma membrane without

Carrier proteins bind specific some molecules that are transported.

They then undergo

conformational change that allows molecules to pass to the other side of the membrane.

Channel proteins form open pores through the membrane, allowing free diffusion of any

molecule of the appropriate size and charge.

In biological systems, this phenomenon is

called facilitated diffusion.

By the 1950's, scientists started to develop synthetic analogs of natural systems that

function on the base of facilitated transport concept.

Facilitated transport that uses a

chemical completing agent immobilized into a synthetic membrane has been the subject

of numerous articles (96-104).

Facilitated transport was accomplished in liquid

membranes, in which two aqueous phases were separated by an organic solvent

containing carrier molecules, such as crown ethers.

Diffusion of metal ions in liquid

membranes is governed by the complex formation with the carrier at the aqueous/organic

interface and the selectivities are generally very high.

The fluxes in the membranes are

very low because they are limited by the convection of the carrier in the organic phase


The biggest disadvantages of using liquid membranes are the low fluxes, the

leaching of the carrier and the poor physical stability.

These problems have prevented

wide-scale application of liquid membranes in industrial separations.

Polymeric facilitated transport has been applied more recently to eliminate liquid

loss. Tsuchida and co-workers have published results based on various metalloporphyrins

cast into polymer films for selective 02 transport (105,106) and N2 transport (107).

Yoshikawa et al. described CO2 transport in these fixed site carrier membranes (108).

Polymeric facilitated membranes have been applied also to ion separations.

For instance,

potassium from sodium and rubidium (97), and potassium from lithium (101).


transport of metal ions resulted in good selectivity with marked improvement in

membrane stability as compared to liquid membranes.

Some other applications include

transport and separation of ethane and ethane through Nafion membranes (109), olefin

(110) and small carbohydrate separation (111).

The first facilitated transport-based system described was that of Scholander (112),

who showed that oxygen diffusion through a filter paper membrane containing a

hemoglobin solution was enhanced.

Oxygen diffusion through a membrane that has been

soaked in methemoglobin, which has no carrier-oxygen binding capacity, showed the

same low diffusion rate that would be expected for simple diffusion of oxygen through


However, in the presence of hemoglobin an additional amount of oxygen is

carried through the membrane due to the hemoglobin-oxygen complex formed in the

The increase in oxygen transport due to the action of hemoglobin as a

carrier is termed the "facilitation effect" (114). Figui

flux pattern for carrier-mediated facilitated diffusion.

Enhanced diffusior

Fl ux

re 1-10 shows a typical nonlinear




Passive diffusion


membrane (113).


As can be seen in Figure 1-10, facilitated diffusion occurs at low concentration of

feed solution. Facilitation effect reaches a maxi

carrier molecules which binds the analyte (115).

mum value due to the saturation of

The basic mechanism for this enhanced

transport is a reversible reaction (see equation 1-7) between an analyte molecule A,

which can enter the membrane phase, with a carrier B, which is immobilized on the

membrane (114).

A+B <=AB


In this process, both the chemical reaction and diffusion occur simultaneously in

the system, resulting in an accelerating transport of the permeate species, A, through the


Membrane Cross-


Concentration gradient in the
case of facilitated diffusion

Concentration gradient in the
case of passive diffusion


Figure 1-11. Schematic representation of how facilitated transport works.

A schematic explanation of how facilitated transport works is shown in Figure 1-


concentration of the immobilized carrier and CAp is the concentration of the permeant A

in the permeate side. According to Fick's

first law of diffusion, the flux of a permeate

molecule across a membrane is directly proportional to the concentration gradient across

the membrane.

One-dimensional representation of the Fick's law (see equation 1-8) has

the following form:



In equation 1-8, J represents the flux, D is diffusion coefficient and represents

concentration gradient of a permeate molecule across the membrane. Inside the

membrane the concentration of the immobilized carrier is higher than the concentration

of the permeate molecule in the feed side.

Due to that the concentration gradient for the

permeate molecule that react with the carrier is higher than the concentration gradient of

a permeate molecule that does not interact with the carrier and is transported by passive


This results in an enhanced flux of molecules that interact with the carrier,

compared with those that do not interact.

Single Pore Polymeric Membranes

Ion channels and pores are crucial for functioning of a living organism (116,117).

Channels and pores are the principal nanodevices mediating the communication of a cell

with other cells where ion channels serve as extremely sensitive electromechanical

devices that regulate electric potential, ionic flow, and molecular transport across cellular

membranes (116).

Emulating the function and structure of these natural nanodevices


It has been shown that a protein nanopore (e.g.,the a-hemolysin channel), which

is embedded into a lipid bilayer membrane, can function as a biosensor for biomolecules,

for example, DNA (118).

The sensing procedure is based on directing the biomolecule to

the pore by means of an electric field.

When passing through the pore, a biomolecule

brings about its temporary blockage, which is observed as a change in the ion current


The ion current blockade depends on the structure and chemistry of the

biomolecule (e.g., the DNA sequence) (118), which is the basis of its detection.

biological pore is however quite fragile.


A more realistic approach to applying this idea

on an industrial scale would involve replacing the protein channel with a durable,

synthetic nanopore.

Recent research has shown that a single conical-shaped pore

generated in a polymeric foil presents similar transport properties to those of natural

biological channels (119-123).

In these studies, they have prepared model nanoporous

systems of known geometry and chemistry to study the relationship between the structure

and transport properties of nanopores, as well as, creating abiotic analogues of biological


Such an approach enables one to focus on the basic physical and chemical

phenomena underlying biochannels function.

A special emphasize was given to the family of voltage-gated channels.

current rectification and the dependence of ion current fluctuations on voltage across the

membrane are fingerprints of this type of channel (117).

were prepared by the track-etch technique.

The synthetic pores studied

This technique is based on irradiation a

polymer foil with swift heavy ions and subsequent chemical development of the latent


penetrates the foil and produces one latent track (37). Subsequently, one latent track after

chemical development results in the formation of one pore. Counting the number of ions,

which penetrate the foil gives a possibility to prepare membranes with a designed number

of pores.

Controlling the irradiation down to one ion enables preparing a macroscopic

sample containing just one pore.

The Department of Materials Research, GSI

Darmstadt, possesses a unique world wide facility suitable for single-ion irradiation (37).

A membrane with a single pore creates an optimal system for fundamental studies of ion

transport through nanopores, because averaging effects resulting from ion transport

through many pores can be avoided.

To prepare voltage-gated nanopores, an asymmetric pore geometry has been used,

because it was found that biological voltage-gated channels are asymmetric (124-126).

The conically shaped nanopores in polymer membranes were shown to rectify ion current

and exhibit ion current fluctuations of similar statistical properties as the ion current

through biological voltage-gated channels (119,121).

The application of a single conical gold tube membrane to protein biosensing will

be presented in chapter

The steps involved in preparing these membranes are

presented here, and are as follows:

Irradiation with heavy ions

Polyethylene terephthalate (PET) (Hostaphan RN12, Hoechst) and polyimide

(Kapton HN50, Du Pont) foils, having a thickness of 12 jm are irradiated with single

swift heavy ions at normal incidence.

Figure 1-12 shows the chemical formula of PET

anti ltr ntnrn

0 H


Figure 1-12. Chemical formula of the PET (A) and Kapton (B).

Gold, xenon and uranium ions of energy 11.4 MeV per nucleon are used at the

linear accelerator UNILAC (GSI, Darmstadt).

At this energy the penetration range of

ions in PET foil is larger than the thickness of foils and the energy loss of ions along the

track is well above the energy threshold, which assures a homogeneous etching (127).

Single ion irradiation is performed by defocusing the ion beam and placing a metal mask

with an aperture of 0.3 mm in front of the polymer foils.

The ions pass through the

aperture in a discrete way and as soon as one ion reaches the detector placed behind the

sample, the beam is switched off by a beam chopper within several microseconds.

Ion track etching

Chemical etching of single-ion irradiated foils is performed in a conductivity cell,

connected to a voltage source and picoamperometer (see Figure 1-13).

FI i!-



To obtain conical pores, etching is performed only from one side.

The other side

of the membrane is protected against etching by a stopping medium, which neutralizes

the etchant (121,122,128,129).

For etching of PET, 9 M NaOH has been used, therefore

an acidic solution plays the role of a stopping medium.

Ion tracks in Kapton were

developed in sodium hypochlorite with 13 % active chlorine content.

A stopping

medium of 1 M potassium iodide was used, which serves to reduce the OC1"

in etching, to Cf (122,129).

ions, active

The chemical stopping is supported by an electrical one.

The platinum two electrode system is configured in a way that the anode is placed on the

etching side, and the cathode is placed within the neutralizing side (128).

At the very

beginning of the etching process the two halves of the conductivity cell are not connected

with each other and the ion current measured is zero.

When the pore is etched through,

the ion current increases gradually, indicating increase of the pore diameter.

The etching

process is stopped by washing the pore with a stopping medium and water.

The big opening of the pore, D, has been determined by scanning electron


For conical pores in PET, D

- 600 nm and for pores in Kapton D

2 pm.

The difference in D values results from difference in non-specific etching of the two

polymers, the so-called bulk etch rate, which determines the big opening of the pores


The small opening of the conical pores is below scanning electron microscopy

resolution and its diameter d was estimated by measuring a current-voltage characteristic

of a single nanopore in a standard solution of 1 M KC1.

Assuming an ideal conical shape

of the pore its small opening can be calculated using the following equation (128,122):


/r 41 <"


where L is the length of the pore, K stands for the specific conductivity of the

electrolyte, U denotes the voltage applied across the membrane and I is the ion current


This etching process gives the possibility of producing pores with an effective

diameter d as small as 2 nm.

Gold electroless plating

The single conical tube membranes are obtained using a template synthesis


Single pore membranes are electroless plated with gold (40), as described

before in this chapter.

Dissertation Overview

The aims of the research presented in this dissertation are to investigate potential

applications of the nanotube membranes in chemical and biochemical sensing and


Chapter 1 provides background information on the template synthesis method and

two types of templates (porous alumina and polymeric membranes) that are used in this



presents the preparation of a biomimetic ligand-gated ion channel

membrane, based on a microbattery/nanoporous system.

This membrane turns on the

battery and attendant ion current in the presence of a targeted chemical stimulus (a

surfactant in this case).

This microbattery was prepared by depositing anode and cathode

materials on either side of a nanoporous alumina membrane.

The pores in the membrane

were made hydrophobic by reaction with an 18-carbon (C18) alkyl silane.

When placed

between two salt solutions, the pores in the Cls-modified membrane are not wetted by


membrane causing the biomimetic gate to open, which results in a high flux of ions

through the nanoporous membrane, and consequently, the microbattery turns "on"


"on", the microbattery discharges and the discharging current is collected in an external


Chapter 3 presents silica nanotube membranes are used to prepare highly selective

membranes for protein separation.

Two antibodies with different affinities for hevein

have been immobilized on highly ordered porous alumina membranes.

labeled with green fluorescence protein (GFP).

red fluorescence protein (RFP)


Hevein was

The transport of both GFP-Hevein and

- which was used as a control analyte- has been

The transport of both proteins has been recorded simultaneously at two

different emission/excitation wavelengths.

As a control experiment, we used membranes

with an immobilized antibody that does not have any affinity towards GFP-Hevein or

RFP. The alumina membranes were prepared by two step anodization method, having

pore diameters between 50 and 100 nm and thickness from 40 to 200 mm.

Both the

influence of pore diameter and the membrane thickness on the transport of GFP-Hevein

and RFP were studied.

These membranes selectively transport the protein (GFP-Hevein)

that binds to the antibody, relative to the other protein (RFP) that has no affinity for the


Chapter 4 explores another application of the porous alumina membranes, in this

case, serving as protein microarrays.

microarrays are presented. The first

pre-patterned aluminum surface. In

Two methods of preparing alumina-based

method implies growing alumina membranes on a

the other approach, commercial alumina membranes


thickness of the membranes on the fluorescence intensity, and also the capability of these

microarrays to selectively recognize different analytes were studied.

Chapter 5 deals with a new class of artificial ion channels based on a

membrane that contains a single conically shaped nanotube.


These nanotube-based

abiotic ion channels exhibit transport properties analogues to voltage-gated biological


The ion current through a single nanotube fluctuates in time in a voltage-

dependent manner as well as it is rectified. The membrane with a single conically shaped

gold nanotube was prepared by the template method. The nanotube has a large-diameter

opening of- 600 nm and a small-diameter opening of 2 5 nm. In the biosensing

application, the nanotube-containing membrane is placed between the two chambers of a

conductivity cell filled with an electrolyte.

Electrodes present in each half-cell solution

are used to apply a transmembrane potential and measure the resulting ion current

through the nanotube.

The internal surfaces of the nanotube are modified with a specific

biochemical molecular-recognition agent (the "capture" agent, e.g., an antibody) which

interacts specifically with a given biomolecule (the analyte) present in one of the

contacting solution phases.

The binding interaction between the nanotube-bound capture

agent and the solution-phase analyte is transduced as a change in the ion current that

flows through the nanotube.

This new biosensing technology was demonstrated using

both biotin as the capture agent and streptavidin as the analyte, and protein G as the

capture agent and IgG as the analyte.

The detection of a biological warfare agent (ricin)

is also presented.

The results and conclusions of this dissertation are summarized in Chanter 6.



Ion channels are the heart of many biological processes including nerve activity and

muscle contraction.

Channels operate by being either open or closed.

There are several

aspects of the channel environment which affect the channel opening (gating) such as

voltage, a molecular ligand, phosphorylation, or mechanical stimulus (116).


the principles of such natural sensor system is of great importance in sensor development


One example of a natural ligand-gated ion channel is the acetylcholine-gated

ion channel (132), which is closed ("off" state) in the absence of acetylcholine but opens

(and supports an ion current,

"on" state) when acetylcholine binds to the channel.


concept of ion-channel mimetic sensing, originally proposed by Umezawa's group (133),

has been of considerable interest in analytical chemistry (134-136).

In the biological

channel there are no electrodes, and the ion current is driven by an electrochemical

potential difference across the cell membrane (137).

That is, the cell membrane has its

own built-in transmembrane power supply that drives the ion current when the channel


Whether this power supply can be utilized in this way depends on whether the

channel is open or closed.

We describe here the preparation of a biomimetic ligand-gated ion channel

membrane, based on a microbattery/nanoporous system.

To explore this concept we


thin-film battery anode onto one face, and a thin-film battery cathode onto the opposite

face, of the membrane.

Hence, in analogy to the naturally occurring channel case, we

have an ion-channel mimetic membrane with a built in electrochemical potential

difference across the membrane.

We show here that in the absence of the ligand (again, a

hydrophobic ionic surfactant), the membrane is in its "off" state, and the electrochemical

potential difference cannot be utilized to drive a transmembrane ion current.

when the ligand is detected, the membrane switches to its "on"

In contrast,

state and the

transmembrane battery discharges producing a corresponding transmembrane ion current.



Octadecyltrimethoxysilane was obtained from Aldrich.

Dodecyltrimethylammonium chloride, hexadecyltrimethylammonium chloride,

dodecylbenzenesulfonic acid, and the nonionic surfactant Triton X-100 were obtained

from Acros Chemicals.

Octyltrimethylammonium bromide was obtained from from

Fluka and N-Dodecyl-N,N-dimethyl-3-ammonio-1-propanesulfonate from Sigma.


powder 99.9% was obtained from Strem Chemicals and zinc powder (99.33%) from

Fisher Chemicals.

All chemicals were used as received.

MiliQ 18-Mq water was used

for preparing all aqueous solutions.

Commercial porous alumina membranes (-200 nm-

diameter pores, 60 |xm thick) were obtained from Whatman Inc.

Silanization of the Alumina Membranes

A solution that was 5% (v/v) in octadecyltrimethoxysilane (Cis-silane) was

nrenared in ethanol: to this solution was added acetate hbuffer (50mM. nH=5.1 to make


from the solution and rinsed with ethanol. The membrane was sonicated for 10 minutes in

ethanol to remove the physisorbed silanes from the surface. The modified membrane was

cured at 1500C in air for 20 min. The performance of the coating was assessed by

measuring a contact angle of 1300 (+80).

Microbattery Fabrication

As shown schematically in Figure 2-1, the anode and cathode of the battery were

deposited as thin films coating the faces of the Cis-modified membrane.

These films

were deposited by thermal evaporation of either Zn (anode material) or Ag (precursor to

cathode material) from a tungsten boat.

depositor was used.

A Denton Vacuum DV-502 vapor-phase

The pressure inside the deposition chamber was -10-5 Torr.

Deposition time was 10 minutes for both the silver and zinc films.

The Ag film was

deposited first, and then a portion at the surface of this film was converted to AgC1,

which acted as the cathode for the microbattery.

This was accomplished by immersing

the silver-coated membrane in an aqueous solution that was 0.1M in FeC13 and 0.3M in


The Zn film was then deposited on the opposite face of the membrane.

important to point out that these films are porous and are thus permeable to ions and

molecule present in solution phases that contact the membrane videe infra).


0.1 MFeCI
0.3 MH-C


Cq-ndified alunina


OAAgC 8-lnmdified


Figure 2-1. Schematic representation of the microbattery fabrication.

Scanning Electron Microscopy (SEM)

SEM was used to study the surface morphology of the anode and cathode films.

Data were obtained using a JEOL 6400 microscope.

Elemental compositions for the

films were obtained using an Oxford energy dispersive spectrometer (EDS) attached to

the JEOL microscope.

Cell Assembly and Battery Discharge Measurements

After deposition of the electrode films, the membrane was sandwiched between

two pieces of Scotch tape that had 0.47 cm diameter holes punched through them.

holes defined the area of the membrane exposed to the contacting solution phases.



addition, the tape was used to attach a Pt foil lead to the surface of each battery electrode

film in order to make electrical contact with the electrodes.

The membrane was then

mounted between the two halves of a U-tube permeation cell (31,40,43).


shows a schematic representation of a U-tube permeation cell. The electrolyte solution

I -1 1 "fi 11 fl t *E T /r1A 1 1 fl 1 -

... 1.


This was accomplished by injecting the desired volume of a stock surfactant solution into

the 0.1 M NaCI in both half-cells.

The surfactant solutions were also 0.1 M in NaCI.

battery discharge current was monitored using a potentiostat (EG&G model 273)

interfaced to a PC running CorrView and CorrWare software packages (Scribner

Associates, Inc., Southern Pines, North Carolina).

Glass U-cell





Perm eate

Stirring bars



Schematic representation of a U-tube permeation cell.

Results and Discussions

Characterization of the Electrode Films

Surface and cross-sectional SEM images (Figure 2-3) show that both the Ag/AgCl

and the Zn thin films are porous.


This porosity results because the films are deposited as

The cross-sectional images indicate that these particulate films are -500 nm

thick, and that deposition does not propagate down into the pores.

(The particles seen in

the pores in the cross-sectional images were dislodged from the surface films during

EDS spectra for the Ag, Ag/AgC1 and Zn films are shown in Figure 2-4.

The Ag

film (Figure 2-4 A) shows prominent peaks for Ag and Au; the Au peak results because

the surface of the film was sputtered with Au prior to taking the SEM image.


weaker signals are observed for Cu (from the Cu foil tape used to attach the sample to the

SEM stub) and Al and O (from the underlying alumina membrane).

The upper surface of

the Ag film was chemically oxidized to AgC1, which served as the cathode for the

transmembrane microbattery.

Ag/AgCl film

Zn film



Top view


Top view

Cross-sectional (upper) and surface (lower) SEM images of the battery

electrode films that coat the faces of the alumina membrane.

After oxidation a prominent Cl peak is observed in the EDS spectrum (Figure 2-4

fn \ rT^ .- ',7 r .. i r*_ I1. A 1 1 A r i._ n


rue UUCU S *flyrt

LS ^*^|*t ,^;,^ &^

2 ... 4 -'.1 E 8 10 12
*~~~~~ ~ ~ i f i ihi kk AA Jw^f V

I~ *..i ..,
-6 *.t-=i -.

I IIIIIIII ..yr:I..lllllL 461111 5!1~(~1'1 a':.11'11.'. : 8 2
.. .. ] 1.~ :I B :2
r -

Figure 2-4. EDS spectra of the membrane surface after deposition ofAg (A) and after
conversion of the Ag surface to AgCl (B). EDS spectrum of the opposite
membrane surface that had been coated with Zn (C).

Battery Discharge Experiments

Control experiments were first conducted with membranes that were not modified

with the hydrophobic Cis-silane.

As described above, a Zn anode film was applied to one

face of the membrane and a Ag/AgCl cathode film was applied to the opposite face.

membrane was mounted in the U-tube cell, and the electrode films were connected to the

leads of the potentiostat.

However, the half-cells were initially devoid of electrolyte

solution, and as a result, there was no ironically conductive pathway through the

r_ Jr- C "- ....
O~~ ~ ~~~~~~~~ ~ .': 2 ,jf .;i"li~ ":i I:

NaC1) was added to both half-cells.

Because the untreated alumina membrane is so

hydrophilic, electrolyte immediately flooded the pores, allowing for ionic conduction

between the anode and cathode films.

This allowed the transmembrane microbattery to

discharges via the following discharging half-reactions:

Zn -- Zn2+

2AgCl+ 2e

= 0.763 V

= 0.222 V

-> 2Ag+2C1-



As indicated by Eo values, this battery delivers almost 1 V

As shown in Figure

A the current raises immediately to a peak value of -250 pIA and then decays away with


Before this experiment, the face of the membrane coated with the AgCI cathode

film was dark purple in color due to the AgC1.

After this experiment this face of the

membrane was white, indicating that all of the AgCl had been reduced during the battery


This explains why the current ultimately decays to zero (Figure

An analogous experiment was conducted with a C1i-modified membrane that had

the Zn anode and AgCl cathode films on its surfaces (Figure

electrolyte was added to the half-cells at t

In this case,

However, the current obtained is at the

noise level of the potentiostat, indicating that battery discharge is prevented.

As shown in

our prior work, this is because the hydrophobic pores are not water wetted (138), and as a

result there is, again, no ionic-conduction pathway through the membrane.

These results

show that when the hydrophobic microbattery membrane is exposed to NaCl solution, the

membrane is in its "off" state, and transmembrane battery discharge is prevented.

0 0003

0 0002

E 00001


-0 0001



! i i -.


Tune (Sec)

[.iI lIE lhl hl liUllUIMIIII

S000 3000
1000 2000 3000

Time (Sec)


Current-vs.-time response for the transmembrane microbattery applied to an
alumina membrane that was not rendered hydrophobic by silane

functionalization (A).

Analogous current-vs.-time response for the

hydrophobic C18-modified membrane (B).

Figure 2-6 shows current-vs.-time data for the C18-modified membrane before and

after injection of dodecylbenzene sulfonate (DBS) into the half-cell solutions.

injection of DBS, t

Prior to

< 200 s, the membrane was again in its "off" state, and battery

discharge was prevented.


s, DBS was injected to make the DBS concentration

in both half-cells 1 mM.

After injection there was an induction period followed by a

time region in which a low-level discharge current (-1.5 gA) was observed (inset Figure


This low-level discharge current flowed for 1300 s, after which a burst of current,

at a much higher level was observed.

Analogues results were obtained for C18

membranes upon exposure to the cationic surfactant dodecyltrimethylammonium (DTA).

These results are in many ways similar to results obtained in our prior

investigations of the effect of DBS on the ionic resistance of the C18-modified alumina

membrane (138).

First, when 0.1 M KCI was present in both half-cells, without added



membrane resistance to decrease, but still the resistance remained high (> 20 MO).

When the DBS concentration reached 10 jiM a precipitous 4-order of magnitude drop in

resistance was observed.

The key point is that two "on" states were observed a high-

resistance "on" state at low concentrations of DBS followed by a sharp transition to a

low-resistance "on" state at higher concentrations of DBS.

We showed that the sharp transition from the high-resistance "on" state to the low

resistance "on" state is associated with flooding of the pores with the electrolyte solution


That is, at some critical solution-phase concentration of DBS, the quantity of DBS

partitioned into the membrane is sufficiently high that the pores are no longer

hydrophobic, and water cannot be prevented from entering the pores.

In this flooded

state charge is carried through the membrane by migration of ions in the solution-filled


At lower DBS concentrations, the ionic current is presumably carried by surface

migration of the DBS and attendant counterion along the pore walls in pores that are

devoid of water (138).

This accounts for the high resistance of this "on" state.

In the transmembrane microbattery experiments described here, after injection of

DBS into the electrolyte solutions, DBS must diffuse through the porous electrode films

videe infra) and then into the hydrophobic pores.

The two "on" states observed

previously (138) are analogous to the two current-levels observed here.

The low current

"on" state (e.g., inset Figure 2-6) is associated with surface migration in pores devoid of

water, and the sharp transition to the higher current "on" state is associated with flooding

of the pores with water and electrolyte. As be:

some critical level of DBS into the membrane.

fore, flooding occurs when diffusion brings

Finally, the induction period prior to the


in the membrane such that a discharge current above the detection limit for the

potentiostat can be supported.








1000 2000 3000

Time (Se

Figure 2-6.

Current-vs.-time response for a hydrophobic membrane before and after

injection of DBS surfactant solution.

After injection, the DBS concentration in both half-cells was 1 mM. The timescale

of this induction process, however, seemed long; e.g., in Figure 2-6 it takes in excess of


s before the critical membrane level of DBS is achieved.

This suggests that there is

some barrier to transport of DBS into the membrane.

The most likely source of this

barrier is diffusion of DBS through the battery electrode films on the surfaces of the


To explore this issue we attempted to make thinner electrode films; however,

because of the particulate nature of these films, the lateral electronic resistance of these

thinner films was too high, and battery discharge was not observed.

used above) were then thermally evaporated onto these Au films. In essence, the Au

films act as current-collectors for the overlying battery electrode films. We found that

the induction times for these thinner battery electrode films were shorter than for thicker


For example, with the 1 mM DTA solution a membrane with the thinner battery

electrode films showed a response time of 400

s as opposed to 1000

s for the membrane

with the thicker films.

These results show that transport of surfactant through the battery

electrode films is a kinetic barrier in this system.

Effect of Surfactant Concentration on the Response Time

The response time of this device is defined as the induction time period between

injection of surfactant into the half-cell solutions and observation of the low current "on"


Over the concentration range 0.01 mM to 1 mM, the response time is inversely

proportional to the solution-phase DBS concentration (Table 2-1).

This result is

consistent with our model that the response time is associated with transport of DBS

through the electrode films and into the pores of the membrane.

Higher DBS

concentrations yield higher concentrations at the membrane solution interface and thus

higher net fluxes into the membrane.

As a result, the time required to achieve a

concentration of DBS in the membrane sufficient to support the low current "on" state

decreases with increasing solution-phase DBS concentration.

Table 2-1.

Effect of DBS concentration on the response time.

DBS Concentration (mM) [ Response time (s)
2 350
1 1.5
0.1 52
0.01 70


2-2 shows that analogous results are obtained for the cationic surfactant

DTA; however, at both the 0.01 and 0.001 mM levels the response times for DTA are

about a factor of three times higher than for DBS.

These results also fit our diffusional-

transport model because the diameter of the DTA head group is larger than the diameter

of the DBS head group -3.7 nm vs.

The larger diameter for the DTA

head group makes the surface diffusion coefficient smaller, and as a result the diffusional

transport time longer.


Effect of DTA concentration on the response time.

When the solution-phase DBS concentration is increased to

mM the response

time goes up again, indeed, to a value higher than is observed for the lowest DBS

concentration (Table 2-1).

While analogous results are obtained for DTA, the

concentration producing the longest response time (1 mM, Table 2-2) is lower than for


Our initial hypothesis was that these longer response times at the highest surfactant

concentrations were in some way associated with micelle formation; however, the critical

micelle concentrations (CMCs) for DBS and DTA are 1.1 mM (140) and 4.4 mM (141),


Hence, while the concentration that yields the longest response time for

DBS (2 mM) is above the CMC, the concentration that yields the longest response time

for DTA (1 mM) is below the CMC.

It is not yet fully understood why the response time

II i i i i -

DTA Concentration (mM) Response time (s)
1 1000
0.1 140
0.01 216
0.001 600

-2.0nm(13 9).


Investigations of Alkyl Chain Length on the Response Time

During the investigations of hydrophobic alkyl thiol-modified gold nanotube

membranes it was shown that the transmembrane flux increases with the hydrophobicity

of the permeate molecule (44).

This is because the flux is directly proportional to the

partition coefficient for the permeate molecule at the membrane/feed-solution interface


in reference 44).

The same principle should apply for flux of hydrophobic

surfactant molecules into the C18-modified alumina membranes studied here. As a result,

response time should decrease with increasing hydrophobicity of the surfactant. To

explore this issue we investigated response times for three alkyl trimethylammonium

surfactants (Table 2-3).

In agreement with the above analysis, response decreases with

increasing hydrophobicity alkyll chain length) of the surfactant.


Effect of alkyl chain length in alkyl trimethylammonium surfactants on the
response time. Concentration in all cases = 1 mM. These results were

obtained using the C18-modified membrane with thinner battery electrode
Number of carbons in alkyl chain Response time (s)
8 >3600
12 400
16 5


It has been shown in this work that a microbattery /nanoporous membrane acts as a

biomimetic "smart membrane" in the sense of emulating the function of ligand-gated ion

channels; i.e., they can be switched from an "off" state to an "on" state in response to the

presence of a targeted chemical stimulus. Modifying the aluminum oxide with long chain

alkylsilanes makes the membrane pores hydrophobic and the microbattery

is "off'

. In the


response time of the system: the hydrophobicity of the analyte (shorter response time for

more hydrophobic analytes), the nature of the polar head group of the analyte (anionic

surfactants are sensed faster than the cationic ones) and the concentration of the analyte

(shorter response time for lower concentration). This concept could ultimately lead to a

remote sensing technology where the battery discharge current is use to drive a device

(e.g., a buzzer) that signals to the outside world that the ligand has been detected.



Many technically challenging and commercially attractive separation problems can

not be solved with existing membranes because the typically achieved separations of

complex mixtures are only fractionation into substance groups (142).

Membranes with

high selectivities for example for chiral drugs, toxins or complex biomolecules are


The aim of current membrane development is preparing "tailored" membranes

with high selectivity and/or high flux of the analytes of interest across the membranes;

this can be achieved by developing membranes modified with molecular recognition

elements, with high pore density and a narrow pore size distribution.

Active research is

devoted to highly specific membrane separations based on molecular recognition inside

the nanoporous membranes (47,82,152).

We report here a protein separation method based on facilitated transport

selectivity, that utilizes immobilized molecular recognition elements in a porous alumina


We found that by modifying the walls of the alumina membrane pores with

antibodies, proteins that have an affinity towards these antibodies are transported at a

higher rate than proteins that do not interact with the antibodies.

We used two Fab

fragments (1C2 and 1A4) of antibodies that were grown against the protein hevein; 1A4

- ~ -~ -..


Hevea Brasiliensis involved in the inhibition of several chitin-containing fungi (144,145).

Hevein was produced as a green fluorescence protein (GFP) fusion protein (GFP-Hevein)

in insect cells (



In this way GFP-Hevein can be detected by fluorescence

Red fluorescence protein (RFP) was used as a control protein because it

does not have any affinity towards any of the antibodies. Because their excitation and

emission spectra do not overlap (XexGFp=395 nm, XemGFP=510 nm, hexRFp=563 nm,

,emRFp=582 nm), GFP and RFP are very well suited for dual-label experiments (146).



The antibodies ENA 11 His, 1A4 and 1C2 Fab fragments were provided by VTT

Biotechnology, Finland. Red fluorescence protein (rDsRed2 protein) was purchased from

BD Biosciences Clontech.

High purity aluminum foils 100 mm x 500 mm x 0.2 mm,

purity 99.9998%) were obtained from Alfa Aesar, tetraethyl orthosilicate (TEOS) from

Sigma Aldrich and triethoxysilylaldehyde from Gelest.

Fabrication of the Nanoporous Alumina Membranes

We used a two-step anodization method (52,53) to fabricate the porous alumina


Briefly, the aluminum foils were first electropolished at 15 V in a solution

with the following composition: 95 wt. % H3PO4, 5 wt. % H2S04 and 20 g/L CrO3.

solution was heated at 70 OC.

The polished aluminum foil was anodized at 40, 50 and 70

V to obtain membranes having 50, 70 and 100 nm pore diameter, respectively.

We used

5% oxalic acid as the electrolyte solution and the anodization experiments have been

done at 0 C.

The first film of the membrane was dissolved away in an aqueous solution

highly ordered nanoporous alumina membranes.

was dissolved into a saturated HgCl2 solution. 1

The aluminum that was not oxidized

'hen the barrier layer of the alumina

membranes was removed in 5% H3P04 solution.

Scanning electron microscopy was used to measure the pore diameters and the

thicknesses of the alumina membranes prepared.

A JEOL FE-SEM 6438 was used.

Alumina nanoporous membranes (Figure 3-1) were used as templates for immobilization

of the antibodies.

Figure 3-1. Scanning electron micrograph of a porous alumina membrane with pores of
50 nm in diameter and pore density of- 1010 pores/cm2.

Antibody Immobilization

Figure 3-2 shows a schematic representation of the modification steps involved in

antibody immobilization.

Silica nanotubes were deposited inside the pores of the

alumina membranes using a sol-gel method (81.84).

Briefly. a sol-gel silica precursor

was allowed to hydrolyze for 30 minutes.

Alumina membranes were than immersed into

the sol-gel for 1 minute under sonication, after which they were air dried for 10 minutes

at room temperature and cured in the oven for 12 hours at 1500C.


has been used to attach aldehyde terminal groups onto the pores of the membranes.

free amino sites of the antibody Fab fragments react with the aldehyde groups via a

Schiff base chemistry (147,176,177).

The aldehyde-modified silica nanotube membranes

were incubated for 12 hours at 4 OC in a solution containing 0.2 mg/ml antibody Fab

fragments solutions.

All solutions of the antibody Fab fragments, GFP-Hevein and RFP

were made in phosphate buffer saline (PBS) solution having pH=7.4.

After washing

them with PBS, the antibody-modified membranes were incubated for 3 hours in a

blocking solution containing 1% bovine serum albumin (BSA) and 1% Tween-20 in PBS.

Blocking solution was used to block both the unreacted aldehyde groups and nonspecific


The membranes were washed copiously with PBS afterwards and stored in

PBS at 40C until used in transport experiments.

- OH




Si H

Surface of the
silica nanotube


Transport Experiments

The antibody-modified silica nanotube membranes were sandwiched between two

pieces of Scotch tape that had 0.314 cm2 area holes punched through them.

defined the area of the membrane in contact with the solution phases. The

These holes


were then mounted between two halves of a U-tube permeation cell (40,43,44,148). The

feed solutions had equal concentrations in GFP-Hevein and RFP. A volume of 3 ml PBS

was used on both halves of the permeation cell. The rate of transport (flux) was

determined by periodically measuring the fluorescence intensity of the permeate solution

using a Varian Cary Eclipse spectrofluorometer.

GFP-Hevein and RFP were detected


Results and Discussions

Effect of Antibody Affinity on Selectivity Coefficient

Transport plots in Figure 3-3 show the number of picomoles of GFP-Hevein and

RFP transported through the nanotube membranes versus the permeation time.

A single

membrane was cut into three pieces and they were modified with ENA 11 His Fab

fragments (Figure 3-3A), 1C2 Fab fragments (Figure 3-3B) and 1A4 Fab fragments

(Figure 3-3C) respectively. The membrane used for these experiments had pores 70 nm

in diameter and a thickness of 90 upm. A concentration of 10 nM in both GFP-Hevein

and RFP in PBS was used in feed side, with PBS only in the permeate side.

Due to the

fact that ENA 11 His does not have affinity for either GFP-Hevein or RFP, membranes

modified with this antibody were used for control experiments.

As can be seen from

Figure 3-3A RFP is transported by passive diffusion at a higher rate than GFP-Hevein,

28,000 Da (149,150) and Hevein has a molecular weight of 4,000 Da (151).

When the

antibody having an affinity towards Hevein (1A4 or 1C2 Fab fragments) was

immobilized, the membranes transported GFP-Hevein at a much higher rate (Figure 3-3A

and 3-3B) than they transported RFP.


0 500



Time, min


0 500


Time, min




Time, mn.



Because 1A4 and 1C2 Fab fragments bind hevein, these data suggest that

immobilization of these antibodies facilitates the transport of GFP-Hevein versus RFP.

The transport of GFP-Hevein through the membranes is not linear in time because the

concentration gradient decays with time, so the driving force for diffusion decays as well.

Due to that, the fluxes of GFP-Hevein and RFP through the membranes were calculated

with respect to the linear part of the transport plots (the first two points of the plots).

defined a selectivity coefficient for a membrane, s, as the ratio between the flux of GFP-

Hevein and the flux of RFP.

Figure 3-4 shows the influence of the type of the antibody

immobilized on the selectivity coefficient.


.-~*.* 1*,


1A4 ...


Figure 3-4. Effect of the antibody immobilized on the selectivity coefficient.

As can been observed from Figure 3-4, immobilization of an antibody with higher

affinity towards the analyte yields a higher selectivity coefficient.

Immobilization of an

antibody of high binding affinity towards an analyte, results in increase of the analyte



Effect of the Feed Solution Concentration on the Flux and Selectivity Coefficient

An important parameter in characterizing facilitated transport of the molecules

through membranes is the feed solution concentration. A plot of flux versus feed

concentration should have a "Langmuirian" shape (47,82,152).

We investigated the effect

of the feed solution concentration on fluxes of GFP-Hevein and RFP (Figure 3-5). In

these experiments the membranes were modified with 1A4 Fab fragments and they had

pores of 70 nm in diameter and 90 mrn thickness.

55 60 65 70 75 80 85 90 95 100

Feed Concentration, nM


Plot of fluxes of GFP-Hevein and RFP versus feed solution concentration.

The plot showed in Figure 3-5 has a Langmuirian shape for GFP-Hevein.

It can be

observed that at high feed concentration the membrane transports RFP at higher rates

than GFP-Hevein.

At high feed concentration (100 nM), passive diffusion is

predominant. Facilitated transport theory also predicts that the highest selectivity

'^'~~^"""""~'~~l~l""""""~~~~~ 1"11~~-'~""'~~~~


plots of GFP-Hevein and RFP through a 1A4 antibody-modified membrane when using

nM (A), 20 nM (B),

0 nM (C) and 100 nM (D) feed solution concentration.




TnU% nsi

Tnt nin



TrIP 11*1

Tm uu nit

Figure 3-6.

Transport plots of GFP-Hevein and RFP through a 1A4 antibody-modified

membrane when using

nM (A), 20 nM (B),

50 nM (C) and 100 nM (D) feed

solution concentration.

These transport plots were used to calculate the selectivity coefficient (Figure 3-7).

Indeed the selectivity coefficient decreases as the feed solution increases.

nM feed

solution concentration a selectivity coefficient of 30.6 is obtained, and for 100 nM feed

solution concentration we obtained a value of 0.7 of the selectivity coefficient, which

corresponds to massive diffusion of the proteins.





Feed Concentration, nM


Variation of the selectivity coefficient with feed solution concentration.

Effect of the Pore Diameter on the Flux and Selectivity Coefficient

The diameter of the pores of the alumina membranes is another important

parameter that influences the flux of the proteins and selectivity coefficient of a


Figure 3-8 shows the transport plots obtained when membranes with pores of

50 nm (Figure 3-8A), 70 nm (Figure 3-8B) and 100 nm (Figure 3-8C) in diameter were


The membranes with pores of 50 nm and 70 nm in diameter had thickness of 80

jtm and 90 am respectively, and were modified with 1C2 Fab fragments.

The membrane

with pores of 100 nm in diameter had a thickness of 50 pm and was modified with 1A4

Fab fragments.

RFP has been us

(Figure 3-9). Th

In all cases a feed solution concentration of 10 nM in GFP-Hevein and

;ed. The selectivity coefficient decreases as the pore diameter increases

re increase in selectivity coefficient is obtained at the cost of lowering the


means that when using membranes with this pore diameter, facilitated transport is not

predominant; only passive diffusion is observed at high pore diameters.


I -CGFP-He'\ei

Time, min



Time, min.




Time, min



50 nm 70 nm 100

Figure 3-9. Selectivity coefficient variation with the pore diameter.


Highly selective nanotube membranes for protein separations have been prepared.

The separation is based on molecular recognition inside the nanopores of alumina


We have used antibody Fab fragments which as molecular recognition

elements selectively bind and transport proteins.

There are important parameters that

should be taken into account in obtaining a desired protein separation with certain fluxes

and selectivity coefficients.

These parameters are: binding affinity between the antibody

and the antigen, feed solution concentration, pore diameter of the membrane and

membrane thickness.

A higher binding affinity is desired because it leads to a higher

selectivity; although a very high binding affinity is not appropriate because the analyte

should be released from the antibody and transported through the membrane.


selectivity coefficients are obtained at lower feed solution concentration and when using


however, lower fluxes are obtained. High fluxes are important in separation processes;

higher fluxes can possibly be obtained by decreasing the membrane thickness, by

applying pressure (153) or by applying an electric field across the membrane (154).



Microarray technology is an emerging technology which is having a considerable

impact in proteomic research.

Protein microarrays allow the identification and

quantification of a large number of target proteins using a small amount of sample within

one single experiment (155 ).


This technology requires rapid, high throughput protein

Recent studies showed that protein microarrays can be used to screen for protein-

protein interaction (156,157), antibody specificity profiling (158,159 ), immune profiling

(160), and protein-small molecule interactions (161).

Significant challenges exist for protein microarrays which do not exist for gene

arrays (155,162,163,164).

The initial challenge is developing a system capable of

detecting a broad range of concentrations; proteins can exist in a very broad dynamic

range (up to 1010) in any cell.

The second challenge is detecting very low abundance

proteins; for DNA, PCR methods exists for amplification, while for proteins no

amplification method is available yet.

DNA is a very uniform and stable molecule (it

does not lose its activity when stored dry), with a well defined activity prediction based

on primary nucleotide sequence.

These factors are different for proteins.

Proteins exhibit

very diverse individual tertiary molecular structures; further, their 3D structure is

important for their activity, they should always be kept wet to avoid denaturation.

hydrogen bonds, hydrophobic or Van der Waals interactions.

In addition proteins may

have multiple binding sites and can possibly interact with different molecules in the same


Because of these challenges, substrate requirements are more demanding for

protein microarray technology.

Various types of substrate have been explored (165,166)

and the search for new supports with superior performances is still a challenge. There

have been reports on the spotting of protein microarrays using a variety of surfaces and

immobilization chemistries, including but not limited to agarose (167), polyvinylidene

difluoride (168), and polyacrylamide gel pads (169).

Proteins arrays on glass surfaces

coated with aldehyde (161), poly-L-lysine, and gold surfaces derivatized with SAMs

(170) have been reported also.

Application ofnanoporous silicon as support for protein

microarrays has been recently reported (171).

Three dimensional porous surfaces offer

several advantages over the flat surfaces, including higher sensitivity due to the higher

sample loading capacity, broader dynamic range of concentrations and a 3-D

environment that preserves protein activity and accessibility.

Here we report two methods of fabrication of 3-D porous alumina-based

microarrays and show their applications in antibody specificity screening.

Besides the

advantages offered by a 3-D support for microarrays, alumina membranes present a very

well defined morphology, which is important for uniform immobilization of the proteins

and providing reproducible detection of the ligand-binding events.




Chemical Technologies, and triethoxysilylbutyl aldehyde from Gelest. Triton X-100,

tetraethylorthosilicate (TEOS), rhodamine B isothiocyanate, human and mouse IgG, anti-

human IgG labeled with Alexa 488 and anti-mouse IgG labeled with Alexa 594 were

purchased from Sigma Aldrich.

Surface coating polymer FSC-M was obtained from

Shipley and polymer remover (remover 1165, Shipley) was purchased from MicroChem.

All the materials were used as received. The TEM copper grids (400 mesh, PELCO) that

were used as masks were purchased from Ted Pella. The porous alumina membranes that

were used for the second method were bought from Whatman, and they had a nominal

pore diameter of 200 nm.

(Cranston, RI).

Silver plating solution (Ag 1025) was purchased from Technic

Silver wire (2 mm thick) were purchased from Alfa Aesar (Ward Hill,


Fabrication of porous alumina microarrays

Method 1

High purity aluminum foil was first glued to a glass by using an epoxy glue.


the Al foil/glass plate was electropolished into a solution composed of 95 wt. % H3P04,

wt. % H2S04 and 20 g/L Cr03, heated at 70 OC.

Al foil/glass was washed with distilled

water and dried under vacuum at room temperature.

Figure 4-1 shows a schematic representation of the first method of fabrication of

the porous alumina microarrays.

There are four steps involved in this procedure and they

are as follow:

Step 1.

The electropolished Al foil/glass was first spin-coated with a surface

cnatinp nnlvmer (FSC-MV

From the SEM measurements we found that the thickness of

The polymer/Al foil was then inserted into a reactive ion etching apparatus

(Samco Plasma Ion Etching System, model RIE-1C). The pc

for 5 seconds in order to make the surface hydrophilic. This

oxygen plasma with radio frequency (13.56 MHz) of 140 W.

20 Pa oxygen and oxygen flow rate was 30 seem. Copper gr

)lymer surface was etched

was accomplished using

The plasma pressure was

ids (400 mesh) were placed

on top of the polymer with a dilute Triton X solution, which is a wetting agent, and

ensures the copper grid was stick flat to the surface after drying.

Step 3.

The plate with copper grids was etched for 4.5 minutes and then the copper

grids were blown away.

In this way, we obtained Al surfaces patterned with the coated


Cu grid
Al foil

02 plasma etching

Cu grid removing

anodization of Al

Step 2.

Step 4.

The patterned Al/glass system was electrochemically oxidized to form

porous alumina films, only in the areas where the Al was exposed.

The anodization was

carried out in 5% oxalic acid as electrolyte, at 00C, under a constant voltage of 50 V.

In a similar way, patterning of anodic alumina into aluminum was reported (172).

In this case, the aluminum was patterned with silica either by a sol-gel process or by

dielectric evaporation.

In the first case they reported the presence of cracks at the

interface between aluminum and alumina, and in the second case they observed the

growth of tilted pores underneath the silica layer.

Method 2

Figure 4-2 shows a schematic representation of the steps involved in fabrication of

the alumina microarrays by method

A thin Au-Pd layer (approximately 90 nm in

thickness) was sputtered on one side of the alumina membrane.

performed using a Denton Vacuum Desk II Cold Sputter. The

seed layer for electroplating.

Au-Pd sputtering was

Au-Pd layer was used as a

Commercial alumina membranes having 60 atm in

thickness and 200 nm pore diameter were used.

Au-Pd sputtered alumina membrane. Another

A copper TEM grid was placed on top of

mask (aluminum foil) was placed on top

of the copper grid just to have enough Au-Pd material for making the electrical contact.

This assembly was then inserted into the center of the vacuum chamber of a reactive ion

etching apparatus (Samco Plasma Ion Etching System, model RIE-1C) and Ar plasma

was used to etch the Au-Pd seed layer.

The Ar plasma parameters were as follows: 10

mmins, 13.56 MHz, 140 W, 10 Pa Ar, Ar flow rate = 12 sccm.

After etching, an Au-Pd

replica of copper grid was transferred to the membrane.

1Agon etchng

' F 1!

spin coating

Cu grid

Au-Pd layer

Alumina membrane

~:~" x~ "4"" E":; <0'""
0*:": *" ::
A>r -
0x" 0"
~a,,.,~: ~0 -42E:pr



Figure 4-2. Schematic representation of the microarray fabrication by method

Electroplating was accomplished using a EG&G PAR Model 273

galvanostat/potentiostat which was controlled using a CorrWare software package

(Scribner Associates, Inc., Southern Pines, North Carolina).

Electrochemical cells were

prepared from a Teflon cell (17 mm inner diameter) and stainless still plate, which were

held together using screws and o-rings (see Figure 4-3).

To electrodeposit

Ag into the

membrane, the membrane was placed on Teflon tape, Au-Pd sputtered layer side up.

Electrical contact was made to the membrane using copper adhesive tape.

A silver wire

was used as the counter electrode.

Ag was then deposited at

-2 mA cm-

for 8 minutes,

.', "* ^ *


layer to prevent the leakage of plating solution through the membrane and suppress the

lateral growth of the electrodeposits.

electrochemical cell again. In this c;

the open pores faced up. Additional

for 90 minutes.

The spin-coated membrane was place into the

ise, the Ag electroplated layer was side down so that

Ag was then plated into the membrane at -0.50 mA

After electroplating, the membrane was immersed into spin coating

polymer remover for 10 minutes, rinsed with ethanol, and dried at room temperature.

Figure 4-3.

Electrochemical cell setup for silver electrodeposition: A, Ag wire counter

and reference electrode; B, Ag plating solution; C, Cu foil; D, Au-Pd modified
alumina membrane as working electrode; E, stainless steel plate; F, teflon
tape; G, O-ring seal.

Membrane modification for sensitivity studies

For sensitivity experiments, five membranes with 0.5, 1.2, 50, 60 and 90 pm

thickness have been prepared.

They were prepared by electrochemically oxidation of

aluminum, using the two step anodization method (52).

Briefly, high purity (99.9998%)

aluminum foils were electropolished at 15 V in a solution with the following

oxalic acid as the electrolyte.

The anodization was conducted at 00C, for 15 hours.

first film of the membrane was dissolved away in an aqueous solution that was 0.2 M in

CrO3 and 0.4 M in H3PO4, at 60-70 OC.

The second anodization step was carried out in

exactly the same conditions (referring to the voltage applied and the electrolyte solution

used) as the first step.

The time of the anodization in the second step varied for the five

membranes from 20 minutes to 16 hours, resulting in formation of highly ordered

nanoporous alumina membranes having pore diameter of 75 nm and thicknesses between

0.5 and 90 jrm.

The aluminum that was not oxidized was dissolved into a saturated

HgC12 solution, except for the case of 0.5 and 1.2 jpm thick membranes due to their more

susceptibility to fragment into the small pieces.

A sol-gel template synthesis method was used to deposit silica nanotubes (with a

wall thickness

- 3 nm) within the pores of the alumina films (173,174 ).

First, a sol-gel

silica precursor was prepared by mixing absolute ethanol, TEOS and 1 M HCI (50:50:1).

This solution was allowed to hydrolyze for 30 minutes.

Alumina template membranes

were than immersed into the sol-gel for 1 minute under sonication, after which they were

air dried for 10 minutes at room temperature and cured in the oven for 12 hours at 1500C.

The inside walls of the silica nanotubes were reacted then with APTES, a silane with an

amino terminal group.

The silica nanotube membranes were immersed into an ethanol-

based solution which contained 6% APTES and 6% pH 5.1 acetate buffer solution.

membranes were kept in solution for 10 minutes under vacuum followed by 20 minutes

in ambient air at room temperature.

They were dried under nitrogen and cured at 120-

~. A -~ -~ -- -

The amino modified silica nanotube membranes were immersed in a solution of 1 %

(wt) rhodamine B isothiocyanate in DMF for 16 hours in the vacuum, under nitrogen.

The rhodamine B modified membranes were washed in DMF, chloroform and ethanol.

The washing was performed for 10 minutes under sonication in each solvent.




, = 570 nm X = 595 nm







Silica Nanotube




Figure 4-4. Modification steps for sensitivity studies.

As a control experiment we modified a piece of glass with rhodamine B in the same

conditions as the alumina membranes.

The glass was initially cleaned in a piranha

solution (3:1 H2SO4:30% H202) at 900C and washed copiously with deionized water.

Fluorescence spectra of the alumina membranes and glass modified with

rhodamine B were taken using a Zeiss fluorescence microscope. The Rhodamine B dye

was excited using 570 nm light and the emission was monitored using a 590 nm band

pass filter.

The dye was excited while simultaneously monitoring the emission with a




Microarray modification for selectivity studies

The alumina microarrays made by method 2 were used to investigate their

selectivity, in terms of screening for antibody specificity.

Again, a silica thin film was

deposited on the pore walls of alumina membrane-based microarrays.

These were

immersed into a ethanolic solution that was 5% in an aqueous acetate buffer with a pH of

5.1, and 5% in triethoxysilylbutyl aldehyde.

The aldehyde groups react readily with the

primary amines on the proteins to form a Schiffs base linkage (176,177).

This approach

was used to covalently attach the capture proteins, which in this case were human and

mouse IgG's.

The proteins were spotted on the alumina microarrays using a 10 X microscope

connected to a monitor, and a manual microinjection system (Brinkman,

A volume of 10 jiL of protein solution (0.2 mg/ml in PBS pH=

Westbury, NY).

.4) was back loaded into

a femtotip (Fisher Scientific, Pittsburgh, PA) and a compensation pressure of 50 psi


The tip was positioned using a micromanipulator until the tip touches the

alumina surface.

Due to the porous nature of the alumina, the dye is pulled into the

islands through capillary action without the need for addition pressures.

Once the surface

was saturated with protein solution the tip was reposition to another spot and filled in a

similar manner.

After 12 hours incubation at 40C, the arrays were immersed into a

blocking PBS buffer solution that contained 1% BSA and 0.1 % Tween-20 for 3 hours.

This step is necessary not only for blocking the unreacted aldehyde groups, but also for

reducing the non-specific adsorption of the proteins (161).

After washing thoroughly

tta+1, Df ld V~nr rln ni-nrn'd : ,r ,.nraltn ;nn *k~O n~nn~ n nalun an nnn~nrnn a~~n +b +nca nr

Alexa 594 dye, both having a concentration of img/ml in PBS).

After 12 hours of

incubation, the arrays were washed three times with PBS and then twice with deionized


Fluorescence microscope imaging was performed in order to evaluate the

selectivity of the alumina-based microarrays.

The Alexa 488 dye was excited with 495

nm light and the emission monitored using 515 nm band pass filter.

The Alexa 594 dye

was excited using 590 nm light and the emission monitored using a 590 nm band pass

filter. After individual image acquisition, the fluorescence images for each dye were

overlaid. Also an optical image of the surface was acquired using reflected light from the


Results and Discussions

Microarrays fabricated by method 1

Figure 4-5 shows scanning electron micrographs (SEM) of the porous alumina

microarrays fabricated by method 1 at a low (A) and higher (B) magnification.

As can be

seen in the SEM image, the porous alumina films are very distinct areas on the

polymer/Al surface.

The pores of the alumina film are not highly ordered in this case,

due to the fact that the anodization process took place only in one step and for a very

short period of time (20 minutes).

Uniformly cylindrically pores can be obtained by

using a two-step anodization method (52).

Figure 4-5. Scanning electron micrographs of the porous alumina microarrays
fabricated by method 1 at a low (A) and higher (B) magnification.

Microarrays made by method 2

Figure 4-6 A shows the SEM image of the silver patterned porous membrane.

silver metal was not electrodeposited throughout the whole length of the membrane; it

was deposited only along on a distance that represents 5% of the membrane thickness

(see Figure 4-6 B).

The possibility of using commercially available alumina membranes

presents an advantage of this method.

For quantitative studies, however, because of poor

homogeneity of the pores diameter in these alumina membranes, one will have to use

membranes prepared in house with highly ordered pores.

These membranes will have to

be patterned by method

Figure 4-6. Scanning electron micrographs of the porous alumina microarrays fabricated

by method

A, Ag patterned on porous alumina and B, Ag rods after

dissolving the membrane

Effect of the silica on the sensitivity measurements

Although silanes can be attached directly on the alumina surfaces, we found that

the fluorescence signal is enhanced if a thin film of silica is deposited primarily on the

pore walls of the membrane (Figure 4-7).

The silica film introduces a higher density of

hydroxyl groups on the surfaces, which provides a higher reactive surface area for further


The fluorescence spectra in Figure 4-7 show that for samples modified

with silica, the fluorescence signal is approximately 7 times higher than for samples

modified initially only with the silanes.





600 700

Wavelength (nm)

Figure 4-7

. Fluorescence spectra for a rhodamine B-APTES-alumina sample with (green)

and without (red) silica.


The graph in Figure 4-8 shows the relative fluorescence coefficient for rhodamine

B-modified membranes of 0.5, 1.2, 50, 60 and 90 am thickness.

defined the relative

fluorescence coefficient, a, as the ratio between the fluorescence intensity of the dye

modified membrane and the glass slide modified in the same way.

As we expected, the

3D structure of the alumina membranes leads to a higher sample loading capacity,

yielding an enhanced signal. Depending on the thickness of the membrane, the signal can

be enhanced as much as 416 times for a 90 um thick membrane.

glass 0.5 1.2 50 60 90

Membrane thickness, pm

Figure 4-8. Relative fluorescence coefficient for rhodamine B-modified membranes of
different thicknes.


As an application for the porous alumina-based microarrays we have screened the

arrays for antibody specificity.

After capture proteins were immobilized (human and

mouse IgG's), the arrays were probed with a mixture of the target proteins (anti-human

IgG-Alexa 488 and anti-mouse IgG-Alexa 594).

The left side image of Figure 4-9

represents an optical image of a 350 jtm x 350 jim section of the microarrays showing the

spots where we immobilized the capture proteins, and the right side image shows the

fluorescence image acquired after the mobilization of the target proteins.

It can be seen

from these images only the spots containing the capture proteins were highly fluorescent,

indicating that the proteins were immobilized and able to retain their functional properties

on the porous surfaces.

The spots where no capture proteins were immobilized are

lightly visible, indicating some nonspecific binding on the alumina surface.

Human IgG





Human IgG

Figure 4-9: Optical (left) and fluorescence (right) image of a 350 upm

x 350 am area of

the microarrays after immobilization of the target proteins.

Investigations of the fluorescence intensity profile (Figures 4-10 and 4-11) denoted

cross-reactivity between the human and anti-mouse IgG, and mouse and anti-human IgG


The uneven peaks in the intensity profile graphs are the results of both an

inhomogeneous delivery of capture proteins using the manual injection system and the

disordered structure of porous alumina.

These drawbacks can easily be overcome by

using a automatic injection system and a very highly ordered pore alumina support. That

is, high uniformity of the porous support is one general demand for quantitative protein

immobilization; both geometry (pore size) and morphology (pore shape, level of

branching) affect the physical properties of the protein microarrays and thus their

performances and characteristics (171).

One important feature of our microarrays is that the arrays are very well defined on

the platform.

They are separated from each other by either Al (method 1) or Ag (method

This eliminates the tendency of the samples to spread out, which is a main issue in

-150 -100 -50 0 50 100 150

X Axis (m)

Figure 4-10. Excitation with 495 nm light: A, 2D fluorescence image; B. 3D fluorescence
image; C, fluorescence intensity profile.

S -150 -100 -50 0 50 100 150 2

30 -150 -100 -50 0 50 100 150 2'

X Axis (pm)

Figure 4-11. Excitation with 590 nm light: A, 2D fluorescence image; B. 3D fluorescence
image; C, fluorescence intensity profile.


In summary, we report here two methods of fabrication of 3-D porous alumina-

based microarrays.

showed the advantages of using these arrays in terms of

sensitivity and the importance of using silica nanotubes for signal enhancement.

application of these microarrays in antibody specificity screening has been shown also.

The arrays obtained by method

being opened at both sides, can be incorporated into

microfluidic devices. The electrodeposited Ag rods confer to the membrane a greater

mechanical stability.

J J4 4 4 4* 1- 4 4 44 > 4 4.1-n

samples can be spotted on the same surface.

Furthermore, the sample loading capacity

can be controlled by varying the thickness and the pore diameter of the alumina


The uniformity of the spot intensity profile can be improved by using

alumina membranes with very highly ordered pore diameter distribution.

The ability to make protein arrays on a surface with very well defined features and

morphology should increase the capabilities of researchers to study protein interactions

on a whole proteome scale using the array technology.




There has been a big interest in constructing single molecule sensors based on


The principle of the sensor operation is based on the nanometer opening of

the pore which is comparable to the size of molecules to be detected (178).

When a

molecule enters the pore, the pore is temporarily blocked, which can be observed as a

significant temporary reduction in the ion current passing through the pore.

operates therefore as a Coulter counter on a single molecule level (178). T]

The device

his type of

sensor has been constructed on the basis of a protein a-hemolysin and its functioning was

demonstrated for DNA analysis (118,179, 180).

The nanopore sensor enabled

determination of the length distribution as well as chemical composition of DNA strands

in a solution, which built hopes for single-nanopore super fast DNA sequencing


A further major advance was made in the group of Hagan Bayley in

engineering a biosensor that is capable of identifying individual DNA strands with single-

base resolution.

Each biosensor element consists of an individual DNA oligonucleotide

covalently attached within the lumen of the a-hemolysin pore to form a "DNA-


. The other single strand the analyte is in the electrolyte solution.


system could distinguish between complementary and non-complementary DNA strands,

making it specific for a given DNA sequence.

This biological pore-bilayer system is


industrial scale would involve replacing the protein channel with a durable, robust,

synthetic nanopore.

Detecting single DNA molecules and characterizing their

distribution was demonstrated with several types of solid state nanopores but none of

them was equipped with recognition sites specific for a given biomolecule (183-187).

Here we present a 3-dimensional nano-immunoassay based on a single pore

system, capable of probing protein-protein interactions and detecting warfare agents.

principle of operation of this device is very simple and based on an intuitive and checked

experimentally fact that transport properties of a nanopore depend very strongly on the

pore walls surface characteristic (119, 188,189).

If an analyte to be detected binds to the

recognition sites placed on the pore walls and the pore has an opening of several

nanometers, the transport characteristic of a nanopore, expressed for example in a form of

current-voltage (I-V) characteristic, will be significantly changed.

Basing the detection

signal on I-V curves rather than time series will significantly simplify the recording as

well as data analysis process.

I-V curve represents average transport properties and as

such is much less demanding concerning the noise-free environment for recording than

time series, which is the main detecting signal for Coulter counter based devices (178).

As a base for the 3-dimensional nano-immunoassay we chose polymeric

membranes covered electrolessly with gold.

Gold surface enables easy modification of

chemistry of the pore walls by application of a thiol chemistry (44,190).

The pores in

polymer membranes were prepared by the track etching technique, which is based on

irradiation of polymer films with heavy ions and subsequent development of the latent

tracks by chemical etching (37).

The technique gives amazing freedom in preparation of


our 3-dimensional nano-immunoassay we chose asymmetric, conical shape of the pore


A conical pore has a much lower resistance than an equivalent cylindrical

pore of the same limiting diameter.

Additional advantage of using asymmetric pores is

that we limit the interactions zone in the pore, which makes the sensor's response faster

and is expected to lower the detection limit.

The pores were subsequently covered

electrolessly with gold (40), which resulted in formation of gold tubes (123).

principles of the nanodevice operation were shown first with the system biotin-

streptavidin, which is known to have a very high binding constant, and the binding is

regarded as practically irreversible (191).

We also checked applicability of the device for

sensing protein-protein interactions on the example of protein G modified Au tubes.

Protein G is a cell surface-associated protein isolated from Goward Group G Streptococci

and binds with high affinity immunoglobulins (IgG's) (192,193).

In our experiments we

have used cat IgG which has no affinity for protein G, and horse IgG which has a strong

affinity to the protein G (194,195).

It is also shown the potential of this nanopore system

in building sensors for warfare agents, on the example of ricin.



We used 12 jam thick polyethylene terephthalate (Hostaphan RN 12, Hoechst) foils,

irradiated with single swift heavy ions (

(UNILAC, GSI Darmstadt). To obtain

Au, Xe, U) (196) of 2.2 GeV kinetic energy

conical pores, the single ion irradiated polymer

foils were mounted between two chambers of a conductivity cell and etched from one

side in 9 M NaOH, as described elsewhere (120,122,128). The chemical etching was

resulted in increase of the pore diameter.

The diameter of the large pore opening was

estimated on the basis of the so called bulk etch rate, which for PET at 9 M NaOH and

room temperature is 2.13 nm/min (120).

For example,

2 hours etching results in 520 nm

pore diameter.

The diameter of the small opening was obtained on the basis of

conductivity measurements assuming a conical shape of the pore (37).

prepared had a diameter of-40 nm.

The pores we

Plating the nanopores with gold resulted in final

diameters varying between 5 and 20 nm, function of the gold deposition time.

The big

diameter of the pores did not change significantly.

Electroless plating of PET membranes

The electroless plating was performed according to the procedure described

elsewhere (40).

The plating process was performed at 3.6 OC, and pH 9.9.


after 2.5 hours of plating, the gold layer has an approximative thickness of 4 nm.


Lysozyme, streptavidin, bovine serum albumin (BSA), protein G

cat IgG and horse IgG were purchased from Sigma Aldrich.

- biotin labeled,

EZ-Link Biotin-HPDP or

(N-(6-(Biotinamido)hexyl)-3'-(2'-pyridyldithio)-propionamide was bought from Pierce.

Ricin Toxoid and Biotinylated Anti-Ricin IgG were bought from Toxin Technology, Inc.,

Sarasota, FL.

Ricin Toxoid has been toxoided using glutaraldehyde crosslinking and has

less than 1% of the original toxicity.

Experimental Setup

The single conical-Au-nanotube membrane was mounted between two halves of a

nair n,,1,, 44..4... n.ll /1 d)0\ nn n A ,I A nfl'*! .l.-:n4-.*..n nran ,-,,s-.A :-*. tnnl-. Ll t nnll nal.,-<.4rt