«.9: a... ‘4—

“an.

 

 

 

 

 

i
-
.5
Ii
fag)-
-. e;-
’9'";
h. 4:.“ , ;
”57;”? :L‘ ‘f..'l‘!§a?:g ’
m 3~vnfiV?-‘%‘fiv;s “gr
.t! :. stratum: & N.
. r" .‘ g y 4
w fl-r‘vr *:
.rt‘rflifirlflh
“9.;qu ‘
a ”41-3.
i“ v
5; 3" 45‘ ‘3'“ ' .
, .. _ ‘ 2 .
“3,-5.1, ' 5%:5'5'5'2552'“ *‘a'r’ 75“"
fit? -:'.._...2:?~‘~ "’5'“ IE
\ “33$.“ 'vfir ”.3." w
'L n . -
X‘. ’. «1153:..uxé—flw
fizvfimzuzggmh“ ,
2%.;‘..:¢L"'2‘l 32:. 14x1"
3' ~.:fw-‘.~;..kmyn v
' A“. W ‘2.“ ,1;
.
n; - > -
9v 1‘“ ' AM '
uf tr.» 5%
, . . . _
‘9!" “:7 .33: "-. an.
..-, ,...V W -. A, 9...
‘1‘? ##WA‘: .. .. n.“
‘ 4.. “3......L " 1.1:" 1.;
a» $3.“- ' ,. 5“.”
€

,~
‘-
1w

«rd
mg.” an 7.2 “r w :m > "1.5.1;
. ‘ “ "EasEC-Eh

r' . 1'3“, .. («1 ,
11“”:- . 1 W ‘4 , '32. . *3: :12 - Sam. “.4...
31H" "‘ “" "’ ; '4" . .. - Efifiénv'ggifix V m f , .. &”’”}Za}%“:?~ Fig“:
'i‘ a £99.; 1 m». $11, an»: 5.3. 1 ‘ 4 ‘ .3 “QE'9m \ .2.- .m-r- ..-.-
.w ‘ _ w hcAfgu-iémfi. ‘1“ .31.: " . ' it??? '

v4 ,_.__ "

.uits‘; A
1‘1!” 4"- ‘
on: f .

X

 

THEScs

This is to certify that the

thesis entitled

Fundamental Studies of Organic/Inorganic Carposite Thin Films
As Selective Coatings for Mass Sensors

presented by

Gregory 0. Noonan

has been accepted towards fulfillment
of the requirements for

M.S. degree in Cherm stry

MMMM

Ugfééor professor
Date @Hk/ g; /7 74

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

Illilllllllllllilllllll'llliillllillll

293 01033 6778

 

LIBRARY
MIchIgan State
UnlverSIty

 

 

 

PLACE II RETURN BOX to roman this checkout from your record. ‘
TO AVOID FINES return on or before date duo.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I'M Affirmative WM OpportunIty Infiltwon
7 7 W

pus-9.1

FUNDAMENTAL STUDIES OF ORGANIC/INORGANIC COMPOSITE THIN FILMS
AS SELECTIVE COATINGS FOR MASS SENSORS

By

Gregory 0. Noonan

A THESIS
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degreee of

MASTER OF SCIENCE

Department of Chemistry

1994

ABSTRACT

FUNDAMENTAL STUDIES OF ORGANIC/INORGANIC COMPOSITE THIN FILMS
AS SELECTIVE COATINGS FOR MASS SENSORS

By

Gregory 0. Noonan

The mass sensing properties of organic/inorganic composite films have been
studied through the use of surface acoustic wave (SAW) devices ,quartz crystal
microbalances (QCM), and Fourier transform infrared spectroscopy.

Zirconium acetate polymer oxide glasses were prepared by hydrolysis of
zirconium (IV) n-propoxide in excess acetic acid. X-ray photoelectron spectroscopy
(XPS) and FTIR were used to determine structure of the films. The zirconium acetate
films consist of Zr-O backbone coordinated with bridging and chelating bidentate ligands.
The films responded to alcohols while rejecting alkanes. It also shows molecular sieving
properties which leads to the exclusion of larger molecules from the pores of the film.

Poly(dimethylmethylsilicone) and poly(phenylmethyldiphenylsilicone) films were
compared using SAW devices. FTIR spectroscopy using an attenuated total reflectance
attachment was also used to observe viscoelastic changes of both films. The responses of
both films was correlated to solvatochromic and chromatographic data. Any deviations

from the expected trends were attributed to the FTIR observed viscoelastic changes.

In memory of Donald R. Shaw for his

support, pride, and encouragement.

iii

ACKNOWLEDGEMENTS

I would like to thank Dr. Jeffrey Ledford for his support and guidance in
completing this research. Without his ideas, encouragement, constant focusing of my
research, and unyielding patience this body of work would not have been completed. I
want also like to thank past and present members of the Ledford group: Kathy, Tom, Paul,
Radu, and Jefl‘ R. for both scientific and emotional support during this time. Thank you
Per, Mike, Ed, Maria, and the new addition, Sara T., for your fi'iendship, basketball, and
home maintenance over the last two years. I also wish to acknowledge the other graduate
students Matt, Chris, Sara, Dana, Tim, and Elizabeth for the needed distractions from
graduate research.

I would like to thank my parents, stepparents, brother, and sister for supporting
my decision to return to graduate school. I want thank the Ouimette family for their
encouragement in everything I have done. I want to acknowledge the MSU Vet school
contingent: Tom and Kelley, Scott, Allison, Lisa, I-Iiedi, Kay and Jeff, and Vicki and Tim
for their friendship before and during my program. Thank you Joe and Sue, for a place to
live and a “friendly ear” whenever it was needed. I want to recognize Jake for helping me
stay sane through walks in the woods and playing ball. You truly are a “man’s best
fi’iend”. Finally I would like to thank Jane, a very important friend. It is not an
exaggeration to say that without you I would not be writing this acknowledgment. You
made many sacrifices, but still gave me the support and encouragement I needed. I have
learned many things from you about what is and isn’t important, and I hope that you are

proud of me because I am proud of you.

This research was sponsored in part by MARTIN MARIETTA ENERGY
SYSTEMS, INC., and performed under Subcontract 19K-MP607C for the Oak ridge K-

25 Site, which is managed for the US. DEPARTMENT OF ENERGY under Contract
DE-AC05-84OR21400.

TABLE OF CONTENTS

Page
List of Tables ............................................................................................................... vi
List of Figures ............................................................................................................. vii
Chapter 1: Introduction to the Study of Mass-Based Chemical Sensors ....................... l
1.1. Concepts of Chemical Sensing .............................................................. l
1.2. Overview of Piezoelectric Devices ........................................................ 2
1.2.1. Piezoelectric Effect ........................................................ 2
1.2.2. Quartz Crystal Microbalance (QCM) ............................. 5
1.2.3. Surface Acoustic Wave (SAW) Devices ......................... 10
1.3. Response Mechanisms of Acoustic Wave Devices ................................. 12
1.3.1. Partitioning .................................................................... 12
1.3.2. Viscoelastic Effects on Sensor Response ........................ 15
1.4. Overview of Sol-Gel Processing ........................................................... 17
1.4.1. Sol-Gel Chemistry ......................................................... 17
1.4.2. Preparation of Metal Carboxylate Thin Films ................. 20
1.4.3. Use of Sol-Gel Materials in Mass Sensors ...................... 22
1.5. Characterization of Thin Film Sensors ................................................... 23
1.5.1. X-ray Photoelectron Spectroscopy (XPS) ...................... 24
1.5.2. FTIR/Attenuated Total Reflectance (ATR) .................... 26
1.5.3. Sensor Test Apparatus ................................................... 28
1.6. Focus of Research ................................................................................ 29
References ................................................................................ 31
Chapter 2: Structure and Mass Sensing Properties of Acetate Thin Films .................... 34
2.1. Introduction .......................................................................................... 34
2.2. Experimental ........................................................................................ 35
2.3. Results ................................................................................................. 41
2.4. Discussion ............................................................................................ 50
2.5. Conclusions .......................................................................................... 61

References ................................................................................ 62

Chapter 3: Study of Polysiloxanes as Selective Coatings for SAW-based Sensors ........ 63

3.1. Introduction .......................................................................................... 64

3 .2. Experimental ........................................................................................ 66

3.3. Results ................................................................................................. 70

3.4. Discussion ............................................................................................ 82

3.5. Conclusions .......................................................................................... 87
References ................................................................................ 88

Chapter 4: Future Work .............................................................................................. 89
4.1. Tailoring Metal Carboxylate Polymer Oxide Glasses ............................. 89

4.2. Molecular Recognition Centers in Polymer Oxide Glasses ..................... 93
References ................................................................................ 98

vii

Table

Table 2.1.

Table 2.2

Table 3.1.

Table 3.2

List of Tables

Page
Probe analytes. ........................................................................................ 36
Analyte responses on coated and uncoated QCM. .................................... 47
List of analyte boiling points, dipole moments, and sensor response. ....... 67
Response ratios for probe analytes. .......................................................... 76

viii

List of Figures

Figure Page
Figure 1.1. Fundamental thickness shear mode vibration. ........................................... 6
Figure 1.2. ATR cell used for analyte/film interaction studies ..................................... 28
Figure 1.3. Gas handling and sensor test cell schematic. ............................................. 29
Figure 2.1. Schematic diagram of sensor test apparatus. ............................................ 40

Figure 2.2. FTIR spectra of (a) acetic acid, (b) zirconium (IV) n-propoxide in n-
propanol, and (c) zirconium acetate film. ................................................. 43

Figure 2.3. FTIR spectrum of carboxylate region obtained for (a) sodium acetate
and (b) zirconium acetate film. ................................................................. 45

Figure 2.4. Variation in QCM response (Hz/mmole) versus analyte boiling point
for uncoated quartz crystal. ..................................................................... 46

Figure 2.5. Variation in QCM response (Hz/mmole) versus analyte boiling point
for zirconium acetate coated quartz crystal. ............................................. 49

Figure 2.6. Variation in hydroxyl stretch intensity for methanol (—) and ethanol
(---) as a function of time following analyte injection. .............................. 51

Figure 2.7. Variation in relative intensity of the hydroxyl stretch for n-propanol
(—) and C-H stretch for heptane (--) as a function of time following
analyte injection ....................................................................................... 52

Figure 2.8. FTIR spectra of zirconium acetate film (a) before and (b) after heating. ..... 54

Figure 2.9

Figure 2.10.

Figure 3.1.
Figure 3.2.
Figure 3.3
Figure 3.4.
Figure 3.5
Figure 3.6.

Figure 3.7.

Figure 3.8.

Figure 3.9.

Figure 4.1.

Variation in normalized QCM response (zirconium acetate coated

crystal/uncoated crystal) versus analyte boiling point. .............................. 59
Variation in normalized QCM response (zirconium acetate coated

crystal/uncoated crystal) versus analyte molar volume .............................. 60
General structure of polysiloxane polymer. .............................................. 64
Response curve for blank SAW device ..................................................... 72
Response curve for OV-101 coated SAW device. .................................... 73
Response curve for OV-25 coated SAW device. ...................................... 74
Response ratio curve for OV-25/OV101 comparison. .............................. 77
FTIR/ATR spectra of methanol on OV-lOl. . .......................................... 78

FTIR/ATR spectra of analyte induced morphology changes for
nonane on OV-lOl. ................................................................................. 8O

FTIR/ATR spectra of analyte induced morphology changes for

pyridine on OV—25 ................................................................................... 81
FTIR/ATR spectra of analyte induced morphology changes for

nonane on OV-25. ................................................................................... 83
FTIR spectrum of zirconium valerate/benzoate film. ................................ 91

Chapter 1

Introduction to the Study of Mass-Based Chemical Sensors

1.1. Concepts of Chemical Sensing

A chemical sensor has been defined as a device which provides direct information
about the chemical composition of its environment. It consists of a physical transducer
and a chemically selective layer [1]. This definition is often narrowed further to include
only devices that reversibly respond to changes in the chemical environment. Using this
definition, indicator tubes and test strips are usually not considered to be chemical sensors.
Also, analytical instrumentation that requires several discrete steps to obtain data are not
within the standard definition of chemical sensors.

In the development of chemical sensors it is necessary to consider the operation of
the transducer as well as the interactions taking place in the chemically selective coating.
The transducer (e. g., mass, optical, thermal, and/or electrochemical device) defines the
physical or chemical parameter to be monitored while the chemical layer enhances the
selectivity and sensitivity of the device. In many cases, transducers are available that
inherently provide the desired level of sensitivity. Understanding chemical interactions and
synthesizing chemically selective materials may be considered the frontier of sensor

development.

Acoustic wave or mass sensors respond to changes at the surface or coating of the
device. These changes can be caused by the addition or removal of mass, changes in the
density or viscosity of the coating or even changes in the conductivity of the overlayer.
The large number of events that can produce a response from a mass sensor can be a
disadvantage and a benefit. One major disadvantage is the difficulty in determining the
fundamental processes that generate a response. However, it is the multiple response
mechanism which allows the use of mass sensors for the study of a variety of analyte/film
interaction types and material properties [2]. Other advantages of mass sensors are their
simplicity of construction and operation, light weight, and low power requirements.
Additionally, mass sensors are relatively inexpensive and have a high degree of sensitivity
for a broad range of compounds. This high sensitivity can lead to problems with
interferents and thus dictates the development of chemically selective coatings and a

detailed understanding of analyte/film interactions.

1.2. Overview of Piezoelectric Devices

1.2.1. Piezoelectric Effect

The piezoelectric effect was first reported by Jacque and Pierre Curie in 1880 [3].
They observed that with certain compounds the application of a mechanical force
produced an electrical signal. A short time later it was also noted that the application of

an electrical signal produced a deformation in the crystal. Since these initial discoveries,

the piezoelectric effect has been documented in a large number of compounds such as
rochelle salt (NaKC4H406 - 4H20), lead zirconium titanate, and or-SiOZ.

To fully explore the piezoelectric effect, it is first necessary to understand the
relationship between stress and strain in solids and the displacement of particles within a
solid. Displacement is defined as the vector indicating the difference between equilibrium
and perturbation. Strain is commonly used to define relative displacement and the two are

related by:

S=Vs~u (1.1)

where S is the strain, u is the displacement vector, and Vs is the symmetric gradient
operator. In any elastic solid the application of a strain produces a restoring force defined
as stress. In the case of a piezoelectric solid the strain also generates an electric field.

Stress and strain are related by:

T=cS (1.2)

or the inverse relationship:

S =sT (1.3)

T is a second rank tensor representing stress, c is a matrix of material stifi‘ness constants,
and s, in the inverse relationship, is called the compliance matrix. Elements of the
compliance matrix are derived from stiffness constants that depend on the symmetry of the

solid.

In the case of an acoustic wave it is not only necessary to consider a single force,
but also a time dependent force based on the propagation of the wave. It is also important
in a time dependent situation to account for losses due to dampening of the wave.

Accounting for time dependency and dampening effects, equation 1.3 can be modified to:
T=cS+ndS/dt (1.4)

where r] is a matrix of viscosity constants.

Materials which display piezoelectric characteristics do so because of the lack of a
center of inversion. At equilibrium, before strain, the dipoles cancel resulting in a net
charge of zero within the solid. Application of a strain distortion in the form of an
extension or compression of the structure leads to a macroscopic charge separation within
the solid. Of the 32 point groups known, 20 lack inversion centers, however not all 20
exhibit appreciable piezoelectric characteristics. The strain and displacement can be

equated to the stress and electrical forces acting on the crystal by:

D=eE+dT (1.5)

s=d'e+sT (1.6)

where E is the electrical field gradient, 8 is the electrical permittivity matrix, and d',d are
piezoelectric strain coefficient matrices. Equations 1.5 and 1.6 show that the magnitude
of the measurable piezoelectric effect is determined by the values of the piezoelectric

coefficients.

After considering the physical properties leading to piezoelectricity, it is important
to consider the behavior of the wave created in these materials. The Christofl‘el equation

is used to determine the propagation of waves within solids:

k21‘~-v-= p 1.7
u 1 A2”, ( )

I‘ij is the Christoffel matrix, which is a function of the piezoelectric coupling constants and
the electrical permittivity of the solid, and vi is the particle velocity. Solution of the
Christofi‘el matrix leads to simultaneous equations describing the modes of propagation of
the acoustic waves. In anisotropic solids three modes of mutually orthogonal wave
propagation are established. Displacements along the direction of propagation are
referred to as longitudinal waves, while perpendicular displacements are referred to as
shear waves [4].

or-SiOz is the material most often used for mass sensor applications. Quartz is
physically and chemically rugged, and is fairly inexpensive. Both quartz crystal
microbalances (QCM) and surface acoustic wave (SAW) devices are employed in mass
sensing applications. QCMs utilize a thickness shear mode wave to measure the mass
change at the crystal surface while SAW devices depend on the propagation of Rayleigh

waves.

1.2.2. Quartz Crystal Microbalance (QCM)

A quartz crystal will exhibit numerous modes of vibration when placed in an

electronic feedback loop. These include, but are not limited to, longitudinal (extensional),

lateral (shear and flexural) and torsional (twist) vibrations. There is also the possibility of
overtones and the coupling of modes. Removal of unwanted modes and enhancement of
waves best suited for sensor applications are afl‘ected by the geometric cut of the crystal
and placement of the electrodes. The singly rotated Y-cut enhances the shear wave mode

which has the greatest sensitivity to the addition of mass (Figure 1.1).

 

 

 

 

 

/_
/ / 7‘
/
/ / \
/ /
/ / /
/ / /
/ / /
/ / /
/ / /
/ / /
/ / /
/ / / /
i" 4..— 1 / /
\ \ \ ‘\ /
\ —-——> /
Figure 1.1. Fundamental thickness shear mode vibration.

A specific class of rotated Y-cut crystals known as AT-cut are most often used for mass
sensor applications. They exhibit the lowest temperature dependence near room
temperature [5]. Plating of electrodes on either face of a crystal allows for the application
of an oscillating electric field across the device. The electric field generates an acoustic
shear wave through the crystal and produces a standing wave with nodes at either face.
Since the piezoelectric material, orientation and wave mode fix the velocity, equation 1.8

defines the resonant frequencies that may exist for a given thickness.

mt
T (1.8)
Where k is the wave vector, b is the crystal thickness, and n is an integer. This expression
simply indicates that the thickness of a bulk wave device must be an integral number of
half wavelengths for an acoustic wave to be resonant.

In 1959, Sauerbrey developed an equation to relate the frequency change of a
quartz crystal to the mass added at the surface of the crystal [6]. In developing this
relationship Sauerbrey assumed that the added material simply increased the thickness of
the crystal. The resulting expression (equation 1.9) does not consider the density,

viscosity or shear modulus of the added layer.
2
df = —2fq Am/(pqvq) (1.9)

pq and vq are the density and shear wave velocity of quartz , and fq is the firndamental

frequency of the quartz oscillator. Equation 1.9 is written in many forms including:

M=Cfo (1.10)

where
Cf =2fq2/(pqvq) (1.11)

is the mass sensitivity or calibration constant of a QCM, and mf is the mass added to the
crystal [4]. Although the assumptions made by Sauerbrey are physically unrealistic,
experimenatal data showed the Sauerbrey equation to be accurate for layers no greater

than 2% of the crystal mass [7-9].

As the use of QCMs as thickness monitors became more widespread, new
expressions were developed to more accurately describe the change in frequency with the
increase in mass. The first equation to improve on Sauerbrey's assumptions used the areal
density (ml), defined as the mass per unit area of a deposited film, and assumed it was

linearly proportional to the frequency change.
v
mf =.p_q_?.[_l-_.l_:l (1.12)

Equation 1.12 can also be written in terms of film thickness and density as follows:

Where fc is the fi'equency of the coated crystal. These equations were shown to be
accurate for a 10% change in the base fi'equency of the QCM [10,11]. Although empirical
data showed the above equations appeared to be more accurate, there was still no
consideration of the overlayer physical properties.

Lu and Lewis [12] developed an expression which considered the differences in the
shear modulus, density and thickness of the quartz and film. The shear mode acoustic

impedence was defined as:

I

Z. 49.11.)? (1.14)

Of

l

Zf=(Pfl»lf)2 (1-15)

for quartz and film, respectively. This allowed Lu and Lewis to write:
p ,1 f =(pqvq/21thc)tan"{Ztan[1t(fq -f,)/fq]} (1.16)

In the case where film and quartz acoustic impedences are matching, Z=1 and equation

1.15 simplifies to equation 1.12. For unmatched impedances and very small mass changes:

At“
”a <<l (1.17)

and equation 1.16 reduces to the original Sauerbrey expression (equation 1.9). Therefore,
equation 1.16 is a more general expression describing the relatonship between the change
in frequency for a change in mass. The Sauerbrey expression and equation 1.12 are
specific limits of equation 1.16.

Perhaps the most 118le expression describing the frequency dependence on the
mass and physical properties of an overlayer was derived by Kanazawa [13]. The relation,
given in equation 1.18, was developed to describe the response of a QCM in contact with

liquids. However, the expression is applicable to any viscous overlayer.

A/ = —/.%(p1n./upu)% (1.18)

TIL is the absolute viscosity, )1 is the shear modulus, and all other terms being as desribed
above. Experimental data from Kanazawa and other researchers using 5 and 9 MHz

QCMs supported the validity of the above equation [14].

10

1.2.3. Surface Acoustic Wave (SAW) Devices

A second type of mass sensor which utilizes the piezoelectric effect is the surface
acoustic wave (SAW) device. Surface waves were first described by Rayleigh in the study
of the propagation of shock waves fi'om earthquakes along the earth's crust [15] and are
ofien referred to as Rayleigh waves. Rayleigh waves are actually a specific type of surface
acoustic wave and are the most common type used in sensor applications. Their
propagation through the piezoelectric material is described by the stress-strain
relationships given previously, however it is the specific particle displacement that
distinguishes them as Rayleigh waves.

Rayleigh waves are confined largely to the surface, approximately one acoustic
wavelength thick, and consist of a combination of two waves of equal phase velocity - a
longitudinal wave whose displacement is parallel to the direction of propagation and a
shear wave with displacement perpendicular to the interface plane. The combination of
these waves produces an elliptical particle displacement at the surface of the material. As
with bulk acoustic wave devices, the velocity of the wave is determined by the
piezoelectric material and the specific cut of the crystal.

Similar to QCM devices, an oscillating radio fi'equency must be applied to the
crystal to launch a SAW. However, unlike the QCM, the transducers to produce Rayleigh
waves are placed on the same face of the piezoelectric crystal. An interdigitated wave.
The periodicity, or center to center distance between twtransducer (IDT), developed by

White et al. [16], can be used to launch and detect a SAW. The dimensions of the lDTs

11

establish many of the characteristics of the acoustic 0 consecutive fingers, determines the
wavelength of the wave. Therefore, the fiequency of the SAW on any given material is
related to the spacing of the IDTs.

Since the acoustic wave is confined to the surface, any matter present at the
surface of the crystal interacts strongly with the wave. Unlike bulk oscillators, the
frequency of the acoustic wave is not altered by interaction with an added mass.
However, the phase of the surface wave is affected by a change in the mass at the surface.
The phase change is caused by variation in the velocity of the propagating wave.
Although the frequency does not change, the SAW device can be placed in an oscillator
feedback circuit that relates the decrease in velocity to frequency change [17]. Wolhtjen
and Dessy [17] demonstrated that the oscillator circuit produced a better signal to noise
ratio than direct velocity or amplitude measurements.

Since the recorded fi'equency change is not directly related to the change in mass
at the crystal surface, the Sauerbrey equation does not apply for SAW devices. Instead

the following equation [18] is used:

 

M = (’61 +k2 )jbzhp' —kzjb2h[ :2. (£‘1-2lfl' )] (1.19)

Where k1 and k2 are material constants for the substrate, f 0 is the unperturbed resonance
frequency, h is film thickness, p' is film density, VR unperturbed Rayleigh wave velocity, N
is the Lame constant, and u’ is the overlayer modulus. The first term of equation 1.19
depends on the mass per unit area of the overlayer film, while the second term depends on

the mechanical properties of the polymer coating. For a thin (_<. 0.1 pm) lossless polymer

12

above its glass transition temperature (T g) the frequency change from the film physical

properties is considered negligible and equation 1.19 reduces to equation 1.20.

A! 5(1, +1, )fo’hp' (1.20)

1.3. Response Mechanisms of Acoustic Wave Devices

Acoustic wave devices offer the necessary sensitivity to be useful for chemical
sensing applications. However, the uncoated quartz surfaces or gold electrodes do not
selectively adsorb specific analytes. Therefore it is necessary to modify the surface to
increase the selectivity of the device. This modification commonly involves covalently
bound sites, Langmuir-Blodgett films or organic and inorganic polymeric films.

In the case of polymer films, the response of the acoustic wave device is normally
described in terms of two distinct interactions. First, mass added upon adsorption of the
analyte afl‘ects the propagation of the acoustic wave as dicussed above. Second, the
adsorbed analyte may cause the polymer to swell or contract leading to changes in the
viscoelastic properties of the film. Viscoelastic changes also effect the propagation of the

wave and lead to changes in the resonant fiequency, amplitude or velocity of the wave.

1.3.1. Partitioning

Partition coefficients (K) are a measure of the equilibrium between analyte in the

vapor phase and that sorbed in the stationary phase and are described by equation 1.21.

13

K=C%V (1.21)

C3 is the analyte concentration in the stationary phase and CV is the concentration in the
vapor phase. K values are conventionally determined through gas—liquid chromatography
(GLC) specific retention volumes and describe the affinity of a particular stationary phase
for a compound of interest.

King [19] described the response of a coated QCM in terms of the partitioning of
the probe analytes into the polymer coating. He observed that the extent of partitioning
increased with analyte boiling point. Recently Grate et a1. [20] proposed the use of SAW
frequency shifts to determine analyte partition coefficients (KSAW)- He related Ks AW to

the fi'equncy shift of the device by:
Nu =AfszKSAw/Ps (122)

where f v is the sensor frequency shift, f s is the frequency shift of the coating, and ps is
the density of the polymer film. Their experiments found that KSAW and KGLC values
differed by roughly a factor of 4, however similar trends were noted for both systems.
This suggested that K values determined from sensor measurements could be used to
describe trends in analyte film interactions on SAW devices.

Three models describing the interaction of vapor with coating material presently
exist. The boiling point model, the most basic of the three, uses molecular weight,
density, vapor pressure, and vapor activity coefficients to describe interaction of analytes
with films. This model does not consider deviations from ideal vapor-polymer behavior.

The solubility parameter model attempts to account for deviations from ideality through

14

the use of a Hildebrand solubility parameter. However, this method also does not
adequately describe the interactions of vapor and polymer coating. Grate and Abraham
[21] described the use of solubility parameters in linear solvation energy relationships
(LSERs) to quantify coating material solubility properties. LSERs account for dispersion
forces (induced-dipole/induced-dipole), dipole induction interactions (dipole/induced-
dipole), dipole orientation interactions (dipole orientation interactions), and hydrogen-

bonding interactions using the following expression:
logK=c~I-a0112'I +bB§I +llong6+ rR2+s1t; (1.23)

The a2“ and B2B parameter indicate the hydrogen—bond acid strength and base strength,
respectively. They are derived using values of equilibrium constants for the complexation
of acids and bases by reference acids and bases in an inert solvent. Dispersion interactions
(longé) are a combination of the energy required to form a cavity in the film, the energy
obtained when the vapor fills the cavity, the subsequent film reorganization energy around
the vapor, and the resultant attractive interactions between the vapor and the film. R2 is a
measure of the ability of a solute to interact with a solvent through 11 and 1t electron pairs.
The ability of a compound to stabilize a neighboring charge or dipole is measured by 1:2".
A more detailed description of each variable and collections of available solute parameters
may be found in the work of Abraham et a]. [22-3 2].

From expression (1.22), KSAW values can be calculated for each analyte and film.
Once these values and the appropriate analyte constants (orzH , [32H , long6, R2, and 1:2")

are known, LSERs can be developed to determine the values of the constants a, b, c, I, r,

15

and s which describe the solubility interactions taking place between the vapor molecules
and the polymer phase on the SAW device. The magnitude of these coemcients is
believed to describe the interaction mechanisms between a film and analyte vapor.

Patrash and Zellers [33] used partition coefficients determined by SAW analysis to
determine LSER values for a variety of films. They also showed that the response of an
untested analyte could be estimated within 25% from the LSER data. In addition, Patrash
and Zellers compared the LSER model to the boiling point and solubility parameter

models and found the LSER model to be more accurate.

1.3.2. Viscoelastic Effects on Sensor Response

Along with the mass effect, the propagation of acoustic waves also depends on the
physical properties of the applied coating. Wohltjen [18] noted this dependence and
accounted for polymer physical properties through the second term of equation 1.19.
Further work has established that polymer modulus changes caused by temperature
changes or vapor sorption could result in a shift in the frequency of the SAW device [34-
3536]. However, in both the coating of SAW devices and in measuring analyte film
interactions, this term is ofiened ignored [20,37].

Grate er a1. [38] proposed that the frequency shift due to polymer modulus
changes dominate the response of coated SAW device to probe analytes. Using
partitioning data fi'om an earlier study [20], Grate developed equation 1.24 to estimate the
change in frequency due to mass loading and vapor induced polymer swelling. The first

term of equation 1.24 describes the response due to mass loading of vapor sorbed by the

l6

polymer film. The second term estimates the fi'equency shift caused by vapor induced

 

swelling of the SAW coating.
Af.=[~sc;xac).[__a.%c;x chng) a...)
S S

,03 is the density of the polymer coating when in equilibrium with the sorbed vapor, pL is
the density of the vapor in the liquid state, a is the polymer thermal expansion coefficient,
and As AW is the frequency change of a polymer coated SAW due to thermally induced
swelling. Grate assumed that fi'equency shifts accompanying vapor induced swelling were
comparable in magnitude to shifts due to thermal swelling.

Ballantine [3 9] evaluated the validity of equation 1.24 in studying the fi'equency
response and attenuation of poly(isobutylene) (PIB) coated SAW devices. Using airbrush
and drop coating techniques, Ballantine varied film thickness from 8 to 250 kHz.
Ballantine did not observe significant viscoelastic effects for airbrushed films of less than
200 kHz thickness. However, for drop coated films resonance effects were noted at
periodic film thicknesses. This difference can be attributed to the different morphologies
observed for the two types of films. The airbrushed polymers appeared as overlapping
plates, while the drop coated films were uniform over the surface of the crystal.

It is now widely accepted that polymer swelling can occur upon the sorption of
certain analytes and that this swelling affects the propagation of the acoustic wave.
Presently, no quantitative model to assess the importance of this efi‘ect has been
developed. The lack of an appropriate model has led researchers using solubility

interactions to ignore the efi'ect of viscoelastic changes [33]. Considerable research will

17

be required to obtain an understanding of vapor induced modulus changes and their effect
on acoustic wave devices. Unfortunately, viscoelastic changes are likely to be analyte and

film dependent, making a universal theory difficult to develop.

1.4. Overview of Sol-Gel Processing

Sol-gel processing is a relatively new approach for the production of glasses and
ceramics. Typical preparation techniques use metal alkoxide precursors in a low
temperature process to produce metal oxides and/or organic-inorganic composite
materials. The low temperature processing of the sols and gels allows for greater control
of glass properties and incorporation of inorganic and organic guest molecules. In
addition, the rheology of the sols allows the glasses to be cast in many forms including
fibers, thin films, and monoliths. Once cast, the optical clarity, extensive pore structure,
and stability of the glasses make them useful materials for chemical sensing applications.
The specific synthetic route employed depends on the desired properties of the gel. Since
sol-gel chemistry has been extensively reviewed [40-4142], only a brief introduction will

be presented below.

1.4.1. Sol-Gel Chemistry

Most sol-gel preparations involve the hydrolysis of a metal alkoxide followed by a

condensation and polymerisation step. A general synthetic route involving the reaction of

18

a metal alkoxide is shown in reaction (1). The addition of water causes the hydrolysis of

the metal alkoxide:

M-OR+H20——>M-OH+ROH

Once hydroxy groups are generated two competing reactions, olation or oxolation may
occur. Oxolation is shown in reaction (2) and forms hydroxy bridges through the

elimination of of solvent molecules.

M—OH+M-OID(——)M—OH-M+XOH (2)

X in the above reaction can be either a hydrogen or an alkyl group. Olation forms oxygen

bridges through the the elimination of water or alcohol as shown in reaction 3.

M—OH+M-OX——)M-—O-M+XOH (3)

As in reaction (2), X can be either hydrogen or an alkyl group.

The above three reactions can be described as 8N2 type nucleophilic substitutions.
The mechanism involves the addition of a nucleophile to the metal center (a), followed by
the transfer of a proton toward an OR group (b), and finally departure of the positively

charged leaving group (c). These three steps are outlined in the reaction scheme below:

H5+

XOH +MOR —> O-M-OBR —-> x0 —M-O/ -—-> X0 -M + ROH
/ \R
X
(a) (b) (c)

19

The rate of the nucleophilic reaction depends on the charge of the metal atom (5M) and
the ability of the metal atom to increase its coordination sphere (N). This latter factor
accounts for the general rule that the reactivity of analogous metal alkoxides increases
going down the periodic table. The size and shape of the alkoxide group also affects the
chemical reactivity of the metal precursor. Large or bulky ligands can decrease the
reaction rate by sterically hindering attachment of the nucleophile. In addition, the ability
of the ligand to form a stable leaving group will also affect the reactivity.

Most of the initial sol-gel work utilized fairly unreactive silicon alkoxides. The
high electronegativity, small size and fourfold coordination of silicon forces the use of acid
and base catalysts to increase the rates of hydrolysis and condensation. On the other hand,
the larger size, smaller charge, and ability to increase N causes transition metal alkoxides
to be extremely reactive. Thus, steps to slow the rates of hydrolysis and condensation are
necessary to avoid the formation of insoluble oxides.

Pioneering work in the area of sol-gel synthesis was performed by Yamane [43],
Yoldas [44,45], and Hench [46,47]. In general, synthesis involves the mixing of metal
alkoxides with alcohol, followed by the addition of excess water and either an acid or base
catalyst. This mixture is normally vigorously stirred and refluxed for several hours at
slightly elevated temperatures. Through drying, the gel is allowed to form over a period
that varies from several hours to several months. Yamane [43] correlated processing
conditions with properties of gels and reported that gelation time decreased, while density
and pore size increased with increasing ammonia concentrations. Yoldas [44,45] studied

alumina gels and determined that acid catalysts must be noncomplexing and sufficiently

20

strong to produce a charge effect. Hench [46,47] showed that the addition of drying
control chemical additives (DCCAs) such as, formamide, glycerol, and organic acids,

could be used to control pore size distribution in gels.

1.4.2. Preparation of Metal Carboxylate Thin Films

As mentioned above the reactions of transition metal alkoxide complexes are rapid
compared to the rates of hydrolysis and condensation for silicon alkoxides. The increased
rates are attributed to the metals' small charge, large size and ability to extend its
coordination. Because of this reactivity, transition metal alkoxides can react to form
insoluble metal oxide precipitates. To prevent precipitation chemical additives such as
solvents, B-diketonates, or organic acids are normally used. In most instances they are
nucleophilic molecules that react with the alkoxide to form new molecular species with
difi‘erent structure, functionality, and reactivity. These substitutions produce a decoupling
of hydrolysis and condensation leading to the growth of particles and the formation of
gels.

Carboxylic acids, especially acetic acid, are commonly used chemical control
agents. Langkammerer [48] and Haslam [49] first reported the use of carboxylic acids
chemical additives to produce metal carboxylate films. More recently carboxylic acid
modification has been investigated by Berglund [50,51] and Livage [52].

A variety of reactions may occur upon the addition of carboxylic acids to transition
metal alkoxides. Perhaps the most common reaction products are metal alkoxide-

carboxylate species; however, additional ligands (e. g., esters, hydroxides, alcohols, and

21

water) may be attached to the metal center [40,42]. In the initial step, the alkoxide ligand

is displaced, via a nucleophilic substitution by the carboxylic acid.
(no)4 M + R'COOH ——+(rt'coo)1vr(oa)3 + ROH (4)

The addition of water to the metal alkoxide, carboxylic acid solution produces hydrolysis

shown in reaction 5:
(Ir'coo)1vr(oa)3 + 1120———>M(OH)(R'COO)(OR)2 (5)
The hydrolysis step is then followed by condensation:

2 M(OH)(R'COO)(OR)2 ——>

(R. C00)(OR)2 M - O - M(OR)2 (R' coo) (6)

which continues to form polymeric gels. In the presence of excess acid there is no
evidence for the presence of esters, alkoxides or alcohols coordinated to the metal.
Indeed IR spectra clearly indicate the presence of bidentate chelating/bridging carboxylate
ligands [53].

The present synthetic route to metal carboxylate films increases gelation time
yielding reproducible films which are fracture free and optically transparent. The
incorporation of carboxylic acids also allows greater control over the chemical properties
of the final products. For example the addition of acetic acid produces films with difi‘ering

solubility properties than those synthesized from valeric acid [50]. Additionally, the use of

22

carboxylic acids is applicable to a wide number of transition metal alkoxide starting

materials [50].

1.4.3. Use of Sol-Gel Materials in Mass Sensors

The sol-gel synthetic process is used to produce aerogels, xerogels, monoliths,
ceramics and thin films. Films were initially used as protective or reflection resistant
coatings. More recently researchers have taken advantage of the porous, inert properties
of the films in applying them to the area of chenrical sensing. Most of the research
involving sol-gel films as chemical sensors has utilized the glasses as immobilization media
for optically active centers [54,55]. However, some work applying sol-gel films to mass
sensing has also been carried out.

Frye et a1. [56] first used a sol-gel matrix in conjunction with a mass sensor device
in 1988. The silicate sol-gel coated SAW devices were not used as chemical sensors, but
rather to compare N2 adsorption isotherms between bulk and thin film silicate glasses.
Pore size, surface area and percent porosity were different in the two forms. Thin films
were shown to be less porous than bulk samples produced using the same synthetic steps.
This result is not surprising considering the more rapid solvent removal from a thin film
sample. Nevertheless, these results confirmed that physical properties of polymer oxide
glasses cannot be inferred from one form to another. Frye et a1. [57] again used SAW
devices to study pore size distribution, surface area, diffusion coefficients and aging effects
on four component sol-gel films. Aging was used to produce larger polymeric species and

thus more porous films.

23

Also in 1988, Bein et a]. [58] embedded zeolite crystals into silicate glasses . The
films were densified at 570 K in order to collapse the pore structure of the silicate glass.
The destruction of the silicate pore structure produced a film in which only the channels of
the zeolite were available for diffusion. This zeolite pore structure imparted molecular
sieving properties to the silicate thin film. IR results indicated that pyridine molecules
could diffuse into the zeolite and interact with the structural hydroxyls while the larger
perfluorotributylamine molecule could not enter the pores. Bein er a]. [59] extended the
zeolite membrane work to mass sensing by coating the films onto SAW devices and
measuring their response to methanol, propanol and isooctane. The alcohols, having
smaller kinetic diameters, adsorbed at levels around 500-800 ng/cm2 while absorption of
isooctane was less than 5 ng/cmz. Bein's research established the ability of SAW devices
to monitor analyte interactions with sol-gel films. However, it must be noted that none of
the intrinsic properties of the sol-gel matrix were examined in this work. The sol-gel film

served only to immobilize zeolite crystals.

1.5. Characterization of Thin Film Sensors

The intermolecular interactions which contribute to sensor response occur at the
surface of the thin film. Therefore, characterisation of the surface is necessary to
understand the nature of these interactions. X-ray photoelectron spectroscopy (XPS) is a
surface analytical technique which gives quantitative and qualitative information about the
sample. Attenuated total reflectance (ATR) is a sampling technique that enhances the

ability of IR spectroscopy to examine analyte/thin film interactions.

24

1.5.1. X-ray Photoelectron Spectroscopy (XPS)

The firndamental process for X-ray photoelectron spectroscopy involves the
irradiation of the sample with photons of energy hv (e.g., 1253.6 or 1486.6 eV). These
photons produce photoelectrons with kinetic energy (BK). Since the energy of the
incident photon is also known it is possible to determine the binding energy of the

electron. The energy conservation equation for this process is:
mus; =EK+E{(k) (1.25)

where E1]- is the initial state energy, Bil-(k) is the total final state energy after ejection of
the photoelectron fi'om the kth level. The binding energy of an electron is defined as the
energy necessary to remove it to infinity with a kinetic energy of zero. In XPS
measurements EVB(k), the binding energy of an electron in the kth level referred to the

vacuum level is defined by:
V .
EB(k)=E{ +E;. (1.26)
Ifwe substitute equation 1.25 into 1.26 we get the photoelectric equation.
hv = EK = agar) (1.27)

Binding energies in XPS are expressed relative to a reference level. In the case of
insulating solids a well defined feature such as the C Is binding energy (284.6 eV) is often

used.

25

Non-equivalent atoms of the same element can give rise to core level peaks with
measurably different binding energies. These differences, called chemical shifts, arise fi'om
difl‘erences in formal charge, molecular environment, and lattice energy. These difl‘erent
conditions lead to dissimilar coulombic interactions between electrons and attractive
potentials from neighboring nuclei. This redistributes the valence electrons and affects the
potential of the core electrons. This potential change alters the binding energies of
electrons.

Signal intensities, Ii, are of interest in quantitative chemical analysis. Relative
concentrations of elements are calculated using expressions similar to equation 1.26:

m = Gbelbflbla (1.26)
[B] Gagaxanalb
where o is the photoelectron cross section, C is the fraction of photoelectric events which
take place without intrinsic plasmon excitation, )1 is the mean fi'ee path of the
photoelectron in the solid, 1] is the kinetic energy dependent spectrometer transmission
function, and I is the area of the photoemission peak.

XPS allows for quantitative analysis of the thin film surfaces. In our sol-gel
research, these ratios allow us to compare the quantity of metal centers to backbone and
ligand oxygen and the number of ligands per metal center. In conjunction with FTIR data

these techniques allow for detailed characterization of our sol-gel polymer films.

26

1.5.2. FTIR/Attenuated Total Reflectance (ATR)

An attenuated total reflectance (ATR) attachment utilizes total internal reflectance
to obtain multiple interactions with the sample, leading to a higher signal-to-noise ratio.
Total internal reflection can occur when light traveling through an optically dense medium
reaches a boundary with an optically rarer medium. If the angle of incidence (Oi) is
greater then the critical angle (9‘), all of the incident radiation will be internally reflected.

(9c is determined from equation 1.27:

sin e, = “%l (1.27)

where nland 112 are the refractive indices of the optically dense and optically rare medium,
respectively.

The term total internal reflectance suggests that there is zero energy flux into the
rarer medium, however a finite decaying electric field extends some distance across the
boundary. This field is called the evanescent field or wave and its penetration depth is

given as follows:

dp = l (1.23)

21:01? sin2 01 41:2, )1/2

 

The use of FTIR/ATR to examine the structure of thin film polymeric materials is
well established [60,61]. ATR has been used to depth profile and characterize polymer

blends [62] and study polymer formation reactions [63]. It has also been used to

27

determine reaction rates in a wide variety of biological and chemical systems [64]. In
addition, polymer coated internal reflectance elements have been used to extract solutes
from aqueous solution for analysis by FTIR [65,66]. ATR has also been utilized to study
the molecular level interactions of polymers and surfactants [67]. Rates of diffusion of
aqueous samples into polymer films has also been measured [68,69]. All of these diflirsion
systems involved difi‘usion into or from aqueous solution; none of the research focused on
determining gas phase difi‘usion rates. Fieldson et al. [69] suggested the technique could
be used to measure diffusion rates of gas or vapors or to examine polymer changes during
difl‘usion. However, they did not establish the ability to investigate such effects.

ATR cells have been used to expose vapor phase analytes to a sol-gel film.
Dulebohn et al. [70] determined bonding interactions between CO and a rhodium doped
titanium sol-gel film. FTIR/ATR spectra confirmed the reversible bonding of CO to the
rhodium center. Through application of a pulse injection system, analyte residence time
within the ATR cell should also ofi‘er information into analyte difl‘usion and analyte/film
interactions. Additionally, band shifts could offer insight into the type of interactions
causing retention of the analyte. Finally, viscoelastic changes or morphology changes
induced by solvent vapor could affect in the infrared spectra of the polymers. Figure 1.2

shows the ATR cell used for our pulsed experiments.

28

Gas Out Gas In

Temperature
Control

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1.2. ATR cell used for analyte/film interaction studies.

1.5.3. Sensor Test Apparatus

There are two common experimental methods to expose the acoustic sensor to the
analyte of interest. Analyte can be introduced to the sensor cell as either a constant stream
or a pulse. Constant stream methods rely on the establishment of equilibrium before
sensor readings are recorded. Pulse measurements normally give a single maximum
frequency shift which is recorded as the sensor response. The choice of pulse or constant
flow exposure depends on the intended analytical application of the sensing device.

The use of pulse injection dictates specific design considerations for a sensor cell
and gas handling system. As with gas chromatography, the analyte must be vaporized
upon injection into the gas stream and be carried to the sensor test cell as a plug of vapor.
To insure this, a heated injector block, heated transfer line and minimum cell volume are

necessary. A schematic of the gas system and sensor cell appears in Figure 1.3.

29

Gas Flow —9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Zeolite
Injector
N2 MassFlow
Controller

 

 

 

 

 

 

 

 

U L I Sensor Cell

 

Figure 1.3. Gas handling and sensor test cell schematic.

1.6. Focus of Research

In the study and development of chemical sensor coatings the ability to control the
chemical and physical properties of the films is of great utility. Tailoring film structure
allows us to investigate specific modifications and relate them to changes in the
film/analyte interactions.

Polysiloxanes are a classic organicfrnorganic polymer which is available with a
large variety of chemical and physical properties. The Si-O-Si backbone can be

derivatized with a large number of functional groups to enhance specific solubility

30

interactions. Additionally, data concerning chromatographic, solvatochromic, and
thermodynamic characteristitcs of film/analyte interaction are available for comparison to
sensor results. Such a large quantity of data is useful in helping to establish the
importance of solubility interactions and viscoelastic effects on mass sensor response.

More recently our interest has focused on the use of transtion metal carboxylate
thin films as coatings for chemical sensors. As noted previously, very little work utilizing
sol-gel derived thin films as mass sensor coatings has been reported. Considering the
ability to easily alter the ligands and thus the chemical properties of the films, polymer
oxide glasses may indeed offer the versatility necessary to produce selective coatings.
Additionally, changes in processing conditions, such as aging, can also lead to changes in
the physical properties of these films.

Sol-gel films offer advantages as matrices for the incorporation of guest molecules.
The optical clarity of the metal carboxylate glasses makes them ideal materials for optical
sensing applications. While work on silicates as optical matrix is proceeding at a rapid
rate, very few researchers have taken advatage of the chemical control possible with sol-
gel transition metal film synthesis.

Nearly all of the sol-gel research published to date has focused on the use of guest
molecules to supply the necessary selectivity to the system. Little if any work has been
done to deterrrrine how the matrix affects analyte interactions. Sol-gel synthesis offers the
ability to control the synthesis from molecular precursor to final product, but researchers

have not taken advantage of this control.

31

References

1. Janata, J .; Bezegh, A. Anal. Chem. 1988, 60, 62R.

2. Frye, G. C.; Martin, S. J. Appl. Spec. Rev. 1991, 26(1&2), 73.

3. Curie, J .; Curie, P. Bull. Soc. Min. Paris 1880, 3, 90.

4. Bastiaans, G. J. in Chemical Sensors“, Edmunds, G. Ed.; Chapman Hall: New
York, NY, 1988, 295.

5. Lu, C. in Applications of Piezoelectric Quartz Crystal Microbalances; Lu, C.;
Czandema, A. W., Eds; Elsevier: Amsterdam, 1984.

6. Sauerbrey, G. Z. Z Phys. 1959, 115, 206.

7. Mueller, R M.; White, W. Rev. Sci. Instum. 1968, 39, 291.

8. Mueller, R M.; White, W. Rev. Sci. Instum. 1969, 40, 1646.

9. Stockbridge, C. D. in Vacuum Microbalance Techniques, vol. 5; Behrndt, K. H.
Ed.; Plenum Press: New York, NY, 1966, 135.

10. Behrndt, K. H. J. Vac. Sci. Technol. 1971, 8, 622.

11. Denison, D. R. J. Vac. Sci. Technol. 1973, 10, 126.

12. Lu, G; Lewis, 0. J. Appl. Phys. 1972, 43, 4385.

13. Kanazawa, K. K.; Gordon, J. G. Anal. Chim. Acta 1985, 175, 99.

14. Kanazawa, K. K.; Gordon, J. G. Anal. Chem. 1985, 57, 1770.

15. Lord Rayleigh London Math. Soc. Proc. 1887, 17, 14.

16. White, R. M.; Voltmer, F. W. Appl. Phys. Lett. 1965, 7, 314.

17. Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51(9), 1458.

18. Wohltjen, H. Sens. &Act. 1984, 5, 307.

19. King, W. H., Jr. Anal. Chem. 1964, 36, 1735.

20. Grate, J. W.; Snow; A.; Ballantine; D. 8., Jr.; Wohltjen, H.; Abraham, M. H.;
McGill, R. A.; Sasson, P. Anal. Chem. 1988, 60(9), 869.

21. Grate, J. W.; Abraham, M. H. Sens. &Act. B. 1991, 3, 85.

22. Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chem. Soc.
Perkin. Trans. 2 1990, 1451.

23. Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1990,518,329.

24. Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J. J.; Taylor, P. J. J. Chem.
Soc. Perkin. Trans. 2 1990, 521.

25. Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Duce, P. P.; Morris, J. J.; Taylor, P.
J. J. Chem. Soc. Perkin. Trans. 2 1989, 699.

26. Abraham, M. H.; Duce, P. P.; Grellier, P. L.; Prior, D. V.; Morris, J. J.; Taylor, P.
J. Tetrahedron Lett. 1989, 29(13), 1587.

27. Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J. J .; Taylor, P. J .; Laurence,
C.; Berthelot, M. Tetrahedron Lett. 1989, 30(19), 2571.

28. Abraham, M. H.; Grellier, P. L.; McGill, R. A.; Doherty, R. M.; Karnlet, M. J.;
Hall, T. N.; Taft, R. W.; Carr, P. W.; Koros, W. J. Polymer 1987, 28, 1363.

29. Abraham, M. H.; Grellier, P. L.; McGill, R. A. J. Chem. Soc. Perkin. Trans. 2

1988, 523.

30.

31.

32.
33.
34.
35.
36.
37.

38.

39.
40.

41.
42.
43.

45.
46.

47.

48.

49.
50.

51.

52.
53.
54.

55.
56.

32

Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Duce, P. P.; Morris, J. J.; Taylor, P.
J J. Chem. Soc. Perkin. Trans. 2 1989, 699.

Kamlet, M. J.; Abboud, J-L, M., Abraham, M. H.; Taft, R. W. J. Org. Chem.
1983, 48, 2877.

Patte, F.; Etcheto, M.; Laffort, P. Anal. Chem. 1982, 54, 2239.

Patrash, S. J.; Zellers, E. T. Anal. Chem. 1993, 65(15), 2055.

Ballantine, D. 8., Jr.; Wohltjen, H. IEEE Ultrason. Sympos. 1988, 559.

Bartley, D. L.; Dominquez, D. D. Anal. Chem. 1990, 62(15), 3069.

Frye, G. C.; Martin, S. J.; Ricco, A. J. Sens. Mater. 1989, 1-6, 335.

Ballantine, D. 8.; Jr.; Rose, 8. L.; Grate, J. W.; Wohltjen, H. Anal. Chem. 1986,
58(14), 3058.

Grate, J. W.; Klusty, M.; McGill, R. A.; Abraham, M. H.; Whiting, G.; Andonian-
Haftvan, J. Anal. Chem. 1992, 64(6), 610.

Ballantine, D. S., Jr. Anal. Chem. 1992, 64(24), 3069.

Brinker, C. J.; Scherer, G. W. Sol-Gel Science, The Physics and Chemistry of
Sol-Gel Processing Academic Press, Inc.: San Diego, CA, 1990.

Livage, J.; Sanchez, C. NewJ. Chem. 1990, 14, 513.

Livage, J.; Henry, M.; Sanchez, C. Prog. Solid State Chem. 1988, 18, 259.
Yamane, M.; A50, 8.; Sakaino, T. J. Mater. Sci. 1978, 13, 865.

Yoldas, B. E. J. Mater. Sci. 1975, 10, 1856.

Yoldas, B. E. Ceram. Bull. 1975, 54(3), 289.

Hench, L. L. in Science of Ceramic Chemical Processing, Hench, L. L.; Ulrich, D.
R., Eds; John Wiley & Sons: New York, NY, 1986, 52.

Hench, L. L.; Orcel, G.; Nogues, L. in Better Ceramics Through Chemistry II;
Brinker, C. J.; Clark, D. E.; Ulrich, D. R. Eds; Mater. Res. Soc. Symp. Proc., 73:
Pittsburgh, PA, 1986, 35.

Langkammerer, C. M. US. Patent No. 2 621 193 1952.

Haslam, J. H. US. Patent No. 2 621 195 1952.

Gagliardi, C. D.; Berglund, K. A. in Processing Science of Advance Ceramics;
Aksay, A. 1.; McVay, G. L.; Ulrich, D. R., Eds; Mater. Res. Soc. Symp. Proc.,
155: Pittsburgh, PA, 1989, 127.

Gagliardi, C. D.; Dunuwila, D.; Berglund, K. A. in Better Ceramics Through
Chemistry IV; Zelinsky, B. J. J.; Brinker, C. J.; Clark, D. E.; Ulrich, D. R., Eds;
Mater. Res. Soc. Symp. Proc., 180: Pittsburgh, PA, 1990, 801.

Sanchez, C.; Livage, J .; Henry, M.; Babonneau, F. J. Non-Cryst. Solids 1988,
100, 65.

Doeufi‘, 8.; Henry, M.; Sanchez, C.; Livage, J. J. Non-Cryst. Solids 1987, 89,
206.

Lev, O. Analusis 1992, 20, 543.

Levy, D. J. Non-Cryst. Solids 1992, 147&148, 739.

Frye, G. C.; Ricco, A. J.; Martin, S. J.; Brinker, C. J. in Better Ceramics Through
Chemistry 111; Brinker, C. J.; Clark, D. 13.; Ulrich, D. R., Eds; Mater. Res. Soc.
Symp. Proc., 121: Pittsburgh, PA, 1988, 349.

57.

58.

59.

60.
61.

62.
63.

65.

66.
67.
68.
69.
70.

33

Frye, G. C.; Martin, S. J.; Ricco, A. J.; Brinker, C. J. in Chemical Sensors and
Microinstmmentation; ACS Symp. Ser. 403, 1989, 208.

Bein, T.; Brown, K.; Enzel, P.; Brinker, C. J. in Better Ceramics Through
Chemistry 111; Brinker, C. J.; Clark, D. E.; Ulrich, D. R., Eds; Mater. Res. Soc.
Symp. Proc., 121: Pittsburgh, PA, 1988, 761.

Bein, T.; Brown, K.; Frye, G. C.; Brinker, C. J. .I. Am. Chem. Soc. 1989, 111,
7640.

Holland-Moritz, K.; Siesler, H. W. Appl. Spectros. Rev. 1976, 11(1), 1.

Smith, A. L. Applied Infiared Spectroscopy John Wiley & Sons: New York, NY,
1979.

Popli, R.; Dwivedli, A. M. J. Appl. Poly. Sci. 1989, 37, 2469.

Kuhn, K. J.; Hahn, B.; Perec, V. Urban, M. W. Appl. Spectrosc. 1987, 41(5),
843.

Miller, M. P. Appl. Spectrosc. Rev. 1987, 23(3&4), 329.

Heinrich, P.; Wyzgol, R.; Schrader, B.; Hatzilazaru, A.; Lubbers, D. W. Appl.
Spectrosc. 1990, 44(10), 1641.

Meuse, C. W.; Tomellini, S. A. Anal. Lett. 1989, 22(9), 2065.

Evanson, K. W.; Urban, M. W.; J. Appl. Polym. Sci. 1991, 42, 2287.

Xu, J. R.; Balik, C. M. Appl. Spectrosc. 1988, 42(8), 1543.

Fieldson, G. T.; Barbari, T. A. Polymer 1993, 34(6), 1146.

Dulebohn, J. 1.; Haefner, S. C.; Berglund, K. A.; Dunbar, K. R. Chem. Mater.
1992, 4, 506.

34

Chapter 2

Structure and Mass Sensing Properties of Zirconium Acetate Thin Films

2.1. Introduction

The sol-gel process is an attractive approach for the preparation of sensing
materials because porous, optically transparent, thin films of high purity can be synthesized
at low temperatures. Polymer oxide glasses have been used to construct chemical sensors
based on optical [1-5], mass [6-10], and conductimetric [11,12] transduction methods. In
most cases, porous glass-based sensors have utilized recognition centers such as
porphyrins [3], cyclodextrins [13], cryptates [4] and organometallic compounds [2] to
prepare selective composite devices. In addition, Bein et al. [9,10] have reported that
zeolite/glass composite films exhibit molecular exclusion properties. Much of the previous
work with porous glass sensors has considered the gel to be simply a immobilization
matrix for a more selective interaction site. Little effort has been devoted to using the
versatility of sol-gel chemistry to prepare glass films which can enhance the selectivity,
sensitivity, or stability of the recognition center.

We have recently reported that variation of the carboxylic acid used to prepare
transition metal carboxylate thin films can significantly affect the chemical properties of
the glasses [14]. For example, film hydrophobicity can be controlled by selection of the

hydrocarbon chain length attached to the carboxylate ligand. This suggests that control of

34

35

sol-gel synthesis conditions can lead to glass films with tunable properties that may
complement the selectivity of molecular recognition centers. In this work, Fourier
transform infrared spectrosc0py (FTIR) and X-ray photoelectron spectroscopy (XPS)
have been used to determine the structure of a zirconium acetate thin film prepared using
Zirconium(IV) n-propoxide and acetic acid. The information obtained from spectroscopic
measurements has been compared to the mass sensor response of the film to simple

alcohols and alkanes to evaluate the effect of film structure on sensor properties.

2.2. Experimental

Materials. Zirconium(IV) n-propoxide (70% in n-propanol), acetic acid, and n-
propanol were obtained from Aldrich Chemical Company and used without further
purification. The set of probe analytes used in this work is given in Table 2.1. Analytes
were chosen to study analyte/film interactions for n-alkanes and alcohols with difl‘erent
steric limitations. The interaction of water with the film was examined to determine film
hydrophilicity. Analytes were obtained from either EM Science or Aldrich Chemical
companies and used as received.

Film Preparation. The method used to synthesize the film was based on
techniques developed in these laboratories for the preparation of transition metal
carboxylate thin films [14,16]. Each solution and film was prepared in triplicate.
Reactions were carried out at room temperature in capped vials. Molar ratios of acetic
acidzwaterzalkoxide of 3:13.921 were used. The synthesis involved mixing acetic acid with

the zirconium n—propoxide, diluting the acetic acid/alkoxide solution with n-propanol (12

36

Table 2.1. Probe analytes.

 

 

Boiling Point Molar Volume
Analyte Symbol (°C) (cm3/mole)3
methanol 0 65 40.4
hexane El 69 130.5
ethanol 0 78 58.4
iso-propanol $ 82 76.8
tert-butanol 0 82 94
n-propanol O 97 74.7
heptane El 98 146.5
sec-butanol $ 99 92
water A 100 18
iso-butanol O 108 92.8
n—butanol O 1 17 91 .5
octane Cl 126 162.3

 

a - Reference [15].

37

mL added to ~ 2 mL of solution), and finally adding water. A vortex mixer was used to
vigorously stir solutions following the addition of each reactant. The reaction between
acetic acid and zirconium n-propoxide was exothermic and the resulting zirconium
solutions were clear and colorless. Films were immediately spray coated onto the desired
substrate (quartz, KBr, or AMT IR) using the n-propanol diluted solutions. The films
were air-dried overnight and stored in a desiccator. All films prepared in this study were
transparent, colorless, and exhibited complete and uniform coverage of all substrates.

Fourier Transform Infrared Spectroscopy (FTIR). FTIR spectra of films spray
coated onto KBr windows were obtained using a Mattson Instruments Galaxy 3020
Fourier transform infi'ared spectrometer. The mid-IR region (4000-400 cm’l) was
examined with a resolution of 2 cm'l. Data acquisition and processing were performed
using an Enhanced First software package.

FTIR/ATR experiments were performed using a Mattson Instruments Galaxy 5020
FTIR spectrometer equipped with a wideband mercury cadmium telluride (MCT)
detector. A commercial attenuated total reflectance (ATR) accessory (Specac, Inc.)
designed for HPLC detection has been modified to perform analyte/film studies. The
attachment consists of a 500 11L cell in contact with an AMTIR internal reflectance
element (IRE), contained in a temperature controlled water bath held at 30°C. The
fittings of the cell have been changed to accommodate 1/4" o.d. (1/8” i.d.) stainless steel
tubing. The inlet of the ATR cell is connected to a gas chromatograph injector by a 4"
long, 1/8" i.d. stainless steel transfer line. The injector and transfer line were heated to

250°C. Analytes were injected as 0.5 11L pulses into a 20 cm3/min flow of dry nitrogen

38

(99.95%, AGA Gas Co.). Analyte/film interaction studies were performed using a
polymer film coated onto the AMTIR IRE. The thinnest coating that could be detected by
the FTIR was used. FTIR/ATR spectra were collected with 4 cm'1 resolution.

A software package supplied by Mattson (FIRST Macros) was modified to collect
time dependent FTIR/ATR data. The program uses a spectrum of the polymer coated
ATR element as the background for firrther data collection. Eighty spectra (each
spectrum is the sum of 10 scans taken at one second intervals) were obtained following
the injection of each analyte. Each spectrum is integrated over a wavenumber interval
characteristic of the analyte. Plots of integrated peak area versus elapsed time are used to
determine the kinetics of analyte/polymer interaction.

X-ray Photoelectron Spectroscopy (XPS). Films cast on quartz slides (used as
received) were analyzed with a Perkin-Elmer surface science instrument equipped with a
model 10-3 60 precision energy analyzer and an omnifocus small spot lens. All spectra
were collected using a non-monochromatized Mg anode (1253.6 eV) operated at a power
of300 W (15 kV and 20 mA) with an analyzer pass energy of50 eV. The C Is, 0 1s and
Zr 3d were scanned for each film. Binding energies were referenced to ligand alkyl and
adventitious carbon (C ls = 284.6 eV). Empirically derived sensitivity factors [17] were
used for quantitative XPS calculations involving the C15 and Ols. For the Zr 3d,
sensitivity factors were derived from the analysis of Zr02 (Aldrich Chemical Company,
99.99+%). XPS peaks were fitted with 20% Lorentzian-Gaussian mix Voigt functions

using a non-linear least squares curve fitting program [18]. Each reported result is based

39

on the analysis of at least three films. It must be noted that the use of XPS to evaluate
film composition assumes that the films are homogeneous.

Sensor Test Apparatus. A schematic diagram of the sensor test apparatus is
shown in Figure 2.1. The approximate volume of the cell is 8 mL. Analytes were
delivered to the cell as a pulse using a gas chromatograph injector operated at 250°C.
Typical operating conditions are 20 cm3/min flow of dry nitrogen (99.95%, AGA Gas Co.)
and test cell temperature of 30°C. The temperature of the test cell was controlled by
resistance heating the transfer line to 150°C. The transfer line was constructed from a
3.5" long, 1/16" i.d. stainless steel tube. The injection volume used for the set of analytes
varied from 0.1 1.1L to 3.0 11L in order to maintain frequency shifts of less than 150 Hz. At
least three injections of each analyte were made to ensure reproducibility. Analyte
injection order was varied to minimize the effect of analyte order on sensor response. The
peak shape for each analyte consists of a sharp leading edge that appears 15 seconds after
analyte injection, a narrow peak maximum (approximately 2 sec), and a tailing return to
baseline. The sharp rising edge indicates that each analyte reaches the sensor cell as a
pulse. The time required for the frequency to return to baseline was less than 2 minutes
for the alkanes and approximately 6 minutes for the alcohols.

Quartz Crystal Microbalance (QCM). Bulk acoustic wave measurements were
performed using 6 MHz quartz crystal microbalances fi'om Maxtek Inc. (Torrance, Calif).
A model TM-100 Thickness Monitor (Maxtek Inc.) was used to determine the mass of
the polymer coating and the response of the coated device to the probe analytes. The

output voltage of the TM-100 thickness monitor (3.06 mV = 1 Hz) was recorded on a

4o

 

 

fi 1 fi 1

 

Gas Handling .
System Injector
1 1 L o

 

 

 

 

 

 

 

 

 

 

 

 

 

Teflon Aluminum

Gasket\/< >/ / Blocks
C 0 >

 

 

 

 

 

 

 

 

 

1

Data
Acquisition

 

 

J

Figure 2.1. Schematic diagram of sensor test apparatus.

41

Hewlett Packard 3394A integrator. The sensitivity of the QCM is 53 Hz/ug.

Quartz oscillators were soaked in iso-propanol, rinsed with methanol, and dried in
air prior to use. Cleaned devices were placed in the oscillator circuit and the thickness
monitor set to zero. The oscillators were spray coated with the polymer solution and
allowed to dry. Films of approximately 0.4 um thickness (~ 61 ug,) were used in this
study. Mass sensor responses are reported as the maximum frequency shift in hertz per
micromole of analyte injected (Hz/umole) to normalize the response. Responses are
plotted against analyte boiling point to facilitate the study of condensation effects on
sensor response. A smooth increase in response versus analyte boiling point is expected
for simple condensation, while selective analyte/film interactions result in deviations fiom

the smooth trend.

2.3. Results

FTIR Spectra. The FTIR spectra of acetic acid, zirconium(IV) n-propoxide (in n-
propanol), and the zirconium acetate film are shown in Figure 2.2. The assignment of
spectral features for acetic acid (Figure 2.2a) is straightforward [19,20]. Weak methyl C-
H stretching absorbances are observed near 2900 cm'1 superimposed on an O-H
absorbance centered at 3000 cm'l. The strong absorbance observed at 1711 cm'1 is due
to a dirneric carboxylic C=O asymmetric stretch. Asymmetric and symmetric CH3
deformation modes are observed at 1414 and 1280 cm'l, respectively. Bending and
stretching vibrations of the C-O-H groups are observed between 1440 and 900 cm'l.

Skeletal vibrations are evident at 1382 and 625 cm'l.

42

The FTIR spectrum obtained for zirconium(IV) n-propoxide in n-propanol is
shown in Figure 2.2b. Spectral features of the alkoxide are similar to those of the parent
alcohol with many absorbances shifted to lower fi'equency and additional features present
due to (C-O)Zr and (Zr-O)C stretches [21]. The major absorbances are the C-H
stretching vibrations at 2960, 2934, and 2874 cm'1 and a feature at 1133 cm'1 which we
attribute to a combination of (C-O)Zr and skeletal stretches. The peak observed at 1005
cm'1 is due to (C-O)Zr stretching and features observed at 597, 549, and 505 cm'1 are
attributed to alkoxide (Zr-O)C stretches. These features are similar to those reported by
Barraclough [21] for zirconium(IV) iso-propoxide: a C-O absorbance at 1011 cm'1 and
features between 590 and 520 cm'1 due to alkoxide Zr-O stretches. Interference from n-
propanol in the zirconium(IV) n-propoxide spectrum does not allow the assignment of
features to bridging OR groups attached to the metal center. The peak observed at 3340
cm'1 is the O-H stretch of the n-propanol solvent.

Figure 2.2c shows the FTIR spectrum measured for the zirconium acetate film.
Three weak features characteristic of methyl stretches of the acetate ligand are observed at
3017, 2933, and 2980 cm]. Less intense bands observed between 1400 and 700 cm‘1 are
due to C-C stretching and C-H bending vibrations of the acetate ligand. A broad band due
to hydroxyl groups is observed near 3400 cm]. The low fiequency region of the
spectrum shows two major features at 645 and 466 cm'1 which we attribute to Zr-O
stretches. Atik and Aegerter [22] have attributed bands at 666 and 363 cm"1 to Zr-O
stretches in a zirconium acetate film.

Figure 2.3 shows the carboxylate stretch region of the FTIR spectra measured for

the zirconium acetate film and sodium acetate. The asymmetric and symmetric COO

43

 

 

 

 

 

 

 

 

 

 

 

S
‘33?
”H
8 m
V <1- m
on ‘9 to
m V
no
(c) a l
m
m
8 2
3“:
(D
N
cu
o
c
(U
.o
S.
o
u) (b)
.o
<
H
r-I
l\
r-I
O
m
8 fl
8 2.5
H m
,2, s
(a)
l l I l l
3600 2500 1600 900 400

Frequency (cm'l)

Figure 2.2. FTIR spectra of (a) acetic acid, (b) zirconium (IV) n-propoxide in n-
propanol, and (c) zirconium acetate film.

44

stretch fi'equencies for sodium acetate (1580 and 1418 cm'l, Av = 162 cm’l) are similar in
relative intensity and position to those reported by Spinner [23]. In the film spectrum, the
symmetric COO stretch is as broad as the asymmetric stretch; however, only one well-
resolved maximum is observed at 1449 cm’l. The separation between the asymmetric and
symmetric COO stretches (Av = 113 cm‘l) is reported with respect to this maximum
which is on the high energy side of the peak.

XPS Spectra. The XPS Zr 3d,/2 binding energy measured for the zirconium film
(182.2 eV) is typical of values measured for tetravalent zirconium [24]. The XPS C Is
spectra measured for the film consists of peaks at 284.6 eV (reference peak) and 288.9 eV
due to alkyl and carboxylate carbon, respectively. No features due to alkoxide or ether
carbon (~ 286 eV) are observed in the spectra. The XPS 0 1s spectra measured for the
film consists of peaks at 530.0 eV and 531.4 eV that may be attributed to backbone and
ligand (acetate, hydroxyl, or water) oxygen, respectively. The carboxyl carbon/zirconium,
backbone oxygen/zirconium, and ligand oxygen/zirconium atomic ratios calculated from C
Is, 0 Is, and Zr 3d XPS peak intensities were 1.1, 1.5, and 2.5, respectively.

QCM Response. Uncoated Crystal. Figure 2.4 and Table 2.2 show the variation
in sensor response (I-Iz/umole) as a function of analyte boiling point for the uncoated
quartz oscillator. The response of the uncoated device increases with increasing analyte
boiling point.

Zirconium Acetate Film. The variation in sensor response as a firnction of analyte
boiling point for the zirconium acetate film is shown in Figure 2.5. In contrast to the
uncoated crystal results, the response for the zirconium acetate film does not increase

simply with analyte boiling point. The fi'equency shifts observed for the unbranched

45

 

 

 

 

N Os
6 s
3 (b)
E ..
a 3
oo
E
(a)
l l l l
1764 1600 1444 1296
Frequency (cm")

Figure 2.3. FTIR spectrum of carboxylate region obtained for (a) sodium acetate and
(b) zirconium acetate film.

46

 

 

 

 

6
o
5-
o 4‘
E" e '3
E 3-
‘6“ o
8
a e
m 2_ .
6‘2 0
. :1
A
1‘ o
q 0 D e
0 . . . , . , .
60 80 100 120 140
Analyte Boiling Point (°C)

Figure 2.4. Variation in QCM response (Hz/umole) versus analyte boiling point for
uncoated quartz crystal.

47

Table 2.2 Analyte responses on coated and uncoated QCM.

 

 

Response Response Ratio
(Hz/umole) (coated/uncoated)
Analyte Blank Acetate Normalized

methanol 0.49 23 46

hexane 0.33 ' 0.17 0.52
ethanol 0.89 25 28
iso—propanol 0.42 12 29
tert-butanol 2. l 3 .8 l .8
n-propanol 1 .8 24 13

heptane 1.6 1.1 0.69
sec-butanol 3.8 18 4.7
water 1 .3 26 20
iso-butanol 2.7 19 7.0
n-butanol 5.5 29 5.3

octane 3.6 2.3 0.64

 

 

48

primary alcohols (open circles) increase by less than 30% from methanol (b.p. 65°C) to n-

butanol (b.p. 117°C). It should be noted that for the uncoated crystal the response for n-
butanol is over a factor of ten greater than the methanol response. The response measured
for iso-butanol (open circle, b.p. 108°C), a branched primary alcohol, is lower than that
observed for the other primary alcohols. The primary alcohol responses are also
significantly higher than the values observed for alkanes (open squares) with comparable
boiling points. For the secondary alcohols, iso-propanol and sec-butane] (crossed circles),
the measured responses are significantly lower than the values obtained for unbranched
primary alcohols of comparable boiling point. The response for tert-butanol (solid circle)
is lower than all the other alcohols. Alkane responses increase from 0.2 (hexane) to 2.3
(octane) Hz/umole. In addition, the alkane fiequency shifts observed for the zirconium
acetate film are approximately 30% lower than the values recorded for the uncoated
crystal.

The water response of the acetate film (open triangle) is similar to that observed
for the primary alcohols and significantly higher than alkanes of comparable boiling point.
This response is considerably enhanced compared to the uncoated crystal.

FTIR/ATR Studies. For the uncoated IRE, FTIR/ATR kinetics plots observed
for the analytes exhibit a steady increase in purge time and integrated peak area with
increasing analyte boiling point. This is the trend expected for condensation of analytes.

Figure 2.6 shows the variation in integrated area of the hydroxyl region for
methanol (—) and ethanol (--) on a zirconium acetate film versus purge time. The area of

the absorbance peak measured for methanol is 5 times the area determined for ethanol. In

49

 

 

 

 

30
O
25- O A
. O O
A 20-
% 9 0
g .,
a 15-
0
a r 9
o
o.
8 10-
M
51
O
‘ U
U
o-——EL , . T . , .
60 80 100 120 140
Analyte Boiling Point (°C)

Figure 2.5. Variation in QCM response (HzJumole) versus analyte boiling point for
zirconium acetate coated quartz crystal.

50

addition, the purge time of methanol is twice that observed for ethanol. For alcohols with
boiling points higher than ethanol (> 78°C), the peak areas and purge times increase with
increasing alcohol boiling point. This may be attributed to condensation effects noted for
the uncoated IRE. Normalized peak areas of the hydroxyl stretch of n-propanol and the
CH stretch of heptane observed on the zirconium acetate film are shown as a function of
time in Figure 2.7. The purge time observed for n-propanol (b.p. 97°C) is twice that
recorded for heptane (b.p. 98°C). The purge time obtained for heptane is representative

of the values observed for hexane and octane.

2.4. Discussion

Composition of Zirconium Acetate Films. The variety of reactions that occur in
the polymer solutions could result in the coordination of several different ligands
(e.g.,alkoxides, carboxylates, esters, hydroxides, alcohols and water) to the metal centers
in the cast films. The predominance of carboxylate ligand features observed in the FTIR
spectra of the film (Figure 2.2c) suggests that most of the ligands are acetate groups. This
is consistent with qualitative XPS results that show C Is peaks due only to alkyl and
carboxylate carbon. In addition, no features attributable to ester or alkoxide ligands are
observed in the FTIR spectrum of the film. It should be noted that previous work on
zirconium acetate gels [22] showed that acetate and alkoxide ligands were bound to the
polymer backbone. It is possible that the excess acid used in our preparation increases the
reactivity of the alkoxide ligands during the hydrolysis and condensation steps of the

reaction. The enhanced reactivity would lead to a decrease in the number of alkoxide

51

 

f

1

ii

Integrated Peak Area

 

 

 

 

\
o I I I I k U I I

l
0 50 100 150 200
Time (see)

Figure 2.6. Variation in hydroxyl stretch intensity for methanol (—) and ethanol (--) as
a function of time following analyte injection.

52

 

 

 

 

I I ' I

1.0 b 3

0.8 r ‘. -
‘ié 0.6 e .' -
a) .
a. ‘.
E 0.4 - .‘ 3
o ; .
z . ‘1

0.2 L .' -

0.0 a L 1 a I

0 50 100
Time (sec.)

Figure 27. Variation in relative intensity of the hydroxyl stretch for n-propanol (——)
and OH stretch for heptane (---) as a function of time following analyte
injection.

53

ligands present in the cast films. Doeuff et al. [25] reported that gelatinous precipitates
prepared from excess acetic acid and titanium butoxide (Ac/T i = 10) contained no
alkoxide ligands.

The atomic ratios derived from XPS intensity ratios provide a quantitative
assessment of the number and type of ligands present in the film. The extent of
carboxylate incorporation observed for zirconium acetate film (1.1 acetates/zirconium) is
similar to results obtained fi'om TGA analyses of zirconium-based acetate gels [26]. Since
each carboxylate carbon is associated with two ligand oxygens, a film containing only
carboxylate ligands is expected to have a ligand oxygen/carboxyl carbon ratio close to
two. The ligand oxygen/zirconium atomic ratio calculated for the zirconium acetate film
(2.5) suggests that there are approximately 0.3 oxygen ligands per zirconium center that
cannot be attributed to acetate. FTIR and XPS analyses indicate that the only other
possible oxygen containing ligands present in the film are hydroxyl ligands or molecular
water. Water has a characteristic absorbance near 1640 cm'1 [19]. Unfortunately, the
strong carboxylate absorbances in this region make it impossible to use this feature to
distinguish between hydroxyl ligands and molecular water. Figure 2.8 compares the
spectra of the zirconium acetate film before (a) and after (b) heating at low temperature
(< 100°C). It is obvious that a significant portion of the hydroxyl features are lost. Since
it seems unlikely that hydroxyl ligands would be lost at such low temperatures, we
attribute the excess ligand oxygen primarily to water absorbed in the film. This is
consistent with previous work which indicated that zirconium acetate films were

hydrophilic and water soluble [l6].

54

 

 

 

 

 

Absorbance
E

 

 

 

 

 

(a)

 

 

 

 

l l l l
4000 3000 2000 1000

Frequency (cm'l)

 

Figure 2.8. FTIR spectra of zirconium acetate film (a) before and (b) after heating.

55

Zirconium/Acetate Coordination. Previous studies of metal carboxylate
coordination have compared the splitting between asymmetric and symmetric COO
stretches with the splitting observed in the ionic form of the ligand to ascertain carboxylate
coordination modes [27]. Splitting much larger than that observed for the ionic species
tend to indicate monodentate ligands while smaller separations are considered
characteristic of bidentate ligands. It has also been suggested that splitting significantly
smaller than the ionic value are indicative of bidentate chelating coordination while
separations closer to ionic values are typical of bidentate bridging coordination [28]. This
reasoning has been used to detemrine acetate ligand coordination in the evolution of
titanium xerogels [25] and in titanium, zirconiurrr, and hafirium acetate solutions before
and after the addition of water [16].

The asymmetric-symmetric splitting observed for the acetate film (Av = 113 cm'l)
is lower than the separation measured for sodium acetate (Av = 162 cm'l). This indicates
that a majority of the acetate ligands are bidentate. Since the asymmetric and symmetric
peaks are fairly broad, both bidentate bridging and chelating ligands are probably present.
Atik and Aegerter [22] report values of 1578 and 1452 cm‘1 (Av = 126 cm'l) for the
asymmetric and symmetric COO stretches of a zirconium acetate film, a separation also
indicative of bidentate ligands.

Structure of Zirconium Acetate Films. Using the XPS estimates of backbone
oxygen and carboxylate ligand concentrations in the film, a general picture of the film
structure can be proposed. Our results suggest that the zirconium acetate film consists of

highly crosslinked networks of zirconium and oxygen, with most zirconium centers

56

covalently bonded to three backbone oxygen atoms (assuming a backbone oxygen
coordination number of 2). In addition, if we assign a -2 charge to backbone oxygen and
a -1 charge to the carboxylate group, the atomic ratios calculated from XPS lead to a total
negative charge of -4.2. With experimental error, this is the value expected for a neutral
polymer involving a Zr4+ metal center. This further validates the use of XPS to determine
polymer composition. It must be noted that film stoichiometry determined by XPS cannot
be used to determine the zirconium coordination number. It is likely that the acetate films
have significant interchain interactions that lead to a coordination number greater than 5.

Effect of Coating on Sensor Response. Uncoated Crystal. The increase in
response observed for the uncoated quartz oscillator with increasing analyte boiling point
(Figure 2.4) indicates that the uncoated crystal does not selectively interact with any
analyte class. These results also suggest that physisorption (condensation) of the analytes
significantly influences the sensor response trend. This is consistent with results reported
by King [29] which showed that a larger amount of partitioning would occur for higher
boiling analytes exposed to coated quartz crystals. While partitioning cannot occur on the
uncoated device, condensation of the higher boiling analytes is expected to yield similar
results. Thus in our discussion of zirconium acetate film response, the film will be
considered selective for a particular analyte only if the response is significantly difi‘erent
from that expected for condensation.

Zirconium Acetate Coatings. Comparison of the frequency shifts observed for
primary alcohols and normal alkanes (Figure 2.5, Table 2.2) suggests that the acetate film

interacts preferentially with alcohols. We attribute this enhanced alcohol response to

57

hydrogen bonding interactions between the alcohol hydroxyl group and the O atoms (Zr-O
backbone and acetate ligands) in the film. The relatively small number of hydroxyl groups
present in the fihn may also participate in hydrogen bonding interactions. The enhanced
interaction between alcohols and the polymer may serve as the driving force for migration
of alcohols into the pores of the film.

Further discussion of the alcohol responses observed for the zirconium acetate film
is facilitated by correcting the responses to compensate for increasing analyte boiling
point. We behave that this correction can be accomplished by dividing the response of the
acetate film by the values obtained for the uncoated crystal. Figure 2.9 and Table 2.2
show the variation in the acetate coated/uncoated crystal response ratio as a fimction of
analyte boiling point. It should be noted that the alkane response ratios are essentially
independent of hydrocarbon boiling point. This suggests that the procedure has minimized
the effect of condensation on our evaluation of sensor response. With the exception of
tert-butanol and sec-butanol, the ratios calculated for the alcohols decrease dramatically
from methanol to n-butanol. We attribute this trend to molecular sieving properties of the
zirconium acetate film. Larger alcohols cannot diffuse into the pores of the film as readily
as smaller alcohols. Further evidence of the molecular sieving effect is obtained by
examining the variation in response ratio as a function of molar volume (cm3/mole) shown
in Figure 2.10. Indeed, with the exception of iso-propanol, the response ratio is inversely
proportional to the molar volumes of the primary alcohols (Table 2.1). Note that the
molar volume of tert-butanol is larger than any other alcohol examined in this study. The

assertion that pore size effects dominate the alcohol response of the acetate film is

58

consistent with FTIR/ATR results for methanol and ethanol (Figure 2.6) which show that
significantly more methanol diffirses into the film compared to ethanol. In addition, the
longer residence time measured for methanol may indicate that methanol diffuses further
into the film than ethanol.

The response ratio calculated for iso-propanol is significantly higher than the value
obtained for n—propanol. In part we attribute this deviation to the low response recorded
for iso—propanol on the uncoated crystal. This low response would tend to artificially
increase the ratio response reported for iso-propanol.

The low alkane response observed for the zirconium acetate film may be attributed
to the limited interaction possible between the polar acetate film and nonpolar
hydrocarbons. The suppressed alkane response, compared to the uncoated crystal, also
suggests that the hydrocarbons do not diffuse into the acetate film. This is consistent with
the molecular sieving properties proposed to explain the alcohol responses. Note that the
molar volumes of the alkanes are significantly larger than the values reported for the
alcohols examined in this study (Table 2.2). In addition, FTIR/ATR results obtained for
heptane and n-propanol (Figure 2.7) show that the residence time for heptane is
significantly shorter than that observed for n-propanol. Thus we believe that the alkanes

condense on the surface of the acetate film, but do not migrate into the pores of the glass.

59

 

 

 

 

50
O
,
4o -
30
g 7 O 9
M
o d
a
o
g 20-
M
,
O
10 '-
O
. a O
0 n . 1:1 1:1
I I ' I ' l
60 30 100 120
Analyte Boiling Point (°C)

Figure 2.9 Variation in normalized QCM response (zirconium acetate coated
crystal/uncoated crystal) versus analyte boiling point.

60

 

50

404

30-
O 9

Response Ratio

10-

o
- 0
e

 

 

0 1'1 El [:1
r ' r ' r f r 1 f 1 ' r
40 60 80 100 120 140 160

 

Analyte Molar Volume (cm3/mole)

Figure 2.10. Variation in normalized QCM response (zirconium acetate coated
crystal/uncoated crystal) versus analyte molar volume.

61

2.5. Conclusions

Zirconium acetate thin films have been prepared using sol-gel techniques. FTIR
and XPS results indicate that the polymer is a highly crosslinked network consisting of a
Zr-O backbone modified by approximately one bidentate acetate ligand per Zr center.

The zirconium acetate film selectively interacts with alcohols through hydrogen
bonding interactions involving the alcohol hydroxyl group and the oxygen atoms in the
acetate polymer. The responses for most alcohols can be explained in terms of pore size

efi‘ects of the zirconium acetate film.

62

References

1. Avnir, D.; Levy, D.; Reisfeld, R. J. Chem. Phys. 1984, 88, 5956.

2. Dulebohn, J .; Haefirer, S.; Berglund, K.A.; Dunbar, K. Chem. Mater. 1992, 4, 506.

3. Lessard, R.B.; Wallace, M.M.; Oertling, W.A.; Chang, C.K.; Berglund, K.A.;
Nocera, D.G. in Processing Science of Advanced Ceramics; Aksay, A.I.; McVay,
G.L.; Ulrich, DR, Eds; Mater. Res. Symp. Proc. 155: Pittsburgh, PA, 1989,
109.

4. Dulebohn, J.I.; VanVlierberge, B.; Berglund, K.A.; Lessard, R.B.; Yu, J .; Nocera,
D.G. in Better Ceramics Through Chemistry IV; Zelinsky, B.J.J.; Brinker, C.J.;
Clark, BE; Ulrich, D.R., Eds; Mater. Res. Soc. Proc. 180: Pittsburgh, PA, 1990;
733.

5. Newsham, M.D.; Cerreta, M.K.; Berglund, K.A.; Nocera, D.G. in Better Ceramics
Through Chemistry 111; Brinker, 01.; Clark, BE; Ulrich, D.R., Eds; Mater. Res.
Soc. Proc. 121: Pittsburgh, PA, 1988; 627.

6. Yan, Y.; Bein, T. J. Chem. Phys. 1992, 96, 9387.

7. Frye, G.C.; Martin, S.J.; Ricco, A.J.; Brinker, CI. in Chemical Sensors and
Microinstrumentation; ACS Symp. Ser. 403, ACS: Washingston, DC, 1989, 208.

8. Frye, G.C.; Martin, S.J.; Ricco, A.J.; Brinker, CI in Better Ceramics Through
Chemistry III; Brinker, C.J.; Clark, BE; Ulrich, D.R., Eds; Mater. Res. Soc.
Proc. 121: Pittsburgh, PA, 1988;349.

9. Bein, T.; Brown, K.; Frye, G.C.; Brinker, C]. J. Amer. Chem Soc. 1989, 111,
7640.

10. Bein, T.; Brown, K.; Enzel, P.; Brinker, CI. in Better Ceramics Through
Chemistry III; Brinker, C.J.; Clark, BE; Ulrich, D.R., Eds.;Mat. Res. Soc. Symp.
Proc., 121: 1988, 761.

11. Yoshimura, N.; Sato, S.; Itoi, M.; Taguchi, H. Sozai Busseigahu Zasshi 1990, 3,
47.

12. Inubushi, A.; Masuda, S.; Okubo, M.; Matsumoto, A.; Sadamura, H.; Suzuki, K.
in High Tech Ceramics; Vincenzini, P., Eds; Elsevier Science Publishers:
Amsterdam, 1987, 2165.

13. Berglund, K.A.; Nocera, D.G. unpublished work.

14. Gagliardi, C.D.; Dunuwila, D.; Berglund, K.A. in Better Ceramics Through
Chemistry IV; Zelinsky, B.J.J.; Brinker, C.J.; Clark, BE; Ulrich, D.R., Eds;
Mater. Res. Soc. Proc. 180: Pittsburgh, PA, 1990; 801.

15. Handbook of Solubility Parameters and Other Cohesion Parameters;2nd Ed.;
Barton, A.F.M., Ed; CRC Press: Boca Raton, Florida, 1991.

16. Gagliardi, C.D.; Berglund, K.A. in Processing Science of Advanced Ceramics;
Aksay, A.I.; McVay, G.L.; Ulrich, D.R., Eds, Mater. Res. Symp. Proc. 155:
Pittsburgh, PA, 1989 127.

17. Wagner, C.D.; Davis, L.E.; Zeller, M.V.; Taylor, J.A.; Gale, L.H. Surf Interface

Anal. 1981, 3, 211.

18.

19.

20.

21.

22.
23.
24.

25.
26.
27.
28.
29.

63

Software provided by Dr. Andrew Proctor, University of Pittsburgh, Pittsburgh,
PA.

Colthup, N.B.; Daly, L.H.; Wiberley, S.E. Introduction to Infiared and Raman
Spectroscopy; 3rd Ed. Academic Press: San Diego, 1990.

Silverstein, R.M.; Bassler, C.G.; Morrill, T.C. Spectrometric Identification of
Organic Compounds; 3rd Ed. John Wiley & Sons, Inc.: New York, 1974.
Barraclough, C.G.; Bradley, D.C.; Lewis, J.; Thomas, I.M. J. Chem. Soc. 1961,
2601.

Atik, M.; Aegerter, MA. J. Non-Cryst. Sol. 1992, 147&148, 813.

Spinner, E. J. Chem. Soc. 1964, 4217.

Handbook of X-ray Photoelectron Spectroscopy; Wagner, C.D.; Riggs, W.M.;
Davis, L.E.; Joulder, J.F.; Muilenburg, G.E., Eds; Perkin Elmer, 1979.

Doeuff, S.; Henry, M.; Sanchez, C.; Livage, J. J. Non-Cryst. Solids 1987, 89, 206.
Leaustic, A.; Riman, RE. J. Non-Cryst. Sol. 1991, 135, 259.

Mehrotra, R.C.; Bohra, R. Metal Carboxylates, Academic Press: London, 1983.
Deacon, G.B.; Phillips, R.J. Coord Chem. Rev. 1980, 33, 227.

King, W.H.,Jr. Anal. Chem. 1964, 36, 1735.

64

Chapter 3

Study of Polysiloxanes as Selective Coatings for SAW-Based Sensors

3.1. Introduction

Chromatographic stationary phases are commonly used coatings for mass sensor
applications [1]. The phases are readily available with a wide variety of chemical and
physical properties. Their synthesis is normally well controlled, leading to reproducible
molecular weight and chain length distributions between samples. Perhaps most
importantly, there exists a large body of data on the chromatographic selectivity of these
phases. Allowing some predictive capability for their application as mass sensor coatings.

Polysiloxanes are one class of organic polymers traditionally used as gas
chromatographic stationary phases [2] and more recently utilized as sensor coatings

[1,3,4]. The general siloxane structure is shown in Figure 1.

 

 

 

 

 

r ” r
B E
- J X .. a Y

Figure 3.1. General structure of polysiloxane polymer.

Where A, B, D, E can represent the same or different fimctional groups. The ability to
easily change functional groups makes it possible to tailor the physical and chemical

64

65

properties of the polymer. Additionally, the extensive use of siloxane polymers as
stationary phases has prompted research devoted to understanding how their chemical
composition influences polymer selectivity [5-8]. This is useful when researching the
interactions which contribute to mass sensor response.

Mass sensor responses cannot be solely attributed to the addition of mass at the
surface of the crystal. Polymer morphology changes caused by changing temperatures or
solvent induced swelling can lead to a change in the amplitude or velocity of the
propagating wave [9-11]. However, a useful quantitative model for determining
viscoelastic effect does not yet exist. Therefore, recent work describing the response of
SAW devices has focused on using linear solvation energy relationships (LSERs) to
describe and predict analyte responses [3,4]. These studies have ignored the contribution
of viscoelastic changes to the response of the sensor. However, morphology changes may
interfere with accurate determination of LSER and parameters for more elastic films.

In this study responses from surface acoustic wave devices coated with OV-101
(dirnethylmethylsilicone oil), a low molecular weight methylsilicone, and OV-25
(phenylmethyldiphenylsilicone), the most polar of the phenyl silicones, are compared. In
addition, a novel method for qualitatively monitoring the presence of analyte induced
morphology changes is presented. FTIR/ATR studies illustrate numerous morphology
changes for OV-25, while few changes are noted for OV-101. Deviations from the
chromatographic [5] and solvatochromic [7,12] predicted trends are explained in light of

these findings.

66

3.2. Experimental

Materials. OV-101 (M.W. 30,000) and OV-25 (M.W. 10,000) were obtained
from Anspec Chemical Company. Table 3.1 shows the boiling points (vide infra) for the
analytes examined in this study. Analytes were chosen to study analyte-film interactions
for molecules with a wide range of chemical functionalities, reactivities, molecular
weights, and boiling points. The analyte/film interactions for a series of homologous
hydrocarbons (n-hexane to n-nonane) were examined to establish the effect of analyte
boiling point on sensor response. Thiophene, and pyridine, were examined to determine
the selectivity for aromatic heterocycles. Benzene, 2-pentanone, pyridine, nitropropane,
and 1-butanol were studied because they are used to determine McReynolds constants for
chromatographic stationary phases. Analytes were obtained from Aldrich, EM Science or
J. T. Baker Chemical Companies and used as received.

Sensor Test Apparatus. A schematic diagram of the 13.5 mL cell and
experimental apparatus has been shown previously [13]. Analytes were delivered to the
cell as a pulse using a Varian autosampler Series 8000 and a Varian gas chromatograph
injector operated at 300°C. The temperature of the test cell was controlled by resistance
heating of the transfer line. Typical transfer line temperatures are in the range of 3 15-33 5°
C. Typical operating conditions are 40 cc/min flow of dry N2 and test cell temperatures of
60°C. Normal autoinjection volume range of analytes are limited to 1.0 11L to 4.0 11L
based on the expected sensor response. Normally, six injections of each analyte were

made on three coated SAW devices to ensure reproducibility.

67

 

 

Table 3.1. List of analyte boiling points, dipole moments and sensor response.
Response
(I-Iz/umole)
Boiling Dipole
Point Moment
Analyte (°C) (debyes) Blank SAW OV-101 OV-25
pentane 31 0 9 39 41
methanol 65 1.70 40 18 56
hexane 69 0 29 185 297
ethanol 78 1 .69 98 53 167
benzene 80 0 75 345 909
iso-propanol 82 1.66 104 86 221
thiophene 84 0.55 144 373 1212
n-propanol 97 1 .68 289 175 485
heptane 98 0 124 456 744
2-pentanone 100 2.69 333 502 1760
toluene l 1 1 0.36 305 846 2682
pyridine l 16 2.19 1177 1226 3 508
n-butanol 118 1.66 1947 1211 2191
nitropropane 121 3.66 1587 1742 5130
octane 126 0 451 1344 1532
chlorobenzene 131 1.69 2185 2324 5799
nonane 151 0 1827 5004 6024

 

68

Surface Acoustic Wave (SAW) Device. Data were obtained using a 200 MHz
SAW resonator array microbalance (F emtometrics, Costa Mesa, CA). The 200 MHz
device has been described previously [14]. The array device consists of 6 individual
resonators mounted on a single platform that is directly inserted into the circuit board.
The difference between each resonator and a separate reference device is determined and
this beat frequency is the measured signal. Power is supplied and frequency counted by an
external unit provided by the manufacturer.

Injection System. A Varian series 8000 autoinjector system was mounted on a
steel plate above a Varian injector that was removed from a gas chromatograph that was
subsequently mounted beneath the same steel plate. The pneumatic pressure is provided
from a standard laboratory gas cylinder, while purified gas with controlled flow rates are
supplied to the injector by Maxtech Mass Flow Controllers. Both the actuators and the
cell are using nitrogen gas.

Data Acquisition and Manipulation. The fi'equency of the resonators were
collected using a digital counting board (MetraByte, Taunton, MA) and stored in a
Gateway 2000 486-66 personal computer. The software used to collect the data was
written by the manufacturer, and is capable of collecting data for all 6 channels at
resolutions of 1-100 Hz (0.2-6 sec). The software also allows for the displaying and
printing of data, and for manual peak determinations. Additionally, the standard software
has been modified by F emtometrics to allow for timed, multiple runs to allow for the full
utilization of the Varian autoinjector capabilities. In this study, data analysis was

performed using custom software written using LabView for Windows (v2.52, National

69

Instruments, Austin, TX). This software takes the data output file and separates it into
each individual channel, and then simultaneously displays the response of all 6 channels on
a single page. Additionally, the software counts peaks, determines peak heights, peak
areas, and the relevant analytical information. It also allows for the user to select either
Savitsky—Goolay smoothing or median filtering of the data, and some control over the
sensitivity of the peak determination. An output file that contain peak height, area,
analytical information, and relevant processing information (such as filtering, sensitivity,
etc.) can be automatically generated whenever the user is satisfied with the level of data
analysis. Data presented in this paper were collected with a resolution of 10 Hz, which
provides 5-7 points across the peak maximum

Coating Procedure. OV-lOl and OV-25 films are prepared by spray casting from
dilute toluene and acetone solutions (~200 ppm), respectively. The resonators were
cleaned by soaking in iso-propanol, and then air dried. An airbrush is used to deliver the
polymer solution to the entire surface of the crystal. The films were then allowed to air
dry and stored in a desiccator. The change in frequency was constantly monitored during
the coating procedure. The coating thicknesses used in this study were 150 kHz i 10
kHz.

Fourier Transform Infrared Spectroscopy (FTIR). FTIR/ATR experiments
were performed using a Mattson Instruments Galaxy 5020 FTIR spectrometer equipped
with a wideband mercury cadmium telluride (MCT) detector. A commercial attenuated
total reflectance (ATR) accessory (Specac, Inc.) designed for I-IPLC detection has been

modified to perform analyte/film studies. The attachment consists of a 500 uL cell in

70

contact with a zinc selenide (ZnSe) internal reflectance element (IRE), contained in a
temperature controlled water bath held at 60°C. The fittings of the cell have been changed
to accommodate 1/8" stainless steel inlet and outlet fittings that are connected to a gas
handling system. The inlet of the ATR cell is connected to a gas chromatograph injector
by a 4" long, 1/8" id stainless steel transfer line. The injector and transfer line were heated
to 250°C. Analytes were injected as 0.5 11L pulses. The polymers were prepared by drop
coating 400 11L of 1000 ppm polymer solution on to the IRE and allowing the solvent to
air dry. The resulting polymer films were approximately 0.5 pm thick. FTIR/ATR film
spectra were collected with 4 cm'1 resolution.

A software package supplied by Mattson (FIRST Macros) was modified to collect
time dependent FTIR/ATR data. The program uses a spectrum of the polymer coated
IRE as the background for firrther data collection. Eighty spectra (each spectrum is the
sum of 10 scans taken at one second intervals) were obtained following the injection of
each analyte. Each spectrum can be integrated over a wavenumber interval characteristic
of the analyte. Plots of integrated peak area versus elapsed time are used to determine the
kinetics of analyte-polymer interaction. Additionally, changes in analyte absorption bands
or polymer background may indicate modes of analyte/film interaction or changes in

polymer structure or morphology.

3.3. Results

SAW Response. Uncoated Device. Figure 3.2 (Table 3.1) shows the variation in

sensor response (Hz/umole) as a function of analyte boiling point for the uncoated device.

71

Frequency change per mole was chosen to account for differences in the volume of analyte
injected. While analyte boiling point emphasizes the effect of physisorption on sensor
response. The uncoated SAW device exhibits an increase in response with increasing
analyte boiling point. Alkanes produce slightly lower responses than analytes with
comparable boiling points. Analyte response, with the exception of octane, increase
dramatically at pyridine (b.p. 116°C) and above.

OV-101 Film. Figure 3.3 shows the variation in sensor response as a function of
analyte boiling point for the OV-101 coated SAW device. As with the uncoated device,
the response of the OV-101 film increases with increasing analyte boiling point. However,
it exhibits a smoother trend, and does not show the sudden increase noted for the
uncoated device. The alcohols, with the exception of n-butanol, define the lower limit of
the smooth increase. All of the analytes, with the exception of the alcohols, produce a
greater response on the OV-101 coated device.

OV-25 Film. Figure 3.4 shows the response of OV-25 film plotted versus analyte
boiling point. As with the uncoated and OV-lOl coated sensors there is a general increase
in response with increasing analyte boiling point. However, benzene and thiophene (b.p.
80°C and 84°C, respectively) exhibit significant enhanced responses compared to analytes
of similar boiling points. Alcohols and alkanes produce responses slightly lower than the
remaining analytes. The response of the OV-25 coated sensor is greater for every analyte
than that of the uncoated device.

OV-25/OV-101 Ratios. Table 3.2 lists the response ratio for OV-25/OV-101, while

Figure 3.5 shows the ratios plotted versus analyte boiling point. The alkanes (squares)

2500

2000 -

: fl

Response (Hz/umole)

500 -

Figure 3.2.

72

 

‘s”
l

s—s
O
O
O
1

El

 

El

 

 

/
O-M ' u

r
20 40 60 80 100

U

1

r
120

Analyte Boiling Point (°C)

Response curve for uncoated SAW device.

' U
140

 

160

73

 

 

 

5000 - O
4000 -
1.3 .
0
§ 3000-
;“ .
8
0
g- 2000 -
d)
M O
. /\O
(D
1000 - o/
/
. (I) d)
o \o
0 -—r0Fi_'rO-I—cp s ' 1 I 1 l s

 

20 40 60 80 100 120 140 160
Analyte Boiling Point (°C)

Figure 3 .3. Response curve for OV-101 coated SAW device.

74

7000

 

6000- A/

5000 -

l

A
§
1

 

D)
O
O
O
1

Response (Hz/umole)

\

 

 

 

 

 

2000 -
A l
. A
A

1000 - \

- A

AA
0 - —FA-1—-—_I—-_l— t F I I U I f I I
20 4O 60 80 100 120 140 160

Analyte Boiling Point (°C)

Figure 3.4. Response curve for OV-25 coated SAW device.

75

exhibit the lowest increase from OV-101 to OV-25 with a ratio of 1.3 i 0.3. The alcohol
(circle) response ratios are similar within experimental error except for n-butanol (b.p.
117°C). n-Butanol exhibits a ratio of 1.81. All of the other analytes (triangles) produce
response ratios of 2.5 or greater.

FTIR/ATR. The diffirsion time of analytes through the ATR cell was not
significantly difl‘erent for the coated or uncoated IRE. Therefore, differences in diffusion
rates were not determined fi'om the FTIR/ATR results. The residence time of the analyte
seemed to be governed by analyte boiling point and not analyte/film interactions.
Although analyte kinetics were not evident from the ATR experiments, changes in
polymer film absorption bands were observed for some analytes. These changes were
evident after analyte injection, and were attributed to vapor induced polymer morphology
changes.

OV-101 FTIR/A TR. OV-101 exhibited the smallest amount of polymer changes.
Indeed, the majority of analytes did not produce any changes in the polymer spectrum.
Figure 3.6 shows a series of spectra collected during the injection of methanol across an
OV-101 coated IRE. Figure 3.6a is the initial spectrum taken before methanol has been
injected. After injection methanol absorption bands are evident in spectrum 2 which is
shown in 3.6b. By spectrum 3, which is shown in 3.6c, methanol has completely purged
from the cell.

Certain analytes did produce OV-101 polymer changes, these included pyridine,
nitropropane, octane, chlorobenzene, and nonane. Figure 3.7 illustrates a series of spectra
taken during injection of nonane across a OV-101 coated IRE. Figure 3.7a is simply a

reference spectrum of OV-101 taken prior to any analysis. Figure 3.7b is the spectrum

Table 3.2.

76

Response ratios for probe analytes.

 

Response Ratio

 

 

Analyte Uncoated/OV- 1 01 OV—25/OV- 1 01
pentane 4.43 1 .05
methanol 0.45 3.1 1
hexane 6.3 1 1 .61
ethanol 0.54 3.14
benzene 4.61 2.64
iso-propanol 0.82 2.57
thiophene 2.59 3.25
n-propanol 0.60 2.78
heptane 3.68 1 .63
2-pentanone 1.51 3.51
toluene 2.77 3. 17
pyridine 1.04 2.86
n-butanol 0.62 1.81
nitropropane 1 . 10 2.94
octane 2.98 1 . l4
chlorobenzene 1.06 2.50
nonane 2.74 1.20

 

77

 

3.5- A

Response Ratio

1.5-

E El

 

 

l ' l ' l

l 0 - D
. q- I I I I I I I I
0 40 60 80 100 120 140 160

2

Analyte Boiling Point (°C)

Figure 3.5. OV25/OV-101 response ratio.

 

78

 

(C)

Absorbance

(b)

i (a)

 

 

1 l 1
4000 3000 2000
Frequency (cm-1)

Figure 3.6. FT IR/ATR spectra of methanol on OV-101.

1
1000

 

 

 

 

79

before the injection of nonane. It is important to remember that the polymer coated IRE is
used as the background for the entire run. Therefore, any changes in the polymer
structure should produce absorption bands in the following scans. Adsorption bands at
3000 cm“1 in figure 3.7c indicate nonane in present in the ATR cell. Also in this spectrum,
changes in polymer morphology are evident from the negative features around 1050 cm".
Figure 3.7d shows the polymer film spectrum returned to original features once the
nonane is purged from the cell. All of the analytes that produced morphology changes on
OV-101 exhibited similar profiles to that of nonane.

OV-25 F TTR/A TR. Unlike OV-101, OV-25 exhibited polymer absorption changes
for a majority of the compounds injected. Indeed, the only analytes that did not affect the
polymer were pentane, methanol, hexane, ethanol, and iso-propanol. Of the analytes
which showed viscoelastic effects all, except nonane, produced positive shifts in the
polymer absorption bands. Additionally, in contrast to OV-101, the polymer changes
persisted after the analytes had purged fi'om the cell. An example of the changes noted for
OV-25 is shown for pyridine in figure 3.8a-f. Figure 3.8 (a) and (b) are reference spectra
of OV-25 and pyridine, respectively. Although certain bands at 1800 cm'1 and below do
overlap, their shape and relative responses do differ between OV-25 and pyridine. Figures
3.8 (c) through (1) are spectrum 1, 2, 4, and 80 of the pyridine kinetic run. In 3.8d the
relative peak absorbances match those of pyridine and are therefore attributed to the
analyte. However, by the fourth spectrum (figure 3.8e), the band profiles and ratios are
identical to those of the OV-25 film. Finally, at spectrum 80, nearly 15 minutes after the

injection of pyridine, the polymer bands are still present.

80

 

 

(C)
(b)

Absorbance

 

 

 

 

 

1M f m

1 1 1 1
4000 3000 2000 1000

Frequency (cm'l)

 

 

 

Figure 3.7. FTIR/ATR spectra of analyte induced morphology changes for nonane
on OV-lOl.

81

 

1(I) l I I

1 L M
(e) 1... M
(d) 3., M1

(C)

 

 

 

 

Absorbance

 

 

 

W Quintin

(a)
#11».— 1 1

1
4000 3000 2000 1000
Frequency (cm'l)

 

 

 

 

Figure 3 .8. FTIR/ATR spectra of analyte induced morphology changes for pyridine on
OV-25.

82

As mentioned above, nonane does not produce positive peaks on OV-25 as noted
for the other analytes. In fact, nonane produces negative polymer peaks while present in
the cell. However, these peaks reduce in magnitude and eventually reverse as nonane
purges fi'om the ATR cell. This change in the polymer background is illustrated in Figure
3.9a-d. Figure 3.9a shows the initial spectrum of the nonane kinetics run. A slight
negative peak is visible just above 1000 cm". Upon injection of luL of nonane this
negative band greatly increases. The presence of nonane is evident from the features just
below 3000 cm'l. By spectrum 4 of the series (figure 3%) the nonane is beginning to
purge item the cell and the negative polymer peaks are decreasing in magnitude.
Spectrum 20 (figure 3.9d) shows all the nonane purged form the cell, but the polymer

feature is now positive. Polymer peaks slowly return toward baseline in the continuing

spectra.

3.4. Discussion

Effect of Coating on Sensor Response. Uncoated Crystal. The general increase
in response observed for the uncoated SAW device with increasing analyte boiling point
(Figure 3.2) indicates that physisorption (condensation) of the analytes significantly
influences the sensor response trend. This is consistent with results reported by King [15]
which showed that a larger amount of partitioning would occur for higher boiling analytes
exposed to coated quartz crystals. While partitioning cannot occur on the uncoated

device, condensation of the higher boiling analytes is expected to yield similar results.

83

 

f-

I
£93-1L, - II_

(C)

 

5

 

Absorbance

 

 

 

 

 

,0», A) u,

£9111_
1 I I I I 1 J ATV?“
4000 3000 2000 1000

Frequency (cm'l)

 

 

 

Figure 3.9. FTIR/ATR spectra of analyte induced morphology changes for nonane on
OV-25.

84

Notwithstanding, physisorption is not the only interaction contributing to the
uncoated sensor response. If the response profile were due solely to condensation we
would not expect the sharp rise at pyridine and above, or the slight enhancement of other
more polar analytes over alkanes. Indeed, we would expect a trend more similar to that
observed for the OV-101 coated device. The sharp rise at pyridine is attributed to a
combination of increased physisorption and the more polar analytes interacting more
strongly with the polar quartz surface. This type of interaction with an uncoated SAW
device has been observed previously [16].

OV-101 Coated. As stated above the OV-101 coated SAW device exhibits the
smoothest increase in response with increasing analyte boiling point. OV-101 is
considered to be a nonpolar stationary phase which separates compounds based solely on
boiling point [2]. Solvatochromic parameters calculated from chromatographic data and
linear solvation energy relationship calculations firrther establish that PDMS interact solely
through dispersion forces [7,12]. This type of interaction should produce a response
curve based almost entirely on analyte boiling point due to physisorption. With the
exception of small deviations this is exactly what is observed for the OV-101 coated
device. The slight deviations from this trend are caused by the alcohols defining the lower
limit of response on the OV-101 coated sensor. This is actually in agreement with
chromatographic findings which suggest that alcohols do not show eflicient partitioning
on OV-101 [6]. In addition, comparison of retention volume of ethanol and a series of

alkanes shows extremely poor retention of the ethanol [17].

85

OV-25 Coated. OV-25 is the most polar of the phenyl siloxanes, however
chromatographic data suggests that a phenyl content of no greater than 50 % is sufficient
for better separation. Decrease in retention volume has been observed with increasing
phenyl content above 50%. However, solvatochromic investigations do show an increase
in the 7t* character of the stationary phase with increasing phenyl content. This increased
p* should lead to an increase in the response to analytes containing this parameter. The
OV-25 coated device only shows enhanced responses for benzene and thiophene and not
for the other aromatic. The lack of response maybe due to an artifact of condensation
effects. The response due to condensation may be large compared to that of the 11:“
interaction. Therefore, any possible enhancement is masked by the already large response.

Condensation is an important factor in the response of the OV-25 coated sensor.
However, the response of OV-25 is cannot be completely explained by physisorption. The
alkane series defines the response expected for simple condensation, while the remaining
non-alcohols exhibit enhanced response. All of the analytes which display enhanced
response exhibit considerable polarizability coefficients [18-23]. Solvatochromic
parameters determined fiom chromatographic data suggest OV-25 interacts through
dispersion forces, polarizability and even hydrogen bond accepting characteristics [7,12].
This suggests that polarizability is a major contributor to the response of our OV-25
coated SAW device. The alcohol response is somewhat puzzling since alcohol have
polarizability factors comparable to those exhibited by the higher responding analytes.
However, the response is not contrary to data observed during chromatographic
separations of alcohols. Indeed retention volume of ethanol falls considerably below that

of other comparable analytes [17]. Alcohols also tend to exhibit anomalous results

86

during separation by OV-25 stationary phase. The remaining analytes exhibit considerable
polarizability character.

OV-101 and OV-25 Comparison. The comparison of OV-25 with OV—101 sensor
responses (Figure 3.5) indicates that OV-25 exhibits larger responses for the entire analyte
set. This increased response is fully expected for the analytes which interact through
induced dipole forces. However, alkanes which interact solely through dispersion forces
do not exhibit efficient partitioning in OV-25 during chromatographic separation [24].
The alkanes Indeed, retention times of n-alkanes decrease dramatically from OV-101 to
OV-25 [17]. Therefore we expect OV-101 to exhibit larger responses to the alkane series
than the OV-25 coated device.

Although in need of firrther study, we attribute the increase in response of the OV-
25 not only to increased retention of analyte, but also to larger viscoelastic changes. Our
FTIR/ATR results display significant changes in the polymer structure during the presence
of the analyte and even after it has purged from the cell. Grate et al. [25] proposed that
SAW responses were predominantly due to polymer morphology changes. Grate also
showed that an increase in polymer modulus should lead to an increase in the frequency
shift of the SAW device. Therefore, a polymer which exhibits a greater viscoelastic effect
should produce a greater to nonselectively retained analytes.

Ballantine [26], and Bartley and Dominguez [10] observed little or no viscoelastic
responses for the systems they examined. These conflicting results suggest that responses
due to polymer morphology changes are film and analyte dependent. Therefore, it is not

surprising that large changes were observed for OV-25, but OV-101 exhibited swelling

87

only for analytes that showed a large amount of physisorption. Indeed, Patrash and

Zellers recently [4] reported a greater relaxation time for the OV-25 polymer.

3.5 Conclusions

The selectivity of OV-101 and OV-25 polysiloxane films have been investigated
using SAW devices and FTIR/ATR spectroscopy. The results for OV-101 and OV-25
agree fairly well with what is predicted from solvatochromic and chromatographic data.
The responses which do not agree are attributed to the analyte induced viscoelastic
changes observed in OV-25. FTIR/ATR analysis is able to observe these polymer

morphology changes.

88

References

1. Guilbault, G. G.; Jordan, J. M. CRC Crit. Rev. Anal. Chem. 1988 19, 1, 1-28.

2. Haken, J. K.; J. Chromatogr. 1977, 300, 247-288.

3. Grate, J. W.; Abraham, M. H. Sens. &Act. B. 1991, 3, 85-11.

4. Patrash, S. J.; Zellers, E. T. Anal. Chem. 1993, 65(15), 2055-2066.

5. Yancey, J.A. J. Chromatog. Sci. 1985, 23, 161.

6. Yancey, J .A. J. Chromatog. Sci. 1985, 24, 117.

7. Brady, J. A.; Bjorkman, D.; Herter, C. D.; Carr, P. W. Anal. Chem. 1984, 56,
278-283.

8. Ito, M. M.; Kato, J.; Takagi, S.; Nakashiro, E.; Sato, T.; Yamada, Y.; Saito, H.;
Narniki, T.; Takamura, 1.; Wakatsuki, K.; Suzuki, T, Endo, T. J. Am. Chem. Soc.
1988, 110, 5147.

9. Ballantine, D. 8., Jr.; Wohltjen, H. IEEE Ultrason. Syrnpos. 1988, 559.

10. Bartley, D. L.; Dominquez, D. D. Anal. Chem. 1990, 62(15), 3069.

11. Frye, G. C.; Martin, S. J.; Ricco, A. J. Sens. Mater. 1989, 1-6, 335.

12. Li, J.; Zhang, Y.; Carr, P. W. Anal. Chem. 1992, 64, 210.

13. Townsend, E. B.; Ledford, J. S. submitted Anal. Chem. 1994.

14. Bowers, W.D; Chuan, R.L.; Duong, T.M. Rev. Sci. Instrum. 1991, 62(6), 1624-
1629.

15. King, W.H.,Jr. Anal. Chem. 1964, 36, 1735.

16. Martin, S. J.; Ricco, A. J.; Ginley, D. S.; Zipperian, T. E. IEEE Trans. Ultra,
Ferroe. Freq. Contr. 1987, 34(2), 142.

17. Parcher, J. F.; Hansbrough, J. R.; Koury, A. M. J. Chromatogr. Sci. 1978, 16,
183.

18. Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chem. Soc.
Perkin. Trans. 2 1990, 1451.

19. Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1990, 518, 329.

20. Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J. J.; Taylor, P. J. J. Chem.
Soc. Perkin. Trans. 2 1990, 521.

21. Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Duce, P. P.; Morris, J. J.; Taylor, P.
J. J. Chem. Soc. Perkin. Trans. 2 1989, 699.

22. Abraham, M. H.; Duce, P. P.; Grellier, P. L.; Prior, D. V.; Morris, J. J.; Taylor, P.
J. Tetrahedron Lett. 1989, 29(13), 1587.

23. Li, J.; Dallas, A. J.; Carr, P. W. J. Chromatogr. 1990, 517, 103.

24. Kong, I. M.; Hawkes, S. J. J. Chromatogr. Sci. 1976, 14, 279.

25. Grate, J. W.; Klusty, M.; McGill, R. A.; Abraham, M. H.; Whiting, G.; Andonian-
Haftvan, J. Anal. Chem. 1992, 64(6), 610.

26. Ballantine, D. 8., Jr. Anal. Chem. 1992, 64(24), 3069.

Chapter 4

Future Work

4.1. Tailoring Metal Carboxylate Polymer Oxide Glasses

When choosing polymers for use as chemical mass sensor coatings one needs to
consider the properties of the target analytes. Polymers containing functional groups that
enhance the chemical interactions with the probe compounds should lead to a more
selective response. Additionally,if polymers possess similar physical properties differences
in sensor response are more readily attribited chemical effects intead of physical changes in
the polymer film.

Transition metal carboxylate films are relatively new materials which are easily
synthesized. The straightforward synthetic process allows chemical composition and
physical properties of the films to be readily altered. It is this ability to tailor the metal
carboxylate characteristics that make them usefirl, interesting materials for sensor
development.

As mentioned previously, production of metal carboxylate films through carboxylic
acid modification of the alkoxide results in high quality films. The final polymer consists
of a metal-oxygen backbone with bidenate chelating/bridging carboxylate groups attached
to the metal center. Carboxylate films have been synthesized using carboxylic acids of

increasing chain length up to oxanoic acid [1]. It has been shown that the choice of acid

89

90

can effect properties ( e. g., solubility and hydrophobicity) of the final films [1]. However,
the abiliy to tailor the chemical properties of metal carboxylate films using various
carboxylate ligands has not been fully explored.

We intend to develop transition metal carboxylate chemistry necessary to produce
thin films with ligands that emphasize specific intermolecular interactions. The carboxylic
acids we intend to incorporate are p-arninobenzoic, p-fluorobenzoic, glutaric, valeric, and
4-oxopentanoic acid. Although each acid exhibits a number of solubility interactions, each
was chosen to enhance a specific chemical property. For example p-fluorobenzoic acid
was chosen for its large polarizability value although it also exhibits dispersion and dipole
interactions.

Recently we have attempted synthesizing zirconium benzoate films fi'om zirconium
n-propoxide and benzoic and valeric acid in the usual method [Chapter 2]. Benzoic acid
was used in a 1:1 ratio with the alkoxide, and the valeric acid was used in excess to
produce the conventional 9:1 acid ratio. The method produced a clear colorless solution
which was cast to form a crack free colorless film. FTIR spectra of the dried film (Figure
4.1) confirmed the presence of the benzoate ligand, and showed no indication of dimerized
carboxylic acid. Unfortunately, preliminary sensor results indicate little difference between
films containing valerate and those containing benzoate ligands. This suggests that the
valeric acid dominates the chemical properties of the film containing benzoate ligands.

Our benzoic acid results suggest that films must be synthesized using only a single
carboxylic acid. Initial results using only benzoic acid indicate that the synthetic route

used for valeric acid films will not be suitable for other carboxylic acids. Benzoic acid,

91

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4000 3000 2000 1 000

Figure 4.1. FTIR spectrum of zirconium valerate/benzoate film.

92

when mixed with zirconium n-propoxide in n-propanol produces a white precipitate on
addition of water. Attempts to cast films without the addition of water, with a R3 of 1,
produced fractured films and a white precipitate upon evaporation of the n-propanol.
FTIR results suggest some substitution by benzoate ligands does occur, however the
extent of ligation is still under investigation. These results indicate that synthesis of metal
caboxylate thin films will require alternative routes.

One alternative procedure which may lead to the production of high quality fihns is
the use of additional solvents in the hydrolysis and condensation steps. Hench [1,2] has
established that drying can be controlled through the addition of drying control chemical
additives (DCCAs). These additives reduce the drying rate allowing the polymer network
to form before the solvent evaporates. This procedure has been successfully used for the
synthesis of monolithic silicates. Lessard et al. [3] modified the Hench procedure, using
different solvent additives, to produce silicate monoliths of greater density and, good
optical quality. In addition, different molecular precursors have been obtained depending
on the solvent used. Zirconium propoxide dissolve in cyclohexane leads to the formation
of monolithic gels [4]. The cyclohexane slows hydrolysis and condensation rates allowing
the gels to form.

The addition of solvents to metal alkoxides can assist in forming gels by altering
the rates of hydrolysis and condensation. However, a major difficulty in the synthesis of
metal carboxylate thin films is not synthesizing the polymer, but producing a high quality
fiacture free film. Shrinking of the films during drying can often cause them to crack and

flake. For the production of monoliths this cracking is can be controlled by controlling the

93

evaporation rate of the solvents. However, for thin films evaporation is not easily
controlled. Instead, we propose allowing the gels to dry rapidly and then redissolving
them in an appropriate solvent. Once dissolved the polymers could be recast as thin films.
This method is commonly used for organic polymers. Although the physical properties of
the polymers maybe altered their chemical composition remains unchanged.

The overall goal of this portion of research is to synthesize films with difi‘erent
solubility parameters. Each acid was chosen to enhance a single property in the
solvatochromic parameters used in LSER. This systematic study should allow a better
understanding of the importance of these interactions in mass sensor response. In addition
to calculating LSER, the production of the films should lead to some important insight on
the nature and importance of the carboxylic acid. Presently little is known about the exact
nature of this additive. Film physical properties such as pore size, polymer chain length,
and average molecular weight are also of interest in such a study. Attempts to understand
and control these properties will also be made since each property will have influence the

response of mass sensors.

4.2. Molecular Recognition Centers in Polymer Oxide Glasses

Tailoring the backbone and organic ligands of sol-gel films will allow some control
of the selectivity. However, it is unlikely that analyte specificity is obtainable through this
method of sol-gel modification. This is especially true in the area of mass sensors. Since

mass sensor response is largely influenced by dispersion forces, diffirsion, and

94

physisorption. Therefore, it is necessary to incorporate a more selective recognition
center into the design of a chemical sensor.

Molecular recognition has been simply defined as a high level of control over
chemical interactions [5]. However, a more accurate definition is a process involving both
binding and selection of substrate by a given receptor molecule, and possibly a given
function. It implies a structurally well defined pattern of intermolecular interactions [6].
Molecular inclusion compounds, a specific type of molecular recognition center, are able
to surround all or part of the guest, establishing numerous non-covalent binding
interactions. This arrangement generally achieves higher recognition through the large
contact area of the host and guest. A large variety of inclusion and recognition centers
have been developed [67-8].

The porosity, optical clarity and non-reactivity of sol-gel glasses makes them ideal
matrices for the incorporation of host molecules. Indeed, a variety of optical centers,
molecular recognition and molecular inclusion compounds have been entrapped in sol-gel
glasses [9,10]. Much of the research has focused on probing the properties of the gels
with the entrapped species [1 1-1213]. However, the addition of compounds such as
enzymes [14,15], cryptate complexes [16], porphyrins [17] and cyclodextrins [18]
concentrated on the development of chemical sensors.

The first successful addition of an organic molecule to a sol-gel matrix was
performed by Avnir et al. [19]. Researchers entrapped rhodamine 6G (R6G), an organic
laser dye, in a silica sol-gel glass to investigate its use as a solid state laser matrix. R6G

was simply added to the glass before gelation, and the glass dried as usual. Avnir showed

95

that the incorporated dye maintained its absorption and emission properties within the
silica matrix. Attempts to leach the dye by soaking the gel in methanol and water
established that the dye was not simply adsorbed at the exposed walls of the pores, but
trapped within the rigid glass, most likely within the pore structure. The retention of the
optical properties of the dye was encouraging for a wide range of applications.

More recent work focused on the development of chemical sensor has established
the utility of sol-gel matrices as supports for chemical sensors. Dulebohn et al. [16]
entrapped an europium cryptate complex in a titanium film. Using the target analyte as a
light harvesting center (LHC), he demonstrated the ability to use absorption energy
transfer emission (AETE) to produce a signal in sol-gel films. Titanium films have also
been used as the matrix for porphyrins in order to detect aqueous carbon monoxide [17].
Narang et al. [18] incorporated dansyl-glysine-B—cyclodextrin into a silica glass to study
the interaction with the target analyte bomeol. Dansyl, a fluorophore, is highly fluorescent
in nonpolar enviromnents was tethered to the rim of the CD and resides within the CD
cavity. Exposure to bomeol displaces the dansyl from the cavity which led to a dirrrinished
fluorescence signal.

As mentioned above, Narang et al. [18] established the ability to utilize inclusion
phenomena in sol-gel matrices. However, the use of transition metal films in place of
silicates offers greater versatility and control of film properties. This work will focus on
the incorporation of CD complexes into transition metal carboxylate thin films. Unlike the
silicate glasses, the incorporation of CD into these polymers is not straightforward. Past

attempts at entrapment have deactivated the inclusion properties of the cyclodextrin by

96

filling the cavity with carboxylic acid during synthesis [20]. Since the CD is simply
entrapped and not covalently bound to the metal any attempt at rinsing the excess valeric
acid also removes the CD. Attempts at binding the CD to the zirconium center once the
film has been cast show that although the CD does adsorb to the film, it is a reversible
process.

We hope to covalently attach the CD to the metal center during polymerization of
the oxide glass. Although the CD may still become filled during hydrolysis and
condensation it should be possible to empty the cavity while retaining the CD in the film.
The easiest method of attaching the CD to the metal center is through derivatization of the
CD's bottom rim. Ideally, the modification of the secondary alcohol to a carboxylic acid
group should provide a tethering site to the inorganic backbone. However, if this proves
too difficult the use of the tosylate has been proposed for reaction in the sol-gel matrix
[21].

Most recognition centers are not perfectly selective, interferences fi'om similarly
sized or charged species often produce false signals. The use of sol-gel films as matrices
for the inclusion compounds could lead to a reduction in interferences. By controlling the
solubility properties of the films it is possible to incorporate a second dimension of
selectivity.

Another level of confidence is presently being explored in the areas of array
devices and pattern recognition algorithms. SAW devices are presently available as six

array devices. Coating each array with films of different solubility properties or containing

97

different recognition centers leads to differences in response. It is the development and

characterization of film-analyte interactions that will allow the use of sensor array devices.

98

References

1. Hench, L. L. in Science of Ceramic Chemical Processing, Hench, L. L.; Ulrich, D.
R., Eds; John Wiley & Sons: New York, NY, 1986, 52.

2. Hench, L. L.; Orcel, G.; Nogues, L. in Better Ceramics Through Chemistry II;
Brinker, C. J.; Clark, D. E.; Ulrich, D. R. Eds; Mater. Res. Soc. Symp. Proc. Vol.
73, Pittsburgh, PA, 1986, 35.

3. Lessard, R. B.; Wallace, M. M.; Oertling, W. A.; Chang, C. K.; Berglund, K. A.;
Nocera, D. G. in Processing Science of Advanced Ceramics; Aksay, A. I.; McVay,
G. L.; Ulrich, D. R., Eds; Mat. Res. Soc. Symp.Proc. 155: Pittsburgh, PA, 1989;
109.

4. Kundu, D.; Ganguli, D. J. Mater. Sci. Lett. 1986, 5, 293.

5. Bein, T. in Supramolecular Architecture, Bein, T. Ed.; ACS Symp. Ser. 499:
Washington, DC, 1992, 1.

6. Lehn, J-H. Chemica Scripta 1988, 28, 237.

7. Galan, A.; Mendoza, J.; Toirin, C.; Bruix, M.; Deslongchamps, G.; Rebek, J. J.
Am. Chem. Soc. 1991, 113, 9424.

9. Katz, H. E. in Inclusion Compounds; Acedemic Press, Orlando, FL. 1991, 391.

9. Lev, O. Analusis 1992, 20, 543.

10. Avnir, D.; Braun, S.; Ottolenghi, M in Suprarnolecular Architecture, Bein, T. Ed.;
ACS Symp. Ser. 499: Washington, DC, 1992, 385.

11. Avnir, D.; Kaufinan, V. R. Langmuir 1986, 2, 717.

12. Kaufman, V. R.; Avnir, D.; Pines-Rojanski, D.; Huppert, D. J. Non-Cryst. Solids
1988, 99, 379.

13. Pouxviel, J. C.; Dunn, B.; Zink, J. I. J. Phys. Chem. 1989, 93, 2134.

14. Braun, S.; Shtelzer, S.; Rappaport, 8; Avnir, D.; Ottolenghi, M. J. Non-Cryst.
Solids 1992, 147&148, 739.

15. Ellerby, L. M.; Nishida, C. R.; Nrshida, F.; Yamnaka, S. A.; Dunn, B.; Valentine,
J. S.; Zink, J. 1. Science 1992, 255, 1113.

16. Dulebohn, J. 1.; Van Vlierberge, B.; Berglund, K. A.; Lessard, R. B.; Yu, J.;
Nocera, D. G. in Better Ceramics Through Chemistry IV; Zelinsky, B. J. J .;
Brinker, C. J.; Clark, D. E.; Ulrich, D. R. Eds; Mat. Res. Soc. Symp.Proc. 180:
Pittsburgh, PA, 1989; 733.

17. Gagliardi, C. D.; Dunuwila, D.; Chang, C. K.; Berglund, K. A. in Better
Ceramics Through Chemistry V; Hampden-Srnith, M. J .; Klemperer, W. G.;
Brinker, C. J ., Eds; Mat. Res. Soc. Symp. Proc., 271: Pittsburgh, PA, 1992; 645.

18. Narang, U.; Dunbar, R. A.; Bright, F. V.; Prasad, P. N. Appl. Spectrosc. 1993,
47, 1700.

19. Avnir, D.; Levy, D.; Reisfeld, R J. Phys. Chem. 1984, 88, 5956.

20. Private communication with Dr. K. A. Berglund.

21. Technical Proposal CFMR 1992-1993.

onN STATE UNIV. LIBRARIES
lllllllHIIHHlIlIHlIlllllllllllllllllll
12 3010336778

|
9

nICHI
E3