PLACE N RETURN BOX to monthl- chockout from your mood.
TO AVOID FINES Mum on or baton dd. duo.

DATE DUE DATE DUE DATE DUE

* amimco
- 2 3 a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[:3
JI | |

MSU Is An Afflnndlvo Wan-l Oppommuy Inultwon
W

 

 

 

m1

FUNDAMENTAL STUDIES OF AN ALYTE/FILM INTERACTIONS AND THEIR
APPLICATIONS FOR CHEMICAL SENSOR DEVICES

By

Edwin B. Townsend IV

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCF OR OF PHILOSOPHY

Department of Chemistry

1995

ABSTRACT

FUNDAMENTAL STUDIES OF ANALYTE/FILM INTERACTIONS AND THEIR
APPLICATIONS FOR CHEMICAL SENSOR DEVICES

By

EDWIN B. TOWNSEND IV

Coated Surface Acoustic Wave (SAW) devices typically measure vapor
concentrations during vapor/film equilibrium. However, many practical sensor
applications require the analysis of pulsed sample streams. Thus, the research described
was motivated by the desire to deveIOp SAW chemical sensor systems to detect
environmentally important compounds utilizing pulsed sample introduction.

A semi-automated pulsed injection and data analysis system utilizing 200 MHz
SAW resonators has been constructed. A cell consisting of separate sample splitting
chamber, sample transfer lines, and sensor device areas allows temperature control over
the individual regions. Injections involving hexane, tn'chlorobenzene, and 1% (w/v) of
biphenyl in hexane show that all three temperatures play an important role in the
magnitude and time dependency of the SAW response.

The interactions of various classes of compounds with poly(isobutylene) [PIB]
coated SAW devices and quartz crystal microbalances (QCM) have been examined. For
the pulsed system, a new parameter (“pulse coefficient”) has been developed that is
analogous to the partition coefficient in a continuous flow system. Using the pulse
coefficients, Linear Solvation Energy Relationships (LSER) were calculated that provided
insight into analyte/film interactions. Pulse coefficients and LSER calculations compare
favorably with values reported for continuous flow SAW and GLC experiments.

PIB films coated on SAW devices and FI'IR/ATR internal reflection elements

were exposed to nitromethane, isooctane, and perchloroethylene vapors. A correlation

was found between a spectroscopic indication of vapor induced polymer swelling and the
responses of the polymer coated SAW devices. Sorption of perchloroethylene and
isooctane caused PIB films to swell significantly while sorption of nitromethane swelled
PIB films to a lesser degree. The observed polymer swelling occurs nearly
simultaneously with analyte sorption, and both IR and SAW results indicate that the
polymer swelling changes were reversible.

The responses of uncoated, PIB, poly(vinyl alcohol), and OV-275 coated SAW
devices to 4% (w/v) biphenyl or 500 ppm pentachlorobiphenyl (PCB) in acetone,
methanol, or hexane have been determined. SAW results obtained from the PIB coated
devices indicate that dispersion forces are most important for both biphenyl and PCB.
The nature of the solvent does not affect the sensor response to biphenyl. However, the
response to PCB in hexane is larger than for PCB in methanol and acetone, regardless of

polymer coating.

For Mom and Dad

who showed me the path
For Maria,

who has chosen to share her path with mine
For Sara

a wonderful addition to our path

iv

ACKNOWLEDGMENTS

There are many who played important roles during the course of my graduate
school experience. First and foremost was Prof. Jeff Ledford, who dispensed advice,
guidance, and encouragment along with whatever else he felt was needed. Believe it or
not, these words were appreciated. I learned more from you about the philosophy of
science and life than I ever thought existed. I would like to thank Prof. Chris Chang, who
allowed me the freedom to orchastrate the TAC program however I saw fit. Finally, Prof.
Stan Crouch also deserves special mention. I admire and appreciate your
professionalism. Finally, Profs. Blanchard, Harrison, and Galloway along with Drs.
Funkhauser and Atkinson have always been available when I needed them.

There were others in the department who made significant contributions to my
experience. To Beth Thomas and Cathy Allison; I don’t think I could have survived
without you. Everybody should be lucky enough to have you on their side. I can’t thank
you enough for all the help you’ve given me throughout the years. Dick Menke, my
master cell constructor. Very little of this work could have been acomplished without
your mechanical skills.

I have several friends who deserve special comment:

Kurt & Tracy K.- Thanks for everything. I’m sure that via Maria and Tracy, the
phone companies will always be profitable.

Laura “Admiral” P. - Thank you for your support and encouragement. You
deserve a lot from life, and I know you’ll get it. Thanks for showing me what a very
positive, upbeat outlook can do for a person.

Jim “Odie” R. - I’m not sure that anything I c0uld write would be printable. Lets
just say that you’re the most unique, but no less valued, entry on my list.

Stephen M. - Thanks for all of your computer advice. I hope everything works

out for you.

Kris K.- My favorite pool and bar partner. When are you going to get those
acknowledgements done?

Greg N. - I hope that you’re able to find someone who can help you during your
final months as much as you helped me. Despite our “ideological” differences, I very
much value your friendship. If only you couldn’t just go run 6 miles at 7 minute pace
whenever you wanted...

Jim N. - From you I have learned truth, honesty, and integrity. Mr. Moral fiber
and all that stuff. I also want to remind you about the job you’ve promised me.

I have made many other friends during my time in my graduate school. The
“original” group of cronies, Karen W-L, Jon W., Dave G., Gary 8., Jeff 6., Berly G.,
Jason R., Sue R., Mark 8., Mary 8., Eric H., and the others who started me off
(in)correctly. The second generation of friends included Per “Calbert” A., Mike T., Paul
“Gamemaster” P., Kathy S. (& family), Jeff R., Tim “Rabbit” M., Dana S., and all the
others. I’m sure that I’ve inadvertantly forgot a few and I apologize. I wish you all well
and thank you for your contributions to my sanity and well-being.

A dissertation could never be completed without the support of family. Even
though I’m not sure they always understood why some events were occuring, they were
supportive nonetheless. From my side, thanks go to Mom & Dad, Beef, Mary, and Drew,
and all of my grandparents. On Maria’s side, thanks go to Joanne, Michele, John, and
Michael, along with Joe, Amy, & the kids. John, if it weren’t for your inspiration, I don’t
know if I’d still be running. Thanks for the hobby (oh, excuse me, way of life).

I owe everyting to the two most important women in my life, Maria and Sara. The
contributions that you make to my life are indescribable. Maria, the only way I can thank

you is to make sure you get yours. And you will.

Now, Get to Work!!!

vi

This work has been performed by a contractor of the US. Government under
Contract DE-ACOS-840R21400. The work performed at Michigan State University was
performed under subcontract 19KMP607C to Martin Marietta Energy Systems Inc. A
US. Patent has been applied for to cover various aspects of this development effort.
Accordingly the US. Government retains a nonexclusive, royalty-free license to publish
or reproduce the published form of this contribution, or allow others to do so, for US.

Government purposes.

vii

TABLE OF CONTENTS

Page
List of Tables ................................................................................................................... vi
List of Figures .................................................................................................................. vii
Chapter 1. Introduction to Surface Acoustic Wave Sensors ........................................ 1
1.1 Focus of Research .................................................................................... 1
1.2 Acoustic Devices ..................................................................................... 3
1.2.1 Introduction to Acoustic Devices ............................................... 3
1.2.2 Quartz Crystal Microbalance ...................................................... 4
1.2.3 Surface Acoustic Wave Devices ................................................ 6
1.3 Factors Contributing to SAW Response .................................................. 12
1.3.1 Mass Loading ............................................................................. 12
1.3.2 Mechanical Properties ................................................................ 14
1.4 Coatings for SAW Devices ...................................................................... 18
1.4.1 Introduction ................................................................................ 18
1.4.2 Coordination Chemistry, Charge Transfer, Host/Guest
Mechanisms ................................................................................ 19
1.4.3 Solubility .................................................................................... 20
1.5 Solubility Models and the Interpretation of SAW Response ................... 21
1.5.1 Partitioning ................................................................................. 21
1.5.2 Boiling Point Model and Hildebrand Solubility Parameters ...... 23
1.5.3 Linear Solvation Energy Relationships ...................................... 24
1.5.4 Application of LSERs to SAW Response .................................. 27
1.6 References ................................................................................................ 29
Chapter 2. Effect of Silanization and Operating Temperature on the Response of
uncoated and Poly(isobutylene) Coated Surface Acoustic Wave
(SAW) Sensors ........................................................................................... 34
2.1 Abstract .................................................................................................... 34
2.2 Introduction .............................................................................................. 35
2.3 Experimental ............................................................................................ 37
2.4 Results and Discussion ............................................................................ 40
2.5 References ................................................................................................ 48

viii

Chapter 3.

3.1
3.2
3.3
3.4
3.5

Chapter 4.

4.1
4.2
4.3
4.4
4.5
4.6

Chapter 5.

5.1
5.2
5.3
5.4
5.5
5.6
5.7

Chapter 6.

6.1
6.2
6.3
6.4
6.5
6.6

Development and Performance of a Semi-automated 200 MHz
Surface Acoustic Wave Resonator Array-Based Sensor Test

Apparatus .................................................................................................... 51
Abstract .................................................................................................... 51
Introduction...... ........................................................................................ 5 1
Experimental............................................ ................................................ 53
Results and Discussion ............................................................................ 58
References ................................................................................................ 70

Study of Poly(isobutylene) Coated Piezoelectric Mass Sensors in a

Pulsed Injection System .............................................................................. 72
Abstract .................................................................................................... 73
Introduction .............................................................................................. 73
Experimental ............................................................................................ 77
Results and Discussion ............................................................................ 85
Conclusion ............................................................................................... 101
References ................................................................................................ 102

An FTIR/ATR Study of Sorbent Induced Polymer Swelling:

Application to the Study of Surface Acoustic Wave Sensors ..................... 105
Abstract .................................................................................................... 105
Introduction .............................................................................................. 105
Experimental ............................................................................................ 107
Results ...................................................................................................... 1 12
Discussion ................................................................................................ 125
Conclusion ............................................................................................... 133
References ................................................................................................ 135

Detection of Biphenyl and Pentachlorobiphenyl in Organic Solvents

Using Polymer Coated Surface Acoustic Wave Devices ........................... 137
Abstract .................................................................................................... 137
Introduction .............................................................................................. 138
Experimental ............................................................................................ 140
Results ...................................................................................................... 146
Discussion ................................................................................................ 153
References ................................................................................................ 162

ix

Table

Table 2.1.

Table 2.2.

Table 4.1.

Table 4.2.

Table 4.3.

Table 4.4.

Table 4.5.

Table 5.1.

Table 5.2.

Table 5.3.

List of Tables

Page
Analytes and their responses at 30°C .......................................................... 42
Analytes and their responses at 60°C .......................................................... 47
List of analytes, boiling points, and the solvation energy parameters
used in this work ......................................................................................... 78
Response ratios, pulse coefficients, and partition coefficients for the
blank and P13 coated QCM and SAW devices. ......................................... 91
Comparison of LogPSAw, LogPQCM, and LogKSAw with LogKSAw
and LogKGLc values from the literature. ..................................................... 96
Results of LSER fits using the QCM and SAW data and different
solvation parameters. .................................................................................. 98
Comparison of the contributions from each of the common
statistically significant interaction parameters to logKgAw and
logPSAw for poly(isobutylene). .................................................................... 100
AssigMents of the major IR bands for poly(isobutylene), isooctane,
perchloroethylene, and nitromethane. ......................................................... 118
Comparison of vapor induced polymer swelling estimates to the
change in integrated intensity of the C-H bending band of PIB (1235
cm“). ........................................................................................................... 121
List of analytes, boiling points, refractive indicies, and solvation
energy parameters. ...................................................................................... 130

Table 6.1. The average measured and calculated responses (Hz) to the solvent,
biphenyl/solvent mixtures, and biphenyl measured using uncoated
and coated SAW devices held at 30°C and 60°C ....................................... 148

Table 6.2. The average measured and calculated responses (Hz) to the solvent,
2,3,3’,4,4’-pentachlorobiphenyl(PCB)/solvent mixtures, and PCB
measured using uncoated and coated SAW devices held at 75°C .............. 152

Table 6.3. List of boiling points and solvation parameters for compounds used
in this work and selected related species. The LSER coefficients
available for the polymer films examined are also included ...................... 155

xi

Figure

Figure 1.1.

Figure 1.2.
Figure 1.3.

Figure 1.4.

Figure 1.5.

Figure 2.1.

Figure 3.1.

Figure 3.2.

Figure 3.3.

List of Figures

General overview of a SAW device and measurement scheme. An
alternating rf voltage, applied at one transducer, strains the
piezoelectric substrate and generates the SAW. The wave travels
along the surface of the crystal and any surface perturbations will
affect the velocity and amplitude of the wave. The wave is
subsequently detected at the other side of the device by the second
set of IDTs. Any change in wave velocity is indirectly measured as a

frequency change using an oscillator circuit ...........................................
The dual delay line SAW device ............................................................

The SAW resonator ................................................................................

An illustration of the contribution of mass and viscoelastic effects to

SAW response ........................................................................................

The partitioning phenomena between molecules in the vapor phase

and the film on the SAW sensor substrate ..............................................

Schematic diagram of sensor test apparatus ...........................................

View of the cell bottom showing the platform array device inserted.
Individually mounted SAW devices are mounted perpendicular to

the devices on the platform .....................................................................

Schematic diagram of the complete sensor test apparatus ......................

Output created by the LabView software written to process 6

channels of SAW data ............................................................................

xii

Page

....7

....10

....ll

....15

....22

....39

....55

....56

....59

Figure 3.4. The effect of sensor cell temperature on the response of an uncoated
device to injections of 3 1L of hexane followed by 3 uL of a 1%
(w/v) mixture of biphenyl in hexane. Sensor cell temperatures = (a)
60°C, (b) 45°C, and (c) 30°C. Splitting chamber temperature =
250°C. Transfer line temperature = 50°C. Carrier gas flow rate = 80
cc/min ......................................................................................................... 61

Figure 3.5. Effect of various combinations of splitting chamber/transfer line
temperatures on the response of an uncoated device to injections of 3
uL of hexane followed by 3 uL of a 1% (w/v) mixture of biphenyl in
hexane. Splitting chamber/transfer line temperatures = (a) 100/38°C,
(b) 200/49°C, (c) 300/63°C, and (d) 400/77°C. Sensor cell
temperature = 30°C. Carrier gas flow rate = 80 ccan ............................. 63

Figure 3.6. The effect of transfer line temperature on the uncoated device
response to 0.1 [1L injections of trichlorobenzene. Transfer line
temperatures = (a) 75°C and (b)150°C. Splitting chamber
temperature = 400°C. Sensor cell temperature = 30°C. Carrier gas
flow rate = 80 cc/min .................................................................................. 66

Figure 3.7. Effect of carrier gas flow rate on the uncoated device response to
injections of 3 [LL of hexane followed by 3 uL of a 1% (w/v) mixture
of biphenyl in hexane. Carrier gas flow rate = (a) 20 cc/min, (b) 50
cc/min, and (c) 100 cc/min. Splitting chamber temperature = 250°C.
Transfer line temperature = 50°C. Sensor cell temperature = 30°C .......... 67

Figure 3.8. Calibration curves for hexane injected on a blank ( ) and
poly(isobutylene) coated (A) device ........................................................... 69

Figure 4.1. Schematic diagram of the SAW sensor test cell. The sample is
introduced as a pulse and then split equally into 6 streams. Each of
the streams is delivered directly onto the SAW surface. (a) is one of
the six delivery arms, (b) is the mixing chamber, and (c) is one of the
6 resonators mounted on a gold stage ......................................................... 80

Figure 4.2. Schematic diagram of the QCM sensor test cell. The QCM is
positioned in the center of the cell, and the sample is introduced as a
pulse and then split equally into 2 streams. The streams are
delivered to the left and the right of the device .......................................... 83

xiii

Figure 4.3. Variation of uncoated QCM device response (Hz/umole) versus
analyte boiling point ................................................................................... 86

Figure 4.4. Variation of uncoated SAW device response (Hz/nanomole) versus
analyte boiling point ................................................................................... 88

Figure 4.5. Variation in the response of poly(isobutylene) coated QCM device
response (Hz/umole) as a function of analyte boiling point ....................... 90

Figure 4.6. Variation in the response of poly(isobutylene) coated SAW device
(Hz/nanomole) as a function of analyte boiling point for the pulse
injection (O) and continuous flow (+) methods. The response for
nonane (78 Hz/nanomole) has not been included in this plot (see
text) ............................................................................................................. 92

Figure 5.1. Normalized PIB coated SAW device responses to (a) nitromethane,
(b) isooctane, and (c) perchloroethylene using continuous flow and
pulsed injection sample introduction. Equilibrium and pulse
responses have been normalized to analyte concentrations of 0.02
g/L and 0.01 g/L, respectively .................................................................... 113

Figure 5.2. Variation in the normalized equilibrium frequency shifts versus time
following t”2 for a PIB coated SAW device exposed to nitromethane
(A), perchloroethylene ( ), and isooctane (O). (a) Rising and (b)
Falling edges ............................................................................................... 114

Figure 5 .3. IR spectra of (a) nitromethane, (b) perchloroethylene, (c) isooctane,
and (d) poly(isobutylene) ............................................................................ 117

Figure 5.4. (a) IR spectra of a PIB film before exposure to nitromethane (solid
line) and at maximum analyte intensity (dashed line) using pulsed
sample introduction. (b) IR spectra before exposure to nitromethane
(solid line) and during vapor/film equilibrium (dashed line) using
continuous flow sample introduction ......................................................... 119

Figure 5.5. (a) IR spectra of a PIB film before exposure to isooctane (solid line)
and at maximum analyte intensity (dashed line) using pulsed sample
introduction. (b) IR spectra before exposure to isooctane (solid line)
and during vapor/film equilibrium (dashed line) using continuous
flow sample introduction ............................................................................ 122

xiv

Figure 5.6. (a) IR spectra of a PIB film before exposure to perchloroethylene
(solid line) and at maximum analyte intensity (dashed line) using
pulsed sample introduction. (b) IR spectra before exposure to
perchloroethylene (solid line) and during vapor/film equilibrium
(dashed line) using continuous flow sample introduction .......................... 123

Figure 5.7. IR spectra obtained as a function of time for a PIB film exposed to a
continuous vapor stream of perchloroethylene ........................................... 124

Figure 6.1. Overview of the sensor test apparatus ........................................................ 142

Figure 6.2. Typical response profiles obtained for the solvent and the
biphenyl/solvent mixtures using poly(isobutylene), poly(vinyl
alcohol), and OV—275 coated devices held at 30°C .................................... 147

Figure 6.3. Typical response profiles obtained for the solvent and the
biphenyl/solvent mixtures using poly(isobutylene), poly(vinyl
alcohol), and OV-275 coated devices held at 60°C .................................... 149

Figure 6.4. Typical response profiles obtained for the solvent and the
PCB/solvent mixtures using poly(isobutylene), poly(vinyl alcohol),
and OV-275 coated devices held at 75°C ................................................... 151

XV

Chapter 1

Introduction to Surface Acoustic Wave Chemical Sensors

1.1 Focus of Research

Surface acoustic wave (SAW) devices belong to a general class of acoustic
microsensors that are sensitive to surface mass changes. The potential of these devices is
best illustrated by noting that a SAW device operating at 200 MHz exhibits a, 904 Hz
frequency shift when perturbed by 1 ng of surface mass. The small size and extreme
sensitivity of these devices has motivated researchers to investigate their potential use as
chemical sensors. The chemical selectivity of SAW devices is enhanced through the
application of a coating to the device surface. Using a good coating, Ballantine and
Wohltjen [l] have suggested that it is possible to detect vapors at the 10-100 ppb
concentration level, with selectivity of 1000:1 or more over commonly encountered
interferences, with a linear dynamic range of 34 orders of magnitude .

Typically, the coated device is exposed to a continuous stream of vapor. We are
exploring an alternate method of sample introduction, the pulsed method, where the
analyte of interest reaches the SAW device as a plug, interacts with the coated surface,
desorbs, and is rapidly swept out of a test cell. This importance of this style of sample
introduction technique in relation to mass sensors has been discussed previously [2-10].
For example, J anghorbani and Freund have successfully used a quartz crystal

microbalance (QCM) as a GC detector [8].

2

Edmonds and West have used pulsed sample introduction in combination with QCMs
coated with gas chromatographic phases to detect various organic species [9]. Grate and
Abraham have pointed out that this type of sample introduction can be used in
conjunction with laser or thermal desorption [2]. Grate et al. [10] later used thermal
desorption of analyte in preconcentrator tubes to deliver pulses of gas to an array of SAW
devices.

The research described in this dissertation was motivated by the desire to develop
SAW chemical sensors that detect environmentally important compounds utilizing pulsed
sample introduction. Chapters 2-6 detail knowledge gained regarding the different
aspects of sensor operation in the pulsed sampling environment. Chapter 2 explores the
effects of temperature and surface Silanization on uncoated and poly(isobutylene) coated
mass sensor responses. Chapter 3 details a new pulsed sensor system and a SAW cell
used to illustrate the effects of various operating parameters and to provide new insight
into phenomena occurring in pulsed sampling sensor systems. Chapter 4 demonstrates
that data obtained from poly(isobutylene) coated acoustic devices in pulsed injection
systems can be used in combination with Linear Solvation Energy Relationships to obtain
molecular level insight into analyte/film interactions. Chapter 5 demonstrates that
FTIR/ATR spectroscopy can be used to probe phenomena that occur during the sorption
of an analyte into a film and how these processes can influence SAW response. Chapter

6 demonstrates the influence of solvent matrix on SAW device response.

1.2 Acoustic Devices

1.2.1. Introduction to Acoustic Devices

Piezoelectricity was first observed in 1880 by Pierre and Jacque Curie [11], who
found that, for particular compounds, the application of a mechanical force resulted in the
generation of an electrostatic voltage. They also found that the application of an electric
field to these compounds induced crystal deformations. Since these initial findings, a
large number of materials have been found that exhibit the piezoelectric effect. Quartz is
the piezoelectric material used most often for chemical sensing applications because of its
chemical inertness and the insensitivity of its wave properties to changes in temperature.

Piezoelectric materials may support a wide range of acoustic waves. Waves can
travel either through the bulk or along the surface of the material, or by reflections along
multiple surfaces. Transverse or shear waves have particle displacements that are normal
to the direction of wave propagation, and can be subsequently polarized so that the
displacements are parallel or normal to the surface of the sensor. Surface acoustic (or
Rayleigh) waves are confined to an area at the surface of the crystal that is roughly one
acoustic wavelength thick. Surface particles affected by these Rayleigh waves move in an
elliptical path, with both surface-parallel and surface normal components.

Acoustic microsensors normally consist of a piezoelectric material in combination
with some type of metal transducer. The transducers launch acoustic waves into the

crystal at high frequencies. The types of waves generated depend on the crystal

4
orientation, thickness, and the geometry of the metal transducers. There are four basic
kinds of acoustic devices, so named because of the types of waves generated and
detected. They are the quartz crystal microbalance, the surface acoustic wave, flexural
plate wave, and acoustic plate mode devices. Because this dissertation includes only
quartz crystal microbalances and surface acoustic wave devices, the discussion is limited

to those two devices.
1.2.2. Quartz Crystal Microbalance

The first type of microbalance to be developed using the piezoelectric properties
of materials was the quartz crystal microbalance (QCM). The dominant acoustic wave
generated and monitored in a QCM is the bulk transverse wave, which travels
perpendicular to the sensing surface. The thickness of the QCM deterrrrines the
wavelengths of the fundamental and the harmonic resonances. Typical resonant
frequencies for the fundamental mode are between 5 and 10 MHz.

The use of QCMs as detectors was first demonstrated by Sauerbrey [12], who
derived an expression that related a change in thickness of the crystal (caused by an
increase in mass at the surface of the crystal) to shifts in frequency according to equation

(1), where pq and vq are the density and shear wave velocity of quartz , and fq is the

fundamental frequency of the quartz oscillator.

Af=“2fq2 Am/(ptIVQ) (1)

5

This expression assumes that the material deposited on the QCM surface only increases
the thickness of the crystal. The consequences of changes in density, viscosity or shear
modulus of the added layer are ignored. However, the utility of this expression was
illustrated by adding mass to the crystal using vacuum metal evaporation [12] and later
verified by others [13-15].

Kanazawa has derived the most recent expression that describes the frequency
dependence of the QCM on both the mass and physical properties of an overlayer [16].
The relationship is shown in equation (2), and was developed to describe the response of

a QCM in contact with liquids, but is applicable to any viscous overlayer,

3 l
Af =" o/2(PLflL/“PH)/2 (2)

where f0 is the frequency of the uncoated crystal, TIL is the absolute viscosity of the
liquid, p. is the shear modulus, p is the density of quartz, and pL is the overlayer density.

This equation has been supported using results obtained from 5 and 9 MHz QCMs [l7].
QCMs have been used as GC detectors [18], for monitoring trace amounts of
atmospheric gases [19], or for polymer film characterization [20]. Many reviews have
been published that discuss coating materials and the use of QCMs to detect analytes at
low concentrations [19-27]. Sensors designed using guest/host chemistry have also been
reported [28-30]. Chemometric techniques have been used to analyze the responses

obtained from QCM arrays [31,32].

1.2.3. Surface Acoustic Wave Devices

Surface waves were first described by Lord Rayleigh [33] and are often referred to
as Rayleigh waves. Devices utilizing these types of waves were created [34] after White
and Voltrner developed the interdigital transducer (IDT) [35]. Wohltjen and Dessey first
introduced the concept of SAW chemical sensors in 1979, and subsequently described
operational concerns and demonstrated possible applications [34,36-39]. The concepts of
coated SAWs as detectors for gas chromatography, to probe film properties, and for gas
phase analyte detection were also introduced.

Typically, a SAW device is constructed using ST-cut quartz and two sets of
interdigital transducers (IDTs) formed by photolithographic patterning of a thin metal
film (see Figure 1.1). When an alternating potential is applied to one transducer, an
alternating strain is generated in the substrate due to its piezoelectric nature. If this
alternating potential is at the appropriate frequency, the strain launches a Rayleigh (SAW)
wave which travels along the surface of the substrate. The SAW has most of its energy
confined within one wavelength of the surface and thus, is strongly affected by any
material in contact with the surface. Consequently, the properties of the material have an
effect on the velocity and amplitude of the SAW wave.

SAW devices normally operate in the 3-300 MHz range. The actual resonant
frequency depends upon the spacing between the IDTs, as illustrated by equation (3)

where f0 is the resonant frequency, v is the velocity of the SAW wave, and d is the

distance between the transducers.

Surface Acoustic Wave

f4:— Polyrner Film /—=|__—

 

 

Piezoelectric

Substrate
Transducer

 

 

 

LZ

 

 

Frequency
Counter

Figure 1.1. General overview of a SAW device and measurement scheme. An
alternating rf voltage, applied at one transducer, strains the piezoelectric
substrate and generates the SAW. The wave travels along the surface of
the crystal and any surface perturbations will affect the velocity and
amplitude of the wave. The wave is subsequently detected at the other
side of the device by the second set of IDTs. Any change in wave velocity
is indirectly measured as a frequency change using an oscillator circuit.

f0 = (3)

31
d

This expression illustrates that the closer the spacing between the IDTs, the higher the
resonant device frequency. As will be discussed in Section 1.3.1, the theoretical
sensitivity of the device increases with the square of the operating frequency. Thus, it
appears that increased sensitivity can be achieved with smaller devices.

Velocity and amplitude measurements can be made using a signal generator and a
vector voltmeter [40]. _ An rf voltage at a fixed frequency matched to the resonant
frequency of the device is supplied to the input transducer by a signal generator. A vector
voltmeter monitors the change in insertion loss and any phase shift between input and
output ports. Attenuation is evaluated from the insertion loss while the change in wave
velocity is obtained from the phase change. The frequency shift can be calculated from
the change in wave velocity.

However, a simpler and more common measurement method involves the indirect
determination of acoustic velocities by using the SAW device as the frequency control
element of an oscillator circuit. In this case the SAW device is used as an amplifier
feedback path. Fractional changes in wave velocity lead to equivalent changes in
oscillation frequency. Thus, monitoring the oscillation frequency provides a precise,
indirect measurement of acoustic wave velocities. Devices based on these circuits are

relatively simple and inexpensive to construct.

9

There are two major surface acoustic wave device configurations. One is the dual
delay line device (illustrated in Figure 1.2) and the other is termed a SAW resonator
(illustrated in Figure 1.3). The dual delay line device is made so that the IDTs are placed
at opposite ends of the surface. It is called a delay line device because the electrical
signal at the input transducer reaches the output transducer 5 orders of magnitude slower
as an acoustic wave than it would as an electromagnetic signal [23]. One disadvantage of
the dual delay line device is that the surface wave is launched in both directions from the
transducer, thus half the wave is lost.

SAW resonators place transducers inside reflector arrays. These reflector arrays
reflect the SAW back towards the center of the device from the edges of the crystal to
create a standing wave. The standing wave effectively traps the acoustic energy through
constructive interference of the reflected waves. Because the acoustic energy is not lost at
the ends of the crystal, the resonator (as opposed to the delay line) is considered a high Q
device [41,42]. The configuration of the reflector arrays permits only 1 standing wave to
exist at a frequency corresponding to the spacing of the transducers. Thus, efficient
signal transmission (and oscillation) is restricted to the frequency of the standing wave.
This decreases the amount of noise compared to the delay line system. Consequently, the
most significant advantage of a resonator device is the improved mass detection limits
achieved by increasing the signal/noise ratio. Other factors affecting the sensitivity of

these devices will be discussed in Section 1.3.1.

10

 

1.5 cm

 

 

 

Figure 1.2. The dual delay line SAW device.

Reflector Arrays
IDTs

‘ elocity Correctors

F

    

. F

0.17 cm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—

 

 

1.0 cm

A

Figure 1.3. The SAW resonator.

12

1.3 Factors Contributing to SAW Response

1.3.] Mass Loading

In 1978, Tiersten and Sinha [43] derived an equation relating SAW velocity to the
properties of acoustically thin, elastic films. Wohltjen [39] later applied the Tiersten
expression as shown in equation (4) to interpret polymer coated SAW responses to vapor
sorption. In this interpretation, Af is the frequency shift (hertz) observed following

sorption of gas phase molecules into the film, fo is the fundamental frequency of the
device (hertz), k, and k2 are material constants of the ST cut quartz substrate (-8.7 x 10'8
and -3.9 x 10'8 mzs/kg) [44], h is the film thickness (meters), p is the film density

(kg/m3), u is the shear modulus of the film material (N/m2), l is the Lamé constant, and

V, is the Rayleigh wave velocity in the substrate (3300 m/s for quartz).

 

2 2 4].l{l||.l]
Af—(k +k )fh—kf (4)
l 2 Op 20 V2U+2“

The first term describes the frequency shift due to mass loading and the second term
represents the influence of film mechanical properties. If the shear modulus of the
coating is small compared to the square of the wave velocity (4WV%< 100), then the
second term is negligible and equation (5) simplifies to an expression resembling the

Sauerbrey equation (1):

13

Af = (k1+k2)f3hp (5)

This equation assumes that the film is an isotropic, nonconducting,
nonpiezoelectric polymer film held above its glass transition temperature with a thickness
less than one percent of the acoustic wavelength. Ideally, since hp is the mass per unit

area, Af is proportional to the change in surface mass.

Expression (5) clearly shows that the response of the SAW device increases with
the square of the operating frequency. This offers exciting opportunities for
rrricrosensors, because the operating frequency increases as the spacing between the IDTs
decrease. Consequently, smaller devices should be more sensitive. However, this is not
strictly true, because as device frequency increases, sensing area decreases. Grate and
Klusty [42] have shown that as the sensing area decreases, the thickness of the applied
polymer coating must also decrease. Consequently the amount of vapor/unit area that can
be collected also decreases. Thus, increasing device frequency is offset by decreased
sensing area when considering SAW sensitivities.

The fact that device sensitivity increases with the square of the operating
frequency also applies to the QCM. However, the operating frequency of the QCM
increases with decreasing crystal thickness. Thus, there is a practical, physical limitation
for the QCM because as the device is made thinner, it becomes much more difficult to
manipulate. Because the frequency of the SAW device depends on the spacing of the

IDTs rather than the crystal thickness, much higher frequencies can be obtained without

l4
sacrificing ruggedness. For this reason, high frequency SAW devices are normally

chosen over QCMs for gas phase sensor applications.
1.3.2 Mechanical Properties

Consideration of the mechanical contributions to SAW response is best separated
into two cases. The first is the frequency decrease observed when a polymer film is
applied to the surface of a bare SAW, and the second is the frequency decrease obtained
when a vapor is sorbed by the polymer. These two instances are shown in Figure 1.4.

The inherent sensitivities of acoustic wave devices to the physical properties of
viscoelastic thin films were noted in early models for SAW sensors based on perturbation
analysis [39,45]. These models predicted that the resonant frequency of a SAW device
would be altered by both the mass and the shear modulus when a thin, nonconducting,
isotropic film was applied to its surface.

Using Langmuir-Blodgett layers of known mass/area, Grate and Klusty [42] have
demonstrated that when a film is applied to a bare crystal, the mass of the coating is the
primary source of the frequency decrease. In theory, the modulus effect is only 10-15%
of the mass effect if the modulus of the film is 109 N/m2 (typical of glass polymers) at a
given SAW device operating frequency. Grate and Klusty suggest that if the modulus is
106 N/mz, then the modulus contribution to SAW response should be negligible [42,46].
Rubbery polymers typically have a film modulus of 106 N/mz, but that value can increase
to approximately 109 N/m2 at the high frequencies employed by typical SAW devices

[47].

 

 

 

 

15

 

 

 

 

Uncoated SAW
’
Mass Polymer Film
ms
4::
E
m
>.
8
93 ' SAW with Polymer
a
5
LL.
Modulus
V
Modulus Vapor Sorption
Afv
6 Mass
1 t__.
’
Figure 1.4. An illustration of the contribution of mass and viscoelastic effects to SAW

response.

16

The sorption of vapor into a polymer is known to expand the polymer volume by
the free volume of the sorbed molecules [48]. Effectively, vapor sorption acts to dilute
the polymer chain segments with increasing volume. Increasing the polymer volume
decreases the chain-chain interactions because the polymer chain segments are farther
apart and the intermolecular interactions are decreased. Thus, it is reasonable to conclude
that vapor sorption changes polymer modulus by increasing the free volume of the
polymer.

In the case of vapor sorption on a coated SAW device, it was initially presumed
that for a polymer well above its glass transition temperature (T g), the second term in
equation 3 was negligible and that the frequency response was dominated by the mass
loading term. However, recent investigations have provided substantial evidence that
mechanical effects are not negligible and may, in fact, represent a significant contribution
to the frequency response to vapors [46,49].

Martin et al. [50] found that changes in film viscoelastic properties affect SAW
response, and proposed that the response could be described by the Maxwell model. In
this model, the resistance of the polymer chains moving against each other is represented
by the shear viscosity, and the polymer chains reorienting and reorganizing is the shear
stiffness. In the case of vapor sorption, the vapor diffuses into the free volume of the
polymer, causing the shear viscosity to decrease and a change in the shear stiffness. This
changes the elasticity of the polymer, which alters the SAW response.

Grate et al. [46] began to assemble a more complete explanation of the effects of

viscoelastic thin films on acoustic wave devices while investigating differences between

17

partition coefficients determined from GLC measurements (Kcm) and partition
coefficients from SAW measurements (KSAw) [51]. These authors reported that KSAw
was approximately 4 times larger than Kocc for six analytes sorbed into fluoropolyol,
poly(epichlorohydrin), and poly(isobutylene) polymers. In an effort to explain this
finding, they proposed that changes in thin film mechanical properties resulting from
vapor induced polymer swelling makes a significant contribution to the responses of
polymer-coated SAW devices [46]. These researchers assumed that the change in free
volume of the polymer caused by vapor sorption would be similar to the change in free
volume induced by thermal expansion. This is reasonable because new volume created
by polymer thermal expansion above the glass transition temperature is characterized as
free volume [52].

Thus, using known thermal expansion rates for. polymer films and the
experimentally determined effects of thin film expansion on SAW device resonant
frequencies, the volume effect was estimated to be 10-20 kHz per percent of volume
increase [46]. Volume increases caused by vapor sorption were estimated from test vapor
concentrations, Konc values, and liquid densities of vapors. These estimates suggested
polymer swelling of 0.3-3% by certain vapors. Thus, estimated swelling can be
multiplied by a volume effect calibration to calculate the frequency decreases that occur
in response to vapor induced swelling. It was reported that polymer coated sensor
responses can be modeled (equation 6) by summing the response expected from mass

loading and the response expected from swelling-induced modulus changes.

18

AfsCVKGLC ]+[AfsCVKGLC )(ASAW ) (6)
ps PL (1

Af=[

All of the terms are as defined before, and Cv is the vapor concentration, pS 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, Otis the polymer thermal expansion coefficient, and AS AW is the

frequency change of a polymer coated SAW due to thermally induced swelling.

Martin et al. [49] have recently interpreted the dynamics and response of polymer-
coated SAW devices in terms of film viscoelastic properties and film resonance effects.
They have proposed that there are two distinct regimes of SAW response. The first, when
the film is considered to be acoustically thin, and the second, when the film is considered
to be acoustically thick. Using the information presented by Martin et al. [49], values for
polymer modulus caused by either vapor sorption or a temperature change may now be
determined, and the resultant velocity and attenuation changes can be modeled using the
appropriate equations. In this dissertation, all films are calculated to be acoustically thin.
This suggests that changes in wave velocity (frequency) are proportional to changes in the

analyte concentration in the film.
1.4 Coatings for SAW Devices

1.4.] Introduction

There are several review articles that discuss coatings and the selectivity of

piezoelectric chemical sensors [53-56]. Interaction mechanisms for chemically selective

19
layers have also been reviewed extensively [2,21,24,27,57-63]. Films typically interact
with vapor through coordination chemistry, charge transfer, host/guest mechanisms, or

solubility interactions.
1.4.2 Coordination Chemistry, ChargeTransfer, Host/Guest Mechanisms

A compromise between selectivity and reversibility can be found in coordination
chemistry or charge-transfer complex formation. The potential use of coordination
interactions has been discussed by Nieuwenhuizen et al. [58,64]. Selectivity in
coordination and charge transfer chemistry can also be influenced by the choice of the
central metal ion or the ligands from both an electronic and a steric point of view. For
example, these interactions have been used to detect N 02 using metal phthalocyanines
[23,65]. Zellers et al. [66—68] have reported SAW sensors which respond to select olefrns
by coordination to Pt complexes. Lewis acid/Lewis base interactions have been utilized
in the detection of SO; (Lewis acid) using basic amines [21].

In an alternate approach, a chemical reaction between the vapor of interest and the
coating leads to selectivity. For example, Snow and Wohltjen have developed an
irreversible cyclopentadiene sensor [69]. Zellers et al. [70] have demonstrated styrene
and vinyl acetate sensors. These sensors are not intrinsically reversible but can be made
so by regenerating the device using a reagent [71].

A special case of complexation for potential use in sensors involves cage or
inclusion compounds. Dickert has been a pioneer in the application of cage molecules

and guest/host interactions for the construction of piezoelectric based chemical sensors

20
[72-79]. Cyclodextrins [80] and calixarenes [81] will complex hydrophobic compounds
of appropriate size within their hydrophobic cavities. The cavity provides a site where
solubility interactions occur in a particular steric arrangement. For example, modified

cyclodextrins have been applied to QCM devices for the detection of benzene [82].

1.4.3 Solubility

Typically, polymeric materials have been used in organic vapor sensing
applications because vapor sorption in rubbery polymers is rapid and reversible; they tend
to form adherent thin films, and some selectivity can be achieved by varying the chemical
structure of the film [24]. Vapor selectivity is determined only by slight differences in the
partition coefficients between the polymer coating and the various vapors. To a great
extent, the partition coefficient is influenced by the solubility properties of the coating
and the vapor of interest. Grate and Abraham [2] have extensively discussed the
importance of solubility for vapor/polymer interactions.

In practice, it is not possible to make a material that will interact via a single
solubility property. For example, all organic materials will undergo significant dispersion
interactions. Because solubility interactions are not completely selective, the use of
sensor arrays with pattern recognition has been investigated as a means of improving
selectivity [31,32,83,84].

There has been interest in using chromatographic stationary phases as coatings for
SAW devices. McCallum has summarized work involving the interaction of seven

analytes with nine chromatographic stationary phases [60]. Several researchers have used

21
Carbowax phases as coatings for SAW devices [19,21,63,85-88]. Methyl- and phenyl

siloxanes have also been exarrrined [89-91].

1.5 Solubility Models and the Interpretation of SAW Response

1.5.1 Partitioning

Traditionally, the partition coefficient (K) is used as a measure of the amount of
analyte that sorbs into a film. The similarity between the partitioning of analytes into
stationary phases in gas-liquid chromatography and the partitioning between analytes and
sensor phases (see Figure 1.5) was recognized early for quartz crystal microbalances
[18,92,93] and was more recently applied to the study of SAWs [46,51]. The partition
coefficient, K, is defined in equation (7).

Cs

K:—
Cv

(7)

CS represents the amount of analyte in the stationary phase and CV is the concentration of

the analyte in the vapor phase.

22

Sensor
Substrate

 

Figure 1.5. The partitioning phenomena between molecules in the vapor phase and the
film on the SAW sensor substrate.

23
1.5.2 Boiling Point Model and Hildebrand Solubility Parameters

The simplest model that attempts to account for the interactions between the
vapor and the film is the boiling point model [94]. This model relates the partition

coefficient to the analyte saturation vapor pressure as given in equation (8).

K— PlRT

- (3)
Y 2M 1P2

K is the partition coefficient, p1 is the density of the stationary phase, M1 is the
molecular weight of the stationary phase, 72 is the vapor activity coefficient, and p2 is the

analyte saturation vapor pressure. Expression (8) was first used to describe the retention
of vapors in GLC [94]. Using an approximation arising from Trouton’s rule and the
Clausius-Clapeyron equation, the log of the partition coefficient can be shown to be
directly proportional to the analyte boiling point. The weakness of this model is that it
fails to account for deviations from ideality.

Deviations from the ideal vapor-polymer solution behavior described above are
observed in most real systems. A model using Hildebrand solubility parameters was
developed that intended to take into account non-ideal behavior. The regular solution
theory developed by Hildebrand assumes that deviations from ideality arise from the
nonzero heat of mixing, which can be related to the infinite-dilution activity coefficient
and to the solubility parameters of the polymer, 5; and solute vapor 62 by equation (9),

where V2 is the vapor molar volume.

24

2

10 = — 9
g“ 2.303RT 2.303RT U

 

The solubility parameter of a substance is defined as the square root of the molar
vaporization energy per unit volume and, as such, is a measure of cohesive energy (or
enthalpy). Since solubility parameter values generally increase with increasing polar
constituents in a molecule, solubility parameters are commonly used as indices of overall
polarity. In general, the more similar the solubility parameter values of two materials, the
lower their heat of mixing and the greater their mutual solubility. The solubility

parameter model predicts the following relationship:

(10)

 

 

2
91’” _V2(51‘52)
Mlpz 2.303RT

log K = log[

which indicates that for situations where 81 and 82 differ appreciably (y > 1), the partition
coefficient will be reduced relative to the ideal case.

This model fails when specific oriented chemical interactions are the dominant
mode of intermolecular interaction. In addition, regular solution theory is most
applicable for nonpolar or slightly polar materials. Thus, this model still provides only a

limited picture of vapor/film interactions.
1.5.3 Linear Salvation Energy Relationships

Linear solvation energy relationships are a combination of Linear Free Energy

Relationships (LFER) [95,96], which attempt to determine the correlation between

25

structure and reactivity, and Quantitative Structure-Activity Relationships (QSAR),
whose premise is that microscopic structural features of a chemical compound can be
related to its macrosc0pic properties [97]. The first application of LFER examined the
effect of aromatic substitution on the acidity of a number of benzoic acids [98]. Current
applications of LFERs have been reviewed recently [99,100]. In general, QSAR analysis
involves a regression equation that correlates selected microscopic features of a set of
chemical compounds with a specific property. QSAR is now used predominantly to
predict and understand biological, pharmacological, and physical properties [101-105].

Early examples of LSER explained solvent effects on various free energy based
properties [106-108]. Two solvation equations (Equations 11 and 12) have been
developed [109,110] that have been applied to various processes that involve transfer of a
series of solutes from the gas phase to a condensed phase or transfer of a series of solutes

from one condensed phase to another.

log SP = c + a2a¥+b2fl¥ +llong6+rR2+s1t£I (11)

log SP = c + azaf’ +1323? +vi +rRz+s1c5i (12)

SP is a standard property for a series of solutes in a given phase. For example, SP might

be the partition coefficient for a number of solutes on a given stationary phase, or a
biological property for a series of solutes. 2a? , 2B5! , long, R2, 1t? , and VK are

called molecular descriptors, and are combined with the standard property in a multiple

26
linear regression to determine the constants. These constants can be used to characterize

the solvent phase.
2a? and 2B? are measures of the solute hydrogen-bond acidity and basicity,

and thus the constant a would be a measure of the solvent hydrogen bond basicity and the
constant b would be a measure of the solvent hydrogen bond acidity. The values for these
descriptors are obtained from hydrogen-bond complexation constants of acids by
reference bases (such a pyridine) in a reference solvent (such as carbon tetrachloride) or
complexation of bases by reference acids (such as 4-fluorophenol) in carbon tetrachloride
[111,112], and more recently from chromatographic or partition measurements [113].

logL16 is a descriptor derived from the solute gas-liquid partition coefficient on
hexadecane at 298K [114]. They have been obtained by gas chromatography and several
hundred values are now available [111-114]. V, is McGowan characteristic volume and
can be calculated for any solute based on its chemical structure [115]. Both of these
descriptors are considered to be a measure of dispersion interactions, and represent the
energy of the endoergic creation of a cavity within the solvent, the incorporation of the
solute in the solvent, and the reorganization of the solvent around the solute and the
associated interactions.

R2 is an excess molar refraction term thatis determined using the refractive index
of a compound, i.e., the molar refraction in excess of the molar refraction (MR) of an

alkane of the same characteristic volume (V,) [116].

RS = MR, - MR, for an alkane of the same V,

27

By subtracting the molar refraction for an alkane of the same characteristic volume, the
dispersive component of molar refraction (already accounted for by logLK’) is removed.
This descriptor represents the tendency of a solute to interact through 1! or n electrons,
and thus r is a measure of a solvents’ ability to stabilize 1: or n electron interactions.

Rf is a measure of the compounds dipolarity/polarizability. It is obtained
experimentally from gas-liquid chromatographic measurements [109,110]. This
parameter is important with regard to the dipole/dipole and dipole/induced-dipole

interactions. For non-protonic, aliphatic solutes with a single dominant dipole, 1:? values

are approximately proportional to molecular dipole moments.
1.5.4 Application of LSERs to SAW Response

Grate and Abraham proposed [2,46] using KSAw in combination with Linear
Solvation Energy Relationships to help elucidate the chemical interactions occurring

between a vapor and a film. This expression is given in equation (14).
log KSAW = c + a2a§+bzfig+llong6+rR2+srt£1 (14)

They also discussed the prediction of SAW responses through LSER expressions
that were obtained based on GLC data. It was hypothesized that a films’ performance as
a GLC phase should be an accurate predictor of the film as a SAW coating. They detailed
this approach for SAW devices and also suggested possible coatings that might interact

through specific terms.

28

It is important to realize that models based on solubility and partitioning are
simply logical places to start when attempting to understand vapor/film interactions. All
of these models calculate partition coefficients assuming SAW device response to be
strictly mass loading. Grate et al. [46] have shown that vapor induced changes in film
properties are important. However, this was accomplished using only 6 analytes and 3
films. Thus, the effect of vapor induced changes in film mechanical properties for
individual analytes is not well defined.

However, Patrash ,and Zellers [117] have recently examined OV-275
(poly[bis(cyanoallyl)-siloxane]), OV-25 ([25% methyl] poly(methylphenylsiloxane)),
poly(isobutylene), and poly(phenyl ether) as SAW coatings and exposed each to 37
analytes and then determined the LSER coefficients for the films. They found that
interaction predictions from GLC data were reasonable approximations for the SAW

device.

1.6

5.1.5;er

99°99

10.

11.
12.
13.
14.
15.

16.
17.
18.
19.
20.
21.
22.
23.
24.
25.

29

References

Ballantine, D. 8., Jr.; Wohltjen, H. Anal. Chem. 1989, 61, 704A.
Grate, J. W.; Abraham, M. H. Sens. & Act. B. 1991, 3, 85.
Suleiman, A; Guilbault, G. G. Anal. Chim. Acta 1984, 162, 97.
Thompson, M.; Stone, D. C. Anal. Chem. 1990, 62, 1895.

Hecki, W. M.; Marassi, F. M.; Kallury, K. M. R.; Stone, D. C.; Thompson, M.
Anal. Chem. 1990, 62, 32.

Thompson, M.; Stone, D. C.; Nisman, R. Anal. Chim. Acta 1991, 248, 143.
Yan, Y.; Bein, T. J. Phys. Chem. 1992, 96, 9387.

Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325.

Edmonds, T. E.; West, T. S. Anal. Chim. Acta 1980, 117, 147.

Grate, J. W.; Rose-Pehrsson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H.
Anal. Chem. 1993, 65, 1868.

Curie, J.; Curie, P. Bull. Soc. Min. Paris 1880, 3, 90.
Sauerbrey, G. Z Phys. 1959, 155, 206.

Mueller, R. M.; White, W. Rev. Sci. Instum. 1968, 39, 291.
Mueller, R. M.; White, W. Rev. Sci. Instum. 1969, 40, 1646.

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

Kanazawa, K. K.; Gordon, J. G. Anal. Chim. Acta 1985, 175, 99.
Kanazawa, K. K.; Gordon, J. G. Anal. Chem. 1985, 57, 1770.
Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325.

Edmonds, T. E.; West, T. S. Anal. Chim. Acta 1980, 117, 147.

Martin, S. J.; Frye, G. C. Proc. 1991 IEEE Ultrason. Symp. 1991, 393.
Hiavay, J.; Guilbault, G. G. Anal. Chem. 1977, 49, 1890.

Guilbault, G. G.; Jordan, J. M. CRC Crit. Rev. Anal. Chem. 1988, 19, 1.
Grate, J. W.; Martin, S. J.; White, R. M. Anal. Chem. 1993, 65, 940A.
Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 987A.
Alder,,J. F.; McCallum, J. J. Analyst 1983, 108, 1169.

26.

27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.

45.
46.

47.

48.

49.
50.

3O

Munoz Leyva, J. A.; Hidalgo De Cisneros, J. L.; Garcia Gomez DE Barreda, D.;
Beerra, Fraidias, A. J. Talanta 1994, 41, 159.

Ward, M. D.; Buttry, D. A. Science 1990, 249, 1000.

Dickert, F. L.; Bauer, P.A. Adv. Mater. 1991, 3, 436.

Dickert, F.L.; Schuster, 0. Adv. Mater. 1993, 5, 826.

Dickert, F. L.; Haunschild, A. Adv. Mater. 1993, 5, 887.

Carey, W. P.; Kowalski, B. R. Anal. Chem. 1986, 58, 3077.

Carey, W. P.; Beebe, K. R.; Kowalski, B. R. Anal. Chem. 1987, 59, 1529.
Lord Rayleigh London Math. Soc. Proc. 1887, 17, 14.

Wohltjen, H. US. Patent 4,312,228, Jan. 26, 1982.

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

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1458.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51 , 1465.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1470.

Wohltjen, H. Sens. Actuators 1984, 5, 307.

Frye, G. C.; Martin, S. J. Appl. Spectrosc. Rev. 1991, 26, 73.

Bowers, W. D.; Chuan, R. L.; Duong, T. M. Rev. Sci. Instrum. 1991, 62, 1624.
Grate, J. W; Klusty, M. Anal. Chem. 1991, 63, 1719.

Tiersten, H. F.; Sinha, B. K. J. Appl. Phys. 1978, 49, 87.

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

Tiersten, H. F.; Sinha, B. K. J. Appl. Phys. 1978, 49, 87.

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

Ferry, J. D. Viscoelastic Properties of Polymers, 3rd ed.; John Wiley and Sons:
New York, 1980. -

Rogers, C. E. in Polymer Permeability; Comyn, 1., Ed.; Elsevier: New York,
1985; p 11-73.

Martin, S. J .; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201.

Martin, S. J.; Ricco, A. J .; Frye, G. C. 1990 Solid State Sensor & Actuator
Workshop, ACS Spring Meeting, San Francisco.

51.

52.

53.
54.
55.

56.
57.
58.
59.

61.
62.
63.

65.

66.

67.

68.
69.
70.

71.
72.

31

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

Roe, R.-J. Encyclopedia of Polymer Science and Engineering, 2nd ed.; John
Wiley and Sons, Inc.: New York, 1987; Vol. 7, 531-544.

Janata, J. Anal. Chem. 1992, 64, 196R.
Janata, J .; Josowicz, M.; De Vaney, D. M. Anal. Chem. 1994, 66, 207R.

Bastians, G. J. in T. E. Edmonds (Ed), Chemical Sensors, Blackie: London, 1988,
Chp. 14, p. 295.

Janata, J ., Principles of Chemical Sensors, Plenum: New York, 1989.
Mierzwinski, A.; Witkiewicz, Z. Environ. Pollut. 1989, 5 7, 181.
Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261.
D’Aminco, A.; Verona, E. Sens. Actuators 1989, 17, 55.

McCallum, J. J. Analyst(London) 1989, 114, 1173.

Fox, C. G.; Alder, J. F. Analyst (London) 1989, 114, 997.

Katritzky, A. R.; Offerman, R. J. CRC Anal. Chem. 1989, 21(2), 83.
Alder, J. F.; McCallum, J. J. Analyst (London) 1983, 108, 1169.
Nieuwenhuizen, M. S.; Barendsz, A. W. Sens. Actuators 1987, 11, 45.

Snow, A. W.; Barger, W. R. in A. B. P. Lever and C. C. Leznoff (Eds),
Phthalocyanines Properties and Applications, VCH Publishers: New York, 1989,
Chp. 5. PP. 341.

Zellers, E. T. In Chemical Sensors and Microinstrumentation; Murray, R. W.,
Ed.; ACS Symposium Series 403; American Chemical Society: Washington, DC,
1989; Pp.176-190.

Zellers, E. T.; Hassold, N. C.; White, R. M.; Rappaport, S. M. Anal. Chem.
1990, 62, 1227.

Zellers, E. T.; Zhang, G.-Z. Anal. Chem. 1992, 64, 1277.
Snow, A. W.; Wohltjen, H. Anal. Chem. 1984, 56, 1411.

Zellers, T. E.; White, R. M.; Rappoport, S. M.; Wenzel, S. W. Transducers ’87,
459.

Zhang, G.-Z.; Zellers, E. T. Anal. Chem. 1993, 65, 1340.
Dickert, F. L.; Bauer, P.A. Adv. Mater. 1991, 3, 436.

73.

74.

75.

76.
77.
78.

79.
80.

81.
82.

83.

84.

85.
86.
87.
88.
89.

90.

91.

92.
93.

32

Dickert, F. L.; Haunschild, A.; Hofmann, P.; Mages, G. Sens. Actuators B 1992,
6, 25.

Dickert, F. L.; Baumler, U. P. A.; Zwissler, G. K. Synth. Met. 1993, 61, 47.

Dickert, F. L.; Bruckdorfer, Th.; Feigl, H.; Haunschild, A.; Kuschow, V.;
Oberveier, E.; Bulst, W.-E.; Knauer, U.; Mages, G. Sens. Actuators B. 1993, 13-
14, 297.

Dickert, F. L.; Haunschild, A. Adv. Mater. 1993, 5, 887.
Dickert, F. L.; Haunschild, A.; Maune, V. Sens. Actuators B 1993, 12, 169.

Shurig, V.; Novotny, H. P.; Schmalzing, D. Angew. Chem. Int. Ed. Engl. 1989,
28, 736.

Dickert, F.L.; Schuster, 0. Adv. Mater. 1993, 5, 826.

Stoddart, R.; Zarycki, R. Cyclodextrins, Royal Society of Chemistry: London,
1989.

Gutsche, C. D. Calixarenes, Royal Society of Chemistry: London, 1989.

Lai, C. S.-I., Moody, G. J .; Thomas, J. D. R.; Mulligan, D. C.; Stoddart, J. F.;
Zarzycki, R. J. Chem. Soc., Perkin Trans. 2 1988, 319.

Ballantine, D. 8.; Rose, S. L.; Grate, J. W.; Wohltjen, H. Anal. Chem. 1986, 58,
3058.

Ema, K.; Yokoyama, M.; Nakamoto, T.; Moriizumi, T. Sens. Actuators 1989, 18,
291.

Ho, M. H.; Guilbault, G. G. Reitz, B. Anal. Chem. 1983, 55, 1830.
Ho, M. H.; Guilbault, G. G.; Reitz, B. Anal. Chem. 1980, 52, 1489.
Lopez-Roman, A.; Guilbault, G. G. Anal. Lett. 1972, 15, 225.
Tomita, Y.; Ho, M. H.; Guilbault, G. G. Anal. Chem. 1979, 51, 1475.

Brady, J. E.; Bjorkman, D.; Herter, C. D.; Carr, P. W. Anal. Chem. 1984, 56,
278.

Lucklum, R.; Henning, B.; Hauptrnann, P.; Schierbaum, K. D.; Vaihinger, S.;
Gopel, W Sens. Actuators A 1991, 25-2 7, 705.

Schierbaum, K. D.; Gerlach, A.; Haug, M.; Gopel, W. Sens. Actuators A 1992,
31, 130.

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

Karasek, F. W.; Guy, P.; Hill Jr., H. H.; Tiernay, J. M. J. Chromatogr. 1976,
124, 179-186.

94.
95.
96.
97.

98.
99.

100.
101.

102.
103.
104.
105.
106.
107.
108.

109.

110.
111.

112.

113.
114.

115.

33

Littlewood, A. B. Gas Chromatography, Academic Press: New York, 1970; 44.
Burkhardt, G. N. Nature (London) 1935, I7, 125.
Hammett, L. P. Chem. Rev. 1935, 17, 125.

Chemial Applications of Atomic and Molecular Electrostatic Potentials; Politzer,
P., Truhlar, D. G., Eds.; Plenum: New York, 1981.

Hammett, L. P. J. Am. Chem. Soc. 1937, 59, 96.

Exner, 0. Correlation Analysis of Chemical Data; Shorter, J ., Ed.; Plenum:
New York, 1988.

Exner, O. Prog. Phys. Org. Chem. 1990, 18, 129.

Weinstein, H.; Osman, R. Green, J. P.; Topiol, S. in Chemial Applications of
Atomic and Molecular Electrostatic Potentials; Politzer, P., Truhlar, D. G., Eds.;
Plenum: New York, 1981, 309.

Hansch, C.; Fujita, T. H. J. Am. Chem. Soc. 1964, 86, 1616.

Hall, L. H.; Kier, L. B. Rev. Comput. Chem. 1991, 2, 367.

Bersuker, I. B.; Dimoglo, A. S. Rev. Comput. Chem. 1991, 2, 423.

Hansch, C. Acc. Chem. Res. 1993, 26, 147.

Abboud, J.-L. M.; Kamlet, M. J.; Taft, R. W. J. Am. Chem. Soc. 1977, 99, 8325.
Karnlet, M. J .; Taft, R. W. Prog. Org. Chem. 1983, 48, 2877.

Karnlet, M. J.; Abraham, M. H.; Doherty, R. M.; Taft, R. W. J. Am. Chem. Soc.
1984, 106, 464.

Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1991, 587, 213.

Abraham, M. H.; Whiting, G. S. J. Chromatogr. 1992, 594, 229.

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.

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

Abraham, M. H. Chem. Soc. Rev. 1993, 22, 73.

Abraham, M. H.; Grellier, P. L.; McGill, R. A. J. Chem. Soc. Perkin Trans. 2
1987, 797.

Abraham, M. H.; McGowan, J. C. Chromatographia 1987, 23, 243.

34

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

117. Patrash, S. J.; Zellers, E. T. Anal. Chem. 1993, 65, 2055.

Chapter 2

Effect of Silanization and Operating Temperature on the Response of
Uncoated and Poly(isobutylene) Coated Surface Acoustic Wave (SAW)

Sensors

2.1 Abstract

The effects of Silanization and Operating temperature (30 and 60°C) on the
response of uncoated and poly(isobutylene) coated surface acoustic wave (SAW) devices
to alkanes, alcohols, and water have been investigated using pulsed sample introduction.
Silanization of the SAW surface with dichlorodimethylsilane “deactivates” the surface
primarily through a decrease in the active surface area of the device. Silanization also
appears to increase the surface area of the poly(isobutylene) coating and decrease the
contribution of vapor induced changes in thin film mechanical properties to sensor
response. Increasing the operating temperature of an uncoated or poly(isobutylene)
coated, unmodified SAW device decreases the response to alkanes and alcohols, but
increases the response to water. The increase in water response has been ascribed to an
increase in the number of silanol and silylether groups that are available for adsorption on
the quartz surface held at 60°C. At 30°C, these sites are partially occupied by trace water
' adsorbed from the carrier gas. For the uncoated, silanized SAW device, the response to

all analytes except l-butanol decreases with increasing temperature. The increase in 1-

35

36
butanol response has been attributed to adsorption at additional silylether and tracesilanol
sites available on the higher temperature surface. Increasing the Operating temperature
decreases the magnitude of the response of the poly(isobutylene) coated, silanized SAW

devices to all analytes examined.
2.2 Introduction

Considerable effort has been devoted to the development of vapor sensors based
on surface acoustic wave (SAW) devices [1-17]. Most sensors utilize the sorption
properties of a thin film coating of polymeric material to improve the analytical
capabilities of the device. The importance of sorption phenomena has led to the
application of linear solvation energy relationships (LSERs) to describe analyte/film
interaction mechanisms at a molecular level [18—20] and to predict suitable coatings for
detection of specific analytes [18,21]. In addition to these chemical aspects, construction
of a practical sensor device requires consideration of thin film physical properties such as
adhesion and thermal stability. These properties are influenced in part by polymer/device
surface interactions. Consequently, a number of investigators have examined the effect of
surface modification and operating temperature on sensor response.

Several studies have shown that pretreatment of the SAW device surface (usually
SiO2) improves the uniformity, stability, and performance of sensor coatings. Grate et al.
[22] reported that higher quality films could be prepared on devices cleaned using a
nitrogen plasma versus solvent rinsing. Ballantine [23] used silanized devices to study

the influence of poly(isobutylene) film morphology on the sensor response to isooctane.

37
McGill et al. [24] modified SAW surfaces with diphenyltetramethyldisilazane in order to
maintain poly(isobutylene) adhesion and rrrinimize the impact Of interfacial adsorption on
sensor responses to water. Dickert et al. [25] have suggested that long-chain hydrocarbon
substituents of organosilyl compounds form assembled films that can include n-decyl
tethers attached to cyclodextrins. These authors attributed a decrease in response time for
the modified device to alignment of the included cyclodextrin cavities.

Theoperating temperature of the device affects the physical properties of the film
as well as the sensitivity and selectivity of the sensor. Grate et al. [22] showed that
thermal expansion of the polymer film decreases the device frequency. They proposed
that similar sorbent induced swelling phenomena led to larger sensor responses than
would be expected from only mass loading. Martin et al. [16] have recently presented an
extensive treatise on the effect of film thickness and operating temperature on the
properties of poly(isobutylene) coated devices. It is also well known that increasing
temperature decreases the magnitude of the sensor response [9,22,26,27]. For example,
Grate et al. [9] have shown that the response of a fluoropolyol-coated sensor to dimethyl
methylphosphonate (DMMP) decreases exponentially, with increasing temperature.
These authors also reported that increasing temperature decreases the selectivity of the
sensor.

Previous studies have shown that surface modification (Silanization) and operating
temperature can significantly affect the response of SAW devices. However, earlier work
focused primarily on detailed studies of the equilibrium response for a limited number of

analytes. The present study is part of a larger effort devoted to the development of

38
practical sensor systems based on pulsed sample introduction. The importance of pulsed
sample introduction in chemical sensing has been discussed extensively [18,20,28-36]. In
this study, the influence of Silanization and operating temperature on the response of
uncoated and poly(isobutylene) coated SAW devices to pulses of alkanes, alcohols, and

water is reported.
2.3 Experimental

Chemicals. Poly(isobutylene) (abbreviated PIB) was obtained from Aldrich
chemical company (MW = 380,000). The alkanes and alcohols were obtained from
Aldrich chemical company and used as received. Water was distilled and deionized prior
to use.

SAW System. A schematic diagram of the sensor test apparatus is shown in
Figure 2.1. The approximate volume of the test cell is 8 mL. Analytes were delivered to
the cell as a pulse using a gas chromatograph injector operated at 240°C. The flow rate
was 20 mL/min flow of N2 (99.95%, AGA Gas Co.) dried over molecular sieves. The
temperature of the cell was controlled by resistance heating of nichrome wire wrapped
around the transfer line. The estimated water content of the carrier stream is < 5 ppm
[37].

SAW data were obtained using 158 MHz dual delay line devices manufactured by
Microsensor Systems, Inc. (Bowling Green, KY). The SAW device is permanently
mounted on a conventional 12 pin gold TO-8 header which has bonded gold wire

connections. The device exhibits a resonant frequency shift of approximately 365 Hz

 

 

 

 

( \
Gas Handling
System
L J

SAW
Figure 2.1.

 

 

39

 

Injector

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

 

 

K

Data
Acquisition

 

/

Schematic diagram of sensor test apparatus.

Aluminum
locks

40

when perturbed by a surface mass change of 1 nanogram. An RF electronics module
(CHM-158, Microsensor Systems, Inc.) is used to power the sample and reference
devices. The output of the electronics module consists of the individual SAW oscillator
frequencies as well as the difference frequency between the sample and reference devices.

Software developed in-house is used to convert data from the Microsensor
Systems format to one compatible with Microsoft Excel (Microsoft Corporation,
Redmond, Washington). The responses reported in this work are the difference
frequencies between sample and reference devices contained in the sensor test cell. Both
delay lines of the sample device were silanized and/or coated with the polymer. The
reference device was a clean, uncoated device that was sealed using a metal lid and epoxy
resin (Torr Seal, Varian Vacuum Corporation).

Two methods were used to clean the SAW devices prior to Silanization or coating.
In the first procedure, the device was rinsed with acetone and methanol, soaked in
isopropanol, and then allowed to air dry. The second method involved the same solution
and drying steps, followed by a 30 minute treatment in an ozone cleaner (UV Ozone
Photoreactor PR-lOO; UVP, Inc., New Jersey). SAW devices that were either rinsed or
rinsed and treated with ozone are designated “unmodified”. Both types of unmodified
devices showed similar sensor responses. The surface of the quartz was silanized by
exposing the device to a 10% solution of dichlorodimethylsilane in toluene for 5 nrinutes
followed by methanol rinsing, isopropanol soaking, and air drying. PIB films are
prepared by spray casting from dilute toluene or chloroform solutions (~1000 ppm).

Approximately 85 kHz (200 ng, 0.027 um) films were used in this study.

41

Operating Procedures. A typical experiment involved 30 seconds of baseline
data collection followed by three to six injections of analyte. Injections were separated by
a length of time necessary for the frequency to return to baseline. Analytes were injected
in random order to minimize any possible effects of exposure history on response data. In
addition, some data were collected at 60°C followed by 30°C, while other data sets were
collected in the reverse order. Injection volumes were varied from 5 [LL for analytes with
low boiling points to 0.05 uL for high boiling point samples to produce frequency shifts
less than 25 kHz. Sensor responses are reported as the maximum frequency shift in hertz
per microgram (or nanomole) of analyte injected to account for differences in the analyte
injection volume. For the blanks and each film, data were collected on at least three
devices at both temperatures. The device to device SAW response values have typical

RSDs of 15%.

2.4 Results and Discussion

Effect of Silanization on Device Response. Uncoated. Table 2.1 shows the
responses of the unmodified and silanized devices to alkanes, alcohols, and water at 30°
C. Within an analyte class, the responses measured for both types of uncoated SAW
devices increase with increasing analyte boiling point. This suggests that the response is
strongly affected by analyte physisorption (condensation). Dickert and Haunschild [38]

have also shown an increase in sensor response with increasing alkane boiling point

42

 

 

n3 2... was e5 25 85 8d n: 2: be?
§ 28 ewe 8a SN SN 3.» me E 3:53-.
Rd a: we.” 24 and owe Sn ”.8 a 358.82
So we 43 42. ”3 $6 ed 36 me 3552:
o2 88 E. 23 as w: 3” 8e 5 23:8
3w ”9 4.8 w; 9; Se 9% 3a 3 258
EN EN 2N m2 EN EN one we we 83%..
new 4.3 Ni 4.9m ewe e3. a mom 8 2.22

BOEOSWENE Eaumihm 22:05.3: Swami—Em 29:23:}: Ewan: 2oEo§E~m gum Gov .Eom 8252

e355 BEBE: 825% Beeeee: 3:8
8.80 m3 888::

 

 

 

 

.eemne O08 a 838mg :2: nee 833.2

AN 2an

43
(pentane, hexane, and heptane) using a silanized 430 MHz SAW resonator. King [26]
reported that higher boiling analytes are able to achieve a larger level of partitioning into
a coated sensor device. Patrash and Zellers [19] also reported that an increase in analyte
boiling point increases partition coefficients calculated using SAW responses in
continuous flow experiments. While it is not possible for partitioning to occur for
uncoated devices, it is expected that analyte condensation would produce similar results.

Silanization decreases the magnitude of the responses obtained with the uncoated
device. These findings can be explained in terms of the effect of Silanization on the
physical and chemical properties of the SAW device surface. We believe that Silanization
lowers the active surface area of the SAW device. Since the response of the uncoated
device is strongly influenced by condensation, a decrease in surface area should decrease
the response observed with the silanized sensor. Silanization is also known to reduce the
surface area of chromatographic supports [39]. It should be noted that dichlorosilanes
can polymerize over the quartz surface, thus similar experiments with
monochlorosilanization reagents may not yield the same results.

Silanization is also known to deactivate silica surfaces by replacing the active
proton on the silanol functionality with a less active, silyl derivative [40,41]. This change
in chemical functionality is expected to decrease the sensor response to analytes that can
hydrogen bond to the device surface (e. g., alcohols and water). McGill et al. [24]
attributed a decrease in interfacial water adsorption on phenyl silane modified SAW
devices to a decrease in the number of silanol groups present on the silanized surface.

Silanization should improve the chemical affinity of the device for alkanes. However, the

44
pronounced decrease in alkane response observed following Silanization suggests that the
decrease in active surface area influences the sensor response more than the change in
chemical properties of the SAW surface.

It should also be noted that Silanization of the device does not appear to
preferentially decrease the responses to alcohols or water compared to the alkanes. Thus,
for the short vapor exposure times typical of pulsed sample injection, changes in chemical
affinity of the device surface may have less impact on the response magnitude than the
decrease in active surface area.

Poly(isobutylene) Coated Silanized. Table 2.1 also shows the responses obtained
for the PIB coated unmodified and PIB coated silanized SAW devices operated at 30°C.
As anticipated, the responses observed for the PIB coated devices are greater than those
obtained with either the unmodified or silanized device.

Alkanes are expected to interact strongly with the PIB film through dispersion
forces [18-20,42]. Our results indicate that Silanization has little effect on the responses
to hexane, heptane, and octane; however, the nonane response is smaller on the PIB
coated silanized device. The difference in nonane response observed for the two PIB
coated devices can be understood in terms of mechanical phenomena that have been
reported for PIB coated SAW devices [16,22,23]. Ballantine [23] reported that the
contribution of mechanical effects depends on the coating thickness and the amount of
analyte sorbed in the film. We believe that the combination of condensation and sorption
during our pulsed sample introduction produces sufficient nonane concentration in the

film to cause such effects. However, our results indicate that the magnitude of these

45

effects are lower on the PIB coated silanized device. It has been suggested that
Silanization of the SAW device surface before coating enhances the wetting and adhesion
of polymer films [23,24]. In addition, Dickert and Haunschild [38] have noted that
hydrophobic coatings do not adhere well to quartz surfaces. We propose that the
increased adhesion of the PIB film to the silanized surface decreases the mechanical
perturbation of the film produced by vapor sorption. This decreases the magnitude of
moduli changes for the PIB coated silanized device and results in a lower response to
nonane. Similar phenomena may occur for alkanes with lower boiling points; however,
the analyte concentration in the film resulting from our pulse injection may be too low to
Observe any significant effects.

Silanization increases the response of the PIB coated device to alcohols. We
ascribe these findings to an increase in PIB surface area resulting from improved wetting
and adhesion of the polymer to the surface of the silanized device. Since alcohols are not
expected to interact with PIB as strongly as alkanes [18-20], the increased surface area
may have a more significant effect on alcohol sorption. Alcohols are also known to self-
associate in hydrophobic solvents [43,44]. Thus, increased sorption and self-association
both contribute to the enhanced alcohol response observed with the PIB coated silanized
device.

The predominant role of dispersion forces in the interaction of analytes with PIB
suggests that the PIB coated devices should not show a strong response to water. Indeed,
with the exception of methanol, water gives the lowest response of the analytes. McGill

et al. [24] have reported that PIB coated SAW devices exposed to concentrated water

46

vapor streams ([H2O] > 6194 mg/m3) exhibit fast mass loading responses (decrease in
frequency) followed by slowly developing anomalous responses (increase in frequency).
For lower concentration water streams, only anomalous responses were observed. These
researchers also reported that Silanization eliminated the anomalous response and resulted
in the low water vapor responses expected for a PIB coated device. We have not
observed anomalous responses for any of the analytes. Apparently, the short
analyte/surface contact time in a pulsed experiment precludes observation of these slowly
developing phenomena.

Effect of Operating Temperature on Device Response. Uncoated. Table 2.2
shows the responses determined using an unmodified device operated at 60°C.
Comparison of these responses with those obtained at 30°C (Table 1) indicates that
increasing the sensor temperature decreases the responses obtained for alkanes and
alcohols. Similar results have been obtained for various polymer-coated sensors
[9,22,26,27]. In contrast, water shows a larger response at 60°C. This observation can be
explained by the effect of temperature on the number of surface silanol and silylether
groups that are available for hydrogen bonding interactions. For the device held at 30°C,
we believe that trace water present in our carrier gas (< 5 ppm, see Experimental) adsorbs
on some of the silanol and silylether groups and essentially decreases the number of
active adsorption sites available to the water vapor pulse. Increasing the sensor
temperature to 60°C shifts the adsorption equilibrium and, consequently, increases the

number of active sites available for interaction with water from the vapor pulse. Thus the

47

 

 

w—d owd V0.0 Vfim N— .o 3.6 ~06 5mm 8— .583
Gad mm— 06— m: WNM w.mv wvd van 5: 3:535-—
5._ Mg EN 4.9. Rd E... Re 2.2 S 65852
A: .o vmd _ —.o end :6 mad 8.0 Nwd no 35505
cm— 95 02 02: mod N.m~ wfio wNm ~m~ 0525:
v. g N ha WWN NNN and awé aNd v. g m ©N_ 0:800
mvé New mNd ”no avd hwé 3d mdm ma ocean..—
Obg WON no; m.NN o— .o ~a.— th OYw Go 0:83:
29:98:}: Eflmqim o_oEo§=§m Enema—hm oBEOSEEZ EEménm 29:28.33 an}: 610% 26m 85.22..
Bneefi 8585:: e855... 3585:: ”5:8
e280 m5 388:

 

 

 

 

.Aeemfiwv Deco 3 830mm»: :05 new moi—«Ea

.N.N 2an

48
increase in active site density with increasing sensor temperature leads to a larger
response to water.

For the silanized SAW device, comparison of the responses measured at 30°C
(Table 2.1) and 60°C (Table 2.2) indicate that increasing the sensor temperature decreases
the response to all alkanes and alcohols except 1-butanol. As described above, a larger
number of silylether (and trace silanol) groups are available on the device surface held at
60°C. These additional sites can participate in hydrogen bonding interactions with 1-
butanol. We believe that the adsorption of l-butanol is enhanced by dispersion
interactions between the hydrocarbon portion of the molecule and the methyl groups ofthe
silane. In principle, similar phenomena could occur for isopropanol. However, the
interaction of isopropanol with the additional polar sites may be sterically hindered.

Poly(isobutylene) Coated. The responses obtained for PIB coated devices
operated at 60°C are also given in Table 2.2. Comparison of these data with the
appropriate data from Table I shows that increasing the sensor temperature decreases the
response to alkanes and alcohols. Similar response decreases have been reported for
several polymer-coated sensors [9,22,26,27]. However, the responses to water measured
with the uncoated, unmodified and the PIB coated, unmodified devices are similar. This
suggests that the PIB film does not completely cover the unmodified device surface.
Thus, the increase in water response with increasing temperature can be attributed to the
increase in available hydrogen bonding sites as described previously. McGill et al. [24]
have also shown that an unmodified device coated with a 250 kHz PIB film responds to

water.

E“
u:

PWflP‘MPP’Nt“

11.
12.
13.
14.
15.
l6.
17.

18.
19.
20.

21.
22.

23.
24.

49

References

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1458.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1465.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1470.

Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 940A.
Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 987A.
Ballantine, D. 8., Jr.; Wohltjen, H. Anal. Chem. 1989, 61 , 704A.

Fox, C. G.; Alder, J. F. Analyst (London) 1989, 114, 997.
Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261.

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

Ballantine, D. 8.; Rose, S. L.; Grate, J. W.; Wohltjen, H. Anal. Chem. 1986, 58,
3058.

Wohltjen, H. Sens. Actuators 1984, 5, 307.

Snow, A. W.; Wohltjen, H. Anal. Chem. 1984, 56, 1411.

Frye, G. C.; Martin, S. J. Appl. Spectrosc. Rev. 1991, 26, 73.
McCallum, J. J. Analyst(London) 1989, 114, 1173.

Katritzky, A. R.; Offerman, R. J. CRC Anal. Chem. 1989, 21(2), 83.
Martin, S. J .; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201.

Martin, S. J .; Ricco, A. J.; Ginley, D. S.; Zipperian, T. E. IEEE Trans. Ultrason.
1987, UFFC-34, 142-147.

Grate, J. W.; Abraham, M. H. Sens. & Act. B. 1991, 3, 85.
Patrash, S. J .; Zellers, E. T. Anal. Chem. 1993, 65, 2055.

EB. Townsend IV, J.S. Ledford and DP. Hoffmann, submitted to Sensors and
Actuators B.

McGill, R.A.; Abraham, M.H.; Grate, J .W. Chemtech 1994, 24(9), 27.

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

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

McGill, R.A.; Grate, J.W.; Anderson, M.A. Interfacial Design and Chemical
Sensing, ACS Symposium Series 561, Anterican Chemical Society, Washington,
DC, 1994, 280.

25.

26.
27.
28.

29.

30.
31.
32.

33.
34.
35.
36.
37.
38.
39.

40.

41.

42.

43.

50

Dickert, F. L.; Bruckdorfer, Th.; Feigl, H.; Haunschild, A.; Kuschow, V.;
Oberveier, E.; Bulst, W.-E.; Knauer, U.; Mages, G. Sens. Actuators B. 1993, 13-
14, 297.

King Jr., W. H. Anal. Chem. 1964, 36, 1735.
Guilbault, G. G.; Jordan, J. M. CRC Crit. Rev. Anal. Chem. 1988, 19, 1.

EB. Townsend IV, J .S. Ledford and DP. Hoffmann, accepted for publication in
Review of Scientific Instruments.

Grate, J. W.; Rose-Pehrsson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H.
Anal. Chem. 1993, 65, 1868.

Suleiman, A; Guilbault, G. G. Anal. Chim. Acta 1984, 162, 97.
Thompson, M.; Stone, D. C. Anal. Chem. 1990, 62, 1895.

Hecki, W. M.; Marassi, F. M.; Kallury, K. M. R.; Stone, D. C.; Thompson, M.
Anal. Chem. 1990, 62, 32.

Thompson, M.; Stone, D. C.; Nisman, R. Anal. Chim. Acta 1991, 248, 143.
Yan, Y.; Bein, T. J. Phys. Chem. 1992, 96, 9387.

Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325.

Edmonds, T. E.; West, T. S. Anal. Chim. Acta 1980, 117, 147.

Private communication, Supelco technical support.

Dickert, F. L.; Haunschild, A. Adv. Mater. 1993, 5, 887.

R.L. Grob in Modern Practice of Gas Chromatography, 2nd Edition, R.L. Grob,
Ed., John Wiley & Sons, New York, New York, 1985.

ML. Lee, R]. Yang, and K. D. Battle, Open Tubular Column Gas
Chromatography, John Wiley & Sons, New York, New York, 1984.

R. P. Evershed in Handbook of Derivatives for Chromatography, 2nd Edition, K.
Blau and J. Halket, Ed., John Wiley & Sons, New York, New York, 1993.

Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. 8., Jr.; Jurs, P. C. Anal. Chem.
1988, 60, 2801.

Barton, A.F.M. Pure Appl. Chem. 1985, 5 7, 905.
Myers, M. E.; Abu-Isa, I. A. J. Appl. Poly. Sci. 1986, 32, 3515.

Chapter 3

Development and Performance of a Semi-automated 200 MHz Surface

Acoustic Wave Resonator Array-Based Sensor Test Apparatus

3.1 Abstract

A semi-automated pulse injection and data analysis system utilizing 200 MHz
surface acoustic wave resonators has been constructed. A cell consisting of separate
sample splitting chamber, sample transfer lines, and sensor device areas has been
constructed to allow individual temperature control over the regions. Experiments
involving injections of n-hexane, trichlorobenzene. and 1% (w/v) of biphenyl in n-hexane
show that all three temperatures play an important role in the magnitude and time
dependency of the SAW response. Flow rate is also shown to be an important factor.
The capabilities Of the system have been demonstrated with uncoated and

poly(isobutylene) coated devices.
3.2 Introduction

Surface acoustic wave (SAW) devices have been widely studied as mass sensors
[1-15]. The use of coatings to enhance the selectivity and sensitivity of the devices is
well established [4,5], however, it has become apparent that a single film and device is
not adequate for many practical applications. This limitation has led to the use of SAW
arrays to obtain more information about the composition of a sample. SAW array data
sets are well suited to various data handling, manipulation, and recognition techniques

that increase the information content of the results. Pattern recognition techniques have

51

52
been applied to mass sensors with some success [16-23] and the application of neural
networks has also been suggested [18,23].

There has been considerable interest in the development of SAW technology and
associated instrumentation since Wohltjen demonstrated many capabilities of the device
in 1979 [1-3]. For example, Grate et al. [24] developed an automated continuous vapor-
generation and data collection system for chemical microsensors. Nederlof and
Nieuwenhuizen [25] detailed their automated continuous flow system for testing SAW
gas sensors. Munoz Leyva et al. [26] described a test system that can be used to study the
effects of pressure, temperature, and relative humidity on the response of a piezoelectric
sensor. Grate et al. [19] reported efforts to develop a smart sensor system for the
detection of organophosphorus and organosulfur compounds using an array of SAW
sensors in conjunction with automated sample preconcentration. Recently, Ricco and
Martin [27] have developed multiple frequency SAW devices for sensing and materials
characterization.

SAW responses are usually measured with devices that have equilibrated with
sample vapor. This method of testing produces a great deal of fundamental information,
but many practical sensor applications require the analysis of pulsed sample streams [28-
35]. The details of such analyses vary with the specific application. However, in general,
a pulse experiment involves a more rapid sequence of analyte sorption and desorption
than is commonly associated with equilibrium methods. Grate and Abraham [33] have
suggested that this type of sample introduction can be used in conjunction with laser or
thermal desorption. Grate et al. [19] have also used thermal desorption of analyte from
preconcentrator tubes to deliver pulses of gas to an array of SAW devices. In addition to
these practical applications, we have recently reported that Linear Solvation Energy
Relationships (LSER) derived from data collected using pulsed sample introduction can

provide fundamental insight into coating properties [36].

53

This work is part Of a larger study devoted to the development of practical sensor
systems for environmental analysis [36,37]. In this paper we describe a sensor test
system designed for convenient, unattended collection of response data from SAW
resonator arrays. The system employs an autosampler and custom software for data
acquisition, manipulation, and storage. We have also developed a sensor test cell that has
separate temperature controlled regions for sample splitting, sample transfer, and SAW
device containment. The test cell contains a six array 200 MHz SAW resonator device
that could ultimately be used for pattern recognition studies. The availability of six
channels also allows rapid collection of the large amounts of data needed for fundamental
studies of sensor coatings using LSER methods [33,36,38]. Responses measured for
uncoated and poly(isobutylene) coated devices have been used to demonstrate the

capabilities of the system.
3.3 Experimental

Chemicals. Poly(isobutylene) (MW = 380,000) was obtained from Aldrich
chemical company. n-Hexane, trichlorobenzene, and biphenyl were obtained from
Aldrich chemical company and used as received.

Surface Acoustic Wave (SAW) Device. Data were obtained using a six element
array of 200 MHz SAW resonator microbalances (Femtometrics, Costa Mesa, CA). The
200 MHz resonator has been described previously [39]. The frequency difference
between each resonator and a separate reference device (beat frequency) is the measured
signal. Power is supplied and frequency counted by an external unit provided by the
manufacturer. The SAW resonator exhibits a 904 Hz frequency shift when perturbed by a
l nanogram mass change.

Prior to use or coating, the SAW devices were rinsed with acetone and methanol,
soaked in isopropanol, and dried in air. Poly(isobutylene) coated devices were prepared

by spray casting from dilute (~500 ppm) toluene or chloroform solutions of the polymer.

Typi

calm

IESOE

54
Typical film thickness was 100 kHz, which corresponds to a 0.012 um thick film. This
film thickness is calculated using the density of the polymer, the mass deposited as
calculated from the frequency shift, and the active area of the SAW device.

Test Cell. Two resonator configurations are available from Femtometrics: single
resonators mounted on individual gold platforms and the six element array device used in
this work. Figure 3.1 shows a schematic diagram of the bottom part of the test cell with
the array platform inserted. The cell bottom is constructed from brass. Electrical contact
between the array and the Femtometrics control board is made through a rectangular piece
of circuit board using TO-8 pins. High vacuum epoxy (Torr Seal, Varian Vacuum
Company) is used to seal all openings in the bottom of the cell (circuit board holes around
pins, unused circuit board holes, and circuit board/cell interface). The cell was designed
to accommodate either type of SAW device. In the configuration using single devices,
the SAWs are installed perpendicular to the devices mounted on the single platform. The
volume of the cell bottom is approximately 15 mL.

Figure 3.2 shows a schematic diagram of the complete test cell (top and bottom).
The top and bottom are joined using six stainless steel bolts and a thin (~ 1/32”) Teflon
gasket. The top of the cell is a solid brass block with six holes for the transfer lines
machined through directly above the SAW devices. Additionally, a l/ ” id. path was
machined around the outer edge of the block to allow for water cooling. One end of each
1/16” i.d. (1/8” o.d.) stainless steel transfer line is silver soldered into the top of the brass
block. The other end of each transfer line is silver soldered into a stainless steel splitting
chamber (internal volume ~ 1 mL). Mounted directly above the splitting chamber is a
stainless steel housing designed to accommodate a 1/ ” diameter heater cartridge. The
splitting chamber is attached to a GC injector via a 1/16” id (1/8” od) stainless steel
transfer line and a 1/4” Swagelock connector. The total volume of the system (injector,
splitting chamber, transfer lines, and cell) is approximately 20 mL. Equal splitting in the

chamber was verified by examining the response of n-hexane on 6 blank devices.

55

duets—m 05 :o 803% 05 9 3:20:28qu
@2508 2a 803% 3<m 32:58 335232: 60:85 83% >38 Shauna o5 wagon.“ 8289 :8 95 mo 32>

 

ems >83,
m e8 vegan;

Adaaz

I

7-6 cm Chamber
Temperature

Transfer Line
Temperature

Flow Out
V~

 

 

Cell Top

I

.0

 

 

 

 

56

Flow In

"—— Heater Cartridge

/ Splitting Chamber
Flow Out
I

 

{Water Bath

Water Bath

I
I
\

60‘“

Cell Bottom

Figure 3.2. Schematic diagram of the complete sensor test cell.

57

Heat is supplied to the cell top using a 200 W, 1/ ” x 1 1/ ” heater cartridge
(Omega) attached to the splitting chamber and nichrome wire wrapped around the transfer
lines. The heated areas of the cell top are insulated using ceramic tape (Wilt, Lake
Pleasant, NY) and fiberglass insulation. Temperature is monitored in three places using
K-type thermocouples: directly underneath the splitting chamber, at the bottom of one of
the transfer lines, and next to the array platform. Transfer line temperatures can range
from 70° to 160°C while the splitting chamber temperature can go as high as 500°C.
Typical operating temperatures are splitting chamber between 250° and 400°C, transfer
line at 100°C, and sensor cell at 30°C. The lid of the cell is either heated or cooled
(depending on the temperature of the transfer lines) with a water bath in order to maintain
a 30°C sensor cell temperature. With a heated water bath, the temperature of the sensor
cell can reach 90°C.

Analytes were delivered to the test cell as a pulse using a Varian autosampler
Series 8000 and a Varian gas chromatograph injector operated at 250°C. The
autosampler was mounted on a steel plate above the injector to provide mechanical
stability to the apparatus. The pneumatic pressure for the autosampler and the carrier gas
flow for the injector are provided by a standard laboratory gas cylinder of N2 (99.95%,
AGA Gas Co.). The carrier gas is dried over molecular sieves and the flow rate is
controlled by Porter Mass Flow Controllers. Typical carrier gas flow rates vary from 50
to 100 cc/min. We have used either 1.0 or 10.0 uL syringes for sample introduction.
Three injections of each analyte were made on at least three devices to ensure
reproducibility. Sensor responses are reported as the maximum frequency shift in hertz.
We do not believe that peak area is a valid measure of our sensor response since such a
measurement convolutes the amount of analyte sorbed with the residence time of the
analyte on the surface of the device. RSDs for SAW responses were approximately 10%.

Data Acquisition and Manipulation. The beat frequencies of the resonators

were collected using a digital counting board (MetraByte, Taunton, MA) and stored in a

58

Gateway 2000 386-25 personal computer. The Femtometrics supplied software is
capable of collecting data for all 6 channels at a resolution of either 100 (1pt/0.25ec), 10
(1pt/0.8sec), 5 (1pt/1.53ec), 2 (1pt/3.0sec), or 1 (1pt/6.0sec) Hz. Data presented in this
paper were collected with a resolution of 10 Hz in order to provide at least three data
points across a typical peak maximum. The standard software has been modified by
Femtometrics to allow for timed, multiple runs that allow full utilization of the Varian
autoinjector capabilities.

Data analysis was performed using custom software written using LabView for
Windows (v2.52, National Instruments, Austin, TX). Figure 3.3 shows an example of the
software output. The software separates the Femtometrics file into individual channels,
and then simultaneously displays the response of all 6 channels on a single page. In
addition, the software determines peaks, peak heights, peak areas, averages, standard
deviations, and % RSDs. The user can control the sensitivity of the peak determination
and select either none, Savitsky-Goolay smoothing, or median filtering of the data. An
output file that contains peak height, peak 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.

3.4. Results and Discussion

We have reported that the response of an uncoated resonator can be understood
primarily in terms of analyte physisorption [36]. In contrast, polymer coatings often
exhibit selective sorption and mechanical effects [4,5,40,41] that may complicate
characterization of a sensor test system. For example, poly(isobutylene) is well known
for its ability to selectively interact with alkanes through dispersion interactions
[33,36,38] and its susceptibility to viscoelastic effects [40,41,42]. As a result of such
possible complications, most of the data presented in the paper were obtained using

uncoated SAW devices.

59

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

       

 

 

 

 

”-4 N I},
p I i i :- u-uv-cin- an" '
“b -'. T-.. “Own “y.

 

 

 

 

 

 

 

 

 

 

.l
1'n*.-f-fi..+
r --!I-o-t-fll- anal I DID-mutt--

urn—v.- m- """" nun-n”! : I he... u- "I.” hunt-1m

 

 

Figure 3.3. Output created by the LabView software written to process 6 channels of
SAW data.

. 60

In our system, the major instrument variables that can be controlled are the
temperatures of the sensor cell, transfer line, and splitting chamber as well as the carrier
gas flow rate. We anticipate that the sensor cell temperature will influence the analyte
residence time and adsorption/desorption equilibria in the cell. The temperature of the
splitting chamber and transfer line will influence analyte arrival time by affecting the
partitioning of the analytes during sample transfer from the injector to the sensor. Finally,
the carrier gas flow rate will also affect the arrival and residence times of the analyte. in
the cell.

Effect of Sensor Cell Temperature. Figure 3.4 shows the effect of cell
temperature on the response of an uncoated device to an injection of 3.0 uL of hexane
followed by 3.0 uL of 1% (w/v) biphenyl in hexane. For a cell temperature of 60°C, the
response to pure hexane is significantly lower than that observed for the biphenyl-hexane
mixture. This difference can be explained in terms of the relative sensitivity of the blank
to hexane and biphenyl. We have previously reported that enhanced adsorption of
compounds with higher boiling points increases the sensitivity of the uncoated device to
such analytes [36,38]. In addition, King [43] proposed that analytes with higher boiling
points are able to achieve a higher level of partitioning into a film coated on a
piezoelectric sensor. While partitioning cannot occur on the uncoated device, we believe
it is reasonable to expect adsorption effects to produce similar results. Thus we anticipate
that the response to biphenyl (bp = 256°C) would be much greater than to hexane (bp =
69°C). The large response to biphenyl leads to a larger response to the biphenyl-hexane
mixture.

Most studies of coated SAW devices employ sensor temperatures less than 60°C
in order to increase sensor response and prolong coating life. Figures 3.4b and 3.4c show
that decreasing the cell temperature increases the magnitudes of the peaks observed
following hexane and biphenyl-hexane injection. This can be attributed to greater analyte

adsorption on the cooler device surface. The most significant change in sensor response

61

 

246

245 -

244 -

 

 

243'- *IJ ‘ --+ a AT

 

242 -

241-

 

 

Beat Frequency (kHZ)

240-

239 -

 

 

 

238 - "" ___-

 

 

 

237 ‘ ' ‘
0 200 400 600 800 1000 1200 1400

True (3)

Figure 3.4. The effect of sensor cell temperature on the response of an uncoated
device to injections of 3 [IL of hexane followed by 3 u]. of a 1% (w/v)
mixture of biphenyl in hexane. Sensor cell temperatures = (a) 60°C, (b) 45
°C, and (c) 30°C. Splitting chamber temperature = 250°C. Transfer line
temperature = 50°C. Carrier gas flow rate = 80 cc/rnin.

62

is the peak splitting Observed for the biphenyl-hexane mixture. We believe that the
narrow peak (designated henceforth as “hexane/biphenyl peak”) is caused by rapid
adsorption, desorption, and purging processes involving hexane and some of the
biphenyl. After the hexane/biphenyl vapor responsible for the narrow peak leaves the
cell, the biphenyl which initially adsorbed on the walls of the cell desorbs and can interact
with the SAW device. The broadening observed on decreasing the cell temperature
reflects the longer times required for such processes to occur in a cooler cell. Eventually,
the remaining biphenyl is purged from the cell and the response returns to baseline. It is
possible that the relatively large cell (volume ~15 mL) used for these studies is in part
responsible for the degree of splitting shown in Figure 3.4. Larger surface area cells
would be expected to show more pronounced adsorption effects in this experiment.
However, the cell size is dictated by the dimensions of the array and the ease of array
handling. Slightly smaller cell volumes are possible, but they require serrriperrnanent
device encapsulation and/or less effective flow splitting. Furthermore, experiments with
these systems showed the same splitting phenomenon illustrated in Figure 3.4.

Eflect of Splitting Chamber/1' ransfer Line Temperatures. Portable sensors are
generally less elaborate than laboratory sensor test systems. One possible simplification
for portable devices is to minimize the number of heat sources used in the system. Under
such conditions, a thermal gradient will exist between the sample injector and the sensor
device. We have simulated this situation by using only the splitting chamber heating
cartridge to supply heat to the cell top. In this case, heat flow from the splitting chamber
will increase the transfer line temperature. Figure 3.5 shows the effect of increasing
splitting chamber/transfer line temperature on the response profiles obtained following
injections of 3.0 ML of hexane and 3.0 uL of 1% (w/v) biphenyl in hexane. For all of the
temperature combinations, the SAW response magnitude for pure hexane is essentially

constant. This suggests that the combination of heat from the splitting chamber and the

262

260

258

256

254

252

Beat Frequency (kHz)

250

248

246

Figure 3.5.

63

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50 100 150 200 250 300 350 400
Time (3)

Effect of various combinations of splitting chamber/transfer line
temperatures on the response of an uncoated device to injections of 3 pl.
of hexane followed by 3 [LL of a 1% (w/v) mixture of biphenyl in hexane.
Splitting chamber/transfer line temperatures- — (a) 100/38°C, (b) 200/49°C,
(c) 300/63°C, and (d) 400/77°C. Sensor cell temperature: 30°C. Carrier
gas flow rate: 80 cc/min.

64

transfer line is sufficient to keep hexane in the vapor phase. However, this is not the
casefor the biphenyl/hexane mixture. Note that for a splitting chamber/transfer line
temperature of 100°/38°C (Figure 3.5a), the hexane and hexane/biphenyl peaks have
approximately the same magnitude. This suggests that the biphenyl condenses in the
chamber and transfer lines to such an extent that essentially none of the biphenyl reaches
the sensor with the hexane. Increasing the splitting chamber/transfer line temperature
increases the magnitude of the hexane/biphenyl peak. We attribute this to an increase in
the amount of biphenyl arriving with the hexane. Finally, we note that the biphenyl peak
shape is dramatically affected by the splitting chamber/transfer line temperature. This
observation may be understood in terms of the arrival time of biphenyl and the desorption
kinetics of the adsorbed analyte. For lower temperature combinations (Figures 3.5a and
3.5b), most of the biphenyl arrives after the hexane, adsorbs on the device surface and cell
walls, and then slowly desorbs as described in the previous section. For higher
temperatures, more biphenyl adsorbs and desorbs with hexane. The tailing observed for
approximately 80 seconds after the hexane/biphenyl peak (Figure 3.5d) can be attributed
to slow desorption of the remaining biphenyl from the surface of the SAW device.

The separation between the hexane/biphenyl peak and the peak attributed to pure
biphenyl suggests a method for the determination of trace amounts of high boiling
compounds in low boiling solvent matrices. Indeed, adsorption and thermal desorption
methods have been used in combination with selective coatings to produce advanced
sensor systems. Kindlund et al. [44] reported that water can be separated from organic
species using thermal desorption in combination with a silicone oil preconcentrator.
Grate et al. [19] have also demonstrated the advantages of this approach in a
comprehensive study of a smart sensor system that employs thermally desorbed Tenax
preconcentrator tubes and an array of four coated SAW devices.

Effect of Transfer Line Temperature. The effect of transfer line temperature on

sensor response is most clearly illustrated with data obtained for a single component

65

sample. Figure 3.6 shows the effect of transfer line temperature on the response of an
uncoated SAW device to 0.1 [.11. injections of trichlorobenzene (bp = 214°C). Both
curves show a sharp increase in beat frequency approximately 2-3 seconds after injection.
However, curve (a) is significantly broader than curve (b). Both curves show similar
sloped falling edges. We believe that the shape of the response profile can be explained
in terms of sample dispersion in the transfer lines and the sample concentration in the
cell. For a transfer line temperature of 75°C, analyte dispersion in the lines results in a
low, constant sample concentration in the test cell for a significant time interval. The
result is a plateau in the response curve between 82 and 85 seconds. Increasing the
transfer line temperature from 75° to 150°C produces a narrower response profile and a
larger change in beat frequency compared to the corresponding features observed for the
lower transfer line temperature. This can be attributed to less sample dispersion in the
high temperature transfer lines and, consequently, a larger fraction of the TCB reaches the
cell in a shorter period of time. We must note that preliminary experiments with low
volume pulse injection systems that did not utilize heated transfer lines produced broad or
irreversible response profiles that could be mistaken for selective analyte/film
interactions.

Efiect of Carrier Gas F low Rate. Figure 3.7 shows the effect of carrier gas flow
rate on the response of the uncoated device to injections of 3.0 uL of hexane followed by
3.0 [1L of 1% (w/v) biphenyl in hexane. For a flow rate of 20 cc/min, both the hexane
and biphenyl peaks are broadened by significant sample dispersion in the transfer lines
and slow adsorption/desorption processes in the cell. Increasing the flow rate decreases
the analyte dispersion and increases the rate of the adsorption/desorption processes.
Consequently, the separation between the hexane/biphenyl and biphenyl peaks and the
width of the biphenyl peak decreases.

248

246

244

242

Beat Frequency (kHz)

240

238

Figure 3.6.

66

 

 

 

 

 

 

 

b
a
J n I 1 L L
80 90 100 1 10
Tire (s)

The effect of transfer line temperature on the uncoated device response to
0.1 tLL injections of trichlorobenzene. Transfer line temperatures = (a) 75°
C and (b)150°C. Splitting chamber temperature = 400°C. Sensor cell
temperature = 30°C. Carrier gas flow rate = 80 cc/min.

N
D)
\0

Beat Frequency (kHz)

238

Figure 3.7.

67

 

 

 

 

 

 

Mk:

 

 

 

 

200 400 600 800 1000 1200
Time (3)

Effect of carrier gas flow rate on the uncoated device response to
injections of 3 [AL of hexane followed by 3 [AL of a 1% (w/v) mixture of
biphenyl in hexane. Carrier gas flow rate = (a) 20 cc/min, (b) 50 cc/min,
and (c) 100 cc/min. Splitting chamber temperature = 250°C. Transfer line
temperature = 50°C. Sensor cell temperature = 30°C.

68

Efi‘ect of Analyte Concentration. The response of a sensor to changes in analyte
concentration is of interest for practical analytical devices and the fundamental study of
sorption phenomena. Figure 3.8 shows the effect of hexane injection amount on the
response of uncoated and 100 kHz poly(isobutylene) coated devices. Our results indicate
that the uncoated device response increases linearly with hexane amount. However, the
response curve obtained for the poly(isobutylene) coated device is nonlinear. Nonlinear
calibration curves are not uncommon for coated devices. Wohltjen and Dessey [2]
reported that the response of a DC 970 coated sensor used as a GC detector was nonlinear
for 0-20 ug injections of o-chlorotoluene. Grate et al. [9] found that a fluoropolyol
coated device gave a nonlinear response to dimethyl methylphosphonate in the
concentration range of 1-10000 ug/L. Later, Grate et al. [19] reported a nonlinear
response for the same vapor and film in a different SAW system over a vapor
concentration rangeiof 0—1500 mg/m3. Our experience with pulse injection systems has
indicated that the concentration range over which sensor response is linear depends on a
number of factors such as cell design, operating conditions, polymer coating, and analyte.
However, we do not mean to imply that coated SAW device responses are inherently
nonlinear. Snow and Wohltjen [12] demonstrated linearity for a poly(ethylene maleate)
coated device exposed to 100-1000 ppm of cyclopentadiene. Frye et al. [45] showed that
polystyrene coated SAW device responses were linear over a wide range of methanol
partial pressures. Bowers et al. [39] reported that the response of a triethanolarnine
coated resonator to HCl was linear in the concentration range of 018-1 ppm. Figure 3.8
also shows that the poly(isobutylene) coated device is more sensitive than the uncoated

sensor. Similar response enhancements have been observed by a number of investigators.

Beat Frequency (Hz)

69

 

A
O
O
O
l

 

Figure 3.8.

~
-
*

 

Fl' I I T T r T T fl I I I

1
1000 2000 3000 4000 5000 6000 7000

ug of Hexane

Calibration curves for hexane injected on a blank (Cl) and
poly(isobutylene) coated (A) device.

 

S"
u:

PWSQMPPN?‘

y—a
.9

11.
12.
13.
14.
15.
16.
17.

18.
19.

20.
21.

22.

23.

24.
25.

70

References

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1458.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51 , 1465.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51 , 1470.

Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 940A.
Grate, J. W.; Martin, S. J.; White, R. M. Anal. Chem. 1993, 65, 987A.
Ballantine, D. 8., Jr.; Wohltjen, H. Anal. Chem. 1989, 61, 704A.

Fox, C. G.; Alder, J. F. Analyst (London) 1989, 114, 997.
Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261.

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

Ballantine, D. 8.; Rose, S. L.; Grate, J. W.; Wohltjen, H. Anal. Chem. 1986, 58,
3058.

Wohltjen, H. Sens. Actuators 1984, 5, 307.

Snow, A. W.; Wohltjen, H. Anal. Chem. 1984, 56, 1411.

Frye, G. C.; Martin, S. J. Appl. Spectrosc. Rev. 1991, 26, 73.
McCallum, J. J. Analyst(London) 1989, 114, 1173.

Katritzky, A. R.; Offerman, R. J. CRC Anal. Chem. 1989, 21(2), 83.
Carey, W. P.; Kowalski, B. R. Anal. Chem. 1986, 58, 3077.

Carey, W.P.; Beebe, K.R.; Kowalski, B.R.; Illman, D.L.; T. Hirschfeld, T. Anal.
Chem. 1986, 58, 149.

Carey, W. P.; Beebe, K. R.; Kowalski, B. R. Anal. Chem. 1987, 59, 1529.

Grate, J. W.; Rose-Pehrsson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H.
Anal. Chem. 1993, 65, 1868.

Katrinsky, A. R.; Savage, G. P.; Pilarska, M. Talanta 1991, 38, 201.

Nanto, H.; Kawai, T.; Sokooshi, H.; Usuda, T. Sens. Actuators B 1993, 13-14,
718.

Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. S., Jr.; Jurs, P. C. Anal. Chem.
1988, 60, 2801.

Zellers, E. T.; Pan, T-S.; Patrash, S. J .; Han, M.; Batterrnan, S. A. Sens. Actuators
B 1993, 12, 123.

Grate, J. W.; Ballantine, D. 8., Jr.; Wohltjen, H. Sens. and Act. 1987, 11, 173.
Nederlof, A. J.; Nieuwenhuizen, M. S. Rev. Sci. Instrum. 1993, 64, 501.

26.

27.
28.
29.
30.

31.
32.
33.
34.
35.
36.

37.

38.
39.
40.
41.

42.
43.

45.

71

Munoz Leyva, J. A.; Hidalgo De Cisneros, J. L.; Garcia Gomez DE Barreda, D.;
Beerra, Fraidias, A. J. Talanta 1994, 41 , 159.

Ricco, A. J.; Martin, S. J. Sens. Actuators B. 1993, 10, 123.
Suleiman, A.; Guilbault, G. G. Anal. Chim. Acta 1984, 162, 97.
Thompson, M.; Stone, D. C. Anal. Chem. 1990, 62, 1895.

Hecki, W. M.; Marassi, F. M.; Kallury, K. M. R.; Stone, D. C.; Thompson, M.
Anal. Chem. 1990, 62, 32.

Thompson, M.; Stone, D. C.; Nisman, R. Anal. Chim. Acta 1991, 248, 143.
Yan, Y.; Bein, T. J. Phys. Chem. 1992, 96, 9387.

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

Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325.

Edmonds, T. E.; West, T. S. Anal. Chim. Acta 1980, 117, 147.

BB. Townsend IV, J .S. Ledford, and DP. Hoffmann, submitted to Sensors and
Actuators.

E.B. Townsend IV, J .S. Ledford, and DP. Hoffmann, submitted to Anal. Chim.
Acta.

Patrash, S. J .; Zellers, E. T. Anal. Chem. 1993, 65, 2055.
Bowers, W. D.; Chuan, R. L.; Duong, T. M. Rev. Sci. Instrum. 1991, 62, 1624.
Martin, S. J .; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201.

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

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

‘ King Jr., w. H. Anal. Chem. 1964, 36, 1735.

Kindlund, A.; H. Sundgren, H.; Lundstrom, I. Sens. &Act. B. 1984, 5, 1.
Frye, G. C.; Martin, S. J.; Ricco, A. J. Sens. Mater. 1989, I, 335.

Chapter 4

Study of Poly(isobutylene) Coated Piezoelectric Mass Sensors in a Pulsed

Injection System

4.1 Abstract

The interactions of various classes of compounds with poly(isobutylene) coated
surface acoustic wave (SAW) devices and quartz crystal microbalances (QCM) in a
pulsed injection system have been examined. For the pulsed system, a new parameter
(designated “pulse coefficient”) has been developed that is analogous to the partition
coefficient in a continuous flow system. Calculated pulse coefficients were similar in
trend and slightly larger than calculated SAW coefficients and literature SAW and GLC
partition coefficients reported for the same analytes. Using the pulse coefficients, Linear
Solvation Energy Relationships (LSER) were calculated for the SAW and QCM. The
results indicate that a poly(isobutylene) film interacts with analytes primarily through
dispersion interactions. Thus, an enhanced response to alkanes is observed because of the
dominant role of dispersion forces in the intermolecular interactions for both the polymer
and the alkanes. LSER coefficients calculated for the pulsed injection system compare

favorably with values reported for continuous flow SAW and GLC experiments.

72

73

4.2 Introduction

Surface acoustic wave (SAW) devices have been widely studied as mass sensors
[1-16]. Typical devices employ polymer coatings that are able to interact selectively with
the analyte of interest. If this coating is a soft, amorphous polymer above its glass
transition temperature, then the frequency shift of the sensor resulting from the addition

of a thin film coating is as described by Wohltjen [11]:

 

2 2 411 1+9
Af= k +k f h —k f h 4.1
(1 2)o P 20 [V122 [1+2ul] ( )

where Af is the frequency shift (hertz), f0 is the fundamental frequency of the device
(hertz), k, and k2 are material constants of the ST cut quartz substrate (-8.7 x 10‘8 and -
3.9 x 10‘8 mzs/kg) [17], h is the film thickness (meters), p is the film density (kg/m3), u
is the shear modulus of the film material (N/mz), A is the Lame constant, and V, is the

Rayleigh wave velocity in the substrate (3300 m/s). The first term provides the frequency
shift due to mass loading and the second term details the contribution from the
mechanical properties of the film. If the shear modulus of the coating is small compared
to the square of the wave velocity (411/V§< 100), then the second term is considered

negligible and equation (4.1) simplifies to the more familiar expression:

Af = (k1 +k2)r3hp (4.2)

 

74
This equation assumes that the film is an isotrOpic, nonconducting,
nonpiezoelectric polymer film held above its glass transition temperature with a thickness
less than one percent of the acoustic wavelength [11]. Similar expressions have been
derived for the effect of vapor sorption on the frequency of the device.
Grate et al. [9] used partition coefficients to describe vapor sorption into materials

used as SAW coatings. The frequency shift upon vapor sorption, Afv, and the partition

coefficient, KSAW’ are related by the following expression:

AfszKSAW
Ps

 

A1, = (4.3)

where CV is the vapor concentration, Afs is the frequency shift caused by the polymer
coating, and pg is the density of the polymer coating. This expression assumes that the

response of the device is entirely due to mass loading [9].

The sorption of the vapor into the film depends largely on the solubility properties
of both the film and the vapor [18]. Dominant interaction modes involve dispersion
forces (induced-dipole/induced-dipole), dipole induction interactions (dipole/induced-
dipole), dipole orientation interactions, and hydrogen-bonding. Originally, Hildebrand
solubility parameters were thought to describe these interactions adequately, but more
recently, solvation parameters have been proposed as better models [9,18]. If the
response for several analytes that have different combinations of solvation parameters are
known, the following equation can be used to describe the solubility properties of the

film.

75

log K = c + 320? +b2B£1+llong6 +rR2 +sn§i (4.4)

Bag is a measure of hydrogen-bond acid strength, 2135’ is a measure of hydrogen-
bond base strength, logLl6 describes dispersion interactions, R2 is a quantitative measure
of the ability of a solute to interact with a solvent through n and 1t electron pairs, and rt?

measures the ability of a compound to stabilize a neighboring charge or dipole. A more
detailed description of each variable and collections of solute parameters may be found in
the extensive work of Abraham et al. [18-36]. From expression (4.3), KSAw values can be
calculated for each analyte and film. Once KSAw values and the appropriate solvation
parameters are known, linear solvation energy relationships (LSERs) can be developed to
determine the values of the coefficients a, b, l, r, and s in equation (4.4). The relative
magnitude of these coefficients describes the contribution of specific analyte/film
interaction mechanisms to analyte partitioning [10,18].

In previous work, continuous flow sample introduction systems have been used to
collect stw data for comparison with KG”; values [9,37]. For the same temperature,
logKSAw values were greater than logKGLc. This was attributed to differences in vapor
concentration, temperature, and/or changes in the viscoelastic properties of the polymer
film induced by vapor sorption [37]. Kcm determinations were also combined with
solute solvation parameters to explain phase interactions for a SAW coating [9,18]. More
recently, Patrash and Zellers [38] have used a continuous flow system to examine the

behavior of SAW devices coated with OV-275, OV-25, poly(phenyl ether), and

76
poly(isobutylene). These authors also reported detailed interpretations of SAW responses
in terms of boiling point, solubility parameter, and LSER models.

Pulsed sample introduction has also been utilized widely in the study Of mass
sensors [18,39-45]. Janghorbani and Freund have successfully used a QCM as a GC
detector [44]. Edmonds and West have used pulsed sample introduction in combination
with QCMs coated with GLC phases to detect various organic species [45]. Grate and
Abraham have pointed out that this type of sample introduction can be used in
conjunction with laser or thermal desorption [l8]. Grate et al. [46] have used thermal
desorption of analyte from preconcentrator tubes to deliver pulses of gas to an array of
SAW devices. While the specific details of pulsed introduction methods may vary, the
analyte of interest usually reaches the SAW device as a plug, interacts with the coated
surface, desorbs, and is rapidly swept out of the test cell.

In continuous flow sample introduction systems, the response is usually measured
after the film has equilibrated with the analyte. The response magnitude is believed to
depend only on the partition coefficient of the analyte into the film and any modulus
changes that accompany analyte sorption. The pulse measurement is a dynamic one that
may convolute analyte diffusion with partitioning phenomena and thus the applicability
of equilibrium solubility models is questionable. Indeed, recent work has emphasized the
importance of the response kinetics of quartz crystal microbalances when challenged with
a vapor [47]. Freeman et al. [47] have shown that on a QCM, the rate constants for the

response of AT100, OV-101, Carbowax 20M, and ethyl cellulose coatings to the same

77
vapor were different. It was also shown that the response of an ethyl cellulose coated
sensor to three different vapors resulted in three different rate constants.

In this work, the response of poly(isobutylene) [abbreviated PIB] coated mass
sensors to various analytes containing a variety of functional groups has been determined
using a pulse injection system. To compare the results from the pulsed system to the more
conventional continuous flow system, a new parameter, termed the “pulse coefficient”

(PSAw or PQCM), is calculated using equation 4.5.

AfSCVPSAW
13s

Af:

V

 

(4.5)

This new parameter is analogous to the more traditional KSAw. However, the pulse
parameter does not represent a partition coefficient and may actually be a combination of
contributions from both vapor/film diffusion kinetics and partitioning effects. We would
like to determine if the film response kinetics reported by Freeman et al. [47] play an
important role for our pulse measurements. Finally, the pulse coefficient is used in
combination with analyte solvation parameters to develop LSER expressions for the PIB

coated mass sensors.

4.3 Experimental

Materials. Poly(isobutylene) (MW = 380,000) was obtained from Aldrich
chemical company. Analytes were obtained from either Aldrich or Baker chemical
companies and used as received Table 4.1 shows the boiling points and solvation
parameters for the analytes examined in this study. Analytes were chosen to examine

analyte/film interactions for molecules with a wide range of chemical functionalities,

78

 

 

8d 8o.o NM: .4 8d 8d 5 65:8 8
2.8 28 8m.“ 88 :8 5 22% a
8d 8o.o Red 8.0 8.o e2 238 M:
88 88 8mm 88 8d 5 683588228 2
85 $3 83 :8 8d ed 63882.8 2
as 83 52 was Rd E 3.58; 2
.38 See 83 and 88 e: 688.3 3
mg See 83 .2 .o 88 n _ _ 6839 2
88 m: .o 22 Ea 8d 2: Basia 2
$8 83 83 3o «8 82 cases _ 2
8d 8o.o m: .m 85 8d 8 Bean: 2
98 ammo 83 88 8d 5 68382285 a
one See 23 28 8d 4m 2.6825 w
one 28 82 one 98 8 358.82 a
2 .o none 33 8d 8d 5 655836 e
N3 23 82 E .o 88 8 83:8 m
8.0 8o.o 83 8d 8d 8 use: a
$8 8.8 83 :8 .98 me 6552: m
88 ES 83 $8 85 en 2268 N
«do 82 on: 28 85 3 Sea _
we we Eda Maw new 6°C 38 285. 88:2
22:9 Engage—om 8036885 fiwaBm 8mm Ewegm Eo< mam—Em
8886.5 88-x 28-:

 

.8m-w: x83 £5 5 e8: mcouoEaSn >328 gag—Om 05 new .353 mazes moi—«ea we 85

44V 2an

79

molecular weights, and boiling points. Furan, thiophene, pyridine, and pyrrole were
studied to determine the selectivity for aromatic heterocycles. Benzene, 2—pentanone,
pyridine, l-butanol, and nitropropane were studied because they are used to determine
McReynolds constants for chromatographic stationary phases [48]. In the context of
LSER expressions, the alkanes mainly interact through dispersion interactions.
Trichloroethylene and perchloroethylene interact through dispersion, polarizability, and
dipole mechanisms. Furan, benzene, thiophene, 2-pentanone, toluene, pyridine, and
nitropropane interact through dispersion, polarizability, dipole, and H-bond base type
interactions. Acetone, methanol, isopropanol, water, and l-butanol interact through all
five mechanisms.

Surface Acoustic Wave (SAW) System. Data were obtained using a 200 MHz
SAW resonator array microbalance (Femtometrics, Costa Mesa, CA) that has been
described previously [49]. The array device consists of 6 individual resonators mounted
on a single stage that is directly inserted into the circuit board. The frequency difference
between each resonator and the separate reference device (beat frequency) is the
measured signal. Power is supplied and frequency counted by an external unit provided
by the manufacturer. The SAW device exhibits a 904 Hz frequency shift when perturbed
by a 1 nanogram mass change.

A schematic diagram of the 13.5 mL cell and experimental apparatus is shown in
Figure 4.1. 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 pneumatic

80

 

Pulse
Introduction

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.1. Schematic diagram of the SAW sensor test cell. The sample is introduced
as a pulse and then split equally into 6 streams. Each of the streams is
delivered directly onto the SAW surface. (a) is one of the six delivery
arms, (b) is the mixing chamber, and (c) is one of the 6 resonators
mounted on a gold stage.

81
pressure is provided from a standard laboratory gas cylinder, while purified gas with

controlled flow rates are supplied to the injector by Porter Mass Flow Controllers. Both
the actuators and the cell use nitrogen gas. The temperature of the test cell was controlled
by resistance heating of the transfer line. Transfer line temperatures are approximately

120°C. Typical operating conditions are 40 cc/min flow of N2 (99.95%, AGA Gas Co.)

dried over molecular sieves and test cell temperatures of 30°C. Normally, six injections
of each analyte were made on at least three devices to ensure reproducibility. Typical
RSDS over all devices were 10%.

The vapor generation system and operating procedures have been detailed
elsewhere [50]. Analyte concentration is controlled by adjusting the relative flow rates of
carrier gas through a bubbler containing the analyte and a dilution gas line. Total
flowrates of dry N2 carrier gas (99.95%, AGA Gas Co.) are maintained at 100 cc/min.
Vapor concentrations are measured using an in-line 5 cc sample loop connected to a
Varian 3700 GC equipped with a flame ionization detector. Each run consists of a two
minute nitrogen purge, vapor exposure for a period of time necessary for the SAW
response to equilibrate, and a nitrogen purge equal to the time necessary for vapor/film
equilibration. Data are collected at 2 Hz resolution. Poly(isobutylene) films are prepared
by spray casting from dilute toluene or chloroform solutions (~500 ppm): Typical film
thicknesses were 100 kHz, which corresponds to 0.012 pm. The devices were cleaned by
rinsing with acetone and methanol, soaking in isopropanol, and drying in air.

SAW Data Acquisition and Manipulation. The beat frequency of each

resonator was collected using a digital counting board (MetraByte, Taunton, MA) and

82 .
stored in a Gateway 2000 386-25 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 standard software has been modified by
Femtometrics to allow for timed, multiple runs that fully utilize the capabilities of the
Varian autoinjector. In this study, data analysis was performed using custom software
written using LabView for Windows (v2.52, National Instruments, Austin, TX). This
software separates the data output file into individual channels and displays the response
of all 6 channels simultaneously on a single page. Additionally, the software counts
peaks, determines peak heights, peak areas, and then calculates averages, standard
deviations, and %RSDs. It also allows user selection of either none, Savitsky-Goolay
smoothing, or median filtering of the data, and provides some control over the sensitivity
of the peak determination. The capabilities of the software are described in further detail
elsewhere [51].

Quartz Crystal Microbalance (QCM). Quartz crystal microbalance
measurements were performed using 6 MHz quartz crystal microbalances from Maxtek
Inc. (Torrance, Calif). A model TM-100 Thickness Monitor (Maxtek Inc.) was used to
determine the mass of polymer coating and the response of the coated device to analytes.
The output voltage of the thickness monitor was recorded using a Hewlett-Packard
3394A integrator. The 6 MHz device exhibits a 53 Hz frequency shift when perturbed by
a 1 microgram mass change.

The QCM sensor test apparatus shown in Figure 4.2 is similar to the SAW

apparatus except that the volume of the cell is 8 mL. Analytes were delivered to the cell

83

 

( \
Pulse
Introduction

k J

 

 

 

 

 

Figure 4.2.

Schematic diagram of the QCM sensor test cell. The QCM is positioned
in the center of the cell, and the sample is introduced as a pulse and then
split equally into 2 streams. The streams are delivered to the left and the
right of the device.

84

as a pulse using a gas chromatograph injector operated at 240°C. Typical operating

conditions are 20 mL/min flow of N2 (99.95%, AGA Gas Co.) dried over molecular sieve

and a test cell temperature of 30°C. The temperature of the test cell was controlled by
resistance heating of the transfer line. For the QCM measurements, at least three
injections of each analyte were made to ensure reproducibility. For the blanks and each
film, data were collected on at least three devices. The QCM data have typical RSDs of
6%.

QCM crystals were cleaned following procedures similar to those used for the
SAW devices. Cleaned crystals were placed in the oscillator circuit and the thickness
monitor set to zero. Crystals were spray coated with the polymer solution and allowed to
dry in air. Films of approximame 0.038 um thickness (~3.54 pg) were used in this
study.In both types of acoustic measurements, analyte injection order was varied to
minimize the effect of analyte order on sensor response. Mass sensor responses are
reported as the maximum frequency shift in hertz per micromole (QCM) or nanomole
(SAW) of analyte injected to normalize the response for each analyte. Responses are
plotted against analyte boiling point to determine the effect of condensation on sensor
response.

Linear Salvation Energy Relationship (LSER) Calculations. Linear solvation
energy relationships were developed from equation (4.4) by multiple linear regression
analysis using the analyte solvation parameters (Table 4.1) and the pulse coefficients

calculated according to equation (4.3). Note that in most of this study PSAw or PQCM is

85

calculated instead of stw. The vapor concentration (CV) used in equation (4.3) is taken
as the mass of analyte injected divided by the volume of the test cell. A Gateway 2000
4DX2-66V computer running Microsoft Excel (Microsoft Corporation, Redmond, WA)
was used to calculate the multiple linear regressions. The solvation parameters not found
in the literature were calculated in a manner consistent with those suggested by Abraham

et al. [19-36].

4.4 Results and Discussion

Response of Uncoated Mass Sensors. QCM. Figure 4.3 shows the variation in
sensor response as a function of analyte boiling point for the uncoated QCM device.
With the exception of trichloroethylene (bp=87°C) and perchloroethylene (bp=12l°C),
the response of the uncoated QCM increases steadily with analyte boiling point. The
large responses observed for trichloroethylene (T CE) and perchloroethylene (PCB) may
be explained in terms of the surface structure of the QCM devices used in this work.
Much of the QCM surface is covered with a gold film used to make electrical contact
with the quartz piezoelectric crystal. It is well known that chlorine interacts strongly with
gold. Thus we attribute the high responses measured for TCE and PCB to the strong

interaction of the chloride substituent with the gold surface.

86

 

 

 

8

6.
E
C
E d
1
E
:5 4"
i E
a ' %‘EE
0‘ E
E
U 2-
C7 3: I:

I: II
~ 1:
a; :0:
:e:
ae= 9%
O U I U T I l I I '1 '

 

l I
20 4O 6O 80 100 120 140 160

Analyte Boiling Point (°C)

Figure 4.3. Variation of uncoated QCM device response (Hz/umole) versus analyte
boiling point.

87

SAW. The variation in sensor response as a function of analyte boiling point
measured for the uncoated SAW device is shown in Figure 4.4. With the exception of
isopropanol (bp=82°C), water (bp=100°C), and 1-butanol (bp=117°C), the response
generally increases with analyte boiling point. Isopropanol, water, and l-butanol exhibit
larger frequency shifts than analytes of comparable boiling point. This enhanced
response may be attributed to the attraction between these polar molecules and the polar
SAW device surface ($02), and the fact that alcohols tend to self associate. Martin et al.
[52] have also shown that polar molecules bind more strongly to a polar quartz surface
and that analyte-analyte interactions (self association) are significant forpolar molecules.

The general trend of increase in frequency shift with increasing boiling point
observed for the QCM and SAW devices indicates that the sensor response is strongly
affected by analyte physisorption. King [53] proposed that analytes with higher boiling
points are able to achieve a higher level of partitioning into a film. While it is not
possible for partitioning to occur on an uncoated device, it is reasonable to expect
condensation effects to produce similar results. Patrash and Zellers [38] have also shown
that an increase in analyte boiling point increases the SAW response in continuous flow
experiments.

Response of Poly(isobutylene) Coated Devices. QCM. Figure 4.5 shows the
variation in sensor response as a function of analyte boiling point for the PIB coated
QCM. The response trend is similar to that of the uncoated device; however, the
response magnitudes are approximately three times larger. Table 4.2 shows the PIB

coated/blank QCM response ratio calculated for each of the analytes. Responses for

88

 

 

3-J

SAW Response (Hz/nanomole)

E E

E 11,:

=B= #3. Q
O-J I U I f I I I I l I I I l
20 40 6O 80 100 120 140 160

 

 

 

Analyte Boiling Point (°C)

Figure 4.4. Variation of uncoated SAW device response (Hz/nanomole) versus analyte
boiling point.

89

hydrocarbons such as hexane, cyclohexane, octane, and nonane are significantly enhanced
for the PIB coated QCM. Isopropanol, water, and l-butanol responses are similar to the
values measured on the blank, while methanol is significantly lower.

As noted for the blank QCM (Figure 4.3), the responses measured for TCE and
PCB are significantly greater than values observed for analytes of comparable boiling
point. However, these results cannot be attributed to adsorption on exposed gold surfaces
since the measured frequency shifts are more than three times those observed for the
blank (Table 4.2). Equilibrium data (vide infra) collected for TCB and PCB using SAW
devices do not indicate that these analytes have anomalously high diffusion rates in the
PIB film. Unfortunately, the exact reason for the high response to TCE and PCB is not
clear and these findings are under further investigation.

SAW. Figure 4.6 shows the variation in sensor response as a function of analyte
boiling point for the PIB coated SAW device. The response for nonane (78
Hz/nanomole) has not been included in the plot in order to emphasize the response trend
obtained for analytes of lower boiling point. As noted for the blank, the response of the
PIB coated SAW generally increases with increasing analyte boiling point. However, the
alcohol and water responses are no longer enhanced relative to the other analyte
responses. In fact, only perchloroethylene and nonane could be considered enhanced
relative to compounds with similar boiling points. Table 4.2 shows that the average
response of the PIB coated SAW device is approximately nine times that of the blank

device. The responses measured for acetone, methanol, isoprOpanol, l-butanol, and

9O

3O

 

20-

15- E

QCM Response (Hz/ umole)

 

 

 

10-
‘ :0:
5" II DIE: n:
, :9 1% no:
.9. :9: o- 0-
.9.
O U l ' Ifl-U I I I I l U

I T l
20 40 6O 80 100 120 140 160

Analyte Boiling Point (°C)

Figure 4.5. Variation in the response of poly(isobutylene) coated QCM device
response (Hz/pmole) as a function of analyte boiling point.

91

00$" mm Omdn $3" 3 Oman

 

 

 

 

 

 

one CN.m 0.3 NH 0:0:0:
8.0 3d 0.3 v; 0.050
5d 84. no.~ 0.0 N: 0:800
omfi 8.: bm.m 0.0 wd 053508020000
NON E .m and ad m.— 0:008:02:
vw.m mw.m mod N; A: 3:82.-—
m_.v 3N mé 0; 0520.8
9mm cod 3d o.n fin 0:028
and and n3 w; 0:0:8:0Q-N
9am Ema ad ad 00.03
vwd cod m.o_ Em 0:800:
Rd and mod c._ _ Wm 0535080305
own vwd ma QM 05.325
ww.~ “Nd ed N; 3:300:02
Nam cod how 0.... 05.8833
wm.m SN v.2 w.m 0:00:09
mud mod and 0.3 m.n 0:88:
2: «Wm o_.~ wd ed 3:052:
va C .N w._ vd 0:9000
3N omd 0.» 9m :83
3200002 35.0002 300.002 22m 0:85 Zoo 0.55 385..
m3 m5 m3 3 3<m mum 0 200 an

 

.8038 3<m 0:0 200 08000 mi 0:0 0:33 05 :8 8:20Eoo0 528:: 0:0 08085000 02:: .0000: 080%0M

.Né 030,—.

SAW Response (Hz/nanomole)

Figure 4.6.

92

 

18-
15-

12-

9‘ i
6- §§+ E

i §+§
3- +

3:

=0 egFo-o
0 III 'U'VIVIU

I
20 4O 60 80 100 120 140 160

 

 

 

 

Analyte Boiling Point (°C)

Variation in the response of poly(isobutylene) coated SAW device
(Hz/nanomole) as a function of analyte boiling point for the pulse injection
(0) and continuous flow (+) methods. The response for nonane (78
Hz/nanomole) has not been included in this plot (see text).

93

nitropropane are comparable to those obtained for the blank device while 2-pentanone
and all of the alkanes (excluding octane) show large response enhancements.

The PIB coated SAW response observed for nonane is more than four times the
value measured for any other analyte examined in this study. The high boiling point of
nonane and the large solubility of alkanes in PIB should lead to considerable sorption of
the molecule into the polymer film. However, extensive nonane sorption cannot
completely explain the high nonane SAW response because we would also expect to
observe a similar enhancement in the QCM data. We believe that the large SAW
response to nonane may be attributed to viscoelastic effects that have been reported for
PIB coated SAW devices [16,17,37]. Martin et al. [16] have shown that vapor sorption
changes the modulii of the film. Ballantine [17] has shown that viscoelastic effects
depend on the amount of analyte sorbed in the film and the thickness of the polymer
coating. Apparently, condensation and sorption during our pulsed sample introduction
results in sufficient nonane content to observe these effects in the SAW data. Similar
phenomena may also occur for analytes with lower boiling points; however, the analyte
concentration in the films must be too low to observe any significant effects.

Previously, Jurs et al. [54] reported that PIB was the most sensitive SAW coating
among 10 coatings tested for isooctane. Our results indicate that there is a large
enhancement .in alkane response from the blank device to the PIB coated device.
However, comparison of alkane responses with values measured for analytes of similar

boiling point suggests that selective detection of alkanes in mixtures would be difficult.

94

Additionally, both Jurs et al. [54] and Grate et al. [37] have shown that for PIB, the
frequency shift of toluene was similar to isooctane. In our work, the response to octane
(bp=126°C) is larger than toluene. However, interpolation using our results (Figure 4.6)
predicts that the responses of isooctane (bp=98°C) and toluene would be similar.

Jurs et al. [54] also reported that the frequency shift caused by water was similar
to l—butanol (both of which gave opposite difference responses). In our data, the
magnitude of the water response is low (as noted by Jurs et al. [54]), but the l-butanol
response is greater than that due to water. The most likely explanation for the difference
between the 1-butanol response reported in our work and by Jurs et al. [54] is that they
used an uncoated delay line as the reference, while we use a sealed device. Jurs et al.
[54] reported that their reference signal changed significantly in response to some vapors.
Since PIB does not sorb most vapors strongly, they noted that their difference
measurements for PIB may be less accurate.

In order to determine if the trend in SAW responses is affected by the method of
sample introduction, eight analytes selected from different regions of the response curve
were presented to the sensor under continuous flow conditions in order to obtain their
equilibrium response. These results are also presented in Figure 4.6 (crosses). For the
analytes selected, the responses measured using pulsed sample introduction are similar to
those obtained from the equilibrium method. This suggests that the analyte diffusion
rates in PIB are comparable and thus the method of sample introduction (pulse or

continuous) has little impact on the response trend.

95

Evaluation of LSER Expressions. Pulse Coefficients. Table 4.2 shows the pulse
coefficients calculated for the PIB coated QCM (PQCM) and the pulse and selected
partition coefficients calculated for the PIB coated SAW (PSAw, KSAw). The pulse
coefficients for the SAW are larger than those calculated for the QCM. In general, the '
magnitude of the coefficients increases with increasing boiling point and magnitude
changes are similar for PQCM and PSAw. We attribute the differences between PQCM and
PSAw to the location of the sensor device relative to the inlet gas stream and the resulting
effect on our calculation of vapor concentration (Cv) in equation (4.3). The SAW devices
are located directly underneath the gas stream, while the QCM is positioned in the center
of the cell. This makes our calculation assumption for Cv more accurate and more
similar to a continuous vapor stream in the QCM system, while the actual vapor
concentration above the SAW at the time of the measurement maybe much larger than
our assumed concentration. This would result in higher PSAw values for our
measurements.

Comparison of the pulse and selected partition coefficients determined in this
work with the partition coefficients for similar vapors reported by Grate et al. [9,37] and
Patrash and Zellers [38] is shown in Table 4.3. In most instances, the relative magnitudes
and trends of the coefficients are similar. However, the absolute magnitudes of the pulse
coefficients are greater than the partition coefficient values. We believe that this can be
attributed to the fact that the hot vapor (due to the heated injector) presented in the pulsed
experiments is more soluble in the film than the room temperature vapor presented by

continuous flow.

96

.92 :0 005030 0:09: .003 :0 005050 0:0Q0 20:20:80:

220 0:02 80 8 as: 05 000 a? 085% 300.. 08:30 3% £2 02 8 as: 0:0: 80 5:3 02.3% 5000

 

 

 

 

 

3.0 00.0 00.0 2082
00.0 2.0 :0 0.0.0 2.0 $0 3033.:
2.0 2.0 8.0 00:00::
£0 02.0 .0000 00.0 00.0 8.0 3.0 0820.:
00.0 :00 000 00.0 053588300:
20 00.0 :00 65:88.:
3.0 00.0 00.0 8002.220
8.0 2.0 $0 835:
000 2.0 8.0 00.0 830::
00.: 00.: 00.0 2.0 655...:
20000002 3:000:02 23000002 23%0002 $20002 2200.002 2:05.
0.00 e a: 0:: 0.00 e 0:: 0.00 e a: 0.00 e a: 00:0 0 a:
05.3 .3 :0 3.20 0:02 02:00 0 0300: 0.53 0.:

 

0:30:05 05 80:: 0020:, 3330.: 0:0 3,330.: 5:3 Ema—mo: 0:0 £00.33 .3<mmw04 :0 50:09:00 m4. 030:.

97

LSER Coefiicients. Table 4.4 shows the results of the multiple linear regression
analyses performed on data obtained from both the PIB coated QCM and SAW devices.
Initially, regressions were performed using all the analytes. This resulted in LSER
equations with very large a coefficients, suggesting that the film acts as a hydrogen bond
donor. Since PIB is a nonpolar polymer, one would not expect hydrogen bond donation
to be an important interaction mode for the film. Closer examination of the data reveals
that the alcohols may be unfairly influencing the fit. The influence of these polar
hydrogen bonding compounds can be seen in their high on and low dispersion solvation
parameters. Grate et al. [37] have also reported that alcohols incorrectly influence
regression analysis and removed alcohol data to produce improved regressions. Patrash
and Zellers [38] reported that alcohols deviated significantly from predictions based on
the boiling point model and performed a separate regression analysis for these analytes.

We have also examined the effect of removing alcohol data from our regression
calculations. Following the removal of the alcohol data, the a coefficient is statistically

insignificant (Pstat=0.06). Furthermore, the improved correlation coefficients (0.91) for

both the SAW and QCM regressions, despite utilizing a relatively small number of
analytes, demonstrates the influence of the alcohols on the fit. These fits indicate that
PIB cannot act as a hydrogen bond acceptor, solute polarizability is not important, and
dispersion interactions are the main mode for analyte/film interactions. Similar findings
were reported by Grate and Abraham [18].

Table 4.4 also shows a comparison of the film coefficients determined in this

study with similar work in the literature. Patrash and Zellers [38] found that both the

98

2:05:80: 5 0:0 0030:, 000:: 0

 

 

 

2:: 00:: $0.: 30:- 80:- 6.0.00: 823::
:00: 0.. .0 232$.
8.: 0: .: E0: 5:. 6.8: 222::
0:: :.: 2:: 3:. 60:0: 828::
:0: 0a a 23:20:
0:0,: 000.: 8.: 00.: :3. 00: 08:02: 6.3 0::
$0.: 08.: 20.: 20:. :0... 50: :3: 6.00: a:
. :02 200:0 05. 083:
0:: x Z: 0:: x S: 0:: x 0.:
S: I l 3.: I 8.: 9.: 2002 :5 0 2:0
30:: 0.0:: 20:: a 0.0 $8.9 $2 0 0.0 0:: x 0.0 .
2.: 3.: 2.: :..: 00:. 0.0.: :2 8:020»: :30:
£30: :00
9:: x 0.. 0 0:: x 00 38::
0:: I I 0.0.: I 8.0. 8.: 2:02 :5 a 0:0
59 00:: 0:: x :0: 60:: 9:: x 3: 88::
00.: 2.: 8:- 5.: 8.: :0 0:: 8:800: .000:
0:30:03.
8205000 :0:0_0::00 : : ~ 0 0 0 m0t0m

 

00:22:80: :05030: 0:20:20 0:0 0:00 22m 0:0 200 05 $50 02.: mmmq :0 33:3:

6.0 030:.

99

polarizablity and dipole coefficients did not make significant contributions to the fit and
did not include their values. Comparing our data at 30°C to that of Patrash and Zellers
[38], we find that there is a consistent difference between the dispersion terms in each
case. Abraham et al. [27,28] performed LSER calculations for squalene, which may be
considered a close GLC analog to PIB. In one instance, Abraham et al. [27,] found that
squalene did not have significant on or [3 contributions, but later [28] reported significant
or contributions for squalene. When comparing our values to those of Abraham et al.
[27,28], it is important to emphasize that squalene is simply an approximation to PIB, and
that. the values for squalene were obtained at much higher temperatures. However,
comparison of the data still indicates that both expressions predict dispersion to be the
major mode of interaction. Other variations in the calculated LSER coefficients can be
attributed to differences in the number and classes of compounds in the analyte sets.

Table 4.5 shows the calculated contributions from the common statistically
significant parameters to the logKgAw and logPSAw term for analytes present in both our
experiments and those of Patrash and Zellers [38]. The sum of the calculated
contributions for the analytes do not equal 100% because not all of the contribution are
presented. There should be some differences in the percent contribution from each type
of interaction because the styles of the fits are different. Patrash and Zellers [38] fit only 4
parameters in the LSER expression. Nevertheless, the dispersion interactions for the

hydrocarbons agree to within approximately 12%.

100

 

 

Table 4.5. Comparison of the contributions from each of the common statistically
significant interaction parameters to logKSAw and logPSAw for
poly(isobutylene).

% contribution to logPSAw and longAw from
Analyte This worka Ref [38]b Th” Ref [38]b
worka
Methanol 36 20 33 44
Hexane 0 0 76 84
Isopropanol 24 12 53 65
Benzene 0 0 72 83
Cyclohexane O O 75 88
Trichloroethylene 0 O 74 85
Toluene O O 94 85
Pyridine O O 64 78
l-butanol 20 10 52. 73
Nonane O O 75 90 .

 

 

 

aUsing the LSER fit obtained at 30°C with 100 kHz PIB films on 200 MHz resonator

SAW devices. This work determined that b, r, and 5 terms are not significant. bUsing the
LSER fit obtained at 25°C with 200 kHz PIB films on 158 MHz SAW devices. Patrash
and Zellers determined that both the r and 3 terms are not significant.

101

4.5 Conclusion

Our work has shown that mass sensor responses obtained using a pulsed sample
introduction technique can provide meaningful chemical insight into analyte/film
interaction mechanisms for PIB films. In the absence of other complicating factors, we
anticipate that similar success can be achieved in.the study of any coating provided the
relative analyte diffusion rates in the material of interest are comparable. Pulsed response
data can be used to calculate pulse coefficients that are analogous to partition coefficients
obtained from equilibrium data. We have shown that these pulse coefficients can be used
to determine useful LSER expressions. This was demonstrated by examining the
calculated contributions and types of interactions suggested by LSER coefficients. These
contributions are similar to results obtained by others using equilibrium data. The
calculated LSER expression for the film, used in conjunction with the analyte parameters,
clearly shows the major mode of analyte/PIB interaction to be dispersion forces. This is
also consistent with previously reported research. The magnitudes of the pulse
coefficients are larger than partition coefficients determined in this study and in the
literature. This variation may be attributed to our different method of calculation for the
analyte vapor concentration and the increased solubility of hot gases in the film. Finally,
it should be noted that this experimental approach is simple, convenient, and capable of
providing fundamental insight into analyte/film interactions that occur in practical pulse

sampling applications of mass sensors.

:5
as

PWNP‘MPP’P?‘

11.
12.
13.
14.
15.
16

17.
18.
19.
20.
21.

22.
23.
24.

25.
26.

102
References

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1458.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51, 1465.

Wohltjen, H.; Dessy, R. Anal. Chem. 1979, 51 , 1470.

Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 940A.
Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 987A.
Ballantine, D. 8., Jr.; Wohltjen, H. Anal. Chem. 1989, 61, 704A.

Fox, C. G.; Alder, J. F. Analyst (London) 1989, 114, 997.
Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261.

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

Ballantine, D. 8.; Rose, S. L.; Grate, J. W.; Wohltjen, H. Anal. Chem. 1986, 58,
3058.

Wohltjen, H. Sens. Actuators 1984, 5, 307.

Snow, A. W.; Wohltjen, H. Anal. Chem. 1984, 56, 1411.

Frye, G. C.; Martin, S. J. Appl. Spectrosc. Rev. 1991, 26, 73.
McCallum, J. J. Analyst(London) 1989, 114, 1173.

Katritzky, A. R.; Offerman, R. J. CRC Anal. Chem. 1989, 21(2), 83.
Martin, S. J .; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201.
Ballantine, D. 3., Jr. Anal. Chem. 1992, 64(24), 3069.

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

Abraham, M. H. Pure Appl. Chem. 1993, 65(12), 2503.

Abraham, M. H. Chem. Soc. Rev. 1993, 22, 73.

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

Abboud, J .-L. M.; Kamlet, M. J .; Taft, R. W. J. Am. Chem. Soc. 1977, 99, 8325.
Abraham, M. H.; McGowan, J. C. Chromatographia 1987, 23, 243.

Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1991, 587, 213.

Taft, R. W.; Abboud, J-L. M.; Kamlet, M. J. J. Am. Chem. Soc. 1981, 103, 1080.

Abraham, M. H.; Grellier, P. L.; McGill, R. A. J. Chem. Soc. Perkin Trans. 2
1987, 797.

27.

28.

29.

30.

31.

32.

33.

34.
35.

36.
37.

38.
39.
40.
41.

42.
43.

45.
46.

47.

48.

103

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

Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1991, 587, 229.

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

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.

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

Abraham, M. H.; Duce, P. P.; Grellier, P. L.; Prior, D. V.; Morris, J. J .; Taylor, P.
J. Tetrahedron Lett. 1989, 30, 2571.

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

Abraham, M. H.; Fuchs, R. J. Chem. Soc. Perkin. Trans. 2 1988, 523.

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

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

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

Patrash, S. J .; Zellers, E. T. Anal. Chem. 1993, 65, 2055.
Suleiman, A; Guilbault, G. G. Anal. Chim. Acta 1984, 162, 97.
Thompson, M.; Stone, D. C. Anal. Chem. 1990, 62, 1895.

Hecki, W. M.; Marassi, F. M.; Kallury, K. M. R.; Stone, D. C.; Thompson, M.
Anal. Chem. 1990, 62, 32.

Thompson, M.; Stone, D. C.; Nisman, R. Anal. Chim. Acta 1991, 248, 143.
Yan, Y.; Bein, T. J. Phys. Chem. 1992, 96, 9387.

Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325.

Edmonds, T. 15.; West, T. S. Anal. Chim. Acta 1980, 117, 147.

Grate, J. W.; Rose-Pehrsson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H.
Anal. Chem. 1993, 65, 1868.

Freeman, N. J.; May, I. P.; Weir, D. J. J. Chem. Soc., Faraday Trans. 1994, 90,
751.

McReynolds, W.O. J. Chromatogr. Sci. 1970, 8, 685.

49.
50.
51.

52.

53.
54.

104

Bowers, W. D.; Chuan, R. L.; Duong, T. M. Rev. Sci. Instrum. 1991, 62, 1624.
EB. Townsend IV, Ph.D. Thesis, Michigan State University, 1995.

EB. Townsend IV, J .S. Ledford, D.P. Hoffmann, accepted for publication in Rev.
Sci. Instrum.

Martin, S. J .; Ricco, A. J .; Ginley, D. S.; Zipperian, T. E. IEEE Trans. Ultrason.
1987, UFFC-34, 142.

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

Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. 8., Jr.; Jurs, P. C. Anal. Chem.
1988, 60, 2801.

Chapter 5

An FI‘IR/ATR Study of Sorbent Induced Polymer Swelling: Application to

the Study of Surface Acoustic Wave Sensors

5.1 Abstract

Poly(isobutylene) films coated on surface acoustic wave (SAW) devices and
FI‘IR/ATR internal reflection elements were exposed to nitromethane, isooctane, and
perchloroethylene vapors. A correlation was found between the spectroscopic indication
of vapor induced polymer swelling and the responses of polymer coated SAW devices to
the same vapors. The sorption of perchloroethylene and isooctane caused PIB films to
swell significantly while sorption of nitromethane swelled PIB films to a lesser degree.
This correlates to literature swelling estimates determined for the same or similar vapors.
It was noted that observed polymer swelling occurs nearly simultaneously with analyte
sorption. Finally, both IR and SAW results indicate that the polymer swelling changes

were reversible.
5.2 Introduction

Surface Acoustic Wave (SAW) devices have been widely studied as vapor sensors
[1 -15]. The response of such devices coated with thin elastic films has been described
using an expression that includes contributions from mass loading and mechanical

properties of the film [16]. Initially, the influence of thin film mechanical properties on

105

106
the sorption induced frequency shift of the device was thought to be negligible compared
to mass loading effects. Thus, polymer-coated SAW devices were often considered to be
simply mass sensors. However, it is now widely reported that changes in mechanical
properties contribute significantly to the response of SAW-based vapor sensors
[14,15,17-19].

Vapor sorption expands polymer volume [20] and, essentially, dilutes the film.
This dilution reduces the polymer inter— and intra- chain interactions and decreases
polymer modulus. Grate et al. [17] have suggested that for a polymer coated SAW
device, temperature or vapor induced polymer swelling decreases the film modulus and
SAW device frequency. Recently, Martin et al. [19] have provided extensive insight into
film modulus changes caused by temperature and vapor sorption. They presented models
that describe changes in velocity and attenuation for acoustically thin and thick polymer
film coatings. These authors also described the dependence of the film thickness and film
density on the amount and specific volume of the sorbed vapor.

We are interested in using FUR/attenuated total reflection (ATR) spectroscopy
[21,22] to examine analyte/film interactions that occur in acoustic sensor coatings.
FTIR/ATR is often used for solution analysis. Polymer thin films have been examined
using various types of contact samplers [23]. FTIR/ATR can also be used to study
Polymer films coated onto the internal reflectance element (IRE). In addition, polymer
Coated IRES have been used to study liquid diffusion in thin polymer films [24-31] and

Sorption induced polymer swelling [32].

107

In this chapter, the use of FTIR/ATR to monitor changes in poly(isobutylene)
band intensities caused by the sorption of analyte vapor is reported. These changes are
interpreted in terms of swelling processes that occur in the polymer thin film. The impact
of these swelling processes on SAW device response obtained using both pulsed and
continuous flow sample introduction is discussed. Our choice of analytes has been
guided by previous work [17] with sorption induced poly(isobutylene) swelling. Grate et
al. [17] estimated that nitromethane should produce small _ swelling effects for
poly(isobutylene), while isooctane and dichloroethane should cause large amounts of
vapor induced polymer swelling. We selected nitromethane and isooctane from this set
of analytes and added perchloroethylene for our study. We anticipate that

perchloroethylene should behave similarly to dichloroethane.
5.3 Experimental

Chemicals. Poly(isobutylene) (abbreviated PIB) was obtained from Aldrich
chemical company (MW = 380,000). Analytes were obtained from Aldrich chemical
company and used as received.

Sensor Test Apparatus. SAW Device. Data were obtained using a 200 MHz
SAW resonator array microbalance (Femtometrics, Costa Mesa, CA) that has been
described previously [13]. The array device consists of 6 individual resonators mounted
on a single stage. The frequency difference (beat frequency) between each resonator and
the separate reference device is the measured signal. Power is supplied and frequency

counted by an external unit provided by the manufacturer. The SAW device exhibits a

108
904 Hz frequency shift when perturbed by a 1 nanogram mass change. The SAW array
was contained in a 15 mL brass test cell that has been described previously [33].

Sample Delivery. For pulsed injections, analytes were delivered using a Varian
autosampler Series 8000 and a Varian gas chromatograph injector operated at 250°C.
Typical operating parameters have been given previously [33]. In this study, a 40 cc/min
flow of N2 (99.95%, AGA Gas Co.) dried over molecular sieves and a test cell
temperature of 30°C were used. Normally, six 0.2 uL injections of each analyte
(separated by 5 minutes) were made on at least three devices to ensure reproducibility.

Equilibrium measurements were performed using a vapor generation system
following operating procedures that have been detailed elsewhere [34]. A series of 4 port
and 6 port valves enable the flow to be directed to either the SAW device or the ATR
accessory. Analyte concentration is controlled by adjusting the relative flow rates of N2
carrier gas (99.95%, AGA Gas Co., dried over molecular sieves) through a bubbler
containing the analyte and a dilution gas line. The total flow rate is maintained at 100
cc/min. Vapor concentrations are determined using an in-line 5 cc sample loop connected
to a Varian 3700 GC equipped with a flame ionization detector. Each run consists of a
two minute nitrogen purge, vapor exposure for a period of time necessary for the SAW
response to equilibrate, and a nitrogen purge equal to the time necessary for vapor/film
equilibration.

Equilibrium SAW response data were obtained using 0.02 - 0.04 g/L vapor
streams. To simplify comparison, equilibrium response magnitudes presented in this

paper have been normalized to constant vapor concentration (0.02 g/L). Pulse responses

109
were obtained using 0.2 “L injections of analyte. These injection volumes correspond to
0.009 - 0.02 g/L vapor concentrations if we assume 0.2 ML sample volumes in a 15 mL
cell. Pulse response magnitudes have been normalized to 0.01 g/L vapor concentrations.
We must emphasize that pulse and equilibrium responses cannot be compared directly
since pulse vapor concentrations are never actually constant.

Data Acquisition and Analysis. The beat frequency of each resonator was
collected using a digital counting board (MetraByte, Taunton, MA) and stored in a
Gateway 2000 386-25 personal computer. The software used to collect the data was
provided by the manufacturer. The standard software has been modified by Femtometrics
to allow for timed, multiple runs that fully utilize the capabilities of the Varian
autosampler. The pulse and equilibrium data were collected at resolutions of 100 Hz (1
pth.2 sec acquisition rate) and 2 Hz (1 pt/3 see data acquisition rate), respectively.
Typical RSDs were 10% for the pulse and less than 5% for the equilibrium
measurements.

Data analysis was performed using custom software written in LabView for
Windows (v2.52, National Instruments, Austin, TX). This sofMare separates the data
output file into individual channels 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 then calculates averages, standard deviations, and %RSDs. It
also allows median filtering of the data and provides some control over the sensitivity of

the peak determination.

110

FTIR/ATR. Fl'IR/ATR experiments were performed using a Mattson
Instruments Galaxy 5020 FTIR spectrometer equipped with a wideband mercury
cadmium telluride (MCI‘) detector. FTIR/ATR spectra were collected with 4 cm'I
resolution. A commercial ATR accessory (Specac, Inc.) designed for HPLC detection has
been modified to perform analyte/film studies. The attachment consists of a 500 uL cell
in contact with a ZnSe internal reflection element (IRE). The cell temperature is
maintained at 30°C using a temperature controlled‘recirculating water bath.

Sample Delivery. The fittings of the ATR cell have been altered to accommodate
1/ " o.d. (1/8” i.d.) stainless steel tubing. For equilibrium measurements, the inlet of the
cell was connected directly to an outlet of the vapor generation system and analyte
concentrations were identical to those used in the equilibrium SAW experiments (0.02 -
0.04 g/L). For pulse injections, the inlet of the ATR cell was connected to a Varian gas
chromatograph injector by a 4" long, 1/8" i.d. stainless steel transfer line. The transfer
line was heated using nichrome wire. The injector and transfer line were heated to 250°C
and 120°C, respectively. Analytes were injected into a 20 cc/min flow of N2 (99.95%,
AGA Gas Co., dried over molecular sieves). The injection volume for each analyte (l
pL) was the minimum amount necessary to produce useful spectra; however, due to the
small volume of the ATR. cell, the resulting maximum analyte concentrations were
approximately 100 times higher than in the case of the continuous flow experiments (see
Table 5.2). As in the case of the SAW measurements, we must note that pulse

concentrations are only approximate.

111

Data Acquisition and Analysis. Two software packages supplied by Mattson
(FIRST Macros) were used to collect FI'IR/ATR data. The initial background spectrum
of either the polymer film or the blank crystal was collected using a First software
package. EnhancedFirst 1.6.2 was modified to collect time dependent Fl‘IR/ATR data.
The program used either the spectrum of the polymer coated IRE or a spectrum of the
blank crystal as the background for further data collection. At least one spectrum was
collected before the gas stream containing the analyte was introduced into the cell.
Approximately eighty spectra (each spectrum is the sum of 10 scans taken at one second
intervals) were obtained following the injection of each analyte. For continuous flow
studies, up to 180 scans were collected.

Films. PIB films for the SAW devices were prepared by spray casting from dilute
toluene solutions (~1000 ppm). Typical film thicknesses were 100 kHz, which
corresponds to 0.012 um. Prior to coating, the SAW devices were cleaned by rinsing
with acetone and methanol, soaking in isopropanol, and drying in air. The ZnSe IRE was
drop coated using the identical 1000 ppm PIB solution. The thinnest coating that could
be detected by the FTIR was used. The calculated film thickness was 0.45 um (vide
infra). We have used drop coating on the IRE to facilitate the reproducible preparation of
PIB films. However, it is possible that the different methods of film preparation lead to
small differences in polymer film morphology in our SAW and FI‘IR/ATR experiments.
Ballantine [35] has reported that the morphology of PIB films on SAW devices is

influenced by coating technique.

112

54' R000

SAW Responses. Figure 5.1 shows typical responses obtained from PIB coated
SAW devices exposed to each vapor using pulsed and continuous flow sample
introduction. The responses measured for nitromethane are significantly lower than the
responses observed for isooctane and perchloroethylene, regardless of the sample
introduction method. For continuous flow sample introduction, the frequency shift
measured for isooctane is greater than the perchloroethylene value; however, the pulse
responses for these two analytes are similar.

Since pulse measurements involve a limited analyte/film contact time, the relative
responses for analytes may be affected by the diffusion rates of the sorbates in the
polymer film. In principle, analytes that diffuse more slowly through the polymer could
give artificially low pulse responses compared to equilibrium values. Some insight into
the diffusion rates of sorbates in the film can be obtained from the time dependencies of
sorption and desorption processes measured with the equilibrium vapor streams. Figure
5.2 shows the normalized SAW responses to each vapor measured as a function of time
(tm), where t is the elapsed time after the SAW device beat frequency begins to increase

in response to analyte vapor introduction (Figure 5.2a) or decrease due to analyte vapor

Response

Figure 5.1.

113

 

D

 

Continuous Flow Pulse

 

 

 

_J g _&)

SkHz

500$

 

 

 

 

 

 

Time (3)

Normalized PIB coated SAW device responses to (a) nitromethane, (b)
isooctane, and (c) perchloroethylene using continuous flow and pulsed
injection sample introduction. Equilibrium and pulse responses have been
normalized to analyte concentrations of 0.02 g/L and 0.01 g/L,
respectively.

114

 

 

0.4- 9

Normalized SAW Response

.0
N
r

 

 

(a)

 

0.0 v I ' 1 ' I ' I '

 

10¢

DC

0))

0.8-

0.6-

Normalized SAW Response
t>

 

 

 

 

Time (s) "2

Figure 5.2. Variation in the normalized equilibrium frequency shifts versus time
following t”2 for a PIB coated SAW device exposed to nitromethane (A),
perchloroethylene (CI), and isooctane (O). (a) Rising and (b) Falling
edges.

115

purging (Figure 5.2b). The rising edge profiles (Figure 5.2a) show considerable scatter in
the normalized response to the analytes for t”2 < 53; however, the isooctane response is
always larger than the value observed for perchloroethylene. Responses obtained at
intermediate times (55 < t"2 < 155) suggest that nitromethane equilibrates most rapidly
with the PIB film, followed by isooctane and perchloroethylene. The normalized
responses obtained during analyte vapor purging are shown in Figure 5.2b. As noted in
the rising edge profiles, some scatter in the normalized responses is observed; however,
in most instances the values measured for isooctane are lower than the perchloroethylene.
response. The normalized nitromethane responses are comparable to the values obtained
for isooctane. Finally, we should note that the device frequencies return to within 2% of

the starting values after the analyte is purged from the cell. ‘
FTIR/ATR Spectra. Analytes and PIB. Ideally, in order to use Fl'IR/ATR to
examine the effect of analyte sorption on the intensity of polymer IR bands, a polymer
band that does not significantly overlap with intense analyte bands should be available.
Figure 5.3 shows the IR spectra measured for nitromethane, isooctane, and
Perchloroethylene vapor streams using an uncoated IRE and a PIB spectrum obtained
uSing a polymer coated IRE. Specific band assignments for the major vibrational modes
0f the analytes and PIB are given in Table 5.1. It is evident that there are no
perchloroethylene or nitromethane bands that interfere with the C-H stretching bands
Observed for PIB (3100-2750 cm'l). However, as expected, isooctane has intense C-H
StI‘etch bands that would make it difficult to use this region to examine changes in the PIB

Spfixtrum arising from isooctane sorption. Further comparison of the PIB IR spectrum

116

with the spectra measured for the analytes suggests that the PIB CH2 bending mode (1235
cm") is the most intense PIB band that does not overlap with strong bands from any of
the analytes.

Nitromethane Sorption. Figure 5.4 shows IR spectra obtained with a PIB coated
IRE before (solid line) and after (dashed line) exposure to nitromethane using pulsed
(Figure 5.4a) and continuous flow (Figure 5.4b) sample introduction. It should be noted
that the CH2 bending mode (1235 cm") is the only band in this region that is due
exclusively to PIB. Comparison of the intensity of this band before and during pulsed
nitromethane exposure indicates that vapor sorption has little effect on the PIB band
intensity. The spectrum measured for a PIB film equilibrated with nitromethane vapor
shows a weaker N02 asymmetric stretch feature (1553 cm’l) compared to the pulsed
experiments due to the decreased vapor concentration (see Experimental Section and
Table 5.2). The intensities of the PIB bands were unchanged during the entire vapor
exposure time (approximately 30 minutes). The intensities of the OH stretching bands

for PIB (not shown) are also unaffected by nitromethane sorption.

117

 

(d)

 

 

 

 

g__

(C) _ ‘J

(b) J Mlm .. .__.

(a) I

I n I n I l I 1 I I I L

4000 3500 3000 2500 2000 1500 1000

 

 

Absorbance

 

 

 

 

 

 

Frequency (cm'l)

Figure 5.3. IR spectra of (a) nitromethane, (b) perchloroethylene, (c) isooctane, and
(d) poly(isobutylene).

 

118

x00: :30 m0@

 

£0 :0 :8:

£0 :0 :02

£0 a: :02 £0 a: 32

£0 a: . 0:: £0 a: 2.:

£0 a: 2.: £0 m> :2:

oz as, a: £0 m> :0: £0 m> :50
Noz m> 4:: 0.0 a: :0: £0 0, £0 a.» wow: £0 at, 2::
N02 20> :02 0-0 00> 2: £0 00> 9:: £0 at, 3::

::0:::w:mm< A758 0:0m ::0E:wmmm< 9-88 0:0m 80:83:90. 9.800 0:0m 20:83:04 A783 0:0m
0:050:82: 053500002000: 0:0:00000 350030330:

 

.Fmdfl 0:050:55: 0:0 05350052000: 0:883: A0:0_b0000_vboa :8 00:00 E 00:08 05 .«o 0805:3000: ._.n 030,—.

Absorbance

Absorbance

Figure 5.4.

119

 

 

 

 

 

 

 

 

1300 12(1) 1100

 

 

Frequency (cm‘l)

 

 

 

 

 

 

 

 

 

0.0 k h I 4 l . L A A h
1600 1500 1400 1300 1200 1100
Frequency (cm'l)

(a) IR spectra of a PIB film before exposure to nitromethane (solid line)
and at maximum analyte intensity (dashed line) using pulsed sample
introduction. (b) IR spectra before exposure to nitromethane (solid line)
and during vapor/film equilibrium (dashed line) using continuous flow
sample introduction.

120

Isooctane Sorption. The effect of isooctane sorption on the IR spectra measured
for a PIB coated IRE is shown in Figure 5.5. For pulsed sample introduction (Figure
5.5a), intense bands characteristic of isooctane are observed at 1470 cm'1 and 1369 cm".
In addition, the intensity of the 5, CH2 band of PIB (1235 cm!) decreases significantly in
the presence of isooctane. An intense isooctane band should also be observed at 1392
cm'1 (5: CH2); however, we believe that the sum of the increase in this analyte band and
the decrease in the intensity of the identical mode in PIB produces little change in this
part of the IR spectrum. Equilibrium exposure to isooctane (Figure 5.5b) also leads to the
observation of bands characteristic of the analyte and a decrease in the 5, CH2 band of
PIB. For both pulsed and continuous sample introduction experiments, the spectra
obtained after isooctane was purged from the film were identical to the initial film
spectra.

Perchloroethylene Sorption. Figure 5.6 shows the effect of perchloroethylene
sorption on the IR spectra measured for the PIB coated IRE. As noted for isooctane,
sorption of perchloroethylene leads to a decrease in the PIB 5, CH2 band intensity. The
intensities of the OH stretching bands for PIB (not shown) also decrease on
perchloroethylene sorption. As noted for isooctane, the spectra obtained after
perchloroethylene was purged from the film were identical to the original film spectra.

Figure 5 .7 shows FTIR spectra measured during the sorption of perchloroethylene
into a PIB film under continuous flow conditions. The intensities of the

perchloroethylene bands increase with increasing exposure time, while the polymer bands

121

60%.. .00 030:0 0 :: 00:000.: :05 5000:0080 0535000030000 6w 0.0 0 :0 00000 00003000 0 0:_0> :
0:00:00 :05: w: :0000? 000—00 w:€=0 8000:0050 05 0050 0008080000 00002800 00 030:: :03? 0005. .

 

 

 

 

 

 

 

:- 0.: and. :0 oeoaaoeoseo: 6.: no oeoeoeozoe
2. 0.: 0- 0d 0.000000% NA m0 058000
7 ON 0 ON 0:050:89: mod 04 0:052:02:
m0: 0 20-2 0
0w:0:0 :0 .0330 0:00 0w:0:0 ox: 3&0 0:00 :0m0> o0 .mfizesm 3R0 0:00 0000>
05:2,: .
0005 30E 0:250:00 :0E»~0m 0000800m
A780 mmmC m: .00 0:00

w:_0:0n :0 05 .00 50:00:: 00:0:w00:_ E 0m§:0 05 0: 200008000 @5030 008.200 002.05 .800.» .00 :00t00800 .Nfi 030,—.

122

 

0.3 r (a)

_ “ps—

-__——- "’

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o 5
0 l
g I
.D I
‘3‘ I
2
.160). 1500:1400. 1300‘ 1200‘ 1100
Frequency (cm'l)
i (b)
0.4 " ‘1'.
0
O
5
.D
‘53.
2
'1600 1500 1400 1300 1200 1100
Frequency (cm'l)
Figure 5.5. (a) IR spectra of a PIB film before exposure to isooctane (solid line) and at

maximum analyte intensity (dashed line) using pulsed sample
introduction. (b) IR spectra before exposure to isooctane (solid line) and
during vapor/film equilibrium (dashed line) using continuous flow sample

introduction.

123

 

1.0 (a)

0.6 -

Absorbance

 

 

 

    

.0

    

 

l l I l J; J

J r
1200 11(1) 1000 900 800

. r . ’ 1 4
1600 1500 1400 1300
Frequency (cm‘l)

 

0.6
(b)

0.5 -

0.4 .

Absorbance

 

 

 

 

ii
a
0‘ A

0.

‘un.

 

 

 

oil
I I I a i n .J L I J

0'9601 1500 1400 1300 1200 1100 1000 900 800
Frequency (cm'l)

 

Figure 5.6. (a) IR spectra of a PIB film before exposure to perchloroethylene (solid
line) and at maximum analyte intensity (dashed line) using pulsed sample
introduction. (b) IR spectra before exposure to perchloroethylene (solid
line) and during vapor/film equilibrium (dashed line) using continuous flow
sample introduction.

124

05350802800 .00 E000: :000> 0:280:00 0 0: 000098 :00 Ba 0 :00 080 00 500:3 0. 00 00:00:00 00000: M: fin 0:03”—

A:.::00 >0:0:U0:m
8m 80 82 8: 8S 8mm 88 gm 8%

  

 

 

 

 

 

 

 

 

 

. _ . _ . _ . .:.:
gwfigfia rNo

. w

m

-0: a.

m

9
0::9w5000bfih rod
wd

125

show a corresponding decrease in intensity with time. Similar results were observed for

isooctane and for both methods of sample introduction.

5.5 Discussion

FTIR/ATR Studies. In an FTIR/ATR experiment that employs a polymer coated
IRE, decreases in polymer IR band intensities can occur due to a decrease in the ATR
sampling depth and/or a decrease in the polymer concentration within the sampled
volume. In principle, several vapor induced phenomena could lead to such decreases.
Before examining these possibilities in detail, we must first consider the effect of ATR
sampling depth and PIB film thickness on our measurements. Using refractive indices of

2.4 and 1.505 [38] for the ZnSe crystal and the PIB film, respectively, the penetration

depth of 1200 cm'1 radiation is calculated to be 1.68 pm (0 = 45°). The quantity of
polymer deposited onto the IRE is sufficient to produce a 0.45 pm thick film if it is
assumed that the polymer uniformly coats the ZnSe crystal. However, we must note that
the morphology of the PIB films used in this study is not known and thus it is possible
that the polymer is not evenly spread on the IRE. In such a case, it is likely that the film
thickness is greater (and less) than the penetration depth of the evanescent wave in
several areas of the coating.

Perhapsthe most obvious explanation for our findings is a loss of contact between
the PIB film and the IRE. Since polymer band intensities return to their initial values

after the sorbate is purged from the film, we do not believe that an irreversible loss of

126
polymer/[RE contact (i.e., polymer agglomeration or film lifting) can explain our results.
It also seems unlikely that the PIB film redistributes over the IRE after sorbate is purged
from the film. Indeed, changes in polymer IR band intensities are closely associated with
the presence of sorbate (Figure 7). We believe that the most important vapor induced
phenomena that could produce the IR band intensity changes observed in this study are
refractive index effects and polymer swelling.

Refractive Index Efi‘ects. The penetration depth of the evanescent wave depends
on the refractive index of the sampled medium [22]. For example, if vapor sorption
decreases the refractive index of the sorbate/film system, then the ATR sampling depth
and IR absorbance should decrease. The refractive index of the sorbate/film system

(11 mix) can be calculated from an expression involving the fractional densities of the

sorbate and film and the refractive indices of the pure components:

vap pm
1.lmix = 31.1145. vap + [31%)an (5-1)
vap PIB

where pvap, ppm, nvap, and 11pm are the densities and refractive indices of the pure vapor

and film; p35: and pig, are the densities of vapor and polymer in the film after the
film volume has been increased by vapor induced swelling. Since the refractive index of

perchloroethylene is comparable to that of the PIB film (T able 5.2), we do not anticipate

that perchloroethylene sorption will significantly affect the refractive index of the system.

127
For equilibrium measurements with nitromethane and isooctane, we have used a

calculation method similar to that described by Grate et al. [17] to determine the values of

vap
mix

and pig: used in eq 5.1. The mass of vapor in the film was obtained from the
KGLC values taken from Grate et al. [17] and the experimental vapor concentration (Table
5.2). The swelled polymer volume was calculated by adding the volume of analyte in the
film to the original film volume. The analyte volume was determined from the mass
sorbed and the liquid density. We calculate the contribution of nitromethane and
isooctane to the refractive index of the vapor/polymer mixture to be 8.7 x 10‘4 and 7.7 x
103, respectively. The contribution from the PIB film to the refractive index (assuming a
swollen film) for nitromethane is 1.505 and for isooctane is 1.497. Thus, the refractive
indicies of the mixtures are estimated to be 1.506 (nitromethane/PIB) and 1.505
(isooctane/PIB). These calculations indicate that vapor induced changes in the refractive
index of the system and, consequently, changes in the penetration depth of the evanescent
wave, are insignificant. Thus, we do not believe that polymer band intensity decreases
accompanying vapor sorption are a result of changes in the refractive index of the
analyte/polymer system.

Polymer Swelling. If the polymer film swells upon vapor sorption, the density of
the polymer decreases. For regions of the film that are thicker than the ATR sampling
depth, such polymer swelling would decrease the intensity of the polymer IR bands.

Since refractive index effects are minimal, we believe that the decreases in polymer band

intensities observed in this work are primarily the result of vapor induced polymer

128
swelling. We are not aware of any other similar ATR studies of vapor induced polymer
swelling. However, Balik and Xu [30] have used optical microscopy and FI‘IR/ATR to
study water sorption into a latex polymer. They used an optical microscope to determine
that water swells the polymer and noted a corresponding decrease in polymer IR band
intensities during exposure of the film to water.

Table 5.2 shows the maximum change in. PIB 5, CH2 band intensity during
exposure of the polymer film to the analyte vapors under continuous flow and pulsed
sample introduction conditions. For both vapor introduction methods, isooctane produces
a larger intensity decrease than perchloroethylene and there is essentially no change in
polymer band intensities during the sorption of nitromethane. Table 5.2 also shows the
estimated percent polymer swelling caused by sorption of various concentrations of vapor
reported by Grate et al. [17]. If we assume that perchloroethylene produces a change
similar to that estimated for dichloroethane, the trend in vapor induced polymer IR band
intensity changes is similar to the trend of vapor induced polymer swelling estimates.

FI‘IR/ATR Interpretation of SAW Device Responses. Several models have
been proposed to describe the response of polymer coated SAW devices to vapor
sorption. The boiling point model relates the vapor/film partition coefficient to the
analyte saturation vapor pressure and predicts that the log of the partition coefficient is
directly proportional to the analyte boiling point [39]. Grate and Abraham [6,40]
proposed the use of Linear Solvation Energy Relationships (LSERs) to elucidate chemical
interactions occurring between a vapor and a film. Later, Patrash and Zellers [39]

reported that LSERs could be used to understand and predict the responses of OV—275

129
(poly[bis(cyanoallyl)—siloxane]), OV-25 ([25% methyl] poly(methylphenylsiloxane)),
poly(isobutylene), and poly(phenyl ether) coated SAW devices to a large number of
analytes. We have also demonstrated that pulse responses obtained from a PIB coated
SAW device can be interpreted using LSERs [41]. Each of the previous studies of PIB
[39—41] have shown that this non-polar, hydrophobic polymer interacts with analytes
primarily through dispersion forces.

Scheme 1 compares the analyte response order predicted using the boiling point
and LSER models to the experimentally determined response order. The solvation
parameters used for the LSER calculations are given in Table 5.3. Film coefficients were
taken from Patrash and Zellers [39]. Both models fail to predict the response trend
observed for these analytes accurately. Since the boiling point model does not consider
any specific analyte/fihn interactions, it is not surprising to find that this model does not
predict the correct response trend. In the case of the more sophisticated LSER model, it
must be noted that predictions are appropriate for situations in which the SAW response
is dominated by mass loading. If vapor sorption changes the mechanical properties of the
film, the ability of the LSER model to predict sensor response is questionable. As noted
above (Table 5.2), the IR band intensity decreases indicate that the amount of vapor
induced polymer swelling decreases from isooctane > perchloroethylene > nitromethane.
Thus, for this limited set of analytes, polymer swelling indicated by FTIR/ATR

measurements correlates with the measured SAW device frequency shifts.

130

 

 

 

wmd 000.0 owm.m cod cod can; :N: 05350802800
cod 80.0 00: .m 8.0 8.0 an: mm 800000:
mad Smd Nam: 36 cod Nov: 2: 0:052:80:
:we 0: 2:02 mnw Mew 5:: 0.0 05:: 0:05.
0305 ban—00000: 0:008:00: 5w:0:m 000m 5w:0::m 0mo< 0.00003: w:0_0m
8:02:20 28-: 23-:

 

0:002:00: 80H 3:0:0 :000200 0:0 00:20:: 0.00000: .8500 wfizon £00305 .:0 :03

.m.n 030E

131

Scheme 1. Experimental and predicted order of SAW device response magnitudes.

 

 

Model Order
Measured isooctane > perchloroethylene > nitromethane
Boiling Point perchloroethylene > nitromethane > isooctane
LSER perchloroethylene > isooctane > nitromethane

 

The FTIR/ATR results for isooctane and perchloroethylene are particularly useful
for understanding the relative equilibrium SAW responses to these analytes. Our
FI'IR/ATR data indicate that isooctane swells the PIB film to a greater extent than
perchloroethylene. However, the LSER model predicts that more perchloroethylene sorbs
into the PIB film than isooctane. If the extent of swelling indicated by the FI‘IR/ATR
data is proportional to the change in film mechanical properties caused by analyte
sorption, our results suggest that mechanical effects contribute more to the SAW response
for isooctane than for perchloroethylene. Thus, eventhough more perchloroethylene sorbs
into the film, it appears that the isooctane induced mechanical effects lead to a larger
SAW response for isooctane. Similar results have been reported by Martin et al. [19] in a
thorough study of mechanical effects on the response of polymer coated SAW devices.
These authors showed that mass loading accounts for 40% of the response observed for
low concentrations of pentane using a 97 MHz SAW device coated with a 0.70 um-thick
PIB film. For trichloroethylene, mass loading accounted for 67% of the SAW response.
Thus, effects other than mass loading (e. g., changes in mechanical properties) account for

60 and 33% of the responses to pentane and trichloroethylene, respectively. We believe

 

132
that our findings for isooctane and perchloroethylene are consistent with the work of
Martin et al. [19].

Pulsed vs. Equilibrium Measurements. Many sensor applications require
pulsed sample introduction. However, as noted previously, pulsed sampling may not
allow the analyte to equilibrate with the sensor coating. Indeed, Freeman et al. [43]
reported that an ethyl cellulose coated QCM showed different response rates to different
vapors. They also reported different rate constants for the responses of AT100, OV-lOl,
Carbowax 20M, and ethyl cellulose coatings to the same vapor [43].

We have found that the relative SAW responses of isooctane and
perchloroethylene depend on the sample introduction method (Figure 5.1). It appears that
the pulse isooctane response is lower than expected on the basis of equilibrium
measurements. The sorption and desorption time dependencies shown in Figure 5.2
clearly indicate that isooctane does not diffuse through the PIB film more slowly than
perchloroethylene. Thus we do not believe that the differences between pulse and
equilibrium measurements are an artifact of analyte diffusion rates.

In principle, mass loading and mechanical effects may occur on different time
scales. Frye et al. [44] have reported that the SAW response of a l um-thick
polysiloxane film to a low concentration of methanol consists of a rapid initial frequency
decrease followed by a slower decrease. The initial shift was attributed to rapid diffusion
of methanol through the film, while the slower decrease was ascribed to slow film
processes such as polymer rearrangement/relaxation. In our experiments, the short

analyte/film interaction time for the pulse measurement may limit the contribution of

133

slow processes to SAW response. However, our FI‘IR/ATR studies suggest that PIB
swelling coincides closely with vapor sorption in both equilibrium and pulse experiments.

The apparent discrepancy between the pulse SAW response observed for
isooctane and perchloroethylene and the SAW response expected on the basis of pulse
FI‘IR/ATR indications of film swelling may be a consequence of the differences in film
thickness and analyte concentration used in the two experiments. The PIB films used for
the FTIR/ATR measurements are thicker than the films used for the SAW experiments
(0.45 pm vs. 0.012 tun). In addition, the pulse analyte vapor concentrations required for
the FI‘IR/ATR experiments are significantly higher than the equilibrium vapor
concentrations (Table 5.2). For pulse experiments, the interaction of larger amounts of
vapor with thicker polymer films (FTIR/ATR) may not produce the same phenomena as
in the case of small vapor concentrations and thin films (SAW). In future pulse studies,

we hope to make SAW and Fl‘IR measurements under more similar conditions.

5.6 Conclusion

We have shown that FI'IR/ATR spectroscopy can be used to monitor vapor
induced changes in PIB film structure. Our initial experiments have focused on the
relatively simple study of polymer swelling because the effect of this phenomenon on
SAW device response is of significant current interest. Several logical extensions of this
work are apparent. It has been suggested that more sensitive acoustic sensors can be
constructed by emphasizing the effect of analyte sorption on the mechanical properties of

polymer thin films versus simple mass detection. If suitable polymers for these new

134
sensors exhibit structure dependent IR spectra, FTIR/ATR studies may provide molecular
level insight into the processes that contribute to the device response. Obviously, the
mechanisms and operating parameters of reaction based sensors could be examined in
greater detail using IR. In the study of more highly functionalized polymer coatings,
strong analyte-film interactions could be probed using changes in analyte or polymer IR
bands. Finally, irreversible sensor response caused by either permanent vapor sorption or

altered film structure could be studied.

5.7

9999392"

10.
11.
12.
13.
14.
15.
l6.
l7.

l8.
19.
20.

21.
22.

23.

24.

135

References

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

Wohltjen, H.; Dessey, R. Anal. Chem. 1979, 51(9), 1465.
Ballantine, D. S., Jr.; Wohltjen, H. Anal. Chem. 1989, 61(11), 704A.
Fox, C. G.; Alder, J. F. Analyst 1989, 114, 997.

Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261.

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.

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

Wohltjen, H. Sens. Actuators 1984, 5, 307.

Snow, A.; Wohltjen, H. Anal. Chem. 1984, 56(8), 1411.

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

McCallum, J. J. Analyst 1989, 114, 1173.

Katritzky, A. R.; Offerman, R. J. CRC Anal. Chem. 1989, 21(2), 83.

Bowers, W. D.; Chuan R. L.; Duong, T. M. Rev. Sci. Instrum. 1991, 62, 1624.
Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 940A-948A.
Grate, J. W.; Martin, S. J .; White R. M. Anal. Chem. 1993, 65, 987A-996A.
Tiersten, H. F.; Sinha, B. K. J. Appl. Phys. 1978, 49(1), 87.

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

Bartley, D. L.; Dominquez, D. D. Anal. Chem. 1990, 62, 1649.
Martin, S. J .; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201.

Rogers, C. E. in Polymer Permeability; Comyn, J ., Ed.; Elsevier: New York,
1985; p 11—73.

Fahrenfort, J. Spectrochim. Acta 1961, I 7, 698.

Harrick, N. J. Internal Reflection Spectroscopy, John Wiley and Sons: New
York, 1967.

Internal Reflection Spectroscopy: Theory and Applications; Mirabella, F. M., J r.,
Ed.; Marcel Dekker, Inc.: New York, 1993.

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

25.

26.

27.
28.
29.
30.
31.
32.
33.

34.
35.
36 .

37.

38.
39.
40.
41.

42 .
43.

136

Farinas, K. C.; Doh, L.; Venkatrarnan, S.; Potts, R. O. Macromolecules, 1994,
27, 5220.

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

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

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

Fu, F. -N.; Fuller, M. P.; Singy, B. R. Appl. Spectrosc. 1993, 47, 98.
Balik, C. M.; Xu, J. R. J. Appl. Polym. Sci. 1994, 52, 975.
Marshbanks, T.L.; Ahn, D.J.; Franses, E.I. Langmuir 1994, 10, 276.
Stuart, B. H.; Williams, D. R. Polymer 1994, 35, 1326.

Townsend, E. B. IV; Ledford, J.S.; Hoffmann, D.P., accepted for publication in
Review of Scientific Instruments.

Townsend, E. B. IV, Ph.D. Thesis, Michigan State University, 1995.
Ballantine, D. S. Jr. Anal. Chem. 1992, 64, 3069.

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

Silverstein, R. M.; Baffler, G. C.; Morrill, T. C. Spectometric Identification of
Organic Compounds, 3rd ed., John Wiley & Sons: New York, 1974.

Aldrich Chemical Company.
Patrash S. J.; Zellers, E. T. Anal. Chem. 1993, 65(15), 2055.
Grate, J. W.; Abraham, M. H. Sens. & Act. B. 1991, 3, 85.

Townsend, E. B. IV, Ledford, J. S.; Hoffmann, D. P. submitted to Sensor and
Actuators.

Abraham, M. H. Chem. Soc. Rev. 1993, 22, 73-83.

Freeman, N. J .; May, I. P.; Weir, D. J. J. Chem. Soc., Faraday Trans. 1994, 90,
75 1-754.

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

Chapter 6

Detection of Biphenyl and Pentachlorobiphenyl in Organic Solvents Using

Polymer Coated Surface Acoustic Wave Devices

6.1 Abstract

The responses of uncoated, poly(isobutylene), poly(vinyl alcohol), and OV-275
coated surface acoustic wave (SAW) devices to 4% (w/v) biphenyl or 500 ppm
pentachlorobiphenyl (PCB) in acetone, methanol, or n-hexane have been determined
using pulsed sample injection. Linear Solvation Energy Relationships (LSERs) are used
to interpret uncoated and coated SAW device responses to the solvent and each analyte.
The importance of dispersion forces in the interactions of PIB lead to large responses to
hexane, while the hydrogen bond acid and polarity of PVA and OV—275 lead to large
responses to methanol and acetone. For both biphenyl and PCB, the large response
obtained with the PIB coated device suggests that dispersion forces are important for
these analytes. PCB responses observed with PVA and OV-275 coated devices are not
significantly larger than those obtained with an uncoated device. The nature of the
solvent does not affect the sensor response to biphenyl. However, the response to PCB in

hexane is larger than for PCB in methanol and acetone, regardless of polymer coating.

137

138

6.2 Introduction

Piezoelectric devices such as the surface acoustic wave (SAW) resonator have
been widely employed for the construction of mass sensors because sub-
nanogramchanges in the mass of the device surface produce readily measurable signals.
By themselves, such devices are not chemically selective and thus a coating that interacts
specifically with the target analyte is applied to the surface of the device [1-8]. Ballantine
and Wohltjen [3] have noted that, when a good coating is available, it is possible to detect
vapors at the 10-100 ppb concentration level, with selectivity of 1000:] and a dynamic
range of 34 orders of magnitude. However, it can be difficult to identify reversible
analyte/coating systems that exhibit the requisite high selectivity. In addition, coatings
that show 1000:1 selectivities may be inadequate for the detection of trace levels of one
species in another if both molecules produce significant responses.

Limitations in sensor selectivity have lead to the development of array devices and
Chemometric techniques [9-17] as well as various sample clean-up methods [17,18] in an
effort to improve the quality of mixture analysis. For example, Rose-Pehrsson et al. [10]
used pattern recognition techniques to analyze data obtained from polymer coated SAW
devices exposed to hazardous and nonhazardous vapors as either single or two component
vapors. Their analysis produced unique fingerprints that could be used to detect toxic
organophosphorus compounds in simple mixtures. They also reported that responses to
mixtures were within 30% of the values estimated from the sum of individual vapors.

Zellers et al. [15] also assumed SAW response additivity in order to simulate binary and

139
ternary mixtures using the data of Rose-Pehrsson et al. [10]. They subsequently applied
extended disjoint principal-components regression analysis to improve the identification
and quantification of components in the simulated mixtures.

Most previous reports of mixture analysis employed data obtained with sensors
that had equilibrated with vapor streams. However, many practical applications of
chemical sensors involve the analysis of pulsed vapor samples [9,17-25]. We have
previously described the performance of a SAW array-based sensor test apparatus
designed to analyze pulsed sample streams [‘26]. We have also shown that, under certain
conditions, this method of sample introduction allows the separation of solvent matrix
from analytes that have high boiling points. Similar phenomena have been observed by
Kindlund et al. [18], who reported that water can be separated from organic species using
thermal desorption in combination with a silicone oil preconcentrator. Grate et al. [17]
also observed vapor separation during the application of a smart sensor system consisting
of Tenax preconcentrator tubes and an array of four coated SAW devices. In addition,
these authors noted that the presence of 2-propanol significantly reduced the sensitivity of
four polymer coated SAW devices to the nerve gas soman [17]. This reduction was
attributed to either a decrease in the collection efficiency of the preconcentrator tubes or a
decreased sensOr sensitivity to the test vapor in the presence of solvent, or a combination
of both.

We are interested in the development of portable sensor systems for the detection
of compounds of interest in waste management. Such analyses often require detection of

the target analyte in the presence of excess organic solvent. In addition, typical

140
applications involve analyses of liquid and/or solid samples using various extraction or
thermal desorption sampling methods. Thermal desorption generally produces pulses of
sample vapor, while extraction and sample cleanup approaches require liquid sample
injection. In this study, we have used a pulsed sensor system [26] for the analysis of 4%
(w/v) biphenyl or 500 ppm pentachlorobiphenyl in hexane, acetone, and methanol using
uncoated, poly(isobutylene), poly(vinyl alcohol), and OV-275 coated surface acoustic
wave devices. Biphenyl and the three solvents were selected because of similarities to a
number of environmentally important analytes and solvents, respectively. The films were
chosen to provide a range of analyte(solvent)/fi1m interactions and as such, would be

logical candidates for SAW sensor array applications.

6.3 Experimental

Materials. Poly(isobutylene) (designated PIB, MW =380,000) and poly(vinyl
alcohol) (designated PVA, MW =50,000) were obtained from Aldrich (Milwaukee, WI).
Poly[bis(cyanoallyl)-siloxane] (designated OV-275) was obtained from Anspec (Ann
Arbor, MI). 2,3,3’,4,4’-pentachlorobiphenyl (designated PCB) was obtained from Ultra
Scientific (Kingstown, RI). Biphenyl and all solvents were obtained from Aldrich and
used as received. In this paper, n-hexane, acetone, and methanol are considered solvents,
while the biphenyl and PCB will be referred to as analytes. All of the biphenyl mixtures
were 4% (w/v) biphenyl/solvent, while the PCB mixtures were 500 ppm PCB/solvent.

SAW Test System. Data were obtained utilizing a SAW package available from

Femtometrics (Costa Mesa, CA). This package consists of six individual 200 MHz SAW

 

141
resonator microbalances which exhibit a 904 Hz frequency shift when perturbed by a l
nanogram mass change. The frequency difference (beat frequency) between each
resonator and a separate reference device is the measured signal. Power is supplied and
frequency counted using an external unit provided by the manufacturer.

The pulsed sensor test apparatus has been discussed in detail elsewhere [26]. A
general overview of the apparatus is given in Figure 6.1. Analytes were delivered to the
sensor cell using a Varian gas chromatograph injector operated at 250°C and either a
standard syringe or a Varian autosampler Series 8000. The autosampler actuators and
injector use nitrogen gas (99.95%, AGA Gas Co.). The nitrogen carrier stream used for
the injector was dried over molecular sieves and the flow rate was controlled using Porter
mass flow controllers. The splitting chamber was heated using a Variac and heater
cartridge. Additional heat was applied to the transfer lines through resistive heating using
Nichrome wire wrapped around the lines. The splitting chamber, transfer line, and cell
temperatures were monitored using K—type thermocouples. The lid of the cell is cooled or
heated (depending on the temperature of the transfer lines) with a recirculating water bath
in order to provide the desired sensor cell temperature. The thermocouple used to
monitor the cell temperature was placed near the SAW surface, underneath one of the

transfer line outlets.

142

Injector

 

 

(

 

Splitting Chamber

\

 

 

/// \\\—

Transfer
Lines

 

 

Sensor Cell

 

 

Figure 6.1

Overview of the sensor test apparatus.

 

143

In all cases, the splitting chamber temperature was maintained at 400°C. For the
biphenyl/solvent mixtures, the carrier gas flow rate was 100 mL/min, the transfer line
temperature was 120°C, and the cell temperature was either 30 or 60°C. As we have
described previously [26] cell temperature affects the response profile obtained for
mixtures of low and high boiling point compounds. At 30°C, the response for the
biphenyl/solvent mixtures consists of two peaks: the first peak due to sorption/desorption
of a mixture of solvent and biphenyl and the second peak due to sorption/desorption of
biphenyl that was initially adsorbed on the walls of the cell. For a cell temperature of
60°C, a single peak is observed following mixture injection. Analyses of samples
containing extremely high boiling point compounds such as PCBs are complicated by
severe analyte condensation in test systems held at low temperature. Thus, all PCB
analyses reported in this work are performed using high temperature cells that result in a
single peak following injection of the PCB/solvent mixture. The conditions used for the
PCB analyses were transfer line and cell temperatures of 160 and 75°C, respectively, and
a carrier gas flow rate of 200 mIJmin. We should note that these conditions are close to
the upper limits of our test system. ‘

Polymer Coatings. All SAW devices were pretreated for 10 minutes in an argon
plasma (Harrick Plasma Cleaner, Harrick Scientific Corporation, Ossining, NY) before
coating. 1000 ppm solutions of PIB in chloroform and OV-275 in acetone were
airbrushed onto the surface of the crystal. PVA was dissolved in water to make a 10%
solution, and then a portion of that solution was diluted 50/50 with methanol and

airbrushed onto the surface of the SAW device. The frequency change for all devices

 

144
after coating was 50 i 5 kHz. It should be noted that equivalent frequency shifts
following coating does not necessarily imply that the surface coverage and/or morphology
of the films are identical. Several researchers have noted problems with the spreading or
wettability of films on SAW devices [27-30]. For example, poly(siloxanes) are reported
to have poor wetting properties on SAW device surfaces [29].

Data Collection and Analysis. For each analyte/solvent mixture, a run consisted
of a 1 [JL injection of the solvent followed by a 1 [IL injection of mixture. The solvent
and mixture injections were separated by approximately 1-2 minutes. At least four
injections of solvent and the mixture were made on two devices to ensure reproducibility.
Injection order was varied to minimize the effect of solvent order on sensor response.

The beat frequency of each resonator was collected using a digital counting board
(MetraByte, Taunton, MA) and stored in a Gateway 2000 386-25 personal computer. The
software used to collect the data was modified by the manufacturer to take advantage of
the capabilities of the autoinjector. All data were collected using an acquisition rate of 5
pt/sec (100 Hz resolution). Data analysis was performed using custom software written
in LabView for Windows (v2.52, National Instruments, Austin, TX). The software
separates the Femtometrics data file into individual channels, and then simultaneously
displays the response of all 6 channels. The software also determines peaks and peak
heights for each injection.

Sensor responses to pure solvents are reported as the maximum frequency shift in
hertz observed following sample injection. Evaluation of solvent and analyte responses

for mixtures is less straightforward. The frequency shift due to one component in a

145
mixture is often calculated from the difference in the frequency shift measured for the
mixture and the other pure component. However, this approach may not be valid for data
collected using a pulsed injection system since differences in solvent and analyte boiling
points can lead to significantly different response profiles. For data obtained with high
temperature cells (one peak following mixture injection), we have removed the
contribution of the solvent to the mixture response by subtracting the entire solvent peak
from the mixture peak profile. This was accomplished using the software package Origin
(MicroCal Software, Northhapton, MA) by aligning the rising edges of the solvent and
mixture peaks and taking the difference between the beat frequencies at the same instance
in time relative to the start of each peak. Thus, responses reported for biphenyl and PCB
(designated “biphenyl/mix” or “PCB/mix,” respectively) are the maximum frequency
shifts measured from response profiles obtained using this peak subtraction procedure. It
should be noted that this calculation also assumes additive SAW responses. As noted
above, two peaks are observed following injection of biphenyl/solvent mixtures into a
system that employs a 30°C cell temperature. The biphenyl response from the first peak
(designated as “biphenyl/mix) is determined using solvent subtraction. The response to

biphenyl measured from the second peak is designated “biphenyl/pure.”

 

146
6.4 Results

Figure 6.2 shows typical response profiles obtained after the injection of 1 [LL of
solvent followed by 1 pl. of 4% (w/v) biphenyl/solvent using the coated SAW devices
held at 30°C. The phenomena responsible for the response profiles have been described
above (see Experimental). The average responses to the solvents and biphenyl/solvent
mixtures measured using uncoated and coated devices are given in Table 6.1. For each
pure solvent, the uncoated device gives the lowest response. The PIB coated device
shows the largest response to hexane, while the PVA and OV-275 coated devices show
the largest response for methanol. The PIB coated device gives the largest mixture
response. The mixture responses measured with PVA and OV-275 coated devices are
similar. Table 6.1 also shows the biphenyl/mix responses calculated using solvent
subtraction (see Experimental). The results indicate that the biphenyl/mix responses
determined for a given film are essentially independent of solvent. The PIB coated device
shows the largest biphenyl/mix response. The biphenyl/mix responses determined for
PVA and OV-275 coated devices are similar. The biphenyl/pure peak magnitudes are
also largest on the PIB coated device and essentially independent of solvent for all
polymer coatings.

Figure 6.3 shows typical response profiles measured following the injection of 1
ul. of solvent and 1 [1L of 4% (w/v) biphenyl/solvent using polymer coated SAW devices
held at 60°C. The peaks obtained following mixture injection are broader than the pure

solvent peaks. The average response magnitudes are given in Table 6.1. The hexane,

147

.002” 00 0.0: 000300 00.000 n>N->O 0:0 50000.0
3:530: .A0:0_b:00wcb00 wEm: 00:03:. E020§>=0i£ 05 0:0 E0300 0.0 :0.“ 005030 005000 00:003.. 303:.

0:5.

 

 

0.7;

 

 

3:050:

a); .

J;

1410.}:

/\A

3:235

_ _
_ _

0. 8m
0:0. o.

 

 

 

0:0.00<

 

 

/ l/A
50200
/

50300:»:0005

 

0:008:

 

 

Aouanbard

.3 2:05

148

 

Table 6.1. The average measured and calculated responses (Hz) to the solvent,
biphenyl/solvent mixtures, and biphenyl measured using uncoated and
coated SAW devices held at 30°C and 60°C.a

30°C 60°C
Hexane Acetone Methanol Hexane Acetone Methanol
Uncoated
Solvent 900 (17) 1500 (18) 2300 (19) 500 (20) 600 (10) 700 (14)
Mixture 7200(5) 8000 (12) 8500 (3) 3100 (14) 3500 (5) 3700 (5)
Biphenyl/Mix" 6400 (14) 6200 (13) 6000 (8) 3100 (16) 3500 (7) 3700 (3)
Biphenyl/Purec 900 (14) 900 (10) 900 (12)
PIB Coated ’
Solvent 8300(9) 4400 (20) 5200 (14) 2600 (14) 1100 (3) 1100 (14)
Mixture 35400 (6) 31500 (7) 30900 (7) 18100 (15) 20300 (12) 23300 (9)
Biphenyl/Mix" 29700 (9) 28100 (8) 27200 (8) 18100 (18) 20300 (12) 23300 (10)
Biphenyl/Purec 440002) 4200 ( 16) 4200 (17)
0V-2 75 Coated
Solvent 1200 (16) 6400 (8) 9100 (13) 400 (21) 2100(9) 2400(4)
Mixture 16000 (12) 21700 (7) 21600 (18) 7900 (6) 8500 (12) 10500 (5)
Biphenyl/Mixb 15500 (12) 12300 (9) 12700 (24) 7500 (6) 7800 (15) 8900(7)
BiphenyVPure° 2900 (26) 2800 (33) 2700 (12)
PVA Coated
Solvent 1200(9) 6300 (17) 9600 (7) 300 (22) 2700 (16) 3100 (20)
Mixture 14800 (S) 15800 (3) 20700 (2) 7300 (6) 9400 (9) 9700 (6)
Biphenyl/Mixb 14200 (5) 11100 (8) 12000 (6) 7300 (6) 8500 (8) 8700 (17)
Biphenyl/Purec 2900 (11) 2800 (4) 2600 (11)

 

 

“ %RSDs are given in parenthesis.
b Frequency shift calculated for the biphenyl by peak subtraction (see Experimental).
° Frequency shift of the pure biphenyl peak following the biphenyl/mix spike at 30°C.

149

.050 .0 0.0: 000300 00.000 th->O 0:0 50:00.0
3:33.00 .A0:03.:000_3_00 33: 00:03.: 3028:3533 05 0:0 0:038 05 :0.“ 005050 005000 00:003.. .029;

as...

 

mE

<>m

m>~->0

 

 

 

 

 

_0:0_=02

3. J

4J

 

 

 

 

 

 

 

 

.14. a

 

_/

 

30300300005

OCO~OQ<

E038

000x02

 

 

Kouonbarg

.3 2.0m

150

each film are identical to those observed with the lower cell temperature. Similar to the
30°C results, the PIB coated SAW device shows the largest mixture and biphenyl/mix
responses for each solvent. In addition, variation in solvent has little effect on the
biphenyl/mix responses obtained using each polymer coating.

Figure 6.4 shows typical response profiles obtained with uncoated and coated
SAW devices held at 75°C following a 1 [IL injection of solvent and a 1 [LL injection of
500 ppm PCB/solvent mixture. In all cases, the PCB/hexane peak is broader than the
PCB/acetone or PCB/methanol peak and does not completely return to baseline after 700
seconds. Table 6.2 shows the average frequency shifts measured for the solvent and
PCB/solvent mixtures. The pure solvent responses measured with the uncoated and
coated SAW devices are lower than the values obtained with a 60°C cell; however, the
trends in solvent response are identical. Note that these response decreases can be
attributed to both the increased cell temperature and the increased flow rate (100 vs.
200mIJmin). The responses measured for each PCB/solvent mixture on the PIB coated
device are larger than the responses measured for either the uncoated, PVA or OV-275
coated devices. Examination of the calculated PCB response in Table 6.2 (see
Experimental) shows similar frequency shifts for the uncoated, PVA and OV-275 coated

devices.

 

151

 

.Uomb .0 0.0.. 000300 00.000 mh~->0 0:0 50:00.0

3:33.00 A0:0.b:00m.3.00 «Em: 00:03.: 8020059. 0... 0:0 E0300 0... 00.. 00:.050 00.00:: 00:09:00 .0033... .00 0:03".

2::

 

 

 

 

02000::

3:: 3: :
0 J7: {8)3:
J
33..

 

 

 

 

J3

Kouanbard

0.: 8mm

5020050; 4:11;“.130300 J .

3:05.02 o:0.oo< 2.0on

 

 

 

 

 

i

 

 

mg- >0

 

 

 

152

The average measured and calculated responses (Hz) to the solvent,
2,3,3’,4,4’-pentachlorobiphenyl(PCB)/solvent mixtures, and PCB
measured using uncoated and coated SAW devices held at 75°C.ll

Table 6.2.

 

Hexane Acetone Methanol
Uncoated
Solvent 400 (18) 500 (16) 600 (19)
Mixture 2300 (10) 1700 (15) 1300 (28)
PCB/Mix” 2300 (14) 1700 (24) 1100 (52)
PIB Coated
Solvent 2300 (21) 1000 (25) 800 (13)
Mixture 4800 (10) 3000(16) 2000 (29)
PCB/Mixb 4400 (23) 2700 (20) 2000 (11)
0V-2 75 Coated
Solvent 300 (21) 1700 (13) 1900 (12)
Mixture 2100 (9) 2300 (10) 2400 (7)
PCB/Mix” 2100 (11) 1400 (38) 1100 (34)
PVA Coated
Solvent 300 (30) 2000 (25) 2100 (18)
Mixture 2400 (10) 3000 (15) 2600 (9)
PCB/Mix" 2400 (13) 1500 (26) 1400 (27)

 

 

%RSDs are given in parenthesis.

Frequency shift calculated for the PCB by peak subtraction (see

Experimental).

15 3
6.5 Discussion

Influence of Salvation Phenomena on Sensor Response. The partitioning of
vapor into a thin film acoustic sensor coating is often discussed in terms of expressions
developed primarily to describe partitioning in gas chromatography [24,25,30-32]. Linear
Solvation Energy Relationships (LSER) have been particularly useful for characterizing
new materials [33] and GC stationary phases [34-36], understanding solvation processes
[37,38], and interpreting SAW responses [32-40]. Patrash and Zellers [41] have reported
that LSER coefficients can be used to describe the responses of SAW devices that have
equilibrated with analyte vapor. Recently, we have demonstrated the validity of LSER
expressions for the study of PIB coated devices using a pulsed injection system [42]. Our
results suggest that if the relative analyte diffusion rates in a film are comparable, valid
LSER coefficients can be determined for any coating using data collected with pulsed
sample introduction.

LSERs describe analyte/film interactions in terms of five analyte solvation

parameters and film coefficients:

log K = c + azafi’ +1226? +llogL16 +rR2 +3051 (6.1)

Ba"; is a measure of hydrogen-bond acid strength, 2B"; is a measure of hydrogen-bond
base strength, long describes dispersion interactions, R2 is a quantitative measure of the

ability of a solute to interact with a solvent through It and 1t electron pairs, and “H2

 

154

measures the ability of a compound to stabilize a neighboring charge or dipole. Solvation
parameters for the solvents, biphenyl, and selected chlorinated benzenes are given in
Table 6.3. Parameters for PCBs are not available. The relative magnitude of the film
coefficients (a, b, l, r, and s) determines the importance of specific vapor/film interaction
mechanisms for a film. The coefficients reported for PIB and OV~275 are given in Table
6.3. For PIB, the LSER coefficients indicate that the polymer should interact with vapors
primarily through dispersion interactions. The large a term is an artifact of the fit, and is
often reported for PIB [41,42]. The coefficients for OV-275 indicate that it is highly
polar (s coefficient) and interacts as a hydrogen bond base (a coefficient). The small I-
coefficient suggests that dispersion forces will be less important for OV-275. Film
coefficients for PVA are unavailable; however, we anticipate that this polymer should
interact with analytes predominantly through hydrogen bond acidity and, to a lesser
extent, dispersion forces.

Pure Solvent Responses. The solvent responses obtained with the uncoated and
polymer coated SAW devices can be readily explained in terms of solvation interactions.
The active area of the uncoated resonator surface consists of partially hydroxylated quartz
and thus has polar and hydrogen bonding (acid and base) sites that can interact with
adsorbed analytes. This suggests that the uncoated device should interact with hydrogen
bonding and polar analytes more strongly than with nonpolar species. The responses
measured for hexane, acetone, and methanol at 30, 60, and 75°C (Tables 6.1 and 6.2) are

consistent with these expectations. In addition, the decrease in solvent response with

 

155

 

 

Table 6.3. List of boiling points and solvation parameters [29] for compounds used
in this work and selected related species. The LSER coefficients available
for the polymer films examined are also included.

Boiling
Analyte Point (°C) 205’ 2135’ long6 R2 05’
acetone 56 0.04 0.49 1.696 O. 179 0.70
methanol 65 0.43 0.47 0.970 0.278 0.44
hexane 69 0.00 0.00 2.668 0.000 0.00
biphenyl 256 0.00 0.22 6.014 1.360 0.99
benzene 80 0.00 0.14 2.803 0.61 0.59
chlorobenzene 132 0.00 0.07 3.657 0.718 0.65
1,2-dichlorobenzene 181 0.00 0.04 4.518 0.872 0.78
Polymer a b l r s

poly(isobutylene)“ 0.875 0.275 0.869 — —
ov-275“1 3.03 0.616 0.615 — 1.88

 

 

 

156
increasing cell temperature is consistent with previous reports by a number of researchers
[9,32,39,43-45].

The addition of a polymer coating generally increases the response of the device
to the solvents. Notable exceptions to this statement are the hexane responses measured
using PVA and OV-275 coated devices. The solvation parameters for hexane (Table 6.3)
indicate that the only significant interaction mode for the molecule involves dispersion
forces. Thus, it is not surprising to find that the blank and OV-275 coated devices show
similar hexane responses since both surfaces are extremely polar and are not expected to
exhibit significant dispersion interactions. The similarity of the hexane responses
obtained with PVA and OV-275 coated devices suggests that the dispersion coefficient
for PVA is closer to the value reported for OV—275 than for PIB.

The large hexane response measured with PIB coated devices is expected based
on the importance of dispersion interactions for PIB (large I) and hexane (large long).
Similar results for alkanes on PIB coated devices have been reported by others [32,41].
For example, Patrash and Zellers found that alkanes exhibit larger partition coefficients
on PIB compared to similar boiling point compounds and smaller partition coefficients
compared to the same compounds on OV—275 [41]. The solvation parameters for

methanol show that the molecule has strong hydrogen bond accepting and donating

properties (2a? . 213$! ). and thus is able to interact through both donor and acceptor

hydrogen bonds. Acetone is also capable of strong hydrogen bonding interactions (large

Z [351 parameter). Given the film coefficients reported for OV-275 and anticipated for

 

157

PVA, resonators coated with these polymers should show large responses to methanol
and acetone. The small dispersion solvation parameters (logL'é) for methanol and
acetone suggest that PIB coated devices should be less sensitive to these solvents.
Patrash and Zellers [41] have also reported a larger partition coefficient for methanol in
OV-275 compared to PIB. In summary, the responses of PIB, PVA, and OV-275 coated
devices can be readily explained in terms of LSER principles. It should be emphasized
that the ability to use equilibrium solvation parameters to explain pulse response trends
suggests that any differences in analyte/film equilibration times are not significant for
these systems.

Biphenyl Mixture Responses. The biphenyl SAW response obtained from a
mixture can depend on factors such as biphenyl/solvent, solvent/film, and/or
biphenyl/film interactions. Obviously, sensor responses observed using uncoated devices
do not depend on any type of film interaction and thus most clearly define the effect of
analyte/solvent interactions on our measurements. The similarity of biphenyl responses
from different solvents measured using uncoated resonators (Table 6.1) suggests that the
nature of the solvent has little effect on the amount of biphenyl sorption that occurs
during an experiment. Further, these results indicate that the amount of biphenyl
available to a coated device is essentially independent of solvent. Thus we believe that
any differences in biphenyl response obtained with polymer coated SAW devices can be
attributed to analyte(or solvent)/film interactions.

In principle, the nature of a solvent can influence acoustic sensor responses to

mixtures in several ways. Solvent molecules could compete with trace analytes for

 

158

sorption sites in the coating. In addition, it is well known that vapor sorption expands
polymer volume [46] and dilutes the film. The interaction of the analyte with the film
may be affected by the presence of the solvent species or the change in polymer free
volume. In this study, we have chosen solvents that interact with the polymer coatings to
varying degrees. For example, hexane interacts more strongly with PIB than acetone or
methanol. However, our results indicate that, as noted for the uncoated device, the nature
of the solvent does not significantly affect the response of a given polymer to biphenyl
(Table 6.1).

The solvation parameters for biphenyl (Table 6.3) suggest that the molecule can
interact with solvents (and polymer coatings) using dispersion forces as well as polar and
polarizability interactions. The magnitude of the dispersion solvation parameter (longG)
indicates that dispersion interactions should dominate the solvation properties of the
molecule. Thus, it is not surprising to find that the PIB coated device gives the largest
biphenyl response. However, closer examination of the biphenyl solvation parameters
and polymer film coefficients suggests that the OV-275 coated device should also show a
large response to biphenyl.

Patrash and Zellers have successfully used LSER expressions to predict the
equilibrium response of polymer coated acoustic sensors to analytes [41]. Using the film
coefficients for PIB and OV-275 and the solvation parameters for biphenyl (T able 6.3),
we calculate that the logKSAw for biphenyl on PIB is 5.64 versus 5.56 for biphenyl on
OV-275. These results suggest that the biphenyl responses measured using PIB and OV-

275 coated devices should be similar. However, as shown in Table 6.1, the biphenyl

 

159

response measured with the PIB coated resonator is approximately two times the value
obtained using the OV-275 coated device. There are two possible explanations for this
observation. First, the diffusion rate of biphenyl in OV-275 could be significantly lower
than in PIB. This would lead to an anomalously low response to biphenyl using a pulsed
sample delivery system that employs an OV-275 coated device. In essence, the use of an
equilibrium model (LSER) to predict relative pulse responses is inappropriate if different
sensors do not equilibrate with the analyte at the same rate. Second, it is possible that
biphenyl sorption changes the mechanical properties of the PIB film [1,2,39] and,
consequently, produces larger frequency shifts than those expected from analyte
partitioning alone.

PCB Mixture Responses. In contrast to the results obtained for biphenyl, the
PCB responses measured with the uncoated device indicate that solvent does affect the
amount of PCB available for adsorption onto the SAW surface. Unfortunately, the exact
reason for this observation is unclear. In principle, differences in PCB solubilities could
lead to differences in the extent of PCB adsorption that occurs in our experiment. We
might expect PCB condensation to be more pronounced from poorer solvents. However,
the PCB concentration is very low (500 ppm) and hexane is a well known solvent for
PCB test mixtures.

Comparison of the PCB responses measured with the polymer coated devices
shows that solvent also affects the magnitude of the PCB respOnse obtained with these
sensors. However, solvent effects on the polymer coated device responses must be

considered with respect to the uncoated results. Indeed, it should be noted that the

160
solvent dependencies observed for the coated devices are similar in relative magnitude to
the trends observed for the uncoated device. Thus, we do not believe that solvent affects
the PCB/film interaction and the differences in PCB response reflect only variations in
the amount of analyte available to the device.

Unfortunately, solvation parameters for the specific PCB congener used in this
study are unavailable. However, Table 6.3 lists parameters for benzene, chlorobenzene
and 1,2-dichlorobenzene. The increase in dispersion solvation parameter (long) with
increasing chlorine content suggests that a pentachlorobiphenyl compound will have a

larger logLl6 value than biphenyl. We also anticipate that the PCB congener will have

significant R2 and 1t 5’ values. As in the case of biphenyl, the importance of dispersion

interactions should lead to a large PCB response using the PIB coated device. Based on
the film coefficients reported for OV-27S, one anticipates that a device coated with this
polymer should also show a large PCB response. However, the PCB responses measured
using the OV-275 coated device are not significantly greater than those obtained with the
uncoated resonator. We believe that the similarity of PCB responses obtained with the
uncoated and OV-275 (and PVA) coated devices indicates that the PCB does not absorb
into the bulk of the film during our experiment. Since the solvation interactions between
PCB and OV-275 (and presumably PVA) are favorable, we conclude that the diffusion
rate of PCB in OV—275 and PVA is so small that PCB desorbs from the surface of the
OV-275 and PVA coated devices before it has the opportunity to diffuse into the film.

Apparently, the PCB diffusion rate in PIB is high enough to allow significant absorption;

161
however, we cannot eliminate the possibility that PCB induces changes in the mechanical

properties of PIB that lead to an enhanced response.

6.6

PWSP‘V‘PP’N!‘

p—e
.9

p—e
p—e
O

12.
l3.

14.
15.

16.
17.

18.
19.
20.
21.

22.
23.

162

References

Grate, J. W.; Martin, S. J .; White, R. M. Anal. Chem. 1993, 65, 940A-948A.
Grate, J. W.; Martin, S. J.; White R. M. Anal. Chem. 1993, 65, 987A-996A.
Ballantine, D. S., Jr.; Wohltjen, H. Anal. Chem. 1989, 61, 704A-715A.

Fox, C. G.; Alder, J. F. Analyst 1989, 114, 997-1004.

Nieuwenhuizen, M. S.; Venema, A. Sensors and Materials 1989, 5, 261-299.
Frye, G. C.; Martin, S. J. J. Appl. Spec. Rev. 1991, 26, 73-149.

McCallum, J. J. Analyst 1989, 114, 1173-1189.

Katritzky, A. R.; Offerman, R. J. CRC Anal. Chem. 1989, 21, 83-113.

Grate, J. W.; Abraham, M. H. Sens. & Actuators B 1991, 3, 85-111.

Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. 8., Jr.; Jurs, P. C. Anal. Chem.
1988, 60, 2801-2811.

Amati, D.; Am, D.; Blom, N.; Ehrat, M.; Saunois, J .; Widmer, H. M. Sens.
Actuators B 1992, 7, 587-591.

Carey, W. P.; Beebe, K. R.; Kowalski, B. R. Anal. Chem. 1987, 59, 1529-1534.

Carey, W. P.; Beebe, K. R.; Kowalski, B. R.; Illman, D. L; Hirschfeld, T. Anal.
Chem. 1986, 58, 149-153.

Carey, W. P.; Kowalski, B. R. Anal. Chem 1986, 58, 3077-3084.

Zellers, E. T.; Pan, T-S.; Patrash, S. J .; Han, M.; Batterman, S. A. Sens. Actuators
B 1993, 12, 123-133.

Yao, S.; Chen, K.; Nie, L. Anal. Chim. Acta 1994, 289, 47-55.

Grate, J. W.; Rose-Pehrsson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H.
Anal. Chem. 1993, 65, 1868-1881.

Kindlund, A.; Sundgren, H.; Lundstrom, I. Sens. & Act. B. 1984, 5, 1-17.
Suleiman, A.; Guilbault, G. G. Anal. Chim. Acta 1984, 162, 97-102.
Thompson, M. ; Stone, D. C. Anal. Chem. 1990, 62, 1895-1899.

Heckl, W. M.; Marassi, F. M.; Kallury, K. M. R.; Stone, D. C.; Thompson, M.
Anal. Chem. 1990, 62, 32-37.

Thompson, M.; Stone, D. C.; Nisman, R. Anal. Chim. Acta 1991, 248, 143-153.
Yan, Y.; Bein, T. J. Phys. Chem. 1992, 96, 9387-9393.

24.
25.
26.
27.

28.
29.
30.

31.
32.

33.

34.

35.

36.

37 .
38 .
39.

40.

41.
42.

43.

45.

163

Janghorbani, M.; Freund, H. Anal. Chem. 1973, 45, 325-332.
Edmonds, T. 13.; West, T. S. Anal. Chim. Acta 1980, 117, 147-157.
Townsend IV, E. B.; Ledford, J. S.; Hoffmann, D. P., in press Rev. Sci. Instr.

McGill, R. A.; Grate, J. W.; Anderson, M. R. Interfacial Design and Chemical
Sensing, ACS Symposium Series 561, 1994, 280-294.

Ballantine, D. S. Jr. Anal. Chem. 1992, 4, 3069-3076.
McGill, R. A.; Abraham, M. H.; Grate, J. W. ChemTech 1994, 24, 27-37.

McCallum, J. J .; Fielden, P. R.; Volkan, M.; Alder, J. F. Anal. Chim. Acta 1984,
162, 75-83.

Snow, A.; Wohltjen, H. Anal. Chem. 1984, 56, 1411-1416.

Grate, J. W.; Snow; A.; Ballantine; D. 8., Jr.; Wohltjen, H.; Abraham, M. H.;
McGill, R. A.; Sasson, P. Anal. Chem. 1988, 60, 869-875.

Abraham, M. H.; Du, C. M.; Grate, J. W.; McGill, R. A.; Shuely, W. J. J. Chem.
Soc., Chem. Commun. 1993, 1863-1864.

Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1991, 587, 229-236.

Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chem. Soc.
Perkin. 2 1990, 1451-1460.

Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chromatogr.
1990, 518, 329-348.

Abraham, M. H. Pure Appl. Chem. 1993, 65, 2503-2512.
Abraham, M. H. Chem. Soc. Rev. 1993, 22, 73-83.

Grate, J. W.; Klusty, M.; McGill, R. A.; Abraham, M. H.; Whiting, G.; Andonian-
Haftvan, J. Anal. Chem. 1992, 64, 610-624.

Grate, J. W.; McGill, R. A.; Abraham, M. H. IEEE Ultrasonics Symp. 1992, 275-
279.

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

Townsend IV, E. B.; Ledford, J. S.; Hoffmann, D. P., submitted to Sens.
Actuators.

King, Jr., W. H. Anal. Chem. 1964, 36, 1735-1739.
Guilbault, G. G.; Jordan, J. M. CRC Crit. Rev. Anal. Chem. 1988, 19, 1-28.

Townsend IV, E. B.; Ledford, J. S.; Hoffmann, D. P. submitted to Anal. Chim.
Acta.

 

164

46. Rogers, C. E. in Polymer Permeability; Comyn, J., Ed.; Elsevier: New York,
1985; p 11-73.

 

"I1111111111!111111111111“