Citation
Studies in time-resolved optoacoustic spectroscopy

Material Information

Title:
Studies in time-resolved optoacoustic spectroscopy
Creator:
Voigtman, Edward George, 1949-
Copyright Date:
1979
Language:
English
Physical Description:
x, 156 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Capacitance ( jstor )
Dye lasers ( jstor )
Dyes ( jstor )
Electric potential ( jstor )
Electronics ( jstor )
Lasers ( jstor )
Microphones ( jstor )
Signals ( jstor )
Subroutines ( jstor )
Transducers ( jstor )
Acoustic emission ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Bibliography: leaves 150-155.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Edward G. Voigtman Jr.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
023005629 ( AlephBibNum )
05530168 ( OCLC )
AAK2891 ( NOTIS )

Downloads

This item has the following downloads:


Full Text











STUDIES IN TIME-RESOLVFD OPTOACOUSTIC SPECTROSCOPY


By

EDWARD G. VOIGTMAN JR.














A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE
UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY





UNIVERSITY OF FLORIDA















ACKNOWLEDGMENTS


I would like to acknowledge Dr. Martin Vala for his help and

encouragement throughout these exasperating experiments. His constant

optimism and enthusiasm were of incalculable value in seeing the

experiments through.

I would like to thank my wife Janiece Leach, whose love and

understanding were my only joy. Without her faith in me and her

personal sacrifices, nothing would have been accomplished.

I would also like to thank my fellow graduate students, Joe

Baiardo (who helped with the FFT subroutine translation), Bob Brittain,

Rodger Capps, and Dave Powell (who goaded me into lengthening the dye

laser cavity to effect tuning). Their fellowship was crucial to main-

taining my sanity in the face of unrelenting artifact production by

the TROAS apparatus. I would also like to thank Ed Whitehead for his

excellent machining and Rudy Strohschein for his expert glassblowing.

In addition, I would like to thank the staff of the departmental elec-

tronics shoo (Joe Miller, Russ Pierce, and Bill Wells) and Dr. J. D.

Winefordner, Jr.,for the loan of various necessary pieces of equipment

and laser dye. I also would like to thank Dr. J. Eyler for the loan of

Laser Dye 473.





















The words of Mercury are harsh after the songs of Apollo.

Love's Labor Lost















TABLE OF CONTENTS


CHAPTER PAGE

ACKNOWLEDGMENTS . . .... . . . ... .. ii

PREFACE . . . . . . . . . . . iii

LIST OF TABLES. . . . . . . . .... . vi

LIST OF FIGURES . . . ... . . . . vii

ABSTRACT. . . .... .. ......... . .. ix

ONE INTRODUCTION. ......... . ........ 1

TWO THEORETICAL CONSIDERATIONS. . . .... .. . . 19

Introduction and Direction . . . . . .. 19
The Extended Solution of the Heat-Flow Model. .. .. 20
Excitation Source Selection . . . . . . 26
Sample Cell Construction. . . . .... . .. 27
Sample Selection. . . . . . . . . 32
Pressure Transducer Selection . . . . . .. 33
Pressure Transducer Electronics . . . . . 38

THREE THE EXPERIMENTAL APPARATUS. . . . .... .... 41

The Dye Laser System . . . . 41
Sample Cells and the Vacuum System. . . . . 47
Pressure Transducers and their Associated
Electronics . .. .... ...... 56
Pressure Transducer Electronics . . ... ..... 61
Microcomputer Interface Electronics . . . .. 64
The Microcomputer System. . . . ... ...... 67

FOUR SAMPLE SELECTION. . .. . . . . . 69

FIVE EXPERIMENTAL ARRANGEMENTS . . . .... .. 77

SIX RESULTS . ..... . ... .. . . 108

SEVEN SINGLET TIME-RESOLVED OPTOACOUSTIC SPECTROSCOPY . . 122








APPENDICES

ONE RELEVANT OAS RESULTS . . . . . . . 126

TWO TROAS NOISE EQUIVALENT POWER. . . . . . ... 132

THREE CAPACITANCE MICROPHONE PREAMPLIFIER NOISE MODEL . 134

FOUR MICROCOMPUTER PROGRAM AND I/O LOCATIONS ...... 141

REFERENCES . . . . . .. . . . . 150

BIOGRAPHICAL SKETCH . . . . . . . 156















LIST OF TABLES


TABLE PAGE

1 A comparison of light sources considered for use
in TROAS experiments . . . . . . .... 28

2 Pressure transducers which may be used in TROAS
experiments. . . . . . . . .... 35

3 Capacitance microphones considered for use in
TROAS experiments. . . . . . .. .. . 36

4 Performance characteristics of the Candela dye
laser system . . . . . . . . . 46

5 Sample cells. . . . ... . . . . .. 48

6 Capacitance microphone comparison . . . .... 60

7 Relevant triplet properties of the compounds studied
by triplet TROAS . . . . . . . ... .72

8 A summary of the TROAS experiments performed. .... . 78

9 Peripheral systems memory locations . ... ..... .. 148















LIST OF FIGURES


FIGURE PAGE

1 Electronic states and rate constants typically of
importance in the photophysics of polyatomic
molecules . . . . . . . . ... . 7

2 The idealized TROAS spectrum resulting from direct
triplet excitation. . . . ... . . 12

3 Schematic representation of a typical TROAS
apparatus .... . . . . ... . 14

4 Schematic representation of a versatile OAS apparatus
with dye laser excitation . . . . .... .17

5 Optocoupled trigger circuit for the Candela dye
laser . . . . . . . . . . . 44

6 TROAS sample cell number 2 . . . ..... .. . 51

7 TROAS sample cell number 3 .. . . .. . . 53

8 TROAS sample cell number 4 . . . ... . 55

9 TROAS sample cell number 7 . . . . .58

10 Pressure transducer impedance conversion circuit . 53

11 Schematic representation of the microcomputer inter-
face system and electronic subsystems used in the
TROAS experiments . . . . . . ..... 66

12 Structures of the sample substances used in the
TROAS experiments . . . . . . ... .74

13 The TROAS spectrum of dithione obtained with a Phase-R
DL-1200 dye laser, Rhodamine 6G due lasing broad-
band, and a capacitance electret microphone . . 82

14 The TROAS spectrum of dithione obtained with a Xe
flashlamp (unfiltered) and a capacitance
electret microphone . ........ . . .. 85








15 The TROAS spectra of naphthaline in helium (upper
trace) and in air (lower trace) . ....... 90

16 The TROAS spectrum of iodine at 25. C in air at
atmospheric pressure. . . . . . . ... 94

17 The TROAS spectrum of 9-bromoanthracene in air at
atmospheric pressure with the piezoelectric
transducer used to detect the signal. . . .. 96

18 An artifact TROAS spectrum obtained with air at 25. C
and atmospheric pressure in the sample cell . . 99

19 The output of a piezoelectric transducer directly
exposed to a laser light pulse of high intensity. 102

20 The TROAS spectrum obtained with an empty (10-4 torr)
cell .. .... . . . . . . . 110

21 A voltage preamplifier and externally polarized
capacitance microphone circuit. . . . . .. 136

22 The voltage preamplifier and externally polarized
capacitance microphone circuit of Figure 21 with
appropriate noise sources added . . . ... 138


viii








Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


STUDIES IN TIME-RESOLVED OPTOACOUSTIC SPECTROSCOPY

By

Edward G. Voigtman Jr.

March 1979

Chairman: Martin T. Vala
Major Department: Chemistry

The approach of time-resolved optoacoustic spectroscopy is applied

to the study of the radiationless decay processes of electronically

excited polyatomic molecules in the gas phase. The loss-free theory

of time-resolved optoacoustic spectroscopy (TROAS) is extended by the

addition of the two major model-independent energy loss mechanisms

(viscosity and thermal conductivity) so that acoustic mode amplitudes,

mode amplitude ratios, and noise equivalent power may be calculated

for TROAS experiments.

Restrictions on design of experimental apparatus imposed by the

theory of TROAS are considered, and the necessary properties of

experimental apparatus sufficient to yield TROAS spectra are deduced.

Limitations on sample selection are also considered and lead to the

decision to study the following compounds: biacetyl, naphthalene,

2-chloronaphthalene, anthracene, 1-chloroanthracene, 9-bromoanthracene,

iodine, nitrogen dioxide, and 2,2,4,4-tetramethyl-1,3-cyclobutanedi-

thione (dithione).

A microcomputer-based apparatus was designed and constructed in

concordance with the theoretical constraints imposed by the theory of

TROAS and numerous experiments were performed. The predicted pressure








rise was partially observed as were the predicted radial and longi-

tudinal acoustic modes, but no triplet yield information could be

obtained by direct triplet excitation of the sample compounds due to

the existence of irremovable, mimicking artifacts produced by the

sample cell upon exposure to the highly intense (5 megawatts maximum)

dye laser excitation source pulses. Additional experiments deter-

mined the nature of the four types of artifacts concomitant with

direct triplet excitation TROAS. The possibility of obtaining triplet

yield information by singlet excitation TROAS is examined and found to

afford excellent prospects for polyatomic molecules in the gas phase,

without interference from the mimicking artifacts of triplet TROAS.














CHAPTER ONE
INTRODUCTION


The optoacoustic effect was discovered in 1880 by Bell [1] when

he observed that a great variety of substances emitted audible sounds

when exposed to periodically fluctuating illumination in their absorp-

tion regions. Bell's spectrophone was based on the observation that

the sound intensity was proportional to the sample's absorptivity.

He hoped it would find some use in absorption measurements, particularly

in the infrared region. Though a number of Bell's contemporaries

studied the effect [2-5], interest waned until 1938 when Veingerov [6]

rediscovered the effect and employed it in the study of IR absorption

by gases. Despite a few scattered uses [7-10] for such applications

as IR gas analysis, IR detection instrumentation, and microwave

detection, it was largely neglected until 1946 when Gorelik [11] noted

that the spectrophone could be used to measure the vibrational re-

laxation times of gases. Again the technique lay dormant until

Kaiser [12] and Delaney [13] in 1959 proposed a reaction kinetics-

based model for the optoacoustic effect. Also in 1959, Gerlovin [14]

applied the spectrophone technique in the visible and UV.

It will be useful for our later discussion to note the more im-

portant contributions to the development of the optoacoustic technique

over the last two decades. In 1963, White [15] observed that elastic

waves are produced in solids following absorption of pulsed light.

Several other investigators have made similar observations [16-17].





2

In 1967, Hey [18], using broadband illumination and the spectrophone

technique, measured the relaxation rates of dyes in solution. In

1969, Seybold et al. [19], using steady state illumination and a

capillary rise technique, made similar studies. Also in 1969, Callis

et al. [20], using broadband pulsed illumination and capacitance

microphone volume change detection, were able to determine triplet

yields of anthracene in ethanol. In 1971, Kreuzer [21] greatly im-

proved the spectrophone technique, as it applies to gas phase absorbance

measurements, by employing a laser as the illumination source. Since

then, the technique has found wide use in such applications as air

pollutant detection [22-24], in situ aerosol absorbance measurements

[25], and the study of photochemical reaction mechanisms in the gas

phase [26-27]. In 1977, Patel and Kerl [28] achieved absorptivities of

109 to 10 cm for the gas NO. Significantly increased sensitivi-

ties may also be achieved by placing the spectrophone inside a laser

cavity [29].

Since 1973, the gas phase optoacoustic effect has also been used

in the study of NO2 and SO2 [30], the quenching of iodine atoms by

oxygen [31], and in investigations of radiationless transitions in

biacetyl [32] and the azabenzenes [33]. The quenching of the first vi-

brational level of the lowest excited oxygen singlet has been studied

by Parker and Ritke [34-35]. Hunter and coworkers [36-37] have also

used the spectrophone technique in the study of electronically excited

benzene and biacetyl. These studies were significant for several

reasons. First, important information was obtained which would have

been difficult, if not impossible, to obtain by other techniques. Sec-

ondly, a variety of auxiliary techniques were developed and led to an








increase in the usefulness of the basic spectrophone. Among these were

lock-in amplifier signal detection, carrier-modulated pressure detectors,

and dual excitation source techniques [38].

The source of the heat in a simple (IR source) spectrophone experi-

ment is collisionally-induced vibrational to translational relaxation

[39]. For condensed phases the mechanism is substantially more complex.

In 1973, Parker [40] observed that the glass windows on his optoacoustic

sample cell generated an in-phase signal when illuminated even though

the gas in the cell (Ne, N2, 02) was transparent to the excitation

wavelengths used. With the aid of a thermal diffusion model of the

window and adjacent gas, he established that the signal he had observed

was produced by thermal diffusion from the window into an adjacent

boundary gas layer. At about this time, Rosencwaig and others, using

capacitance microphone detection, extended the spectrophone technique

to include solids, liquids, smears, gels, and biological materials

[41-43]. In 1976, Rosencwaig and Gersho [44] developed the "thermal

piston" model to explain the optoacoustic effect exhibited by condensed

phase materials. These authors suggested that diffusion of heat into

the surrounding gas (to a depth of the thermal diffusion length) causes

the boundary layer to fluctuate and thereby drive the remainder of the

gas. Hence,chopped source illumination could be expected to produce

fluctuations at the same frequency in the boundary layer. Comparison

of their theory with experimental results suggested that this was the

primary source of sounds produced by condensed phases. In 1978,

McDonald and Wetsel [45] extended the theory (the composite piston

model) by including the effect of the expansion and contraction of the

condensed phase. They proposed that the boundary layer was mechanically








driven, thereby generating sound. As noted previously, however,

elastic waves may also be produced in condensed phases by the absorption

of intensity-modulated light. These elastic waves may be detected

directly [46] using piezoelectric transducers. Thus,three mechanisms

have been proposed to account for the production of sounds by condensed

phases in "the" optoacoustic effect: the thermal piston model, the

mechanical piston model, and light-generated elastic waves.

With a few notable exceptions videe infra), recent work on the opto-

acoustic effect applied to chemical systems has dealt with the spectral

distribution of the released heat and not with the kinetics of the

release process. Work in this laboratory was begun a number of years

ago to develop a theory of time-resolved optoacoustic spectroscopy which

was capable of describing the kinetics of radiationless decay processes

(heat release) after electronic excitation in polyatomic molecules. The

theory was developed by Wrobel [47]. It was motivated by the need for

detailed information concerning nonradiative deactivation pathways in

electronically excited polyatomic molecules under approximately

collision-free conditions. In general, phosphorimetry and fluorimetry

are incapable of providing all the data necessary to obtain accurate

triplet yields, without which it is impossible to test the various

theories of radiationless processes in polyatomic molecules. Time-

resolved optoacoustic spectroscopy (TROAS) is intended to complement

these other techniques, particularly for species with low radiative

yields, and thereby lessen the need for the motley assortment of

ingenious methods which have been devised to furnish triplet yields

[48-56].





5

To understand the need for a complementary technique, it is helpful

to consider the various pathways of deactivation available to an elec-

tronically excited molecule. Irradiation of a molecule results in its

excitation from its ground state (So) to a vibrational level of some

excited singlet state, Sx. With few exceptions, rapid radiationless

relaxation ensues populating the lowest vibrational level (v = 0) of the

lowest excited singlet, S( (the superscript denotes vibrational level

and the subscript electronic state). Three deactivation processes may

now occur which depopulate SO (see Figure 1): fluorescence with a rate

kf, radiationless decay with a rate kS, and intersystem crossing to the

triplet with a rate klSC. The triplet vibrational level so populated

may relax directly to SO via radiationless decay with a rate kTv, or it
0 T1
may relax via rapid radiationless decay to T1 with a rate kTv,TO. Inter-
0 1
system crossing back to SO is negligible. Finally, TO may relax by

phosphorescence with a rate k or by radiationless decay with a rate kT.

Since decay directly from the vibrational manifold of T1 to SO is not

usually possible (Kasha's rule), kTv may be neglected. In addition,

vibrational relaxation in T1 is usually fast relative to kT and kp.

The five remaining rate constants (kf, ks, klSC, k kT) may be

used to define the fluorescence quantum yield f, the phosphorescence

quantum yield Dp, and the triplet yield t as follows:


4f = kf/k* where k = k kf + kISC is the observed

fluorescence rate,


*p = kp t/k* where k* = k + kT is the observed phosphorescence
p ~ ~ tpP T


rate, and





























Figure 1. Electronic states and rate constants typically of im-
portance in the photophysics of polyatomic molecules.





7














x -







Isc
"1 k

/ k v 0
/ 1 1

kf I
I 'T
/p
Ik kTv


kT
Sh / /, / ,


h/
S/ /


/ /




So








t = klSC/k*;


Thus,fluorimetry and phosphorimetry experiments yield the four quantities

D, f (kp ) the observed fluorescence lifetime, and (k*)- the ob-
Spp
served phosphorescence lifetime, but these are obviously insufficient

to determine the five rate constants noted above.

Clearly, then, another independent experimental quantity is nec-

essary. Although a number of techniques have been developed to meet

this need [48-56], each such technique possesses certain disadvantages

which prevent it from being a practical complementary technique to

fluorimetry and phosphorimetry. In fact, it is common to simply assume

kS is negligible (Kasha's intersystem crossing rule [57]), although there

is no unassailable basis for such an assumption. Consequently, it is

desirable to find a general technique which is easily capable of pro-

viding the requisite independent quantity and which preferably does so

by measurement of the radiationless rate constants. A clue as to the

nature of this desired technique is provided by noting that radiation-

less processes are accompanied by the release of heat. Therefore, the

general technique should ideally involve monitoring the heat released

during radiationless decay processes subsequent to excitation of the

sample. This is the basis of optoacoustic spectroscopy.

In 1974, Wrobel [47] developed the theory of gas phase optoacoustic

spectroscopy which was specifically designed to accommodate tunable

pulsed dye laser excitation of the electronic states of polyatomic

molecules. In this theory, the pressure fluctuations produced in the

sample vapor were determined from the conservation equations (mass,

momentum, and energy), the equation of state, and the rate equations








associated with a selected multielectronic state model (SO,T1 or

SOT S ). The fundamental result of TROAS (valid for the simple

So,T' model) is the following relation between the pressure fluctua-
tions expected to occur in the gas phase and the nonradiative rate

constants:

-( k)t
P1 )qOOn v [k Ok v TO h VTO (1 e ( )
u0,0 1 1'T T 1 T 1


(kTv+ kv 0 k*)(k*)
1 TTi p p


+ (kTv ,O h v ,0 / (k v 0+ kTv)
,'T1 T'Il t'T1 '

-(kTV TO+ kTv)t
kTO h O/ (kTv+kT k*))(1 -e 1' 1 1 )]
T1 1 T1 IT 'T1 p


where y is the heat capacity ratio, n v is the cell-averaged initial
number density of molecules excited into T, q0,0 is an expansion co-

efficient dependent on the model-specific heat source term, and POO
is the amplitude of the pressure rise expected to occur.

Normally, the vibrational relaxation rate kTv ,T is much larger
than the triplet deactivation rate k*. The above equation then

simplifies to

__O IC -(kv ,T0+k v)t
P',0 = (y-l)qon [ h TV .Tv (1 -e 1 1 )


IC ISC -k*t
+ DT Tv hv 0 (1 -e )]
I 1 1







ISC IC
where T v is the intersystem crossing yield kT/k*, and i v is the

internal conversion yield kTv ,O/(kTv TO +kTv). Thus, the expected
S1 1' '1 I
pressure rise is the sum of two exponentially rising components with

different amplitudes. Note, however, that kTv TO is much larger than
I' 1
k*. Hence, the pressure rise will appear to be a step rise with an
p
exponential rise building upon it. This situation is depicted in

Figure 2 [47]. If the amplitude of the slow rise is denoted p' and

that of the fast rise is denoted p%, then from Figure 2 the following

equation is obtained:


ISC
+TO = (hV v, TO/hvTO) ps/p
1l1 1 1


Since k* is obtained experimentally from the time constant of the slow
p
rise, the radiationless rate constant k- may be obtained directly and

independently of phosphorimetry and fluorimetry experiments. Unfortu-

nately, Wrobel [47] was not able to experimentally determine the

validity of this result,so additional experiments have proven necessary.

Figure 3 illustrates schematically the experimental setup of a

TROAS apparatus. A light pulse from a flashlamp-pumped tunable dye laser

passes through the cylindrical sample cell and strikes the laser energy

detector. Absorbed energy converted to heat via radiationless decay

processes heats the sample gas mixture (molecular vapor plus carrier

gas) producing a pressure jump and acoustic oscillations. These pres-

sure variations are converted to electrical signals by a sensitive

broadband pressure transducer. The electrical signals are amplified and

acquired by a microcomputer-controlled transient waveform recorder.

The microcomputer also synchronizes data acquisition by controlling



























I-


Lb Od
O'b A u i(-,) / ,d









14
















Ti




sC
0.






IC a >
uu
cIn
a)4
o /c














CIRD c"I 0v


Ou







In S. .
10. M
UU 1-
T- _r -i oai,
or i --- --- C S' c ---




















ia,
f 0 ;ur ()i
fi ui 3a



a. S



5i s


i~ c3 I __I

iyi u

U3 u a i -
'/'~~ ~ ^__ E t- ----- 0
'7 i ~3 ~^ -- o 0 3 --- -s ^i
i-* _ U *" C
^~ o -i- u
i- c i-













i oa r -- i -


I l "0 (Ur CU\
I 01 I) r- !
a u/ r Q. C S
-1C = CT \ <
-3~ ~ 0 CT \ <1 -








laser firing, light pulse peak-detector reset, etc. It also performs

data processing tasks (such as Fast Fourier Transforms) and outputs

data to a chart recorder as necessary. The oscilloscope displays

acquired data prior to processing.

This experimental arrangement differs markedly from that of con-

ventional optoacoustic spectroscopy (OAS) as shown in Figure 4. The

excitation light source is usually a monochromator and Xe lamp combina-

tion with light modulation provided by a rotating slotted disk. Alter-

natively, a tunable continuous wave (CW) dye laser and either an

acousto-optic modulator or chopping wheel provides the requisite

modulated source intensity. In any event, the spectrum obtained must

be corrected for variations of source intensity as a function of source

wavelength. The second major difference between OAS and TROAS is the

use of a lock-in amplifier to amplify and detect the pressure trans-

ducer signal (after the preamplifier stage, of course). The resulting

extremely narrow equivalent noise bandwidth allows the use of OAS in the

detection of signals in extraordinarily dilute samples [28]. An

especially good collection of research papers concerning all relevant

aspects of OAS has recently been published [58].

In OAS, accurate waveform information is not necessary; in fact,

a lock-in amplifier provides only amplitude and phase shift information

in the usual case. This is adequate for sample detection purposes but

not well suited for the study of such processes as triplet deactivation

pathways. Further details concerning this use of OAS may be found in

[32].

The theory of TROAS was developed by Wrobel [47] and, therefore,

will not be repeated here, although Chapter Two concerns certain




















































SI

L

o o-
C-c


-J


CC-


i
CO 1.
I a-
1-
0~ C.
U
I- 0
u~I~
CJ U
z(
r





18

extensions and corrections to the loss-free theory. Chapter Three

describes the experimental apparatus used in this research. In

Chapter Four, the sample selection criteria developed in Chapter Two

are used to select promising polyatomic molecules for study. Chapter

Five details the numerous experimental arrangements and modifications

used. Chapter Six discusses the obtained results, and Chapter Seven

contains a discussion of singlet TROAS and its advantages.














CHAPTER TWO
THEORETICAL CONSIDERATIONS


Introduction and Direction


Several papers have appeared in which pulsed light sources were

combined with the optoacoustic effect in order to study various vibra-

tional and electronic state properties of polyatomic molecules. Al-

though the pulsed optoacoustic effect is certainly shown in the works

of Parker [34], Aoki and Katayama [59], and Grabiner et al. [60], the

interpretations provided are either of limited scope or unnecessarily

complex. The theory of TROAS is intended to bring together these

diverse observations in a relatively simple and coherent manner.

Unfortunately, the particular solutions given in Wrobel [47] for

radiationless decay from the lowest singlet and triplet states of poly-

atomic molecules are incomplete due to neglect of all energy loss

mechanisms, exclusive of those incorporated in the model-specific source

term, in the sample gas mixture and sample cell. The most important of

these dissipative processes are viscosity and thermal conductivity

losses [61]. By inclusion of these two loss mechanisms, it is possible

to calculate realistic acoustic mode amplitudes, mode amplitude ratios,

and an important figure of merit, the noise equivalent power (NEP).

Restrictions imposed by the extended theory are then applied to sample

cell design, selection of appropriate pressure transducers, and excita-

tion light source requirements.








The Extended Solution of the Heat-Flow Model


Consider a molecular vapor at equilibrium in a cylindrical sample

cell of length L and radius A. The equilibrium density, pressure, and

temperature are PO, pO, and TO. Radiationless relaxation processes

following excitation of the vapor in an absorption region cause fluctua-

tions in p, p, T and cause the vapor to assume an acoustic velocity

7(F,t). The conservation equations and equation of state constrain the

behavior of the vapor

R
p = p T equation of state,


d- = -p v- conservation of mass,
dt

dv -
S= -7 p conservation of momentum,
du i

Sdu -P.. 7-(I + TR) conservation of energy,
dt ij ax R


where R is the ideal gas constant, M is the molecular weight of the

sample, u is the internal energy per gram of sample, p is the pressure

tensor, T is the nonradiative energy flux vector, TR is the radiative

energy flux vector, the convective derivative is given by


d + (( 7)
dt at


and the Einstein summation convention is used.

This solvable system of equations may be simplified by noting that

the fluctuations induced by released heat are small relative to the








equilibrium values. This acoustic approximation linearizess" the

system of equations and hence avoids the possibility of shock waves [62].

Thus, p = p' + PO p = p' + pO, T = T' + TO, and v is small. Products

of primed quantities are neglected. If viscosity and thermal con-

ductivity are also neglected, the system of equations simplifies to

yield a single separable equation [47,63]



2-,- c2 p' = (-l1) ^ (2.1)
at2

where the heat source term H is given by

due
H -v R PO (2.2)


and ue is the energy per gram stored in electronically excited

molecules.

The heat source term is given explicitly only when the physical

model (e.g.,a simple two-state model of triplet decay) is selected.

With appropriate heat source terms, equation (2.1) underlies both OAS

and TROAS. As stated previously, the loss-free solutions of (2.1)

are inadequate for the calculation of mode amplitudes, etc. Appendix One

contains a brief summary of Kreuzer's approach [63] which includes the

two major energy loss mechanisms--viscosity and thermal conductivity.

These results will be used as needed to complete the treatment of TROAS.

The fundamental effect of including viscosity and thermal con-

ductivity is the introduction of a finite quality factor Q. for each

acoustic resonance wj. The use of a Fourier Transform technique to effect

a solution of (2.1) is natural for OAS since the frequency parameter a

may represent the frequency at which the illumination source is








modulated. It may be, for instance, the number of pulses per second

(times two pi) delivered by a simple chopping wheel interposed between

illumination source and sample cell. If the analysis of Kreuzer [63]

is to be carried over to TROAS, it must be possible to identify a

suitable w.

It is assumed in the theory of TROAS that the illumination source

is an energy pulse of negligible duration. This situation is approxi-

mately met when the light pulse duration is much less than the smallest

relevant time constant in the physical model. As a practical matter,

it is satisfactory to have the light pulse duration, tp, substantially

less than the nonradiative lifetime of the spin-forbidden processes

being examined. Such spin-forbidden processes will almost always be

involved due to intersystem crossing even if only an excited singlet

state were initially excited. In the event no "slow" (e.g., forbidden)

states lie below the desired "fast" (e.g.,allowed) state, the system is

probably better treated by conventional spectroscopic methods. The

time constant, tT, of the thermal damping subsequent to pulse excita-

tion and relaxation will, for reasonable physical models, always be

much greater than t .

The high modulation frequency case for OAS is given by the con-

dition [63]


S tT >> 1 (2.3)

For the simple chopping wheel type of intensity modulation, the light

pulse duration is


-1/
/2







assuming a 50% duty cycle for the chopping wheel. Hence,

-1 1
W tt >> 1 implies tT >> > a/2

so the light pulse duration is short relative to the thermal damping

time. Precisely this situation occurs in TROAS. A reasonable estimate

for w in TROAS is, therefore, given by t' More accurately, w will
p
be assumed to be the spectral bandwidth required to encompass most of

the spectral energy in the Fourier Transform of the light pulse [38].

For example, assume the light pulse to be a rectangular pulse of dura-

tion t centered at t = 0. Then the Fourier Transform is proportional

to [64]


tp since ( t p/27) .


The significant portion of the spectrum is in the range
-1 -1
jI/2 I p p
it is possible to calculate acoustic mode quality factors Qj, mode

amplitudes Aj(w), and NEP for TROAS. Expressions for Q. and Aj(i)

are given in Appendix One and an expression for NEP is derived in

Appendix Two.

One possible complication arises in the quantity for the absorbance,

a. The relation between absorbed light pulse intensity I and resultant

heat produced H is


H(r,t) = a I(F,t) (2.4)


if the intensity is not high enough to saturate the absorbance and if

the time variation of I is slow relative to the rate of relaxation.

The second condition is usually violated in the study of triplet








relaxation via TROAS. Hence, in a more realistic fashion, the relation

between H and I should be [63]


H(r,w) = c(w) I(T,w) (2.5)

The only immediate effect of this for TROAS is that the heat produced

lags the light intensity variation. This will not be considered

further.

Combining the above relation with the expression for corrected

acoustic mode amplitudes (A1.20) from Appendix One yields


pA.(F,-) a(w) I(Fr,) d V
iw(-Y ) (2.6)
SV 2 [1- (a/.) -i( /mj Q )]


The integral determines the coupling between the normal acoustic modes

and the beam intensity since I(F,m) may be expanded in the same ortho-

normal set of functions used to express the pressure p(F,i). Hence,

the orthogonality equation (A1.7) from Appendix One implies that

a normal mode can be excited if and only if the intensity has a cor-

responding nonvanishing component. Intuitively, one expects longi-

tudinal modes to be excited if significant attenuation of the beam

occurs in passage through the cell. In a uniformly illuminated cell

(e.g.,intracavity in a dye laser) longitudinal modes should not be

excited. It is also reasonable to suppose that radial modes will be

least strongly pumped in the uniform illumination case. The former

expectation is verified by calculator simulations by Wrobel [47] while

Kamm [61] has explicitly given the relationship between excitation

beam width (a) and acoustic mode excitation:








p(r,t) = (-i ) 2 -2 P e-
J (1 /m iw/ Qj) c J (aj)


Jg(l,j r/A) e (2.7)

where

a 2 -2
(j ( (c (2,j)2 (2.8)

For the lowest frequency radial mode (j 1) excited at l, the above

equation simplifies to

(Y-1)Q1 p el-
Al(W ) Vc 1,1 r/) (2.9)
S c JO (al,l)

Note that the radial dependence of the amplitude of the lowest radial
mode is relatively weak, residing solely in u,.

In TROAS, the pressure rise mode (j =0) contains the significant
information. The other acoustic modes obscure the pressure rise and

waste energy that would preferably be pumped solely into the j=0 mode.

Examination of (2.9) above indicates that the radial modes will be

minimally pumped when e is minimized. This occurs at a=A, i.e.,
when the excitation beam width equals the sample cell radius.

Although other dissipative mechanisms external to the source term,

such as wave reflection, wave scattering, microphone conversion, and
volumetric losses, are present [61], they are of minor importance in

comparison with viscosity and thermal conductivity effects. Conse-

quently, we are now in position to consider the characteristics required

of each component of a TROAS experiment. The modified TROAS theory
restricts the selection of various components of the apparatus and

also determines to a large extent just what substances may be studied.








Excitation Source Selection


The "ideal" illumination source for TROAS should have the follow-

ing properties:

(1) high constant energy per output pulse (preferably greater

than 100 mJ),

(2) short, stable pulse duration tp (such that tp is much

less than other significant system time constants),

(3) high pulse repetition rate (for possible signal averaging

purposes),

(4) low beam divergence (for efficient energy coupling to

the cylindrical sample cell),

(5) moderately large beam width (0.5-3.0 cm typically),

(6) nearly monochromatic light output (no more than several

Angstroms spectral bandwidth to enable excitation of a

single state),

(7) tunable operation in the visible and near ultraviolet

(to excite electronic states),

(8) ease of alignment,

(9) reliable operation over more than 104 high energy pulses

(higher if signal averaging is used).

These conditions are only met by tunable pulsed dye lasers. A compromise

between conditions 1 and 3 must be made in the selection of a suitable

dye laser system since the optical pumping sources have limited duty

cycles, pumping capacities, or heat dissipation capabilities. Since the

desired pressure rise mode is proportional to pulse energy, high pulse

energy is especially desirable; a coaxial flashlamp-pumped dye laser will








provide energies of several Joules per pulse. Unfortunately, the

repetition rate for this type of laser is approximately 0.1 Hz so

signal averaging is laborious. In fact, the rather limited tube life

(something above 104 shots), tube expense (approximately $500.), tube

failure mode (rapid disintegration), and photo-decomposition of the

dye solutions after a few score shots effectively rule out signal aver-

aging. Linear flashlamp-pumped dye lasers have repetition rates up to

30 Hz and pulse energies of 100 mJ maximum. Nitrogen laser-pumped dye

lasers have similar repetition rates but significantly less energetic

output pulses. They are not recommended for TROAS in the general case.

See Table 1 for a comparison of the above three dye laser types and the

monochromator and Xe lamp combination.


Sample Cell Construction


The sample cell should be an inert, high heat capacity, isothermal

cylinder having transparent end windows and several inlets for pressure

transducer(s), pump-out ports, sample admission, etc. A cylindrical

geometry is chosen to provide maximum coupling between excitation beam

and sample vapor. As previously shown, the desired pressure rise mode

(j =0) is maximally pumped when the excitation beam width equals the

cell radius. The inlets should be as small as possible to avoid ex-

cessive wave reflection losses. If one pressure transducer is used, it

is probably best positioned midway between the end windows since the

solution of the equations of TROAS is greatly facilitated for such a

choice [47] and beats in longitudinal modes are eliminated. The end

windows present a problem in that they should have high thermal conduc-

tivity and yet be at the same temperature as the body of the cell. The















-0 C. E


25-.


=a) u- E
C_0
a)



a/) U) -J
a) 23 Ci ; r
*- i- = c= 2: s - 5. S.-
C n3 r- 0 C r
a) C*J 0 0 0 i- 2 Ln ~ a3 a)
E i- Cej" V 0l 0 N 4- 4-.
5-
.-c
CL L



xu










a) I
m vi 4





S.-



C -a
5- --
ina E 11*






< m3 v7Sn
'41

C a
CLr



0; aJ )= 2. N7 .
1- 5)0 0 ~
:I Ec
0) C V 2- ) aE
CO E O Oir O -
S- *i- 3 0 S T



0 .-I. Qm C S V 0 o N. '4-. '-


*0
a):


11-

25-
a) C

*0 0-5)


V7 a) LO-
o a- -
0/ <0. CJ .- E .
a) 2 '0 S) 2 5- -r C




,a O f aS C





.0 o-) LO C C O -






4-1
0 -C -0Q










C)
C S- A .
V1 Ir-Q (U Q;I













ar; = cuE



U 0 2 M -
3 CDula 3 N






0 -+i- CjE ^ ^ "- U
"I CI C










C n A 0
0. 0.5- A1 = )-
oC (-o 0n o0 C0 i O .-I 4-I
C, I C cI S C










C C C1 S- C C
0 0 Sa >, .


















CL CU34- 41^ =
< -l A ) -
xa n3 == ^ f /







.0 0 = A i> 5












a aa)
S- S.- A E )
= S C 0S ;_ CS a)- E >
0 c- m aD







- 4-- C i- 5-C -7 a) 4 4-I







C. C 1 C3 0 -I 0) c Q)
S = C= = =
S.3 05- (U)0 4- C a
I-S. S0 5) 2_ 2 ^ C
C i 0. c H- 1
Q. o tU +- E r3 4- -
Q. vl 3 0 S- != ^
tU~5 >i -1 -0 S -= 0 E

*- *- /i J E *' ^n *r



^J QJ i- Q- (3 (3 C: C --
aj C u a; a 3 i- v )

a L j Q. as aa m -







high thermal conductivity is necessary so that the windows can dissipate

any heat produced by adsorbed substances. Otherwise, this heat will

pump spurious longitudinal modes [65] despite the windows being at nodes

of these modes [66]. If the windows cool significantly, condensation

may occur and the previous problem is greatly exacerbated. Sapphire is

an excellent choice of window material though expensive. Double

windows of Suprasil or Pyrex are satisfactory if light intensity can be

compromised approximately 8%. The double windows must have an adequate

dead air space between them, and they must be parallel to avoid scatter-

ing light onto the cell walls and pressure transducer.

In the absence of thermal conduction, the expected pressure rise

term is given by Wrobel [47]:


O,o t- = ( -1) kT(k )-1 (1 -e L)E/Vc (2.10)

where kT(k*)1 is the nonradiative yield, a is the sample absorbance,

L is the cell length, E is the excitation source energy per pulse, and

Vc is the cell volume. As a practical matter, a will be small since all

samples should be run at as low a pressure as possible (consistent with

the continuum model of sound conduction) to minimize the quenching ef-

fect of collisions. For a small, 1 -e-'L is approximately equal to aL.

Thus, equation (2.10) simplifies to


P',0 It_ L/' A2L = a/7rA2 (2.11)

where A is the cell radius. Thus,the pressure rise is independent of

cell length but inversely proportional to A2. Nevertheless, L should

be as short as possible to lessen scattered light problems and to allow








placement of the cell in the sample compartment of a UV-visible spectro-

photometer for in situ gas phase absorbance measurements.

Cell radius is much more important for a variety of reasons. Pres-

sure rise amplitude is proportional to A-2 so the cell radius should be

minimized. The lower limit on radius is the excitation beam width.

The NEP is also proportional to A so again the radius should be mini-

mized. Three time constants also depend on cell radius; these are

designated tT, td, and tw and are described below.

The time constant for the thermal reequilibration following release

of heat in the cell is approximately given by


tT = A2/c (2.12)


where K is the thermal diffusivity. This upper time limit is not

critical and usually is of the order of 1 to 10 seconds.

The acoustic delay time td is the time required for sound waves

to reach the pressure transducer from the nearest illuminated sample

region. It is a lower time limit given by


td = (A-a)/c (2.13)

Note that td = 0 for a = A.

The "mean free path time" tw is the time required for collisional

deactivation at the cell wall (ignoring deactivation at the windows).

This is given approximately by


tw. A2/ic (2.14)

where E is the mean free path in the sample gas mixture. This is an

important upper time limit.








The time limits defined by these time constants may be modified in

several ways. First, since both td and tw are inversely proportional

to the speed of sound in the sample gas, it is possible to add a trans-

parent carrier gas such as He, Ar, SF6 or a freon to the sample to

change the speed of sound in the mixture. The speed of sound in an

ideal gas mixture is


c = (yRT/M)1/2 (2.15)

where M is the average molecular weight of the sample mixture. Thus,

by employing a heavy carrier gas (e.g. SF6) the speed of sound can be

decreased and the upper limit due to the "mean free path time," t,,

increased. Unfortunately, at the same time the lower limit due to the

acoustic time delay, td, is also increased. Thus, it is necessary to

choose the carrier gas with care to select the upper and lower time

limits which are most convenient for study of the molecule under

consideration.

The second possibility is to use the fact that both < and t are

inversely proportional to pressure [47]. Note, however, that the

pressure rise amplitude is proportional to the partial pressure of the

absorbing species and that collisional deactivation processes such as

triplet-triplet annihilation become more important at higher pressures.

The use of an inert monatomic carrier gas also increases the signal

amplitude through the heat capacity ratio term. A sample calculation

is given in Chapter Four.








Sample Selection

The pressure rise which occurs after radiationless decay is given

approximately by [47]


p' z (Y-1) kT(k*)-1 E po/-A2RT (2.16)

where E is the pulse energy, kT(k*)-1 is the nonradiative yield, E is

the molar absorption coefficient, and pO is the sample partial pressure.

It is assumed that a is small so that the following relation is valid:


1 -e-'L aL = E poL/RT (2.17)

Equation (2.16) is the principal constraint on sample selection.

A well-chosen sample should have a high vapor pressure at room tempera-

ture, a high singlet-triplet absorption coefficient, and a low radiative

yield. The laser output pulse energy must be high in the triplet

absorption region, the cell radius relatively small, and an inert

monatomic carrier gas should be used to increase the y-1 term and to

insure that the acoustic approximation is valid.

In addition, the sample should not be readily quenched by collisions

(especially with the carrier gas), should not readily decompose, and

should not undergo any sort of photochemistry. It should also have

triplet and singlet manifolds well separated relative to the spectral

bandwidth of the excitation light source. The triplet lifetime must

also fit in the time window given by


td < t < minimum of (tT,tw) (2.18)

Another limitation is imposed by the frequency response of the

pressure transducer and associated electronics. An upper time limit








due to low frequency roll-off of the pressure detection system response

(typically 6 db per octave) may limit observation times to substantially

less than the thermal damping time tT unless care is taken in the

design of the pressure detection system. Even worse is the lower time

limit given approximately by


t, = (2fh)-1 (2.19)

where fh is the -3 db high frequency cut-point. High frequency roll-

off for a capacitance microphone is typically 12 db per octave [67].

Since the desired pressure rise is actually a sum of two disparate

exponential rises followed by the thermal reequilibration exponential

decay, it is necessary to have high frequency response. Otherwise, the

pressure detection system will be rise-time limited and thus unable to

distinguish "fast" heat from "slow." Thus, the triplet lifetime is

constrained by another time window; it must, therefore, fit into the

intersection of the allowed time frames.


Pressure Transducer Selection

The "ideal" pressure transducer for TROAS should have the following

properties:

(1) high acoustic sensitivity (preferably above 1 mV/Pa),

(2) high, flat frequency response including dc (to facilitate

static pressure calibration),

(3) low acoustic impedance (concordant with gas phase acoustics),

(4) high immunity to elastic waves in the mounting substrate

(e.g., vibrations and light-induced artifacts in the

substrate),








(5) immunity to stray light,

(6) linear response to absolute pressure changes,

(7) low output impedance (for high noise immunity and ease

of amplification),

(8) long-term output stability,

(9) wide operating temperature range and low temperature

drift,

(10) small, rugged, chemically inert construction (for ease of

mounting),

(11) low output drift in reactive environments,

(12) ability to withstand vacuum pump-down,

(13) relatively low cost.

No single pressure transducer available meets all of these condi-

tions or even the first dozen of them. In the previous TROAS study [47],

the pressure transducer was a PitranTM (pressure sensitive transistor)

developed and marketed by Stolab, Inc. Several different pressure

transducer types were used in the present study: capacitance micro-

phones, piezoelectric disks, and piezoelectric cylinders. These types

of transducers are compared with respect to the above properties in

Table 2. A more detailed comparison of capacitance microphones is

presented in Table 3. Properties not listed are similar for the various

microphone types.

Although capacitance microphones are well suited to gas phase

acoustics applications in general, they have several disadvantages

relative to TROAS applications. First, it is usually necessary to

exclude oxygen from the sample cell to avoid triplet quenching. This

is achieved by pump-down of the cell and refill with sample and carrier
























U J
O W
10 0



u 0
u












u




WE

C w

NU
a

: I


0)
4-1
0
r 3
WU -
S V0 3 0)
o o 0 r
- 3 -


N
z


O
1
0T 0

0- i
1 rL
S> I

CM 0 3 3 0
0 i- 0 0


N






U 0- 0









W 0-


1- U -
S 0 =
0 0 *




U U 0
0 C- -t



Q" P S- ^
J 0 r 3Q C
S- U ~


N







SI

0


0 2
LAD


E au

0 0 3




-00
0O V V Wj
0, u -

V-VO 3 *,- 0 v -3

V0E - 4- ci E


C) h

0 -)





4- 4-
0 0
CD




0. a.

3 0
0 0


4-I
4T- Ci- 0
sV 0
*o r


C-

0: -
WU 0
WJE 4-1
> c

4-L S
U 3
.0 > 0
00 0


-0


0 )
m o o


. 0 C 0 -o
a) Ln 3 o
> C\i 5- S




















C7o




























S1) a)
It 5- Nsi
E 4-' 0
I U -
* W 1
.0 -- r-
-T -U 0.
-J CL












CL





.0 I. -
-0
0 C )

ajX S-













r-
SL.




OWO
U -) 0..
C-)-- C




















*0
UCN


UT
o Li












L.
U
-r-
is


0.





N

E



s- N
S I S. 3

o CO o
EO o -- c >







4-1


a- n
.. N


E
tu o

COD I
L. N



5a -- CD
W) I L=













U. lI
0- V
S0 0 >








.0-
C -
E N -



4 (C3) 10
0 0

















a- ) 0
S N-

OC 0
a- o



















0 1.. U -
*r L W, a
^u "iC- 3 a)
o~ c: m d)en -
ua eC V Ll-LL > 4-








gas. Well-designed capacitance microphones, however, have a trapped

gas volume beneath the metallized diaphragm and a deliberate slow leak

to the external environment. For optimum sensitivity the restoring

force on the diaphragm due to the compliance of the trapped volume

should equal that due to tension in the diaphragm [67]. The second

disadvantage is operating temperature range. Although externally

polarized microphones may tolerate operating temperatures of several

hundred degrees C, electret elements depolarize at elevated

temperatures [68].

Piezoelectric transducers are not well suited to gas phase

acoustics because of their severe acoustic impedance mismatch with

gases. The acoustic impedance Za of a medium is the product of the

sound velocity and the density of the medium. Typical values for

solids are in the range 107 to 108 kg m-2sec- ; for gases, about

5 x 102 kg m-2sec- [69]. As noted by Farrow et al. [69], the acoustic

transmission coefficient at is given by


at = (4Zai Za2)/ [(Zai +Za2) (2.20)


where Z is the acoustic impedance of medium i with sound waves
al
normally incident at the media boundary. The transmission coefficient

(at) is the ratio of transmitted wave amplitude to reflected wave

amplitude; gas-solid interfaces have typical values of 3.x10-5 while

solid-solid interfaces may be close to unity (0.9) [69]. Thus,

piezoelectric transducers are susceptible to vibrations and shocks in

the mounting substrate and are relatively insensitive to gas phase

sound waves.








Pressure Transducer Electronics


The output signal of the pressure transducer must be amplified and

possibly conditioned before being further processed. The two primary

pressure transducer choices are capacitance microphones and piezo-

electric ceramics both of which have extremely high output impedances

(greater than 108 ohms). If a voltage preamplifier is used, it must be

located as close to the transducer as possible to avoid power line noise

and cable microphonics. Since the sample cell (and transducer) may need

to be heated to produce an adequate sample partial pressure, the pre-

amplifier must have good temperature stability, wide operating tempera-

ture range, and low thermal noise.

Methods which allow use of a connecting cable between transducer

and preamplifier are generally deficient in other aspects. Charge

preamplifiers are usually noisier (ceteris paribus) than voltage pre-

amplifiers [70]. Parametric amplification of audio frequencies is too

complex to justify the minimal improvement in noise performance expected.

Carrier current or carrier voltage modulation schemes such as AM, FM,

and pulse frequency modulation are all usable if properly designed and

well implemented [71-72]. These schemes are usually, however, difficult

to implement. Voltage preamplification is, therefore, generally the

simplest and most practical method to use.

The actual preamplifier circuit configuration is less important

than the "front end." It is crucial that the front end provide high

gain and add as little noise as possible [64]. The field effect tran-

sistor (FET) is preferred for voltage amplifiers driven from source

impedances above approximately 104 ohms due to the extremely low input








current noise typical of FETs (10-13-10-16 A/Hz1/2). The so-called

1/f noise of junction FETs (JFETs) is substantially lower than that of

MOSFETs for ultrasonic frequencies below approximately 1 MHz [73].

Discrete low-noise JFETs are also quieter than JFETs in IC op amps [74].

The dominant noise parameter for a JFET is input noise current In with

input noise voltage Vn of secondary importance. The optimum source

impedance is V /In. Noise performance will be degraded if nonreactive

components are used to match source impedance to the above optimum

source impedance.

The dominant noise sources in a JFET are channel generated thermal

noise, gate leakage current shot noise, and generation-recombination

surface effect noise [75]. The dominant noise source for high fre-

quency operation is the thermal noise [75]


V2 = 4 k TB / gf (2.21)


where B is bandwidth and gfs is forward transconductance. Low noise

JFETs, therefore, have low leakage and high gain.

The most convenient circuit configuration is probably the simple

source follower. This circuit provides extremely high selectable input

impedance, low output impedance, and high, flat frequency response. If

the input is capacitively coupled, a rather long time constant results

which may, in some circumstances, limit the upper time limit of observa-

tion of the transducer output. This RC time constant is particularly

important for piezoelectric transducers [62].

The effect of the preamplifier input impedance is interesting.

Total noise voltage decreases with increasing input resistance if the


__





40


input resistance is above approximately 5.x108 ohms and if the JFET
14 1/2
input noise current I is below about 2.xl14 A/Hz /. This result is

based on microcomputer calculations based on a composite noise source

model. Details are found in Appendix Three.














CHAPTER THREE
THE EXPERIMENTAL APPARATUS


The apparatus used in the present research may be divided into five

subsystems:

(1) a pulsed, tunable dye laser system with photodetector,

(2) sample cells and vacuum system,

(3) pressure transducers and associated electronics,

(4) microcomputer interface electronics and transient

waveform recorder,

(5) a microcomputer system with adequate software.

The subsystems have been listed in decreasing order of importance.

Previous TROAS studies [47] were unsuccessful primarily due to short-

comings in the dye laser system although inadequacies in the remaining

subsystems were sufficient to compel a complete redesign of the ap-

paratus. The subsystems adopted for this study are described below in

detail.


The Dye Laser System


The dye laser system chosen was a Candela Corporation ED 625-U

flashlamp driver unit and CL 625 coaxial flashlamp. A Phase-R Corpora-

tion high voltage supply (25 KV maximum) provided the 18-22 KV

necessary to fire the flashlamp and cause lasing in the dye. The laser

mirror mount and grating mount were also Phase-R products. The dye

solutions were turbulently pumped by a Micropump (Model 10-84-316-852)








high flow centrifugal pump and filtered and bubble trapped by a Pall

filter assembly (resin-free polypropylene 0.45 pm filter cartridge and

polypropylene housing). Tuning was effected using an inexpensive

Edmund Scientific diffraction grating (5000 A blaze, 600 1/mm) and a

40% reflectivity (420 nm -650 nm) front reflector from Candela. The

dye reservoir was a machined stainless steel tank not equipped with heat

exchange coils. The windows on the coaxial flashlamp were antireflec-

tion coated in the visible on the outer faces only. The laser cavity

length was approximately 2 meters. The system will lase broadband

(70-100 A) and not tune if the cavity is shortened by a factor of three.

The dye solutions were prepared from laser grade dyes supplied by

the Exciton Corporation. Concentrations of 10 10-3 M were used with

200 proof ethanol or Spectro-grade methanol as the solvent. Solution

volume was one liter. Dyes used successfully were, in Exciton's nomen-

clature, Rhodamines 6G, B, and 110; Coumarins 540, 480, 450, 440;

Fluorol 555; Kiton Red S; and Laser Dye 473.

The spark gap trigger in the flashlamp driver unit was pressurized

with He or dry N2 at 8-12 psi. The laser driver unit could be triggered

either under manual pushbutton control or via an optocoupler circuit

under microcomputer control. The circuit used is shown in Figure 5.

A commercial optocoupler was not used because of their relatively low

(less than 7500 V) breakdown voltages. No significant EMI or RFI was

produced by a laser firing sequence.

Alignment of the optical cavity was achieved with the aid of an

adjustable aperature and a He-Ne alignment laser. The aperature was

used to coaxially align the dye laser optics, dye cavity, and sample

cell. Three passes of the He-Ne 6328. A beam through the system are


































0
*-
0C

41


U



r-
o-0







*r















4- -

















u




CL
13

































U
C)










- -


















.. --
U-
><-u






44













4.1



E




E
VS


OL







v






N-
0i--





E "3

1: 4-1 *r
e 4








sufficient to produce excellent alignment. The He-Ne laser is approxi-

mately 5 meters in front of the dye laser cavity. Fine adjustment, if

necessary, is accomplished by use of the micrometers on the optics

mounts. Performance characteristics of the dye laser system are shown

in Table 4.

Several comments on this dye laser system are in order. First, the

system would lase under very poor conditions; the cavity may be a cheap

mirror and a Suprasil window. An energy meter must be used to optimize

system performance. Second, the dye solution heats up when turbulently

pumped and pulse energy falls about 20%; the dye reservoir should have

heat exchanging coils for optimum performance. Third, the power density

is high enough (almost 1 MW/cm ) to "burn" the diffraction grating so

inexpensive 600 I/mm grating replicas are recommended. Fourth, fresh

dye solutions should be used whenever possible because the first few

score of shots are the most energetic. When the system is almost per-

fectly aligned and fresh dye is used, a sharp, snapping sound accompanies

the laser pulse. The grating will most likely be damaged when this

excellent lasing occurs. Finally, if the laser did not lase when

aligned, it was realigned once or twice. This required no more than

ten minutes. If it still did not lase, the dye solution was changed.

Flushing the system with 190 proof ethanol followed by one flush with

200 proof ethanol eliminates dye contamination problems.

The laser power was monitored with a silicon PIN photodiode (MRD-

501) with a current limited, regulated 22.5 volt bias supply and 3.01 K

ohm photocurrent conversion resistor. With the aid of machined Teflon

attenuators, the detector system was used to detect pulse intensities

as high as 5. MW. In addition, a Gen-Tec ED-200 joulemeter was used to

obtain accurate pulse energies for comparison purposes.























Table 4. Performance characteristics of


Property


the Candela dye laser system.


Achieved performance


Pulse energy 2.5 J

Pulse duration 5.x10-7 sec

Pulse repetition rate 0.1 Hz

Spectral bandwidth 0.2 nm

Tuning range 410. nm to 640. nm








Sample Cells and the Vacuum System


Six different sample cells were constructed for use in the present

TROAS studies. In addition, the sample cell used by Wrobel [47] was

altered and used in the initial experiments. The salient features of

these cells are given in Table 5. Initial attempts to comply with the

constraints imposed by the theory of TROAS were loosened in later cell

designs when consistent results could not be obtained. Specific

information concerning each cell is given below.

Cell #1 was the aluminum cell used by Wrobel [47]. The cell was

modified to accept a capacitance microphone and a copper tube was cold

soldered inside (with gallium) to increase corrosion resistance. It

was not vacuum-tight after the modifications and was inconvenient to

heat.

Cell #2 was the primary sample cell used. It was machined from a

9.5" long, 2.5" diameter rod of stainless steel which was drilled,

reamed, and ballized to 1.002" inner diameter. The ends were machined

for "O"-ring seals and 2" diameter, 1/8" thick Suprasil windows.

Pump-out ports 1" from each end were Cajun Ultratorr connectors stain-

less steel-soldered into the cell. Two pressure transducer ports were

milled into the cell to allow use of one while improvements were tested

on the other port.

The cell was temperature controlled to 0.1 C by an Oven Indus-

tries, Incorporated,proportional temperature controller driving a

doubly insulated heating tape. Feedback output was provided by a

thermistor mounted in a reamed hole 0.050" from the inner cell wall.

Cell temperatures were measured with a thermometer when necessary.













Table 5. Sample cells.


Sample cell Length and Vacuum Suitable
umber c Composition radius V pressure
(cm) y transducers

1 aluminum with L = 16.5 He leak Pitran,
Suprasil A = 0.64 tight capacitance
windows microphone

2 stainless L = 23.45 He leak piezoelectric
steel with A = 1.27 tight transducer,
Suprasil capacitance
windows microphones

3 Pyrex tubing O.D. =0.94 He leak piezoelectric
Teflon valve I.D. =0.61 tight transducer

4 Pyrex flask A = 2.83 He leak piezoelectric
and side arm tight

5 stainless L = 5.0 not He capacitance
steel with A = 0.36 leak tight microphone,
Suprasil piezoelectric
windows transducer

6 Pyrex test L = 15.0 not He capacitance
tube A = 0.76 leak tight microphone

7 window glass L = 4.80 not He capacitance
and Pyrex A = 8.40 leak tight microphone
cylinder








The cell was rigidly mounted to a 2' by 3' by 1" thick drilled and

tapped (1/4-20, 2" centers) aluminum plate which served as optical

bench, mounting platform, and system ground. The cell may be isolated

from ground with Teflon straps (to avoid ground loops). See Figure 6.

Cell #3 was a Pyrex and Teflon 5 mm valve with one "arm" sealed.

See Figure 7. A shielded piezoelectric transducer on the outside of

the glass was intended to pick up elastic waves in the glass produced

by an illuminated internal sample melted or sublimed onto the inner

wall of the glass tube.

Cell #4 was a Florence flask with a Pyrex and Teflon 5 mm valve and

tungsten feed-through electrodes in a separate Pyrex apparatus. This

cell was used for solutions and gas phase measurements with a suspended

piezoelectric transducer. See Figure 8.

Cell #5 was a Cajun Ultratorr 3/8" tee with 3/8" diameter, 1/16"

thick Suprasil windows in the arms of the tee and the pressure trans-

ducer in the stem. Minimal machining was required to enable the tee to

accept either a capacitance microphone or a piezoelectric disk. A

spring-loaded electrode was used when the piezoelectric transducer was

in place and the signal was brought out through a vacuum-tight BNC

bulkhead feed-through connector soldered to a brass 3/8" nut from a

similar Cajun 3/8" tee. The cell was vacuum tight though without need

to be evacuated.

Cell #6 was a Pyrex test tube with a Teflon tape-wrapped capacitance

microphone snugly inserted into the end. The cell was illuminated per-

pendicular to the long axis and heated uniformly with a heat gun.

Cell #7 was a Pyrex cylinder section with 1/4" sidearm and 1/2"

drilled hole for mounting the capacitance microphone. Ordinary window


























































V-


-0









i
Q:
I-

10




























































































ul 3
CO
S. C'J -0
S C
3. *i
1.0






























Figure 7. TROAS sample cell number 3.



















Vacuum System


5 mm Pyrex Valve


Teflon Stopcock
With Ethylene-
Propylene "O"-rings



Molded Aluminum Enclosure






Piezoelectric
Transducer


Sample
Area





























Figure 8. TROAS sample cell number 4.












to
Differential Amplifier


Tungsten Electrodes



Spring Loaded Nickel Silver
Contacts


to Teflon 5 mm Valve and
Vacuum System


Piezoelectric
Transducer








glass plates were carefully epoxied onto the ends of the cylinder to

produce a highly transparent cell. See Figure 9.

The vacuum system used must be capable of removing oxygen from the

sample cell to avoid the possibility of triplet quenching. A liquid

nitrogen-trapped rotary oil pump system is sufficient to attain 10-4

torr which is adequate to exclude oxygen from liquid samples by repeated

freeze-pump-thaw cycles. Solids may be deoxygenated by solution in a

suitable solvent and subsequent pump-down. The system pressure was

measured with a Wallace & Tiernan gauge (0-800 torr in units of 1 torr).

The vacuum system was tested for leaks with a Veeco MA-2 mass spec-

trometer helium leak detector. Pressures to 10-4 torr were routinely

obtained in the system.


Pressure Transducers and their Associated Electronics


Several hundred man-hours were spent in the design and construction

of improved circuitry and transducer mounting techniques. Among the

techniques tried without success were laboratory-made, externally polar-

ized capacitance microphones and FM carrier current modulation by

microphone capacitance changes. The Pitran pressure sensitive trans-

ducer was rejected due to moderate sensitivity (380 pV/mtorr), high

temperature drift (400 mV/C), differential pressure mode operation, high

acoustic impedance, extremely high susceptibility to damage (especially

in mounting), and sensitivity to mounting substrate vibrations.

The most important parameters to be considered in selection of

pressure transducers for TROAS are sensitivity, acoustic impedance,

and flat, broadband frequency response. Unfortunately, these parameters

are usually dependent on each other. Consider the capacitance






























Figure 9. TROAS sample cell number 7.























Pyrex Cylinder and
Pump-out Port












t jMicrophone
I Port


Window Glass Plates








microphone. The resonant frequency of a standard unpinned disk micro-

phone (approximately 80% of the high frequency cut-point) is


resonant = (fr/m)1/2 (3.1)

where the restoring force fr is roughly proportional to acoustic im-

pedance and diaphragm mass m is proportional to the square of the

diaphragm diameter. Thus, high frequency response requires low sensi-

tivity and small, rigid, thin diaphragms. Two high quality microphones

are compared in Table 6. The microphone actually used is also

described in Table 6 [76].

Laboratory-made microphones worked, though not as well as commer-

cial capacitance microphones, probably due to the method of construction

[67]. The microphones were constructed of aluminized mylar without a

trapped gas volume (to allow pump-down).

The piezoelectric transducers used were 1/2" disks of PZT-5H from

Vernitron, Inc. Also used were two sizes of piezoelectric cylinders

and a thin, high frequency response (5 MHz) disk also of PZT-5H. One

immediate problem with piezoelectric transducers is calibration. The

output voltage of the transducer is [62]

K' Ap Ap
V 1T p (3.2)
0 i + iwT C (


where K is the modulus of the piezoelectric, Ap is transducer area, C

is the transducer and electronics shunt capacitance, Ap is the pressure

change, T is the RC time constant of the transducer and amplifier, and

w is the frequency of the pressure change. For w much greater than

-, the output voltage is proportional to pressure change. For static
,the output voltage is proportional to pressure change. For static

















Table 6. Capacitance microphone comparison.


Radio Shack B & K Dynasciences
Property 33-1056 4144 814

Acoustic 3.2 mV/Pa 50. mV/Pa 0.56 mV/Pa
sensitivity

Resonant 6. KHz 8.3 KHz 90. KHz
frequency

Frequency 20 Hz 10 Hz 50 Hz -
response 12 KHz about 15 KHz 120 KHz

Corrosion low low high
resistance

References [76] [63] [72]








or low frequency pressure changes this is not the case. Calibration of

the piezoelectric ceramic transducer used in this research is described

in Chapter Four.


Pressure Transducer Electronics


Many preamplifiers and amplifiers were designed, breadboarded,

modified, constructed, tested, and usually rejected. These included

preamplifiers with discrete JFETs, discrete bipolar transistors, and

bipolar, Bi-mos, or JFET input op amps. Although the noise model

(Appendix Four) indicates that the discrete JFET-input voltage follower-

with-gain circuit is optimal, this circuit is not easily implemented in

the vicinity of the heated cell. Unfortunately, this is necessary

because of the very high output impedance of the pressure transducer.

The circuit eventually adopted was a source follower (1010 ohms input

impedance) used to provide impedance conversion. The circuit used a

discrete JFET (2N5486) in an ac-coupled configuration with input time

constant of 103 seconds. Voltage gain was obtained with a Tektronix

Model 26A2 differential amplifier which allows switch selection of

gains from 102 to 105, adjustable bandwidth from dc to 1 MHz, and dif-

ferential mode inputs. This was especially convenient when used with

the data acquisition system and Fast Fourier Transform program because

the bandwidth may be adjusted to avoid aliasing. Differential mode

operation was necessary to eliminate power line hum pick-up. See

Figure 10 for the circuit schematic.

Having selected a pressure transducer and associated preamplifier,

the electrical noise of the system was measured under no-signal condi-

tions. For the system described, this was 11. pV rms determined by






















































c T
i








averaging 25 acquired noise spectra and assuming a crest factor of 4

[74]. The system bandwidth was 0.1 Hz to 1. MHz with shorted inputs

and gain of 5.x103. With the sensitivity of the Radio Shack capacitance

microphone given as 3.2 mV/Pa and the conversion factor of 1 Pa =

7.50061 mtorr, the sensitivity is thus 4.2x10-4 V/mtorr. If the limit

of detection is considered to be a signal to noise ratio (S/N) of unity

(no signal averaging), then the pressure required to equal electrical

noise is 0.20 mtorr. This pressure equivalent noise level is very im-

portant in determining the suitability of samples for TROAS studies.

By direct comparison of the piezoelectric transducer with the Radio

Shack capacitance microphone, the pressure equivalent noise level of

the former is found to be approximately 20. mtorr. This comparison was

carried out in cell #2 with NO2 (a strongly absorbing substance).


Microcomputer Interface Electronics

The electronics necessary to interface the microcomputer to the

various amplifiers, chart recorder, peak detector, et cetera is shown

in schematic form in Figure 11. The circuitry consists of relatively

standard, independent subsystems interfaced to the microcomputer I/O

lines which in turn are provided by Motorola 6820 Peripheral Interface

Adaptor (PIA) LSI chips tied to the microcomputer address, data, and

control busses. The functions of these subsystems are listed below:

(1) chart recorder output (0-1 volts) via an 8-bit D/a

converter,

(2) four input channel multiplexer- selectable active filter-

gain controlled amplifier- offset and level shifter- 8-bit

A/D converter,

































































u
0-C
L
4-

O
a,









































Ua



40
an
u





0-1-






.4-I
u,
o
C
r 1





a-





0=

5-


















U-)
;:

a>'r
L/)







66



sU
aru
cu u
C i, 3
*i- ~ ~ -1 3- Q.
Svi c m

-^~Q O.'llC f




oE5-- ICi
0 I S S









m u a 0
au





-lsC Q O 5. Q--
__ __ 0 CO D 1 *r-






ca a ca
L C



























uco
i -C ca


!- ,c- D Q 0 to S c= !--















ri 4-C i CA< .
(Ua n --- :a






I- I n c c


















cm c0
>~~1 GJ 0 (T *r a o -_
*-100 r- T----C __ /

















4 -- 7 -01-
4 c L5i
_ CF






















'a. In rQ
14 14























aM-
0)r ^^ 'p ---
**o- l ^ \ I^^ ^^





'--T --* 1-1 4-i -

+-)~O -I t/ (_ O
4-- ro
cn C:11 /\11 1 a
r IT 1+ -I ___________


aj

C ---- j - --


vi -l- LIO I I 1 i.

*r ( o I I I 1 / 01 -a I /- ---
'r l-- I I I 10 *r I Q /*
*^ QI I I 0 -L I I

ii ^/Tr 7^
___ //

^V^L^Ym "
_____ ^ C
c r-< o ^ - __ -
*r -- ^ t -- ^ < - s c






r i -- N4a c2 m








(3) transient recorder and stepping motor control lines

(LSTTL or n-MOS),

(4) photodetector peak detector with CMOS 4016 transmission

gate reset,

(5) photodetector linear response circuit with additional

CMOS 40014 Schmidt trigger optional output.

The stepping motor control lines and linear CW light detection cir-

cuitry are not used in TROAS; the circuitry was designed to allow con-

ventional OAS with a Spectra-Physics Model 164 Argon Ion Laser and a

Model 375 CW dye laser. The flashlamp-pumped dye laser, sample cell,

light detector, and Spectra-Physics lasers were coaxially aligned, and

the interface electronics and software facilitated rapid changeover from

TROAS mode to OAS mode or vice versa.

The data acquisition system incorporated a Biomation 805 Waveform

Recorder with selectable sampling rates to 5 MHz, 2048 words of 8 bits

each memory storage, pretrigger recording, sensitivity to 0.1 volts

full scale, and microcomputer compatible data readout. A Tracor

Northern NS-570 Digital Signal Averager with 20 psec (minimum) dwell

time per channel, 9 or 12 bit data conversion, and automatic normaliza-

tion of summed spectra was used in the earliest attempts at TROAS.

The signal average was found to be too slow and susceptible to RFI

generated by a laser firing sequence.


The Microcomputer System


The Commodore 2001-8 PET microcomputer system was chosen for use

in the experiment for many reasons. The most important are hardware

and software compatibility with previously used KIM-1 microcomputer


~








systems; 6502 microprocessor-based, efficient architecture; fast 8 K

BASIC with assembly language subroutines; built-in cassette tape drive

and video monitor; low cost (S795).

The main program used in the TROAS studies is listed in Appendix

Four. The listing contains no "remarks" because of memory restrictions

imposed by the relatively small amount of available random access

memory (8K of RAM). The program consists of independent subroutines

called from the keyboard by number. These are briefly described

below:

(1) initialization and status of microcomputer interface

subsystems,

(2) data acquisition from transient recorder and control of

laser firing, light peak detector readout,

(3) chart recorder output of raw data,

(4) data storage in named cassette files,

(5) 256 point Fast Fourier Transform subroutine based on the

Fortran version by Bell [77],

(6) video display of numerical data concerning FFT spectral

intensities and powers,

(7) line spectrum output to chart recorder of FFT results,

(8) assembly language subroutine to eliminate the effects of

the time delay (about 100 usec) in the laser firing circuit

optocoupler,

(9) least squares exponential curve fit subroutine.














CHAPTER FOUR
SAMPLE SELECTION


In order to determine the validity of the theory of TROAS, suitable

calibration standards must be run. These compounds should meet the con-

ditions imposed by the theory and those arising from necessary compro-

mises in the experimental apparatus. Previous attempts by Wrobel [47]

failed to verify the fundamental two-component pressure rise expected

to occur immediately after absorption of light by the sample. The

observation of the pulsed optoacoustic effect by other investigators

has been previously mentioned; those observations do not, however,

constitute evidence for the validity of the theory of TROAS.

The first compound used in the TROAS experiments reported herein

was 2,2,4,4-tetramethyl-l,3-cyclobutanedithione (dithione). Oithione

is a deep red, crystalline solid with distinct camphoraceous odor.

This compound and the related dione have been the objects of several

theoretical and experimental investigations by Vala and coworkers [78-

80]. Although the UV-visible spectrum of dithione is still without a

definitive interpretation, several important facts are known. First,

several definite triplet peaks at 5943 A, 5926-5922 A, 5836 A, and

5738 A are easily accessible; in fact, the dye laser output power with

Rhodamine 6G dye spans these wavelengths and is highest at about 5900 A.

The second important property is a relatively high molar absorption

coefficient (approximately 1.5 1 mole- cm- ). The third property is








the significant vapor pressure (approximately 0.50 torr at room tempera-

ture). These last two results are due to Powell [81].

The pressure rise expected from dithione may be calculated using

equation (2.16). The expected rise p' is 24. mtorr assuming E = 1.5

1 mole-1 cm-1, PO = 0.50 torr, T = 300 K, y = 1.4, k = 0, E = 1 J, and

A = 1.27 cm (for cell #2). This is much larger than the pressure

equivalent noise level obtainable using the Radio Shack capacitance

microphone (0.20 mtorr) and approximately equal to that of the system

if a piezoelectric transducer is used (20. mtorr).

Since only the piezoelectric transducer can tolerate vacuum pump-

down, a carrier gas at atmospheric pressure must be used with the

dithione to avoid damaging the capacitance microphone. Helium gas is

an excellent choice because of its high purity, inertness, high sound

velocity, and high heat capacity ratio. The effect of trapped oxygen

in the microphone would be to convert slow heat to fast heat by triplet

quenching. This problem will be deferred until later.

The three time constants tT, td, and tw may be calculated from

equations (2.12) -(2.14) respectively. Under the above conditions,

these are tT = 0.73 sec, td = 6.6 usec, and tw = 0.86 sec. For air

at STP, they are t- = 2.7 sec, td = 19. isec, and t = 7.3 sec. The

acoustic delay times td were calculated assuming a = A/2 = 0.635 cm.

With the frequency response of the microphone given as 20. Hz to

12. KHz, the corresponding allowed time window is 42. usec to 25. msec.

Thus, the pressure detector limits the triplet lifetime which can be

observed at both extremes. If the oxygen in the trapped volume in the

microphone quenches the triplet, the result may well be solely fast

heat, i.e., heat fast relative to the 42. usec limit. It is also known








from previous studies [82] that the triplet lifetime of dithione is

less than 1 msec. In addition, dithione definitely photochemically

decomposes and also thermally decomposes to the mixed ketone and other

unknown species. This decomposition to a viscous red oil occurs at

temperatures below the reported decomposition temperature [83]. The

dithione used in these experiments was supplied by the research labora-

tories of Tennessee Eastman. In general, impurities caused no problems

in the TROAS experiments so that compounds were used as supplied, which

was in no case less than 95% purity.

In order to overcome the problem of decomposition and to obtain

known radiationless rate constants, several compounds exhibiting the

internal heavy atom effect were studied. These were naphthalene,

2-chloronaphthalene, anthracene, l-chloroanthracene, and 9-bromo-

anthracene. These compounds have been studied for many years and

consequently several of the triplet energies, molar absorption coeffi-

cients, and radiationless rate constants are known or can be estimated

reasonably accurately. The lowest triplet in naphthalene, at 77 K in

a glass [84], is 4695 A. It may be expected to shift somewhat in the

gas phase. Other relevant data (and references [85 -87]) are given in

Table 7. Figure 12 shows the structures of the organic molecules

studied. Unfortunately, the molar absorption coefficients are very

small for all of these compounds. It was hoped that the relatively high

vapor pressures of several of the compounds, particularly naphthalene

(10 torr @ 35.8 C [85]),would compensate for the low absorptivities.

In addition to the above compounds, both 19 and NO2 were tried.

Both of these substances absorb strongly throughout the visible, have

been studied extensively, and have complicated spectra. Iodine has

















<. L CO
C 0


4- ^ 1
0
Co
C, m cj
CO C) C '
*co )a ri
gu r -


0 S


CL L' S..

0*
Pi






-, ~-0
I.- CI
- S-J 40

QH -


g0










I IC
























0>,
+' re ao












Co
























M
3. :
I- ,-*






U-



Co
i a)













(U





Cl'C

C C~
-005
ui









u-o
*- C-
o a



jit








C-r0
H C
.c~
4- C


L

4




U co
0


rc~
Co

-ci




'. rn C r^




LO Lo J
coooc c LO



zi)CJ C oi M =
m 0'] 0l_ CD
LCO Lo L
i-PI 0 U3
V)-L Co


0
S-

C -- 1

U U



0 0
-4-







fl0 CO LC)



M m C co
C--_ c .
Ip,-,


a) ar
s: ra oj D

C C1 o- C o U I aC
C L? U c-U (U
C- C C 0 c s E
- < 0 U0 i U 0-I
3 5 S J -C C
CD- u -C 4 nC
0 'rJI
< r C rca


CC-.J m
~oQ I LO

I-II
C9 C
T C C 1CD
C=1--I Co 0 S

X -. o

tvi Lrtj 's UT u o






C\J, I--I -u





xn -






o o





























3,
+3
a
dI
E

S..
di
a
..



Cl
C
O
ci
4:






0
di



















/1-
1
CU
a
3,


-0
V)






3

di
1)






a
2

"3





+i

-a

+3

rC
vl








ai
'a





4-3

U
a
01
03








di



U-






- ____________~~~~~_____________ __
















c I
uu



00


U /U


S0=0


0=
RJ\


ac
= i-



u
Y


(N~
II



~KJ


-u

u

8-
2Q







been studied by conventional OAS techniques with interesting results

[31]. When moist, iodine attacks aluminum very rapidly and stainless

steel (316) slowly. It also attacks the silver metallization on the

PZT-5H piezoelectric transducers yielding silver iodide and inter-

mittant operation. Nitrogen dioxide does not appreciably attack either

aluminum or stainless steel. It was used to compare the sensitivities

of the capacitance microphone and the piezoelectric transducer. Reac-

tive substances reportedly cause drift problems with externally biased

capacitance microphones [87]. It should be noted that neither iodine

nor NO2 is suitable for a rigorous TROAS study because neither of them

is adequately modeled by a simple two-state model.

Biacetyl (CH3COCOCH3) is an excellent compound to study because of

its unusual properties--high phosphorescence quantum yield (0.149) [88],

low triplet energy (57 Kcal/mole), high vapor pressure, and unusual

transition from small molecule behavior to large molecule behavior at

an accessible wavelength (4450 A [89], 4430 A [32]). It is a small

enough molecule to be accessible to detailed theoretical studies and

relatively easy to study since both fluorescence (from S1) and phos-

phorescence (from T1) are readily observed and spectrally distinct.

In addition, the photochemistry of biacetyl has been extensively

studied [90-91] especially by Noyes and coworkers [92-94]. It is used

routinely as a gas phase emission standard and as a triplet energy

acceptor in the study of the photochemistry of other molecules [95].

It is also known [96] that biacetyl, at pressures below 10 torr,

exhibits essentially no collisions during the lifetime of the lowest

excited singlet state 51 (T = 10-8 sec).





76


Among the disadvantages of using biacetyl for TROAS studies are

its reactivity, low S0 T1 oscillator strength (1.4 x 10-7 [87]), and

foul odor (rancid butter). On the whole, the advantages probably

outweigh the disadvantages insofar as TROAS is concerned.















CHAPTER FIVE
EXPERIMENTAL ARRANGEMENTS


The major components of the attempted TROAS experiments are

presented (in concise form) in rough chronological order in Table 8.

The arrangement of the components was essentially that of Figure 2

though, of course, the initial experiments were performed without the

aid of either microcomputer system or transient waveform recorder.

Consequently, the complexity of the experimental apparatus increased

(as necessary) to obtain results and eliminate interference.

The previous experiments had been performed in the near-UV (320-

420 nm) where the performance of the Phase-R DL-1200 coaxial flashlamp-

pumped dye laser was very poor, at best [47]. It was decided, therefore,

to use this system in the visible region only, where dye performance is

much better and alignment of the system is far easier. With Rhodamine

6G dye, the maximum output energy per pulse in the Phase-R system was

guaranteed to be at least 250 mJ as opposed to the obtained maximum

near-UV output of 150 pJ [47]. Poor laser performance, in general, and

low pulse energy, in particular, were felt to be the primary impediments

to successful completion of the previous work.

As mentioned in the previous chapter, the maximum energy output of

Rhodamine 6G occurs at approximately 590. nm which is almost ideal for

excitation of the lowest triplet state of dithione. In addition to

favorably placed triplet bands, dithione has a molar absorption co-

efficient of about 1.5 1 mole-cm- and a vapor pressure of about 0.5

77


























M me =


C3
In




















S11
s-~









T3 0
C3






4a







Sa




L.

0





4-1







S-
C
(a





















U
5-
-o-
Si
(-




















C-
0



C-





















'-S
Ga





=I
i-


u S- c.


U
-SD

0-
(a
(S3 C.



-(5-
.2 SD (a
r0 (a
(a -





Q-z CMa
C- :


i- CMi CMJ

















c C =
o a C
r 00
o o o
r- -=
r
C) C) C)
r 5


SD
=



5-




0 C
SD Cr SD


SD
3:


S-

m r a,
2: (D 2:



SD S SD
C C C,
C C
C C


Lr) O Ln
C) = 0
cc CO cc


.D
S1
r w

-o

-I


-- ..- -
C.- C- 0


5-.







-C

a)


C\j C CM C C \M C C\J


4-1 5) S
0 0 0


SD
=
SD
in

iS C
C C
Vc >
O>> -
= 0


CL -

= S C)



0,->




U LO*r-
Z -
o CM
o LD C


F- : (















V)
in
Ci

in

Ci

Ci


c
1..








CT
Li








o i
J














v,
Ci i








-I--
cn


CiC)
0

uin -


41/
i








S-U
LV
3 3



LIT)





v,
a;
U
s-


0. ~ ~ r 0. 0. 0. Li0 Li. 0 i L


0/ C) a) Q)ao. Qa)a. Ci (D C) a) w a)
1 1i Ci i/i i i i i Ci i i
S- -C - -






u : c s= r = r = = c r
010
a) Ci i i C a) C) a) a)

x ea 0 0 a 0 v I 0 Ci d 0d M






a)

3

i Ci ci O ) C ) Ci cM i C ji Ci C



CL
Cil




~0










a) a aa)
I = I iiJ aM =C
0 i M ci = -C i n C


Ci I Li Q. 4 l C D +
i 3-IiI (T 3 *' <


'j (3cC i: '1: .


i

l--



























O O C CO C D CO CO
o o Co0 0 0 cc 0 30 .o c.
C C C 3 C 3 3 3


Ci
0



Ci



Co

Ci








torr at 300 K. Therefore, a small quantity of solid dithione was placed

in cell #1 with the capacitance microphone (hereafter CM) connected

directly to the input of the Tracor Northern NS-570 digital signal

average. Whenever possible, samples were run at room temperature

first to see if a spectrum could be obtained. If no spectrum was ob-

tained, the cell was heated via a doubly-insulated heating tape driven

by a feedback controlled, fully proportional temperature controller.

Dithione samples were heated as high as 85. C to increase sample partial

pressure while not greatly increasing the likelihood of thermal decom-

position. Temperatures were also limited to avoid thermal depolariza-

tion of the electret microphone diaphragm and to avoid significantly

increasing the integral JFET preamplifier's thermal noise.

Samples were also usually run with air as a carrier gas in the

initial attempts with each sample since the effects of oxygen quenching

of the excited triplet would be confined to converting slow heat to

fast heat. The CM was only used with air at atmospheric pressure as

the carrier gas because it is not possible to pump down the CM without

causing diaphragm rupture as the trapped air beneath the diaphragm

expanded.

Despite these limitations, initial results were encouraging (see

Figure 13). Unfortunately, the Phase-R system refused to tune or even

lase narrowband; it also suffered numerous breakdowns culminating in

the decision to switch to the Candela laser system. Additional problems

encountered during the initial experiments included sample cell cor-

rosion, possible oxygen quenching of the triplet (in an unexpected

fashion, such as excitation of the metastable la, state), possibly

inadequate low frequency response of the transducer and/or the digital

























0 =
Q.

-u




S-1
Sa)

aj i-



CL
0O
0cm
.0
Ia 0








U l-

C .R
03



















-ci
uci












-C SZ
o" *





2 U )
cOs-
*0 0 =










C- *- aO
-' 0.
.0 3 >
0 uci
4-'





C r
0 -3 -








Eci


u '4-
aj e 0o


(0 0,


5-.




i- ci









en
LL






















































ou




H-


- asJnssaJd








signal average, false triggering of the signal average by the Phase-R

generated RFI/EMI, and thermal decomposition of the dithione after

several hours in the heated cell.

While the Candela laser system was on order, the next attempt was

made using an Avco nitrogen laser-pumped dye laser equipped with elec-

tronic wavelength scanning drive. No results were obtained because the

output pulse energy (200. uJ maximum) is insufficient to produce a

pressure rise signal greater than the pressure equivalent noise level

of about 0.20 mtorr. Therefore, a xenon flashlamp (Vivitar) with

wavelength selective filters (Veb Jenaer Glaswerk Schott & Gen., Jena)

was used to provide pulse excitation of the dithione. A typical result

is shown in Figure 14. The relatively slow data acquisition rate of the

signal average (20. usec/channel minimum dwell time) and CM allowed no

conclusions to be drawn concerning the crucial initial portion of the

spectra; it was decided, therefore, to upgrade the data acquisition

system, associated electronics, sample ceil, and pressure transducer.

In addition, an Ithaco model 1391 low noise, charge preamplifier was

purchased to provide amplification for the recently acquired piezo-

electric disks and cylinders to be used as fast pressure transducers.

Lack of adequate UV optics and pulse energy measurement equipment

led to the decision to use the Candela system only in the visible

region. Dithione was once again tried in hopes of improving upon the

previous promising results. The piezoelectric transducer (hereafter PT)

was used in addition to using the CM. These experiments and the Xe

flashlamp ones previously performed were done using sample cell #2.

This cell was far easier to load, clean, heat, and mount than cell #1.

It could also be pumped dcwn for use with the PT elements. Unfortunately,



































S-





L






Qr
Cq








S-
C-r


(o L))






LO






-C





















u-

X 3




1U
U-
















LC.
4- U






















U-
c)



ra




-r- r



4-



















































w


I-


- ajnssaJd








no significant results were obtained despite modifications to the pre-

amplifier circuitry and heating of the sample cell. It was decided,

therefore, to conserve the small remaining reserves of dithione for a

later attempt.

The next sample substance studied was biacetyl. It was deemed a

reasonable candidate for study by TROAS techniques because of the favor-

able properties listed in Table 7 of the previous chapter. The only

anticipated problems were possibly inadequate triplet molar absorption

coefficient, possible photochemistry or decomposition, and the definite

strong rancid butter odor of biacetyl even in minute concentrations in

air. A technical grade sample was repeatedly freeze-pump-thawed (liquid

nitrogen and 10-4 torr liquid nitrogen-trapped pump) to remove dissolved

oxygen and placed in sample cell #2. The biacetyl sample exhibited its

characteristic oxygen-free phosphorescence [32]. Several experiments

were performed at room temperature using the PT and several different

excitation sources. Broadband laser excitation below the triplet (T2)

at 4430 A gave a null result. Broadband (70-100 A) excitation above

T2 resulted in excitation of the lowest excited singlet with consequent

triplet production via intersystem crossing. The result was large

amplitude oscillations superposed on a small amplitude exponential

pulse. Electronic filtering of the transducer output with an active

fourth-order Butterworth filter (equal component Sallen-Key) removed

most of the oscillatory features at the expense of the risetime of the

buried exponential waveform.

Similar results were obtained when unfiltered broadband Xe flash-

lamp excitation was employed. Triplet excitation via intersystem

crossing from the easily excited lowest singlet is achieved at light








intensities much lower than those routinely obtained with the Candela

system [97]. In fact, triplet-triplet annihilation processes become

important at low excitation source intensities [97]. Unfortunately,

the resulting TROAS spectrum cannot be interpreted without the direct

triplet excitation TROAS result unless the information it supplies is

available from other sources.

It was not possible to tune the Candela system to only excite the

triplet because it refused to tune or lase narrowband (2 A) just as the

Phase-R had. This problem was finally traced to the laser cavity

length being too short. Attempts to make the system tune with a laser

cavity of about 2/3 m were invariably unsuccessful while the system

worked perfectly with a cavity of 2 m. Alignment did not seem to be at

fault though a longer cavity must necessarily emphasize the effects of

the dispersion element (diffraction grating) and the alignment is easier.

An unexpected problem with biacetyl was its ability to dissolve

ketone-resistant "0"-rings with consequent oxygen contamination of the

degassed sample. To avoid these problems, five naphthalene and anthra-

cene derivatives were selected for study. It was hoped that the

stability and relative inertness of the compounds would enable the

experiment to proceed. The molar absorption coefficient of the

2-chloronaphthalene triplet is approximately 10-3 1 mole- cm The

other compounds studied are likely to have similar extremely small

values. It is exceptionally difficult to obtain such values by optical

means. Unfortunately, this means that the expected pressure rise from

these compounds may be below the pressure equivalent noise levels of

the system regardless of pressure transducer used. Without the molar

absorption coefficients this simply is not known. Nevertheless, the








compounds were tried for several reasons: favorable triplet lifetimes,

reasonable vapor pressures, photochemical stability, accessible triplet

states in the visible, and thermal stability.

Naphthalene was the most studied of the aromatic nydrocarbons. As

usual, the initial results were quite encouraging (Figure 15) though

incorrect because the exponential decay is independent of carrier gas

composition. This cannot be explained if the decay is actually due to

thermal reequilibration as it should be. Experiments were done at room

temperature and again at approximately 85 C. Experiments were performed

with piezoelectric transducers, electret capacitance microphone, and a

laboratory-made externally polarized capacitance microphone. Naphthalene

was run neat, with air, and with helium at pressures below 760 torr.

Although the triplet wavelengths are known from 77 K glass phase

results, they may be shifted in the gas phase. Therefore, the TROAS

experiments were carried out by manually scanning the laser from longer

wavelengths (below the triplet in energy) to shorter wavelengths (above

the triplet). The dye laser was usually scanned over about 500 A in

increments of 10 A or less.

The effect of wavelength scanning was quite unexpected. Instead

of a relatively sharp onset of the pulsed optoacoustic effect, the

acquired spectra had maximum amplitude near the dye output energy maxi-

mum. The spectra were reduced in amplitude as an extreme in the dye

gain curve was approached and invariably the spectra had become no-

signal baseline immediately prior to the laser reaching the ends of its

tuning range. This was observed with dithione (Rhodamine 6G), naphtha-

lene (Laser Dye 473), and anthracene (Kiton Red S).





















=
O
Ln Q


O xl




oa
3 0








415
S.- -(







U M












*- ai
C -u
C Oa
C M















(3) f
ae ca
























10-0
-a>






mc
ci0

S- -0





3-:

C L






U.-
11 LW



0)




cra -






















































































- anssid


U


a)




Full Text

PAGE 1

STUDIES IN TIME-RESOLVED OPTOACOUSTIC SPECTROSCOPY By :dward g. voigtman jr. A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1979

PAGE 2

ACKNOWLEDGMENTS I would like to acknowledge Dr. Martin Vala for his help and encouragement throughout these exasperating experiments. His constant optimism and enthusiasm were of incalculable value in seeing the experiments through. I would like to thank my wife Janiece Leach, whose love and understanding were my only joy. Without her faith in me and her personal sacrifices, nothing would have been accomplished. I would also like to thank my fellow graduate students, Joe Baiardo (who helped with the FFT subroutine translation), Bob Brittain, Rodger Capps, and Dave Powell (who goaded me into lengthening the dye laser cavity to effect tuning). Their fellowship was crucial to maintaining my sanity in the face of unrelenting artifact production by the TROAS apparatus. I would also like to thank Ed Whitehead for his excellent machining and Rudy Strohschein for his expert glassblowing. In addition, I would like to thank the staff of the departmental electronics shoo (Joe Miller, Russ Pierce, and Bill Wells) and Dr. J. D. Winefordner, Jr., for the loan of various necessary pieces of equipment and laser dye. I also would like to thank Dr. J. Eyler for the loan of Laser Dye 473.

PAGE 3

The words of Mercury are harsh after the songs of Apollo. Love's Labor Lost l n

PAGE 4

TABLE OF CONTENTS CHAPTER PAGE ACKNOWLEDGMENTS H PREFACE iii LIST OF TABLES yi LIST OF FIGURES vii ABSTRACT ix ONE INTRODUCTION 1 TWO THEORETICAL CONSIDERATIONS 19 Introduction and Direction 19 The Extended Solution of the Heat-Flow Model 20 Excitation Source Selection 26 Sample Cell Construction 27 Sample Selection 32 Pressure Transducer Selection 33 Pressure Transducer Electronics 38 THREE THE EXPERIMENTAL APPARATUS 41 The Dye Laser System 41 Sample Cells and the Vacuum System 47 Pressure Transducers and their Associated Electronics 56 Pressure Transducer Electronics 61 Microcomputer Interface Electronics 64 The Microcomputer System 67 FOUR SAMPLE SELECTION 69 FIVE EXPERIMENTAL ARRANGEMENTS 77 SIX RESULTS 108 SEVEN SINGLET TIME-RESOLVED OPTOACOUSTIC SPECTROSCOPY .... 122 IV

PAGE 5

APPENDICES ONE RELEVANT OAS RESULTS 126 TWO TROAS NOISE EQUIVALENT POWER 132 THREE CAPACITANCE MICROPHONE PREAMPLIFIER NOISE MODEL ... 134 FOUR MICROCOMPUTER PROGRAM AND I/O LOCATIONS 141 REFERENCES 150 BIOGRAPHICAL SKETCH 156

PAGE 6

LIST OF TABLES FABLE PAGE 1 A comparison of light sources considered for use in TROAS experiments 28 2 Pressure transducers which may be used in TROAS experiments 35 3 Capacitance microphones considered for use in TROAS experiments 36 4 Performance characteristics of the Candela dye laser system 46 5 Sample cells 48 6 Capacitance microphone comparison 60 7 Relevant triplet properties of the compounds studied by triplet TROAS 72 8 A summary of the TROAS experiments performed 78 9 Peripheral systems memory locations 148 v i

PAGE 7

LIST OF FIGURES FIGURE PAGE 1 Electronic states and rate constants typically of importance in the photophysics of polyatomic molecules 7 2 The idealized TROAS spectrum resulting from direct triplet excitation 12 3 Schematic representation of a typical TROAS apparatus 14 4 Schematic representation of a versatile OAS apparatus with dye laser excitation 17 5 Optocoupled trigger circuit for the Candela dye laser 44 6 TROAS sample cell number 2 51 7 TROAS sample cell number 3 53 8 TROAS sample cell number 4 55 9 TROAS sample cell number 7 58 10 Pressure transducer impedance conversion circuit ... 53 11 Schematic representation of the microcomputer interface system and electronic subsystems used in tne TROAS experiments 56 12 Structures of the sample substances used in the TROAS experiments 74 13 The TROAS spectrum of dithione obtained with a Phase-R DL-1200 dye laser, Rhodamine 6G due lasing broadband, and a capacitance electret microphone .... 32 14 The TROAS spectrum of dithione obtained with a Xe flashlamp (unfiltered) and a capacitance electret microphone 85 vi 1

PAGE 8

15 The TROAS spectra of naphthaline in helium (upper trace) and in air (lower trace) 90 16 The TROAS spectrum of iodine at 25. C in air at atmospheric pressure 94 17 The TROAS spectrum of 9-bromoanthracene in air at atmospheric pressure with the piezoelectric transducer used to detect the signal 96 18 An artifact TROAS spectrum obtained with air at 25. C and atmospheric pressure in the sample cell ... . 99 19 The output of a piezoelectric transducer directly exposed to a laser light pulse of high intensity. . 102 20 The TROAS spectrum obtained with an empty (10~ 4 torr) cell no 21 A voltage preamplifier and externally polarized capacitance microphone circuit 136 22 The voltage preamplifier and externally polarized capacitance microphone circuit of Figure 21 with appropriate noise sources added 138 V 1 1 1

PAGE 9

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy STUDIES IN TIME-RESOLVED OPTOACOUSTIC SPECTROSCOPY By Edward G. Voigtman Jr. March 1979 Chairman: Martin T. Vala Major Department: Chemistry The approach of time-resolved optoacoustic spectroscopy is applied to the study of the radiationless decay processes of electronically excited polyatomic molecules in the gas phase. The loss-free theory of time-resolved optoacoustic spectroscopy (TROAS) is extended by the addition of the two major model -independent energy loss mechanisms (viscosity and thermal conductivity) so that acoustic mode amplitudes, mode amplitude ratios, and noise equivalent power may be calculated for TROAS experiments. Restrictions on design of experimental apparatus imposed by the theory of TROAS are considered, and the necessary properties of experimental apparatus sufficient to yield TROAS spectra are deduced. Limitations on sample selection are also considered and lead to the decision to study the following compounds: biacetyl, naphthalene, 2-chloronaphthalene, anthracene, 1-chloroanthracene, 9-bromoanthracene, iodine, nitrogen dioxide, and 2,2,4,4-tetramethyl-l ,2-cyclobutanedithione (dithione). A microcomputer-oased apparatus was designed and constructed in concordance with the theoretical constraints imposed by the theory of TROAS and numerous experiments were performed. The predicted pressure 1 x

PAGE 10

rise was partially observed as were the predicted radial and longitudinal acoustic modes, but no triplet yield information could be obtained by direct triplet excitation of the sample compounds due to the existence of irremovable, mimicking artifacts produced by the sample cell upon exposure to the highly intense (5 megawatts maximum) dye laser excitation source pulses. Additional experiments determined the nature of the four types of artifacts concomitant with direct triplet excitation TROAS. The possibility of obtaining triplet yield information by singlet excitation TROAS is examined and found to afford excellent prospects for polyatomic molecules in the gas phase, without interference from the mimicking artifacts of triplet TROAS.

PAGE 11

CHAPTER ONE INTRODUCTION The optoacoustic effect was discovered in 1880 by Bell [1] when he observed that a great variety of substances emitted audible sounds when exposed to periodically fluctuating illumination in their absorption regions. Bell's spectrophone was based on the observation that the sound intensity was proportional to the sample's absorptivity. He hoped it would find some use in absorption measurements, particularly in the infrared region. Though a number of Bell's contemporaries studied the effect [2-5], interest waned until 1938 when Veingerov [6] rediscovered the effect and employed it in the study of IR absorption by gases. Despite a few scattered uses [7-10] for sucn applications as IR gas analysis, IR detection instrumentation, and microwave detection, it was largely neglected until 1946 when Gorelik [11] noted that the spectrophone could be used to measure the vibrational relaxation times of gases. Again the technique lay dormant until Kaiser [12] and Delaney [13] in 1959 proposed a reaction kineticsbased model for the optoacoustic effect. Also in 1959, Gerlovin [14] applied the spectrophone technique in the visible and UV. It will be useful for our later discussion to note the more important contributions to the development of the optoacoustic technique over the last two decades. In 1963, White [15] observed that elastic waves are produced in solids following absorption of pulsed light. Several other investigators have made similar observations [16-17].

PAGE 12

In 1967, Hey [18], using broadband illumination and the spectrophone technique, measured the relaxation rates of dyes in solution. In 1969, Seybold et al. [19], using steady state illumination and a capillary rise technique, made similar studies. Also in 1969, Callis et al. [20], using broadband pulsed illumination and capacitance microphone volume change detection, were able to determine triplet yields of anthracene in ethanol . In 1971, Kreuzer [21] greatly improved the spectrophone technique, as it applies to gas phase absorbance measurements, by employing a laser as the illumination source. Since then, the technique has found wide use in such applications as air pollutant detection [22-24], in situ aerosol absorbance measurements [25], and the study of photochemical reaction mechanisms in the gas phase [26-27]. In 1977, Patel and Kerl [28] achieved absorpti vi ties of 10 to 10 cm for the gas NO. Significantly increased sensitivities may also be achieved by placing the spectrophone inside a laser cavity [29]. Since 1973, the gas phase optoacoustic effect has also been used in the study of N0 2 and S0 2 [30], the quenching of iodine atoms by oxygen [31], and in investigations of radiationless transitions in biacetyl [32] and the azabenzenes [33]. The quenching of the first vibrational level of the lowest excited oxygen singlet has been studied by Parker and Ritke [34-35]. Hunter and coworkers [36-37] have also used the spectrophone technique in the study of electronically excited benzene and biacetyl. These studies were significant for several reasons. First, important information was obtained which would have been difficult, if not impossible, to obtain by other techniques. Secondly, a variety of auxiliary techniques were developed and led to an

PAGE 13

increase in the usefulness of the basic spectrophone. Among these were lock-in amplifier signal detection, carrier-modulated pressure detectors, and dual excitation source techniques [38]. The source of the heat in a simple (IR source) spectrophone experiment is collisionally-induced vibrational to translational relaxation [39]. For condensed phases the mechanism is substantially more complex. In 1973, Parker [40] observed that the glass windows on his optoacoustic sample cell generated an in-phase signal when illuminated even though the gas in the cell (Ne, ^, 0o) was transparent to the excitation wavelengths used. With the aid of a thermal diffusion model of the window and adjacent gas, he established that the signal he had observed was produced by thermal diffusion from the window into an adjacent boundary gas layer. At about this time, Rosencwaig and others, using capacitance microphone detection, extended the spectrophone technique to include solids, liquids, smears, gels, and biological materials [41-43]. In 1976, Rosencwaig and Gersho [44] developed the "thermal piston" model to explain the optoacoustic effect exhibited by condensed phase materials. These authors suggested that diffusion of heat into the surrounding gas (to a depth of the thermal diffusion length) causes the boundary layer to fluctuate and thereby drive the remainder of the gas. Hence, chopped source illumination could be expected to produce fluctuations at the same frequency in the boundary layer. Comparison of their theory with experimental results suggested that this was the primary source of sounds produced by condensed phases. In 1978, McDonald and Wetsel [45] extended the theory (the composite piston model) by including the effect of the expansion and contraction of the condensed phase. They proposed that the boundary layer was mechanically

PAGE 14

driven, thereby generating sound. As noted previously, however, elastic waves may also be produced in condensed phases by the absorption of intensity-modulated light. These elastic waves may be detected directly [46] using piezoelectric transducers. Thus, three mechanisms have been proposed to account for the production of sounds by condensed phases in "the" optoacoustic effect: the thermal piston model, the mechanical piston model, and light-generated elastic waves. With a few notable exceptions (vide infra), recent work on the optoacoustic effect applied to chemical systems has dealt with the spectral distribution of the released heat and not with the kinetics of the release process. Work in this laboratory was begun a number of years ago to develop a theory of time-resolved optoacoustic spectroscopy which was capable of describing the kinetics of radiationless decay processes (heat release) after electronic excitation in polyatomic molecules. The theory was developed by Wrobel [47]. It was motivated by the need for detailed information concerning nonradiative deactivation pathways in electronically excited polyatomic molecules under approximately collision-free conditions. In general, phosphorimetry and fluorimetry are incapable of providing all the data necessary to obtain accurate triplet yields, without which it is impossible to test the various theories of radiationless processes in polyatomic molecules. Timeresolved optoacoustic spectroscopy (TRCAS) is intended to complement these other techniques, particularly for species with low radiative yields, and thereby lessen the need for the motley assortment of ingenious methods which have been devised to furnish triplet yields [48-56].

PAGE 15

To understand the need for a complementary technique, it is helpful to consider the various pathways of deactivation available to an electronically excited molecule. Irradiation of a molecule results in its excitation from its ground state (S Q ) to a vibrational level of some excited singlet state, S . With few exceptions, rapid radiationless relaxation ensues populating the lowest vibrational level (v = 0) of the lowest excited singlet, sj (the superscript denotes vibrational level and the subscript electronic state). Three deactivation processes may now occur which depopulate S, (see Figure 1): fluorescence with a rate k f , radiationless decay with a rate k s , and intersystem crossing to the triplet with a rate k,^. The triplet vibrational level so populated may relax directly to S Q via radiationless decay with a rate kyV, or it may relax via rapid radiationless decay to T 1 ? with a rate k T v T 0. Intern ] ' system crossing back to S-j is negligible. Finally, T^ may relax by phosphorescence with a rate k or by radiationless decay with a rate k T . P I Since decay directly from the vibrational manifold of T-, to S Q is not usually possible (Kasha's rule), k T v may be neglected. In addition, 'l vibrational relaxation in J } is usually fast relative to k y and k . The five remaining rate constants (k f , k s , k ISC , k , k T ) may be used to define the fluorescence quantum yield $-, the phosphorescence quantum yield $ and the triplet yield $. as follows: 9f = k f /k* where k* = k f + k s + k ISC is the observed fluorescence rate, *p = V^t^p wher s k* k + k T is the observed phosphorescence rate, and

PAGE 16

Figure 1. Electronic states and rate constants typically of importance in the photophysics of polyatomic molecules.

PAGE 18


PAGE 19

associated with a selected multielectronic state model ($ ,t! or S 0' T r S l '' The fundamental result of TROAS (valid for the simple Sq.T^ model) is the following relation between the pressure fluctuations expected to occur in the gas phase and the nonradiative rate constants: P 0,0 = (*-l)q , n?v ^ T 0k T V jT 0hv T (1 e" ^ ') / + (k T v T 0h v T v T / (k T v T + k T v) 'r'i 'r'i 'r'i t i -(k T v T + k T v)t k T 0hv 0/ (k T v + k T v T 0-k*))(l -e 'r'i T l )] '1 'i h 't'i p where y is the heat capacity ratio, niv is the cell -averaged initial number density of molecules excited into T^, q n n is an expansion coi U ,U efficient dependent on the model -specific heat source term, and P' is the amplitude of the pressure rise expected to occur. Normally, the vibrational relaxation rate k,v T is much larqer T T T 1 than the triplet deactivation rate k* The above equation then simplifies to _n IC _ (k T v T + k T v)t P o,o = (y 1)q o,oV l> T v h V,T (1 ' e ] ] ] ) IC ISC ~^*^ + * v *o h v (1 -e p )] 'l 'l 'l

PAGE 20

10 ISC . , . ic where * v is the intersystem crossing yield k T /k*, and $ v is the T l ' p T l internal conversion yield k T v T 0/(k T v ,0 + k T v). Thus, the expected 'r'i t t'i t i pressure rise is the sum of two exponentially rising components with different amplitudes. Note, however, that k T v T is much larger than 1 1 ' ' 1 k*. Hence, the pressure rise will appear to be a step rise with an exponential rise building upon it. This situation is depicted in Figure 2 [47]. If the amplitude of the slow rise is denoted p' and that of the fast rise is denoted pi, then from Figure 2 the following equation is obtained: & = (hv v T 0/hv 0) p'/p' . T l 'r'l 'l s r Since k* is obtained experimentally from the time constant of the slow rise, the radiationless rate constant k^ may be obtained directly and independently of phosphorimetry and fluorimetry experiments. Unfortunately, Wrobel [47] was not able to experimentally determine the validity of this result, so additional experiments have proven necessary. Figure 3 illustrates schematically the experimental setup of a TROAS apparatus. A light pulse from a flash! amp-pumped tunable dye laser passes through the cylindrical sample cell and strikes the laser energy detector. Absorbed energy converted to heat via radiationless decay processes heats the sample gas mixture (molecular vapor plus carrier gas) producing a pressure jump and acoustic oscillations. These pressure variations are converted to electrical signals by a sensitive broadband pressure transducer. The electrical signals are amplified and acquired by a microcomputer-controlled transient waveform recorder. The microcomputer also synchronizes data acquisition by controlling

PAGE 21

o

PAGE 22

12 (TO b A 1 U( L -A.) /°'? d

PAGE 24

14 CU

PAGE 25

15 laser firing, light pulse peak-detector reset, etc. It also performs data processing tasks (such as Fast Fourier Transforms) and outputs data to a chart recorder as necessary. The oscilloscope displays acquired data prior to processing. This experimental arrangement differs markedly from that of conventional optoacoustic spectroscopy (OAS) as shown in Figure 4. The excitation light source is usually a monochromator and Xe lamp combination with light modulation provided by a rotating slotted disk. Alternatively, a tunable continuous wave (CW) dye laser and either an acousto-optic modulator or chopping wheel provides the requisite modulated source intensity. In any event, the spectrum obtained must be corrected for variations of source intensity as a function of source wavelength. The second major difference between OAS and TROAS is the use of a lock-in amplifier to amplify and detect the pressure transducer signal (after the preamplifier stage, of course). The resulting extremely narrow equivalent noise bandwidth allows the use of OAS in the detection of signals in extraordinarily dilute samples [28]. An especially good collection of research papers concerning all relevant aspects of OAS has recently been published [58]. In OAS, accurate waveform information is not necessary; in fact, a lock-in amplifier provides only amplitude and phase shift information in the usual case. This is adequate for sample detection purposes but not well suited for the study of such processes as triplet deactivation pathways. Further details concerning this use of OAS may be found in [32]. The theory of TROAS was developed by Wrobel [47] and, therefore, will not be repeated here, although Chapter Two concerns certain

PAGE 27

17 I 4-^ -rCJ r— O CL r*

PAGE 28

13 extensions and corrections to the loss-free theory. Chapter Three describes the experimental apparatus used in this research. In Chapter Four, the sample selection criteria developed in Chapter Two are used to select promising polyatomic molecules for study. Chapter Five details the numerous experimental arrangements and modifications used. Chapter Six discusses the obtained results, and Chapter Seven contains a discussion of singlet TROAS and its advantages.

PAGE 29

CHAPTER TWO THEORETICAL CONSIDERATIONS Introduction and Direction Several papers have appeared in which pulsed light sources were combined with the optoacoustic effect in order to study various vibrational and electronic state properties of polyatomic molecules. Although the pulsed optoacoustic effect is certainly shown in the works of Parker [34], Aoki and Katayama [59], and Grabiner et al . [60], the interpretations provided are either of limited scope or unnecessarily complex. The theory of TROAS is intended to bring together these diverse observations in a relatively simple and coherent manner. Unfortunately, the particular solutions given in Wrobel [47] for radiationless decay from the lowest singlet and triplet states of polyatomic molecules are incomplete due to neglect of all energy loss mechanisms, exclusive of those incorporated in the model-specific source term, in the sample gas mixture and sample cell. The most important of these dissipative processes are viscosity and thermal conductivity losses [61]. By inclusion of these two loss mechanisms, it is possible to calculate realistic acoustic mode amplitudes, mode amplitude ratios, and an important figure of merit, the noise equivalent power (NEP). Restrictions imposed by the extended theory are then applied to sample cell design, selection of appropriate pressure transducers, and excitation light source requirements. 1'

PAGE 30

20 The Extended Solution of the Heat-Flow Model Consider a molecular vapor at equilibrium in a cylindrical sample cell of length L and radius A. The equilibrium density, pressure, and temperature are p Q , p and T Q . Radiationless relaxation processes following excitation of the vapor in an absorption region cause fluctuations in p, p, T and cause the vapor to assume an acoustic velocity v(r,t). The conservation equations and equation of state constrain the behavior of the vapor P = -ft p T equation of state, dp * rft -p v'v conservation of mass, dv *+ p^ = -v • p conservation of momentum, p -tz = -P.,t— " V *( I + J R ) conservation of energy, where R is the ideal gas constant, M is the molecular weight of the sample, u is the internal energy per gram of sample, p* is the pressure tensor, I is the nonradiative energy flux vector, T D is the radiative energy flux vector, the convective derivative is given by d _ 3 ,dt = 3t + (V and the Einstein summation convention is used. This solvable system of equations may be simplified by noting that the fluctuations induced by released heat are small relative to the

PAGE 31

21 equilibrium values. This acoustic approximation "linearizes" the system of equations and hence avoids the possibility of shock waves [62], Thus, p = p' + p Q , p = p' + p Q , T = T' + T Q , and v is small. Products of primed quantities are neglected. If viscosity and thermal conductivity are also neglected, the system of equations simplifies to yield a single separable equation [47,63] 2 f-c 2 7 2 p' . ( T -l)jtt (2.,) at where the heat source term H is given by H ~= " V ' r Rp 0^r (2-2) and u e is the energy per gram stored in electronically excited molecules. The heat source term is given explicitly only when the physical model (e.g., a simple two-state model of triplet decay) is selected. With appropriate heat source terms, equation (2.1) underlies both OAS and TROAS. As stated previously, the loss-free solutions of (2.1) are inadequate for the calculation of mode amplitudes, etc. Appendix One contains a brief summary of Kreuzer's approach [63] which includes the two major energy loss mechanisms—viscosity and thermal conductivity. These results will be used as needed to complete the treatment of TROAS. The fundamental effect of including viscosity and thermal conductivity is the introduction of a finite quality factor Q. for each acoustic resonance uj.. The use of a Fourier Transform technique to effect a solution of (2.1) is natural for OAS since the frequency parameter u may represent the frequency at which the illumination source is

PAGE 32

22 modulated. It may be, for instance, the number of pulses per second (times two pi) delivered by a simple chopping wheel interposed between illumination source and sample cell. If the analysis of Kreuzer [63] is to be carried over to TROAS, it must be possible to identify a suitable to. It is assumed in the theory of TROAS that the illumination source is an energy pulse of negligible duration. This situation is approximately met when the light pulse duration is much less than the smallest relevant time constant in the physical model. As a practical matter, it is satisfactory to have the light pulse duration, t , substantially less than the nonradiative lifetime of the spin-forbidden processes being examined. Such spin-forbidden processes will almost always be involved due to intersystem crossing even if only an excited singlet state were initially excited. In the event no "slow" (e.g., forbidden) states lie below the desired "fast" (e.g., allowed) state, the system is probably better treated by conventional spectroscopic methods. The time constant, tj, of the thermal damping subsequent to pulse excitation and relaxation will, for reasonable physical models, always be much greater than t . P The high modulation frequency case for OAS is given by the condition [63] w t T » 1 . (2.3) For the simple chopping wheel type of intensity modulation, the light pulse duration is a)" 1 /2

PAGE 33

23 assuming a 50% duty cycle for the chopping wheel. Hence, u t. » 1 implies t T >> u" > co" /2 so the light pulse duration is short relative to the thermal damping time. Precisely this situation occurs in TROAS. A reasonable estimate for oj in TROAS is, therefore, given by t . More accurately, oj will p J be assumed to be the spectral bandwidth required to encompass most of the spectral energy in the Fourier Transform of the light pulse [38]. For example, assume the light pulse to be a rectangular pulse of duration t centered at t = 0. Then the Fourier Transform is proportional to [64] t sine (u t /2ir) . The significant portion of the spectrum is in the range ju/2:r |
PAGE 34

relaxation via TROAS. Hence, in a more realistic fashion, the relation between H and I should be [63] H(r,w) = a(u) I(r,u) . (2.5) The only immediate effect of this for TROAS is that the heat produced lags the light intensity variation. This will not be considered further. Combining the above relation with the expression for corrected acoustic mode amplitudes (A1.20) from Appendix One yields A ( u ) = zMlzU _J 1Z (2.6) J v c o)^ [i -u/ Uj r-tu. q.)] The integral determines the coupling between the normal acoustic modes and the beam intensity since I(F,oj) may be expanded in the same orthonormal set of functions used to express the pressure p(F,w). Hence, the orthogonality equation (A1.7) from Appendix One implies that a normal mode can be excited if and only if the intensity has a corresponding nonvanishing component. Intuitively, one expects longitudinal modes to be excited if significant attenuation of the beam occurs in passage through the cell. In a uniformly illuminated cell (e.g., intracavity in a dye laser) longitudinal modes should not be excited. It is also reasonable to suppose that radial modes will be least strongly pumped in the uniform illumination case. The former expectation is verified by calculator simulations by Wrobel [47] while Kamm [61] has explicitly given the relationship between excitation beam width (a) and acoustic mode excitation:

PAGE 35

25 •iodN, y-1 P e " y j p(r,t) z(=^) r-^r— — : v-? J to. (1 -co /to-ioj/co. Q-) C J„ (a, *J J J J vJ I j J J (a lfj r/A) e ^ (2.7; where »j . (|) 2 (Zc^f 2 (2.8) For the lowest frequency radial mode ( j = 1 ) excited at to-,, the above equation simplifies to (y-UQ, p e " y i W = -x1 tit; — J < a i,i r/A) (2 9) 1 C J Q (a 1}1 ) Note that the radial dependence of the amplitude of the lowest radial mode is relatively weak, residing solely in p.. In TROAS, the pressure rise mode (j = 0) contains the significant information. The other acoustic modes obscure the pressure rise and waste energy that would preferably be pumped solely into the j =0 mode. Examination of (2.9) above indicates that the radial modes will be • " M l minimally pumped when e is minimized. This occurs at a = A, i.e., when the excitation beam width equals the sample cell radius. Although other dissipative mechanisms external to the source term, such as wave reflection, wave scattering, microphone conversion, and volumetric losses, are present [61], they are of minor importance in comparison with viscosity and thermal conductivity effects. Consequently, we are now in position to consider the characteristics required of each component of a TROAS experiment. The modified TROAS theory restricts the selection of various components of the apparatus and also determines to a large extent just what substances may be studied.

PAGE 36

26 Excitation Source Selection The "ideal" illumination source for TROAS should have the following properties: (1) high constant energy per output pulse (preferably greater than 100 mJ), (2) short, stable pulse duration t (such that t p is much less than other significant system time constants), (3) high pulse repetition rate (for possible signal averaging purposes) , (4) low beam divergence (for efficient energy coupling to the cylindrical sample cell), (5) moderately large beam width (0.5-3.0 cm typically), (6) nearly monochromatic light output (no more than several Angstroms spectral bandwidth to enable excitation of a single state) , (7) tunable operation in the visible and near ultraviolet (to excite electronic states), (8) ease of alignment, (9) reliable operation over more than 10 high energy pulses (higher if signal averaging is used). These conditions are only met by tunable pulsed dye lasers. A compromise between conditions 1 and 3 must be made in the selection of a suitable dye laser system since the optical pumping sources have limited duty cycles, pumping capacities, or heat dissipation capabilities. Since the desired pressure rise mode is proportional to pulse energy, high pulse energy is especially desirable; a coaxial flashlamp-pumped dye laser will

PAGE 37

27 provide energies of several Joules per pulse. Unfortunately, the repetition rate for this type of laser is approximately 0.1 Hz so signal averaging is laborious. In fact, the rather limited tube life (something above TCP shots), tube expense (approximately $500.), tube failure mode (rapid disintegration), and photo-decomposition of the dye solutions after a few score shots effectively rule out signal averaging. Linear flashlamp-pumped dye lasers have repetition rates up to 30 Hz and pulse energies of 100 mJ maximum. Nitrogen laser-pumped dye lasers have similar repetition rates but significantly less energetic output pulses. They are not recommended for TR0AS in the general case. See Table 1 for a comparison of the above three dye laser types and the monochromator and Xe lamp combination. Sample Cell Construction The sample cell should be an inert, high heat capacity, isothermal cylinder having transparent end windows and several inlets for pressure transducer(s), pump-out ports, sample admission, etc. A cylindrical geometry is chosen to provide maximum coupling between excitation beam and sample vapor. As previously shown, the desired pressure rise mode (j=0) is maximally pumped when the excitation beam width equals the cell radius. The inlets should be as small as possible to avoid excessive wave reflection losses. If one pressure transducer is used, it is probably best positioned midway between the end windows since the solution of the equations of TR0AS is greatly facilitated for such a choice [47] and beats in longitudinal modes are eliminated. The end windows present a problem in that they should have high thermal conductivity and yet be at the same temperature as the body of the cell. The

PAGE 38

t— T3 O •
PAGE 39

29 high thermal conductivity is necessary so that the windows can dissipate any heat produced by adsorbed substances. Otherwise, this heat will pump spurious longitudinal modes [65] despite the windows being at nodes of these modes [66]. If the windows cool significantly, condensation may occur and the previous problem is greatly exacerbated. Sapphire is an excellent choice of window material though expensive. Double windows of Suprasil or Pyrex are satisfactory if light intensity can be compromised approximately 8%. The double windows must have an adequate dead air space between them, and they must be parallel to avoid scattering light onto the cell walls and pressure transducer. In the absence of thermal conduction, the expected pressure rise term is given by Wrobel [47]: P6 f o|t-H= fr-1) V k p) _1 < ] e " aL )E/V c (2.10) where ky(k*)~ is the nonradiative yield, a is the sample absorbance, L is the cell length, E is the excitation source energy per pulse, and V c is the cell volume. As a practical matter, a will be small since all samples should be run at as low a pressure as possible (consistent with the continuum model of sound conduction) to minimize the quenching effect of collisions. For a small, 1 e is approximately equal to al_. Thus, equation (2.10) simplifies to P 0,o|t~> ^L/uA 2 L = aAA 2 (2.11) where A is the cell radius. Thus, the pressure rise is independent of cell length but inversely proportional to A 2 . Nevertheless, L should be as short as possible to lessen scattered light problems and to allow

PAGE 40

30 placement of the cell in the sample compartment of a UV-visible spectrophotometer for in situ gas phase absorbance measurements. Cell radius is much more important for a variety of reasons. Pres-? sure rise amplitude is proportional to A so the cell radius should be minimized. The lower limit on radius is the excitation beam width. The NEP is also proportional to A so again the radius should be minimized. Three time constants also depend on cell radius; these are designated tj, t d , and t w and are described below. The time constant for the thermal reequilibration following release of heat in the cell is approximately given by t T = A 2 /k (2.12) where k is the thermal diffusivity. This upper time limit is not critical and usually is of the order of 1 to 10 seconds. The acoustic delay time t d is the time required for sound waves to reach the pressure transducer from the nearest illuminated sample region. It is a lower time limit given by t d = (A-a)/c . (2.13) Note that t. = for a = A. The "mean free path time" t w is the time required for colli si onal deactivation at the cell wall (ignoring deactivation at the windows). This is given approximately by t w s A 2 /Sc (2.14) where £ is the mean free path in the sample gas mixture. This is an important upper time limit.

PAGE 41

31 The time limits defined by these time constants may be modified in several ways. First, since both t d and t w are inversely proportional to the speed of sound in the sample gas, it is possible to add a transparent carrier gas such as He, Ar, SFg or a freon to the sample to change the speed of sound in the mixture. The speed of sound in an ideal gas mixture is c = (yRT/M) 1/2 (2.15) where M is the average molecular weight of the sample mixture. Thus, by employing a heavy carrier gas (e.g. SF 5 ) the speed of sound can be decreased and the upper limit due to the "mean free path time," t increased. Unfortunately, at the same time the lower limit due to the acoustic time delay, t d , is also increased. Thus, it is necessary to choose the carrier gas with care to select the upper and lower time limits which are most convenient for study of the molecule under consideration. The second possibility is to use the fact that both K and g are inversely proportional to pressure [47]. Note, however, that the pressure rise amplitude is proportional to the partial pressure of the absorbing species and that collisional deactivation processes such as triplet-triplet annihilation become more important at higher pressures. The use of an inert monatomic carrier gas also increases the signal amplitude through the heat capacity ratio term. A sample calculation is given in Chapter Four.

PAGE 42

32 Sample Selection The pressure rise which occurs after radiationless decay is given approximately by [47] P' 2 (Y-1) k T (k*) _1 E I p Q /^A 2 RT (2.16) where E is the pulse energy, ^(k*) -1 is the nonradiative yield, e is the molar absorption coefficient, and p Q is the sample partial pressure. It is assumed that a is small so that the following relation is valid: 1 e " a = ctL = I p L/RT . (2.17) Equation (2.16) is the principal constraint on sample selection. A well -chosen sample should have a high vapor pressure at room temperature, a high singlet-triplet absorption coefficient, and a low radiative yield. The laser output pulse energy must be high in the triplet absorption region, the cell radius relatively small, and an inert monatomic carrier gas should be used to increase the y-1 term and to insure that the acoustic approximation is valid. In addition, the sample should not be readily quenched by collisions (especially with the carrier gas), should not readily decompose, and should not undergo any sort of photochemistry. It should also have triplet and singlet manifolds well separated relative to the spectral bandwidth of the excitation light source. The triplet lifetime must also fit in the time window given by t d < t < minimum of (t-ptj . (2.18) Another limitation is imposed by the frequency response of the pressure transducer and associated electronics. An upper time limit

PAGE 43

33 due to low frequency roll-off of the pressure detection system response (typically 6 db per octave) may limit observation times to substantially less than the thermal damping time tj unless care is taken in the design of the pressure detection system. Even worse is the lower time limit given approximately by t £ 5 (2f h ) _1 (2.19) where f^ is the -3 db high frequency cut-point. High frequency rolloff for a capacitance microphone is typically 12 db per octave [67]. Since the desired pressure rise is actually a sum of two disparate exponential rises followed by the thermal reequilibration exponential decay, it is necessary to have high frequency response. Otherwise, the pressure detection system will be rise-time limited and thus unable to distinguish "fast" heat from "slow." Thus, the triplet lifetime is constrained by another time window; it must, therefore, fit into the intersection of the allowed time frames. Pressure Transducer Selection The "ideal" pressure transducer for TROAS should have the following properties: (1) high acoustic sensitivity (preferably above 1 mV/Pa), (2) high, flat frequency response including dc (to facilitate static pressure calibration), (3) low acoustic impedance (concordant with gas phase acoustics), (4) high immunity to elastic waves in the mounting substrate (e.g., vibrations and light-induced artifacts in the substrate) ,

PAGE 44

34 (5) immunity to stray light, (6) linear response to absolute pressure changes, (7) low output impedance (for high noise immunity and ease of amplification), (8) long-term output stability, (9) wide operating temperature range and low temperature drift, (10) small, rugged, chemically inert construction (for ease of mounting) , (11) low output drift in reactive environments, (12) ability to withstand vacuum pump-down, (13) relatively low cost. No single pressure transducer available meets all of these conditions or even the first dozen of them. In the previous TROAS study [47], the pressure transducer was a Pitran™ (pressure sensitive transistor) developed and marketed by Stolab, Inc. Several different pressure transducer types were used in the present study: capacitance microphones, piezoelectric disks, and piezoelectric cylinders. These types of transducers are compared with respect to the above properties in Table 2. A more detailed comparison of capacitance microphones is presented in Table 3. Properties not listed are similar for the various microphone types. Although capacitance microphones are well suited to gas phase acoustics applications in general, they have several disadvantages relative to TROAS applications. First, it is usually necessary to exclude oxygen from the sample cell to avoid triplet quenching. This is achieved by pump-down of the cell and refill with sample and carrier

PAGE 45

35 Q. O ra -iO ,— <— X

PAGE 46

36 Q)

PAGE 47

37 gas. Wei 1 -designed capacitance microphones, however, have a trapped gas volume beneath the metallized diaphragm and a deliberate slow leak to the external environment. For optimum sensitivity the restoring force on the diaphragm due to the compliance of the trapped volume should equal that due to tension in the diaphragm [67]. The second disadvantage is operating temperature range. Although externally polarized microphones may tolerate operating temperatures of several hundred degrees C, electret elements depolarize at elevated temperatures [68]. Piezoelectric transducers are not well suited to gas phase acoustics because of their severe acoustic impedance mismatch with gases. The acoustic impedance Z a of a medium is the product of the sound velocity and the density of the medium. Typical values for solids are in the range 10 to 10 kg m sec ; for gases, about 5 x 10 2 kg m" 2 sec _1 [69]. As noted by F transmission coefficient a 4. is given by 2 -? -1 5 x 10 kg m c sec [69]. As noted by Farrow et al. [69], the acoustic a t = < 4Z ai Z a2>^(Z ai+ Z a2 ) 2 ] (2-20) where Z . is the acoustic impedance of medium i with sound waves a I normally incident at the media boundary. The transmission coefficient (a.) is the ratio of transmitted wave amplitude to reflected wave amplitude; gas-solid interfaces have typical values of 3.xl0 while solid-solid interfaces may be close to unity (0.9) [69]. Thus, piezoelectric transducers are susceptible to vibrations and shocks in the mounting substrate and are relatively insensitive to gas phase sound waves.

PAGE 48

38 Pressure Transducer Electronics The output signal of the pressure transducer must be amplified and possibly conditioned before being further processed. The two primary pressure transducer choices are capacitance microphones and piezoelectric ceramics both of which have extremely high output impedances (greater than 10 ohms). If a voltage preamplifier is used, it must be located as close to the transducer as possible to avoid power line noise and cable microphonics. Since the sample cell (and transducer) may need to be heated to produce an adequate sample partial pressure, the preamplifier must have good temperature stability, wide operating temperature range, and low thermal noise. Methods which allow use of a connecting cable between transducer and preamplifier are generally deficient in other aspects. Charge preamplifiers are usually noisier (ceteris paribus) than voltage preamplifiers [70]. Parametric amplification of audio frequencies is too complex to justify the minimal improvement in noise performance expected. Carrier current or carrier voltage modulation schemes such as AM, FM, and pulse frequency modulation are all usable if properly designed and well implemented [71-72]. These schemes are usually, however, difficult to implement. Voltage preampl ification is, therefore, generally the simplest and most practical method to use. The actual preamplifier circuit configuration is less important than the "front end." It is crucial that the front end provide high gain and add as little noise as possible [64]. The field effect transistor (FET) is preferred for voltage amplifiers driven from source impedances above approximately 10 ohms due to the extremely low input

PAGE 49

39 current noise typical of FETs (10" 13 -1C~ 16 A/Hz 1/2 ). The so-called 1/f noise of junction FETs (JFETs) is substantially lower than that of MOSFETs for ultrasonic frequencies below approximately 1 MHz [73]. Discrete low-noise JFETs are also quieter than JFETs in IC op amps [74], The dominant noise parameter for a JFET is input noise current I with input noise voltage V of secondary importance. The optimum source impedance is V n /I . Noise performance will be degraded if nonreactive components are used to match source impedance to the above optimum source impedance. The dominant noise sources in a JFET are channel generated thermal noise, gate leakage current shot noise, and generation-recombination surface effect noise [75]. The dominant noise source for high frequency operation is the thermal noise [75] V n 2 = 4kTB/g fs (2.21) where B is bandwidth and g fs is forward transconductance. Low noise JFETs, therefore, have low leakage and high gain. The most convenient circuit configuration is probably the simple source follower. This circuit provides extremely high selectable input impedance, low output impedance, and high, flat frequency response. If the input is capacitively coupled, a rather long time constant results which may, in some circumstances, limit the upper time limit of observation of the transducer output. This RC time constant is particularly important for piezoelectric transducers [62], The effect of the preamplifier input impedance is interesting. Total noise voltage decreases with increasing input resistance if the

PAGE 50

40 input resistance is above approximately 5.xl0 8 ohms and if the JFET input noise current I is below about 2.xl0~ 4 A/Hz 1/2 . This result is based on microcomputer calculations based on a composite noise source model. Details are found in Appendix Three.

PAGE 51

CHAPTER THREE THE EXPERIMENTAL APPARATUS The apparatus used in the present research may be divided into five subsystems: (1) a pulsed, tunable dye laser system with photodetector, (2) sample cells and vacuum system, (3) pressure transducers and associated electronics, (4) microcomputer interface electronics and transient waveform recorder, (5) a microcomputer system with adequate software. The subsystems have been listed in decreasing order of importance. Previous TROAS studies [47] were unsuccessful primarily due to shortcomings in the dye laser system although inadequacies in the remaining subsystems were sufficient to compel a complete redesign of the apparatus. The subsystems adopted for this study are described below in detail . The Dye Laser System The dye laser system chosen was a Candela Corporation ED 525-U flashlamp driver unit and CL 625 coaxial flashlamp. A Phase-R Corporation high voltage supply (25 KV maximum) provided the 18-22 KV necessary to fire the flashlamp and cause lasing in the dye. The laser mirror mount and grating mount were also Phase-R products. The dye solutions were turbulently pumped by a Micropump (Model 10-84-316-852) 41

PAGE 52

42 high flow centrifugal pump and filtered and bubble trapped by a Pall filter assembly (resin-free polypropylene 0.45 urn filter cartridge and polypropylene housing). Tuning was effected using an inexpensive Edmund Scientific diffraction grating (5000 A blaze, 600 1/mm) and a 40% reflectivity (420 nm 550 nm) front reflector from Candela. The dye reservoir was a machined stainless steel tank not equipped with heat exchange coils. The windows on the coaxial flashlamp were anti reflection coated in the visible on the outer faces only. The laser cavity length was approximately 2 meters. The system will lase broadband o (70-100 A) and not tune if the cavity is shortened by a factor of three. The dye solutions were prepared from laser grade dyes supplied by the Exciton Corporation. Concentrations of 10 -10 M were used with 200 proof ethanol or Spectro-grade methanol as the solvent. Solution volume was one liter. Dyes used successfully were, in Exciton's nomenclature, Rhodamines 6G, B, and 110; Coumarins 540, 480, 450, 440; Fluorol 555; Kiton Red S; and Laser Dye 473. The spark gap trigger in the flashlamp driver unit was pressurized with He or dry H ? at 8-12 psi. The laser driver unit could be triggered either under manual pushbutton control or via an optocoupler circuit under microcomputer control. The circuit used is shown in Figure 5. A commercial optocoupler was not used because of their relatively low (less than 7500 V) breakdown voltages. No significant EMI or RFI was produced by a laser firing sequence. Alignment of the optical cavity was achieved with the aid of an adjustable aperature and a He-Ne alignment laser. The aperature was used to ccaxially align the dye laser optics, dye cavity, and sample cell. Three passes of the He-Ne 6328. A beam through the system are

PAGE 53

-c a i—
PAGE 54

44

PAGE 55

45 sufficient to produce excellent alignment. The He-Ne laser is approximately 5 meters in front of the dye laser cavity. Fine adjustment, if necessary, is accomplished by use of the micrometers on the optics mounts. Performance characteristics of the dye laser system are shown in Table 4. Several comments on this dye laser system are in order. First, the system would lase under very poor conditions; the cavity may be a cheap mirror and a Suprasil window. An energy meter must be used to optimize system performance. Second, the dye solution heats up when turbulently pumped and pulse energy falls about 20%; the dye reservoir should have heat exchanging coils for optimum performance. Third, the power density is high enough (almost 1 MW/cm 2 ) to "burn" the diffraction grating so inexpensive 600 1/mm grating replicas are recommended. Fourth, fresh dye solutions should be used whenever possible because the first few score of shots are the most energetic. When the system is almost perfectly aligned and fresh dye is used, a sharp, snapping sound accompanies the laser pulse. The grating will most likely be damaged when this excellent lasing occurs. Finally, if the laser did not lase when aligned, it was realigned once or twice. This required no more than ten minutes. If it still did not lase, the dye solution was changed. Flushing the system with 190 proof ethanol followed by one flush with 200 proof ethanol eliminates dye contamination problems. The laser power was monitored with a silicon PIN photodiode (MRD501) with a current limited, regulated 22.5 volt bias supply and 3.01 K ohm photocurrent conversion resistor. With the aid of machined Teflon attenuators, the detector system was used to detect pulse intensities as high as 5. MW. In addition, a Gen-Tec ED-200 joulemeter was used to obtain accurate pulse energies for comparison purposes.

PAGE 56

46 Table 4. Performance characteristics of the Candela dye laser system. Property Achieved performance Pulse energy 2.5 J Pulse duration S.xlO sec Pulse repetition rate 0.1 Hz Spectral bandwidth 0.2 nm Tuning range 410. nm to 640. nm

PAGE 57

47 Sample Cells and the Vacuum System Six different sample cells were constructed for use in the present TROAS studies. In addition, the sample cell used by Wrobel [47] was altered and used in the initial experiments. The salient features of these cells are given in Table 5. Initial attempts to comply with the constraints imposed by the theory of TROAS were loosened in later cell designs when consistent results could not be obtained. Specific information concerning each cell is given below. Cell #1 was the aluminum cell used by Wrobel [47]. The cell was modified to accept a capacitance microphone and a copper tube was cold soldered inside (with gallium) to increase corrosion resistance. It was not vacuum-tight after the modifications and was inconvenient to heat. Cell 42 was the primary sample cell used. It was machined from a 9.5" long, 2.5" diameter rod of stainless steel which was drilled, reamed, and ballized to 1.002" inner diameter. The ends were machined for "0"-ring seals and 2" diameter, 1/3" thick Suprasil windows. Pump-out ports 1" from each end were Cajun Ultratorr connectors stainless steel -soldered into the cell. Two pressure transducer ports were milled into the cell to allow use of one while improvements were tested on the other port. The cell was temperature controlled to ±0.1 C by an Oven Industries, Incorporated, proportional temperature controller driving a doubly insulated heating tape. Feedback output was provided by a thermistor mounted in a reamed hole 0.050" from the inner call wall. Cell temperatures were measured with a thermometer when necessary.

PAGE 58

48

PAGE 59

49 The cell was rigidly mounted to a 2 ' by 3' by 1" thick drilled and tapped (1/4-20, 2" centers) aluminum plate which served as optical bench, mounting platform, and system ground. The cell may be isolated from ground with Teflon straps (to avoid ground loops). See Figure 6. Cell #3 was a Pyrex and Teflon 5 mm valve with one "arm" sealed. See Figure 7. A shielded piezoelectric transducer on the outside of the glass was intended to pick up elastic waves in the glass produced by an illuminated internal sample melted or sublimed onto the inner wall of the glass tube. Cell #4 was a Florence flask with a Pyrex and Teflon 5 mm valve and tungsten feed-through electrodes in a separate Pyrex apparatus. This cell was used for solutions and gas phase measurements with a suspended piezoelectric transducer. See Figure 8. Cell #5 was a Cajun Ultratorr 3/8" tee with 3/8" diameter, 1/16" thick Suprasil windows in the arms of the tee and the pressure transducer in the stem. Minimal machining was required to enable the tee to accept either a capacitance microphone or a piezoelectric disk. A spring-loaded electrode was used when the piezoelectric transducer was in place and the signal was brought out through a vacuum-tight BNC bulkhead feed-through connector soldered to a brass 3/8" nut from a similar Cajun 3/8" tee. The cell was vacuum tight though without need to be evacuated. Cell #6 was a Pyrex test tube with a Teflon tape-wrapped capacitance microphone snugly inserted into the end. The cell was illuminated perpendicular to the long axis and heated uniformly with a heat gun. Cell #7 was a Pyrex cylinder section with 1/4" sidearm and 1/2" drilled hole for mounting the capacitance microphone. Ordinary window

PAGE 61

51 3 O

PAGE 62

Figure 7. TROAS sample cell number 3.

PAGE 63

53 Vacuum System 5 mm Pyrex Valve Sample Area CM Teflon Stopcock With EthylenePropylene "0"-rings Molded Aluminum Enclosure =6 Piezoelectric Transducer

PAGE 64

Figure 8. TROAS sample cell number 4.

PAGE 65

55 to Differential Amplifier Tungsten Electrodes Spring Loaded Nickel Silver Contacts to Teflon 5 mm Valve and Vacuum System

PAGE 66

56 glass plates were carefully epoxied onto the ends of the cylinder to produce a highly transparent cell. See Figure 9. The vacuum system used must be capable of removing oxygen from the sample cell to avoid the possibility of triplet quenching. A liquid nitrogen-trapped rotary oil pump system is sufficient to attain 10" 4 torr which is adequate to exclude oxygen from liquid samples by repeated freeze-pump-thaw cycles. Solids may be deoxygenated by solution in a suitable solvent and subsequent pump-down. The system pressure was measured with a Wallace & Tiernan gauge (0-800 torr in units of 1 torr). The vacuum system was tested for leaks with a Veeco MA-2 mass spectrometer helium leak detector. Pressures to 10" 4 torr were routinely obtained in the system. Pressure Transducers and their Associated Electronics Several hundred man-hours were spent in the design and construction of improved circuitry and transducer mounting techniques. Among the techniques tried without success were laboratory-made, externally polarized capacitance microphones and FM carrier current modulation by microphone capacitance changes. The Pitran pressure sensitive transducer was rejected due to moderate sensitivity (380 yV/mtorr), high temperature drift (400 mV/C), differential pressure mode operation, high acoustic impedance, extremely high susceptibility to damage (especially in mounting), and sensitivity to mounting substrate vibrations. The most important parameters to be considered in selection of pressure transducers for TROAS are sensitivity, acoustic impedance, and flat, broadband frequency response. Unfortunately, these parameters are usually dependent on each other. Consider the capacitance

PAGE 67

Figure 9. TROAS sample cell number 7.

PAGE 68

58 Pyrex Cylinder and Pump-out Port Microphone Port Window Glass Plates

PAGE 69

59 microphone. The resonant frequency of a standard unpinned disk microphone (approximately 80% of the high frequency cut-point) is f resonant = ^M'* (3-D where the restoring force f f is roughly proportional to acoustic impedance and diaphragm mass m is proportional to the square of the diaphragm diameter. Thus, high frequency response requires low sensitivity and small, rigid, thin diaphragms. Two high quality microphones are compared in Table 6. The microphone actually used is also described in Table 6 [76]. Laboratory-made microphones worked, though not as well as commercial capacitance microphones, probably due to the method of construction [67]. The microphones were constructed of aluminized mylar without a trapped gas volume (to allow pump-down). The piezoelectric transducers used were 1/2" disks of PZT-5H from Vernitron, Inc. Also used were two sizes of piezoelectric cylinders and a thin, high frequency response (5 MHz) disk also of PZT-5H. One immediate problem with piezoelectric transducers is calibration. The output voltage of the transducer is [62] K 1 A n a p i + iwx C ^ 3 i} where K is the modulus of the piezoelectric, A is transducer area, C is the transducer and electronics shunt capacitance, Ap is the pressure change, t is the RC time constant of the transducer and amplifier, and u is the frequency of the pressure change. For u much greater than t , the output voltage is proportional to pressure change. For static

PAGE 70

60 Table 6. Capacitance microphone comparison. Property Radio Shack 33-1056 B & K 4144 Dynasciences 814 Acoustic sensitivity

PAGE 71

61 or low frequency pressure changes this is not the case. Calibration of the piezoelectric ceramic transducer used in this research is described in Chapter Four. Pressure Transducer Electronics Many preamplifiers and amplifiers were designed, breadboarded, modified, constructed, tested, and usually rejected. These included preamplifiers with discrete JFETs, discrete bipolar transistors, and bipolar, Bi-mos, or JFET input op amps. Although the noise model (Appendix Four) indicates that the discrete JFET-input voltage followerwith-gain circuit is optimal, this circuit is not easily implemented in the vicinity of the heated cell. Unfortunately, this is necessary because of the very high output impedance of the pressure transducer. The circuit eventually adopted was a source follower (10 10 ohms input impedance) used to provide impedance conversion. The circuit used a discrete JFET (2N5486) in an ac-coupled configuration with input time 3 constant of ICr seconds. Voltage gain was obtained with a Tektronix Model 26A2 differential amplifier which allows switch selection of 2 5 gams from 10 to 10 , adjustable bandwidth from dc to 1 MHz, and differential mode inputs. This was especially convenient when used with the data acquisition system and Fast Fourier Transform program because the bandwidth may be adjusted to avoid aliasing. Differential mode operation was necessary to eliminate power line hum pick-up. See Figure 10 for the circuit schematic. Having selected a pressure transducer and associated preamplifier, the electrical noise of the system was measured under no-signal conditions. For the system described, this was 11. yV rms determined by

PAGE 73

63 A/W ^i ~L -Ayw1> H^ o -WV-

PAGE 74

64 averaging 25 acquired noise spectra and assuming a crest factor of 4 [74]. The system bandwidth was 0.1 Hz to 1. MHz with shorted inputs and gain of 5.xl0. With the sensitivity of the Radio Shack capacitance microphone given as 3.2 mV/Pa and the conversion factor of 1 Pa = 7.50061 mtorr, the sensitivity is thus 4.2xl0" 4 V/mtorr. If the limit of detection is considered to be a signal to noise ratio (S/N) of unity (no signal averaging), then the pressure required to equal electrical noise is 0.20 mtorr. This pressure equivalent noise level is very important in determining the suitability of samples for TR0AS studies. By direct comparison of the piezoelectric transducer with the Radio Shack capacitance microphone, the pressure equivalent noise level of the former is found to be approximately 20. mtorr. This comparison was carried out in cell #2 with N0 2 (a strongly absorbing substance). Microcomputer Interface Electronics The electronics necessary to interface the microcomputer to the various amplifiers, chart recorder, peak detector, et cetera is shown in schematic form in Figure 11. The circuitry consists of relatively standard, independent subsystems interfaced to the microcomputer 1/0 lines which in turn are provided by Motorola 6820 Peripheral Interface Adaptor (PIA) LSI chips tied to the microcomputer address, data, and control busses. The functions of these subsystems are listed below: (1) chart recorder output (0-1 volts) via an 8-bit D/a converter, (2) four input channel multiplexerselectable active filtergain controlled amplifieroffset and level shifter8-bit A/D converter,

PAGE 75

= CO O +J > 0) oo

PAGE 76

V V U CL +-> O £ 3 — I s 66 -a ai •ren _ii. Jk_ E CO O if A 0) 4> -rA nvo cm —

PAGE 77

67 (3) transient recorder and stepping motor control lines (LSTTL or n-MOS), (4) photodetector peak detector with CMOS 4016 transmission gate reset, (5) photodetector linear response circuit with additional CMOS 40014 Schmidt trigger optional output. The stepping motor control lines and linear CW light detection circuitry are not used in TROAS; the circuitry was designed to allow conventional OAS with a Spectra-Physics Model 164 Argon Ion Laser and a Model 375 CW dye laser. The flashlamp-pumped dye laser, sample cell, light detector, and Spectra-Physics lasers were coaxially aligned, and the interface electronics and software facilitated rapid changeover from TROAS mode to OAS mode or vice versa. The data acquisition system incorporated a Biomation 805 Waveform Recorder with selectable sampling rates to 5 MHz, 2048 words of 8 bits each memory storage, pretrigger recording, sensitivity to 0.1 volts full scale, and microcomputer compatible data readout. A Tracor Northern NS-570 Digital Signal Averager with 20 usee (minimum) dwell time per channel, 9 or 12 bit data conversion, and automatic normalization of summed spectra was used in the earliest attempts at TROAS. The signal averager was found to be too slow and susceptible to RFI generated by a laser firing sequence. The Microcomputer Syste m The Commodore 2001-8 PET microcomputer system was chosen for use in the experiment for many reasons. The most important are hardware and software compatibility with previously used KIM-1 microcomputer

PAGE 78

63 systems; 6502 microprocessor-based, efficient architecture; fast 8 K BASIC with assembly language subroutines; built-in cassette tape drive and video monitor; low cost ($795). The main program used in the TROAS studies is listed in Appendix Four. The listing contains no "remarks" because of memory restrictions imposed by the relatively small amount of available random access memory (8K of RAM). The program consists of independent subroutines called from the keyboard by number. These are briefly described below: (1) initialization and status of microcomputer interface subsystems, (2) data acquisition from transient recorder and control of laser firing, light peak detector readout, (3) chart recorder output of raw data, (4) data storage in named cassette files, (5) 256 point Fast Fourier Transform subroutine based on the Fortran version by Bell [77], (6) video display of numerical data concerning FFT spectral intensities and powers, (7) line spectrum output to chart recorder of FFT results, (8) assembly language subroutine to eliminate the effects of the time delay (about 100 usee) in the laser firing circuit optocoupler, (9) least squares exponential curve fit subroutine.

PAGE 79

CHAPTER FOUR SAMPLE SELECTION In order to determine the validity of the theory of TROAS, suitable calibration standards must be run. These compounds should meet the conditions imposed by the theory and those arising from necessary compromises in the experimental apparatus. Previous attempts by Wrobel [47] failed to verify the fundamental two-component pressure rise expected to occur immediately after absorption of light by the sample. The observation of the pulsed optoacoustic effect by other investigators has been previously mentioned; those observations do not, however, constitute evidence for the validity of the theory of TROAS. The first compound used in the TROAS experiments reported herein was 2,2,4,4-tetramethyl-l ,3-cyclobutanedithione (dithione). Dithione is a deep red, crystalline solid with distinct camphoraceous odor. This compound and the related dione have been the objects of several theoretical and experimental investigations by Vala and coworkers [7880]. Although the UV-visible spectrum of dithione is still without a definitive interpretation, several important facts are known. First, several definite triplet peaks at 5943 A, 5926-5922 A, 5836 A, and o 5738 A are easily accessible; in fact, the dye laser output power with Rhodamine 6G dye spans these wavelengths and is highest at about 5900 A. The second important property is a relatively high molar absorption coefficient (approximately 1.5 1 mole" cm ). The third property is 69

PAGE 80

70 the significant vapor pressure (approximately 0.50 torr at room temperature). These last two results are due to Powell [81]. The pressure rise expected from dithione may be calculated using equation (2.16). The expected rise p' is 24. mtorr assuming l = 1.5 1 mole" 1 cm" 1 , p Q = 0.50 torr, T = 300 K, Y = 1 .4, k = 0, E = 1 J, and A = 1.27 cm (for cell #2). This is much larger than the pressure equivalent noise level obtainable using the Radio Shack capacitance microphone (0.20 mtorr) and approximately equal to that of the system if a piezoelectric transducer is used (20. mtorr). Since only the piezoelectric transducer can tolerate vacuum pumpdown, a carrier gas at atmospheric pressure must be used with the dithione to avoid damaging the capacitance microphone. Helium gas is an excellent choice because of its high purity, inertness, high sound velocity, and high heat capacity ratio. The effect of trapped oxygen in the microphone would be to convert slow heat to fast heat by triplet quenching. This problem will be deferred until later. The three time constants tj, t d , and t w may be calculated from equations (2.12) (2.14) respectively. Under the above conditions, these are t= 0.73 sec, t d = 6.6 usee, and t = 0.36 sec. For air at STP, they are t T = 2.7 sec, t , = 19. usee, and t = 7.3 sec. The i a w acoustic delay times t d were calculated assuming a = A/2 = 0.635 cm. With the frequency response of the microphone given as 20. Hz to 12. KHz, the corresponding allowed time window is 42. usee to 25. msec. Thus, the pressure detector limits the triplet lifetime which can be observed at both extremes. If the oxygen in the trapped volume in the microphone quenches the triplet, the result may well be solely fast heat, i.e., heat fast relative to the 42. usee limit. It is also known

PAGE 81

71 from previous studies [82] that the triplet lifetime of dithione is less than 1 msec. In addition, dithione definitely photochemically decomposes and also thermally decomposes to the mixed ketone and other unknown species. This decomposition to a viscous red oil occurs at temperatures below the reported decomposition temperature [83]. The dithione used in these experiments was supplied by the research laboratories of Tennessee Eastman. In general, impurities caused no problems in the TROAS experiments so that compounds were used as supplied, which was in no case less than 95% purity. In order to overcome the problem of decomposition and to obtain known radiationless rate constants, several compounds exhibiting the internal heavy atom effect were studied. These were naphthalene, 2-chloronaphthalene, anthracene, 1-chloroanthracene, and 9-bromoanthracene. These compounds have been studied for many years and consequently several of the triplet energies, molar absorption coefficients, and radiationless rate constants are known or can be estimated reasonably accurately. The lowest triplet in naphthalene, at 77 K in o a glass [84], is 4695 A. It may be expected to shift somewhat in the gas phase. Other relevant data (and references [35-87]) are given in Table 7. Figure 12 shows the structures of the organic molecules studied. Unfortunately, the molar absorption coefficients are very small for all of these compounds. It was hoped that the relatively high vapor pressures of several of the compounds, particularly naphthalene (10 torr @ 35.3 C [85]), would compensate for the low absorptivities. In addition to the above compounds, both l 9 and NO? were tried. Both of these suDStances absorb strongly throughout the visible, have been studied extensively, and have complicated spectra. Iodine has

PAGE 82

72 o

PAGE 84

74 o o o o o o o o 0=0 / 0=0 \ o o o

PAGE 85

75 been studied by conventional OAS techniques with interesting results [31]. When moist, iodine attacks aluminum very rapidly and stainless steel (316) slowly. It also attacks the silver metallization on the PZT-5H piezoelectric transducers yielding silver iodide and intermittant operation. Nitrogen dioxide does not appreciably attack either aluminum or stainless steel. It was used to compare the sensitivities of the capacitance microphone and the piezoelectric transducer. Reactive substances reportedly cause drift problems with externally biased capacitance microphones [87]. It should be noted that neither iodine nor N0 2 is suitable for a rigorous TROAS study because neither of them is adequately modeled by a simple two-state model. Biacetyl (CH 3 C0C0CH 3 ) is an excellent compound to study because of its unusual properties — high phosphorescence quantum yield (0.149) [88], low triplet energy (57 Kcal/mole), high vapor pressure, and unusual transition from small molecule behavior to large molecule behavior at an accessible wavelength (4450 A [89], 4430 A [32]). It is a small enough molecule to be accessible to detailed theoretical studies and relatively easy to study since both fluorescence (from 3-,) and phosphorescence (from T-j ) are readily observed and spectrally distinct. In addition, the photochemistry of biacetyl has been extensively studied [90-91] especially by Noyes and coworkers [92-94]. It is used routinely as a gas phase emission standard and as a triplet energy acceptor in the study of the photochemistry of other molecules [95]. It is also known [96] that biacetyl, at pressures below 10 torr, exhibits essentially no collisions during the lifetime of the lowest excited singlet state S-, (t = 10" 8 sec).

PAGE 86

76 Among the disadvantages of using biacetyl for TROAS studies are its reactivity, low S Q -^ T-, oscillator strength (1.4 x 10" 7 [87]), and foul odor (rancid butter). On the whole, the advantages probably outweigh the disadvantages insofar as TROAS is concerned.

PAGE 87

CHAPTER FIVE EXPERIMENTAL ARRANGEMENTS The major components of the attempted TROAS experiments are presented (in concise form) in rough chronological order in Table 8. The arrangement of the components was essentially that of Figure 2 though, of course, the initial experiments were performed without the aid of either microcomputer system or transient waveform recorder. Consequently, the complexity of the experimental apparatus increased (as necessary) to obtain results and eliminate interferences. The previous experiments had been performed in the near-UV (320420 nm) where the performance of the Phase-R DL-1200 coaxial flashlamppumped dye laser was yery poor, at best [47]. It was decided, therefore, to use this system in the visible region only, where dye performance is much better and alignment of the system is far easier. With Rhodamine 6G dye, the maximum output energy per pulse in the Phase-R system was guaranteed to be at least 250 mJ as opposed to the obtained maximum near-UV output of 150 pJ [47]. Poor laser performance, in general, and low pulse energy, in particular, were felt to be the primary impediments to successful completion of the previous work. As mentioned in the previous chapter, the maximum energy output of Rhodamine 5G occurs at approximately 590. nm which is almost ideal for excitation of the lowest triplet state of dithione. In addition to favorably placed triplet bands, dithione has a molar absorption coefficient of about 1.5 1 mole" cm and a vapor pressure of about 0.5 77

PAGE 88

73 1—

PAGE 90

80 torr at 300 K. Therefore, a small quantity of solid dithione was placed in cell #1 with the capacitance microphone (hereafter CM) connected directly to the input of the Tracor Northern NS-570 digital signal averager. Whenever possible, samples were run at room temperature first to see if a spectrum could be obtained. If no spectrum was obtained, the cell was heated via a doubly-insulated heating tape driven by a feedback controlled, fully proportional temperature controller. Dithione samples were heated as high as 85. C to increase sample partial pressure while not greatly increasing the likelihood of thermal decomposition. Temperatures were also limited to avoid thermal depolarization of the electret microphone diaphragm and to avoid significantly increasing the integral JFET preamplifier's thermal noise. Samples were also usually run with air as a carrier gas in the initial attempts with each sample since the effects of oxygen quenching of the excited triplet would be confined to converting slow heat to fast heat. The CM was only used with air at atmospheric pressure as the carrier gas because it is not possible to pump down the CM without causing diaphragm rupture as the trapped air beneath the diaphragm expanded. Despite these limitations, initial results were encouraging (see Figure 13). Unfortunately, the Phase-R system refused to tune or even lase narrowband; it also suffered numerous breakdowns culminating in the decision to switch to the Candela laser system. Additional problems encountered during the initial experiments included sample cell corrosion, possible oxygen quenching of the triplet (in an unexpected fashion, such as excitation of the metastable A n state), possibly inadequate low frequency response of the transducer and/or the digital

PAGE 91

S 01 TO -rO 0) O c CM O r— SZ 0) on +-> C_) rT3 CD -C S_ . O. +J r— ra ai I 01 — a> x> •rOJ 2 o cd O 40) en o Q. C oi -t— CD (/> en 01(8(0 f— 3 (O 0! 0) .c o .c (— U3 -^)

PAGE 92

82

PAGE 93

signal averager, false triggering of the signal averager by the Phase-R generated RFI/EMI, and thermal decomposition of the dithione after several hours in the heated cell. While the Candela laser system was on order, the next attempt was made using an Avco nitrogen laser-pumped dye laser equipped with electronic wavelength scanning drive. No results were obtained because the output pulse energy (200. pJ maximum) is insufficient to produce a pressure rise signal greater than the pressure equivalent noise level of about 0.20 mtorr. Therefore, a xenon flashlamp (Vivitar) with wavelength selective filters (Veb Jena er Glaswerk Schott & Gen., Jena) was used to provide pulse excitation of the dithione. A typical result is shown in Figure 14. The relatively slow data acquisition rate of the signal averager (20. usec/channel minimum dwell time) and CM allowed no conclusions to be drawn concerning the crucial initial portion of the spectra; it was decided, therefore, to upgrade the data acquisition system, associated electronics, sample cell, and pressure transducer. In addition, an Ithaco model 1391 low noise, charge preamplifier was purchased to provide amplification for the recently acquired piezoelectric disks and cylinders to be used as fast pressure transducers. Lack of adequate UV optics and pulse energy measurement equipment led to the decision to use the Candela system only in the visible region. Dithione was once again tried in hopes of improving upon the previous promising results. The piezoelectric transducer (hereafter PT) was used in addition to using the CM. These experiments and the Xe flashlamp ones previously performed were done using sample cell #2. This cell was far easier to load, clean, heat, and mount than cell #1. It could also be pumped down for use with the PT elements. Unfortunately,

PAGE 94

fO

PAGE 95

85 aansssad

PAGE 96

86 no significant results were obtained despite modifications to the preamplifier circuitry and heating of the sample cell. It was decided, therefore, to conserve the small remaining reserves of dithione for a later attempt. The next sample substance studied was bi acetyl. It was deemed a reasonable candidate for study by TROAS techniques because of the favorable properties listed in Table 7 of the previous chapter. The only anticipated problems were possibly inadequate triplet molar absorption coefficient, possible photochemistry or decomposition, and the definite strong rancid butter odor of biacetyl even in minute concentrations in air. A technical grade sample was repeatedly freeze-pump-thawed (liquid -4 nitrogen and 10 torr liquid nitrogen-trapped pump) to remove dissolved oxygen and placed in sample cell #2. The biacetyl sample exhibited its characteristic oxygen-free phosphorescence [32]. Several experiments were performed at room temperature using the PT and several different excitation sources. Broadband laser excitation below the triplet (T ? ) ° o at 4430 A gave a null result. Broadband (70-100 A) excitation above T 2 resulted in excitation of the lowest excited singlet with consequent triplet production via intersystem crossing. The result was large amplitude oscillations superposed en a small amplitude exponential pulse. Electronic filtering of the transducer output with an active fourth-order Butterworth filter (equal component Sallen-Key) removed most of the oscillatory features at the expense of the risetime of the buried exponential waveform. Similar results were obtained when unfiltered broadband Xe flashlamp excitation was employed. Triplet excitation via intersystem crossing from the easily excited lowest singlet is achieved at light

PAGE 97

87 intensities much lower than those routinely obtained with the Candela system [97]. In fact, triplet-triplet annihilation processes become important at low excitation source intensities [97]. Unfortunately, the resulting TROAS spectrum cannot be interpreted without the direct triplet excitation TROAS result unless the information it supplies is available from other sources. It was not possible to tune the Candela system to only excite the triplet because it refused to tune or lase narrowband (2 A) just as the Phase-R had. This problem was finally traced to the laser cavity length being too short. Attempts to make the system tune with a laser cavity of about 2/3 m were invariably unsuccessful while the system worked perfectly with a cavity of 2 m. Alignment did not seem to be at fault though a longer cavity must necessarily emphasize the effects of the dispersion element (diffraction grating) and the alignment is easier. An unexpected problem with biacetyl was its ability to dissolve ketone-resistant "0"-rings with consequent oxygen contamination of the degassed sample. To avoid these problems, five naphthalene and anthracene derivatives were selected for study. It was hoped that the stability and relative inertness of the compounds would enable the experiment to proceed. The molar absorption coefficient of the 2-chloronaphthalene triplet is approximately 10" 3 1 mole^cm -1 . The other compounds studied are likely to have similar extremely small values. It is exceptionally difficult to obtain such values by optical means. Unfortunately, this means that the expected pressure rise from these compounds may be below the pressure equivalent noise levels of the system regardless of pressure transducer used. Without the molar absorption coefficients this simply is not known. Nevertheless, the

PAGE 98

88 compounds were tried for several reasons: favorable triplet lifetimes, reasonable vapor pressures, photochemical stability, accessible triplet states in the visible, and thermal stability. Naphthalene was the most studied of the aromatic Hydrocarbons. As usual, the initial results were quite encouraging (Figure 15) though incorrect because the exponential decay is independent of carrier gas composition. This cannot be explained if the decay is actually due to thermal reequil ibration as it should be. Experiments were done at room temperature and again at approximately 85 C. Experiments were performed with piezoelectric transducers, electret capacitance microphone, and a laboratory-made externally polarized capacitance microphone. Naphthalene was run neat, with air, and with helium at pressures below 760 torr. Although the triplet wavelengths are known from 77 K glass phase results, they may be shifted in the gas phase. Therefore, the TROAS experiments were carried out by manually scanning the laser from longer wavelengths (below the triplet in energy) to shorter wavelengths (above the triplet). The dye laser was usually scanned over about 500 A in o increments of 10 A or less. The effect of wavelength scanning was quite unexpected. Instead of a relatively sharp onset of the pulsed optoacoustic effect, the acquired spectra had maximum amplitude near the dye output energy maximum. The spectra were reduced in amplitude as an extreme in the dye gain curve was approached and invariably the spectra had become nosignal baseline immediately prior to the laser reaching the ends of its tuning range. This was observed with dithione (Rhodamine 6G), naphthalene (Laser Dye 473), and anthracene (Kiton Red S).

PAGE 99

2 C_> «J O -M 00 X S01 •iOl T3 •r(T3 -rS> -a o) o C Q. i(O S Q. OJ oi -a u i— c (C i — ro SOJ -Q -t-> <_> -a (T3 S_ O oi aj s_ Q. S_ -Q Q. 0) 3 x: *— ' Q.OO E O *s3 E i— 4_> O) i — , Oi Q S_ C OI OI •rS_ VI O) 13 0) 2 -J C O) 00 _C i — O) +•> ra S-tJ= 3 2 4-> LO .e s_ a. oi oi O S. W c a. n3 (C OI •a O) S_ O) 10 «j c C_3 ro 00 O O • 0) h0) -M

PAGE 100

90 sjnssajd

PAGE 101

91 The behavior strongly suggests that the obtained spectrum is proportional to laser pulse energy but is independent of laser wavelength over several hundred nanometers in the visible. Accordingly, it was tentatively assumed that the obtained spectra were at least partially due to an artifact induced by the laser light pulse or firing sequence. Anthracene was tried next to see if the same wavelength dependence of the spectrum amplitude would be obtained. As noted above, similar results were obtained. If the cell was carefully cleaned with detergent followed by acetone rinses and then concentrated nitric acid and ethanol, a spectrum could still be obtained. Addition of sample compound caused an increase in signal amplitude and a definite increase in the complexity of the spectrum. This was taken to imply that some part of the spectrum was valid pulsed optoacoustic effect and the remainder was an artifact. Previous experience with the Phase-R system had shown that significant amounts of RFI/EMI were generated by a laser firing sequence. Shielding was finally effected by constructing an enclosure of brass screen (2 layers), wood, and woven RFI shielding strip. The enclosure was well grounded and an AM/FM radio inside would not play. Therefore, great care was taken to properly ground and shield the Candela system. The Candela laser was mounted on a 1/4" thick aluminum plate and covered with a 1/16" thick solid aluminum enclosure. The laser was run on a separate electrical circuit and computer fired via an optocoupler. These procedures largely eliminated RFI/EMI generated by a laser firing sequence except for an occasional spike produced in the spectrum at the moment of laser firing. With the aid of the microcomputer-controlled transient recorder and a few lines of

PAGE 102

92 assembly language code, it was possible to rule out RFI/EMI as the cause of all but the occasional spike. The spike was found to precede the transducer output of the spectrum by several microseconds and was eventually eliminated by modification of the amplifier circuitry. To determine whether a real pulsed optoacoustic effect exists, iodine was run in cell #2. Although iodine cannot be modeled with a simple two state model, it does absorb strongly in the visible, has a vapor pressure of 0.53 torr and 300 K, and is capable of exciting the metastable, long-lived (45 minutes) oxygen a state [87]. Thus, the behavior of iodine vapor differs in air from its behavior in helium. The reactivity of iodine, especially in the presence of traces of moisture, precluded use of a capacitance microphone largely constructed of aluminum. Although the silver metallization on the PT elements was attacked by the iodine vapor, intermittent operation was not especially troublesome after the silver was gone. An iodine spectrum is shown in Figure 16. The cell was at room temperature (25 C) and spectra were obtained neat, with air at various pressures, and with helium at various pressures. Figure 16 shows that a pulsed optoacoustic effect is indeed observable. After iodine, the remaining three compounds in the aromatic hydrocarbon series were run. Figure 17 shows a 9-bromoanthracene spectrum obtained using the PT. Similar results were obtained with helium as carrier gas. No results were obtained with 1-chloroanthracene or with 2-chloronaphthalene. Air was used as the carrier gas for the chlorinated aromatic hydrocarbons. The artifact interfered with all of the previous spectra to some extent.

PAGE 103

Q. 0) -It-C X ra -^ 0) O)

PAGE 104

94 Pressure

PAGE 105

2 2 s(U h ra OJ ra C Q. C OlE'r s_ -ra; 03 •r— 1/1 +J ra ai i— c .c i— co c 0) C-P 1), — ra t/i ra Ol 3 00 a; s-a .c _ o +-> o a: <+3 E o -o c o :/! O J S_ CLt•<0) ul s. S > -M .a •.I— O TJ M ro CO (U
PAGE 106

96 Pressure

PAGE 107

97 Figure 18 shews a typical TROAS artifact spectrum. To analyze such results, a Fortran Fast Fourier Transform program was converted to Basic and assembly language (on the 6502-based PET microcomputer) [77]. With the aid of this FFT subroutine, many spectra were obtained and processed at various transient recorder sampling rates. The FFT subroutine was tested for accuracy with summed signal generator input, and it was tested on data calculated from an analytic exponential pulse. The transform calculated from the data via the FFT subroutine was essentially identical to the transform calculated from the analytic Fourier transform. Almost all of the processed spectra showed peaks at about ICO. KHz, 204. KHz, 450. KHz, and 750. KHz. In addition, mechanical resonances of the sample cell may be excited by external sounds (whistling or snapping one's fingers) or by external vibrations (light tapping of the cell or nearby equipment). These frequencies were primarily in the range between 2.5 KHz and 8.0 KHz. Even the sound of the laser driver spark gap when triggering occurs can excite the cell mechanical resonances. Fortunately, cell mechanical resonances were of little importance since the transient recorder was usually done acquiring a spectrum before the spark gap sound could reach the cell. Acoustic shielding was not necessary at any time. In an attempt to avoid the artifact, sample cell #4 was constructed. Despite the switch to (roughly) spherical symmetry, the expected pressure rise should be approximately the same. The acoustic resonances would be expected to occur at different frequencies than in the cylindrical cell. Note that the longitudinal acoustic resonances in the cylindrical cell are harmonically related while the radial modes are

PAGE 108

C S0) •tCD 31 ST3 JS 3 1/1 O i/l C 1/1 (B C o> so> i-M CD CL -Q CJ O •!0) Is> S_ +J IB Q.1— IO 01 (U J(/I O O X Q. E M iB O c/1 4-> CD S<«-u a Q..C Q.«C T3 I — IB — -^ a -Q CD iB .c +-> <-> 1_ 4— CD • O U lo 3 C\J +-> T3 +-> O C >,+-> IB S. fl U) 3 fTJ -p0) (13 O JZ -2^ +J •!QJ — +j ui T3 .a +-> •i•-en i. 3 s -a sz o o a. •tO 4-> 4-> iD O T3 i— ai a> 03 (/) ifl T3
PAGE 109

99 3anss3J(j

PAGE 110

100 proportional in frequency to the roots of the integral Bessel function J-| . Thus, the radial resonances are not harmonically related. In a spherical cell, the modes are related to the spherical harmonics and thus the new cell should aid in determining whether the acoustic modes are excited as expected. Solid anthracene was placed in cell #4 and strongly heated in vacuo. The vapor pressure of anthracene at 25 C is very low (0.38 mtorr) but is 0.57 torr at the approximate temperature used (150. C). Experiments at both temperatures quickly revealed that the PT produces an output signal when struck by light. Experiments with an empty cell and both Candela system and Xe flashlamp gave similar results. The output waveform was found to be an exponential waveform with amplitude approximately proportional to light intensity. Blocking the direct light path from excitation source to transducer by placing a small piece of electrical tape on the outside of cell #4 eliminated the artifact completely. It was known that piezoelectric transducers are susceptible to light generated artifacts; they may, for example, give a response to the shock front-induced luminescence in shock tube experiments (where the transducers are used for pressure measurements) [62]. The amount of light scattered onto the pressure transducer when installed in cell #2 was found to be small. This was determined by removing the transducer and firing the laser repeatedly, under microcomputer control, while looking into the port, ^lery little light was scattered from the side walls of the cell out the side port. Nevertheless, when the piezoelectric transducer is directly exposed to the laser output, the result is shown in Figure 19. The

PAGE 111

V r•r> U

PAGE 112

102 Pressure

PAGE 113

103 signal is so large no amplification was needed. The FFT of this spectrum has components at 204. KHz (very large), 450. KHz (small), 750. KHz (small), and negligible background. Since the nominal resonant frequency of the particular transducers used is 160. KHz, the 204. KHz peak was tentatively identified as the actual resonant frequency of the transducer. This was verified by direct excitation of tne normally mounted PT using an externally mounted PT driven by an audio sine wave generator. No output was obtained until frequencies of 203. KHz, 450. KHz, and 750. KHz were selected on the signal generator. This was not, however, the only artifact. Sample cell #3 was devised to determine whether the artifact was due in part to an elastic wave in the sample cell. Anthracene was melted in vacuo into intimate contact with the glass cell, and a piezoelectric disk was firmly pressed against the outside of the glass. See Figure 7 for details. The PT was electrically shielded and also shielded from the effects of direct light by the small, snugly fitting aluminum enclosure. Nonreproducible results were obtained with both the laser system and broadband flashlamp excitation at room temperature (where the vapor pressure is a negligible 0.38 mtorr). With the aid of nitrogen dioxide, the sensitivity of the PT was compared to that of the CM and found to be a relatively poor 20. mV/mtorr. As with iodine, the study of N0 2 is not feasible because of the complexity of its spectrum. It does not, however, attack aluminum or stainless steel . The matter of the acoustic resonances was still not resolved, and the artifact was not solely due to the direct effect of light. To clarify the situation, a very small cell was constructed (cell #5).

PAGE 114

104 With a small cell, the acoustic resonances shift to higher frequencies, and the mechanical resonances are also shifted upwards in frequency because of the size and mass reduction. It was hoped that the risetime of the pressure rise would be only marginally impaired by the increased importance of wall effects [59]. Sample cell #5 was a modified Cajun Ultratorr 3/8" tee connector. It was machined to accept either a PT disk or a CM. Naphthalene was placed in the unheated cell, and the Candela system was used to try to excite the triplet state. The results were similar to those obtained in the much larger cell #2. To eliminate the light-induced artifact, a stainless steel disk was placed in the pressure transducer port in front of the transducer. The experiments were repeated with and without naphthalene in the cell and with both transducers. The result was an artifact spectrum in each case. Turning the cell so that the laser pulse struck the outside of the cell also gave an exponential pulse artifact spectrum. This result was also obtained with cell #2. It was completely unexpected since previous experiments with the relatively massive cell #2 (mass greater than 5. kg) had involved the use of a variable aperture between the laser output mirror and the sample cell. It was also noted that only a small amount of incident light struck the cell walls. With this information, dithione was placed in cell #2 in an attempt to finally get an artifact-free TROAS spectrum. Through the use of multiple irises, the laser pulse was prevented from striking any portion of the cell other than the Suprasil windows. Again, no significant results were obtained regardless of cell, transducer, carrier gas, or beam diameter. A curious result of the use of the irises was the

PAGE 115

105 need for three of them positioned as far apart as possible in the space between the laser output mirror and the sample cell. The approximately 1" diameter laser beam had to be narrowed to slightly less than 1/5" to significantly reduce the effect of the artifact. It was not sufficient to merely eliminate a line of sight path from the laser cavity to the nonwindow portion of the cell. This, naturally, greatly reduces the laser energy incident upon the sample and cell windows. For a TEM nf) Gaussian beam of diameter 1" with an aperture of 1/5" diameter, the ratio of transmitted energy to incident energy is For a multimode beam, the result is (0.2/1 .0) 2 = 0.040. Thus, between 4% and 8% of the incident energy is passed along to the sample eel 1 . Consequently, irises were dispensed with and a simple test tube was used as cell #6. The CM was snugly inserted into the test tube mouth after solid dithione was placed in the test tube. The CM was wrapped with Teflon tape to achieve the snug fit required to give immunity to extraneous room noises. The cell was heated with a separately mounted heat gun. No results of a reproducible nature could be obtained. Since the curved test tube might perhaps reflect light onto the CM, another highly transparent cell was constructed (see Figure 9). The diameter of the cylindrical cell was more than large enough to allow the full laser beam to traverse the cell length without impinging on

PAGE 116

106 the cell walls. Nevertheless, the irises were used to produce a 1/2" diameter beam to be certain scattered light would not be troublesome. Laser firings showed that light scatter and reflections were negligible, Dithione solid was placed in the cell, and the cell was heated by the previously mentioned heat gun. The CM was used as the pressure transducer. Again the only spectra obtained were contaminated by an artifact. The empty cell gave similar results. Turning the cell so that reflected light definitely struck the front of the CM did not increase the artifact amplitude. Finally, a PT was epoxied to the rear of a small front surface mirror which was then epoxied to a corner of the square glass plate used as the rear window of cell #7. The PT gave a distinct artifact output when a laser pulse traversed the cell whether air or helium was used as carrier gas in the cell. No absorbing substance was placed in the cell . Several details were neglected in the above rather concise description of the experimental arrangements. First, several substances, for example, iodine and naphthalene, were run at times other than those listed above in addition to the described sequence. Second, the charge preamplifier was replaced with various voltage preamplifiers during the course of the experiments. The charge preamplifier was found to be noisy, did not prove immune to cable microphonics, and did not provide adequate gain. Another recurring problem was condensation of sample on the cell windows. This was definitely the cause of an artifact signal. Consequently, the cell was heated until any condensed sample vaporized or melted and drained away.

PAGE 117

107 The laser tuning problem was peculiarly difficult to correct because of its unusual nature. Three different diffraction gratings, two rear mirrors, and eight front reflectors (three laser mirrors and five neutral density filters) were used in attempts to achieve narrowband tuning. Numerous laser dyes failed to tune despite concentration changes, pumping speed variations, and an attempt to modify the solvent pH with acetic acid [56]. A call to the president of Candela Corporation (Dr. Horace Furumoto, one of the developers of the coaxial flashlamp [98-100]) was of little help though his suggestions concerning the addition of a bubble trap/particle filter and the necessity of turbulent dye solution pumping were immediately implemented. Cavity length was simply not suspected since both Candela and Phase-R sell short (about 45 cm) optical benches for use with their systems. Additional steps taken to eliminate the tuning problem, prior to discovering the cause, included use of intracavity irises and even use of a low reflectivity mirror between the grating and flashlamp [101].

PAGE 118

CHAPTER SIX RESULTS That a pulsed optoacoustic signal was observed in these experiments seems established beyond reasonable doubt. On numerous occasions, a blank was run by illuminating a clean, evaculated cell. An additional blank was run with air in the cell and then sample substance was added and allowed to equilibrate. The resultant spectrum was always larger in amplitude and reproducibly more complex. This was especially easy to observe with iodine or nitrogen dioxide as samples but was also observed with dithione, bi acetyl, naphthalene, and 9-bromoanthracene. Figure 20 shows the spectrum obtained in a run with a clean, empty (10 torr) cell under conditions almost identical to those used to obtain the naphthalene spectra of Figure 15. Although the blank was run with four times higher sensitivity than that used for the naphthalene spectra, the artifact is plainly visible even without the aid of increased sensitivity. Unfortunately, the artifact was not reproducible. It was found to vary in amplitude as much as tenfold from one shot to the next. The laser was not especially stable in output pulse energy (about 10%) but was not directly responsible for the gross variations in artifact amplitude. Furthermore, the waveform of the artifact is only roughly an exponential pulse and varies from shot to shot. This rules out any possibility of signal averaging or subtraction of the artifact from the artifact-containing spectrum. 108

PAGE 120

no ajnss3J(j

PAGE 121

in Although differential mode operation with two sample cells is possible, it will not eliminate the artifact. If the cells are arranged coaxially, the second cell will receive less light than the first and will also receive more scattered light. In addition, the second cell cannot present exactly the same environment as the first because it does not contain an absorbing sample. Furthermore, the cells must thermally track to avoid thermal drift in the transducer outputs and any buildup with time of absorbing or scattering films (due to decomposition of sample, for example) must be identical in the cells. Similar restrictions apply to the standard dual-beam configuration since alignment becomes difficult, and the cells must be precisely equidistant from external noise sources such as the flashlamp driver unit spark gap. Nevertheless, Deaton et al. have used the series arrangement in low source intensity OAS [102], and Hordvik and Schlossberg [103] have used dual piezoelectric transducers in differential mode with lock-in amplifier detection. Even so, they were limited by the in-phase spurious response of their system. As stated previously, cell mechanical resonances were not especially troublesome though the vacuum system did have to be disconnected from the cell to avoid pickup of the rotary oil pump noises. Condensation of sample on the cell windows definitely produced an artifact. Both of these artifacts may be easily avoided by proper design. This is not true of the acoustic resonances. It would be best to avoid, as much as possible, putting energy into the acoustic modes since this reduces energy available to produce the concomitant pressure rise. This means that the excitation beam should have a diameter equal to

PAGE 122

112 the cell diameter. But an artifact is produced if any portion of the cell is illuminated. Hence, the acoustic modes are not easily minimized. Electronic filtering is possible, but cumbersome, since six or more sharp notch filters may be needed, and they must be readily tunable since the acoustic mode frequencies are dependent on the speed of sound in the sample. In addition, signal risetime may be degraded unless care is taken. For this reason, an attempt was made to employ the technique of mathematical (or digital) filtering. Briefly stated, this technique involves transforming a time domain spectrum to the frequency domain via the FFT algorithm, removing or modifying selected frequencies and transforming back to the time domain. The advantage of digital filtering is that perfect filters can be rapidly implemented. Any desired transfer function and phase shift can be easily implemented even if they are physically impossible to realize. Thus, the acoustic resonances and cell mechanical resonances could be eliminated without degrading either the low or high frequency response of the acquired signal. This attempt came to naught because the 3 K of RAM on the PET microcomputer proved insufficient to contain the main program and the necessary array storage needed by the FFT algorithm. It should be noted that the FFT subroutine in Appendix Four cannot be used directly to back transform since it computes the 128 point power spectrum of 256 real input spectrum points. Nevertheless, the subroutine contains the complex Cooley-Tukey algorithm needed to perform the back transform. In fact, the necessary modifications were made and even a 128 point input spectrum caused memory overflow problems.

PAGE 123

113 The corrected mode amplitude equation (A1.20) and the signal amplitude equation (A2.1) may be used to compute the ratio of the amplitudes of the lowest radial mode and the pressure rise. The result is A-jO^J/AqCO) = Q^ty (6.1) where u^ = c^ ^c/A is the lowest radial frequency, a. , is the first nontrivial root of J ] (3.832), A is the cell radius, and c is the speed of sound in the sample. For cell #2, M , is 99.9 x 10 3 radians/sec in air at STP. The amplitude ratio is thus 12./(99.9 x 10 3 sec _1 )(2.7 sec) = 4.4 x 10" 5 where Q-j is assumed to be 12. (the approximate measured value for cell #2). The value of Q-j was obtained using a signal generator, oscilloscope, and small speaker attached to cell #2. Values of Q as high as 890 have been reported in the literature [23]. Actually, the amplitude of the acoustic modes depends on the sample composition and is definitely larger than the above ratio suggests. The acoustic resonances are of large amplitude in iodine, moderate in 9-bromoanthracene, small in dithione, and \>ery small in naphthalene. Evidence for the existence of the acoustic modes is not as clear-cut as might be hoped since the artifacts, particularly the 204. KHz transducer resonance, obscure the acoustic modes. The longitudinal and radial mode frequencies are given by 27Tf ln = u» n = ^cn z /L for n z = 1,2,3, . . . , and (6.2; 2rrf ri = u). = a^.c/A for 1 = 1,2,3, .... (6.3)

PAGE 124

114 If an FFT is performed on a typical spectrum, the obtained frequencies will not generally completely match the acoustic mode frequencies given by the above equations although as many as five matchups may occur. The odd thing about the matches is that the spectrum might contain only the lowest two radial modes and the third and fifth longitudinal modes. One explanation for this is that sampled acoustic mode frequencies which are not sampled precisely an integer number of times will appear as a rather broad band (6 to 10 or so) of peaks in the FFT spectrum. This phenomenon is called leakage, and it may be minimized by reducing the sampling rate (for better resolution) and by smoothing the raw spectrum before processing (a process called apodization) [104]. In the present case, the program used is quite similar to that of the above reference, and leakage was definitely shewn to exist by using a signal generator as the input source. Reduction of the sampling rate was not feasible in the present case because of the strong 204. KHz piezoelectric transducer resonance. If the 204. KHz signal is not removed and the sampling rate is sufficiently reduced, aliasing will occur. If filtering is employed, care must be taken to avoid distorting the spectrum in other ways. In particular, the risetime must not be limited by the filter. Whether the acoustic modes are observed is of little significance if the transducer artifacts cannot be eliminated. The experiments outlined in the preceding chapter were not successful in eliminating the artifacts, but they do suggest the causes. Consider first the light generated artifact. Bradley [62], Hordvik and Schlossberg [103], and Farrow et al . [69] note that piezoelectric transducers produce a spurious output signal if exposed to the illumination light source. Bradley

PAGE 125

115 claims the effect is pyroelectric in nature, so presumably the effect is directly proportional to the incident light intensity. This is reasonable if the output voltage is directly proportional to the stress which in turn is proportional to the heat generated in the transducer by absorption. Hordvik and Schlossberg confirm the thermal nature of the effect and the proportionality. It is not possible to baffle the light without possibly setting up new cell resonances (e.g. Helmholtz resonances [105]) and attenuating the acoustic signal. Nevertheless, a small plug of opaque foam (used to protect CMOS or n-MOS integrated circuits normally) was placed in front of the transducer without benefit. The cell (#2) was even lined with brass screen to block the light without greatly impeding the desired pressure rise. This, too, was unsuccessful . The light-induced artifact is not avoidable even by reducing pulse energy because both signal and artifact are directly proportional to source intensity for given pulse width, and they are commensurate in magnitude. Hordvik and Schlossberg [103] state that "in our experiments the sensitivity was not limited by electronic or environmental noise, but rather by the signal generated by radiation scattered from sample inhomogeneities onto the transducer." The samples used there were large crystals of CaF 2 , SrF 2 , and BaF 2 . Dual piezoelectric transducers in differential mode were used (as noted previously) to measure absorption coefficients down to 10" 5 cm" 1 , and they note that 10" 6 cm" 1 is obtainable with 1 W of laser excitation power if scattering can be avoided. Farrow et al . have performed OAS and pulsed OAS studies on opaque solids using piezoelectric detection of the resultant elastic wave in the solid [69]. They note that signal averaging is necessary

PAGE 126

116 even with opaque solids because of the low signal to noise ratio of individual acoustic transients. The excitation pulses were supplied by an acousto-optic modulator following an argon ion laser and pulse width was 1 msec. The front surface mirror technique for coupling elastic waves from sample to transducer while blocking incident light is due to Farrow et al. It was tried with cell #7 to determine whether the cell itself generates the indirect artifact. It does. In fact, while the lasers used by Farrow et al . had an average output power of 190. mW, the average output power of the Candela system used here was greater than 1. MW. Hence, the transducer and front surface mirror combination did not "block the light" and thereby eliminate an artifact. When the Candela system was fired directly onto the mirror, a large exponential waveform was produced by the transducer. The cause is absorbance by the mirror. The results of Hordvik and Schlossberg [103] are important in understanding the nature of the indirect artifact. They note that the piezoelectric transducer may be used to measure the strain produced in the solid sample following illumination. The equation they arrive at is [103] e rr = -e 9e = -aPa Q (l +a)t/27rCp(l -a)r 2 (6.4) where e rr and e Qe are the relevant strains, a is the absorption coefficient, P is the total beam power, a Q is the thermal expansion coefficient, a is Poisson's ratio, C is the specific heat, p is the density, and r is the detector position. Typical values for their experiments were a = 10" 4 cm" 1 , P = 0.1 W, a Q = 10" 5 K" 1 , t = 5xl0" 3 , r = 2 mm, -1-1 3 C = 1 J g K ,p=3g cm" , and a = 0.3. With tnese values,

PAGE 127

117 -12 e rr = -1.23 x 10 which is sufficient to produce a transducer output of several microvolts. It is important to note that the transducer output is directly proportional to the strain so transducer output is proportional to incident laser power and to sample absorption coefficient. The strain (and signal) will vanish if the thermal expansion coefficient is zero. The importance of acoustic impedance matching is now apparent. The piezoelectric transducer is well coupled to the sample cell since both are solids, and the reduced pressure in the sample cell would act to force the transducer more tightly against the sample cell. If the above equation is solved for a Suprasil 2 quartz window, the result is e = _ (lO^cm^mo 6 W)(5.5xl0~ 7 K -1 )(l + 0. 1 7) (5 x 10~ 3 sec) rr 2(0.75 J g" 1 K" 1 )(2.20 g cm" 3 )(l -0.17)(0.2 cm) 2 = -2. xlO" 6 where the absorbance of Suprasil 2 is assumed to be 10" 4 cm _1 . This is definitely too large, but it indicates, nevertheless, that even a Suprasil 2 window may be expected to produce a signal when illuminated by the high intensities available from the Candela laser system. The production of an artifact signal when sample vapor condensed on a window and was illuminated, is explained by the excellent acoustic impedance coupling obtainable even with liquid droplets on the windows (where <* t = 0.2 typically). This mechanism will be designated the strain artifact. Parker [40] has also observed an artifact due to optical absorption by borosilicate glass of light of wavelengths shorter than 2.5 pm. He used a conventional 0AS apparatus with capacitance microphone pressure

PAGE 128

118 transducer and high pressures (to 35 atmospheres) of nonabsorbing gases such as Ne, N 2 , and 0^. His experimental results support the predictions of his theoretical model that a thin boundary layer in the window material absorbs light, produces heat, and by thermal diffusion into the carrier gas, periodic pressure fluctuations are produced. These fluctuations take place in a boundary layer of carrier gas proportional to the thermal diffusion length in the gas. This mechanism was independently arrived at by Rosencwaig and Gersho [44], who used it to explain their results on the OAS of solid samples. They elaborated upon the similar results of Parker. Rosencwaig has also noted that "the cyclical expansion and contraction of the solid cannot be the major source of the signal" [106]. The thermal diffusion length is the distance over which a thermal wave is attenuated by e~ . The Rosencwaig and Gersho model applied to situations where the substrate and carrier gas were thermally thick (i.e., the thermal diffusion length is much less than the thickness of the substance). The sample applies to the model used by Parker. These results were further extended by Aamodt, Murphy, and Parker [107] to allow for thermally thin substrate and gas phase. This extension gives better results but not quite good enough. Accordingly, McDonald and Wetsel , Jr., [45] extended the solutions further to include precisely the effect Rosencwaig omitted. Acoustic waves in the condensed phase sample result in a mechanical motion of the sample surface. This mechanical piston action couples with the previous thermal piston action in what is dubbed the composite piston model. This model is the basis of standard condensed phase OAS experiments which employ capacitance microphones to detect the heat-flow generated

PAGE 129

119 acoustic waves in the carrier gas. Since acoustic waves are set up in the sample, due to cyclic thermal expansion and contraction of the sample, these may be detected directly by an appropriate transducer. This has been called thermoacoustics [108]. Whatever its name, it is much more efficient than the slow thermal piston type of standard experiments because the carrier gas and slow thermal waves may be dispensed with and the acoustic impedance mismatch corrected. The results of Hordvik and Schlossberg [103] are based on detection of the strain produced in single absorbing crystals. Farrow et al. [69] have noted that the acoustic waves (elastic waves) in the condensed phase samples can be detected yery efficiently with piezoelectric transducers precisely because of the excellent acoustic impedance match. That acoustic waves are indeed generated when a solid absorbs pulsed light was demonstrated by White [15]. Brienza and DeMaria [46] have obtained similar results and attribute the acoustic waves to absorption of light with subsequent thermal generation of acoustic waves. The significance of these results is that the cell windows generate two types of spurious response via the composite piston model and the efficiency of most transducers in responding to substrate acoustic waves. Note that using a capacitance microphone will not help significantly because though the microphone may be less sensitive to substrate elastic waves, it will be more sensitive to the carrier gas acoustic waves, which are the predominant problem as the accuracy of the relatively simple Rcsencwaig and Gersho model shows. This is also verified by Dewey, Jr., who notes that in conventional OAS "the dominant source of background noise is absorption at the cell

PAGE 130

120 windows" [65]. He remarks that the windows produce an in-phase signal while surface absorption at the cell walls produces an out-of-phase signal. The result of his analysis of the situation is that the thermal piston action will be minimized for windows of high density, high heat capacity, and high thermal conductivity. This minimizes heat flow into the carrier gas. To avoid the strain artifact, window absorbance must be minimized by choice of material and window thickness, and the thermal expansion coefficient must be minimized. Capacitance microphones are also susceptible to light-induced artifacts as noted by McClelland and Kniseley [109]. They state "microphones are often highly sensitive to scattered light because absorption at the microphone couples efficiently to the microphone diaphragm." The Princeton Applied Research Model 6001 photoacoustic spectrometer uses a capacitance microphone to detect the acoustic signal. It is isolated from the sample chamber via an interconnecting duct with several right-angle bends to trap scattered and reflected light and thereby avoid producing a light-induced artifact. Since all three of the artifacts are proportional in amplitude to the laser pulse intensity, they may be ameliorated by reducing the laser intensity. This is not possible, however, unless a sample substance is found which is markedly superior to the substances chosen for these studies. Several additional experiments were attempted during the course of the previous gas phase experiments. A cylindrical piezoelectric transducer was filled with a saturated solution of anthracene in ethanol and illuminated with both the Candela laser and broadband flashlamp excitation. It was not possible to obtain results similar

PAGE 131

121 to those of Call is et al . [20] because of the artifacts. Placing the cylinder inside cell #2 and filling the cell with an acetone solution of dithione gave similar results. The laboratory-made microphone also produced an artifact. An attempt was made to use cell #4 with solutions of anthracene (saturated) in ethanol and eosin in ethanol . The direct light artifact was apparent, but the decay time constant was four times longer for 200 proof ethanol than for a dilute solution of eosin in ethanol. An experiment was also performed with the front surface mirror and piezoelectric transducer combination. As previously stated, the laser pulse causes an artifact when the mirror is struck. A ruby boule approximately 1 cm thick was epoxied to the front surface mirror. Pink ruby has an absorbance of about 3.3 cm at 5600 A [110]. The absorbance is from approximately 4900 A to 6100 A. Consequently, the laser system was operated with Rhodamine 6G broadband. No increase in signal was observed. Note that if the excitation source intensity is low enough to be completely absorbed by the sample on the mirror, signal averaging may be used to improve the S/N ratio of the transducer output, if necessary. In the experiments of Farrow et al . the average source intensity (190. mW) was low enough to require averaging of as many as 3 10 acoustic transients.

PAGE 132

CHAPTER SEVEN SINGLET TIME-RESOLVED OPTOACOUSTIC SPECTROSCOPY If it is difficult, as Eddington has said, to empty the well of Truth with a leaky bucket, then we must content outselves with the use of a cup. Consider, for example, the results of Callis et al. [20] concerning the triplet yield of anthracene in ethanol solution. Using broadband flash-lamp illumination, capacitance microphone volume change detection, and signal averaging, they obtained average triplet yields of 0.66. This compares favorably with values of 0.72 in ethanol [51] and 0.53 in liquid paraffin [48]. The equation from which their results were obtained is -t = Q s (h v in " *f hv f)/Q E t ( 7 ] ) where Q s is the slow heat, Q is the total heat (= Q s + Q f ), E t is the triplet state energy, hvn is the average exciting photon energy, hv f is the average fluorescent photon energy, and the capacitance microphone signal is directly proportional to the released heat. The equation was derived assuming $ is much less than $.. The model used is the same as that presented in the introduction with k T v neglected and k T v T 'l 'l'h large relative to k* Note the similarity with the result obtained by Wrobel [47] *t = p ;^S;,T^ /p ; h ^ < 7 2 : 122

PAGE 133

123 The above equation (7.2) applies to the case of direct excitation from S Q to T-j. Consider now the case of direct excitation from S Q to S?. Since this is a spin-allowed process, much lower incident light intensities should suffice. This means that all of the artifacts will be reduced in intensity proportionately while the desired pressure rise will be * t of its previously expected value (2.16). It may be possible, therefore, to entirely avoid the artifacts. It must be shown, however, that singlet excitation is capable of producing the desired triplet yield information. For direct excitation into S° from S Q , the normalized expected pressure rise is given by Wrobel [47] as (with = k, = k" r = kjj: 7 8 8 p o,o r , ic jcr -k*t + $ * o hv,0 (1 -e p ) ] (7.3) z 'l 'l ISC l ISC IC where p = * $ o hv T 0, <$ T = k T /k* $ c = k /k* and •> 1 l 1 P ^i st IC Pr = *. hv T + $ o VO . Let hv_0 * E. , hv c = E t t S ] ,T 1 S 1 S ] T 1 t S 1 s hv Q T = D. Then we have the following: V'i P ; /P f = * t *lf V<*t D+4 S° E s> and (7.4) Assume, as Call is et al . [20] did, that * n is much less than * + . From P t the definitions (Chapter One), this implies k is much less than k y . ii ^§C . Hence, $ T is approximately unity. Assume also that D is much less 'l than E t . Then the above equation simplifies to (since E f < E )

PAGE 134

124 ^f = *t E t /$ S° E s • ^ 7 5 ) But this may be rearranged to the following: v»s° = (E s /E t) p ;/ p f • ( 7 g ) IC Note that * t /* s is equal to k ISC /k s . We are able, therefore, to experi mentally determine the ratio of the two unknown quantities in the known observed flurorescence lifetime (i.e., (k*) and * f are obtainable by fluorimetry and k* = k $ + k ISC + * f k£). The values of k and k y are obtainable by phosphorimetry from * and k* (Chapter One) when the r P triplet yield is known. Hence, all five desired quantities are available and the definition of * (Chapter One) directly gives the desired result: * t ^sctk?)" 1 (7-7) The triplet yield equation for singlet TROAS has several immediate consequences. First, because of the approximations made in its derivation, the yield equation is expected to be most accurate when fluorescence yield and particularly phosphorescence yield are small and when their spectral regions heavily overlap (i.e., the singlet-triplet separation is small). Note that this is the most difficult situation to study for phosphorimetry and fluorimetry. Second, despite the approximations, many more compounds are available for study by singlet TROAS than by triplet TROAS, particularly since lower power lasers, such as nitrogen laser-pumped dye lasers, may be employed to excite the singlets. Energies of several mi 1 1 i joules in the near UV would be

PAGE 135

125 especially advantageous, particularly if obtainable at repetition rates concordant with signal averaging. Third, if phosphorimetry data are available, the phosphorescence yield may be used to obtain, via equation ISC (7.4), a more accurate $ T value, which in turn may be used to obtain 'l a more accurate triplet yield value. Despite the lack of results from the triplet TROAS studies, there is e^ery reason to expect, with adequate pressure transducer and associated electronics, that the singlet TROAS technique will prove useful and convenient.

PAGE 136

APPENDIX ONE RELEVANT OAS RESULTS As previously mentioned, the results obtained by Wrobel [47] in his development of the theory of TROAS are incomplete since the equation he chose as fundamental neglects all dissipative processes external to the model-specific source term. This fundamental equation (Al.l below) is a consequence of the conservation equations and equation of state of the sample. Precisely the same equation underlies the theory of OAS. However, as Kreuzer has noted "this equation does not include the effects of acoustic loss produced by heat conduction and viscosity. In order to properly discuss the optoacoustic effect, it is necessary to include these mechanisms" [63]. If the mechanisms are not included, it is not possible to avoid such consequences as unrealistically large or infinite acoustic mode amplitudes, undefined mode amplitude ratio expressions, and noise equivalent powers of essentially zero. In TROAS, similar problems occur. In fact, the fundamental pressure rise equation (2.10) is evaluated in the limit as t->« since no thermal reequilibration is predicted by the theory of TROAS if the loss mechanisms are neglected. Consequently, the losses must be readmitted to the theory. This is elegantly done by Kreuzer [63] using Fourier transform techniques. It is not necessary, therefore, to replace equation (Al.l) below by a more complicated system of tensor equations if the losses are small enough to allow them to be considered perturbations of the loss-free equation. That this is usually true may be seen in the work of Kamrn [61]. 126

PAGE 137

127 Examination of Kreuzer's results shows that the transition from OAS to TROAS is remarkably simple as indeed it should be. Hence, the remainder of this Appendix will contain a brief resume of Kreuzer's results [63]. Consult the original reference for further details. Note that it would be difficult to obtain similar results if the technique employed by Wrobel [47] were the only means available for solving the equations of state and conservation of mass, energy, and momentum. Conversely, it would be difficult to obtain the radiationless decay rates and quantum yields if only the method of Kreuzer were used. Let p = p(r,t) be the pressure, H = H(7,t) the source term, c the speed of sound, and y = C /C y the heat capacity ratio. Then the fundamental equation describing pressure changes induced by released heat H in a closed sample cell is [1,2] v 2 p c" 2 i-f = -[(y-D/c 2 ] U . (Al.l) The time Fourier Transform of both sides of (Al.l) yields (7 2 + J/z 2 ) p (7,a>) = [( T -l)/c 2 ]iu H(? )U ) (A1.2) where p(r,t) = /p(r,w) e" 1wt dw (A1.3! an d H(r,t) = /H(r,o)) e~ i(1)t d u (A1.4] The transform cf the acoustic velocity v(r,t) is v(r,oj) = (iwpj" 1 7p (r,co) . (A1.5;

PAGE 138

128 We now apply the boundary condition that the component of vp normal to the cell walls vanish at the walls. This condition determines the normal mode solutions p. of the homogeneous wave equation (A1.6) corresponding to (A1.2) above (v 2 + k. 2 ) Pj (r) = (A1.6) These acoustic modes are orthogonal and normalized by the following equation /p* Pj dv = V c 5.. (A1.7) where V c is the cell volume. For a cylindrical sample cell of length L and radius A, the general solution (ignoring azimuthal modes) is _ -ito.t Pj(r,t) = P. J Q (k r r) cos (k z z) e J (A1.8) where k 2 = irp /L for n z = 1,2,3,..., and k r = B] /A for «-, n the n-th root of J,(r). The acoustic resonant frequency w. (in radians per second) is *j = c(k z 2 + k r 2 ) 1/2 (Ai.g; Note that to. = 2wf, where f • is the resonant frequency in Hz. J J J The acoustic pressure p(F,u) is thus p(r,w) = e A.(u) p.(r,u>.) (Al.io; j j j j

PAGE 139

129 Substitution of (A1.8) and (ALIO) into (Al.l) using (A1.6), (A1.7) yields the acoustic mode amplitudes A. (id) _i [(y-D/VJ /p*(r, u ) H(r, w ) dV Aj(«) -*$ ^—^ (Al.ll) J J The amplitudes are infinite at oj = a>, because ajj_ energy loss mechanisms were neglected in the derivation of (Al.l); see references [63,111] for details. The two major energy loss processes are viscosity and thermal conductivity [61]. They may be introduced into the theory either at the beginning or mucn more conveniently as perturbations of the already obtained loss-free solution of (Al.l). This procedure will be outlined below; further details may be found in the original source [63]. The mode quality factor Q^ is defined as where the mode energy E. is Ej " lA/i V c /p q c 2 (A1.13) and L , L • are surface and volumetric losses primarily due to viscosity and thermal conductivity. The volumetric loss of the j-th mode is L vi = -A [(y-1)(K/2C J + (2n/3)] V A. 2 (Al .14) J p c M L J where K is the thermal conductivity, C is the heat capacity at P constant pressure, and n is the coefficient of viscosity. The volumetric

PAGE 140

130 losses are described in further detail by Morse and Ingard [111] and by Kamm [61]. They are, however, usually small relative to the surface losses. The surface loss may be considered to take place in surface layers of thickness \ 2 = 2 n /wp (Al .15; £ h 2 = 2k/u (Al .16) where z y is the viscosity loss thickness, i^ is the thermal loss thickness, and the Prandtl number of the sample gas mixture is 9 Prandtl number = (i / 1.) . ( Al .17) The surface loss is then L sj ' lAji 2 /[1/2 R v |V tj | 2 + 1/2 R n | P .| 2 ] dS (A1.18; where R y = { m p Q /Z) ]/Z , and % = [(y-1)/p c 2 ] (W2; and |v t -| = the acoustic velocity component tangential to the walls that would exist in the loss-free case. The surface losses may be qualitatively understood as follows. Near the cell walls, gas expansion and contraction are isothermal. Far from the walls, they are approximately adiabatic. Thermal losses occur in the transition region from isothermal to adiabatic behavior. Viscosity losses occur because of the requirement that the acoustic velocity component tangential to the

PAGE 141

131 walls vanish at the walls rather than be as described by equation (A1.5) Further details are given by Kreuzer [63], Kamm [61], and Morse and Ingard [66]. The Q. may be measured experimentally with good accuracy by pulse excitation of the u>, acoustic modes noting that the damping is described by [65] A.(o).,t) -oo.t/2Q. A^TOT = e J J (A1.19) The corrected mode amplitudes are thus _ iM [(y-D/VJ /p* (F fU ) H(r,a>) dV A,(u) = -1* % i (A1.20) Given the quality factors Q, , it is also possible to give an expression for the power spectrum of the noise that results from acoustic mode excitation by the thermal fluctuations in the sample and carrier gases. This is ,2 W 2 kT iA jn (»)i — -^ — =-2 r (A1 21) V c Uj Q. [(1-coVo.. 2 ) + ( tt / Uj Q.) 2 ] where the average energy per mode is kT and the noise energy in a frequency band of width Aw centered at u is {A. (u) | Aw/pc 2 . Thus, noise energy may be concentrated around the acoustic mode frequencies by increasing the Q. . The average energy per mode (kT) is independent of Qj. This thermal noise determines the fundamental sensitivity limit of OAS and TROAS experiments.

PAGE 142

APPENDIX TWO TROAS NOISE EQUIVALENT POWER In this appendix an expression for noise equivalent power is derived. The treatment is a simple modification of the derivation of noise equivalent power (NEP) given for conventional optoacoustic spectroscopy by Kreuzer [63]. Let the pulsed light source uniformly illuminate the sample cell such that the assumption H = al is satisfied. Then the corrected mode amplitude equation (Al .20) gives the desired signal amplitude A ( w ) ••) M(1 Vii;IV <*-i: The light pulse of energy E and duration t has intensity I given by 1 if ( A2 -2) c p where L/V c is the cell cross-sectional area. Thus, for TROAS (with wx T » 1), equations (A2.1) and (A2.2) yield the following a ow 'ixV L (A2 3) Define NEP by NEP = a L E/ t p (A2.4; 132

PAGE 143

133 Note that for TROAS, co > w . for most acoustic modes. Thus, the noise power spectrum (A1.21) becomes 2 . T 3 o 4o n c k T uj. |A jn U)| 2 ^ . (A2.5) V c Q „ Then (A2.3), (A2.4), (A2.5) imply 2 2 . 2, , 3 P V u 4p n c kT _ (NEP)' 1 = -^— T z —9 ^-J_ . (A2 . 5 ) :y-D j V c Q.. co Hence, we obtain « 4p n c 2 kT V lo. 3 (NEP) 2 = Z —^ ^ £-_Jj (y-IT Q.
PAGE 144

APPENDIX THREE CAPACITANCE MICROPHONE PREAMPLIFIER NOISE MODEL The most common method of capacitance microphone preamplifi cation is the voltage amplifier circuit. For optimum results, the amplifier should provide gain in the first stage; otherwise, the signal to noise ratio will be degraded. A typical circuit is shown in Figure 21. Noise sources for this circuit are included in Figure 22. The calculation of total noise voltage follows that of Kreuzer [63] closely though several additional sources were used to extend the model as necessary [113], [114], [74], [73]. Noise sources in the circuit are due to microphone acoustic noise, bias resistor thermal noise, input resistor noise, feedback loop resistor noise, amplifier input current noise, and amplifier input voltage noise. The square of the noise voltage is the sum of six terms given by " |e n | + |1 n f UC + R d ] ) e p + 4 kTR^ + li j 2 R. t ' n ' f + (4kT/R p )uc e+ R -y 2 + 4kTC ' [(( M / M ) 2 1) 2 Q m % C e 2 + (o)/o) m Q n] ) 2 ]" 1 (A3.1) 134

PAGE 145

o =3 CU CJ CU II S_ S_ — -C CU uhq: CU X » C !4O T3 Q. Q. CU O O Q.I— sa.
PAGE 146

136

PAGE 147

a; ^ c\j O r— c fO O) (T3 > s+-> 3 rr— Ol U ,•!I— •i-a > CU <— i— (_) ra I— i> fO 3 in a) -E 0) 4-> (T3 O) i. to . •iS_ ~ 4Q.O rON r— S_ Q. Q. II E a. « fO I Q.4->

PAGE 148

138

PAGE 149

139 The first term is the amplifier input voltage noise. The second is amplifier input current noise while the third is thermal noise due to the feedback resistors. The fourth term is amplifier input current noise at the inverting input. Term five is thermal noise due to the parallel combination of the bias and input resistors. The sixth term is the microphone acoustic noise due to Brownian motion. The restoring force, mass, damping, and output voltage of a capacitance microphone model are [63] C ' = C m 2v B 2/k m d2 (A3-2) L = m/k m C = l/. m 2 C (A3. 3) R = 5/k m C = l/Q m . m C (A3.4) V G = (Ad/V B CM)p m (A3. 5) where m is the diaphragm mass, 5 is the damping, C is the microphone capacitance, k m is the restoring force, Vg is the bias voltage, d is the electrode spacing, ^ m is the resonant frequency, A is the microphone area, and p^ is the diaphragm acoustic pressure. In addition, the following relations hold R B >y ^C)" 1 > *
PAGE 150

140 where u>, is the lower frequency limit. Typical values for C . C . L and R are 40 pf, 5 pf, 70 H, and 5 Mohm. Realistic values for R e and Rg are 10 3 to 10 ohms with 10 10 ohms a good choice. The input noise values for a typical low noise bipolar input 725 op amp are e n = 8. x 10" 9 VW Z V1 , i n 1 s 1.5 x 10" 13 A/H. 1/2 For the JFET input AD514 op amp the values are 1/2 13 •15 n = 30. x 10 3 V/H z ,/C , i n = i n s io1J io" n A/H 1/2 Calculations indicate that the AD514 is much less noisy than the 725 for source impedances above several hundred thousand ohms.

PAGE 151

APPENDIX FOUR MICROCOMPUTER PROGRAM AND I/O LOCATIONS The last version of the microcomputer programed used in the TROAS research is listed below. TROAS-7 200 DIM A%(256) ,S(256) 300 DATA A2, 07, A9, 00, 8D.4C, 03, 4E,4B 310 DATA 03,2E,4C,03,CA,D0,F7,60 312 DATA 00,00,A9,08,8D,00,A0,CE 314 DATA 5B, 03, DO, FB,EE, 00, AO, 60 320 FOR J=326 TO 858:G0SUB 900:P0KE J,C:NEXT J 330 POKE 1, 53: POKE 2,3 500 ? " ENTER DWELL TIME/CHANNEL IN SECONDS" -.INPUT U 502 ? " ENTER SENSITIVITY & GAIN": INPUT N5.N3 510 FOR 1=1 TO 2:G0SUB 1000:NEXT I 520 ? " SELECT OFFSET, GROUND INPUT, EXT. TRIGGER" : INPUT I.-GOSUB 2000:0F=T 530 ? " 0FFSET=";OF:? 540 ? " UNGROUND INPUT, SELECT TRIGGER" :?: INPUT I:? 510 ? " INPUT OPTION?" 620 INPUT V:ON V GOSUB 1000 ,2000 ,3000 ,4000 ,5000,6000,7000 , 8000,9000: GOTO 610 700 S=0 710 FOR 1=1 TO 256 141

PAGE 152

142 720 C=PEEK(4.5E4) : IF C<128 GOTO 720 730 A%(I)=PEEK(4.4E4):S=S+A%(I) 740 POKE 4.5E4,54:P0KE 4.5E4,62:NEXT I 750 T=S/256: RETURN 800 FOR 1=1 TO 256 810 IF A%(I)>255 THEN A%(I)=0 820 IF A%(I)<0 THEN A%(I)=0 830 POKE 4.8E4,255-A%(I) 840 GOSUB 890:NEXT I: RETURN 850 FOR 1=1 TO 255 STEP 2 860 IF A%(I)>255 THEN A%(I)=255 865 IF A%(I)<0 THEN A%(I)=0 870 POKE 4.3E4,255-A%(I) 880 GOSUB 390:P0KE 4.8E4,255:G0SUB 890:G0SUB 390:G0SUB 890 NEXT I: RETURN 390 FOR J=l TO 600: NEXT J: RETURN 900 READ GS:C=0 910 FOR 1=1 TO 2 920 F$=MID$(G$,I,1):A=ASC(F$)-64 930 IF A>0 THEN A=A+9 940 IF A<0 THEN A=VAL(F$) 950 C=C+A*(31-15*I):NEXT I: RETURN 1000 ?:? " PIA I/O PORT STATUS":? 1010 ?"PIA$#","$9","$A","$B":?:?:? 1020 POKE 4.0E4,0:P0KE 4.5E4,0:POKE 4.9E4,0 1030 POKE 3.9E4,127:P0KE 4.4E4,0:P0KE 4.8E4,255 1040 POKE 4.0E4,4:P0KE 4.5E4,62 :POKE 4.9E4,4

PAGE 153

143 1050 POKE 3.9E4,127:P0KE 4.8E4,0 1060 ? "CRB",PEEK(4.0E4),PEEK(4.5E4),PEEK(4.9E4):?:? 1070 ?,"TR & SM'Y'TR DATA" ,"D/A" : ? 1080 ? " B",PEEK(3.9E4),PEEK(4.4E4),PEEK(4.8E4):?:?:? 1090 POKE 3.8E4,0:P0KE 4.2E4,0:P0KE 4.7E4,0 1100 POKE 3.7E4,0:P0KE 4. 1 E4.255 :POKE 4.6E4,0 1110 POKE 3.8E4,4:P0KE 4.2E4,4:P0KE 4.7E4,4:P0KE 4.7E4.14 1120 ? "CRA",PEEK(3.3E4),PEEK(4.2E4),PEEK(4.7E4):?:? 1130 ?,"A/D","SYS-C0NT" 5 ll FREE":? "1 140 ? " A\PEEK(3.7E4),PEEK(4.1E4),PEEK(4.6E4):?:?:RETURN 2000 POKE 3.9E4,63:P0KE 3.9E4,127:A=PEEK(4. 1 E4) :P0KE 4.1E4.A-5 POKE 4.1E4.A-4 2005 GOSUB 3030 2010 B=PEEK(3.9E4):IF B>127 GOTO 2010 2020 POKE 4.1E4,2:LP=PEEK(3.7E4)-]28 2024 IF LP<10 THEN LP=1 2026 ?:? " LASER PULSE=";LP:? 2030 POKE 4.1E4,A:P0KE 3.9E4,111:? " ACQUIRE?":? 2040 INPUT N: ?: IF N=0 GOTO 2060 2050 GOSUB 700:G0SUB 2070 2060 POKE 3. 9E4, 127: RETURN 2070 06=255:07=0 2080 FOR 1=1 TO 256 2090 IF A35(I)>07 THEN 07=A%(I) 2100 IF A%(I)<06 THEN 06=A%(I) 2110 NEXT 1:08=07 -06 :N4=(08/255)*N5/N3 2120 ? " SIGNAL VOLTAGE-" ;INT( .5+1 E6*N4) ; "MICROVOLTS" :?

PAGE 154

144 2130 ? " NORMED SIGNAL=" ; INT( . 5+( 1 E6*N4/LP ) ) ; "MICROVOLTS" : ?: RETURN 3000 POKE 4.3E4,0:G0SUB 3030:P0KE 4.3E4,255:G0SUB 3030 3010 GOSUB 800 3020 POKE 4.8E4,255:G0SUB 3030:P0KE 4.8E4,0:G0SUB 3030:RETURN 3030 FOR 1=1 TO 3000: NEXT I: RETURN 4000 FOR 1=1 TO 4 4010 OPEN I,1,1,"TR0AS DATA" 4020 FOR J=l TO 54:Z=J+64*( 1-1 ) 4030 PRINT#I,A%(Z) 4040 NEXT J 4050 CLOSE I:NEXT I: RETURN 5000 ?:? " AVERAGE SUBTRACTED?" :?: INPUT W:X=T-OF 5005 FOR J=l TO 128: J2=2*J :M=J-1 :P0KE843,M:X=USR(0) : I=PEEK(844; L=2*I 5010 S(J2-l)=A%(L+l)-0F-W*X:S(J2)=A%(L+2)-0F-W*X:NEXT J 5020 N=128:N2=7:IL=2:LE=1 :JE=N/2 5030 FOR K=l TO N2:AR=u/LE 5040 FOR J=l TO JE:J1=J-1 5050 FOR L=i TO LE:L1=L-1 : JP=2*(IL*J1+L1 ) + l : JQ=JP+IL:AX=L1*AR 5060 WR=COS(AX):WI=-SIN(AX) 507C RE=S(JQ)*WR-S(JQ+1)*WI:FI*S(JQ)*WI+S(JQ+L)*WR 5080 S(JQ)=S(JP)=RE:S(JQ+L)=S(JP+1)-FI:S(JP)=S(JP)+RE:S(JP+1)= S(JP+1)+FI:NEXT L 5090 NEXT J 5100 LE=2*LE:JE=JE/2:IL=2*IL:NEXT K 5110 Nl=N/2

PAGE 155

145 5120 FOR M=2 TO Nl :M1=2*(N-M)+3:M2=2*M-1 :MB=M1+1 :MC=M2+1 5130 DR=(S(MC)+S(MB))/2:DI=(S(Ml)-S(M2))/2 5140 CR=(S(M2)+S(Ml))/2:CI=(S(MC)-S(MB))/2 5150 S(M1)=DR:S(MB)=DI:S(M2)=CR:S(MC)=CI:NEXT M:AR=^/N 5160 FOR K=2 TO Nl :AX=AR*(K-1 ) :WR=COS(AX) :WI=-SIN(AX) :KN=2* (N-K)+3:KI=2*K-1 5170 RE=S(KN)*WR-S(KN+1)*WI:FI=S(KN)*WI+S(KN+1)*WR 5180 BR=S(KI)+RE:BI=S(KI+1)+FI:H1=SQR(BR*BR+BI*BI): BR=S(KI)-RE:BI=S(KI+1)-FI 5190 S(KI)=H1:S(KN)=SQR(BR*BR+BI*BI):NEXT K 5200 S(129)=SQR(S(129)*S(129)+S(130)*S(130)):01=1E4:02=-1E4 5210 FOR 1=1 to 255 STEP 2 5220 IF S(I)<01 THEN 01*S(I) 5230 IF S(I)>02 THEN 02=S(I) 5240 NEXT 1:03=02-01 5250 FOR 1=1 TO 255 STEP 2 5260 A%(I)=INT(.5+255*(S(I)-01)/03):NEXT I 5270 ?:?:" MAX. AMPLITUDE=" ;INT( . 5+03) : ?:RETURN 6000 ?:? " ENTER LOWER THRESHOLD IN %":?: INPUT 1:1 6010 ? " FREQUENCIES ABOVE THRESHOLD":? 6020 ? TAB(3);"HERTZ";TAB(18);"INTENSITY";TAB(33);"P0WER" 6030 Z1=2.55*Z:U1=1/(510*U):G1=0 6035 FOR 1=1 TO 255 STEP 2 : G1=G1+A%( I ) :NEXT I 5040 FOR 1=1 TO 255 STEP 2 5050 IF A%(I)>Z1 THEN GOSUB 6070 6060 NEXT I: RETURN

PAGE 156

146 6070 U2=INT(.5+I*U1):Z2=.1*INT(A%(I)*3.922):Z3=.1*INT(.5+1E3* A%(I)/G1) 6080 ?:? TAB(3);U2;TAB(18);Z2;TAB(33);Z3;RETURN 7000 POKE 4.3E4,0:G0SUB 3030:P0KE 4.8E4,255:G0SUB 3030 7010 GOSUB 850 7020 POKE 4.3E4,255:G0SUB 3030:P0KE 4.8E4,0:G0SUB 3030:RETURN 8000 POKE 1,77:P0KE 2,3:P0KE 3 . 9E4 ,63 : POKE 3.9E4J27 :A=PEEK(4. 1E4) 801 ? " INPUT TR DELAY":INPUT I: POKE 859 , I : J=USR(0 ) : POKE 1,58: POKE 2,3 8020 POKE 4.1E4,A-4:G0T0 2010 9000 ? "INPUT # OF POINTS TO OMIT": INPUT NN:SC=256-NN 9005 S1=0:S2=0:S3=0:S4=0:S5=0 9010 FOR I=NN+1 TO 256 9020 S1=S1+I*U:S2=S2+I*U*I*U:S3=S3+L0G(A%(I)) : S4=S4+L0G( A%( I ) ) * LOG(A%(I)) 9030 S5=S5+I*U*L0G(A%(I)) 9040 NEXT I 9050 SA=S5-S1*S1/SC:SB=S2-S1*S1/SC 9060 S6=SA/SB 9070 SA=S3/SC:SB=S6*S1/SC:S7=EXP(SA-SB) 9080 SA=(S5-S1*S3/SC)+2:SB=S2=S1*S1/SC:SD=S4-S3*S3/SC 9090 S8=SA/(SB+SD) 9100 ? "Y=A EXP(BX)" 9110 ? " A=";S7 9120 ? " B=";S6;"SEC0NDS" 9130 ? " R 2";S8 9140 RETURN

PAGE 157

147 Remarks have been omitted from the above program to conserve RAM memory space which was at a premium. Several points of nomenclature: semicolons separate lines; question marks are seen as abbreviations of the print command; and dollar signs denote hexidecimal notation. The main program and utility subroutines lie below line number 1000. The nine main subroutines are positioned in blocks of 1000 line numbers starting at line number 1000. In lieu of remarks, a brief description of the program is presented. The data statements (300-314) contain simple assembly language subroutines used by the FFT subroutine and the transient recorder trigger subroutine. Lines 320 and 330 place the assembly language program in memory (in the PET's unused, protected, second tape buffer RAM), while lines 500-620 request information concerning transient recorder settings, et cetera, and request the input (main subroutine) option. Lines 700750 acquire 256 data points from the transient recorder and also average them. Lines 800-840 check for illegal data and then output to the D/A driving the chart recorder. Lines 850-380 output the FFT spectrum as a line spectrum while line 890 is a delay subroutine. Lines 900-950 convert the hexidecimal op codes in the data statements to decimal (required by the POKE command). The locations in memory space of the D/A, A/D, transient recorder control lines, et cetera are shown in Table 9. Subroutine 1, at lines 1000-1140 initializes the PIA chips and thereby initializes the transient recorder, chart recorder output, et cetera. The status of the various initialized subsystems is displayed on the built-in video monitor. Subroutine 2, at lines 2000-2130, triggers the laser and transient recorder, reads the photodetector peak output, determines whether the

PAGE 158

148 Table 9. Peripheral systems memory locations 4.2E4

PAGE 159

149 data need be acquired, and outputs the peak-to-peak voltage of any acquired signal. The integer array A% is used throughout the program as a shared array. Subroutine 3, at lines 3000-3030, outputs chart recorder data. Subroutine 4, at lines 4000-4050, allows data to be stored as named files on cassette tape. Subroutine 5, at lines 5000-5270, is the FFT routine. Lines 5000-5010 perform the necessary binary bit inversion required by the Cooley-Tukey algorithm. Lines 5020-5100 compute the complex transform of the sorted real data. Lines 5110-5190 correct the result for the real nature of the data, while line 5200 computes a point missed in the clash of indices. Lines 5210-5270 scale the floating point FFT so that it may be output to the chart recorder and also output the unsealed maximum peak-to-peak amplitude. Further details may be found in [77]. Subroutine 6, at lines 6000-6080, displays the frequencies,, normalized intensities, and powers of spectral peaks in the FFT above a specified intensity. Subroutine 7, at lines 7000-7020, outputs the FFT spectrum to the chart recorder. Subroutine 3, at lines 8000-8020, calls the assembly language routine that triggers the transient recorder. Because of delays in the optocoupled laser trigger, the transient recorder, at dwell times per channel of 5 y sec or less, may finish recording before the laser has fired. Therefore, the laser is triggered first, then a delay loop is entered, and finally the transient recorder is triggered (immediately prior to the actual firing of the laser). Subroutine 9, at 9000-9140, is a simple least squares fit to an exponential curve with provision for omitting an arbitrary initial number of points (the expected step-rise, for example). The subroutine outputs the decay time in seconds.

PAGE 160

REFERENCES 1. A. G. Bell, Philos. Mag., VJ_, 510(1831) 2. E. Mercadier, C. R. Acad. Sci., 9T_, 929(1880) 3. W. H. Preece, Proc. Roy. Soc. (London), 3]_, 506(1881) 4. W. C. Rontgen, Ann. Phys. Chem. , 12, 155(1881) 5. J. Tyndall, Proc. Roy. Soc. (London), 31_, 307(1881) 6. M. L. Veingerov, Dokl . Akad. Nauk. SSSR, ]_9, 687(1938) 7. A. H. Pfund, Science, 90, 326(1939) 3. K. F. Luft, Z. Tech. Phys., 24, 97(1943) 9. M. J. E. Golay, Rev. Sci. Instrum. , 20, 816(1949) 10. W. D. Hershberger, E. T. Bush, and G. W. Leek, RCA Rev., 7, 442(1946) 11. G. Gorelik, Dokl. Akad. Nauk. SSSR, 54, 779(1946) 12. R. Kaiser, Can. J. Phys., 37, 1499(1959) 13. M. E. Delaney, Sci. Prog. (Oxford), 47, 459(1959) 14. Y. I. Gerlovin, Opt. Spectry. , 7_, 352(1959) 15. R. M. White, J. Appl. Phys., 34, 2123 & 3559(1963) 16. C. M. Percival, J. Appl. Phys., 38, 5313(1967) 17. J. C. Bushnell and D. J. McCloskey, J. Appl. Phys., 39, 554(1968! 18. E. Hey, Ph.D. Thesis, University of Heidelberg (1967) 19. P. G. Seybold, M. Gouterman, and J. Callis, Photochem. Photobiol. 9, 229(1969) 20. J. B. Callis, M. Gouterman, and J. D. S. Danielson, Rev. Sci. Instrum. , 40, 1599(1969) 21. L. B. Kreuzer, J. Appl. Phys., 42, 2934(1971) 150

PAGE 162

152 42. A. Rosencwaig, Science, J_8J_, 657(1975) 43. A. Rosencwaig, Anal. Chem. , 4_7_, 592A(1975) 44. A. Rosencwaig and A. Gersho, J. Appl . Phys., 47, 64(1976) 45. F. A. McDonald and G. C. Wetsel , Jr., J. Appl. Phys., 49(4), 2313(1978) — 46. M. J. Brienza and A. J. DeMaria, Appl. Phys. Letters, 1_1_, 44(1967 47. J. J. Wrobel , Ph.D. Dissertation, University of Florida, 1976 48. P. G. Bowers and G. Porter, Proc. Roy. Soc. (London), 299A, 348(1967) 49. A. A. Lamola and G. S. Hammond, J. Chem. Phys., 43, 2129(1965) 50. C. A. Parker and T. A. Joyce, Trans. Faraday Soc, 62, 2785(1966) A. R. Horrocks and F. Wilkinson, Proc. Roy. Soc. (London), 306A, 257(1968) 51 52. M. Gueron, J. Eisinger, and R. G. Shulman, Mol . Phys., 1_4, 111(1968; 53. C. A. Parker, Photol uminescence of Solu tions, Elsevier, Amsterdam, 1968 ' 54. J. Birks, Photophysics of Aromatic Mole cules, Wiley-Interscience, 1970 55. B. Soep, A. Kellmann, M. Martin, and L. Lindqvist, Chem. Phys. Letters, U, 241(1972) 56. K. H. Drexhage, in Dye Lasers (Topics in Applied Physics, Vol. 1), Springer-Verlag, N. Y., Berlin, Heidelberg, 1973 57. S. P. McGlynn, T. Azumi , and M. Kinoshita, Molecular Spectroscopy of the Triplet State . Prentice Hall, Inc., Englewood Cliffs, M. J., 1969 58. Optoacoustic Spectroscopy and Detection , Y.-H. Pao, editor, Academic Press, Inc., N. Y., 1977 59. T. Aoki and M. Katayama, Japan. J. Appl. Phys., J0_, 1303(1971) 60. F. R. Grabiner, D. R. Siebert, and G. W. Flvnn, Chem. Phys. Letters, ]7_(2), 189(1972) 61. R. D. Kamm, J. Appl. Phys., 47(8), 3550(1976) 62. J. N. Bradley, Shock Waves in Chemistry and Physics , J. Wiley and Sons, Inc., N. Y., London, 1962

PAGE 163

153 63. L. Kreuzer, in Qptoacoustic Spectroscop y and Detection, Academic Press, Inc., N. Y., 1977 ~ 64. A. B. Carlson, Communication Systems, McGraw-Hill, Inc., N Y , 1975 65. C. F. Dewey, Jr., in Qptoacoustic Spectroscopy and Detection , Academic Press, Inc. , N. Y. , 1977 ' 66. P. M. Morse and K. V. Ingard, Theoretica l Acoustics, McGraw-Hill, Inc., N. Y., 1968 67. F. W. Fraim and P. V. Murphy, J. Acous. Soc. of Am., 53(6), 1601(1973) — 68. G. M. Sessler and J. E. West, J. Acous. Soc. Am., 53_, 1589(1973) 69. M. M. Farrow, R. K. Burnham, M. Auzanneau, S. L. Olsen, N. Purdie, and E. M. Eyring, Applied Optics, ]]_{!), 1093(1978) 70. Low noise preamplifiers selection guide, Ithaco, Inc., Ithaca, N. Y., Publication number IPS 112 9/76 71. J. J. Zaalberg van Zelst, Philips Tech. Rev., 9(12), 357(1947/48) 72. Precision condenser microphone data sheets, Dynasciences Corp., Chatsworth, Calif., Models 404, 414, 504, 514, 314. 73. FET design ideas and selection guide, Texas Instruments, Inc., Houston, Texas 74. H. W. Ott, Noise Reduction Techniques in Electronic Systems , Wiley-Interscience, N. 77, 1976 75. S. M. Sze, Physics of Semiconductor De vices, Wiley-Interscience, N. Y., 1969 — 76. Radio Shack capacitance electret microphone #33-1056 data sheet 77. J. R. Bell, Introductory Fourier Transf orm Spectroscopy, Academic Press, Inc., N. Y., 1972 ' 78. R. Spafford, J. Baiardo, J. Wrobel , and M. Vala, J. Am. Chem. Soc. 98, 5217(1976) 79. J. Baiardo, R. Spafford, and M. Vala, J. Am. Chem. Soc. 98, 5225(1976) — 80. J. P. Baiardo, private communication 81. D. H. Powell, private communication 32. J. Herst, private communication

PAGE 164

154 83. E. U. Elam and H. E. Davis, J. Organic Chem. , 32, 1562(1967) 84. W. Siebrand, in The Triplet State , Beirut Symposium 1967, Cambridge U. Press, Cambridge, England, 1967 35. 87 90. 91. 92. 93. CRC Handbook of Chemistry and Physics, 44th Edition, 1961 J. Calvert and J. Pitts, Jr., Photochemi stry, J. Wiley and Sons, Inc., N. Y., 1966 M. B. Robin, in Optoacoustic Spectroscopy and Detec tion, Academic Press, Inc., N. Y., 1977 H. W. Sidebottom, C. C. Badcock, J. C. Calvert, B. R. Rabe, and E. K. Damon, J. Am. Chem. Soc, 94(1), 13(1972) E. Drent, R. P. van der Werf, and J. Kommandeur, J. Chem. Phys., 59, 2061(1973) W. A. Noyes, Jr., G. B. Porter, and J. E. Jolly, Chem. Rev., 56, 49(1956) — R. 3. Cundall and A. S. Davies, Progr. React. Kinet. , 4, 149(1967) F. C. Henriques, Jr., and W. A. Noyes, Jr., J. Am. Chem. Soc, 62, 1038(1940) H. Okabe and W. A. Noyes, Jr., J. Am. Chem. Soc, 79, 801(1957) 94. H. Ishikawa and W. A. Noyes, Jr., J. Chem. Phys., 37, 583(1962) 95. C. S. Parmenter and B. L. Ring, J. Chem. Phys., 46, 1998(1967) C. S. Parmenter and H. M. Poland, J. Chem. Phys., 51_(4), 1551(1969) C. C. Badcock, H, W. Sidebottom, J. C. Calvert, B. R. Rabe, and E. K. Damon, J. Am. Chem. Soc, 94(1), 19(1972) H. W. Furumoto and H. L. Ceccon, Appl . Optics, 8, 1613(1969) H. W. Furumoto and H. L. Ceccon, J. Appl. Phys., 40, 4204(1969) 100. H. W. Furumoto and H. L. Ceccon, IEEE J. Quant. Electr. QE-6, 262(1970) 101. F. P. Schafer, in Dye Lasers (Topics in Applied Physics, Vol. 1), Springer-Verlag, N. Y., Berlin, Heidelberg, 1973 102. T. F. Deaton, D. A. Depatie, and T. W. Walker, A-pl. Phys Letters, 26, 300(1975) 103. A. Hordvik and H. Schlossberg, Applied Optics, 160), 101(1977) 96 97 98. 99.

PAGE 165

155 104. W. D. Stanley and S. J. Peterson, Byte, 3(12), 14(1978) 105. D. M. Munroe and H. S. Reichard, American Laboratory, 119, 2/77 106. A. Rosencwaig, Physics Today, 23(9/75) 107. L. C. Aamodt, J. C. Murphy, and J. G. Parker, J. Appl. Phys., 48, 927(1977) — 108. T. G. Muir, C. R. Culbertson, and J. R. Clynch, J. Acous. Soc. Am., 59, 735(1976) 109. J. F. McClelland and R. N. Kniseley, Applied Optics, 15(12), 2967(1976) ~ 110. H. H. Theising and P. J. Caplan, Spectroscopic Calculations for a Multielectron Ion , Interscience Publishers, N. Y., 1966 111. P. M. Morse and K. U. Ingard, in Encyclopedia of Physics (S. Flugae editor), Vol. Xl/1, Springer-Verlag, Berlin, N. Y., 1961 112. L. C. Aamodt and J. C. Murphy, J. Appl. Phys., 49(6), 3036(1978) 113. W. G. Jung, IC Op-amp Cookbook , H. W. Sams and Co., Inc., Indianapolis, Ind. , 1974 114. J. Sherwin, in National Semiconductor's Linear Applications Handbook 2 , N. S., Santa Clara, Calif., T976

PAGE 166

BIOGRAPHICAL SKETCH Edward G. Voigtman, Jr., was born on December 26, 1949, in St. Louis, Missouri. Family instabilities led to his attending nine schools before graduating from Parsippany High School in June 1968. In September 1968 he entered Rensselaer Polytechnic Institute with the intention of majoring in inorganic chemistry. In 1970, he realized that, in order to take up the slack, he would have to intensify his self-teaching efforts, and he also realized that specialization is usually a quick route to stultification. In June 1972, he graduated from R. P. I. with a B. S. degree in chemistry. In September 1972, not knowing any better, he accepted a Graduate School Fellowship to the University of Florida. He has been pursuing his studies at the University of Florida since then. In June 1973, while visiting his friend Richard Lawrence in South Bend, Indiana, he met Janiece Lee Leach. "They were married on May 31, 1975. 156

PAGE 167

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the deqree of Doctor of Philosophy. M. T. Vala, Chairman Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the deqree of Doctor of Philosophy. E. D. Adams " Professor of Physics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully aaequate, in scope and quality, as a dissertation for the deqree of Doctor of Philosophy. Professor of Electrical Engineering I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. kCll/lbn D. C. Wilson " Associate Professor of Mathematics

PAGE 168

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Cx.K £j2 c-v J. R.\ Eyler Assistant Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. C ^ : /fO? S. 0. Colgate y^ Associate Professor of Chemistry This dissertation was submitted to the Graduate Faculty of the Department of Chemistry in the College of Liberal Arts and Sciences and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. March 1979 Dean, Graduate School

PAGE 169

UNIVERSITY OF FLORIDA 3 1262 08553 1084


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ERLE44EQ7_USP2XK INGEST_TIME 2017-07-17T20:21:43Z PACKAGE UF00099117_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES