Title: Multielement atomic fluorescence flame spectrometry
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00098667/00001
 Material Information
Title: Multielement atomic fluorescence flame spectrometry evaluation of a pulsed EIMAC Continuum source and characterization of background molecular fluorescence
Alternate Title: Atomic fluorescence flame spectrometry
Physical Description: vii, 84 leaves : ill. ; 28 cm.
Language: English
Creator: Fowler, William Kelly, 1948-
Publication Date: 1976
Copyright Date: 1976
Subject: Spectrum analysis   ( lcsh )
Fluorescence   ( lcsh )
Spectrometer   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 80-83.
General Note: Typescript.
General Note: Vita.
Statement of Responsibility: by William Kelly Fowler.
 Record Information
Bibliographic ID: UF00098667
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000184717
oclc - 03319218
notis - AAV1292


This item has the following downloads:

PDF ( 3 MBs ) ( PDF )

Full Text







This work is dedicated to my parents, Mr. William B.

Fowler and Mrs. Josephine H. Fowler, who have inspired

me to some degree by filial instruction but far more

profoundly by living example, and whose moral and finan-

cial support were vital to my success as a graduate student.


The author hereby acknowledges the valuable assistance

of each of the many students, faculty, and staff members

who have labored in his behalf and expresses his regret that

the names of these people are too numerous to set forth

here. However, a special word of thanks must be included

for Dr. James D. Winefordner, whose assistance and encourage-

ment have been of special significance to the author during

the years of his graduate studies.


BIOGRAPHICAL SKETCH .... ...........



ACKNOWLEDGMENTS ......... .....









BACKGROUND ..... .....



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



William Kelly Fowler

December, 1976

Chairman: James D. Winefordner
Major Department: Chemistry

An Eimac xenon arc continuum source operated in the

pulsed mode has been evaluated for potential use as a source

of excitation in atomic fluorescence spectrometry. Limits

of detection, precision data, and analytical curves were

determined for several elements in the argon separated air-

acetylene flame, and the limits of detection are compared

to those obtained with a similar lamp operated in the cw

mode. Little or no improvement is evident from this compari-

son; the lamp was found to be relatively unstable in the

pulsed mode, requiring the use of a rather large residual

dc current in conjunction with a relatively low-amplitude

power pulse in order to sustain the arc. Moreover, the

pulsed source yielded a large background signal due to

scattering of incident source radiation in the flame, and

the background fluorescence spectrum contained numerous

large peaks due to fluorescence of molecular species in

the flame and/or to prominent peaks in the spectral dis-

tribution of the pulsed lamp output, transferred to the

signal detector via scatter.

The numerous peaks observed in the fluorescence back-

ground spectra of the pulsed source studies have been

evaluated separately. An attempt was made to identify

each of the observed peaks and bands, with the result that

molecular fluorescence bands were measured and assigned

to PO, NH, CN, and OH. In addition, peaks resulting from

the grating second order of PO fluorescence bands were

observed. The spectral distribution of the source radiation

was measured in both the pulsed and cw modes, but none of

the spectral irregularities detected in the flame fluores-

cence background spectra were found to correspond to

scatter peaks in the source spectral output. Background

fluorescence spectra were recorded under three different

analytically useful sets of flame conditions (fuel-lean,

stoichiometric fuel-rich), and spectra obtained in both the

pulsed and cw modes are compared. A commercially available

filter designed to remove PH3 from acetylene was evaluated

with respect to its ability to reduce the fluorescence

intensity of PO, a species formed in the flame from PH ,

present in the acetylene as an impurity. A general assess-

ment of molecular fluorescence in flames as a source of

interference in atomic fluorescence spectrometry is pre-


Because the flame fluorescence of PO was discovered to

be particularly intense and widely distributed in the spec-

trum, it has been further characterized in a separate study.

PO fluorescence signals generated by phosphate solutions

were compared to those produced only by the PH3 impurity in

the acetylene. The fluorescence and emission intensities

of PO were compared in various flames, and a silicon intensi-

fied target (SIT) tube in conjunction with an optical

multichannel analyzer (OMA) was evaluated for potential

application to the study of molecular fluorescence in flames.

The possibility of utilization of PO fluorescence for the

analysis of phosphorus was investigated; the analytical

precision and limit of detection for phosphorus determina-

tions were estimated from signals produced by standard

solutions of phosphoric acid nebulized into various flames.



The phenomenon of atomic fluorescence was first observed

in sodium vapor by Wood in 1904 (1) but was not seen in

flames until 1924, when Nichols and Howes (2) detected

the fluorescence of several elements in a Bunsen burner

flame. Atomic fluorescence subsequently attracted little

interest until 1962, when Alkemade (3) suggested its use

as an analytical technique. In 1964, Winefordner and

Vickers (4) published the first paper characterizing the

method with regard to theoretical principles and experi-

mental requirements, and since then, numerous studies

embracing virtually every aspect of atomic fluorescence

spectrometry have been published.

For the analysis of elements which are difficult to

excite in emission, atomic fluorescence spectrometry main-

tains several theoretical and practical advantages, with

regard to sensitivity, over the more conventional methods

of atomic absorption and atomic emission spectrometry (4-7).

The analytical sensitivity of atomic absorption spectrometry

is limited by the ability of the signal detection system

to respond measurably to small changes in light level at

the photocathode caused by the absorption process within

the sample cell. An increase in source intensity at the

wavelength of the spectral line may improve the detection

limits for the technique, provided the RMS noise level on

the observed signal is increased by a smaller factor;

however, stable, narrow-line, unreversed sources are re-

quired, unless a very high-quality, narrow spectral bandpass

monochromator is used. To date, only the moderately intense

hollow cathode discharge tubes meet these requirements to

an extent sufficient to allow atomic absorption spectrometry

to compete with atomic fluorescence spectrometry in terms

of detection limits for trace elemental analysis. Attempts

(other than pulsing of the source) to boost the output

intensity of these sources without seriously degrading

their most desirable characteristics and/or introducing

other undesirable features have largely failed (8,9).

On the other hand, source and monochromator require-

ments in atomic fluorescence spectrometry are considerably

less stringent; the monochromator "sees" only fluorescence,

not the excitation source radiation which produces it.

Accordingly, very intense, broad-line sources may be used

to advantage in this technique, because the intensity of

fluorescence is proportional to the source irradiance at

the wavelength of the excitation line. A variety of

excitation sources have been shown to be adaptable to

atomic fluorescence spectrometry (10,11), and several of

these have manifested considerable promise for augmentation

of their output intensities (12,13). Furthermore, it is

possible to employ a single continuum source for all

elements, while utilizing at the same time a relatively

inexpensive, low-dispersion monochromator in the signal

detection system. In practical applications of atomic

fluorescence spectrometry, improvements in signal-to-noise

ratio to be obtained from increases in excitation source

irradiance are eventually limited only by those types of

noise whose magnitudes are directly proportional to the

first power of the source irradiance.

Atomic fluorescence spectrometry has the advantage

over atomic emission spectrometry that the analytical

signal is directly proportional to the concentration of

atoms boosted to the excited state via radiational excita-

tion rather than to the concentration of atoms in the excited

state due to thermal excitation. Thus, the atomic fluores-

cence signal should be less dependent upon flame tempera-

ture, and the technique should before applicable to ele-

ments which are difficult to excite by thermal means.

Despite its potential advantages, atomic fluorescence

spectrometry suffers from a number of important fundamental

limitations. Of the three flame techniques for multi-

element analysis, atomic fluorescence spectrometry is

most affected by the quenching abilities of certain flame

gas constituents (5). Quenching of excited atoms is

caused by non-radiative de-activation of the excited state

resulting from collisions with molecules during the lifetime

of the excited state. Moreover, because the measurement

of a very small quantity of light is involved, any

extraneous source of radiation at the wavelength of

interest which is passing light into the monochromator

(within the proper light-gathering solid angle) along with

the analyte fluorescence may constitute a serious inter-

ference and even if corrected for may still impose enough

noise upon the signal to limit the analytical sensitivity

of the system. Examples of possible extraneous signal

sources include scattering of source radiation by molecules

and atoms,by solid particles,and by unevaporated solvent

droplets in the flame, flame background emission (both

atomic and molecular), and spectral interference from

overlapping fluorescence lines (atomic) and bands (molecu-

lar ). Other limitations affecting the measurement of

atomic fluorescence signal have been described (14).

Because the potential advantages to be derived from

increasing the intensity of available excitation sources

in atomic fluorescence spectrometry have long been recog-

nized, numerous researchers have attempted to increase the

peak outputs of sources conventionally operated in the cw

mode by pulsing these sources with an intermittent duty

factor at high peak input powers, while maintaining the

average input power at the usual level (15). By gating

the signal detection system to coincide with the excitation

source output pulse, the detector can be made to "observe"

a much higher signal level for a given analyte concentra-

tion than is possible with the same source operated in the

cw mode. The signal-to-noise ratio improvement sought in

these studies derives from the fact that the fluorescence

signal is proportional to the peak source intensity (and

is therefore larger than that obtained in the cw mode),

while the background signal, and therefore the background

noise, are limited by the "on" time of the gated detector

(and are therefore smaller than the background signal and

noise obtained in the cw mode) (16). Limitations experi-

enced in the practical application of these techniques to

atomic fluorescence spectrometry have been generally asso-

ciated with the difficulty or expense of implementation

and with the propensity for introduction into the system of

new or special sources of noise (17).

For the analysis of a multielement mixture by atomic

fluorescence spectrometry, a potentially attractive alterna-

tive to the use of conventional line sources of excitation

is the possibility of simultaneous excitation of many

elements with a single continuum source. Available

continuum sources offer good stability, long useful life,

and simplicity of operation as compared to many line sources,

particularly electrodeless discharge lamps and hollow

cathode discharge lamps. However, several studies (17-19)

have indicated that an important limitation associated with

the application of continuum sources to analytical atomic

fluorescence spectrometry is the low output intensity over

atomic spectral lines, especially in the low UV (180-260 nm).

As a result, limits of detection for elements with reso-

nance lines in the low UV are typically one or two orders

of magnitude worse than those obtainable with line sources.

Interference from source light scattered in the flame,

although considerably greater (worse) when a continuum

source is used, owing to the dependence of the scatter

signal in this case upon the square of the spectral band-

pass of the monochromator, can be minimized by proper

choice of flame and flame conditions and can be compensated

for by wavelength scanning in the vicinity of the fluores-

cence line (18).

An evaluation of a pulsed Eimac source of excitation

for use in atomic fluorescence spectrometry is presented in

Chapter II. Limits of detection, analytical curves, and

precision data are reported for several elements, and the

detection limits are compared with those obtained in a

previous study (18) utilizing an Eimac lamp operated in the

cw mode. Signal processing was accomplished with the

photon-counting system of Reference 18.

Chapter III consists of a separate evaluation of the

fluorescence background observed during the study of the

pulsed Eimac source, with special regard to the numerous

spectral irregularities believed to have resulted from

scatter and/or molecular fluorescence, neither of which have

been reported previously in significant quantities under

the particular set of experimental conditions applied to

these studies. An attempt was made to identify each of

the observed peaks and bands. Spectra were recorded under

three different analytically useful sets of flame conditions

in the flame used for the pulsed Eimac study (argon

sheathed air-acetylene flame). Results obtained with the

Eimac lamp operated in the pulsed and aw modes, respec-

tively, are compared, and the spectral distribution of the

lamp radiation was measured for both modes of operation.

A commercially available filter designed to remove phosphine

(PH ) from acetylene (20)was evaluated in terms of its

ability to reduce the fluorescence intensity of phosphorus

monoxide (PO), a species formed in the flame from PH3'

present in the acetylene as an impurity.

Because the fluorescence of PO was discovered to be

the most intense spectral feature in the observed back-

ground fluorescence spectra, and because molecular fluores-

cence in flames has been largely untreated in the literature,

it was decided to characterize further the molecular

fluorescence of PO. This study is described in Chapter IV

of this dissertation. Signals produced by phosphate

solutions were compared to signals generated only by the

PH3 impurity in the acetylene. Fluorescence and emission

intensities are compared in various flames, and a silicon

intensified target (SIT) tube in combination with an

optical multichannel analyzer (OMA) is evaluated for poten-

tial application to the study of molecular fluorescence,

In order to investigate the potential utility of PO fluores-

cence for the analysis of phosphorus, the analytical

precision and limit of detection for phosphorus determin-

tions have been obtained for solutions of phosphoric acid

nebulized into various flames.

The final chapter of this work contains a brief summary

and conclusion, along with a few statements about possi-

bilities for future research in these areas.



The useful photon flux reaching the atomic vapor from a

continuum excitation source can be increased by increasing

the input power to the source, by improving the efficiency

of light gathering of the source optics, and/or by pulsing

the source while synchronously gating the signal detection

system. Unfortunately, there exists at present only limited

room for improvement in signal-to-noise ratio to be achieved

by use of (non-laser) sources with high input powers, be-

cause the output stability of many available sources dete-

riorates rapidly with an increase in operating current.

Moreover, lamp life is generally significantly reduced, and

operational difficulties may be encountered. The light-

gathering efficiency of the source optics, often severely

limited by the geometry of the lamp itself, seems also to

have reached a practical limit for currently available

sources with the design of the Eimac lamp. Therefore,

pulsing of an Eimac continuum source (while synchronously

gating the detector) presently appears to offer the greatest

potential for increasing the usable exciting flux at the

atomic vapor.

A likely fringe benefit expected to result from the

pulsing of continuum sources is enhanced radiant output in

the UV (21). Such enhancement depends upon the ability of

the input power pulse to increase significantly the color

temperature of the source during the lifetime of the pulse.

Evaluation of a pulsed linear flash lamp continuum

source for application to atomic fluorescence spectrometry

was performed by Johnson, Plankey, and Winefordner (17);

however, their work was considerably hampered by experimen-

tal difficulties. The high voltage pulses needed to start

the xenon are flashtube generated a great deal of RF inter-

ference, and the observed noise level was strongly dependent

upon the orientation of the output cables from the pulsed

power supply. Furthermore, the radiative efficiency of the

long capillary flashtube, i.e., the light-gathering effi-

ciency of the optical system, was relatively low. Their re-

sults indicated that the detection limits with the flash lamp

source were approximately an order of magnitude worse than

those obtained with the 150 W Eimac xenon arc lamp operated

in the aw mode.

The work described here attempts to circumvent these

experimental limitations through operation of an Eimac con-

tinuum source in the pulsed mode with the pulse superimposed

upon a constant dc level. Efficiency of radiative transfer

from source to flame is very high with this source, and it

can be pulsed without the use of a high voltage pulse to

initiate the discharge if a residual dc current is per-

mitted to flow through it between pulses. Astoichiometric

argon separated air-acetylene flame was used for all elements

studied in order to facilitate comparison of results with

those of other atomic fluorescence studies utilizing the

Eimac lamp and other sources operated in the cw mode (17-19,

22). Other experimental parameters, such as the monochroma-

tor slit width and the observation height in the flame, were

also chosen with this purpose in view.


Apparatus (see Table I). A block diagram of the system

used in the present study is shown in Figure 1. Light from a

pulsed Eimac xenon arc continuum source was focused by a

single lens of 2 in focal length into the center of an

argon separated air-acetylene flame at a fixed height above

the burner top, and then reflected back through the flame

with a 140 mm diameter, 62 mm focal length spherical mirror

which was coated for enhanced reflection in the UV. A lens

matching the f/5.3 aperture of the monochromator was posi-

tioned within the black cylindrical entrance port of the

monochromator, at a distance of twice its focal length from

both the flame and the entrance slit. To further reduce the

level of stray light reaching the detector, a conical, black-

ened light trap of 3 1/4 in diameter and 5 in length was

placed on the opposite side of the flame. The photomulti-

plier tube was protected from RF and other sources of inter-

ference with a special housing designed for this purpose

(see Table I or Reference 18). The burner/nebulizer assembly

was a Perkin-Elmer unit, modified for use with a

Table I

Experimental Components and Manufacturers






Digital Synchronous


Capillary Burner

Xenon Arc Lamp

Lamp Power Supply (dc)

Pulsed Power Supply

Function Generator



EMI 6256S





VIX 300 UV





Acton, Ma 01720

Gencom Div.,
Plainview, NY 11803

SSR Instruments,
Santa Monica, Ca 90404
(now Princeton Applied
Research, Princton,
NJ 08540)

SSR Instruments,
Santa Monica, Ca 90404

SSR Instruments,
Santa Monica, Ca 90404

Nonrwalk, Ct 06852

Laboratory constructed

EIMAC, Div. of Varian,
San Carlos, Ca 94070

EIMAC, Div. of Varian,
San Carlos, Ca 94070

Velonex, Div. of Pulse
Engineering, Inc.
Santa Clara, Ca 95050

Wavetok, Inc.,
San Diego, Ca 92123



Figure 1. Schematic diagram of the experimental system employed for evaluation of
the pulsed Eimac lamp. S1 and S2 are SPST knife switches; D1 = IN3212,
D2 = IN4722.

capillary-type burner head (23) fitted with a collar de-

signed to provide a sheathing gas around the flame (argon).

A 300 W Eimac xenon arc continuum source enclosed in

a fan-cooled housing was used for excitation of atomic

fluorescence. The lamp contains a built-in parabolic

aluminum-coated reflector which reflects most of the emitted

radiation through a 2.5 cm diameter sapphire window. Power

supplied to this source consisted of periodic pulses

superimposed upon a constant de level. Power pulses were

obtained from a Velonex High Power Pulse Generator with a

plug-in output unit (No. V-1766) which delivers up to

+ 120 V at high current levels. A conventional series-

controlled, constant current de power supply was employed

to maintain a constant residual dc current flowing through

the lamp while it was being pulsed.

The outputs of these two power supplies were wired

in parallel as shown in Figure 1. Switch S1 permitted

protection of diode D1 from the 15 kV starting pulse. Switch

S2 isolated the dc power supply from the pulse generator

and was left open while the lamp was being started. S1

was opened as soon as the arc had been initiated, and S2

was closed immediately prior to switching on the pulse

generator. The purpose of the diodes was to prevent reverse

current surges.

The signal current (current pulses corresponding to

individual photons) from the photomultiplier tube was passed

into an amplifier/discriminator module and then into a

digital synchronous computer; these two instruments com-

prised the photon-counting system used in this study. The

system contained two separate scalers (labeled DATA and

BACKGROUND, respectively) and a variable sampling interval

which could be synchronized with the excitation pulses from

the Eimac lamp. In the mode of operation adopted for this

work, the sampling gate was applied alternately to the

signal produced during the "on" time of the excitation

pulse and to the signal occurring between pulses. The

former signal was collected in the DATA scaler, while the

latter was collected in the BACKGROUND scaler. The outputs

of the two scalers could be summed, subtracted, or presented

individually. Thus, the contributions to the total signal

of flame emission, photomultiplier dark current, and

fluorescence generated by the residual lamp radiation

between pulses could be subtracted out to obtain a pure

pulse fluorescence (plus scatter) signal, while the summed

outputs, recorded with distilled water being aspirated into

the flame, could be used to estimate the noise on the back-

ground observed in fluorescence (see later).

The pulsed output of the high power pulse generator

was synchronized with the photon counter data sampling gate

by simultaneously triggering both instruments from a

function generator, which was set to provide a 300 Hz

square wave trigger signal. A gate triggering delay control

on the photon counter permitted the gate to be finely

tuned to coincide with the lamp output pulse. A gate width

or sampling period of 20 ps was applied to both scalers of

the photon counter, while the pulse width of the lamp

input pulses was fixed at approximately 14 ps (FWIIM). These

settings proved to be optimal in terms of overall system

performance. Attempts to increase the pulse width sharply

reduced the maximum usable pulse power, while further

reduction in the photon counter gate width proportionately

decreased its dynamic range, already severely restricted

in the pulsed made (as compared to operation in the cw mode;

refer to the Results and Discussion section).

Procedure and experimental conditions. Experimental

conditions used throughout this study are summarized in

Table II. All components of the experimental system were

operated according to the directions given in the manu-

facturers' manuals, except as otherwise noted (i.e., the

pulsing of the Eimac lamp). The monochromator slit width,

the observation height in the flame,and the argon-fuel-air

ratio for the flame were established initially at values

intended to represent a signal-to-noise compromise for all

the elements and were held constant throughout the study.

The chemicals used in the preparation of stock solu-

tions were all of reagent grade. All standard solutions

employed in this work were obtained by serial dilution of

the stock solutions.

Table II

Experimental Conditions for Present Atomic
Fluorescence Studies

Monochromator (0.3 m, Czerny-Turner, f/5.3 600 g/mm grating
blazed at 3000 nm.)

Slit width 75 pm

Slit height 1 cm

Bandpass 0.4 nm

Dc Lamp Power Supply


Pulsed Power Supply

Amplitude setting

Pulse width

Repetition rate

Flame Conditions
(flow rates of gases)



6.5 A

S1/2 of full scale

=15 ps

300 Hz

9.7 R/min

1.3 &/min

Ar. 10 R/min

Obvervation height in flame, 2.5 cm (above burner top)

Photomultipier voltage, 1115 V

Photon Counter gate width, 20 us

The analytical calibration curve data presented here

are based upon ten, 10 s measurements, while the precision

data (% RSD) are based upon ten, 1 s measurements (ten,

10 s measurements resulted in an improvement of /TUi in

% RSD).

The detection limits were derived from the ten, 1 s

measurements used for the determination of the % RSD data

and were converted to detection limits corresponding to

10 s counting times; these values were obtained with solu-

tions whose concentrations were roughly 102 times the

detection limit for each element and were estimated by

extrapolation to a signal equal to twice the noise on the


Results and Discussion

In Table III, atomic fluorescence detection limits

are given for 16 elements, along with values for comparison

purposes obtained in a previous study (18) with a 500 W

Eimac continuum source employed in the cw mode. When a

photon-counting detection system is used and the major

noise is shot noise, the RMS background (source-induced

scatter, and emission and fluorescence of flame gas

species) noise can be estimated with sufficient accuracy by

taking the square root of the background signal obtained

when distilled water is aspirated into the flame. For back-

ground shot noise to predominate, it must exceed 5 times

the background, where ( is the background flicker factor.

Values for comparison purposes from Table IV, Reference 2.
Experimental conditions are identical for each element.

bNoise due to background fluorescence resulted in analytically
useless detection limits.

Table III

Limits of Detection Obtained for Several Elements in
the Argon Separated Air/Acetylene Flame by
Atomic Fluorescence Spectrometry with
a Pulsed Continuum Source

Pulsed Continuum
ement Wavelength, nm (this work)

Limit, pg/ml
500 Watt Continuum





















































Thus, if the background flicker factor is about 0.003,

typical for the Eimac-induced scatter background, then the

total noise is predominantly background shot noise as long

as the background count is below = 3xl04 counts. The

analytical calibration curves shown in Figures 2 and 3,

as well as the relative standard deviations listed in

Table IV, lend support to the general validity of this


The analytical calibration curves of Figures 2 and 3

reflect an important instrumental limitation associated

with the utilization of a photon counter with a small duty

factor. Because the maximum usable count rate is eventually

limited by the pulse pair resolution of the photon counter,

a restriction is imposed upon the total number of counts

that can be recorded in a given measurement (time) interval

without incurring an intolerably large counting error due

to pulse overlap. If the actual photon-counting system

"on" time is very much smaller than the measurement interval

(small duty factor), the above restriction is proportionately

more stringent. The "on" time duty factor for the system

employed here was 0.006 (i.e., 20 Is times 300 gate inter-

vals per second), and the instrumental pulse pair resolu-

tion was 12 ns. From these parameters, the practical upper

limit on the total number of counts per 10 s measurement was

estimated to be approximately 105 counts. Therefore, none

of the analytical calibration curves were extended beyond





9 C
2 M S

1 al'
-4 10


Fiue2 nltclclbaincre o d n n
Snotie ihteple ia ore


5 x


- 3

- 1 3 li

Figr 3.O A alyia airto uvsfrMAadZ
obandwt h ule ia ore


Table IV

Precision of Analytical Determinations of Several Elements
by Atomic Fluorescence Spectrometry with
a Pulsed Continuum Source

Relative Standard
Deviation of Analyte
Fluorescence Signal
(% RSD)




Relative Standard
Deviation of
Background (% RSD)

Average Background
Elements Signal, Counts/s





























Wavelengths of lines used here are those listed in Table III.
Analyte concentrations were 2102 times the detection limit for each

this level, even though the linearity limit due to self-

absorption was not reached for any of the included elements,

except possibly zinc. The linearity limit imposed by the

photon counter can be increased by means of neutral density

filters placed in the optical path to reduce the photon

flux reaching the detector.

From Table III, it is apparent that limits of detection

determined with the present pulsed system are roughly

comparable to those obtained with the 500 W Eimac continuum

source operated in the cw mode. That the expected improve-

ment in detection limits largely failed to materialize is

due primarily to the inherent instability of the Eimac lamp

when operated in the pulsed mode, and to the unusually high

level of background scatter radiation and/or molecular

fluorescence observed during the "on" time of the pulse.

A rather high residual de current (6.5 A) was required

to sustain the arc, even at moderate pulsing amplitudes.

The total signal including atomic fluorescence generated by

the flame and the source during a 20 ps interval between

pulses was electronically subtracted from the same signal

produced within the duration of the pulse. Moreover, the

background signal occurring between pulses was added to

the background signal associated with the pulse in order to

determine the noise on the background (for detection limit

calculations). Thus, the signal-to-noise ratio also

depended upon the residual de current flowing through the

lamp between pulses. Visible fluctuations in the lamp

output intensity were observed at all but the lowest of

pulse power settings, but the precision data of Table IV

suggest that the noise amplitude in many cases was not

significantly greater than that which is expected from a

shot noise limited measurement system; consequently, these

source fluctuations must not have contributed significantly

to the total noise at the signal levels used to determine

the precision data. The background fluorescence and

scatter signals observed with the present pulsed system

were substantially larger than those seen in previous ow

studies, and this fact is believed to represent an important

limitation with respect to signal-to-noise ratio in the

present study.

The reasons for the unusually large background signal

observed as fluorescence cannot be clearly ascertained with

the instrumental system used for this work. However, there

is little room for doubt that, at any given wavelength,

scattering of the incident source radiation from particles

or molecules in the flame accounted for a significant pro-

portion of the total background signal. It should be noted

that several previous studies (17,19,22) of atomic fluores-

cence excited by a continuum source have reported little

or no evidence of scatter when the argon separated air-

acetylene flame was used.

The phenomenon of light scatter is not sufficient,

however, to explain the substantial irregularity observed

in the spectral distribution of the background radiation.

Scattering of source radiation might be expected to produce

some contour in the background spectrum if the spectral

distribution of the source radiation contained prominent

emission lines and bands superimposed upon the emitted

continuum. Preliminary studies of the spectral output of

the pulsed Eimac lamp have indicated that a structured

spectrum of this type is indeed emitted. Nevertheless,

many of the most prominent anomalous features in the back-

ground spectrum are known or suspected to have resulted

from the fluorescence of molecular species in the flame.

In a series of experiments conducted prior to this study,

the molecular fluorescence bands occurring between 230

and 270 nm (22) were tentatively linked to the PO molecule.

Apparently, PO is formed in the flame from an impurity in

one of the flame gases, probably the PH3 impurity in

acetylene (20). As a part of this study, a 0.33 M H P
solution was aspirated into the flame and was found to

increase the PO fluorescence intensity by a factor of three.

These bands interfered so extensively with the determina-

tions of Co at 240.7 nm and Fe at 248.3 nm that useful

detection limits for these elements could not be deter-

mined at these wavelengths. Similar interference are

believed to have occurred elsewhere in the spectrum. Indeed,

the observed background irregularities due to source scatter

and/or molecular fluorescence were of sufficient magnitude

and number to cause possible error if off-wavelength correc-

tion for the background were attempted in multielement

analysis as described in Reference 18.

Cursory observations of the spectral radiant intensity

of the Eimac lamp have indicated a shift in the source

output toward the UV resulting from the pulsing of the lamp.

However, the background interference discussed above

were most apparent in the short-wavelength portion of the

spectrum, and these appeared to nullify any improvement

which might otherwise have been expected in the detection

limits of elements with resonance lines at low wavelengths.

Operation of the Eimac source in the pulsed mode did not

seriously decrease the useful lifetime of the source (the

present lamp has been operated for more than 200 h with

no apparent loss in light output and little observable

deterioration of the arc electrodes). Moreover, no RF

or ground-loop-type interference was experienced with the

system as employed in this work.



In the previous chapter, in which a pulsed Eimac

continuum source was evaluated for application to atomic

fluorescence spectrometry, it was noted that the background

spectrum, i.e., the fluorescence signal observed at each

wavelength when only distilled water was aspirated into

the flame, contained numerous "anomalies" which resulted

in sharply elevated background levels at certain wave-

lengths of interest. It was recognized that these back-

ground spectral irregularities could be attributed to either

of two phenomena: (a) molecular fluorescence of molecules

within the flame, and/or (b) prominent peaks in the

source spectral output, transferred to the detector by

scattering of source radiation from particles and molecules

in the flame. Some of these background anomalies were known

to have resulted from molecular fluorescence; however, many

others had never been observed previously and thus could

not be described. Because of the discrete nature of the

digital output signal of the photon-counting system and

the lack of a suitable digital-to-analog converter, it was

not feasible in the work of Chapter II to scan the entire

background fluorescence spectrum over the range of wave-

lengths utilized for multielement atomic fluorescence


Although molecular emission (24-27) and absorption

(27-29) of flame gas constituents have been fairly well

documented, and their effects upon analytical atomic fluores-

cence measurements recognized (4,13,30),molecular fluores-

cence in flames has received but little attention in the

literature. The fluorescence of the CH radical at 431.5

nm has been excited by a dye-laser in an oxyacetylene

flame (31), and several observations of OH flame fluores-

cence have been reported (3,32,33). Molecular fluores-

cence bands were seen, but not identified, by Johnson,

Plankey, and Winefordner (22). The fluorescences of

several alkaline earth compounds (CaOH, SrOH, and BaC1)

have been observed when concentrated solutions of these

compounds were aspirated into flames irradiated by high

intensity continuum sources (34,35).

Reported incidences of observable scatter signals in

atomic fluorescence studies have been linked primarily to

the use of cool flames and sample matrices with high solids

content (18,36,37). Measurable scatter signals have, in

general, not been noted when pure aqueous solutions were

used with the air-acetylene flame.

For these reasons, a study of the background anomalies

described in the previous chapter has been conducted and is

discussed here. Wherever possible, the equipment and

experimental conditions used in the work of the previous

study (Chapter II) have been retained in the present work

in order to preserve the validity of any comparisons which

may be drawn.


Instrumental systems (see Table V). Block diagrams

of the experimental systems used in the present work are

given in Figures 4-6. The excitation source, nebulizer,

burner, monochromator, photomultiplier detector (and its

shielded housing), mirror, light trap, and entrance optics

are exactly as used in the work of the previous chapter.

For fluorescence measurements in the pulsed mode, the ex-

perimental arrangement of Figure 4 was adopted; this arrange-

ment is similar to the one used previously (see Figure 1)

except that a boxcar integrator and strip-chart recorder

perform the signal-processing functions for these studies.

Figure 5, the system used here for the fluorescence mea-

surement in the cw mode, is similar to Figure 4 except

that a lock-in amplifier is used in place of the boxcar

integrator, and the excitation optics have been altered

to accommodate a mechanical chopper in the excitation beam.

In addition, the lamp power supply employed for all cw

measurements was a series-controlled, constant current do

power supply similar to the one described in the previous

used in this study are listed in

Table V

ponents and


Experimental Comn


Boxcar Integrator/Integrator
Modules (A & B)

Lock-in Amplifier



162/164 Princeton Applied Research,
Princeton, NJ 08540

391 Ithaco, Inc., Ithaca
NY 14850

164 Ithaco, Inc., Ithaca
NY 14850

SRG Sargent-Welsh Scientific,
Birmingham, Al 35202

PS 300-1 Eimac, Div. of Varian,
San Carlos, Ca 94070

-- L'Oxhydrique Francaise,
Malakoff, Paris, France

-- (Laboratory constructed)

Current Preamplifier

Strip Chart Recorder

Lamp Power Supply (dc)

Acetylene (PH 3) Filter

Mechanical Chopper

Note. All other instruments
Table I.

Figure 4. Schematic diagram of the system used for all flame background molecular
fluorescence measurements in the pulsed mode. Switches and diodes are
as described previously (Figure 1).


Lens C

Figure 5. Schematic diagram of the system used for
the measurement of flame background molecu-
lar fluorescence in the cw mode.

Figure 6. Schematic diagram of the experimental arrangement employed for the
measurements of the lamp spectral output. Note that either the cw or
the pulsed power supply circuit can be attached to the lamp input


SNeutral Density Filter


I Diaphragm

chapter. It should be noted that the optical configuration

of Figure 5 is not the optimum arrangement for chopped

beam-cw source measurements; the system can be optimized,

in principle, only when (for lenses of equal diameter) the

focal length of lens A is twice that of lens B, or (if the

focal lengths are equivalent) when the diameter of lens B is

twice that of lens A. In the system of Figure 5, lenses A

and B have equivalent (2 in) focal lengths and diameters,

and despite adjustments in their positions to achieve the

best possible result, significant light losses were en-

countered. However, these losses did not detract from the

validity of the background recordings of this work, since an

optimized system was not required to observe the charac-

teristics and features described in this study.

For all measurements except that of fluorescence in

the cw mode, signal processing was accomplished with a dual-

channel boxcar integrator whose output was fed to a strip-

chart recorder. The dual-channel boxcar integrator employed

for these studies was equipped with two separate integrator

modules labeled A and B, respectively, a variable aperture

or sampling interval applicable simultaneously to both

integrators, and a separate aperture delay circuit for each

integrator module. The aperture delays are referenced to

a single trigger signal supplied externally; hence the

instrument is capable of being synchronized to the excita-

tion pulse in the pulsed mode by simultaneously triggering

both the boxcar and the pulsed power supply from the same

trigger source. The outputs of the two integrators can be

read individually or can be subtracted electronically to

yield the difference signal, A B. With the aid of an

oscilloscope, the aperture of the A integrator was set

(using the aperture delay control) to coincide with the

lamp output pulse (for pulsed mode measurements) and the

aperture of the B integrator was positioned at a point in

time approximately half-way between successive A integrator

apertures. Therefore, the total signal fed into the A

integrator consisted of flame emission signal plus photo-

multiplier dark current plus any signal (fluorescence and/

or scatter) generated by the source output pulse, while

the total signal processed by the B integrator consisted

of flame emission signal plus photomultiplier dark current

plus any fluorescence and/or scatter signal produced by

the residual lamp output between pulses. Hence, the

difference signal, A B, was composed entirely of fluores-

cence and scatter signals produced by the lamp output pulse

(as in the previous study), and it was this net signal which

was recorded in all spectra obtained in the pulsed mode.

As in the work of the previous chapter, the aperture was

chosen to be 20 us, and the excitation source pulse width

was fixed at approximately 15 ps. A time constant of 0.8 s

was employed for all work with the boxcar integrator.

For the measurement of fluorescence in the cw mode,

the signal was processed with a lock-in amplifier, together

with a current preamplifier whose function was to amplify

the signal current and convert it to a signal voltage for

use at the imput terminal of the lock-in. The reference

channel of the lock-in was supplied with a reference output

triggering signal from the chopper. The sensitivity was

adjusted to achieve the desired response, and the time

constant was set at 0.4 s. The output of the lock-in was

fed to the same strip-chart recorder as employed with the

boxcar integrator.

To record the spectral distribution of the excitation

source radiation in both the pulsed and cw modes, the

experimental system was modified as shown in Figure 6.

An aluminum block coated with Eastman White Reflectance

Coating no. 6080 (Eastman Kodak Co., Rochester, N.Y. 14650)

was placed on top of the burner assembly and oriented so

as to scatter the source radiation into the monochromator.

The coating contained barium sulfate, and when applied

according to the manufacturer's directions, it formed a

diffuse reflector whose absolute reflectance is 90% or

better at all UV and visible wavelengths above 220 nm (38).

To reduce the radiant flux impinging on the photomultiplier,

a 2.0 neutral density filter was placed in front of the

monochromator entrance port, and a diaphragm with a 0.5 cm

aperture was positioned between the source and the

reflector at a distance of 6 cm from the reflector. The

distance from source to reflector was approximately 30 cm.

As shown in Figure 6, either the pulsed or cw lamp power

supply could be attached to the lamp input terminals, as


Procedure and experimental conditions. A summary of

the experimental conditions applicable to this work is

given in Table VI. All equipment was operated as de-

scribed in the manufacturers' manuals except as previously

noted. The monochromator slit width and the observation

heightin the plane were chosen to be identical to those

used in the previous study (Chapter II).

When required, the acetylene filter (for removal of

PH3) was spliced into the acetylene supply line between the

pressure regulator on the acetylene tank and the flow con-

trol value. In all fluorescence studies, distilled water

was aspirated into the flame.

For recording the lamp spectral distribution using

the experimental arrangement of Figure 2, all instrumental

settings were maintained identical to those used for the

fluorescence studies, except that the signal recorded in the

cw mode was the output signal from the A integrator only,

rather than the difference signal, A B. Thus, the boxcar

integrator was gated with a repetition frequency of 300 Hz

and an aperture width of 20 ps in both the pulsed and cw

modes; this was done to permit a more direct comparison of

the lamp outputs in the two modes.

Table VI

Summary of Experimental Conditions


Slit width 75 pm

Slit height I cm

Bandpass 0.4 nm

Spectral scan rate 5 nm/min

Lamp Power Supply (dc)
(used in pulsed power circuit)


Lamp Power Supply (dc)
(used in cw studies)


6.5 A

20 A

Pulsed Power Supply

Amplitude setting

Pulse width

Repetition rate

Flame conditions



~ 1/2 of full scale

~ 15 us

300 Hz

Air lO.51/min
C2H2 1.3R/min
Ar 10 IR/min



10 L/min

10 e/min



Table VI (continued)

Strip Chart Recorder chart speed 1 in/min

Photomultiplier voltage -1400 V

Boxcar Integrator aperture 20 ps

Observation height in flame 2.5 cm above burner top

Results and Discussion

The spectra recorded for this study are shown in

Figures 7-13. In each case, the baseline (the horizontal

axis) represents the photomultiplierdark signal, i.e., the

signal observed with the monochromator slits closed. The

series of bands lying in the 220-275 nm region of the

fluorescence spectra (Figure 7-11) constitute the y-system

of the PO molecule, arising from the electronic transition

A2E-X2n. The weak band in the 280-295 nm portion of the

spectrum of Figure 7, as well as the much stronger band in

the 305-320 nm region, are both due to the OH radical. This

system, known only as the 306.4 nm system of OH, is due to

the transition A2 +-X2H. The peak or sharp band at 336 nm

is probably the band head of the 336.0 nm system (A H-X E)

of the NH radical. Interference of this band with atomic

fluorescence measurements is, of course, unlikely because

of its narrowness. The band whose maximum is at 388.3 nm

is believed to be due to the cyanogen violet system, i.e.,

the B2E-A2H transition of the CN radical. Another possi-

bility for this assignment, although considered less likely,

is the 390.0 nm system of CH. However, this band is

described (27) as being degraded to the red and having

maxima at 397.1 and 388.9 nm. Of these characteristics,

only a "secondary" maximum at 387.1 nm is evident, which

suggests only that the observed fluorescence in this spec-

tral region may contain a contribution from both species.

1 -v i

a 230 250 270 290


z 3


320 340 360 380



460 480 500 520


Figure 7. Flame background molecular fluorescence spectrum
obtained in the pulsed mode from a fuel-lean
flame. The portions of the 190-600 nm spectral
region which are not shown here contained no
detectable molecular fluorescence.


o 230 250 270 290

z 2-


M 320 340 360 380

u- 3



460 480 500 520


Figure 8. Flame background molecular fluorescence spec-
trum obtained in the pulsed mode from a fuel-
rich flame. As in Figure 7, only selected
portions of the spectrum are shown.

. 1-

8 230 250 270 290




4630 480 500 520


Figure 9. Flame background molecular fluorescence spec-
trum obtained in the pulsed mode from a
stoichiometric flame. As in Figure 7, the
barren portions of the 190-600 nm spectral
region have been excluded.




Selected portions of the flame background
molecular fluorescence spectrum obtained in
the cw mode with a stoichiometric flame.

Figure 10.


320 340 360 380

460 480 500 520


vr 3-




Figure 11.

Selected portions of the flame background
molecular fluorescence spectrum obtained in
the pulsed mode from a stoichiometric flame
while employing the acetylene filter (for
removal of PH3) as described in the text.






Figure 12.

Lamp spectral output
wavelength region of
in the pulsed (solid
line) modes.

in the 190-350 nm
the Eimac lamp operated
line) and cw (dotted



3- 8040 2



440 460 480 500

520 540 560 580


Figure 13. Lamp spectral output in the 350-590 nm wave-
length region of the Eimac lamp operated in
the pulsed (solid line) and cw (dotted line)

The series of bands observable beyond 450 nm are the grating

second-order dispersion o-f the PO y-system bands. Although

observed in second order, these features are nonetheless

capable of adverse influence upon atomic fluorescence

measurements in this spectral vicinity. The above band

assignments are summarized in Table VII.

At any wavelength in the fluorescence spectra which

is well removed from the observed bands, the offset between

the baseline and the spectral trace represents the scatter

signal at that wavelength. There are no "anomalies" in

any of these spectra which can be shown to correspond to

peaks in the spectral distribution of the lamp radiation,

as had been suspected previously. However, a scatter

signal is obviously present at all wavelengths and in atomic

fluorescence spectrometry with a continuum source may ulti-

mately limit the signal-to-noise ratio for many analyses,

as it did in the previous study. In spite of these results,

it may still be possible to observe peaks in the fluores-

cence background, transmitted from the source by scatter,

in those applications of the pulsed Eimac source where

scatter signals are found to be large.

An important objective of this study was to effect a

comparison of the results obtained with the Eimac source

operated in the pulsed and cw modes, respectively. Figure 9

is the pulsed mode fluorescence spectrum which most closely

corresponds to the cw mode fluorescence spectrum of Figure 10.

The fluorescence of the y-system of PO is clearly more


Table VII

Summary of Molecular Fluorescence Band Assignments

Spectral Region (nm) Species Transition Spectroscopic System

220-275 PC A2 _2H y-system

280-295 OH A2Z -X 8 306.4 nm system

305-320 OH AE -_X H 306.4 nm system

323-330 PO B2E-X2n 8-system

336 (max) NH A H-X E 336.0 nm system

380-390 CN B2C-A271 cyanogen violet system

450-550 (PO y-system grating second order)

intense (relative to the fluorescence signals of OH, NH,

and CN) in the pulsed mode than in the cw mode; an apparent

reason for this can be discerned from Figures 12 and 13.

Not only is the spectral output of the pulsed lamp generally

shifted toward the UV as compared to that of the cw source,

but also many of the large peaks in the pulsed lamp spectrum

coincide with the bands of the PO y-system. Hence, the

fluorescence excitation of this molecular band system is

more efficient in the pulsed mode.

Comparing Parts A and B of Figures 12 and 13 further,

it can be seen that the lamp output in the pulsed mode

during the 20 ps sampling period is only a little (~ 3 to 5

times) larger than the lamp output in the cw mode during

the same sampling period. Considering that practical

analytical atomic fluorescence measurements employing a

continuum source in the cw mode utilize a much larger duty

factor and more efficient excitation optics, the above fact

must have played a significantly negative role in the

previous study, in which detection limits obtained in the

pulsed mode were compared to those obtained in the cw mode,

with little or no improvement evident for the pulsed mode.

The reasons for this less-than-favorable intensity ratio

are not completely understood at this time; apparently the

peak level of radiation emitted during the application of

the power pulse represented an excedsively small fraction

of the peak electrical power dissipated within the "on"

time of the pulse. Because accurate current measurements

cannot be made by conventional means in a low-resistance,

high-current circuit, and because the resistance of a

xenon arc does not obey Ohm's Law, it was not feasible to

measure the power input to the pulsed source.

It is interesting to compare Figures 7-9 to ascertain

the effect of varying the fuel/air ratio upon the observed

fluorescence bands. The absence, in the fuel-lean flame,

of the bands assigned to CN and NH is striking. Also, the

OH band at ~285 nm is present to a significant extent only

in the fuel-lean flame, and even the band at ~310 nm appears

to undergo a marked increase in intensity as the fuel/air

ratio is decreased. Further scrutiny of the latter band

reveals that, at the wavelengths corresponding to fluores-

cence maxima in the stoichiometric and fuel-rich spectra

(Figures 9-11 at 306.8 nm and 309.2 nm), there are sharp

"dips" indicating virtually no fluorescence at all in the

fuel-lean flame. Although these particular observations

may have little immediate relevance to the study of molecu-

lar fluorescence as a potential source of interference in

atomic fluorescence spectrometry, they do reveal that

molecular fluorescence in flames may be, in many instances,

strongly dependent upon flame conditions, a fact which could

be of considerable importance in optimizing experimental

conditions for analysis by atomic fluorescence spectrometry.

The fluorescence spectrum of Figure 11 was recorded

under the same set of experimental conditions as adopted

for the spectrum of Figure 9, except that the acetylene

filter for removal of PH3 was inserted into the acetylene

line immediately prior to recording the spectrum of Figure

11. The fluorescence of the y-system of PO is barely

discernible in Figure 11, indicating the excellent filter-

ing capability for the filter used here. Moreover, the

grating second-order bands above 450 nm have disappeared

entirely. These spectra suggest that the "attenuation

factor" for this filter may be as much as 30 or more (based

upon relative peak heights).

At this point, it is appropriate to consider in

general terms what effects these findings may have on

multielement analysis by atomic fluorescence spectrometry

with a continuum source. In applications where it is

desired to obtain a qualitative scan of the entire fluores-

cence spectrum in order to establish the presence or

absence of various metals within a sample, molecular

fluorescence bands could easily obscure the presence of

an atomic line. However, a detailed knowledge of the

molecular fluorescence background associated with the

particular analytical conditions (sample matrix, flame

gases, flame conditions, impurities, etc.) relevant to the

analysis should be effective in minimizing any such

spectral interference.

The effect of molecular fluorescence upon quantitative

multielement analysis is a more complex matter. Because a

molecular fluorescence band which overlaps an atomic

fluorescence line constitutes an additional component of

light viewed by the detector at that wavelength, it is

reasonable to expect an increase in the shot noise ampli-

tude at the output of the photodetector as a result of

any spectral interference from molecular fluorescence.

Moreover, molecular fluorescence is capable of transmitting

source fluctuations to the detector, should source flicker

noise prove to be significant in relation to the total

observed noise level in the system. In these ways, inter-

ference from flame molecular fluorescence is, in principle,

similar to that associated with source radiation scatter-

ing from flame components, except that molecular fluores-

cence is usually localized spectrally, and off-wavelength

correction for it may prove to be difficult or impossible

(as mentioned previously).

An additional noise component may become significant

for some types of molecular fluorescence. It has been

demonstrated in this study that at least some molecular

fluorescence bands display an extremely high sensitivity

to flame conditions, specifically the fuel/air ratio.

Furthermore, it is possible, indeed likely, that the

fluorescence of these species may display a strong depend-

ence upon the location in the flame of the focused source

radiation (taken here to be equivalent to the object

volume element, within the flame, of the monochromator

entrance optics). Therefore, any general instability or

turbulence associated with the flame may lead to measurable

fluctuations in the intensity of the molecular fluorescence

of these species.

If a fluorescing molecular species is introduced into

the flame by the aspiration of a real sample containing that

species (or its precursor), then the molecular fluores-

cence should interfere directly with any overlapping atomic

fluorescence line employed for elemental analysis and

thus cause a proportional quantitative error in the mea-

surement. Correction for this type of error would depend

upon a knowledge of the concentration of the interfering

molecular species (or its precursor) in the sample matrix,

as well as the concentrations of any concomitants affecting

the molecular fluorescence intensity, either positively or


It should be noted that the total effect on atomic

fluorescence measurements of the presence (or absence) in

the flame of any fluorescing molecular species cannot be

established solely from a knowledge of the fluorescence

characteristics of that species,because the absorption and

emission properties of that species also affect the signal-to-

noise ratio (5,13,30) for atomic fluorescence measurements

at any wavelengths susceptible to interference from the

molecular bands(s).



In spite of the increasing development of atomic

fluorescence spectrometry (39), there have been few studies

on molecular fluorescence of species in flames, as indicated

previously. It has been suggested (35) that molecular

fluorescence might be applied to the analysis of non-

metallic elements, such as sulfur, phosphorus, carbon,

nitrogen and the halogens, which are difficult to determine

by atomic fluorescence because their atomic resonance lines

are in the far or vacuum UV. However, molecular spectro-

scopic methods have been used previously for the analysis

of non-metallic elements in flames only via molecular

emission (40) and absorption spectrometry (41,42). Emission

intensities of molecular species in flames are often as high

as or higher than the flame emission intensities of atoms

(43), and it is well-known that the study of molecular

absorption flame spectroscopy yields important information

about molecular species (44-47), background absorption

(48,49), chemical interference in atomic absorption spec-

trometry (50-52), and kinetics and mechanisms of flame

reactions (53). Therefore, studies of molecular fluores-

cence in flames from the points of view of spectroscopists

and analytical chemists are of interest and so have been

carried out in the present work. The fact that the

molecular fluorescence bands of PO were found to constitute

a serious source of interference in atomic fluorescence

spectrometry represents an added incentive for this work.

During the past five years, a number of TV-type

multichannel detectors have been evaluated for application

to analytical atomic spectrometry (54-62). Among these

devices, silicon diode array vidicon camera tubes [or

silicon intensified target (SIT) tubes combined with an

optical multichannel analyzer (OMA) have been used as

detectors for atomic emission (58,60,62), absorption (57,59)

and fluorescence spectrometry (62). The SIT-OMA system

has a number of advantageous characteristics, including a

wide spectral range (200-1100 nm), high quantum efficiency

(>50% in the range 350-800 nm, >20% in the range 250-

100 nm), linear response over a large range of fluxes

(~103), low geometric distortion, feasibility of data

storage, and capability of subtraction of the background

in the spectra. These characteristics are considered in

order to propose the possibility that the SIT-OMA system

might be judiciously applied to the study of molecular

flame spectroscopy. In general, molecular spectra are

broad and weak, and contain a great deal of coincident

background radiation due to the various co-existing species

in the flame or matrix. Therefore, signal averaging of the

data (to obtain better signal-to-noise ratio) and subtrac-

tion of the spectral background are necessitated in the

measurement of the spectra. The characteristics of the

SIT-OMA system mentioned above are suitable for this pur-

pose and have been utilized in the present work to study

the molecular fluorescence of PO in various analytical



The experimental components used in the present studies

are listed in Table VIII, along with the appropriate manu-

facturers, and a block diagram of the experimental system

is shown in Figure 14. An Eimac 150 W xenon arc lamp hav-

ing a diameter of 25 mm was focused at the center of the

flame above the burner top by a lens. Although a differ-

ent model, the do power supply for the Eimac lamp was

similar to those described earlier. The monochromator

wavelength control was calibrated by correlation of read-

ings from the wave-length dial with the atomic lines

emitted by a mercury pen lamp. A Perkin-Elmer burner/

nebulizer assembly was used, with modification of the

burner head as described previously. Two kinds of capil-

lary burner heads were used: one with large diameter

capillaries was used for the air-acetylene and air-

hydrogen flames, and one with smaller diameter capil-

laries was utilized for the nitrous oxide-acetylene and

oxygen/argon-hydrogen flames.


Eimac Div. of Varian,
San Carlos, Ca 94070

Eimac Div. of Varian,
San Carlos, Ca 94070

Acton, Ma 01720

Princeton Applied Research
Corp., Princeton, NJ 08540

Princeton Applied Research
Corp., Princeton, NJO08540

MFE Corporation,
Salem, NH 03079

Perkin Elmer,
Norwalk, Ct 06852


Table VIII

Experimental Components and Manufacturers of Equipment
Used in Flame Molecular Fluorescence
Spectrometer System


Model Number


CW Source

Arc Illuminator

Power Supply













All flame conditions employed are summarized in

Table IX. The height of the optical plane was 2.5 cm above

the burner top in all cases. In the air-hydrogen and

oxygen/argon-hydrogen flames, including their argon

separated counterparts, an inner cone of unevaporated

solvent droplets was observed which extended from I to

3 cm above the burner head. The droplet zone scattered

radiation fromthe excitation source and resulted in a large

scattering background signal. Therefore, observation

heights lower than = 2 cm were not used for measurements

of PO fluorescence spectra in this study. Measurements of

the fluorescence signal were taken as a function of the

height of the optical plane above the burner head, and the

signal was found to decrease as the height was increased

above the tip of the inner cone of solvent droplets.

The SIT detector used in the present study contained

a plastic scintillator fitted onto its faceplate; this was

a required accessory for observations in the UV because of

the poor transmission characteristics below 350 nm of the

fiber-optic faceplate of the intensifier. The detector

was mounted at the exit port of a 0.35 m, f/6.8 Czerny-

Turner monochromator which was modified to accept the

detector housing.

The OMA employed for this work was equipped with

"delay" and "preset" controls for regulation of the number

of machine cycles (scans of either the data or background

Table IX

Flame Types and Conditions

Gaseous Flow Rate (R/min)
aAir C2H2 N20 H2 02 Ar Ar
Flame Type (Separation)

N20-C2H2 4.5 11 -

Ar-N20-C2H2 4.5 11 15

Air-C2H2 14 2.75

Ar-Air-C2H2 14 2.75 15

Air-H2 13 10 -

Ar-Air-H2 13 10 8

02/Ar-H2 10 2.6 13 -

Ar-02/Ar-H2 10 2.6 13 8

Ar- means Ar separated flame.

channels), and therefore, the integration time, applied to

any given analytical determination. The "delay" control

established the number of machine cycles to be counted as

a single data point in the OMA memory circuits, while the

"preset" control fixed the total number of data points to

be integrated in each spectral channel of the OMA.

Emission and fluorescence signals were observed

simultaneously by the present SIT-OMA system, and thus the

signals obtained were not purely fluorescence but rather

fluorescence enhanced by emission. However, fluorescence

spectra were produced by electronically subtracting the

emission from the fluorescence-plus-emission spectra,

utilizing for this purpose the two data storage (memory)

banks of the ORA. Emission was collected in one memory bank

by recording the spectrum with the excitation source shut

off, and fluorescence-plus-emission was collected in the

other memory bank by recording the spectrum with the excita-

tion source turned on.

The target of the image detector in the present system

had 500 vertically oriented spectral channels for the data

(or background) arranged in a horizontal row in the focal

plane of the monochromator so as to intercept a portion of

the linearly dispersed spectrum which appeared at the

exit port of the monochromator. The 500 channels covered

almost 50 nm of the spectrum, which was wavelength-cali-

brated by observing the atomic fluorescence spectrum of a

mixture solution of Cd, Ni, Co, Fe, and Pt. In almost all

cases, 2000 machine cycles of SIT, which corresponded to

a "delay" of 3 and a "preset" of 500 on the OMA, were

chosen for themeasurements of molecular fluorescence spectra.

The time for one machine cycle (one scan of the data or

background channels) was 32.8 ms; hence, the system re-

quired 65.6 s for one fluorescence spectrum. Longer accumu-

lation times resulted in fluorescence signals which exceeded

the memory dynamic range (105) of the OMA, and shorter

accumulation times were not sufficient to obtain a good

signal-to-noise ratio. All spectra were recorded with an

X-Y recorder connected to the X and Y plotting outputs of

the OMA.

All phosphoric acid (85%, sp. gr. 1.7) used was of

guaranteed reagent grade. Deionized water was used as the

reference for the background.

Results and Discussion

Molecular bands of PO have been thoroughly investigated

in vacuum discharge tubes and in flames by both emission

and absorption spectroscopy (27,40,42,45), but not by

fluorescence spectroscopy. It is known that PO shows three

band systems: a continuum band system near 540 nm a

8-system near 324 nm and a y-system near 246 nm (27). As

previously discussed, the a-system and the y-system are

caused by the transitions B2E X211 and A2X2Irepcily

where B2E and A2E are the excited electronic energy states

and X2H is the ground state of the PO molecule.

Molecular fluorescence spectra were measured in the UV

for each of the flames listed in Table IX; aqueous solutions

of phosphoric acid were nebulized into these flames to

obtain fluorescence. Fluorescence spectra observed in the

air-hydrogen flame and in the argon separated air-hydrogen

flame are shown in Figure 15. As can be seen in these

spectra, peak fluorescence signals were observed near 230,

238, 254, and 262 nm. The intensities of themolecular

fluorescence signals were higher in the argon separated

air-hydrogen flame than in the unseparated air-hydrogen

flame. Further, the argon separated flame produced less

background noise than the unseparated flame.

The 8-system of PO near 324 nm was also examined in

the air-hydrogen flame. The B-system bands were very

weak, i.e., less than 600 counts per 2000 machine cycles

at the wavelength of highest intensity (for al M HPO

solution and an 850 Fim slit width) as compared with ~104

counts for the y-system; moreover, the observed signals

were largely emission rather than fluorescence.

In order to compare molecular fluorescence signals in

various flames, the fluorescence signals were measured in

each of the flames listed in Table X. The results indicate

that the signals observed in the air-hydrogen, the air-

acetylene, the oxygen/argon-acetylene, and the corresponding

Figure 15.

Molecular fluorescence spectra of PO in
(A) the air-hydrogen flame and (B) the
argon separated air-hydrogen flame, result-
ing from the aspiration of al1M H3PO4
solution into the flames. Experimental con-
ditions: slit width = 100 pm; OMA "delay"
= 3; OMA presett" = 500.


Ar- means the Ar separated flame.

Counts per 2000 machine cycles (delay 3, preset 500), at 246 nm,
produced by a 1 M H30 solution.

Monochromator entrance slit width.

Total signals, i.e., fluorescence enhanced by emission.

eFluorescence signals due to the PH3 concomritant in acetylene,
which were measured while aspirating only water into the flames.

Table X

Comparison of the Fluorescence Intensities of PO in
Various Flames

Fluorescence Intensity (counts)
Flame Typea 100 pmc 850 pmc

N20-C2H29680 (31050)d

9450 (29040) -
Ar-N20-C2H2 (5250)e








6550 (1800)e

41,070 (ll020)e



argon separated flames are due almost entirely to molecular

fluorescence, while the signals from the nitrous oxide-

acetylene and argon separated nitrous oxide-acetylene flames

are due to fluorescence enhanced by emission. The spectra

observed with the argon separated nitrous oxide-acetylene

flame are shown in Figure 16. The fluorescence spectrum of

Figure 16A was obtained by the subtraction procedure

described earlier. As can be seen in Figure 16, the fluores-

cence signal was about 30% of the total signal in the nitrous

oxide-acetylene flame. The fluorescence signals were

observed to decrease in the following order of the flames:

N20-C2H2 >A-20-C2H2 > air-C2H2 > Ar-air-H2 > Ar-air-C2H2

air-H2 > Ar-Ar/02-H2 > Ar/02-H2' where Ar- means the argon

separated flame, and Ar/02 indicates the mixture of approxi-

mately 80% argon and 20% oxygen which was substituted for


As has been reported previously (42,45), phosphate ion

is decomposed partially into phosphorus atoms and PO mole-

cules in the nitrous oxide-acetylene flame, while POx

(x 3 2) molecules are responsible for the primary absorption

bands in the air-acetylene flame (45). Taking account of

these facts along with the results of this work, the total

population of PO appears to be largest in the nitrous oxide-

acetylene flame. In the low-temperature flames of the air-

hydrogen and argon/oxygen-hydrogen mixtures, the correspond-

ing argon separated flames gave higher fluorescence

The pure fluorescence (A) and fluorescence
enhanced by emission (B) of PO in the argon
separated nitrous oxide-acetylene flame,
obtained from al1M II3PO4 solution aspirated
into the flame. Experimental conditions are
as listed for Figure 15.

Figure 16.



intensities than their unseparated counterparts. This may

be explained as follows: in the unseparated flames, the

recombination of PO with excess oxygen in the entrained

air occurs in the flame, decreasing the population of PO.

The intensity of fluorescence depends upon quenching

effects of the co-existing flame gases, and nitrogen gas is

known to have a large quenching cross section (10). There-

fore, in order to examine the quenching effect of nitrogen

gas on the PO fluorescence, the fluorescence signals of PO

were compared in the air-hydrogen and the argon/oxygen-

hydrogen flames, where the Ar/02 ratio was approximately the

same as that of N2/02 in air. Contrary to expectations,

the air-hydrogen flame, which contains nitrogen gas, pro-

duced higher PO fluorescence intensities than the argon/

oxygen-hydrogen flame (see Table X). This suggests that the

fluorescence of PO depends in a complicated manner upon the

flame temperature and the quenching effects of flame gases.

However, this experiment was duplicated on another system

using acetylene as the fuel gas, and a slightly enhanced

signal was observed for the nitrogen-free flame, i.e., the

reverse situation was observed. The need for more work in

this area is indicated.

In order to utilize the fluorescence of PO for the

analysis of phosphorus, the signal-to-noise ratio, precision,

and limit of detection have been determined for phosphoric

acid nebulized into the various flames. As can be seen in

Figure 15, the argon separated flames yielded lower noise

levels, by a factor of 2 or 3, than their unseparated

counterparts. It should be pointed out that the dominant

source of noise in the SIT-OMA system was the flame rather

than the detector. The experimental results indicated that

the argon separated air-hydrogen flame produced the best

signal-to-noise ratio and limit of detection.

As mentioned above, the nitrous oxide-acetylene and

air-acetylene flames produced higher fluorescence intensi-

ties than other flames, but the background flame noises

(and PO fluorescence due to PH3 in acetylene) were still

present and hence the signal-to-noise ratios associated

with these flames were.10wer. In the present work,

phosphine in acetylene resulted in a fluorescence signal

equal to about 25% of that for al1M H3PO4 solution

aspirated into the air-acetylene flame (see Table X).

The dependence of the fluorescence intensity of PO

on the slit width of the monochromator is shown in Figure 17,

and fluorescence spectra of PO recorded while employing

an 850 pm slit width (Part A) and a 100 pm slit width

(Part B) are shown in Figure 18. As can be seen in the

spectra of Figure 18, the fine structure of the PO band

was lost when the 850 pm slit was used, but the band

intensity was greatly enhanced. The plot of fluorescence

signal versus slit width (Figure 17) reached essentially a

plateau for slit widths greater than 750 pm. The limit of



Figure 17.

Fluorescence intensity of PO vs
monochromator entrance slit width for
an argon separated air-hydrogen flame.
Solution and experimental conditions
are as listed for Figure 15.



Figure 18,

PO fluorescence spectra in the argon
separated air-hydrogen flame. Slit widths
of 850 pm (A) and 100 pim (B) were used.


detection for phosphorus was about 80 gg .P/ml (for a signal-

to-noise ratio of 2), and the precision was better than 5%

RSD for a 0.5 M H PO4 solution. The analytical curve for

PO fluorescence versus phosphoric acid concentration was

linear from 0.01 M to 1 M H3PO4. Therefore, self-absorption

processes for PO in the flame did not appear to be impor-

tant even at very high phosphorus concentrations.

The ability of the SIT-OMA system to discriminate

against noise appears to be significantly inferior when

compared with the more conventional signal processing and

detection systems used in the studies of the previous

chapters. This conclusion is in general agreement with

most.of the other published evaluations of the SIT-OMA




The pulsed Eimac xenon arc continuum source presently

offers little improvement as a source of excitation in

atomic fluorescence spectrometry over the same lamp

operated in the ow mode. A lamp especially suited to

pulsed operation is needed, along with a way to reduce or

eliminate the particularly severe background interference

from molecular fluorescence (and scatter) associated with

the operation of the pulsed continuum source, such as removal

of the interfering species from the flame, if possible,

and/or manipulation of flame conditions.

Flame background molecular fluorescence has been

shown to represent a potentially serious form of inter-

ference in atomic fluorescence spectrometry with a continuum

source, and along with improvements in source intensities

and signal-to-noise ratios these interference should

become more prominent, more numerous and, therefore, more

troublesome. The types of studies initiated here will have

to be extended to other flames and to different observation

heights in the flames to see if there is a way to experi-

mentally minimize this problem. A mathematical expression

giving the total deteriorative effect of the absorption,

emission, and fluorescence of a molecular band upon the

observed atomic fluorescence signal-to-noise ratio should

prove to be of considerable value in assessing the extent

of interference from molecular species in practical analyti-

cal work.

The possibility of direct spectral interference from

fluorescing molecular species introduced into the flame by

aspiration of the sample is also a serious concern. Exten-

sive tests of various commonly occurring sample matrices

(e.g., blood plasma, engine oils, dissolved vegetable matter,

dissolved mineral ores, etc.) will be necessary to provide

some idea of the potential severity of this problem.

The question of possible utilization of flame molecular

fluorescence for the analysis of non-metallic elements and

compounds has only just begun to receive attention from

analytical chemists. Because molecular bands are fairly

broad in comparison to atomic lines, the application of a

sensitive spectrometer with a broad spectral bandpass and

good noise discrimination characteristics should enable

researchers to effectively evaluate this possibility for a

wide variety of non-metallic elements and compounds. Also,

the potential for application of non-resonance molecular

fluorescence (using a broad-band excitation monochromator

or broad-band tunable dye-laser) to this area of research

offers considerable promise for improvements in sensitivity.

Finally, the fluorescence of molecules in flames will

have to undergo an extensive characterization from a more

fundamental point of view. The attempts described here to

study the quenching effect of nitrogen were inconclusive

because different results were obtained in separate experi-

ments. Human and Zeegers (35) found molecular fluorescence

quantum efficiencies to be comparable in magnitude to

efficiencies reported for atoms but were unable to inter-

pret their findings with regard to the yet unexplained

fine structure of the molecular fluorescence bands. They

also learned that absorbed radiant energy underwent exten-

sive re-distribution among the various vibrational and

rotational energy levels before being re-radiated as

fluorescence. These types of studies will have to be

repeated and extended in order to learn more about flame

molecular fluorescence and fundamental flame processes.


1. R. W. Wood, Proc. Am. Acad., XI, 396 (1904).

2. E. L. Nichols and H. L. Howes, Phys. Rev., 23, 472

3. C. T. J. Alkemade in "Proceedings of the Xth Colloquium
Spectroscopicum Internationale," E. R. Lippincott and
M. Margoshes, Ed., Spartan Books, Washington, D.C., 1963.

4. J. D. Winefordner and T. J. Vickers, Anal. Chem., 36,
161 (1964) .

5. D. R. Jenkins, Spectrochim. Acta, Part B, 23, 167 (1967).

6. R. F. Browner, Analyst (London), 99, 617 (1974).

7. R. C. Elser, T. C. O'Hlaver, M. C. Parsons, P. St. John,
C. Veillon,and J. D. Winefordner, "Trace Analysis:
Spectroscopic Methods for Elements," in Chemical
Analysis Series, vol. 46, J. D. Winefordner, Ed.,
Wiley-Interscience, New York, 1976, p. 167.

8. G. Rossi and N. Omenetto, Appl. Spectrosc., 21, 329

9. W. G. Jones and A. Walsh, Spectrochim. Acta, Part B,
16, 249 (1960).

10. J. D. Winefordner, S. G. Schulman,and T. C. O'H~aver,
"Luminescence Spectrometry in Analytical Chemistry,"
in Chemical Analysis Series, Vol. 38, P. J. Elving
and I. M. Kolthoff, Eds., Wiley-Interscience, New
York, 1972, chapters 3 and 4.

11. K. W. Busch and G. H. Morrison, Anal. Chem., 45, 713A
(1973) .

12. R. G. Michel, J. Coleman,and J. D. Winefordner, Spectro-
chim. Acta, Part B, to be published.

13. J. D. Winefordner and R. C. Elser, Anal. Chem., 43,
24A (1971).

14. J. D. Winefordner, V. Svoboda,and L. J. Cline, CRC
Crit. Rev. Anal. Chem., 1, 233 (1970).

15. N. Omenetto, L. M. Fraser,and J. D. Winefordner,
Appl. Spectrosc. Rev., 7, 147 (1973).

16. D. G. Mitchell, Paper presented at the Technicon
International Congress, New York, N.Y., 1970.

17. D. J. Johnson, F. W. Plankey,and J. D. Winefordner,
Anal. Chem., 46, 1898 (1974).

18. D. J. Johnson, F. W. Plankey,and J. D. Winefordner,
Anal. Chem., 47, 1739 (1975).

19. W. K. Fowler, D. O. Knapp,and J. D. Winefordner,
Anal. Chem., 46, 601 (1974).

20. R. Mavrodineanu and H. Boiteux, "Flame Spectroscopy,"
John Wiley & Sons, New York, 1965, pp. 60-63.

21. G. Beck, Rev. Sci. Instrum., 45 318 (1974).

22. D. J. Johnson, F. W. Plankey,and J. D. Winefordner,
Can. J. Spectrosc.,19, 151 (1974).

23. K. M. Aldous, R. F. Browner, R. M. Dagnall,and T. S.
West, Anal. Chem., 42, 939 (1970).

24. F. Burriel-Marti and J. Ramirez-Munoz, "Flame Photometry,"
3rd ed., Elsevier, New York, 1964, p. 40.

25. T. J. Vickers and J. D. Winefordner, "Analytical
Emission Spectroscopy," Part II, in Analytical
Spectroscopy Series, Vol. 1, E. L. Grove, ed.,
Marcel Dekker, New York, 1972, pp. 310-312.

26. E. E. Pickett and S. R. Koirtyohann, Anal. Chem., 41,
28A (1969).

27. R. W. B. Pearse and A. G. Gaydon, "The Identification
of Molecular Spectra," 3rd ed., Chapman and Hall,
London, 1965, pp. 51-320.

28. H. Haraguchi, N. Furuta, E. Yoshimura,and K. Fuwa,
Anal. Chem., to be published.

29. A. G. Gaydon, G. N. Spokes,and J. vanSuchtelen,
Proc. Roy. Soc. (London), A 315, 129 (1970).

30. C. D. West, Anal. Chem., 46, 797 (1974),

31. R. H. Barnes, C. E. Moeller, J. F. Kircher,and C. M.
Verber, Appl. Opt., _12, 2531 (1973).

32. T. Carrington, Paper presented at the 8th Symposium
on Combustion, Pasadena, Ca., 1960.

33. C. Veillon, J. M. Mansfield, M. L. Parsons,and J. D.
Winefordner, Anal. Chem., 38, 204 (1966).

34. D. R. Jenkins, Paper presented at the 2nd International
Atomic Absorption and Fluorescence Conference, Sheffield,
England, 1969.

35. H. G. C. Human and P. J. Th. Zeegers, Spectrochim.
Acta, Part B, 30, 203 (1975).

36. J. I. Dinnin, Anal. Chem., 40, 1825 (1968).

37. J. P. S. Haarsma, J. Vlogtman,and J. Agterdenbos,
Spectrochim. Acta, Part B, 31, 129 (1976).

38. F. Grum and G. W~. Luckey, Appl. Opt., 7, 2289 (1968).

39. J. D. Winefordner, Chemtech, 1975, 123.

40. P. T. Gilbert, "Analytical Flame Spectroscopy," R.
Mavrodineanu, Ed., MacMillan, London, 1970.

41. K. Fuwa and B. L. Vallee,Anal. Chem., 41, 188 (1969).

42. H. Haraguchi and K. Fuwa, Anal. Chem., 48, 784 (1976).

43. R. Herrmann and C. T. J. Alkemade, "Chemical Analysis
by Flame Photometry," trans1. by D. P. T. Gilbert,
Interscience Publishers, New York, 1963, pp. 215-230.

44. A. G. Gaydon, "Dissociation Energies and Spectra of
Diatomic Molecules," 3rd ed., Chapman and Hall,
London, 1968.

45. K. Fuwa, H. Haraguchi, K. Okamoto,and T. Nagata,
Bunseki Kagaku (Japan Anal.) 21, 945 (1972).

46. J. A. Fiorino, R. N. Knisely,and V. A. Fassel, Spectro-
chim. Acta, Part B, 23, 413 (1968).

47. H. Haraguchi and K. Fuwa, Spectrochim. Acta, Part B,
30, 535 (1975).

48. S. R. Koirtyohann and E. E. Pickett, Anal. Chem., 37,
601 (1965) .

49. S. R. Koirtyohann and E. E. Pickett, Anal. Chem., 38,
585 (1966) .

50. K. Fujiwara, H~. Haraguchi,and K. Fuwa, Anal. Chem.,
47, 743 (1975).

51. H. Haraguchi and K. Fuwa, Chem. Lett., 1972, 913.

52. H. Haraguchi and K. Fuwa, Bull. Chem. Soc. Japan,
48, 3056 (1975).

53. H. Haraguchi, M. Shiraishi,and K. Fuwa, Chem. Lett.,
1973, 251.

54. M. Margoshes, Spectrochim. Acta, Part B, 25, 113 (1970).

55. P. W. J. M. Boumans and G. Brouwer, Spectrochim. Acta,
Part B, 27, 247 (1972).

56. G. Horlick and E. G. Codding, Anal. Chem., 45, 1490

57. D. G. Mitchell, D. W. Jackson,and K. M. Aldous, Anal.
Chem., 45, 1215A (1973).

58. K. W. Busch, N. G. Howell,and G. H. Morrison, Anal.
Chem., 46, 575 (1974).

59. K. M. Aldous, D. G. Mitchell,and K. W. Jackson, Anal.
Chem., 47, 1034 (1975).

60. D. O. Knapp, N. Omenetto, L. P. Hart, F. W. Plankey,
and J. D. Winefordner, Anal. Chim. Acta, 69. 455
(1974) .

61. Y. Talmi, Anal. Chem., 47, 658A (1975).

62. T. L. Chester, H. Haraguchi, D. O. Knapp, J. Messman,
and J. D. Winefordner, Appl. Spectrosc., 30, 410 (1976).


William Kelly Fowler was born in Lexington, Kentucky

on September 10, 1948. He attended public schools in

Owensboro, Kentucky until graduation from high school in

June, 1966. After graduation from Centre College of

Kentucky with a Bachelor of Arts degree in Chemistry

(1970), he taught chemistry, physics, and physical science

in high school for one year. In 1971, he entered the

University of Florida as a graduate student in the

Department of Chemistry, college of Arts and Sciences.

In addition to his teaching experience, he has spent several

summers employed in either laboratory or research assistant


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.

ines D. Winefor ~er, Chairman
professor of Che istry

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.

Roger .L Bates
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.

Step n G. Schulman 4
Professor of Pharmaceutical

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 Philosop

Kuang Pang Li
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.

G'erhard M. Schm~id
Associate Professor of Chemistry

This dissertation was submitted to the Graduate Faculty of
the Department of Chemistry in the College of Arts and
Sciences and to the Graduate Council, and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.

December, 1976

Dean, Graduate School

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs