Title: Atomic fluorescence flame spectrometry as a means of chemical analysis
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00097947/00001
 Material Information
Title: Atomic fluorescence flame spectrometry as a means of chemical analysis
Physical Description: vii, 87 l. : illus. ; 28 cm.
Language: English
Creator: Staab, Robert Allan, 1937-
Publication Date: 1964
Copyright Date: 1964
Subject: Fluorescence   ( lcsh )
Radiation   ( lcsh )
Flame photometry   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis - University of Florida.
Bibliography: Bibliography: l. 84-86.
Additional Physical Form: Also available on World Wide Web
General Note: Manuscript copy.
General Note: Vita.
 Record Information
Bibliographic ID: UF00097947
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 - 000423896
oclc - 11024128
notis - ACH2301


This item has the following downloads:

PDF ( 3 MBs ) ( PDF )

Full Text






April, 1964


The author wishes to express his sincere apprecia-

tion to his research director, Dr. J. D. Winefordner, for

the advice, guidance, and encouragement given during this

investigation and throughout the author's graduate studies.

He would also like to thank the other members of his super-

visory committee, Dr. L. A. Arnold, Dr. J. M. Pearce, Dr.

T. W. Stearns, and Dr. R. C. Stoufer for their help.

Special thanks are due many of the author's colleagues

for their helpful suggestions.

Finally, the author wishes to express his indebted-

ness to his wife, Cherry, for her assistance, encouragement,

and the typewriting of this dissertation.


ACKNOWLEDG ,EJTS .............. . ........ ..... .. .

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

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


I. INTRODUCTION .................................

II. HISTORY ..................

III. THEORY.......................................


A. Apparatus .................................

B. Solutions .................................

0. Procedure ...............................

V. RESULTS AND DISOUSSIONS.......................

A. Types of Working Ourves..................

B. Types of Gases Added to Fuel Gas..........

C. Types of Gases Added to Burner Sheath.....

D. Comparison of Types of Flame Spectrometry.

E. Accuracy and Signal-to-Noise Ratio........


A. Shapes of Working Curves..................

B. Improvement of Equipment..................






















Chapter Page

C. Sensitized Fluorescence................ .. 51

D. The Value of Atomic Fluorescence Flame
Speotrometry as a Means of Chemical
Analysis ........ ................ ...... 51

VII. SUMMARY ............................. ....... 53

APPENDIX A, WORKING CURVES............................ 55

RATIO ......................... ....... ...... 74

LITERATURE CITED ................... ................ 84

BIOGRAPHICAL SKETCH.................................. 87


Table Page

1. Operating Conditions for Atomic Fluorescence
Excitation Sources .............................. 24

2. Wavelengths Used for Measurements of Atomic
Flame Fluorescence.................... .......... 25

3. Fuel Gas Flow Rates for Various Elements and
Flames When Using Beckman Total-Consumption
Atomizer-Burners.. ............ ................. 29

4. Flame Temperatures for Various Elements When
Using Beckman Total-Consumption Atomizer-
Burners ...... ................................. .. 35

5. Limit of Detection in p.p.m. for Various
Elements for Atomic Fluorescence Flame Spec-
trometry ........................... ............ 40

6. Limit of Detection in p.p.m. for Various
Elements for Atomic Emission Flame Spectrometry. 41

7. Limit of Detection in p.p.m. for Various
Elements for Atomic Absorption Flame Spec-
trometry......................... ............. . 42

8. Effect of Gases Added to Fuel Gas and Sheath,
Percentage Standard Deviations, and Signal-to-
Noise Ratios for Atomic Fluorescence keasure-
ments of Zn 2139 A................................... 76

9. Effect of Gases Added to Fuel Gas, Percentage
Standard Deviations, and Signal-to-Noise Ratios
for Atomic Fluorescence Measurements of Cd
2288 A............ ........ ...................... 78

10. Effect of Gases Added to Fuel Gas and Sheath,
Percentage Standard Deviations, and Signal-to-
hoise Ratios for Atomic Fluorescence Measure-
ments of Hg 2537 A............................. 80

11. Effect of Gases Added to Fuel Gas, Percentage
Standard Deviations, and Signal-to-Noise Ratios
for Atomic Fluorescence Measurements of T1
3776 A.......................................... 82


Figure Page

1. Block Diagram of Experimental Setup for Atomic
Fluorescence Flame Spectrometry................. 11

2. Beckman Custom Sheath Attachment................ 18

3. Chamber Atomizer Fitted to the Modified Meker
Burner ............. .... ...... ..... ...... 20

4. Position of Excitation and Measurement of Atomic
Fluorescence with the (a) Beckman Total-
Consumption Atomizer-Burner Flame and with the
(b) Meker Burner Flame ......................... 27

5. Atomic Fluorescence Working Curves for Mercury
(Hg 2537 A. Line) in Oxyhydrogen and Oxyacety-
lene Flames Using a Vapor Discharge Tube and an
Electrodeless Discharge Tube.................... 56

6. Atomic Fluorescence Working Curves for Zinc
(Zn 2139 A. Line) Introduced as Aqueous Solu-
tions into Different Flames..................... 57

7. Atomic Fluorescence Working Curves for Cadmium
(Cd 2288 A. Line) Introduced as Aqueous Solu-
tions into Different Flames..................... 58

8. Atomic Fluorescence Working Curves for Mercury
(Hg 2537 A. Line) Introduced as Aqueous Solu-
tions into Different Flames Excited by the
Electrodeless Discharge Tube................... 59

9. Atomic Fluorescence Working Curves for Thallium
(T1 3776 A. Line) Introduced as Aqueous Solu-
tions into Different Flames..................... 60

10. Atomic Fluorescence Working Curves for Zinc
(Zn 2139 A. Line) in Different Solvents Intro-
duced into the Oxyhydrogen Flame................ 61

11. Atomic Fluorescence Working Curves for Zinc
(Zn 2139 A. Line) in Different Solvents Intro-
duced into the Oxyacetylene Flame............... 62

12. Atomic Fluorescence Working Curves for Cadmium
(Cd 2288 A. Line) in Different Solvents Intro-
duced into the Oxyhydrogen Flame................ 63

13. Atomic Fluorescence Working Curves for Cadmium
(Cd 2288 A. Line) in Different Solvents Intro-
duced into the Oxyacetylene Flame............... 64

14. Atomic Fluorescence Working Curves for Mercury
(Hg 2537 A. Line) in Different Solvents Intro-
duced into the Oxyhydrogen Flame Excited by the
Electrodeless Discharge Tube.................... 65

15. Atomic Fluorescence Working Curves for Mercury
(Hg 2537 A. Line) in Different Solvents Intro-
duced into the Oxyacetylene Flame Excited by
the Electrodeless Discharge Tube................ 66

16. Atomic Fluorescence Working Curves for Thallium
(Tl 3776 A. Line) in Different Solvents Intro-
duced into the Oxyhydrogen Flame................ 67

17. Atomic Fluorescence Working Curves for Thallium
(Tl 3776 A. Line) in Different Solvents Intro-
duced into the Oxyacetylene Flame............... 68

18. Atomic Absorption Working Curves for Zinc
(Zn 2139 A. Line) Introduced into Different
Flames.......................................... 69

19. Atomic Absorption Working Curves for Cadmium
(Od 2288 A. Line) Introduced into Different
Flames ...... ... ....... ......................... 70

20. Atomic Absorption Working Curves for Mercury
(Hg 2537 A. Line) Introduced into Different
Flames Using the Electrodeless Discharge Tube... 71

21. Atomic Absorption Working Curves for Thallium
(Tl 3776 A. Line) Introduced into Different
Flames ................................ ......... 72
22. Atomic Emission Working Curves for Thallium
(Tl 3776 A. Line) Introduced into Different
Flames....................... ....... ........ 73





Atomic fluorescence flame spectrometry is a new method

of chemical analysis. The principles on which the method

is based have existed for several decades, but only recently

has atomic fluorescence spectrometry been applied to chemical

analysis. Because the technique is based on fluorescence

emission by an atomic vapor excited by absorption of radia-

tion, the flame was chosen as the best means of conversion

of a solution sample into an atomic vapor and as a "con-

tainer" for this vapor.

Factors which influence atomic emission and atomic

absorption flame spectrometry also affect the results ob-

tained by atomic fluorescence. In addition, the use of a

light source poses several problems peculiar to fluorescence

methods. In this investigation it was attempted to use

atomic fluorescence flame spectrometry as a means of analy-

sis as well as to measure several of the more common factors

which apply to flame spectrometry. All sources which were

available were used for excitation of the atomic vapor.

Several types of flames were compared to determine the best

choice for a given analysis. The use of aqueous and non-

aqueous solvents, the addition of foreign gases to the fuel

gas, and the choice of sheath gas surrounding the flame were

also investigated. Comparisons of the results with existing

flame spectrometric techniques were made to determine

the value of the method for practical analyses.

The historical development of atomic fluorescence

flame spectrometry was traced from the first experiments

in atomic fluorescence in the early twentieth century to

the present date. A brief discussion of the theory of

atomic fluorescence flame spectrometry is given to provide

a basis for understanding which factors would tend to in-

fluence the results to the greatest extent. Future in-

strumental improvements and applications are also discussed.

The elements zinc, cadmium, mercury, and thallium

were found to fluoresce with high enough intensity that the

method of atomic fluorescence flame spectrometry could be

used for their analysis. This technique should be par-

ticularly suited to the routine determinations of mercury.

Other elements should be added to this list as soon as

suitable sources of excitation are available.


Nitchell and Zemansky (19) and Pringsheim (22)

have discussed much of the original work on fluores-

cence of atomic vapors. In the last decade of the nine-

teenth century, the sodium D-lines were found to appear

when sodium vapor was excited by sunlight or by a sodium

flame. A band system of the Na2 molecule (42) was also

observed. A short time later Wiedemann and Schmidt (33),

Wood (40, 41, 46), and Puccianti (23) failed to obtain

the emission of the D-lines from sodium vapor excited

by the D-lines. Wood (43, 46), however, observed the

emission of the Na D-lines from a tube of sodium vapor

excited by the D-lines from a sodium flame; the fluores-

cence radiation obtained consisted of only the D-lines

themselves and showed a decrease in intensity for higher

vapor pressures of sodium.

Wood (40, 41, 43, 46, 47, 48) and Dunoyer (9, 10,

11) thoroughly investigated the fluorescence of sodium

vapor. Dunoyer obtained the fluorescence effect from a

beam of fast-moving sodium atoms at the point of excita-

tion. He showed that the resonance radiation was due to

the atoms and that the time between the absorption and

the subsequent emission of the light was quite short.

He found that the resonance radiation was emitted only


from the point where the exciting radiation crossed the

atomic beam, and little or no spreading of the resonance

radiation occurred in the direction of motion of the

atoms. Dunoyer showed that only the center of the D-line

is useful for producing fluorescence. Sources with re-

versed lines gave a decrease in the amount of fluorescence


Bogros (5) obtained fluorescence of an atomic beam

of lithium by illumination with the radiation of a lithium

flame. In 1912 Wood (44, 45) obtained only the 2537 A.

line in the fluorescence of mercury vapor because the quartz

mercury arc used gave only the unreversed line at 2537 A.

Terenin (31, 32) obtained the fluorescence of cadmium

vapor (the 2288 A. and 3261 A. lines) by using a vacuum

arc in cadmium. The 2288 A. or the 3261 A. fluorescence

line was excited by a source giving radiation of a wave-

length the same as that observed in fluorescence. Filters

were used to eliminate all but the desired excitation

source line. Zinc fluorescence was obtained (21, 28, 34)

by excitation of zinc vapor.

Kany atoms have energy levels of magnitude such that

they will exhibit both resonance radiation and line fluores-

cence. An atom may absorb a line from the source, being

raised to an excited energy level; then it may return to

the lowest state, emitting resonance radiation, or may

drop to a low-lying metastable state, giving line fluores-

cence. The fluorescence of thallium, initially investigated

by Terenin (31, 32), is an example. Absorption of the

3776 A. line can give either the 3776 A. or 5350 A. lines
as fluorescence. The 3776 A. line is the resonance line,

and the 53_0 A. line is due to a transition to the meta-
stable 62P3/2 state. Similarly the 2768 A. thallium line

can give rise to both the 2768 A. and 3530 A. lines.
Terenin also investigated the fluorescence of lead, bis-
muth, arsenic, and antimony (31, 32).

Strutt (29, 30) and later Christensen and Rollefson

(6) excited sodium vapor with the 3303 A. doublet from a

vacuum arc in sodium and obtained the 3303 A. doublet and

the u-iines in the fluorescence. A step-like process was

proposed such that the absorption of the radiation of the

3303 A. lines raises the atom to the 42p1/2,3/2 states.
Strutt obtained resonance line fluorescence (Na D-lines)

as a result of 3303 A. absorption. The mechanism of de-

activation of the excited sodium atom could be described
by either of the two processes below.

(I) 42P-? 42S-- 2p--)32S

(II) 42p 3432D--432p 32S

The latter step in each gives rise to the D-lines, and
the first two steps give emission in the infra-red region.

Strutt (30) also examined the breadth of the resonance

D-lines and found agreement with his predictions based

on the Doppler effect. Wood and Dunoyer (47) found that

excitation with only one D-line gave only that line as

fluorescence. Wood and Iohler (48, 49) later showed

that this is true only for pure sodium vapor. The

presence of foreign gases results in energy transfer

to the other D-line.

Fuchtbauer (14) was the first to investigate the

absorption of radiation by excited atoms and their sub-

sequent fluorescence. In his work on mercury vapor, he

showed that after the necessary excitation by absorption

of the 2537 A. line and subsequently other lines of longer

wavelength, considerable fluorescence was observed. Absorp-

tion of the 2537 A. line and subsequent absorption of the

3131.5 A. and 3125 A. lines will give the 5770 A., 3654 A.,

3131.5 A., and 3125 A. lines as fluorescence in addition
to the resonance radiation. Absorption of the 3131.8 A.

line by atoms excited by the 2537 A. line will give the

2967.5 A., 3131.8 A., and 3663.2 A. lines with the 2537 A.

line. Bender (3) obtained similar effects for cadmium and


Fluorescence of species in flames was first reported

by Nichols and Howes (20) as a result of their work on

photoluminescence of flames. They obtained slight in-

creases in the emission of calcium, strontium, barium,

lithium, and sodium by excitation of the flame supplied

with these elements with various sources, none of which

contained any appreciable amount of the element excited.

A thallium flame excited with a mercury arc containing

thallium showed an increase in intensity of 7%. This

increase was ascribed to the effect of the thallium

lines of the source. Badger (2) thoroughly investigated

flame fluorescence for a wide variety of elements and

noted the slight effect of the flame itself on fluores-


At a more recent date Robinson (26) observed a small

amount of fluorescence of the Mg 2852 A. line when an oxy-

hydrogen flame into which was atomized a 1000 p.p.m. solu-

tion of magnesium was excited by a magnesium hollow cathode

discharge tube. Under similar conditions he did not observe

fluorescence from the Na 5890 A. line, the Ni 3524.5 A.

line, or the Ni 3414 A. line.

At the International Spectroscopy Colloquium in 1962

Alkemade (1) told of the use of atomic fluorescence of

flames as a means of determining quantum yields and re-

ferred to its possible use as a means of analysis. He

briefly compared atomic fluorescence and atomic absorption

flame photometry and described the experimental setup used

to measure the quantum efficiency of the Na 5890 A. line

with respect to flame gas composition. Earlier Boers,

Alkemade, and Smit (4) had used this method for the measure-

ment of the quantum efficiency of the Na 5890 A. line for

sodium vapor in a propane-air flame.


The theory of atomic fluorescence flame spectrometry

has been discussed by Winefordner and Vickers (39).

Atomic fluorescence flame spectrometry is based on the

absorption of radiation of a particular wavelength by

atoms in the vapor state and their subsequent deactivation

by emission of radiation of the same or longer wavelength.

Both the absorbed and emitted wavelengths are characteristic

of the atoms which are undergoing the transitions.

In the case of atomic emission flame spectrometry the

atcms are excited thermally and the emission is measured.

For atomic absorption flame spectrometry the decrease in

intensity of the characteristic radiation from a source

due to absorption by the atomic vapor in the flame is the

measured quantity. Atomic fluorescence flame spectrometry

measures the emission resulting from atoms excited by means

of absorption of radiation. Of course, any thermal emission

present must be subtracted from the total emission measured.

Atomic fluorescence may be divided into four types

(39) according to the manner of excitation and emission.
Resonance fluorescence refers to absorption and emission

at the same wavelength. Absorption of resonance radiation

and emission due to a radiational transition to a meta-

stable state is called direct line fluorescence. Stepwise


line fluorescence occurs when the atom is raised to an

excited level by absorption of resonance radiation and

then emits radiation at some longer wavelength. The

fluorescence emission arises from these radiational de-

activations to the ground state. Sensitized line fluores-

cence refers to collisional activation of an atom by a

foreign atom which has been activated by absorption of

its characteristic resonance radiation.

The intensity of the fluorescence emitted radiation
should depend on the amount of exciting radiation absorbed,

on the conversion efficiency of absorbed to emitted radia-

tion, on the amount of emitted radiation absorbed by ground

state atoms, and on the intensity of the excitation source.

The fluorescence intensity, Ip, for resonance fluorescence

is given (39) by equation 1,

IF = pA (1-e L)e-kL)-kL/2cosh(kL/2) watts/cm?-ster., (1)

where 0 is the quantum efficiency accounting for loss of

energy by processes other than fluorescence; Pois the inci-

dent radiant power, in watts, per unit frequency interval,
in sec.- ; A is the half the base width of the triangle

whose area represents the total power absorbed by the line,

in sec.- ; Af is the area of the flame, in cm.2; kO is the

atomic absorption coefficient at the line center, in cm."-

and L is the average path length of the radiation through

the flame, in cm.

The atomic absorption coefficient at the line center,

k, is directly proportional to No, the ground state con-

centration of the absorbing atoms, in atoms/cm.3 of flame

gases. Thus for a given element in an invariant flame

sample cell the fluorescence intensity, IF, should be

directly proportional to the incident radiant power, P.

Variations in flame conditions would affect I primarily

through changes in L, Af, and A within a small flame tem-

perature range. Using a constant excitation source a plot

of Ip against No, or concentration, should be linear. At

higher concentrations the linearity should disappear since

with increasing No the term (1-e-k L) approaches unity and
e-k/2 approaches zero, causing the curve to go through

a maximum and then decrease. At sufficiently high values

of No the quantum efficiency, A, may decrease due to self-

quenching (19, 22).

A typical experimental arrangement (39) shown in

Figure 1 for atomic fluorescence flame spectrometry would

consist of an intense, unreversed source of radiation which

is focused on a flame containing an atomic vapor of the

element being investigated. At right angles to the beam

of radiation from the source to the flame, the optical

axis of the monochromator is aligned in such a manner that

the two axes intersect at the flame cell. At the exit slit

of the monochromator a photomultiplier tube detector is

mounted and connected to a sensitive amplifier. The ampli-

fied signal is then fed to either a meter or a recorder.

Excitation Source

> Lens

Photo Detector

Flame Cell


d.c. Amplifier

Figure 1. Block Diagram of Experimental Setup for
Atomic Fluorescence Flame Spectrometry




Any source emitting radiation at the wavelength re-

quired may be used providing it is intense and not reversed.

However, a very wide line source will result in a decreased

fluorescence signal compared to a narrow line source of the

same intensity because absorption occurs only over the

width of the absorption line. A continuum source could be

used for excitation if it were sufficiently intense over

the with of the absorption line and if a monochromator of

small spectral band width were used to decrease the scat-

tering signal.

Vhen fluorescence is measured at the same wavelength

as that usea for excitation, scattering of the source radia-

tion is usually the main factor in determining the limit of

detection. In cases where the fluorescence is measured at

a wavelength longer than that used for excitation, scattering

can be eliminated by use of a monochromator or narrow band


Atomizer-burners useful for atomic emission and atomic

absorption flame spectrometry are also useful for atomic

fluorescence. The total-consumption type is common and

easy to use but has the disadvantages of having a small

flame size and of giving light scattering due to unevaporated

solvent droplets. Chamber-type atomizers have less scat-

tering problems due to smaller droplets being introduced

into the flame and do provide a flame cell of greater size,

but they have the disadvantages of being cumbersome as well

as extremely wasteful of sample solution. In addition, the

chamber-type atomizers are used with low temperature

flames which results in incomplete compound dissociation

in many cases.

Any monochromator which is useful for atomic emission

can be used for atomic fluorescence measurements if its

range includes the ultraviolet. Better sensitivity might

be expected, however, from instruments with better dis-

persion and better entrance slits and optics. Below 3000 A.

a sensitive d.c. amplifier can be used because thermal emis-

sion is low, but above this wavelength an a.c. amplifier

tuned to the frequency of a chopper in the beam of the ex-

citing radiation should give the best results. Photomulti-

plier tube detectors and meter or recorder readout systems

would be the same in either case.


A. Apparatus

The experimental components were arranged as described

in the previous section on the theory of atomic fluores-

cence flame spectrometry. All components, individually

described below, when using total-consumption atomizer-

burners, were held in place by means of two triangular

optical benches (one 0.5 meter long, No. 22-702, and one

1.0 meter long, No. 22-704, The Ealing Corp., Cambridge

38, Kass.), which were mounted with the shorter bench

perpendicular to the longer bench at its midpoint on a

piece of 0.5 inch thick Bakelite. All components were

fastened to the optical benches with 13.5 mm. diameter

rods which were held in carriers (No. 22-728, The Ealing

Corp.). These carriers could be positioned at any point

on the optical benches. The monochromator-detector com-

bination and the burner were mounted on the longer optical

bench, and the excitation source was mounted on the shorter

optical bench. Figure 1 illustrates the relative place-

ment of the components.

In the cases where the Meker burner was used the mono-

chromator-detector assembly was mounted on blocks, because

the additional height made necessary by the taller burner

was not obtainable with the optical bench and carrier


arrangement. The burner and source were then mounted on

weighted-base ringstands and secured with universal labora-

tory clamps.

Since most of the elements considered were extremely

toxic (27), a hood made from a 3.5 x 7 x 7 inch box open

at the bottom and fitted with 1.75 inch diameter flexible

plastic-jacketed steel tubing at the top was positioned

over the flame. This tube was connected to the intake of

a centrifugal blower, and the vapors were carried from the

blower by similar tubing to a laboratory hood.

Sources. Zinc and cadmium Osram spectral lamps (The

Ealing Uorp.) with one-inch diameter holes cut in their

outer soft glass envelopes to allow unobstructed passage

of light from the quartz inner bulb were operated on a.c.

at the recommended current of 1.5 amperes using a step-up

transformer and ballast. The holes in the Osran lamps were

made by carefully heating the desired portion of soft glass

by means of an oxyacetylene flame and allowing the inner

vacuum to draw in the softened glass. The inverted glass

shell, which was quite thin, was then cracked and the edges

firepolished. The lamps were mounted in a medium prefocus

socket (No. 26-250, The Ealing Corp.) containing a mounting

pin in the center of its base for positioning in a carrier.

For mercury a Hanovia mercury vapor lamp (No. 90012,

Hanovia Chemical and Mfg. Co., Newark 5, N. J.) operated

with a reactive transformer (No. 23634, Hanovia Chemical

and Mfg. Co.), a Hg-S Osram spectral lamp (The Ealing Corp.)

operated at 1.1 amperes, a Philips Hg spectral lamp (The

Ealing Corp.) run at 0.9 ampere, and a mercury electrode-

less discharge tube were used. The mercury electrodeless

discharge tube (Ophthos Instrument Co., Rockville, Md.)

was held at one end by a thermometer clamp (No. S-19427,

E. H. Sargent and Co., Chicago 36, Ill.) in the center of

a 2450 me. resonant cavity (Ophthos Instrument Co.) having

an observation port. Power was supplied to the cavity by

a coaxial cable (No. 73624G2B, Raytheon Co., Waltham 54,

Mass.) from a 100 watt microwave power generator (Model

PGM-10X1, Raytheon Co.). The resonant cavity and the

Hanovia mercury lamp were held in place by means of uni-

versal laboratory clamps. The Hg-S Osram lamp was mounted

the same as the zinc and cadmium Osram lamps, and the

Philips lamp was placed in a standard candelabra screw-

base socket equipped with a mounting pin for placement in

a carrier.

In the case of thallium both the Osram lamp and the

electrodeless lamp were used. The Osram lamp (The Ealing

Corp.) was operated, as mentioned above, at 1.0 ampere and

the electrodeless lamp (Ophthos Instrument Co.) also was

operated as above. Electrodeless lamps (Ophthoe Instru-

ment Co.) were also used for gallium, indium, antimony,

and selenium.

When using the electrodeless lamps or the Hanovia

mercury vapor lamp, the radiation was focused on the flame

by means of a 5.0 mm. focal length quartz lens. This lens

was mounted in a holder (No. 22-810, The Ealing Corp.)

and then fixed in a carrier and positioned on the 0.5

meter optical bench. For the Osram and Philips spectral

lamps no lenses were used. The lamps were moved as close

to the flame as possible. However, in this case it was

necessary to shield the entrance slit of the monochromator

from the incident source radiation by placing flat black

baffles along the sides of the lamp.

In some cases air cooling of the electrodeless dis-

charge tubes was necessary. This was done by mounting a

small centrifugal blower below the resonant cavity so that

it would force a current of air through the center passage

of the cavity and over the discharge tube.

Atomizer-Burner. Beckman medium-bore total-consumption

atomizer-burners (Nos. 4020 and 4030, Beckman Instruments,

Inc., Aullerton, Calif.) and a Meker burner (No. 3-902,

Fisher Scientific Co., Pittsburgh 19, Pa.) fitted with a

chamber atomizer were used for all studies. The Beckman

burner was mounted on a rod (6 inches long by 13.5 mm.

diameter) and held in a carrier on the optical bench.

Under the atomizer capillary a small piece of aluminum

sheet was positioned to hold the sample cuvettes and was

mounted on a swivel so that it could be moved out from

under the capillary. The oxyacetylene burner was used for

both oxyacetylene and air-hydrogen flames, while the oxy-

hydrogen burner was used only for oxyhydrogen flames. The

latter burner was also fitted with a sheath attachment

Figure 2. Beckman Custom Sheath Attachment

(Custom made, Beckman Instruments, Inc.), shown in

Figure 2, for determining the effect of sheath gases

on fluorescence.

The Meker burner, used with natural gas and air,

was modified by sealing the air ports at the base of

the burner tube, drilling at right angles into the gas

fitting and mounting a stainless steel tube in the hole

so that fuel, air and atomized sample might be premixed.

A Beckman small-bore total-consumption atomizer-burner

(No. 4050, Beckman Instruments, Inc.) and a short piece

of glass tubing equipped with a length of rubber tubing

and a pinchclamp were inserted in a two-hole rubber stopper

which was placed in the lower end of a five-inch long and

one-inch diameter glass tube. The upper end of this tube

was closed with a one-hole rubber stopper holding a piece

of glass tubing bent at a right angle; this glass tubing

and the stainless steel tube added to the Meker burner

were joined with a short length of rubber tubing. The

Meker burner and chamber atomizer are shown in Figure 3.

Acetylene, hydrogen, oxygen, and air were regulated

by pressure reduction valves on the standard gas cylinders

and by a Beckman control panel (lo. 9220, Beckman Instru-

ments, Inc.). The flow rates of all gases were measured

by means of rotameters (No. 4-15-2, Ace Glass Inc., Vine-

land, N. J.) placed in the gas lines between the regulators

and the burner. Air for the ..cker burner was regulated

using the above system, and the natural gas was measured

Figure 3. L arr er Atomizer Fitted to the Modified
Seker Burner

by means of a rotameter and controlled by the gas jet

tap on the laboratory bench. Provision was made to add

gases to the sheath attachment or to the fuel gas just

before it entered the burner by means of a "Y" tube in

the fuel gas line. The sheath gases were controlled by

pressure reduction valves and needle valves on the gas

cylinders, and separate rotameters were used to determine

gas flow rates.

onorLchromator-Detector. A small, inexpensive Bausch

and oiao grating monochromator (No. 33-86-25, Bausch and

Lomb Uptical Co., Rochester 2, N. Y.) with UV grating (No.

33-86-01, Bausch and Lomb Optical Co.), visible grating

(ho. 33-86-02, Bausch and Lomb Optical Co.), and variable

slits (No. 33-86-31, Bausch and Lomb Optical Co.) was used

for all studies in this research. A 5 mm. horizontal slit

made from a piece of 1/16 inch aluminum was placed in front

of the entrance slit of the monochromator in order to limit

the height of the flame affecting the detector. Two pins

(3 inches long and 13.5 mm. diameter) were mounted on a
1/8 inch aluminum plate (5 x 8 inches) on the same axis

as the slits of the monochromator; the monochromator was

screwed to the plate. These pins were placed in carriers

which were used to position the monochromator on the one-

meter optical bench. An upright detector housing (No.

200C316, zldorado Electronics Co., Berkeley, Calif.) con-

taining aa RCA. 1P28 photomultiplier tube was placed at

the exit slit of the monochromator and held in place by

a light tight sleeve. A high gain, high stability d.c.

amplifier with meter readout and with a 10 mv. recorder

output (Model ph-200, Eldorado Electronics Co.) was con-

nected to the photomultiplier tube and used for amplifying

and displaying the input signal. Measurements were taken

on either a 10 mv. recorder (G-11A, Varian Associates,

Palo Alto, Calif.) or a 12.5 mv. recorder (Model-SR, E. H.

Sargent and Co.).

Atomic Absorption. Atomic absorption measurements

were taken using the same components as used for atomic

fluorescence, except that the source and flame were then

optically aligned with the monochromator. The source and

lens were placed so as to focus the source radiation on

the center of the flame at the desired position above the

burner tip. The transmitted radiation was then focused

by means of a second lens on the entrance slit of the mono-


B. Solutions

Stock solutions containing 1000 p.p.m. of the ele-

ments being investigated were prepared as described below.

All other solutions were prepared by successive dilutions.

Zinc, cadmium, and mercury solutions in 0.01 M HC1 were

prepared from ZnC12, CdCl2, and HgC12, respectively.

Thallium and indium solutions were made acidic and were

prepared from T1103 and In(NO ) *3H20. Antimony, gallium,

and selenium solutions were prepared from the metal by dis-

solving in acid and diluting to volume. Solutions of the

elements in 30% and 70% methanol in water solutions were

prepared by pipeting 10 ml. of the 1000 p.p.m. stock aqueous

solution into a 100 ml. volumetric flask, adding the nec-

essary volume of pure methanol to reach the proper methanol

to water ratio and then diluting to 100 ml. with the respec-

tive methanol-water solvent system.

C. Procedure

The metal vapor lamp was optically aligned to give

maximum intensity of source radiation on the flame with a

minimum amount of light reflected into the monochromator.

The lamp was turned on, and the current was adjusted to the

desired value. In the case of the electrodeless discharge

lamps the generator power was set as desired, and the cooling

blower was started if needed. The electrodeless lamps were

started by turning the microwave generator to full power

and igniting by means of a Tesla coil. This was done by

inserting a two-inch glass tube with a wire trimmed flush

with the open end into the resonant cavity observation

port. The wire was an extension of the Tesla coil elec-

trode. The voltage of the Tesla coil was then slowly in-

creased until the lamp started. In several cases it was

necessary to vaporize the compound in the lamp by heating

the lamp in the flame of a Meker burner to facilitate

starting of the lamp. The lamps were run under the con-

ditions listed in Table 1.

Table 1

Operating Conditions
Atomic Fluorescence Excitati

trial Zn



Philips spectral Hg

Hanovia vapor discharge Hg

Ophthos electrodeless discharge Hg






on Sources

t Conditions

1.5 amperes

1.5 amperes

1.1 amperes

1.0 ampere

0.9 ampere


40% power, air cooled

40% power, no cooling

60% power, air cooled

50% power, no cooling

70% power, no cooling

60% power, no cooling

Type Lamp

Osram spec

These values were obtained from either manufacturer's

recommendations or by measurement to determine maximum

intensity without having line reversal or losing stability.

The lamps usually were allowed about 15 minutes to stabi-

lize. Long term stability of the lamps resulted in neg-

ligible error as did the lamp flicker (approximately 2%

peak-to-peak). ; hile the lamp was warming up, the amplifier

circuit and photomultiplier high voltage power supply were

turned on and were allowed to warm up.

The monochromator was set to the wavelength being
measured by introducing a concentrated solution (1000 p.p.m.)

of the metal being analyzed into the flame and adjusting the

wavelength dial to obtain a maximum fluorescence signal.

The wavelengths chosen for the elements investigated are

given in Table 2.

Table 2

Wavelengths Used for
measurement t of Atomic Flame Fluorescence

Element .:avelength, A.

Zn 2137

Cd 2288

Hg 2537
Ga 4033, 4172

In 4102, 4511

T1 3776

Sb 2068, 2529, 2598, etc.

be 2040, 2591

For several reasons which will be discussed later,

fluorescence was not observed in the cases of gallium,

indium, antimony, and selenium. These elements will be

omitted from further discussion of experimental condi-


Before a series of standard solutions could be run

for obtaining a working curve and determining the limit

of detection, the position of excitation and measurement

of fluorescence in the flame, the slit width of the mono-

chromator, and the flame gas composition had to be opti-

mized. The position of the fluorescence in the flame was

determined by varying the burner height with respect to

the entrance slit of the monochromator while the excita-

tion source and the entrance slit of the monochromator

were 4ept at the same level. For zinc and cadmium the

best position in the flame was halfway between the tip

of the reaction zone and the tip of the outer cone of

the flame. These positions are illustrated in Figure 4a.

In the case of the Meker burner all measurements were taken

midway between the reaction zone and the tip of the outer

cone of the flame as shown in Figure 4b. The optimum

slit width was chosen such that the signal-to-noise ratio

(the signal due to fluorescence-to-the-flicker in the back-

ground, which was a result of random scattering of incident

radiation, flicker in the excitation source, and flicker in

the flame continuum) was a maximum without giving readings

impossible to measure with a reasonable degree of accuracy

Hg, Tl

Zn, Cd

Outer Cone

(a) (b)

Figure 4. Position of Excitation and Measurement
of Atomic Fluorescence with the (a) Beck-
man Total-Consumption Atomizer-Burner Flame
and with the (b) Meker Burner Flame

when operating near the limit of detection, i.e., at

high amplifier sensitivities. The exit slit was set to

0.56 of the value of the entrance slit for optimum spectral

resolution. Entrance slit widths for the Leker burner were

all iound to be 0.50 mm., and for the total-consumption

atomizer-burner they were 0.25 mm., 0.20 mm., 0.40 mm.,

and 0.25 mm. respectively for zinc, cadmium, mercury, and


Flame gas composition was chosen by measuring the

signal-to-noise ratio for a variety of flame compositions

and choosing the value which maximized the ratio. In all

cases where the Beckman burner was used, the oxygen or air

flow rate was fixed at 2500 cc./min. in order to hold the

solution flow rate constant for all measurements. The flow

rates of the fuel gases are given in Table 3. In all cases

for the ieker burner, the air flow rate was 2800 cc./min.

and that of the natural gas was 4000 cc./min.

The relative intensity due to the atomic fluorescence

of a particular line was obtained as follows. The sample

solution was introduced into the appropriate flame, and

the sensitivity multiplier of the amplifier was adjusted

to bring the reading into the range of the meter. Then

the solvent being used was introduced and a background

reading was taken at the same instrument setting. The

relative fluorescence intensity was obtained by subtracting

the background signal (due to incident light scattering by

solvent droplets in the flame and to the emission of

background radiation by the flame gaces) from the signal
due to the sample. As the concentration of the samples
introduced changed, the gain of the amplifier was in-
creased or decreased in known increments, and measurements
were obtained over a wide concentration range.

Table 3
Fuel Gas Flow Rates for Various Elements and
Flames When Using Beckman Total-Consumption Atomizer-Eurners
Element Flame Fuel flow rate, cc./min.
Zn H2/air 8000

H2/02 5000
C2H2/02 1000
Cd H2/air 8000

H2/02 5000
02C2/02 1000

Hg H2/air 10000

H2/02 8000
02H2/02 1200
T1 H2/air 8000

H2/02 8000
C2H2/02 1200

If the sample also emitted thermally, then an addi-
tional reading of thermal emission had to be made by
blocking off the source radiation and measuring the thermal
emission of the sample and the background due to the solvent

and taking the difference of the two values. This reading

for thermal emission was then subtracted from the reading

made for atomic fluorescence, because in this case it was

actually a reading of both fluorescence and emission. How-

ever, in the four instances where fluorescence was observed,

only the T1 3776 A. line exhibited measureable emission for

concentrations below 1000 p.p.m. using the experimental

setup described.

For a given element each series of readings using the

same flame was normalized to correct for small experimental

variations from day to day so that an accurate comparison

might be made. The normalizing procedure consisted of

first measuring the intensity of a specific standard solu-

tion, i.e., 10 or 100 p.p.m. aqueous solution, as a control

and then measuring the fluorescence signal of the system

under investigation. All control readings for the same

conditions were averaged. A normalizing factor for the

measurements made under these conditions was computed by

dividing the average control reading by the specific con-

trol reading for that series. All measurements in a series

were then multiplied by the normalizing factor calculated

from the control reading for that series.

The effect of argon, nitrogen, and oxygen as sheath

gases was investigated for zinc and mercury in the oxy-

hydrogen flame. The flow rates of each gas added to the

sheath was 1000 cc./min. I-.easurements of relative fluores-

cence intensity were made for 10 p.p.m. aqueous solutions

of the elements being investigated.

Addition of foreign gases to the flame by premixing

argon or nitrogen with the fuel gas was studied for zinc,

cadmium, mercury, and thallium in both oxyhydrogen and

oxyacetylene flames. The flow rate of the fuel gas was

the same as when no foreign gas was added and the flow

rates oi argon and nitrogen were 870 cc./min. and 1000

ec./min., respectively. Solutions of 10 p.p.m. were used

for zinc, cadmium, and mercury, and 100 p.p.m. solutions

were measured for thallium.

At least two readings of the fluorescence signal for

each concentration were made in preparing the working

curves. x-or percentage standard deviation calculations,

ten measurements of the fluorescence signals were made for

each of the following concentrations: 1, 10, and 100 p.p.m.

For thallium these measurements were made on 10, 100, and

1000 p.p.m. solutions rather than on concentrations of 1,

10, and 100 p.p.m. In the case of gases added to the

sheath and fuel gases, six readings were taken for each

gas added.

Atomic absorption measurements were made using the

experimental setup previously described. Readings were

made of source intensity with solvent and with sample in

the flame and of flame emission background with the source

blocked off from the flame and the monochromator-detector.

In all cases the amplifier gain was so low that no correc-

tion for thermal emission of the sample was necessary.

The absorbance for each sample was measured by taking the


logarithm of the ratio of the signal due to solvent only

in the flame to the signal due to sample in the flame.

The thermal emission signal was subtracted from the

sample signal prior to the calculation of absorbance.


A. Types of Working Curves

From the working curve plot in Figure 5 it is evi-

dent that the electrodeless discharge tube was much more

effective in producing atomic fluorescence for mercury

than was the Hanovia vapor discharge lamp. This was

probably due to the greater line intensity of the eloc-

trodeless lamp. The mercury electrodeless lamps have

been said to be stronger than a d.c. arc, a.c. arc, or

condensed spark and to have a Doppler half-width about

the same as a water-cooled hollow cathode discharge tube

(13). The lines emitted from the electrodeless lamp

essentially are not reversed (7). Other excitation sources

for mercury were tried, such as the Osram and Philips spec-

tral lamps, but in no case was fluorescence observed.

The working curves for thallium, Figures 9, 16, 17,

and 21, involved the use of the thallium electrodeless

discharge tube as an excitation source. The thallium

Osram spectral lamp was tried, but as in the case of the

two mercury lamps of this type, no fluorescence could be

obtained. Evidently, the lines produced were either

reversed or otherwise of insufficient intensity at the

line center. Zinc and cadmium Osram lamps do not fall


in the sane category, since they were extremely effec-

tive in producing atomic fluorescence.

JorLiu, curves for zinc (Figures 6, 10, 11, and 18)

and cadmium (Figures 7, 12, 13, and 19) were prepared

using the respective Osram spectral lamps as excitation

sources. The shape of the atomic fluorescence working

curves in Figures 6, 7, 10, 11, 12, and 13 agree with the

shape predicted by Winefordner and Vickers (39). Regions

of low concentration are linear with curvature appearing

at high concentrations. Individual curves will be dis-

cussed later in greater detail.

Fluorescence was not obtained for the elements

gallium, indium, antimony, and selenium when their solu-

tions in the flame were illuminated by electrodeless dis-

charge tube radiation. Measurement of the radiation from

these lamps showed that the intensity of the desired lines

was considerably below that observed for the sources which

were capable of producing fluorescence; in some cases, the

difference was of the order of 1000 times lower intensity.

Comparison of the working curves obtained for dif-

ferent fuel gases and atomizer-burners for each of the four

fluorescing elements, Figures 6 to 9, shows an interesting

trend. lor the Beckman total-consumption atomizer-burners

the air-hydrogen flame exhibited the highest intensity in

all cases. Generally the oxyhydrogen flame was next in

the ranking with the oxyacetylene flame having the lowest

observed relative fluorescence intensity for these three

flames. The Meker burner flame, using natural gas and air,

gave results which cannot strictly be compared to those

for the total-consumption atomizer-burners because the

flame area and path length through this flame were of much

greater dimensions than comparable measurements for the

Beckman burner flames.

Flame temperatures for the oxyhydrogen and oxyacetylene

flames have been calculated for each element by the method

of Winefordner, Mansfield, and Vickers (36, 37) according

to the position in the flame where fluorescence measure-

ments have been made. Temperature values for the air-

hydrogen flame were obtained by comparison of the above

values with those listed by Dean (8) and approximating the

decrease in temperature. Table 4 lists these temperatures.

Table 4

Flame Temperatures for Various Elements
When Using Beckman Total-Consumption Atomizer-Burners

Flame Temperature, OK.

Zn Cd Hg Ti

H2/air 1930 1930 1780 1780

H2/02 2620 2620 2480 2480

02H2/02 2760 2760 2550 2550

These values seem to indicate a correlation between low

flame temperature and maximum relative fluorescence in-

tensity when compared to the working curves for the ele-

ments in Figures 6 to 9.

The weak mercury 2537 A. fluorescence when using a

Meker burner air-natural gas flame was probably related

to the large quenching effects possible when foreign sub-

stances are present with mercury vapor, as explained by

Pringsheim (22) and Litchell and Zemansky (19). The nature

of this specific flame was different than any of the others

because it nad a lower temperature (8) and a much higher

ratio of flame gases to atomic vapor.

Comparison of the atomic absorption measurements in

Figures 18 to 21 with the working curves for atomic fluores-

cence in Figures 6 to 9 will give an idea of the efficiency

of excitation of fluorescence for each element. Zinc, cad-

mium, and mercury showed high absorbance which would account

for the favorable results obtained by atomic fluorescence

measurements. Thallium, however, showed very poor absorp-

tion when compared to the other elements. This would

partially explain the fact that thallium did not fluoresce

as well as the other elements. The low values for the air-

natural gas fluorescence working curve for mercury can be

explained by comparing it with the working curve for atomic

absorption under the same conditions. Here also the air-

natural gas flame gave poorer results than the other flames.

If an atomic vapor in a flame does not absorb radiation

well, then the fluorescence should also be low.

Different solvents used for analysis by atomic fluores-

cence flame spectrometry should affect the relative fluores-

cence intensity as measured according to the efficiency of

solvent evaporation in the flame and their flow rates

into the flame. This is a result of the viscosities of

the solutions (35). Solutions in water, 30% methanol in

water, and 70% methanol in water were used to obtain the

working curves shown in Figures 10 to 17. For both the

oxyhydrogen and the oxyacetylene flames in the cases of

zinc, cadmium, and thallium, the order of decreasing in-

tensity by solvents was 70% methanol, 30% methanol, and

water. Because the solutions with the higher percentages

of methanol should be the ones which would evaporate

easiest, this seems to be the dominant factor in influ-

encing the position of these curves. In the curves for

mercury in Figures 14 and 15 there seems to be a different

trend indicated with the 301 methanol solutions having the

lowest relative fluorescence intensities and the other two

solvents having higher intensities. In most cases the

water solution of mercury had a higher intensity signal

than the 70b methanol signal.

Data obtained on the viscosities of methanol-water

mixtures indicated that a maximum of viscosity for the

system existed near the composition of 30% methanol. At

the height above the flame used for the analysis ef mer-

cury, the solvent droplets would be more completely evap-

oratea than lower in the flame, so the effect on intensity

produced by solvent viscosity would overshadow that of

solvent evaporation, which was more important for the other


B. Types of Gases Added to Fuel Gas

Addition of foreign gases to the flame should have

three main effects: the temperature of the flame should

be lowered, the fluorescence should be quenched by col-

lisional deactivation of excited atoms, and the flame gases

should be diluted. The two flames on which the effect of

addition of argon and nitrogen to the flame by mixing with

the fuel gas was studied exhibited opposite trends in the

change observed in relative fluorescence intensity. In

the oxyhydrogen flame, the addition of either gas gave

readings lower than those obtained when no gas was added.

Argon ana nitrogen probably acted as quenching species in

the flame with nitrogen having a greater effect due to its

molecular form (1). The data for the two flames can be

found in Tables 8 to 11.

Using the oxyacetylene flame, the trend was reversed;

argon and nitrogen addition resulted in a higher intensity

than the normal flame. In this case the observed increase

could possibly have been due to a decrease in the tempera-

ture of the flame or a dilution of the flame gases, which

could be seen as quenching species. The oxyacetylene flame

will have a higher C02 content than the oxyhydrogen flame.

Alkemade (1) found C02 to be extremely effective in quenching

flame fluorescence of sodium. A dilution of the flame gases

would reduce the probability of collisions between activated

atoms ana U02 molecules. Slightly higher readings were ob-

tained for argon than for nitrogen in the oxyacetylene

flame. This could be explained by reasoning that molecular

nitrogen was much more efficient in quenching fluorescence

by collisional deactivation of the excited atoms than was

atomic argon (1). For thallium in the oxyacetylene flame

the results obtained showed that nitrogen addition gave

the lowest intensities, which was probably due to the

quenching effects of nitrogen on thallium vapor.

C. Types of Gases Added to Burner Sheath

Use of a sheathed oxyhydrogen flame gave no definite

indications of its value for use in atomic fluorescence

flame spectrometry. For zinc analyses, argon in the sheath

generally gave higher intensity values as well as having

higher signal-to-noise ratios for aqueous solutions as seen

in Table 8. The percentage standard deviation, however,

was higher for the use of the sheath for all gases than

when the sheath was not used. In the case of mercury, as

seen in Table 10, no definite increase in fluorescence in-

tensity due to the sheath gases could be seen, but the per-

centage standard deviation was lower for aqueous and 30%

methanol solutions. All gases had about the same effect.

Argon seemed to give a better signal-to-noise ratio when

it was used in the sheath. The sheathed flame was found

to be unstable to the extent that it would frequently be

self-extinguished. The use of the sheathed flames, there-

fore, is not sufficiently justified.

D. Comparison of Types of Flame Spectrometry
Atomic fluorescence flame spectrometry compared
favorably with atomic absorption and atomic emission

flame spectrometry as seen in Tables 5, 6, and 7. The
limit of detection in atomic fluorescence has been defined
as "that concentration of fluorescent species which pro-

duces a photodetector signal due to fluorescence approxi-

mately 0.75 times as large as the root-mean-square noise

signal" (38).

Table 5
Limit of Detection in p.p.m. for Various
Elements for Atomic Fluorescence Flame Spectrometry

Flame Solvent Element

Zn Cd Hg T1
H2/02 H20 0.04 0.1 0.1 1.0

30% CH OH 0.03 0.1 0.1 1.0

70% OH3OH 0.005 0.1 0.1

02H2/02 o20 0.04 0.3 0.1 5.0
30 OCH308 0.005 0.3 0.1 5.0

70A OH30H 0.01 0.2 0.3 ---
H2/air H20 0.02 0.2 0.1 2.0
Natural 6as H20 0.01 0.1 7.0 1.0

A simple comparison of these three tables will show that
atomic fluorescence flame spectrometry is the most sensitive

technique of the three available for analysis of zinc and

mercury, and it only falls slightly short of being the best

method for cadmium. Lower limits of detection were possible

for zinc and cadmium (15), but these have been obtained by

special atomic absorption techniques using a 91 cm. cell.

Atomic absorption and atomic emission gave considerably

lower limits of detection for the analysis of thallium.

Table 6

Limit of Detection in p.p.m.
Elements for Atomic Emission Flame
Element riavelength, A. Flame
Zn 2139 H2/02


5200 C2H2/02
Cd 2288 H2/02


3261 H2/air

Hg 2537 H2/02

Tl 3776 H2/02



for Various
Spectrometry (17)
Solvent Limit
H20 500

organic 10
H20 60o
H20 2500

organic 2.5

H20 10
H20 40

H20 50
H20 0.5
H20 30

organic 6

H20 50
H20 10
H20 0.1

organic 0.06
H20 1

H20 0.2


Table 7

Limit of Detection in p.p.m. for Various
Elements for Atomic Absorption Flame Spectrometry (16, 25)

Flame Solvent Element

Zn Cd Hg T1

H2/02 H20 0.3 1 100 0.2

organic 0.08 0.1 100 --

Natural gas
/air H20 0.03 0.03 10 --

Since atomic absorption flame spectrometry offers the

closest competition for atomic fluorescence, the fact should

be mentioned that the absorption measurements listed were

made using a flame cell 10 to 12 cm. long and radiation line

sources whose cost exceeds that of those used for fluores-

cence. Thus, atomic fluorescence is a more convenient and

inexpensive method of analysis than is atomic absorption for

the cases described.

In order to compare the results obtained for the atomic

fluorescence of thallium in flames with atomic emission flame

spectrometry, measurements were made according to standard

techniques of atomic emission flame spectrometry. The ex-

perimental setup used for atomic emission measurements was

exactly the same as used for the fluorescence studies; however,

in this case the thermally emitted radiation from the outer

cone of the flame was observed (the thallium discharge tube

was turned off in this case). Figure 22 shows the working

curves obtained and indicates that atomic emission was as

good or better than atomic fluorescence for the analysis
of thallium. Since fluorescence measurements were made

using the outer cone of the natural gas-air flame, emis-
sion readings were not taken for this flame.

E. Accuracy and Signal-to-Ioise Ratio
The percentage standard deviation for a number of
measurements at a given concentration was calculated (18)

100 [ (Iy F)2 (2)
l n -1

where r is the percentage standard deviation and IF is
the mean average of n readings of the relative fluores-
cence intensity, IF. The signal-to-noise ratio was calcu-

lated from the root-mean-square noise value according to

S/N = IF/1N (2)

where S/N is the signal-to-noise ratio and N is the root-
mean-square noise figure (0.354 of peak-to-peak noise).

The values obtained for percentage standard deviation
are listed in Tables 8 to 11 for the elements which were
considered. The percentage standard deviations can be
compared with average values of 2% for atomic absorption
(12) and 0.2% to 5% for atomic emission (18). For zinc

the percentage of standard deviations were mostly below

5% for all concentrations while for cadmium and mercury
this range was valid only above concentrations of 10 p.p.m.

A reproducibility of 11% to 27% was obtained for 1 p.p.m,

solutions of cadmium and mercury. Thallium exhibits the

same variation as cadmium and mercury but at concentrations

higher by a factor of ten.

The signal-to-noise readings varied in the opposite

manner of the results obtained for the percentage standard

deviations. Higher values for the signal-to-noise ratio

are desirable because this indicates either a high in-

tensity signal, a low noise level, or a combination of

these two factors. Higher values are also an indication

of the ease with which readings of fluorescence intensity

have been made at those concentrations. High signal-to-

noise ratios were found for all concentrations of zinc

and higher concentrations of cadmium, mercury, and thallium.

High noise levels and low signal levels were present for

low concentrations of cadmium, mercury, and thallium.


A. Shapes of Working Curves
The theoretical shape of a working curve plot of in-

tensity of fluorescence vs. ground state atom concentra-

tion has been predicted by Winefordner and Vickers (39).

At low concentrations the plot is essentially linear; as

the concentration increases the curve reaches a maximum

and then decreases in value. This is the shape which has

been observed in actual practice, as seen in Figures 6 and

7, the working curves for zinc and cadmium. Mercury and

thallium (Figures 8 and 9) did not exhibit this curvature

because their high atomic weights placed these concentra-

tions in molarity at lower values than zinc and cadmium.

The region of the working curves most suitable for

use in practical analysis is that in which linearity exists.

In the curved region of the maximum, an error in the in-

tensity measurement would be magnified on the concentra-

tion scale, and so upper limits of analysis must be set

for these elements. The approximate upper limits for zinc,

cadmium, mercury, and thallium are 10 p.p.m., 100 p.p.m.,

1000 p.p.m., and 1000 p.p.m. respectively for total-

consumption atomizer-burners. Higher limits are probable

in the cases of mercury and thallium (1000 p.p.m. was the

upper limit of the study). Using the Xeker burner the

upper limit is 1000 p.p.m. in all cases.

In order to use flame fluorescence as a routine

analytical method one should first plot an accurate

working curve, using a series of standard solutions of

different concentrations, under the conditions to be used

for analysis, in order to determine the behavior of the

working curve for the element with respect to curvature

at different concentrations. Once this curve has been

plotted one needs only to run solutions varying by powers

of ten in order to prepare a working curve suitable for

a series of analyses.

B. Improvement of Equipment

Sources. Probably the one most important item in

the atomic fluorescence instrument is the source which

produces the excitation. As Winefordner and Vickers (39)

have shown for low sample concentrations,

IF = OP'I,. (A)

where IF is the measured fluorescence intensity, 0 is a

constant for a particular system, PO is the incident radiant

power from the source, and No is the ground state atom con-

centration. Thus, a variation of 1% in the source will be

seen directly as a variation of 1 in the fluorescence in-

tensity. The most stable source will give the best accu-

racy if all other factors are constant.

Also from the equation above one can easily see that

the greater the intensity of the source, the greater will

be the intensity of the fluorescence. One should attempt,

therefore, to use a source of as high stability and in-

tensity as possible.

A source can be intense, but if it exhibits a signifi-

cant amount of line reversal, the very portion of the radia-

tion needed for excitation is lest. Therefore, absence of

reversal of the line source should also be of primary im-


Several sources show promise of being valuable for

use in atomic fluorescence flame spectrometry. A water-

cooled mercury arc which emits unreversed lines should be

extremely suited to use in mercury analysis. The above

arc could be extended to other metals through the use of

amalgams, such as a thallium amalgam arc as used by Nichols

and Howes for excitation of thallium fluorescence (20).

Inclusion of trace quantities of elements in the carbon

rods of a d.c. arc (2) could be used in many cases as a

means of exciting fluorescence, especially if no other

source is available. Finally, an extremely high intensity

source emitting a continuum, e.g., a 1000 watt xenon arc

lamp, would be suitable for excitation of fluorescence of

all elements if the source intensity is high enough at the

exact value of the required line. A 100 watt xenon arc

has been tried as a source, but its intensity was not

sufficiently high.

Atomizer-Burners. Because the burner is essentially

a sample cell, under ideal conditions it would be a large

vessel holding only the atomic vapor of the element under

investigation. Herrmann and Alkemade (18) feel that the

most efficient atomizer-burner for atomic emission flame

spectrometry is the total-consumption direct-injection

burner such as that manufactured by Beckman Instruments.

In flame fluorescence, however, light scattering by un-

evaporated solvent droplets is important, and the Beckman

burner does produce a more coarse spray and therefore more

scattering of source radiation than a chamber-atomizer

coupled to a burner. A chamber-atomizer, therefore, would

be more efficient in reducing the scattering of the source

radiation. Actually, the Beckman burner should be a better

choice because it might give a larger concentration of

atoms in the flame due to a higher solution flow rate.

This would depend upon the specific chamber-atomizer chosen

and its efficiency in introducing the solution vapor into

the flame as compared with the efficiency of the Beckman

burner. The use of nonaqueous solvents results in a con-

siderable increase in sample introduction efficiency (36)

when using total-consumption atomizer-burners. Therefore,

it is doubtful whether chamber-type atomizer-burners would

even have a resultant advantage over total-consumption


The size of the flame with respect to its diameter

should also be considered. According to Winefordner and

Vickers (39) the fluorescence intensity should increase

with the path length through the flame taken by the source

radiation. A wider flame than produced by the Beckman

total-consumption atomizer-burner should give better results.

In the case of a wide flame, however, a lens would be needed

in front of the entrance slit of the monochromator so that

all of the flame would be "seen" by the monochromator.

Each element being analyzed probably has an optimum

flame temperature, so differences in the gases used would

also affect the design of a burner. The gases chosen should

not have reaction products which will quench the fluores-

cence as Alkemade (1) found was true of 002 with respect

to sodium fluorescence in a flame.

Monochromators and Detectors. Because most of the ex-

citation sources used for atomic fluorescence flame spec-

trometry produce broad lines, solvent droplet scattering

will reflect a fraction of this radiation into the mono-

chromator. In some cases the flame itself has an intense

background. Ideally, the monochromator should be capable

of resolving only the fluorescence line itself and elimi-

nating the effects of scattering as well as most of the

flame noise. One would expect the best results, therefore,

from a spectrometer of high resolving power.

Most fluorescence measurements will probably be made

in the ultraviolet region of the spectrum. It is essential,

therefore, that all optics be made of quartz to eliminate

attenuation of the fluorescence or source radiation by

optical absorption. The photomultiplier detector tube,

usually a 1P28, should have a quartz outer envelope. In

the visible region better signal-to-noise ratios would be

possible by using a 1P21 photomultiplier tube which has

a glass envelope (24).

Amplifier. Consideration must also be given to the

amplifiers, which convert the signal from the photomultiplier

tube detector into a form which can be read or recorded. In

the ultraviolet region of the spectrum the thermal emission

is low, and so a d.c. amplifier can be used. Sensitivity

and stability are the two primary factors affecting the

selection of an amplifier. The amplifier must be sensitive

enough to detect the slightest variations above the dark

current output of the photomultiplier detector. Stability

is necessary so that the measurements made will be constant

and reproducible.

When measurements are made in the visible region, the

flame background is high and thermal emission is more

probable. If the fluorescence is modulated at a constant

frequency by means of a chopper placed between the source

and the flame, then a tuned a.c. amplifier can be used to

amplify only the fluorescence. The thermal emission is

primarily a d.c. signal, and so no correction for its

presence need be made. A further refinement of this tech-

nique would be to use a tuned a.c. amplifier, which would

amplify only signals of the chopper frequency and attenuate

all others. Because the fluctuation of the flame back-

ground results in a white noise, the flame background

interference could be reduced even further by a narrow

band amplifier.

C. Sensitized Fluorescence

A specialized technique of atomic fluorescence flame

spectrometry is sensitized fluorescence. This technique

may be of special value in cases where excitation sources

are not available. An example of this would be the intro-

duction of a mixture of thallium and mercury in solution

form into a flame illuminated by a source of mercury

resonance radiation. The thallium atoms would be excited

by energy transfer collisions with mercury atoms activated

by absorption of resonance radiation. This technique was

tried, but no thallium radiation could be detected.

D. The Value of Atomic Fluorescence Flame
Spectrometry as a Means of Chemical Analysis

Atomic fluorescence flame spectrometry definitely has

use as a method of analysis. It should complement the ex-

isting methods of atomic emission and atomic absorption

flame spectrometry. Elements whose resonance lines lie in

the ultraviolet and which give poor results by atomic emis-

sion should be adaptable to analysis by atomic fluorescence

as well as atomic absorption, which was the only alternate

spectrometric method. Also, elements which absorb weakly

in this region, e.g., mercury at 2537 A., may be well suited


to analysis by atomic fluorescence if their quantum effi-

ciencies are favorable. The other important advantage

atomic fluorescence has over atomic absorption is that

fluorescence excitation sources are generally less expen-

sive tian hollow cathode discharge tubes. The value of

atomic fluorescence flame spectrometry for routine analyses

should give it a place in the large number of methods

available to the analyst.


Atomic fluorescence flame spectrometry was applied

to the analysis of solutions of zinc, cadmium, mercury,

and thallium. Solutions of these elements were introduced

into a flame and the resulting atomic vapor was excited by

radiation from an intense, unreversed line source. The

resulting fluorescence emission was measured at right

angles to the source-flame axis by means of a compact

monochromator-detector and a sensitive d.c. amplifier.

The wavelengths used for analysis were 2139 A., 2288 A.,

2537 A., and 3776 A. for zinc, cadmium, mercury, and

thallium, respectively.

Different excitation sources were tried and compared.

Working curves were prepared using air-hydrogen, oxyhydrogen,

and oxyacetylene flames with total-consumption atomizer-

burners and an air-natural gas flame with a Meker burner

and chamber atomizer. These curves were linear over a large

range of concentrations with curvature appearing only at

high values of concentration. The influence of 30% methanol

and 70% methanol solvents as well as water on the shapes of

working curves and limits of detection for each element

was studied in detail. Argon and nitrogen were premixed

with the fuel gas and argon, nitrogen, and oxygen were

added to a burner sheath to observe possible effects on the

fluorescence. Atomic absorption measurements were made

for purposes of rough quantum efficiency comparisons.

Percentage standard deviations and signal-to-noise ratios

were determined for all studies.

The limits of detection obtained for zinc, cadmium,

mercury, and thallium were, respectively, 0.01, 0.1, 0.1,

and 1.0 p.p.m. for aqueous solutions. Low flame tempera-

tures and maximum intensity seemed to be related. The 70%

niethanol solutions resulted in the maximum sensitivity

except in the case of mercury where water was the best

solvent. Gases added to the oxyhydrogen flame lowered

the fluorescence intensity while for the oxyacetylene flame

an increase was noted. Argon had the greatest effect. No

justification was found for using the burner sheath for the

elements studied in this work. Atomic fluorescence flame

spectrometry was found to compare favorably with atomic

absorption flame spectrometry for analysis of zinc, cadmium,

and mercury. However, the percentage standard deviation of

5% was somewhat higher than representative values for atomic

absorption methods.

The history and theory of atomic fluorescence flame

spectrometry were described in brief. Suggestions were

made for improvement of the method by selection of proper

components for the experimental setup. After more extensive

research, development, and application, atomic fluorescence

flame spectrometry should become an accepted laboratory

method of analysis for a number of elements.






o 100-

0.1 1 10 100 1000

0.1 1 10 100 1000

Hg Concentration in p.p.m.

Figure 5. Atomic Fluorescence Working Curves for Mer-
cury (Hg 2537 A. Line) in Oxyhydrogen and
Oxyacetylene Flames Using a Vapor Discharge
Tube and an Electrodeless Discharge Tube

O H2/02, Electrodeless Discharge Tube

C2H2/02, Electrodeless Discharge Tube

H2/02, Vapor Discharge Tube

C C2H2/02, Vapor Discharge Tube



I 100-

p 10-

0.01 0.1 1 10 100 1000

Zn Concentration in p.p.m.

Figure 6. Atomic Fluorescence Working Curves for Zinc
(Zn 2139 A. Line) Introduced as Aqueous
Solutions into Different Flames

0 H2/02


N Natural gas/air
ZnCnetaio nppm
Fiue6 tmcFursec okn uvsLrZn
(Z 19A in)Itoueda qeu
Souin it ifretFae
o H/0



S 100 -



0.1 1 10 100 1000

Od Concentration in p.pm.

Figure 7. Atomic Fluorescence Working Curves for Cad-
mium (Cd 2288 A. Line) Introduced as Aqueous
Solutions into Different Flames

0 H2/02

C 22/02

4D H2/air

a Natural gas/air







a 10-


0.1 1 10 100 1000

Hg Concentration in p.p.m.

Figure 8. Atomic Fluorescence Working Curves for Mer-
cury (Hg 2537 A. Line) Introduced as Aqueous
Solutions into Different Flames Excited by
the Electrodeless Discharge Tube

0 H2/02


0 H2/air

( Natural gas/air




1 10 100 1000

Tl Concentration in p.p.m.

Figure 9. Atomic Fluorescence Working Curves for
Thallium (Tl 3776 A. Line) Introduced
as Aqueous Solutions into Different Flames

0 H2/02



a Natural gas/air






a ,

0.01 0.1 1 10 100 1000

Zn Concentration in p.p.m.

Figure 10. Atomic Fluorescence Working Curves for Zinc
(Zn 2139 A. Line) in Different Solvents
Introduced into the Oxyhydrogen Flame

30% CH3OH

C70% CH30H







0.01 0.1 1 10 100 1000

Zn Concentration in p.p.m.

Figure 11. Atomic Fluorescence Working Curves for Zinc
(Zn 2139 A. Line) in Different Solvents
Introduced into the Oxyacetylene Flame

0 H20

0 30% CH OH

0 70% CH30H



1 100-




0.1 1 10 100 1000

Cd Concentration in p.p.m.

Figure 12. Atomic Fluorescence Working Curves for Cad-
mium (Cd 2288 A. Line) in Different Solvents
Introduced into the Oxyhydrogen Flame

0 H20

30% CH3OH

4 70% CHOH

1 10


Cd Concentration in p.p.m.

Figure 13.

Atomic Fluorescence Working Curves for Cad-
mium (Cd 2288 A. Line) in Different Solvents
Introduced into the Oxyacetylene Flame


30% CH OH

70% CH OH











) 10-

0.1 1 10 100 1000

Hg Concentration in p.p.m.

Figure 14. Atomic Fluorescence Working Curves for Mer-
cury (Hg 2537 A. Line) in Different Solvents
Introduced into the Oxyhydrogen Flame Excit-
ed by the Electrodeless Discharge Tube

0 H20

30% CH30H

70% CH 30OH





I0 10-

0.1 1 10 100 1000

Hg Concentration in p.p.m.

Figure 15. Atomic Fluorescence Working Curves for Mer-
cury (Hg 2537 A. Line) in Different Solvents
Introduced into the Oxyacetylene Flame Excit-
ed by the Electrodeless Discharge Tube

0 H20

@ 30% CH30H

o 70% CH30H






Tl Concentration in p.p.m.

H H20

@ 30% CH 30H

4) 70% CH30H





10 100 1000

10- 001

30% CH30H

0 70% CH30H

10 100

Zn Concentration in p.p.m.

Figure 18.

Atomic Absorption
(Zn 2139 A. Line)
ent Flames

Working Curves for Zinc
Introduced into Differ-



Natural gas/air

1 .00(








0.001 -I
1 10 100 1000

Cd Concentration in p.p.m.

Figure 19. Atomic Absorption Working Curves for Cad-
mium (Cd 2288 A. Line) Introduced into Dif-
ferent Flames

0 H2/02
0 H2/air

3 Natural gas/air




1 10 100 1000

Hg Concentration in p.p.m.

Figure 20. Atomic Absorption Working Curves for Mer-
cury (Eg 2537 A. Line) Introduced into Dif-
ferent Flames Using the Electrodeless Dis-
charge Tube

0 H2/02
o H2/air

g Natural gas/air





0.001 -\
10 100 1000

Tl Concentration in p.p.m.

Figure 21. Atomic Absorption Working Curves for Thal-
lium (Tl 3776 A. Line) Introduced into Dif-
ferent Flames

0 H2/02

0 H H2/air

@ Natural gas/air






1 10 100 1000

T1 Concentration in p.p.m.

Figure 22. Atomic Emission Working Curves for Thal-
lium (Tl 3776'A. Line) Introduced into
Different Flames

O H2/02


C) H2/air



Key to Flame Abbreviations

Symbol Flame

aNe Natural gas/air

aH H2/air

UH H2/02

OH Sheathed H2/02

OA 02H2/02

Key to Solvent Abbreviations

Symbol Solvent

H20 Water

Me OH Methanol

Table 8
Effect of Gases Added to Fuel Gas and Sheath,
Percentage Standard Deviations, and Signal-to-Noise
Ratios for Atomic Fluorescence ileasurements of Zn 2139 A.
Flame Solvent Gas Concentration I % r S/N
in p.p.m.
all H20 None 1 85 3.4 48

10 800 8.9 45
100 4800 1.5 68
aH H20 None 1 116 3.2 33

10 657 1.7 74
100 784 3.4 37
OH H20 None 1 51 5.2 51

10 395 1.5 65
100 890 2.6 71
50% MeOH 1 79 2.5 45

10 537 1.0 62
100 683 2.4 26
70% MeOH 1 125 1.9 24

10 580 1.4 37
100 460 6.0 22
H20 Ar 10 355 1.8 67

N2 359 3.0 51
30% MeOH Ar 408 1.5 58

N2 397 3.1 56
70% MeOH Ar 560 2.0 45

N2 538 1.4 51



30% MeOH

70% MeOH







0 2




30% ieOH

70% keOH


30o 'L.eOH

70% heOH


% G- S/N


8 continued

in p.p.m.

















































Table 9
Effect of Gases Added to Fuel Gas,
Percentage Standard Deviations, and Signal-to-Noise
Ratios for Atomic Fluorescence Keasurements of Cd 2288 A.
Flame Solvent Gas Concentration IF % a- S/N
in p.p.m.
aN H20 None 1 6.5 12 9.2

10 108 4.8 31
100 810 3.6 23
aH H20 None 1 20 22 3.8

10 181 3.4 34
100 434 1.8 49
OH H20 None 1 11 11 4.8

10 102 4.5 58
100 328 4.2 60
30% MeOH 1 17 27 3.4
10 127 4.0 25
100 392 1.6 76
70% MeOH 1 23 20 9.1
10 157 3.4 31
100 435 3.5 42
H20 Ar 10 96 8.7 19
N2 85 3.9 18
30% MeOH Ar 103 7.0 15

N2 99 6.7 32
70% MeOH Ar 152 2.4 29

N2 136 6.1 29

Table 9 continued





30% MeOH

70% M1eOH


30% MeoH

70% fleOH


IF % < S/N

in p.p.m.




















































Table 10

Effect of Gases Added to Fuel Gas and Sheath,
Percentage Standard Deviations, and Signal-to-Noise
Ratios for Atomic Fluorescence Measurements of Hg 2537 A.

me Solvent Gas Concentration IF % r S/N






30% MeOOH

70% MeO O
















30% MeOH

70% MeOH


























































30% MeOH

703' Me0H


Table 10 continued

Gas Concentration
in p.p.m.









30% MeOH

70o MeOH


30% MeOH

70% 1MeOH












































































Table 11
Effect of Gases Added to Fuel Gas,
percentage Standard Leviations, and Signal-to-Noise
2.atios for Atomic Fluorescence Leasurements of Tl 3776 A.
l1ame Solvent Gas Concentration IF % < S/N
in p.p.m.
a H20 None 10 5.8 14 5.5

100 72 4.9 41
aE H20 None 10 4.2 11.5 12

100 35 2.1 28
1000 277 2.3 26
OH H20 lone 10 1.7 9.6 7.1

100 20.6 1.9 35
1000 178 2.0 37
30o Ae0H 10 3.4 7.1 14

100 37 3.1 20
70% MeOH 10 13 -- 7.8

100 75 --- 14

H20 Ar 100 10.7 3.8 13

N2 12.5 5.2 13
30A MeOH Ar 16.7 1.8 20

N2 21.8 2.3 12
70/ MeOH Ar 20.8 5.3 17

N2 21.9 3.3 13
OA H20 None 10 1.2 27.3 2.4

100 17.6 4.4 16
1000 126 3.2 20


Table 11 continued

Gas Concentration


in p.p...



30, ~e0OH

70; Me OH


30% MeOH

70% I:eOH



IF 0- S/N
































(1) Alkemade, C. T. J., International Conference on
Spectroscopy, College Park, Md., June 1962.
(2) Badger, R. M., Z. Phys. 5, 56 (1929).

(3) Benaer, i., Phys. Rev. 36, 1543 (1930).
(4) Boers, A. L., Alkemade, 0. T. J., Smit, J. A.,
Physical 22, 358 (1956).

(5) Bogros, A., Compt. rend. 183, 124 (1926).
(6) Christensen, 0. J., Rollefson, G. K., Phys. Rev.
34, 1157 (1929).
(7) .orliss, C. H., Bozman, W. R., Westfall, F. 0.,
J. Opt. Soc. Am. 4, 398 (1953).
(8) Dean, J., "Flame Photometry," McGraw-Hill, New York,
(9) uunoyer, L., Radium 10, 400 (1913), through Mitchell,
A. C. G., Zemansky, M. W., o. cit.
(10) Dunoyer, L., Compt. rend. 178, 1475 (1924).
(11) Dunoyer, L. Wood, R. W., Phil. Mag. 27, 1025 (1914).
(12) Elwell, W. T., Gidley, J. A. F., "Atomic-Absorption
Spectrophotometry," Pergamon, Oxford, 1961.
(13) Fred, M., Tomkins, F. S., J. Opt. Soc. Am. 47, 1976
(14) Fuchtbauer, C., Phys. Z. 21, 635 (1920).
(15) luwa, K., Vallee, B. L., Anal. Chem. 35, 942 (1963).
(16) Gatehouse, B. M., Willis, J. B., Spectrochim. Acta
,17, 710 (1961).
(17) Gilbert, P. T. Jr., "Symposium on Spectroscopy," Am.
Soc. Testing Materials, Spec. Tech. Publ. No. 269

(18) Herrmann, R., Alkemade, C. T. J., "Flammenphotometrie,"
2nd Edition, Springer Verlag, Berlin, 1960. "Chemical
Analysis by Flame Photometry," P. T. Gilbert trans-
lator, Interscience, New York, 1963.
(19) hitchell, A CG., Zemansky, i. W., "Resonance Radia-
tion and Excited Atoms," University Press, Cambridge,
(20) hichols, E. L., Howes, H. L., Phys. Rev. 23, 472
(21) Ponomarev, N., Terenin, A., Z. Phys. 37, 95 (1926).
(22) Pringsheim, P., "Fluorescence and Phosphorescence,"
Interscience, New York, 1949.

(23) Pucclanti, L., Accad. Lincei Atti 12, 430 (1904)
through Mitchell, A. 0. G., Zemansky, M. W., op. cit.
(24) Radio Corporation of America, "Photosensitive Devices
and Cathode-Ray Tubes." Booklet CRPD-105, Tube Divi-
sion, Harrison, N. J., 1955.
(25) iobinson, J. W., Anal. Chem. 33, 1067 (1961).
(26) Robinson, J. W., Anal. Chim. Acta 24, 154 (1961).

(27) zax, N. I., "Handbook of Dangerous Materials,"
Reinhold, New York, 1951.
(28) zoleillet, P., Compt. rend. 184, 149 (1927).
(29) Strutt, R. J., Proc. Roy. Soc. (London), 91A, 511
(30) Strutt, R. J., ibid. 96A, 272 (1919).
(31) Terenin, A., Z. Phys. 31, 26 (1925).
(32) Terenin, A., ibid. 2, 98 (1926).
(33) Wiedemann, E., Schmidt, G. C., Wied. Ann. 57, 447
(1896) through Mitchell, A. C. G., Zemansky, M. W.,
oP. cit.
(34) Tinans, J. G., Proc. Nat. Acad. Sci. 11, 738 (1925).
(35) Winefordner, J. D., Latz, H. W., Anal. Chem. 33, 1727
(36) Winefordner, J. D., Mansfield, C. T., Vickers, T. J.,
ibid. p5, 1607 (1963).

(37) Winefordner, J. D., Mansfield, C. T., Vickers, T. J.,
ibid. 35, 1611 (1963).
(38) Winefordner, J. D., Staab, R. A., ibid. 36, 165
(39) Winefordner, J. D., Vickers, T. J., ibid. 36, 161
(40) oood, R. W., Phil. Iag. 3, 128 (1902).
(41) Wood, R. W., ibid. 6, 362 (1903).
(42) Wood, R. W., ibid. 10, 408 (1905).
(43) Wood, R. W., ibid. 10, 513 (1905).
(44) Wood, R. W., ibid. 23, 690 (1912).
(45) Wood, R. W., Proc. Phys. Soc. (London) 26, 185 (1914).
(46) Wood, i. W., "Researches in Physical Optics, I and
II," Columbia, New York, 1913 and 1919.
(47) Wood, R. W., Dunoyer, L., Phil. Mag. 27, 1018 (1914).
(48) Wood, R. W., Mohler, F. L., Phys. Rev. 11, 70 (1918).
(49) Wood, R. W., kohler, F. L., Phil. hag. 31, 456 (1919).


Robert Allan Staab was born July 11, 1937, at

Chicago, Illinois. He received his elementary educa-

tion in the public school system of Elmhurst, Illinois,

and after moving to West Palm Beach, Florida, in January,

1953, he was graduated from Palm Beach High School in

1955. He has attended the University of Florida since
September, 1955, and received the Bachelor of Science

in Education degree in June, 1960, and the Master of

Education degree in August, 1961. Since that time he

has pursued his studies toward the degree, Doctor of


Robert Allan Staab is married to the former Cherry

Yvonne Weakley. He is a member of the American Ohemical

Society and Alpha Chi Sigma.

This dissertation was prepared under the direction
of the chairman of the candidate's supervisery committee
and has been approved by all members of that committee.
It was submitted to the Dean of the College of Arts and
Sciences and to the Graduate Council, and was approved
as partial fulfillment of the requirements for the degree
of Doctor of Philosophy.

April 18, 1964

Dean, College of Alrts 8- drSience

Dean, Graduate Sehool

Supervisory Committee:

Cb ir.n

Li I---


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