are present, adsorbed water is converted

Table 1. Analysis of Mixtures of Hydrates to water of crystallization and does not

H20 hIax. Temp contribute an error in the analyses.
-Added, Hz0 Error, after 8 Nevertheless, it is advisable t o be con-
Hydrates Analyzed Mg. Found, Mg, 9% Hours, "C. tinuously alert to this possible source of
NarP207.10H20 214 212 -1 165 error.
NaZCOs.HZ0 74 73 -1.4 165
NasPsOio. 6H20 96 96 0 ACKNOWLEDGMENT
Na4P207.10H20 407 403 -1 100
NarPnO,. 10H20 189 189,190,188 165 The author wishes to express his
Sa2COj.H20 77 74,79,77 gratitude to John Andersen of the
NajOsOto.6Hz0 93 90,93,93 Monsanto Chemical Co. for designing
plus anhydrous phase of and building the temperature-measuring
each hydrate unit used in this work.
YazB40.i. 10H20 378 377 -0.2 500
SazB107.lOHzO 216 216" 0 500 LITERATURE CITED
NakPz07. lOHzO 94 95 4-1
e- (1) Chevenard, P., Woche, X., De la
Na2C08, HzO 73 13 +2.8 165 Tullayl, R., Bull. soc. chim. 5 ,
KzC03.1.5HzO 245 247 +0.8 10 (1944).
MgSOd. i"z0 112 107 -4.5 165 (2) Oroeco, E., Ministerio trabalho ind. e
CoClz. 6Ht0 192 193 $0.5 corn. Inst. nacl. technol. (Rio de
Absolute average 1. 2 Janeiro) 1940,33.
(3) Quimby, 0. T., J. Phys. Chem. 58,603
0 22 mg. of HzO from -ha2B4O7.10HzO phase found in Sa4Pz07phase. (1954).

RECEIVEDfor review July 19, 1956

work reported here, the adsorbed
moisture did not influence the results t o
a detectable degree, but i t is conceivable
that with mixtures in which the various
phases were completely hydrated ad-
sorbed n-ater might contribute to the
total weight loss of the sample. It is
believed that when anhydrous phases
Accepted October 5, 1956. Divisions of
.Analytical and Physical and Inorganic
~~~~~~v~~~~~~~~~~~~~~~~~~~~~
129th hleeting, ACS, ~ ~ lT ~l ~~ , ,~~ , ~ i
1956.

Phosphori metry
A New Method of Analysis
R. J. KEIRS, R. D. BRITT, Jr.,l and W. E. WENTWORTH
Deparfmenf of Chemisfry, Florida Sfafe Universify, Tallahassee, Flu.

bPhosphorimetry is a means of chem-

ical analysis based upon the nature
and intensity of the phosphorescent
light emitted by an appropriately
excited molecule. Many organic mol:
ecules containing multiple bonds, when
in a rigid glass formed a t low tem-
perature by solutions of the material
in suitable solvents, phosphoresce if
excited by radiant energy of suitable
frequency. Each phosphorescence is
unique in regard to its frequency,
lifetime, quantum yield, and vibrational
pattern and such properties are used
for qualitative identification. The cor-
relation of intensity with concentration
can serve as a basis for quantitative
measurement. Mixtures are analyzed
by the use of a resolution phosphoro-
scope. The method has been ap-
plied to the determination of several
compounds and their mixtures, three
of which are described.
Present address, Savannah River
Plant, E. I. du Pont de Xemours & Co.,
Inc., Augusta, Ga.
T HE PHOSPHORESCESCE EMISSIONS
from over 200 compounds have
been reported by molecular spectros-
copists interested in the elucidation
of the energy schemes of molecules.
About 90 such emissions have been
tabulated by Lewis and Kasha (11).
The use of phosphorescence emission
spectra for identification of substances
was first suggested by Lewis and
Kasha; however, this subject has been
developed very little in the past decade.
More recently the idea of using a resolu-
tion phosphoroscope (proposed by M.
Ilasha of this laboratory) greatly ex-
tended the annlytical potentialities of
phosphorescence. B theoretical analy-
sis of the method is to be published.
Phosphorescence is not characteristic
of a specific class of compounds, but
a prime condition for its observation
is high viscosity. Substances that phos-
phoresce may be divided into two
classes, based upon the mechanism
by which their phosphorescences are
produced (11). I n the first group are
mineral, or crystal, phosphors (15).
I n this case the individual molecule
is not phosphorescent, but the ability
to emit a n afterglow is associated with
the return of an electron to a n impurity
site in the crystal, following ionization
through the process of photon absorp-
tion. -4s this type of phosphorescence
cannot be ascribed to a definite sub-
stance, phosphorescences of this clash
were not considered in this investiga-
tion. I n the second class the emission
is attributed to a definite molecular
species. whether the substance is in
a pure crystalline state, absorbed on a
suitable surface, or dissolved in a suit-
able rigid solution. It is with such rigid
transparent solutions that this investi-
gation was concerned.
Phosphorescence is produced in mole-
cules by the absorption of radiant energy
of a frequency within the normal ab-
sorption band of the molecule-usually
in the ultraviolet region of the spec-
trum. Two of the many possible end
results are fluorescence and phosphor-

202 * ANALYTICAL CHEMISTRY

escence produced by the excitation of
an appropriate electron energy level
of the molecule. These two processes
may be shown schematically by a
diagram of molecular electron energy
level (Figure l), similar to that given
hy Jablonski (8) for dyes.
Molecules in an electron ground state
level, G, on absorption of radiant energy.
arrow 1, are for a period less than 1O-lz
second (11) in a vibrational level of an
upper excited singlet state, 82, Sa,etc.
A singlet state in an unsaturated mole-
cule is characterized by a pair of un-
saturation (T) electrons having paired
(antiparallel) spins. From one of the
upper singlet levels a process may occur
in which the absorbed energy is given
up in steps that may involye a nonradia-
SINGLET TRIPLET
LEVELS LEVELS
I
II 1 2
Figure 1 . Molecular energy levels
h\
W I
PHOSPHORESCENCE
DECAY CURVE
TIME
a amino acids. Each phosphorescence
EXCITATION ( I R R A D I A N C E X TIME)
b.
Figure 2.
tive transition (dashed arrow), arrow
2, to the lowest excited singlet level,
SI, follon-ed by a radiative transition,
arrow 3, to one of the vibrational levels
of the ground electron level, G . This
singlet -
latter transition (& + G), which is a
singlet transition, represents
a fluorescence band of the molecule
(10, 11).
By a radiationless route, arrow 4,
some of the excited molecules can pass
from their lowest excited singlet state,
SI,into a vibrational level of a metasta-
ble state, T I . This is the phosphores-
cent state or triplet state described by
Lewis and Kasha (11). A triplet state
in an unsaturated molecule is character-
ized by a pair of unsaturation electrons
having unpaired (parallel) spins. Tran-
sition from the lowest triplet state to
a vibrational level of the ground state,
arrow 5, represents a phosphorescence
band of the molecule. The transition
probability of the triplet singlet
-f
transition governs the lifetime of the
metastable triplet state.
Because the energy level of the low-
est triplet state of such molecules lies
below the excited singlet levels, the
phosphorescent emissions are of longer
wave length than the absorptions; hence
the phosphorescences of most molecules
lie in the convenient visible or near infra-
red spectral region.
Several authors (9, 11, 14) have
pointed out that a low-lying triplet
state can be present in any molecule
having a multiple bond-possibly some
others. As the phosphorescence emis-
sions from over 200 compounds have
been studied thus far, a fertile field is
open for analytical exploitation.
I n order to observe the triplet -P
singlet transition in an appropriate
8
-;
VIBRATIONAL PATTERN
AND
BAND SPACING
FREWENCY - transition, v-i.e., the highest frequency
C triplet level to the lowest vibrational
Characteristics of phosphorescence
organic molecule, it is necessary in most.
instances to have the substance in a
dense phase or medium. The reason
given (14) for this is that the direct
excitation to the triplet state is much
too weak, or improbable, to allow a suf-
ficient population of this state to ob-
serve phosphorescence; hence, a more
devious route must be depended upon.
First, excitation to an upper singlet
state must be accomplished, followed
by an interconversion to a vibrational
level of the lowest triplet state. Then
the removal of vibrational energy of the
triplet state through collisions must
result to permit the population of the
lon est metastable triplet state, fol-
lowed, finally, by a radiative transition
(phosphorescence) to a vibrational level
of the ground state.
The role of the rigid media is explained
(14) to be such as to remove triplet
state vibrational energy but not remove
the triplet state electron energy by
thermal deactivation to cause the mole-
cule to dissipate its energy by a non-
radiative process (arrow 6, Figure 1).
I n order for thermal deactivation to
occur, electron energy must be con-
verted into kinetic energy of surround-
ing molecules. Such a process is in-
efficient compared to the removal of
vibrational energy from the same mole-
cule. The momentum transfer from
an electron to a molecule is inefficient
because of the large mass difference.
However, the extremely high collision
rate in liquids overcomes this relative
inefficiency and results in no observable
phosphorescence because of the com-
plete thermal deactivation of the triplet
state.
Phosphorescence has been observed
from some substances when in fluid
media (S),but, as stated by Kasha (9),
this is possible only if the mean life-
time is intrinsically short (order of
second or less).
Each phosphorescence is unique for
each molecule and as proposed by
Lenis and Kasha (11) may be used t o
identify phosphors in mixtures.
Rybab, Lochet, and Rousset (18) have
used the 0-phosphorescence (12) as a
means of identifying some aromatic
is characteristic of the electron struc-
ture of the molecule as a whole and
not that of specific groups in the mole-
cule (11, 12). It is characterized by
four parameters:
1. The mean lifetime, T , which is
the length of time associated with the de-
crease in radiant povxr from P to P / e ,
(a, Figure 2 ) .
2. The frequency of the vibrationlees
band of the phosphorescence character-
ized by the transition from the lowest
ground level (Figure 2 , c ) .
3. The vibrational pattern or band
spacing, A v , (Figure 2 ) which represent8
VOL. 29, NO. 2, FEBRUARY 1957 203
 the spacing of the vibrational levels of
the ground electron level.
4. The quantum efficiency, @, the
ratio of the quanta of light phosphoresced
t o the quanta ahsorbed.
FACTORS AFFECTING PHOSPHORESCENT
EMISSIONS
Lenis. Lipkin, and Magel (12) found
the lifetime to be nearly independent of
temperatui e for fluorescein dissolved in
horic acid. whereas Pyatnitskir and
T'inokurova (17) reported that the
phosphorescence lifetime of biphenyl
varied from 1 to 0.4 second when tem-
perature ranged from 90" to 130" K.
Pyatuitskii (16') ieported invmiant life-
times for phthalic and benzoic acids in
the same temperature range. Svesh-
nikov ( 1 9 ) reported a deviation from
the exponential decay law for time
intervals that were long compared to
T. and involved a thousandfold varia-
tion in intensity. He also reported a
dependence of T upon the intensity of
the exciting light. Both phenomena
exhibit an apparent decrease in T
during the time in which a high concen-
tration of molecules exists in the phos-
phorescent state. Kasha (9) reports
that Calvin explained this as due to
magnetic quenching resulting from the
interaction of excited triplet-state mole-
cules with the inhomogpneous magnetic
fields produced by the relative high
concentration of neighboring triplet-
state molecules. Sveshnikov and Pet-
rov (20) have reported that the nature
of the solvent map affect the lifetime.
Lewis. Lipkin. and Magel (12) have
reported that when a liquid is cooled,
the intensity of phosphorescence in-
creases as viscosity increascs. If there
were no thermal quenching in the solid
state, both the intensity and the lifetime
of phosphorescence should be in-
dependent of temperature, and they
have shown this to be the case. They
also report another indication that
collisional deactivation does not alter
the triple-state lifetime in rigid media:
that thc decay curves (phosphorescent
radiance I'S. time, Figure 2,a) are ey-
ponential. They state further that a
complete lack of collisional process can-
not lw shown by such curves, but if
- a s little as 10% of the decay process
is a second-order thermal deactivation,
it will lie detected in the curves.
Thcir deviations have always been
tracrd to phosphorescent impurities
and. in fnct. they have used the cy-
ponential form of the decay curve as a
criterion of phosphorescent purity.
The parameter of quantum yield of
phosphorrscence is too difficult to
measure for use in qualitative identifica-
tion, but i t is desired to have as high
a value as possible in quantitative
phosphorinietry, to increase precision
and to extend the range to lower con-
centrations.
204 ANALYTICAL CHEMISTRY
I n addition to these qualitative parani-
eters, the radiant poFl-er of the phos-
phorescence is a function of several
variables :
The radiant power of the excitation
light ( l a ) (Figure 2 , b )
The length of time of excitation
The concentration of the phosphor or
other substance that may absorb radiation
of the frequency used t o excite the phos-
phor or that may absorb any of the phos-
phorescent emission
The interval of time following cessation
of excitation and photometry
Under conditions of the analysis any
parameter that may affect the lifetime
or the quantum yield
The Beer-Lambert la^ cannot be
applied directly to this type absorp-
tion-emission, because the light energy
absorbed and not that transmitted is
of essence. Danckwortt (6) has ap-
plied a modified form of the law to
fluorometry, and a similar approach
must be made in phosphorimetry.
From the Beer-Lambert law above, the
fraction of radiant power transmitted by
a given mediuni is given by T = e-fcz.
I n fluorometry and phosphorimetry the
fraction of radiant power absorbed, A ,
is of value as a function related directly
to concentration, where A = (1 - T)
= 1 - e-6~1.
Hence the phosphorescent radiant
power, P,, may be expressed as P , =
K' (1 - e-fcL). This equation may be
differentiated andexpanded ( I S )toa form
P, = K" X (2.30 e ~ -
l 2.65 e2c212
2.04
+
- ....
E3C313
in which the higher order terms may be
neglected if the ECI value is less than
0.01. Then the phosphorescent radiant
power may be related directly to the
concentration :
P , = Kc
These considerations make it ap-
parent that in a quantitative applica-
tion, standardization of excitation con-
ditions, resolution time of the phos-
phoroscope, solvents, temperature, and
concentration limits must be observed.
APPARATUS
The essential major pieces of ap-
paratus are shown diagrammatically in
Figure 3: an excitation source, resolu-
tion phosphoroscope, sample container,
and receptor to detect the phosphor-
escent emission.
TWOexcitation sources were inde-
pendently employed. One was a Gen-
eral Electric lOO-watt, A-HI, medium
pressure mercury arc (c), with the outer
glass jacket removed, $1 hiill g:ii e a
mercury line spectrum su$ei imposed
upon an ultraviolet continuum. The
second was a General Electric. 1000-
watt, A-H6 n ater-cooled, quartz-jack-
eted high pressure mercury arc, which
yielded a stronger continuum to about
2270 A., together with a broadened
line spectrum. The rl-HG was en-
closed in a water-cooled compartment.
Both sources were positioned relative
to the sample in such a manner that
quartz collimating and condensing lenses
as n-ell as optical filters coukd be em-
ployed.
Figure 3. Resolution phosphoroscope
The phosphoroscope employed was of
a modified Becquerel type (1) and con-
sisted of two slotted disks mounted
to a common shaft and driven at a
selected speed by a synchronous motor
(Figure 3). The disks were so arranged
that openings in one blade were in line
with uncut portions of the second blade,
to prevent excitation emission from
gaining access to the spectrograph or
photometer. This allowed a sample
on the optical axis between the rotating
blades to be illuminated (excited to
phosphorescence), followed by a period
of darkness, during which time its
phosphorescence was allowed to pass
the second disk and thence to the spec-
trograph or photometer.
The resolution timeof such an arrange-
ment is the length of time betrveen the
cutoff of excitation to the sample and the
clearing of the optical path by the second
blade to allow the phosphorescent
emission to pass. This time is a func-
tion of the motor speed, size and spac-
ing of the cuts in the disks, relative
radial position of the disks to each
other, and the size of the slits.
Two sets of blades were constructed
for the preliminary investigation. The
first set, L-10, consisted of two iden-
tical 11-inch disks cut from 14-gage
aluminum stock with 10 equally spaced
slots 11/2 inches deep. The slots sub-
tended 9" of arc; the unopened portions
27". The second set, L-2, had two
equally spaced slots per disk ivith each
slot subtending 58" of arc.
The phosphoroscope serves two func-
tions: Fiist, it allows the study of
the phosphormcence free from any ex-
citation eniission and, secondly. it
affords a unique way of resolving
simple mixtures of phosphorescences
based upon their mean lifetimes. For
example, if ttvo compounds of different
phorphorescence lifetimes are present
in a mixture, the proper choice of disks
and motor speeds may give a resolution
time which will permit the shorter lived
compound to decay to a degree too
u r a k to lie detrcted and the only
measurablr emission ~vould be from
the longer lived c,oniponent.
O1 A stoppered the Dewar. The samples
10
I 4-NITROBIPHENYL
131 n I I
j )V
c ,
400 500
WAVE LENGTH, mp
Figure 4. Emission curves
In another case where the tn-o life-
times are nearly equal, resolution may
possibly be gained by a spectroscopic
resolution of the phosphorescent emis-
sion or by a wlrction of excitation fre-
quency that nil1 excite one phosphor
and not, the other. I n a more difficult
case where one phosphorescewe caii-
not be conipletely isolated, relative
intensitj- values of t'wo sets of con-
ditions, one set optimum for, one
compound and the second set' optimum
for the other. limy be determined and
the coiiccntrstion found through the
solution of two simultaneous equations
from appropriate experimental data.
The three motors employed rotated
a t 1800. 300, and 60 r.p.m. Hence
\Yitli proper choice of blades and motor
a desired resolving tinie can be chosen-
viz., with the L-10 disks and t'lir 1800
r.p.m. motor theoretical resolving times
from 0 to 1.5 X second could be
obtained by suitable positioning of the
disks on the shaft. I n piactice, lion--
ever, this range in resolving t h e is
limited by the height of the slits em-
ployed and is iepresented in the present
equipment by something lees than 8 O
of arc. K i t h the L-10 disk assembly
available. a precise setting was not
afforded in such a limited angular
range and so the blades ivere synimet-
rically arranged and peimanently fixed
on the shaft.
Sample tubes n-ere of the shape shown
in Figure 3. They n ere made from clear
fused quartz, 20 mm. in outside diame-
ter in the loner portion and 11 inches
high. The samples n-ere placed in a
fused quartz Dewar flask that was held
securely in place by an upright metal
cylinder that supported the two slits
shown in Figure 3. The sample tube
was centered in t h e Dewar by means of
a plastic 1%-asherpositioned in the bottom
of the Dewar and a hole in a cork that
n ere maintained a t f7 O K. by keeping
them submersed in the quartz Dewar
filled with liquid nitrogen. A gentle
stream of dry air entering the bottom
of the metal cylinder and leaving a t the
slits kept the outside of the Dewar
free of condensation during a run.
Either a Hilger glass spectrograph
n ith a linear dispersion of 14 A. per nim.
at 4200 A. and an aperture o f f 14 was
used to record the phosphorescent
spectra, or a n Aminco (American In-
strument Co.) 10-210 photomultiplier
photometer i ~ a sused t o measure rela-
tive phosphorescent radiances.
A photomultiplier photometer is pre-
ferred to a spectrograph on the basis
of time, but does not give a record of the
660
*.' ' PLATE CALIBRATION
a
m
c
REL WIAWE
a.
WORKING CURVES
BENZOPHENONE
e
*COIICEN?RATIOW * IO
X IO4 M
b d
Figure 5.
total emission froiii which wave length
maxima and hand spacing may be de-
termined. However, plans include the
use of a grating monochromator be-
tween sample and photomultiplier pho-
tometer, so that such data may be read-
ily ascertained. An ultraviolet mono-
chromator between the excitation source
and sample should allow better exci-
tation frequency selection, which will
permit in an ideal case the excitation
of one phos~~lior in a mixture and not
the other, or in the usual case! nhere
mixtures are concerned. selection of
optimum conditions of excitation and
resolution times to determine the com-
ponents of a mixture quantitatively.
RESOLUTION OF MIXTURES OF BENZALDEHYDE,
BENZOPHENONE, AND 4-NITROBIPHENYL
The three comjiounds-benzalr~eliy~e,
benzophenone, and 4-nitrobiphenyl-
lvere chosen on the hasiq of their phos-
phorescence properties, not n ith the
intent to dev-elop a method of analysis
for such a mixtuie, but rather to test
the potentialities of the method with
niixtures which requiied both phos-
phoroscopic and spectro-copic rciolu-
tion.
Purified saniples \vel e prepared : i d
dissolved in a miled solvent, EPA i l l ) ,
consisting of purified ethyl ether, iso-
pentane, and ethyl alcohol in a volume
ratio of 5 : 5 : 2 . Such wlutioni nhen
cooled to liquid nitrogen tempei ature
gave clear transparent glasses. The
purity of the qolvent is. important.
Some phosphoirscencc can be oherved
WORKING
4 - N l T /CURVE
/
CONCENTRATION X 12 U
C.
I2t
WORKING CURVES
A, -
I2 O2 a 6 e IO le u ;s
CONCENTRATION X I$ M
Plate calibration and working curves
VOL. 29, NO. 2, FEBRUARY 1957 a 205
in old or unpuriiied solvents, and the
presence of small amounts of water
may muse the supercooled solution to
crack and become opaque.
Suectra of the three uure comuo-
by standard fixing, washing, and drying.
Relative phosphorescent radiant flux
values were determined from photo-
graphic densities by means of plate
calibration (4). A standard benzo-
phenone solution served as a constant
radiant energy source for plate calibra-
tion for the wavelength region used for
benzophenone and benzaldehyde. A
solution of 4-nitrobiphenyl was used for
plate calibration in its wave length
region.
The mean lifetime of 4-nitrobiphenyl
has been reported as 0.080 second (14);
those of benzaldehyde and benzophe-
none, approximately 0.006 second (11,
14).
The phosphorescent properties of this
mixture were such that the 4-nitro-
biphenyl could he resolved spectro-
scopically or phosphoroscopically
(on the basis of its slower decay rate).
Benzaldehyde and benzophenone were
resolved spectroscopically. Their mean
lifetimes are too similar to permit
resolution phosphoroscopically with
present equipment. The spectral dis-
tribution of each pure emission is
plotted in Figure 4. These emissions
are somewhat atypical of phosphores-
cent emissions. Usually the lowest
wave length (highest frequency) band
is the most intense (see Figure 2,c).
These bands are a little more M u s e
than most.
Plate calibration data were obtained
by exposing a sample of the phosphor-
escent substances a t various times.
A log-step sector (4) was found im-
practical because of the excessively long
exposure times required when the phos-
phorescent source was moved away
from the spectrograph slit to accommo-
date the log-step sector mechanism.
Instead, successive exposure of varying
times were made and relying upon the
validity of the reciprocity law for pho-
tographic emulsions:
Exposure = intensity X time
plate calibration curves were plotted
from absorbance values evaluated with
a microdensitometer. These data when
plotted as shown in Figure 5,a, gave a
means of determining relative intensity
values of a source of similar frequency
from its photographic density that was
recorded on that plate. Density os.
relative intensity was plotted rather
than the usual density us. log relative
intensity, in that an extrapolation to
zero intensity is possible and thereby
provides a more convenient may of
206 ANALYTICAL CHEMISTRY
Figure 6. Print of spectrographic plate
Table I. Results of Benzaldehyde-Benzophenone Mixtures
Determined Actual Conon.
Replica. Mixture Concn. X lo4 M x1 0 4 ~ % Error
Results for Benzaldehyde
1 1 6.0 5.6 7.1
2 5.9 5.6 5.4
2 1 6.1 5.6 8.9
2 6.1 5.6 8.9
3 1 6.3 5.6 12.5
2 6.0 5.6 7.1
Results for Benzophenone
X 1O'M x 10514
1 2 5.1 4.7 8.5
1 7.0 7.5 6.7
2 1 7.2 7.5 4.0
2 5.0 4.7 6.4
3 1 7.2 7.5 4.0
2 5.1 4.7 8.5
evaluating data in the lower exposure ship. First, appreciable photodecom-
region (8). The working curves for the position occurred. A sample that gave
three pure components were prepared a relative intensity value of 1.07 for the
from the spectrograms produced. Pho- first 15-minute exposure gave 0.7i
tographic density values were evaln- for the second 15-minute interval, and
ated and from the appropriate plate cali- 0.60 for the third &minute interval,
bration curve t h e relative intensity, indicating considerable photochemical
I, that caused that photographic black- decomposition on exposure t o the excita-
ening was determined and plotted tion light. Secondly, the &el value may
against molar concentration to give the have been too high. The extinction
working curves shown in Figure 5,b, coefficient, e, of 4-nitrobiphenyl calcu-
a. lated a t 3650 A. (the principal exciting
Two working curves are shown for line of the mercnry arc when copper
benzaldehyde and benzophenone. Two sulfate filter is used), however, gives
each were needed to resolve a mixture of an &el value of about 0.01 a t the con-
these two substances spectrophot,o- centration used. But other lines, nota-
metrically. Wave length XI, 413 nip, bly 3350 A. and 3150 A,, may have an
is a maximum emission region for ben- effect in addition t o the 3650 A. line
zophenone and a near minimum emis- on the amount of light absorbed, even
sion region for benzaldehyde (see Figure though they were much weaker than
4); wave length if,427 mp is a region of the 3650 A. line. Also, the rcl value
maximum intensity for benzaldehyde of 0.01 is based on room temperature
and a near minimum for benzophenone. data. and the absorption peaks for the
The working curve for 4nitrobiphenyl 4-nitrobiphenyl may sharpen and e
was not a straight line. There may be may increase considerably when con-
two reasons for the deviation from a ditions change from room t o liquid
linear intensity-concentration relation- nitrogen temperature.
 I n order to hold ecl values for the times to give appreciable plate black- it n a b decided to attempt only a qiiulitn-
mixtures of these compounds within ening. Because of the excessive and tive resolution phosphoroscopically of
the limits described, a concentration of impractical exposures required as well the 4-nitrobiphenyl and do a quanti-
10-4M for 4-nitrobiphenyl was em- as serious errors due to decomposition tative study with more practical sub-
ployed, which required 2-hour expnsure of the sample upon such long exposures, stances.
Figure 6 is a print of a plate taken
on a Hilger spectrograph to show quali-
tatively the spectroscopic and phos-
phoroscopic resolution possibilities of
such a mixture.
Spectrum -4is a record produced hy a
71 15-minute exposure with 1-mm. slit of the
w phosphorescence of a 5 X 10-aM benz-
Y4 6' aldehyde solution taken nith L-10 blades
E 5' driven a t 1800 r.p.m.
Spectrum B , obtained similarly, is of a
= 4.
5 X IO-'?A benzophenone solution.
Spectrum C is from a 5 X lO-'M
4-nitrobiphenvl solution taken under simi-
lar conditions, except that a 2-mm. slit
was used.
Spectrum D is the record from a mixture
of the three substances taken under con-
ditions of C, while E is the record of the
same mixture when the L-2 blades driven
at 60 r.p.m. were employed. The ex-
posure time in the latter case was 30
minutes. No trace of the benzaldehyde
or the benzophenone spectrum is seen
in E , even xith a 30-minute exposure.

The phosphorescent components of a

mixture such as benzaldehyde and
benzophenone may be determined spec-
trophotometrically. Even though their
spectra overlap, a maximum and a
minimum of benzophenone, occurring
ob 2 4 6 8 at approximately 413 and 427 mp,
CONCENTRATION x lo5 M respectively, coincide almost exactly
with a minimum and maximum of benz-
Figure 7. Working curves aldehyde at these two wave lengths.
Because the radiant powers of phospho-
rescenre of pure benzaldehyde and pure
benzophenone are proportional t o con-
centration (Figure 5 , b, d ) and upon the
Table II. Trial Data for Establishing Working Curves assumption that the radiances of the
Condition A , Condition B, two substances in a mixture are addi-
Concn. Relative Radiance X IO3 Relative Radiance X IO4 tive, the following pair of equations
X 105M Run 1 Rim 2 Run3 Run 1 Run 2 Run 3 were used to determine t h e concentra-
Benzophenone tions of the two phosphors in a mixture:
3.48
2.78
98.3
88.3
101.1
89.1;
99.8
93.6
42.7
41.3
44.8
36.8
42.8
36.1
+ + + b)
PI = ( ~ I C , a) (k2Cb (1)
2.09 65.1 65.0
45.7
67.1
48.2
29.3 29.4 26.7 p2 = (k3Co + c ) + hcb f d ) (2)
1.39 46.1 20.5 19.8 19.0
0.70 23.0 23.6 24.2 11.0 10.5 9.7 where P I is the phosphorescent radiant
(Solventi (0.32) (0.22) (0.37) (4.0) (2.7) (2.1) power of the mixture found experimentally
a t 413 mp, while P1 is the radiant power
.icetophenone of the mixture found a t 427 mp. k l
Relative Radiance X lo4 and ks are the values of the slopes obtained
from the working curves of benzaldehyde
8.99 79.0 83.6 86.2 90 0 91.4 94.3 at 413 ,m
! XI, and 427 m p , An, respectively
7.19 65.6 68.8 73.7 76.6 76 3 79.9 (Figure 2); a and c are the values of the
5.39 51.5 j2.0 55.1 59.8 59.8 62 0 intercepts of these lines. kl and kr are
3.60 34.2 38.9 38 9 42.6 42.5 44.7 the values of the slo es obtained from the
1.80 19.7 20.2 20.3 23.5 23.4 24.5 working curves of Eenzophenone at 413
(Solvent) (3.5) (3.21 (4.3) (4.2) (2.8) (2.1) mp, XI, and 427 mp, A*, respectively.
c, and cb are the molar concentrations
of benzaldehyde and benzophenone, re-
spectively, in the mixture. The terms
Table 111. Acetophenone-Benzophenone Mixture
(klC. + a) and + c) are the slope-
interce t values for the intensities of pure
benxaldkhyde solutions a t AI and x2;
Av .
Concn. Benzo-
phenone X 106.U
Concn. Aceto-
phenone X lO5M while the terms ( k & + +
b ) and (klCb d )
are the slope intercept values of pure
Relative Radiance X l o 3 ~

De- De- benzophenone a t A1 and AI, respectively.

Rfis Condition A Condition B termined Actual termined Actual Simultaneous solution of the two equa-
I 25 4 31 6 0 77 0 70 2 02 1 80 tions yield concentration values for the
I1 97 2 5 19 3 14 3 48 0 98 0 90 t a o components.
111 22 3 9 13 0 48 0 35 8 22 8 99
Samples were prepared for estahlish-

VOL. 2 9 , NO. 2, FEBRUARY 1957 207

ing working curves of benzaldehyde in
the range 5.0 X to 1.0 X 10-3MJ Table IV. Preparation of Working Curves
and from 5.0 X to 1.0 X IO-‘M
for benzophenone. Two mixtures were
prepared and run in triplicate. Each
replicate together with exposures for Diphenylamine
plate calibration and working curves 8.56 63.2 63.0 62.1 43.2 45.0 43.9
were recorded on separate photographic 6.85 49.4 50.1 49.4 32.0 35.0 34.1
plates. The slit width used in all runs 5.14 40.8 37.7 38.6 25.6 25.6 25.5
was 2 mm. with an exposure time of 10 3.42 27.0 29.9 26.3 17.6 17.9 16.9
minutes. A 1800 r.p.m. synchronous 1.71 15.7 14.1 13.6 8.7 9.3 8.4
(Solvent) (1 . O ) (0.1) (0.1) (0,9) (1.0) (1 . O )
motor was used to drive the ten-slot,
L-10 phosphoroscope. Table I lists Triphenj-lamine
the results of this test together with Relative Radiance X 103
the relative per cent error. 2.85 91.0 90.4 88.8 78.7 77.0 75.0
2.28 74.5 72.1 74.5 60.9 62.0 61.5
1.71 60.4 45.6 55.1 49.7 45.6 44.6
PHOSPHOROSCOPIC RESOLUTION OF 1.14 40.6 37.9 36.9 33.4 31.4 31.3
ACETOPHENONE AND BENZOPHENONE 0.57 21.1 18.1 19.3 15.9 15.7 15.7
A resolution based upon the mean life-
times of the phosphorescent states was
attempted, using acetophenone and
benzophenone as the phosphorescent Table V. Diphenylamine-Triphenylamine Mixtures
components of a mixture. The un- 4v. Relative Concn. Diphenyl- Concn. Triphenyl-
corrected mean lifetimes of these two Radiance X lo3 amine X 106M amine X lO6M
compounds have been evaluated by Mix Condition A Condition B Detd. Actual Detd. Actual
Gilmore, Gibson, and McClure (’7) I 330 16.6 1.98 1.71 0.58 0.57
as 0.008 second for acetophenone and I1 962 75.8 1.00 0.86 2.78 2.85
0.006 second for benzophenone. Here I11 70,s 12.4 8.18 8.66 0.30 0.28
also, the compounds were chosen on a
basis of their phosphorescent properties,
t o test the potentialities of the method
and not necessarily to develop a de-
termination for the specific compounds. 5. c ’ \ ‘\ 2
\ ,
The apparatus and procedure as de- \
,
$

\\
scribed were used, except that the spec- \
\
trograph was replaced by an Aminco \\ ‘,
photomultiplier photometer with a Type
931-A photomultiplier tube to measure T
phosphorescent radiant energies. The 4.c
indicating circuit in the photometer D
was modified t o include auxiliary re-
sistors and capacitors to give the neces-
sary time constant in the circuit to w
prevent the undesirable “kick” of the
meter hand, as each pulse of radiant 0,

energy activated the photomultiplier -I

tube when the slow-speed motor was 3.c

employed to drive the phosphoroscope.
Excitation was accomplished with the
use of an A H 4 mercury arc filtered by
7 cm. of water and 1.7 cm. of naphtha-
lene in cyclohexane (7.54 grams per
liter), The latter filter removed the
unwanted region of the mercury spec- T
‘ I
trum below 316 mp (IO). A dark blue, 2.0
5.40-mm. Corning KO. 5030 glass filter
was used before the photomultiplier to
reduce background radiation that de-
velops seemingly from ice formation
in the liquid nitrogen surrounding the 0 250 300 350
sample and from a very feeble emission WAVE LENGTH, my
from the solvent.
The phosphoroscope was operated Figure 8. Absorption curves
under two sets of conditions to obtain
the necessary data for the determination
of the mixture. Condition A utilized
the L-10 blades and 1800 r.p.m. motor
to give a resolution time of approxi- second. The data obtained under this Prepared mixtures were run using the
mately 0.001 second. The working condition gelded Curves in Figure 7: described Drocedure, with results listed
curves based on data obtained under
these conditions are shown in curves A ,
Figure 7. Condition B utilized the L-2
blades and 300 r.p.m. motor to give a
resolution time of approximately 0.02 ground. zero. The relative radiant energy val-

208 e ANALYTICAL CHEMISTRY

ues (photometer scale readings), cor-
rected for background and sample
tube size, listed for the mixture are the
average values of triplicate determina-
tions. The average error is approui-
mately lo%, excluding the obviously
poor acetophenone result of miuture
111, where its concentration n a s in the
exceedingly lorn range of 3.5 X 10-6M.
DETERMINATION OF DIPHENYLAMINE AND
TRIPHENYLAMINE BY SELECTIVE EXCITATION
To test the resolution potentiality of
the method based upon the selective
excitation of the components of a
mixture, diphenylamine and triphenyl-
amine nere used. The mean lifetimes
and phosphorescent spectra are such
(11) that either a phosphoroscopic or a
spectroscopic resolution could probably
be gained, but the ultraviolet absorption
spectra of the t n o , although similar,
nere just different enough to offer a
challenge for the method on the basis
of selective excitation.
Figure 8 shows the similar absorption
cliaracteristics of the t n o compounds.
Curve D represents diphenylamine.
n liile T represents tiiphenylamine for
the wave length region nhere absorp-
tion may give rise to phosphorescence
I T A was used as solvent. Inspection
of the curves indicates the most prob-
able wave length legion that n ould per-
mit resolution of these two would be in
the 325- to 350-mp range. Here the
absorption, and hence the excitation,
would be small for the diphenylamine.
but appreciable for triphenylamine.
If the excitation source could be filtered
to allow the passage of radiation greater
than 325 mp, but highly absorb radia-
tion with shorter wave length, it would
be possible t o excite the triphenylamine
more strongly than the diphenylamine.
I<asha (10) has reported seYe1L.l selec-
ti7.e filters for the ultraviolet region,
t n o of which have been used here.
The broken curve labeled 1 represents
the rather sharp cutoff in the 310-mp
region due to a solution of potassium
acid phthalate. Cur?-e 2 is an approxi-
mation of the cutoff due to a solution
of naphthalene. Yeither dashed c u n e
is quantitative and they are included
only to help visualize the n a y in n liich
resolution is gained.
The folloning changes nere made in
the npparatus:
The L-10 blades driven I Jan~ 1800
r.p.m. motor were used in the two sets
of conditions used for resolution. Con-
dition -4,under which both coniponents
were stronqly excited, u t i l i z d a 7-cm.
water filter and a 1.7-cm. aqueous so-
lution of potassium acid phthalate
(2.94 grams per liter) t o filter the ex-
citation emission from an A-H4 lamp.
Curves A in Figure 9 u-ere obtained from
data obtained in this n a y . Condition
13, nhich affords a stronger evritsttion
of triphenylamine than diphenylamine,
utilized a 1.7-cm. cyclohexane solution
of naphthalene (7.5 grams per liter)
to replace the phthalate filter used in
Condition A. Curves B in Figure 9
mere obtained in this way.
The data from which these curves
were obtained as well as data from pre-
pared mixtures are listed in Table IV.
The concentrations of the unknowns
were determined as described above.
The ayerage error of the siu mixtures
determined is approuimately 8%.
101
e 9’
x 8’
TRIPHENYLAMINE
Y 7.
:6
$5
u 4
F 3.
$2.
I.
’
I6 io
00 IO
CONCENTRATION X ( I ) Becquerel, E., A n n . chim. et p h y s .
/ 27, 539 (1871).
2
x‘
Figure 9. Working curves
CONCLUSIONS
These experiments indicate that cer-
tain organic compounds can be deter-
mined by analysis of the phosphorescent
emissions excited in the compounds
under appropriate conditions. Where
the method has been challenged by
choosing compounds of very similar
absorption or phosphorescent charac-
teristics, initial results have been grati-
fying. The average relative error found
upon determining the concentrations of
known solutions by this method was
about lo%, excluding an unfavorable
result involving benzophenone in a
mixture with acetophenone, where the
concentration was a t the exceedingly
low value of 3.5 x 10-61M. The data
in Table I substantiate the statement
that the precision determined by run-
ning replica samples is much higher than
the accuracy obtained in these initial
runs, indicating that the accuracy can
be improved by controlling experimen-
tal parameters more carefully. To
this end a calibration of the photo-
multiplier photometer is indicated, as
n ell as selection of a suitable phospho-
rescent substance that will serve as a
standard to duplicate excitation and
photometry conditions from day t o
day.
Mixtures containing more than two
phosphorescent substances can be an-
alyzed n-ithout physical separation by
the proper choice of excitation fre-
quency, resolution time, and emission
frequency. The method is cur1 ently
being applied in areas where its high
sensitivity, low limit of detertahility,
and resolution are advantageous.
ACKNOWLEDGMENT
The authors wish to thank Nichael
Kasha for his proposal of and assist-
ance in this investigation, and the
Office of Ordnance Research, U.S. Army,
which has supported this work.
LITERATURE CITED
30
(2) Boltz, D. F., “Selected Tropics in
l l o d e r n Instrumental Analysis,”
pp. 202-9, Prentice-Hall, S e w
’Iork, 1952.
(3) Boudin, S., J . chim. p h y s . 27, 286
(1930).
( 4 ) Brode, W. R., ‘,Chemical Spectros-
copy,” 2nd ed., J. Wiley, KPW
I o r k , 1943. pp. 100-14.
(5) Bnttolph, L. J., J . Opt. SOC.t l m e r .
29, 124 (1939).
(6) Danckm-ortt, P. W., “Lumineszm7-
Analysis in filtrierten nltrariolettm
Licht,” 4th ed., p. 7, J. W. Cd-
nards, Ann ;Irbor, Nirh., 1944.
( 7 ) Gilmore, E. H., Gibson, G. E . ,
McClure, D. S., J . Chem. Ph?ls.
20, 829 (1952).
(8) Jablonski, -4.,2. Physik. 94, 38
(1935).
(9) Kasha, M,,Chem. Revs. 41, 401
(1947).
(10) Kasha, )I., J . Opt. Sac. A m e r . 38,
929 (1948).
(11) Lewis, G. N., Kasha, If.,J . Am.
Chem. SOC.66.2100 (1944).
(12) Le&, G. ii,, Lipkin, D., IIagel,
T. T., Ibid., 63, 3005 (1941).
(13) Lothian, G. T., J . SOC.C h e m . I n d .
61, 58 (1942).
(14) RIcClure, D. S., J . Chem. Phys. 17,
905 (1949).
(15) Pringsheim, Peter, “Fluorescence and
Phosphorescence,” Chap. V I I , In-
terscience, S e w York, 1949.
(16) Pyatnitskii, B. A , , Doklady Akad.
IL’auk S.S.S.R. 68, 281 (1949).
(17) Pyatnitskil, B. -4., Vinokurova, T. P.,
Ibid., 68, 483 (1949).
(18) Rybab, B., Lochet, R., Rousset,
A . , Compt. rend. 241, 1278 (1955).
(19) Sveshnikov, B. Ya., Conipt. rend.
acud. sci. U.R.S.S. 51, 429 (1946).
(20) Sveshnikov, B. Ya., Petrov, A . A . ,
Dokludy Akad. ,\‘auk S.S.S.R.
71, 461 (1950).
RECEIVED for review April 10, 1956.
Accepted Sovember 7 , 1956.
VOL. 29, NO. 2, FEBRUARY 1957 209