V O L U M E 26, NO.

7, J U L Y 1 9 5 4
The final correction curve may now be constructed by plotting
the values of function g from Table I for the most appropriate
geometry against the values of x obtained by dividing the corresponding values of X by the value of k determined in c. From
the curve thus obtained the proper self-absorption correction
factor mav be read for any thickness. Dividing the specific
activity o f t h e sample by this factor, one obtains a fairly reliable
estimate for the specific activity a t infinite thinness.
DISCUSSION

Several authors (1) obtain satisfactory agreement b e b e e n

their experimental data and self-absorption curves of the type
( 1 - e-kz)/kx in both narrow- and large-angle geometries. According to the authors results, this is due to the circumstance
that curves similar in shape are obtained in all geometries, but
different k-values would be required if this simple formula were
used in different geometries. Figure 1 shows that k cannot be
uniquely defined without taking the geometry into account.
This is confirmed by Figure 2, which shows that almost identical
curves may be obtained with two different geometries by merelv
doubling the value of k . The zero geometry curve, (1 - e+.)/
kz, with k = 212 and the 50% geometry curve, F(kx) = g(0.5,
kz), are practically coincident for kx > 3, and do not differ anywhere by more than 10%. Thus, if the above self-absorption
curve is used, which is the authors zero geometry curve, values
of k up to twice as large as the actual k-values had to be used.
Using this procedure the same value, k = (0.25 zk 0.01) sq. cm.
per mg., is obtained from both the narrow- and large-angle geometry data. This value is in satisfactory agreement with the
value k = 0.26 sq. cm. per mg. obtained from the empirical
formula for light absorber elements, k = 0.022/Em1.33,as quoted
by Siri ( 3 ) ( E , = 0.156 mev for carbon-14). Figure 1 illustrates the fit between the experimental data and the theoretical
curves with the above choice of k .
The procedure proposed here may seem somewhat more complicated than those previously employed, but it appears to be
more consistent and, in the authors experience, more precise.

1161
The error in the extrapolation to zero thickness which is inherent
in all such procedures is still present but, to a certain extent,
reduced by the possibility of adjusting experimental curves obtained in different geometries to the theoretical curves through
the proper choice of the absorption coefficient, k. Standard deviations from 1.2 to 6.8% were obtained in three entirely different
series of experiments in which different samples, mountings, and
instruments were used. I t would be desirable to take backscattering and self-scattering into consideration. However, the
method appears practicable even without these corrections, and
preferable t o previously used methods for carbon-14 counting in
organic substances.
ACKNOWLEDGMENT

The authors are indebted to William M. Stokes of this laboratory for the experimental data on cholesterol digitonide.
LITERATURE CITED

(1) Baker, R. G., and Katz, L., ArlucZeonics.11, No. 2, 14 (1953).

(2) Jahnke, E., and Emde, F., Tables of Functions, pp. 6-9,
New York, Dover Publications, 1945.
(3) Siri, W. E., Isotopic Tracers and Xuclear Radiations, p. 58,
New York, hlcGraw-Hill Book Co., 1949.
(4) Stokes, W. IM., Fish, W.A., and Hickey, F. C., J . Biol. Chem.,
200, 683 (1953).
(5)
.
. , Yankwich. P. E., Norris. T. H. and Huston. J., A x . 4 ~CHEM..

19, 439 (1947).

( 6 ) Yankwich, P. E., and Weigl, J. W., Science, 107, 651 (1948)
(7) Yankwich, P. E., and Weigl, J. W., as quoted by Calvin, M.,
Heidelberger, C., Reid, J. C., Tolbert, B. &
and
I.,
Yankwich,

P. E., Isotropic Carbon, p. 305, New York, John Wiley

8: Sons, 1949.
(8) Ibid.,p. 318.
RECEIVED
far review December 11, 1953. Accepted Bpril 22, 1954. Study
supported by grants-in-aid from the American Heart Association and t h e
American Cancer Society upon recommendation of the Committee on Growth
of the National Research Council.

Rapid Spectrophotometric Determination of Fluoride

With Zirconium-Eriochrome Cyanine R lake
STEPHEN MEGREGIAN
Division o f Dental Public Health, Public Health Service, Washington, D. C.

A simple, rapid technique which can be applied directly

to water samples or to fluoride distillates usually gives
results within 5 minutes with an accuracy within 0.02
mg. per liter of fluoride in the fluoride concentration
range of 0.00 to 1.40 mg. per liter. It can replace the
thorium titration for distillates in which the recognition of the thorium-alizarinend point is often very difficult. With the exception of sulfate, interference due to
ions normally present in water is largely eliminated.
A simple procedure for the correction of sulfate interference is included.

IOLOGICAL samples, which are analyzed for fluoride content, are usually ashed, and the ash is subjected to the Willard
and Winter ( 1 1 ) distillation in order to remove the fluoride from
interfering substances. The fluoride in the distillate is usually
determined by thorium titration. This requires concentration of
the distillate, careful buffering, and a tedious titration in which

the thorium-alizarin end point must be approached with great

care.
Water samples are usually analyzed directly for fluoride by one
of many colorimetric methods, all of which are based on the complex-forming property of the fluoride ion with various cations.
These colorimetric methods are all subject, in varying degree, to
the many interferences normally present in water and lack either
rapidity or accuracy, or both, when applied directly to water
samples.
The method proposed, when applied to distillates, eliminates
the necessity for further manipulation of the distillate once it has
been collected. The fluoride value of the distillate can be obtained with great precision and accuracy within 5 minutes following the completion of the distillation. If used directly on a water
sample, only the sulfate concentration need be known and a correction applied. The total time required to obtain both fluoride
and sulfate readings can usually be limited to 10 minutes.
Eriochrome Cyanine R is listed as No, 722 in Colour Index

ANALYTICAL CHEMISTRY

1162
(6) and is the sodium salt of o-sulfohydroxydimethylfuchson dicarboxylic acid. Its structural formula is given as:

,COONa

'COONa
Thrun (9) developed a colorimetric method for fluoride using
the aluminum lake of Eriochrome Cyanine R a t controlled pH.
Richter ( 4 ) applied the aluminum lake to photometric measurement. Saylor and Larkin (6) used it as an internal indicator in
the titration of fluoride with aluminum. Milton et al. ( 2 ) , used a
closely related dye, Chromazurol-S, C.I. No. 723, as an indicator in the titration of fluoride with thorium. Revinson and
Hartley (3) worked out a photometric method for fluoride using
thorium and Chromazurol-S, and Willard and Horton (IO)
compared many indicators for use in the titration of fluoride with
thorium. They judged Eriochrome Cyanine R and Chromeazurol-S to be inferior to Purpurin sulfonate and Alizarin Red S
for this purpose. A search of the literature revealed no previous
record of a reaction involving zirconium with Eriochrome Cyanine R.
PRELIMINARY STUDIES

The Megregian and Maier ( I ) photometric reagent for determination of fluoride ion concentration was studied to determine
the optimum conditions under which this reaction should be carried out. The results of this study indicated that the reaction
between zirconium, alizarin sulfonate, and fluoride ions could best
be carried out in the region of 0.9N hydrochloric acid. At this
acidity, the reaction rate between zirconium and alizarin was
increased, and the fluoride effect on the reaction was more pronounced. Also, the interferences of alkalinity, chloride, and
ferric iron were largely eliminated, and the degree of sulfate interference was decreased. On the other hand, the reaction became
very sensitive to slight temperature variation, and low concentrations (less than 1.0 p.p.m.) of phosphate increased the reaction
rate between zirconium and alizarin. When both phosphate and
sulfate were present, the interference due to a given concentration
of sulfate was increased.
This study indicated that it was possible to prepare a reagent
which was more sensitive to slight differences in fluoride concentration, less subject to interferences, and required less time for
reaction equilibrium to be reached. However, the advantages
gained with this reagent were outweighed by the uncertainty imposed by the possible presence of trace quantities of phosphates in
water samples and the necessity for close temperature control.
Using the knowledge gained from this study, a search was ini-

tiated for a compound other than alizarin which might undergo a

color reaction with zirconium and fluoride ions without introducing errors due to minor constituents normally present in water.
Over 200 dyes and other organic compounds were screened.
Table I lists the compounds which were found to produce a color
with zirconium and indicates the influence of fluoride on the color
produced.
The choice was narrowed doivn to Eriochrome Cyanine R and
Chromazurol-S. The former was selected for further study
because: the zirconium lake is soluble and stable, equilibrium
between zirconium, Eriochrome Cyanine R, and fluoride ions is
reached very quickly a t room temperatures, the dye, as purchased,
can be used immediately without further treatment, and the
change in absorbance per unit fluoride per centimeter of light
path is very large
Reagents. All of the colorimetric methods proposed for the
measurement of fluoride ion concentration have a common defect-namely, the fluoride ion is not directly involved in the
primary color-producing reaction. In every method proposed
to date, fluoride ion concentration is measured by the amount
of decolorization of the product of two reactants, one of which
can form a complex with fluoride ions and can thus be restrained
from entering into the color-forming reaction. A satisfactory
reagent for this type of reaction should have:

A. Maximum sensitivity to fluoride ions in the desired concentration range

B. A color iange which can he adequately measured by available colorimetric or photometric instruments
C. Nonsensitivity to other ions or substances commonly
occurring with fluoride in the test medium
D. Conformity to Beer's law
E. A low temperature coefficient
F. High buffer capacity a t the optimum pH of the reaction
G. Reagent stability
H. Reaction stability a t equilibrium
I. A minimum time interval for the reaction to reach equilibrium
J. A simple procedure
Sone of the published methods in current use conform to all of
the above ideals. The propofied method has advantages A, B, C
(except for sulfate), D, E, F, G, H, I and J.

Apparatus. Beckman Model B spectrophotometer or equivalent and 10-mm. matched cuvettes.

Preparation of the Reagents. REAGENTA. Eriochrome
Cyanine R (1.800 grams) is dissolved in distilled water and diluted to 1liter.
REAQENT
B. Zirconyl chloride octahydrate (0.265 gram) or
zirconyl nitrate dihydrate (0.220 gram) is dissolved in about
50 ml. of distilled water. Seven hundred milliliters of concentrated hydrochloric acid (reagent grade, specific gravity 1.19)
are added to the zirconium solution, and the mixture is diluted t o
1liter with distilled water.
Reference Solution. Ten milliliters of Reagent A are added
to 100 ml. of distilled water, and then 10.0 ml. of a solution prepared by diluting 7.0 ml. of concentrated hydrochloric acid t o
10.0 ml. with distilled water are added. This solution is used
for setting the reference point (zero) of the photometer.
Reagents A and B cannot be combined as a
single reagent, because prior mixing produces precipitation on prolonged storage.
Procedure for Willard and Winter Distillates.
Table I. Compounds Producing Color Change With Zirconium
A 50.0-ml. portion of distillate, or aliquot diluted
Color in
Effect of
t o 50.0 ml., and containing no more than 0.070 mg.
Compound, (1 M1. 0.5%
Color in
0.1N HC1 + Fluoride,
of fluoride (1.40 p.p.m.) is taken and adjusted to
per 50 MI. S o h )
0 . 1 N HC1 9 Mg. Zr/L.
5 Mg./L.
Remarks
a standard temperature (&2O C.). Five milliliters
Alizarin Blue Sa
Red
Brown
Marked
Zr lake preci itates
Blue intensi%es in presof Reagent A followed by 5.0 ml. of Reagent B are
Lighter blue Weak
Alizarin Sapphire BLN" Blue
added to the sample and mixed well. The photomence of F ions
Alizarol Azurine ECAa
Orange
Red
Marked
Ppt.
eter is set a t zero absorbance a t 527.5 mp. (range
Blue-violet Red
Weak
Ppt.
Benzopurpurin 4B
525 to 530 mp.) with the reference solution and the
Carmine
Pink
Lavender
Marked
Slow reaction, ppt.
Chloranilic acid
Pink
Purple
Very weak . . . . . . . . . . . . .
absorbance of the sample recorded.
C hromazurol-S b
Orange
Lavender
Weak
Marked
Slow
Rapidreaction
reaction, ppt.
The fluoride value of the sample aliquot is read
Curcumin
Yellow
Orange
from
a curve prepared by subjecting standard soluEriochrome Cyanine Rb Orange
Lavender
Marked
Rapid reaction, ppt.
Quinalizarin
Orange
Violet
Marked
Slow reaction, ppt.
tions of fluoride to the procedure and plotting the
Quinizarin-2-sulfonate
Orange
Red
Marked
Rapid reaction, ppt.
absorbance values a t the standard temperature.
Standards should be selected in the range of 0.00
a General Aniline & Film Corp., 435 Hudson St., New York, N. Y.
to 0.070 mg. of fluoride (0.00 to 1.40 p.p.m.).
b Geigy Co., Inc., 89 Barclay St., New York, N. Y.
The resultant curve should be a straight line of

1163

V O L U M E 26, NO. 7, J U L Y 1 9 5 4
negative slope. As stated previously, the color reaction is immediate and stable. There is no waiting period for color to develop, and readings can be taken immediately, or a t any other
desired time without significant change in absorbance, provided
the temperature is within the limits set forth. A new curve must
be prepared for each fresh batch of Reagent A or B.
Procedure for Water Samples. A 50.0-ml. water sample, or aliquot diluted to 50.0 ml. and containing no more than 0.070 mg. of
fluoride (1.40p.p.m.), is adjusted to the temperature of the standard curve (&2O C.). If the sample contains free chlorine, two
drops of 0.1-V arsenite solution are added for each part per million
of chlorine present. Five milliliters of Reagent A followed by
5.0 ml. of Reagent B are added to the sample and mixed well.
The photometer is set at zero absorbance a t 527.5 mp. (range
525 to530 mp.) with the reference solution and the absorbance reading of the sample is taken within 5 minutes after the reagents
have been added. If the absorbance reading of the sample
falls beyond the range of the standard curve, the combined sulfate and fluoride Concentrations of the sample are too high. In
this case the procedure must be repeated with a smaller sample
aliquot.
If the presence of aluminum ion is suspected in the water
sample, the reaction is permitted to continue a n additional 15
minutes and another reading taken. An appreciable drop in
absorbance indicates the presence of aluminum ions as an interference. At this point, holding the reacted sample for 2 hours
bpfore making the absorbance reading will eliminate the interference effect of up to 5.0 p.p.m. of aluminum. Or the water sample
can be subjected to the Willard and Winter distillation and the
distillate analyzed according to the procedure for distillates given.
The correction for sulfate concentration in the sample aliquot
is made as follows: The fluoride value of the absorbance reading
is obtained from the standard curve; this value is applied to
Figure 3 and the sulfate error determined; the sulfate error is then
subtracted from the first value to obtain the corrected fluoride
value of the aliquot; and this value is multiplied by the aliquot
factor to obtain the fluoride concentration of the original sample.

of fluoride. Thus, readings which are made a t temperatures

f 2 " C. from that of the standard curve are well within the
standard deviation of the method.
Interferences. The Willard and Winter distillates of ashed
biological samples can be analyzed directly by this procedure.
The complete analysis of a distillate need not take longer than
5 minutes. The distillate need not be neutralized or concentrated, and buffer must not be added. The only possible interference would be the presence of large quantities of volatile anions,
such as nitrate or unprecipitated chloride, which would come over
during the distillation and produce high results in the subsequent
analysis. If this is suspected, a portion of the distillate can be
neutralized to the phenolphthalein end point with LON sodium
hydroxide and the neutralized aliquot taken for the fluoride
analysis. The effect of large quantities of volatile anions on the
zirconium-Eriochrome Cyanine R fluoride reaction is to increase the acidity a t which the reaction takes place, depressing
the total color produced.
For water samples which are subjected to direct analysis, the
tolerance to various interferences is shown in Table 11.

Table 11. Interferences Normally Present in Potable

Waters

g:;++

L
~

.oo

,
wl":

' J
4

&
\j.
650

,D?

LEb6-H

,wJ>

I
20 0

1
"3 0

M CR35RAMS F

(3 0

8
B,"

h 50.0 "L

Figure 1. Absorption Spectra and Standard Curve for

Zirconium-Eriochrome Cyanine R Reaction

Absorption Spectrum. The absorption spectra and a standard

curve of the zirconium-Eriochrome Cyanine R reaction are presented in Figure 1. The spectral curve for the unreacted Eriochrome Cyanine R (reference solution) is very near that of the
zirconium lake. Therefore, this color system requires the incident light to be of high purity a t the wave length of measurement.
If a filter photometer is used for absorbance measurement, the
filter should have a band pass no greater than 20 mp. a t 527.5 mp.
The wave length of 527.5 mp. was chosen, because, as illustrated
in Figure 1, the standard curve measured a t that value is a
straight line in the range of 0.00 to 1.20 p.p.m. of fluoride with
only a slight deviation from 1.20 to 1.40 p.p.m. Also, the maximum differential exists at, this wave length.
Temperature Effect. The best temperature a t which the reaction should be carried out is in the range of 22 O to 28" C. Within
this range a 3" C. differential between the temperature of the
standard curve and that of a sample is equivalent to 0.01 p.p.m.

Concentration, %./Liter,
Necessary t o Produce Fluoride Error
Equivalent to
Interfering Substance
0.02 mg./liter F
0.10 mg./liter F
Alkalinity (as CaCOd
3300.0
Undetermined
3500.0
Undetermined
in n
All n
2.5
6.0
1.0
2.0
5.05
Free chlorine
Reduce with arsenite '
Compensate
Turbidity, color
When reaction is permitted t o continue for 2 hours.
_"..I

KO interference was noted with calcium ion concentrations of

200 p.p.m. or magnesium ion concentrations of 100 p.p.m. The
effect of very large concentrations of bicarbonate or hydroxide
ions (alkalinity) is to reduce the acidity of the reaction, thereby
increasing the total color produced by the zirconium-Eriochrome
Cyanine R system and resulting in low fluoride values. Very
large concentrations of chloride ion influence the reaction in the
same direction. Ferric iron in high concentration produces a
color reaction with Eriochrome Cyanine R, thus increasing the
apparent color of the system and resulting in low fluoride values.
Sone of the above interferences is normally present in sufficient
quantity in water supplies to interfere with the fluoride determination.
Phosphate and hexametaphosphate (also other polyphosphates)
act on the zirconium-Eriochrome Cyanine R reaction in a way
similar to fluoride, in that they complex the zirconium and inhibit
the color forming reaction. Water samples containing phosphates in excess of that listed in Table I1 should be distilled prior
to fluoride analysis.
Aluminum does not affect the primary color reaction a t the
acidity specified. I t reacts with fluoride to form the complex,
anion (A1 F6)--- similar to (Zr Fa)--. Therefore, if no fluoride is
present, aluminum does not interfere. When both fluoride and
aluminum are present in a sample which has been treated with the
reagents, the fluoride is distributed between the zirconium and
aluminum ions. Under the reaction conditions imposed, the
zirconium slowly removes the fluoride from the aluminum complex and after a sufficient lapse of time (about 2 hours), the aluminum effect is no longer evident.
An interesting phenomenon occurs when aluminum is present
in the sample as the metaaluminate ion-Le., when hydroxide
alkalinity is present in the original water sample. Under these
conditions. the interference effect of a given concentration of

ANALYTICAL CHEMISTRY

1164
aluminum on the zirconium-Eriochrome Cyanine R fluoride
reaction is greatly reduced, and it disappears in 10 to 15 minutes.
A logical explanation for this phenomenon would be that in
solutions containing aluminum, fluoride, and hydroxide ions, the
aldminum exists as the metaaluminate anion and does not complex with the fluoride present. When the reagent is added to this
system, the fluoride, being in the ionic form, can immediately
complex with the zirconium which has been added. By the time
the aluminum ion has been released from the metaaluminate anion
there is very little fluoride left; to form the aluminum complex.
Therefore, it exerts much less interference on the zirconiumEriochrome Cyanine R fluoride reaction. Advantage can be
taken of this phenomenon when the analyst has many samples
containing both aluminum and fluoride. Simply by treating the
water sample with a few drops of 1.0s sodium hydroxide until
the solution is alkaline and adding the reagents as in the regular
procedure, the waiting time for elimination of aluminum interference can be reduced from 2 hours to 15 minutes,

000
100

200

300

100

DO0

$00

P P M 50,

Figure 2.

Effect of Sulfate Interference

Free chlorine and other strong oxidizing agents attack Eriochrome Cyanine R thereby reducing the total color of the system.
They can be easily reduced with sodium arsenite. -4slight excess
of arsenite does not affect the fluoride reaction. If desired, arsenite can be incorporated in Reagent A.
Turbidity and color in the original sample can be compensated
by the following technique: A 50-ml. duplicate of the water
sample is taken and treated with 5.0 ml. of Reagent A. This is
followed by 5.0 ml. of a hydrochloric acid solution of the same
acidity as in Reagent B. The absorbance of this solution is subtracted from the absorbance of the original water sample, which
had been treated with both reagents. The net absorbance is
then used to determine the fluoride concentration.

tration by running a line through A and B to C. Subtract the

value obtained on line C from A to give the fluoride value of the
aliquot. Multiply this value by the aliquot factor to obtain the
fluoride content of the original sample. Lines B and C are expansions of A and B and should be used when the sulfate concentration is in the range of zero to 100 p.p.m.
Any procedure requiring added manipulations to remove interferences would nullify the advantages gained by using a rapid
method for determining fluoride concentration. Thus, sulfate
might be removed from the sample by barium precipitation or by
the usual distillation procedure, but the advantages gained in using the rapid method would be counterbalanced by the time required to effect a separation by these conventional procedures.
Another possibility would be to diminish the sulfate effect on the
reagent by adding a known amount of sulfate to the reagent.
Khen this is done, the reagent becomes less sensitive to fluoride,
and the measurable fluoride range is reduced. Large quantities
of sulfate in the reagent render it practically insensitive to fluoride,
and this cannot be counterbalanced by increasing the zirconium
concentration of the reagent.
Fortunately a method exists whereby the sulfate concentration
of a water sample can be quickly and accurately determined, and
by use of Figure 3 a correction can be applied, which eliminates
the necessity for time-consuming separations without materially
sacrificing the accuracy of the rapid method. A slightly modified
procedure based on Sheens ( 7 ) turbidimetric method for sulfate
is described below.
Reagents. Barium chloride dihydrate crystals, reagent grade,
20 to 30 mesh.
Acid salt solution. Sodium chloride (240 grams) is dissolved
in 900 ml. of water.
Twenty milliliters of concentrated hydrochloric acid are added
and the solution is diluted to 1 liter.
Procedure. A 50-ml. water sample or aliquot which, when
diluted to 50 ml., will contain no more than 5 0 mg. of sulfate
(100 p.p.m.) is transferred to a 250-ml. beaker or Erlenmeyer
flask. Ten milliliters of the acid salt solution are added and
mixed. Approximately 0.4 ml. (measured dry with a small
spoon) of barium chloride crystals are added. Rapid 81%irling of
the container is begun immediately and continued for 30 seconds.
The sample is allowed to stand for 5 minutes, after which the re-

A
FLUORIDE

READING

(ERM.)

SULFATE ERROR AND CORRECTION

Zirconium ions in acid solution complex with sulfate ions to

produce the complex (ZrO(SO&-- anion ( 8 ) . iilthough the
reaction conditions have been adjusted to reduce this effect, the
zirconium sulfate complex is strong enough to prevent some of the
zirconium from reacting with Eriochrome Cyanine R to produce
the full color of the reaction. Therefore, the total color reduction
produced in a water sample containing both sulfate and fluoride
ions is the result of the sum of the two complex-forming reactions.
Figure 2 shows how the presence of both fluoride and sulfate affects the fluoride reading. The data obtained in this figure were
used to construct the nomograph in Figure 3, which can be used
to determine the sulfate error in a given sample.
Figure 3 is used as follows: Obtain the fluoride reading of the
sample aliquot, A . Determine the sulfate concentration of the
aliquot, B. Determine the fluoride error of the sulfate concen-

Figure 3.

Sulfate Correction Nomograph

V O L U M E 26, NO. 7, J U L Y 1 9 5 4

1165
standard deviation throughout the range of fluoride concentrations taken was calculated to be 10.0163 mg. per liter of fluoride.
Biological Samples. Urine and bone samples taken in triplicate were distilled, and the fluoride content of the distillates was
determined according to the following procedure: Each distillate collected A Z ~ Sdiluted to 200 ml. One hundred milliliters
were taken for analysis by the thorium titration procedure, and
the remaining 100 ml. were analyzed by the zirconium-Eriochrome Cyanine R method. The results for urine are expressed
as the net fluoride concentration in mg. per liter after subtraction
of the blank for the method used. The bone results are expressed
as per cent of dry bone. The blank obtained in the zirconiumEriochrome Cyanine R procedure includes the calcium oxide
used in ashing the samples and the perchloric acid and silver perchlorate used in the distillation; the blank obtained by the thorium procedure includes, in addition, that contributed by the buffer and the titration blank. A comparison of the two methods is
shoa n in Table IS'.
Esamination of the individual results discloses the greater precision of the zirconium-Eriochrome Cyanine R method, and the
close agreement between the two methods used. These results
should not be interpreted in terms of fluoride recovery, since the
distillation procedure is not a subject of this study.
Water Samples. I n order to test the applicability of the direct method to water samples, the following waters were analyzed in duplicate by distillation and by the zirconium-Eriochrome Cyanine R direct method. The Britton, S. Dak., and
Bartlett, Tex., samples were analyzed as submitted. The synthetic waters A and B were prepared in the laboratory from distilled water and standard solutions of the various ions. The
mineral analysis of these waters is shown below in parts per
million.

Table 111. Zirconium-Eriochrome Cyanine R Reagent

Stability
Reagent
DaysAge'
1 hour
1
3

16

42
67

Fluoride Concentration of Standards, P.P.X.

0.00
0.50
1.00
1.40
Deviation from Standard Curve. P.P.11.
0.013
0,031
0.019
0.008
0.008
0.002
0.013
0.027
0.008
0,010
0 006
0.004
0.008
0.004
0 015
0.000
o 002
0.002
0.013
n 004
0 008
0.023
0,019
0.004
0.013
0.038
0.008
0.036

Table IV.

Biological Samples
Thorium Titration, E.C.R. Method,
P. P. RI.
P.P.R.1.
0 30
0.05
0.04
0.28
0.29
0.045
0.97
0.91
0.80
0.91
1.03
0.95
0.93
0.92

Sample
Blank
AV.
Composite urine
Av.

Composite urine, 0.50 p.p.m. F - a d d e d

Av.
Composite urine, +l.OO
added

1.45
1.39
1.36
1.40

1.13
1.13
1.13
1.13
1.41
1.36
1.38
1.38

1.79
1.81
1.91
1.84

1.83
1.81
1.83
1.82

0.086
0.082
0,091
0.086

0.085
0.086
0.083
0,085

p.p.m. F -

Av.

Britton, S. Dak.
Total solids
2550
Total alkalinity
260
Total hardness
100
so4 - 1100
330

sulting suspension is transferred to a photometer cuvette and the

absorbance read a t 525 mp.
The sulfate concentration of the aliquot is obtained from a
standard curve, which is prepared by subjecting solutions of
known sulfate concentration in the range of 0.0 t o 5.0 mg. (0.0 to
100 p.p.m.) to the above procedure.

&++

AI+++
Si02

POa--FNa

ANALYTICAL DAT4

Source
Britton. S. Dak.
.4v.
Bartlett, Tex.
Av.
Synthetic
added
p.p.m.)
.4v.
Synthetic
added
P.P.m.)

B
=

25

1.01
1.03

100

0.24
0.19

50

0.29
0.30

Table V.
-

...

0.1

6.0L6.8
860

7.618.5
530

Synthetic B

...

32
75
60
56
1.0
0.2
20
3.0
0.45
79

!SO
480-000
320
0.1-1.0

...
...

Synthetic A

...

20

32
75
60
56
1.0
0.2
20
3.0
1.15
79

Table V is a comparison of the fluoride values obtained by the

two methods. The sulfate correction used in the direct procedure
was determined by applying the sulfate concentration of the
sample aliquot, obtained by the turbidimetric procedure, to the
nomogram in Figure 3.

Reagent Stability and Precision. Replicate portions of four

standards containing 0.00, 0.50, 1.00, and 1.40 mg. of fluoride per
liter were taken on different days and treated with the same
batch of zirconium and Eriochrome Cyanine R reagents. The
daily absorbanre readings obtained for each standard were averaged, and the individual deviations from the averages were determined in terms of fluoride concentration without regard to sign.
Table I11 is a summary of these results. Temperatures varied
betneen 25" and 30" C. Using the data from Table 111. the

Distillation
hll. sample F - concn.
distilled
of
(200 ml.
distillate, Aliquot
collected)
p.p.m.
factor
25
0.81
8
0.80

Bartlett, Tex.
1660
390

FACTORS INFLUENCING THE SELECTION OF REACTION

CONDITIONS

The optimum concentrations of zirconium, Eriochrome Cyanine R, and hydrochloric acid, for use as a fluoride reagent, cannot

Water Samples
E.C.R., Direct=
SO1-- concn.

Net F - ,
p.p.m.
6.48
6.40
6.44
8.08
8.24
8.16

hI1. sample
(diluted
to 00 ml.)
5

of aliquot
(turbidimetric),
p.p.m.
109

Apparent

F-,

p.p.m.
0.92
0.94

Sod-- corr.
(Figure 3),
p.p.m.
-0.27

F-, in
aliquot
p.p.m.
0.65
0.67

Aliquot
factor
10

Net

F-,

p.p.m.
6.50
6.70
6.60
8.30
8.40
8.35

48

0.96
0.97

-0.13

0.83
0.84

10

0.48
0.38
0.43

50

59

0.58
0.60

-0.18

0.40
0.42

0.40
0.42
0.41

1.16
1.20
1.18

25

30

0.69
0.67

-0.10

0.59
0.57

1.18
1.14
1.16

(F0.45

(F1.15

Av.
a Corrected for sulfate.

1166

ANALYTICAL CHEMISTRY

be determined on stoichiometric considerations alone. Although

the maximum color of the lake is developed a t a molar ratio of 1 to
3 in 0.1N hydrochloric acid, the reagent is very sensitive to phosphate interference a t this acidity. Also, i t is weakly buffered,
and slow to reach equilibrium. Increasing the acidity reduces
phosphate interference and increases buffer capacity and reaction
rate. However, the reaction is no longer stoichiometric, and
some reduction in fluoride sensitivity occurs, along with a reduction of the total color produced by the reaction (Figure 4).

(IV) produce a color reaction. In a reaction medium containing

0.2N hydrochloric acid, the absorbance change per cm. of light
path per p.p.m. of zirconium is about 0.400 unit when the zirconium concentration lies in the vicinity of 4.0 p.p.m Thus, by
proper control of reaction acidity, a method could be developed,
using Eriochrome Cyanine R, which would determine small
amounts of zirconium in the presence of many cations which usually interfere in other procedures.
SUJIiMARY

i
'.\

A rapid spectrophotometric determination of fluoride ion concentration uses the reaction between zirconyl ions and Eriochrome Cyanine R for color formation and the subsequent decolorization of the lake by fluoride ions. The reaction is immediate stable, and follon-s Beer's law in the concentration range of
0.00 to 1.40 mg. of fluoride per liter. The standard deviation in
this range is +0.0163 mg. of fluoride per liter using standard
photometer cells with a 10-mm. light path Use of this method,
in place of the familar thorium titration on distillates of biological
samples, will save considerable time and increases precision.
When the method is applied directly to water samples, the interference caused by many ions commonly present in water is largely
eliminated. Only sulfate ion concentration must be known and a
correction applied.
ACKNOWLEDGMENT

The assistance provided by Robert Likins, Isadore Zipkin, and

Anastasia Steere of the staff of the National Institute of Dental
Research in the preparation and titration of the biological samples
is acknowledged.
LITERATURE CITED

Figure 4. Effect of Reaction Acidity on Zirconium-Eriochronie Cyanine R Reagent

Some of the color lost can be regained by increasing the Eriochrome Cyanine R concentration of the reagent above the 1 to 3
ratio, However, too great an excess cannot be tolerated, because
the excess dye absorbs strongly a t the wave length of measurement.
The zirconium concentration determines the amount of initial
color which the reaction will produce, and the effective fluoride
range. I t is limited by the ability of the spectrophotometer to
measure deep colors accurately.
The selection of the reaction conditions, as presented above,
was based on these factors and is believed to be near the optimum
with respect to the various considerations previously listed.
When 5.0 ml. of each of the reagents are added to a 50-ml. sample,
it contains 0.375 mg. of zirconium and 9.0 mg. of Eriochrome
Cyanine R (molar ratio 1 to 4) and the solution is 0.7N in hydrochloric acid. Under these conditions 0.050 mg. of fluoride (1.00
p.p.m.) will produce a change in absorbance of 0.480 unit using
a 1.0-cm. light path. The Beckman Model B spectrophotometer can usually be read with precision to 0.005 absorbance
unit. Thus, it is possible to detect changes in fluoride concentration of 0.57 of fluoride in a 50-ml. sample and to measure with
precision 0.050 i 0.016 mg. of fluoride per liter. By scaling
down proportionately the amounts of the reagents added, it is
possible to determine as little as 0.25 y of fluoride in a 5.0-ml.
sample.
OTHER APPLICATIONS

At low acidities (up to 0.1N) Eriochrome Cyanine R produces a

color with many cations such as tin( IV), thorium(IV), titanium(IV), iron(III), aluminum(III), and lead(I1). As the acidity
is increased, only zirconium(1V) and, to a lesser extent, hafnium-

hlegregian, S.,and Maier, F. J., J . Am. Water Works Assoc.,

44,239-48 (1952).
Milton, R. F., Liddell, H. F., and Chirers, J. E., Analyst, 72,
43 (1947).
Revinson, D., and Hartley. J. H., ASAL. CHEM.,25, 794-7
(1953).
Richter, F., Chem. Tech. (Berlin). 1 , 84-90 (1949).
Rowe, F. S.,"Colour Index," Bradford, England, Society of
Dyers and Colourists, 1924.
Saylor, J. H., and Larkin, 11.E., A N ~ LCHEM.,
.
20, 194 (1948).
Sheen, R. T., Kaler, H. L., and Ross, E. AI., IND. ENG.CHEY.,
ANAL.ED.,7,262 (1935).
Sidgwick, K. V., "The Chemical Elements and Their Compounds," 5'01. I , p. 644, London, Oxford University Press, 1950.
Thrun, W. E., ANAL.CHEM.,
22,918-20 (1950).
Willard, H. H., and Horton, C. A,, Zbid., 1190-4.
Willard. H. H., and Winter, 0. B., IXD.ENG.CHEM.,
ANAL.ED.,
5 , 7 (1933).
RECEIVED
for review Xovember 21, 1953. Accepted March 31, 1954.

Spectrochemical Analysis of Aluminum Alloys

Using Molten Metal Electrodes-Correction
I n the article on "Spectrochemical Analysis of Aluminum
Alloys Using Molten Metal Electrodes" [ASAL.CHEM.,26, 795
(1954)], the second line of the third paragraph on page 795
should read: "in the conventional point-to-plane and point-topointexcitation. . "
Page 797, first column, first paragraph, third line from the end
should read: "a distance of 3 mm."
Page 798, second column, the second and third lines of the
table should read:

Inductance
Capacitance

90 ph.
0.007 pfd.

LEOD. FREDERICKSOK,
JR.
J. R. CHURCHILL