T H E FLAME PHOTOMETER IN DETERMINATION OF SODIUM AND
POTASSIUM*
ELSA C. P R O E H L , M.A., AND W I L L I A M P . N E L S O N , M . D .
From the Research Laboratory and the Medical Service, Cushing
Hospital, Framingham,
Massachusetts
Veterans
Administration
In recent years the determination of sodium and potassium in biologic fluids
has been considerably simplified by the development of direct-reading flame
photometers. The use of individually and commercially constructed models has
been described. 1-5 Mosher et al.b have given a detailed account of the use of the
Beckman DU spectrophotometer with the flame photometer. This communication describes our observations with this instrument and a simplified method of
preparing samples and standard solutions.
THE
INSTRUMENT
The principles and characteristics of the instrument have been adequately
described by Mosher et al.6 The sample for analysis is aspirated through a rightangle capillary tube whence it is atomized into an electrically heated glass chamber. The fine spray is carried into a flame of oxygen and propane gas by a steady
current of ] 5 to 25 pounds of compressed air. The light emitted by the combusted
material is admitted into the spectrophotometer where it is resolved into its
spectrum by means of a quartz prism. Adjustment of the prism permits selection of the wave length characteristic of the element under analysis. The band
of light, whose width is determined by a slit opening, then impinges on a photocell and the resultant current is measured. The current is proportional to the
quantity of the element activated by the flame, and is quantitatively expressed
as per cent emission.
The rate of flow of the sample through the capillary must be constant. It was
found that the tips of all capillaries received from the manufacturer were far too
constricted and consistently caused bubbles, with decreased delivery. The tips
were carefully broken off and filed until a rate of flow of approximately 0.4 ml.
per minute was obtained. It is a recommended practice when placing the sample
under the capillary tube to note if the stream contains any bubbles, which can
usually be discharged from the capillary by the quick removal and replacement
of the sample a number of times. If bubbles persist, the tube is disconnected,
attached to a suction apparatus and cleaned by aspirating dichromate solution
through it.
To obviate any contamination of the sample by dirt or moisture in the compressed air line, a porous stone filter is placed between the air valve and the air
port on the machine.
* Received for publication, April 10, 1950.
Sponsored by the Veterans Administration and published with the approval of the
Chief Medical Director. The statements and conclusions published by the authors are the
result of their own stud}' and do not necessarily reflect the opinion or policy of the Veterans
Administration.
806
FLAME PHOTOMETER
S07
PROCEDURE
Instrument settings and operational procedures are described in the literature
accompanying the Beckman instrument and in excellent detail by Mosher et o/..5
Optimal gas, oxygen and air pressures for our machine were determined by
varying the pressure of one while keeping the other two pressures constant. The
graphs obtained by plotting oxygen pressures against emission readings at constant air and gas pressures are similar to those obtained by Mosher.5 For analysis
of sodium the following pressures are used: oxygen, 20 inches (50 cm.) of water;
gas, 1 cm. of water; air, 20 pounds per square inch (about 1400 Cm. per square
cm.). For potassium analysis the pressures used are: oxygen, 20 inches (50 cm.)
of water; gas, 1.5 cm. of water; air, 20 pounds per square inch (about 1400 Cm.
per square cm.). No measurements were made with natural gas.
The sensitivity knob is rotated to its counterclockwise limit; the fixed switch
is set at 0.1; the shutter is closed for dark current adjustment and opened when
emission readings are made. Dark current checks are made with each reading.
The slit width setting is selected according to the element to be determined
and its concentration in the diluted sample. Concentrations of sodium above
0.5 mEq. per liter require a slit width of 0.08 mm. ± 0.02 mm., while at concentrations of 0.5 mEq./L. and below a larger slit width, up to 0.3 mm., is required.
(The setting is made to give readings between 25 and 75 per cent emission whenever possible, although the scale is apparently usable over its entire range). A
slit of 0.3 mm. is used in all potassium analyses.
The wave length setting for an element is made by placing one of the standards
in position. With the shutter open and the slit width setting made, the wavelength dial is rotated in the vicinity of the wave length characteristic of element
(589.3 niyu for sodium; 767 myu for potassium) and left at the position in which
the needle makes its maximum swing to the left. This position may not always
coincide exactly with the wave length for the element as read on the dial. (Adjustment for this discrepancy can be made on the spectrophotometer but must
be rechecked, particularly after occurrence of any jarring or vibration). All
sodium determinations are run on a given set of samples before changing the
wave length setting for potassium analysis.
Preparation of Sam-pies
All glass equipment used is Pyrex.
Serum. Blood is collected under oil, allowed to coagulate and then centrifuged.
The serum is removed with care to avoid transfer of oil. An aliquot is diluted
1 to 100 with distilled water. The flask is stoppered with glass or with Parafilmcovered cork and shaken thoroughly. Both sodium and potassium analyses are
carried out with this dilution. Both determinations may be made using but
0.2 ml. serum.
Urine. Urine is diluted 1 to 100 with distilled water for sodium analyses. This
dilution serves for any concentration of urine sodium encountered. For potassium
analyses a 1 to 500 dilution will usually bring the final concentration into the
desired range of 0.02 to 0.10 mEq./L. For extremes of urine potassium concentration other dilutions are appropriately selected. A dilution of at least 1 to 100
808
PROEHL AND NELSON
should always be employed to avoid possible phosphate interference as described
below.
Preparation of Standards.
Sodium. A stock sodium chloride solution containing 100 mEq./L. is prepared
from reagent grade crystals dried to constant weight at 95 to 100 C. Dilutions
of the stock solution are made to give working standards. The standards used in
analyzing serum, where the range of values is small, contain 1.20, 1.35, 1.50, and
1.65 mEq./L. For urines the standards contain 0.01, 0.05, 0.1, 0.5, 1.0, 1.5, 2.0,
2.5, and 3.0 mEq./L.
Potassium. The stock solution of potassium chloride contains 10 mEq./L. and
is prepared from reagent grade crystals dried to constant weight at 95 to 100 C.
Working standards for urine potassium analyses cover the range of 0.02 to 0.10
mEq./L. at intervals of 0.01 mEq. A second set of potassium standards, covering
a range of 0.02 to 0.10 mEq./L. and containing, in addition, 1.5 mEq. of sodium
chloride per liter, is made up for the analysis of serum.
Relationship of Concentration and Emission
Standard solutions of sodium and potassium covering the range of 0 to 3
mEq./L. were measured at slit widths of 0.08 mm. and 0.3 mm., respectively. A
blank reading obtained with distilled water was subtracted from each reading in
the standard curve. Per cent emission was plotted against concentration. Figure
1 illustrates the curve for sodium standards; a similar graph was obtained with
potassium standards. Figure 2 demonstrates that the curve is rectilinear when
the potassium standards cover a small range (0 to 0.1 mEq./L.). The graphs obtained by plotting emission values for standard sodium solutions over small
ranges against concentration are also rectilinear. Many standards with only little
differences in their concentration are therefore used. Unknown samples are read
between the nearest standards, one higher and one lower, and evaluated by
interpolating between the emission readings of the standard solutions.
Analysis of Sample
When the gas, oxygen, and air pressures have been appropriately regulated
and are flowing at a steady rate, a blank reading using distilled water is made.
This reading is checked frequently and an increase indicates that either the gas
or oxygen pressure is varying. Duplicate dilutions are made routinely from each
unknown sample, and each is read twice between a higher and a lower standard.
During the course of an hour's operation the position of the standard curve may
shift but its slope remains approximately the same. The average difference between duplicates for a series of 75 determinations of serum sodium was 1.82
mEq./L. with standard deviation of these differences of ± 1.41 mEq./L. Therefore, 97.5 per cent of duplicate determinations may be expected to be within
4.6 mEq. per liter of each other. For a series of 65 serum potassium analyses, the
average difference between duplicates was 0.073 mEq./L. with a standard devia-
809
FLAME PHOTOMETER
0
1.0
2.0
3.0
MEO. SODIUM PER LITER
FIG. 1. Per cent emission of standard solutions of sodium chloride
at various concentrations.
o
in
0.04
0.06
MEO. POTASSIUM PER LITER
FIG. 2. Per cent emission of standard solutions of potassium
chloride at various concentrations.
810
PROEHL AND NELSON
tion of these differences of ± 0.145 mEq./L. Thus 97.5 per cent of duplicates
for potassium should be within 0.363 mEq./L. of each other. Potassium determinations are more difficult to carry out than sodium analyses. Fluctuations in
readings make it a more time-consuming operation to obtain readings that check.
Factors Affecting Sodium and Potassium Activation
Effect of sodium and potassium on each other. Determination of sodium was
unaffected by the presence of potassium in a series of solutions in which the
sodium concentration was maintained at 1.5 mEq./L. with potassium concentrations varying from zero to 150 mEq./L.
The emission values of potassium, however, were significantly increased by
the presence of sodium. Pure sodium chloride solutions give no emission at the
wave length employed for potassium determinations. The changes in per cent
emission of a 0.05 mEq./L. solution of potassium chloride when sodium chloride
was added in concentrations from 0.05 to 2.0 mEq./L. are shown in Figure 3.
It is apparent from Figure 3 that the increment in potassium emission changes
only from 10 per cent to 13 per cent as the ratio of sodium to potassium rises from
20:1 to 40:1 or higher (the range encountered in the great majority of serums).
Therefore, 1.5 mEq. sodium chloride per liter was added to each potassium working standard for serum determinations. When serums are being read against such
standards containing 1.5 mEq./L. sodium at all potassium concentrations, the
error introduced because of difference between Na:K ratios in serum and standard
will depend on how much the serum sodium concentration differs from 150
mEq./L. For example, the error thus introduced in analyzing a serum containing
120 mEq./L. of sodium and 6 mEq/L. of potassium would represent only 0.1
mEq./L. of potassium in the undiluted serum. For higher sodium concentrations
and lower potassium levels the error would be even smaller. The sodium potassium
ratios in urines, however, are subject to much greater fluctuation. Therefore, a
sodium analysis of the urine is first made. A dilution for potassium analysis is
then made, and the emission compared with standards containing no sodium.
An approximate value for potassium concentration is thus obtained. By means
of the estimated sodium-potassium ratio, the per cent emission due to sodium
activation is read from the curve in Figure 3 and subtracted from the value obtained for potassium in the original reading. Using this corrected emission reading
the true potassium concentration of the sample is calculated.
Viscosity. Viscosity of the samples at the dilutions employed in these studies
has no observable effect on the emission values of either sodium or potassium.
A series of twenty serums (Table 1) was analyzed for sodium at dilutions of 1-100
and 1-200. The standard deviation between the paired values for the series was
± 1.10 mEq./L., as compared with ± 1.41 mEq./L. for duplicates run at the
same dilution. Thus, doubling the dilution of protein gave no appreciable change
in the sodium values. Recoveries of added sodium and potassium (Table 2) also
indicate that the rates of vaporization of the sample and standard are equal. In
a series of 29 serums, to which sodium was added, the determination had an
average error of 1.96 per cent. In a series of 21 serums, to which potassium was
added, the determination had an average error of 2.16 per cent.
811
FLAME PHOTOMETER
10:1
0
20M
30= |
SOOIUM CONCENTRATION
POTASSIUM CONCENTRATION
Fici 3. Per cent emission of a 0.05 mEq. per liter solution of potassium chloride when sodium chloride is added in concentrations from
0.05 to 2.00 m E q . per liter.
TABLE 1
SERUM
SODIUM V A L U E S O B T A I N E D
AT D I L U T I O N S
O F 1-100
AND
1-200
DILUTIONS
SAMPLE NO.
1-100
1-200
m.Eq./L
m.Eq./L.
139
13S
139
137
139
136
13S
142
136
138
146
153
139
136
139
137
139
143
13S
130
140
140
140
135
13S
136
134
13S
140
135
144
152
137
140
136
136
137
140
135
130
1
2
3
4
5
6
7
S
9
10
11
12
13
14
15
16
17
18
19
20
Standard deviation == ± 1.10 mEq./L.
812
P R O E H L AND
NELSON
Ammonium phosphate, glucose, and urea. These substances were found to have
no effect on sodium and potassium emission. They were studied in concentrations
well above the maximum found in samples of serum and urine diluted as indicated
TABLE 2
RECOVERY OF ADDED SODIUM AND POTASSIUM
POTASSIUM
Sample
1
2
3
4
5
6
7
S
9
10
11
12
13
14
15
16
17
IS
19
20
• 21
22
23
24
25
26
27
2S
29
Error
Amount
Present*
Amount
Found
mEq./L.
%
mEq./L.
mEq./L.
154.5
0.9S
2.42
04
72
9S
57
66
92
7S
90
66
76
05
93
40
SI
S9
15
22
61
26
27
45
41
OS
49
00
Amount
Present*
Amount
Found
mEq./L.
153
165
167
164
167
162
15S
162
15S
156
167
16S
171
16S
166
169
171
160
160
160
163
164
166
167
166
167
170
170
171
169
167
167
166
160
152.5
161
161
164
164
171
173
16S
164
161
165
163
162
165
162
163
164
170
174
165
173
167
163
Average per cent error = 1.96
0.
1.S2
0.59
1.23
3.4S
0.61
1.S9
5.12
1.79
1.7S
1.16
0.
1.20
4.73
3.50
1.S7
1.25
3.12
0.61
0.61
1.20
1.79
4.SI
1.19
1.76
1.76
4.67
90
SO
35
9S
0
35
44
30
26
IS
44
35
10
53
00
IS
6.18
Error
2.3S
1.27
60
96
14
06
24
0.93
3.41
20
73
04
S4
73
0.1S
0.94
0.39
0.SS
0.
0.
Average per cent error = 2.16
* Sodium present equals original sodium concentration as determined by duplicate
photometric analysis plus 25 m E q . / L . sodium added. Potassium present equals original
potassium concentration determined in duplicate plus 1.0 m E q . / L . potassium added.
previously. To solutions of sodium and potassium chloride, ammonium phosphate
was added in concentrations up to 66 mg. (0.5 mM) per liter; glucose up to 100
mg. (0.55 mM) per liter; and urea up to 700 mg. (11.7 mM) per liter with no
significant change in emission. Thus, the dilutions of serum and urine used in
FLAME
PHOTOMETER
813
these studies are adequate to eliminate any effects these substances might have
at higher concentrations.
COMMENT
Although we feel that the data presented indicate that clinically useful sodium
and potassium determinations can be accomplished by the described technics,
it should be emphasized that difficulty was frequently encountered in obtaining
consistent readings particularly for potassium. These annoying fluctuations were
somtimes obviously due to alteration in gas and oxygen pressures which were
not easily rectified. At other times they could not be explained. During the course
of analysis it is essential constantly to be alert to changes in the response of the
instrument.
SUMMARY
A simplified method of preparing samples and standard solutions for analysis
of sodium and potassium in biologic fluids with the Beckman DU Spectrophotometer and flame photometer attachment is reported. Technics of operation are
described. The problems and methods of overcoming the effects of sodium on
potassium activation are described. The effects of viscosity and interference clue
to phosphate, urea and glucose were eliminated by appropriate dilution. Clinically useful results are obtainable by this method.
Acknowledyments.
T h e authors wish to express appreciation to D r . Allan M . Butler in
whose laboratory part of this work was carried on, and to D r . Jack D . Rosenbaum for his
many helpful suggestions in technic and in the preparation of t h e manuscript.
REFERENCES
1. B A R N E S , R. B . , RICHARDSON, D . , B E R R Y , J . W., AND H O O D , R. L . : F l a m e p h o t o m e t e r ;
A rapid analytic procedure. Incl. E n g . Chem., Anal. Ed., 17: 605-611, 1945.
2. BOWMAN, R. L., AND B E R L I N E R , R. W.: Principles of design and operation of internal
standard flame photometers for sodium and potassium determination. Federation
P r o c , 8: 14-15, 1949.
3. H A L O , P . M . : T h e flame photometer for t h e measurement of sodium and potassium in
biological materials. J. Biol. Chem., 167: 499-510, 1947.
4. M A K I N I S , T . P . , M U I R H E A D , E . E . , J O N E S , F . , AND H I L L , J . M . : Sodium and p o t a s s i u m
determinations in health and disease. J. L a b . and Clin. Med., 32: 120S-1216, 1947.
5. M O S I I E R , R . E . , B O Y L E , A. J , B I R D , E . J., JACOBSON, S. D . , BATCHELOR, T . M., J S E R I ,
L. T., AND M Y E R S , G. B . : T h e use of flame photometry for the q u a n t i t a t i v e determination of sodium and potassium in plasma and urine. Am. J. Clin. P a t h . , 19: 461-470,
1949.