Interference-free Flame Photometry of Calcium

in Serum and Urine

Ralph E. Thiers and Kirsten Hviid

Data are given for the rapid determination of calcium in biologic material by flame
photometry. An “automatic background-subtraction,” photometer added to the
available instrumentation eliminated cation interference, while the use of a chelating
agent (EDTA) prevented interference from anions (chiefly phosphate). The pro-
cedure shows excellent precision and accuracy in comparison with standard chemical
procedures and in the analysis of control sera.

FLAIE PHOTOMETRY is the most sensitive and convenient method of de-

termining sodium and potassium in serum and urine, but it has not
been broadly applied to the routine determination of calcium. Instead,
many clinical laboratories employ precipitation and titration of cal-
cium oxalate (1), while others utilize titration of calcium with chelat-
ing agents (2-4). Flame photometry of calcium, an inherently simpler
method, tends to be inaccurate when applied to serum and even more
so with urine because of interference from both cations and anions (5-
7). Sodium, potassium, and phosphate produce the chief problems.
The first two emit a continuous spectrum of radiation which includes
the wave length of the calcium emission (5); the last reduces the con-
centration of the emitting species (6).
This report describes an investigation of the applicability of a lab-
oratory-built “automatic background-subtraction” photometer (8) to
cancel the interference of cations in the flame photometric determina-
tion of calcium in serum or urine samples, and of a chelating agent,
ammonium (ethylene dinitrilo)tetraacetate (EDTA), to prevent the
interference of anions. Conditions have been found under which both

From the Clinical Chemistry Laboratories of the Department of Biochemistry, Duke 15111-
versity Medical Center, Durham, N. C., and the Peter Bent Brigham Hospital, Boston, Mass.
The work reported here was supported in part by an Institutional Grant to Duke University
Medical Center by the American Cancer Society.
Received for publication Jan. 3, 1961.

35
36 THIERS & HVIID Clinical Chemistry

of these technics can be combined into a single, extremely simple pro-

cedure for the rapid determination of calcium in serum or urine. This
procedure Ilas been evaluated and its routine application is described.

Method
Reagents and Apparatus

EDTA Stock Solution A 0.2M solution of the ammonium salt of

EDTA at vH 8 is prepared by placing 58.4 gm. of EDTA (Fisher Sci-
entific) in a 400-rnl. beaker and adding 100 ml. of distilled 1120. Suffi-
cient concentrated NHOH is then added to bring the powder into solu-
tion and raise the pH to 8. The solution is filtered into a 1-L. volumetric
flask and diluted to volume. Working solutions are prepared by dilu-
tioll.
Stan(iard Stock Solutions Standard stock solutions, 1.000 N in
sodium chloride, 1.000 N in potassium chloride, 0.1000 N in calcium
chloride, and 0.1000 N in magnesium chloride, were prepared from
carefully dried NaCl, KC1, CaC0, and Mg, the last two by addition
of a minimum of HC1. A standard 1.0 M phosphate solution was pre-
pared from reagent NaH2PO4 . 110.
Simulated Serum and Urine Solutions An aqueous solution was
prepared from the standards, to contain the cations of serum and also
the phosphate. Another solution was made to simulate the inorganic
constituents of normal urine. The concentrations in these solutions, in
milliequivalents per liter, were as follows. Simulated serum: Na+,
140; K, 4.5; Ca, 5.0; Mg, 2.4, 119P04, 2.0. Simulated urine:
Nat, 290; K,44; 6.5; 112P04, 50.
Background Correction Solutions Background correction solu-
tions, oie each for serum and urine, were made up from the stock
standards to contain all of the above cations except calcium, in the con-
centrations shown. These solutions are utilized in checking the proper
setting for the controls which provide automatic background correc-
tion, as described below.
Calibrated Standard Solutions Working standard solutions for
the construction of flame photometric curves were made by dilution of
the calcium stock standard to contain 1.0, 3.0, 5.0, 7.0, 9.0, and 11.0
mEq./L. of calcium chloride. Other cations or anions were not included
in these standards.
Control Standards Standardized pooled human sera, treated to
contain accurately known concentrations of calcium (Versatol, War-
ner-Chilcott Co.) were purchased as lyophilized powders and recon-
Vol. 9, No. I, 962 FLAME PHOTOMETRY OF CALCIUM 37

stituted with water to provide known primary standards containing

the organic constituents of serum as well as the salts.
Solutions, standards, and samples were diluted 50-fold by means of
a Seligson pipet (9) before aspiration into the flame.
Flame Photometer (8) A hydrogen-oxygen burner (Beckmaii
No. 4020) and flame assembly (Beckman Model B attachment) was
used throughout this work. The photometer employed was fabricated
around a concave Wallace replica grating, 15,000 lines per inch, 21.25-
cm. focal length (Central Scientific Co. No. 86886, Grade A), on a sym-
metrical, double Paschen-R.unge mounting. The optical diagram is
shown in Fig. 1. One exit slit was set exactly on the first-order calcium
line at 4226.7 A. A pair of exit slits, very close together and on the op-
posite side of the optic axis, were set to pass light from either side of,
and closely adjacent to, the other first-order calcium line at 4226.7 A,
but to block the line itself. Two multiplier phototubes were set to re-
ceive the “line plus background” (L+B) and the “background” (B)
beams of light, respectively. The signals from these tubes were passed
to a dual amplifier with independent bias controls on each channel. In-
dependent controls of multiplier phototube dynode voltages were also
provided. A single switch allowed the output microammeter to display
the phototube current from either multiplier phototube, or the differ-
ence between their currents.

ROWLAND CIRCLC

Fig. 1. Optical diagram of automatic background-correcting flame photometer. P repre-

sents photo multiplier tube measuring line-plus-background intensity; F8, photo multiplier
tube measuring background intensity only; Si, entrance slit; 8,, exit slit at Ca 4226.7 A; an
8,, double exit slits adjacent to Ca 4226.7A.
38 THIERS & HVIID Clinical Chemistry

Procedure

Samples of serum or urine are diluted 50-fold with 0.15M EDTA and
aspirated into the flame of the background-correcting photometer. The
reading of the microammeter on the photometer is directly propor-
tional to the concentration of calcium in the sample, provided that the
instrument has been adjusted by the use of water and background cor-
rection solutions, as described in the following. With the instrument
employed in this work, the slope of the calibration curve is about 50
tAmp./mEq./L. of Ca + in the original undiluted sample.
All of the reagents from the list above are not required for this pro-
cedure. Only EDTA, background correction solution, and calibrated
calcium standards are used routinely. The other reagents were em-
ployed in the experimental verification of t.he method, as described be-
low. Stability of the instrument is such that the adjustment of back-
groulld correction and the calibration curve need to be checked only
once each time the flame is turned on.

Automatic Background Correction

Theory

It llas been shown that the presence of sodium or potassium in a

flame contributes a variable amount of continuous background radia-
tion which adds to the measured intensity of all spectral lines (5, 8).
It has long been the practice in spectrography, and more recently in
flame spectrometry, to measure the background adjacent to the line, on
one or both sides, and to subtract its intensity from that of the meas-
ured line-plus-background value. The arrangement of slits in back-
ground-correcting spectrometers allows the measurement of the inten-
sity of line plus background (IL+B) on one photomultiplier tube and of
the background alone (‘B) on the other. If the background is truly con-
tinuous, the signals from these two tubes may be amplified and com-
bined to represent directly the net intensity of the chosen spectral line
(8).
Accurate automatic background correction requires certain crucial
relationships between the amplified responses of the two different
photomultiplier tubes. These relationships, which have not previously
been made explicit, are illustrated in Fig. 2. For each phototube the
response curve approximates a linear relationship between light in-
tensity, I, and the electrical signal produced, A, such that

Aa+bI (1)
Vol. 8, No. I, 1962 FLAME PHOTOMETRY OF CALCIUM 39

where a and b are constants. Independent controls must be provided

on each pliototube amplifier which can alter the slope, b, of the re-
sponse curve, and which can add or subtract a constant bias current, a,
to all points on the curve. With these four controls (two for each photo-
tube) the two response lines

lL+il_af+B+bL-1-BIL+n (2)
and
An=an+bnln (3)

can be superimposed in any chosen position on the A-I axes. Under

these conditions the slope of the pair of lines governs the sensitivity of
the instrument to the spectral line chosen, and if the two lines are su-
perimposed, the instrument is completely insensitive to a continuous
spectrum, because light affects both tubes equally. In practice these
controls are used to compensate for differential reflectivity of the grat-
ing, different exit slit widths, and asymmetrical instrument geometry
-in effect, practically everything except monochromatic or “line” ra-
(hation and curvature of the response lines of the tubes themselves. In
practice, the display of the difference between the two phototube cur-

;b b*

AMPLIFIER CONTROL PANEL

-J

z
(()
Fig. 2. Adjustment of photo
multiplier tube response curves
for automatic background cor-
rection.
I
0

Ui
U.
-J
0

12

LIGHT INTENSITY, I
40 THIERS & HVIID Clinical Chemistr,

rents on the output meter of the instrument allows very precise ad just-
ment of their superposition, which simultaneously compensates for
all linear variables in the method.

Practice

l)istilled water is aspirated into the flame. The continuous radiation

emitted at and near 4226.7 A by the flame is then low in intensity (for
example, I in Fig. 2). The aB and aL+B controls of the photomultiphier
tubes are then adjusted until the meter shows a difference of zero be-
tween the phototube currents. Under these conditions, the two re-
sponse curves have been moved from whatever random relative posi-
lions they originally occupied until they cross at I. A background
correction solution is then aspirated into the flame. This solution con-
tains all of the light-emitting species expected in the sample except the
one to be measured (in this case calcium), and in concentrations higher
than that expected in any sample. Under these conditions the continu-
ous radiation emitted by the flame is high in intensity (for example, 12
in Fig. 1). The bL+B and bli controls are then manipulated until the
meter reads zero. This adjustment alters the slopes of the response
lines by varying dynode voltage. If this alteration has been by rota-
tion about the common point of the two lines at 1, then the lines will be
superimposed and

aL+B_aB (4)

and
(5)

it is usually necessary to repeat alternately aspiration of water and

background correction solution, with successive adjustments, until
this condition is reached. Matching the curves by means of such con-
trols provides automatic background correction over wide sample
variations and, once set, the controls may require no change or only
minor adjustments over month-long periods.

Application to Serum and Urine

The accuracy of such a correction system, and its mode of operation,
can easily be shown by comparing standard solutions of metal chlo-
rides, as shown in Table 1. Separate, simultaneous readings were
taken on the responses of the two phototubes and on their electronical-
ly subtracted differences. The large signal from Na and K observed
with Samples 2 and 4 is the same on both tubes and does not appear in
Vol. 8, No. I, 1962 FLAME PHOTOMETRY OF CALCIUM 4

Table 1. COMPARISON OF STANDARD 50LUTI0NS OF METAL C111,ORIDES

Signa1 (VAmp.)
Coneentrations of ealiono*
(mEq./L.) I II Iti
(L+B) (5) (L+B-
Sampte Ca Atg Na K BL)

1. Water 0 0 0 0 0 0 0
2. Background correction solution 0 0.050 3.0 0.10 290 290 0
3. Calcium alone 0.100 0 0 0 265 100 165
4. Metals of serum 0.100 0.050 3.0 0.10 470 305 165

8111 samples aspirated into flame.

the subtracted readings in Column III. Only the calcium signal of

Samples 3 and 4 is recorded there.
When analogous data are obtained for diluted samples of serum or
urine, the net reading observed does not coincide with that for an
aqueous solution containing the same concentration of calcium. This
lack of agreement results from the presence of interfering anions
(principally phosphate) in the biological samples. For this reason the
accuracy of the automatic background subtraction technic when ap-
plied to serum and urine is not subject to a direct check, as is possible
with the solutions shown in Table 1. However, the combination of au-
tomatic background correction with a method of eliminating anion in-
terference, in this study, permitted the accuracy of both technics to-
gether to be checked, as described in the following.

Prevention of Anion Interference

Wirtscllafter (10) demonstrated that the addition of 17 mM/L. of
EDTA and 11 mM/L. of KOII to solutions containing 2.4 mEq./L. of
Ca and 5.2 mM/L. of phosphate, partially eliminated the depression of
the calcium signal by this anion. West and Cook (11) showed that a
100-mM solution of the sodium salt of EDTA at pH 9 prevented the in-
terfering effect of phosphate, up to 10 mM/L, upon calcium at 0.6
mM/L. These findings prompted an investigation of the effect of the
ammonium salt of EDTA upon the calcium signal from serum and
urine samples, diluted appropriately for the background-correcting
flame photometer. In such dilutions the calcium concentration is 0.1-
0.3 mEq./L. and the phosphate 0.05-0.15 mM/L. in serum and 1-3
mM/L. in urine.*

8During preparation of this report, work by Herrman and Rick on the application of the
ammonium salt of EDTA to flame photometry of calcium, without background correction
came to our attention.
42 THIERS & HVIID Clinical Chemistry

Samples of urine and serum were diluted 50-fold in various coilcen-

trations of EDTA. The calcium signal was read after setting controls
of the instrument with water solution. The results are shown in Fig.
3. The intensity of the calcium line is depressed in all cases wild Wa-

300

Fig. 3. Effect of EDTA at

- 200 - C pH 8 on the intensity of Ca

4226.7 A with samples of serum

and urine. Curves A and C are

D for sera from different individ-
uals; D and E for urilles from
z
- different individuals. Curve B
j 00 - is a simulated serum containing
0.98 times as much calcium and
1.7 times as much phosphate as
(1) ...-.------.E the serum of Curve A.

0 0.2

CONCENTRATION OF EDTA, IN MOLES/LITER

ter (or isotonic NH4C1) is used as the diluent, and rises with increasing
concentrations of EDTA until it reaches a plateau between 0.15 M and
0.2 M. This phenomenon was observed with all samples of serum or
urine tested.
The accuracy of the prevention of anion interference and of auto-
matic background subtraction was tested directly by comparing the
intensity of the calcium line from two simulated urine solutions of
equal calcium concentration, one of which contained EDTA and phos-
phate, the other of which did not. As shown in Table 2, the presence of
phosphate depressed the calcium signal by a large factor, but the addi-
tion of EDTA prevented any decrease whatsoever. Neither the EDTA,
nor the ]IDTA plus phosphate, showed any signal. Thus, the accuracy
and effectiveness of simultaneous correction technics for both cation
and anion interferences on the flame photometry of calcium in biologi-
cal samples, is clearly demonstrated.

Precision
A sample of pooled serum was divided into 18 aliquots, and each
aliquot was diluted separately for analysis by the procedure described.
Vol. 8, No. I, 1962 FLAME PHOTOMETRY OF CALCIUM 43

Table 2. COMPARISON OF CALCIUM LINE OF SIMULATED URINE SOLUTION WITH AND WITHOUT

EDTA AND PHOSPHATE

Concenfrations of components of simulated urtneg*

(mEq./L.) Duplicate
Sample readin got
No. EDTA H,P04- Na (pAmp.)

1 150 0 0 0 0 0 0
2 150 1.0 0 0 0 0 0
3 0 1.0 0.130 5.8 0.87 95 95
4 0 0 0.130 5.8 0.87 255 260
5 150 1.0 0.130 5.8 0.87 260 255

1n samples aspirated into flame.

Automatica1ly corrected for background.

A sample of urine was treated in the same manner. The average read-
ing for the serum samples was 210 VAmp. (corresponding to 4.2
mEq./L.) and, for the urine samples, was 95 jiAmp. (1.9 mEq./L.).
The over-all range of variation of the readings obtained was ±5 pAmp.
in both cases. Since the meter could be read only to the nearest 5
VAmp., the precision can best be expressed as a range of ±5 1LAmp. or
±0.1 mEq./L. (13). Calculation of average or standard deviation
would be misleading in this case, where the data are not distributed ac-
cording to the Gaussian or “normal” relationship, and, in fact, would
give a result far less than the smallest readable meter division. The
precision of all measurements subsequently undertaken proved to be
the same, ±0.1 mEq./L., and was limited by the readability of the
meter.

Accuracy

For serum, the availability of known independent standards (Ver-

satol) made possible a direct check of accuracy. The results are given
in Table 3. The accuracy attained is excellent for clinical chemical
purposes. Since the standards employed in this determination were
actually solutions of CaC12 in 0.15 M EDTA, these data demonstrate

Table 3. ANALYSIS OF COMMERCIAL LYOPHYLIZED STANDARDS8

Calcium concentration Concentration found

Lot No. (mEq./L.) (mEq./L.)

0003010 5.0 4.9, 5.0

130030 3.5 3.5, 3.6
28679 3.7 3.6, 3.7
1:1 mixture of 0003010 and 28679 4.3 4.2, 4.2

Versatol (General Diagnostics, Warner-Chilcott).

44 THIERS & HVIID Clinical Chemistry

more than the accuracy of the over-all method. They confirm for real
serum what the data of Table 2 checked for simulated urine, that is, the
accuracy of the background subtraction technic applied to a complex
system.
In addition, blind analyses were performed on 10 randomly selected
sera from the routine load of the clinical chemistry laboratory over a
period of 2 weeks. These sera had been analyzed by a standard method
of calcium determination (1) which involves a precipitation of calcium
oxalate and a titration. The calcium in these samples ranged in con-
centration from 2.9 to 7.0 mEq./L. The average difference between
the two methods was 0.01 mEq./L., and no difference greater than
±0.1 mEq./L. was observed.
A known sample of urine, comparable to Versatol for serum, is not
available, and the standard methods for determining calcium are far
less reliable when applied to urine than to serum. While this situation
is tolerable in clinical chemistry, because clinically meaningful altera-
tions in calcium concentration are relatively much larger in urine than
in serum, it precludes a direct check on the accuracy of the flame meth-
od with real urine. The results of recovery experiments that were per-
formed (Table 4) are self-consistent and show the same precision as
was observed with serum samples. Accuracy is demonstrated with
simulated urine (Table 2). It is, therefore reasonable to presume that
the same high accuracy prevails with urine as with serum.

Discussion
The precision of this method and its accuracy can both be expressed
as a range of ±0.1 mEq./L. (13). For normal serum containing 5.0
mEq./L. this is a range of ±2%, which makes the method excellent for
clinical use.
Table 4. RECOVERY OF CALCIUM ADDED TO URINE SAMPLES

Meatured concentration Added Calcium concentration

Sample before audition calcium found*
No. (mEq./L.) (mEq./L.) (mEq./L.)

349 1.2 1.9 2.9 3.0

436 0.2 2.4 2.5 2.6
437 0.5 2.2 2.7 2.7-
438 1.1 2.0 3.0 3.0
440 1.4 1.8 3.2 3.3
441 0.4 2.3 2.7 2.5
442 1.0 2.0 3.0 3.1

Duplicate readings.
Vol. 8, No. I, 1962 FLAME PHOTOMETRY OF CALCIUM 45

The procedure is extremely rapid, and simple. Dilution of stand-

ards or samples with a Seligson pipet requires 1#{189}mm. of operator
time; aspiration of sample and reading of the meter require less than
1/2 mm. A scale can be fastened to the meter to read directly in mEq./L.
This method has completely replaced other methods of determining
calcium in the laboratory of the Peter Bent Brigham hospital and has
been in trouble-free operation about a year.
With all samples of serum tested, a short “leveling-off” of the
curves of Fig. 3 was observed around 0.02M EDTA. At this concen-
tration a perfectly satisfactory correction for the anion interferences
of serum was obtained, because the signal, although not maximal, is
the same in tile presence or absence of the phosphate of serum. Thus,
if calciuni determinations in serum only are needed, the concentration
of EDTA in samples and standards may be decreased to 0.02 M with-
out loss of accuracy.
Flame photometric methods for calcium in serum and urine have
beeii described which utilize neither of the corrections which this study
showed are necessary (5). Such methods must either suffer from in-
ferior accuracy or involve slow and cumbersome separations which
sacrifice the main advantages of the flame method.
No effect of dissolved proteins on the calcium signal could be ob-
served, nor did further addition of phosphate or magnesium alter the
data, even in quantities sufficient to interfere severely with other meth-
ods. Other workers have observed effects of protein on the calcium
signal (5) which are not observed in this method, but the concentra-
tions of protein and calcium aspirated have generally been much high-
er. In addition, the slow precipitation of protein causes turbidity when
water is used to dilute serum. The protein remains in solution when
diluted with EDTA in concentrations greater than 0.015M. The age of
the serum sample has not been observed to affect this method.
It should be noted that the actual concentration of calcium in the
aspirated dilutions of serum or urine is very low-for normal sera
about 0.1 mEq./L. This presumably accounts for the exactly linear
calibration curve obtained, since self-absorption and ionization effects
are constant over this range. Cleaning and handling of apparatus
should be performed with the care normally applied to analytical meth-
ods at low concentrations (14).

References
1. Reiner, M., Ed., Standard Method.o of Clinical Chemistry Vol. 1, Acad. Press, New York
(]953, p. 16.
46 THIERS & HVIID Clinical Chemistry

2. Coolidge, T. B., Anal. Biocheir&. 1, 93 (1960).

3. McAllister, H. C., Jr., and Yarbro, C. L., Clin. Chem. 6, 52 (1960).
4. Seligson, D., Ed., Standard Methods of Clinical Chemistry Vol. II, Acad. Press, New
York, (1958), p. 1.
5. Dean, J. A., Flame Photometry, McGraw-Hill, New York (1960).
6. Fukushima, S., Mikrochimica Acta 4, 596 (1959).
7. Margoshes, M., and Vallee, B. L., In Methods of Biochemical Analysis, P. Glick, Ed., Vol.
3, Interscience, New York, (1959) p. 353.
8. Margoshes, M., and Vallee, B. L., Anal. Chem. 28, 1066 (1956).
9. Seligson, D., Aim. J. Clin. Pat hol. 28, 200 (1957).
10. Wirtschafter, J. D., Science 125, 603 (1957).
11. West, A. C., and Cooke, W. P., Anal. Chem. 32, 1471 (1960).
12. Herrmnn, R., and Rick, W., Naturwissenschaften 46, 492 (1959).
13. Davies, 0. L., Statistical Methods in Besearch and Production) Hafner, New York, (1954),
p. 31.
14. Thiers, H. E., In Methods of Biochemical Analysis, P. Glick, Ed., Vol. 5, Interscience, New
York, (1957), p. 273.