N a

Quantitative Analysis by Spectroscopic

Methods
R. A. WOLFE,AND R. W. SMITH,University of Michigan, Ann Arbor, Mich.
‘0. S. DUFFENDACK,
HE value of the spectro- Spectroscopic methods for estimating the given frequency) with which we
scope in identifying sub- amounts of various elements present in a mix- are concerned and not the photo-
stances and in detect- graphed image. The intensity
ing the presence of small amounts
ture, alloy, or solution are reviewed and the of the light in the source may be
$of certain elements in mixtures continued advance in the precision of the de- e x p e c t e d to depend upon the
and compounds has long been terminations is noted. Further increase in amount of the e l e m e n t in the
recognized. The development of mecision has been attained bv the authors,. by“ source that is giving forth that
m e t h o d s of e s t i m a t i n g t h e making use of methods recently developed for light- we have to
q u a n t i t y of a given element work with conveniently is the
present dates from 1874, when
measuring the intensities of spectral lines, and photographed spectrum, and we
L o c k y e r (10) n o t e d that the by a simple means of standardizing the arc must determine from it the rela-
spectral lines of the more abun- source used in the analysis of metallic alloys. tive intensities of the light of
dant elements present in the elec- The necessity for putting an intensity calibration different wave lengths.. .Accu- .

trodes of an arc can be observed pattern on each photographic plate in order to rate quantitative de-
f a r t h e s t from the electrodes. mands the use of the best avail-
insure uniformity and high precision of measure- able of modern spectral
The development of the method
of perfiistent lines or “raies ul- ment is emphasized. The procedure adopted photometry (17)and afullunder-
times” was begun by Hartley (8) by the authors in making analyses of metallic standing of the photographic -
in 1882 and extended and per- aUoys is described. processes (14).
fected by Pollock and Leonaxd I n order to provide a means
(13) and by de Gramont (6, 8). of i n t e r p o l a t i n g b e t w e e n
This method never proved completely successful on account of the fixed percentages in the method of Gerlach and Schweitzer
the impossibility of determining just when a line disappears a t which certain pairs of lines are equally strong, Scheibe
from the spectrum, especially since this depends as much on and Neuhausser (16) introduced the use of a rotating loga-
the characteristics of the photographic plate and apparatus rithmic sectored disk. The proper method for the use of this
used and on the duration of exposure as upon the abundance disk was pointed out by Twyman and Simeon (91) and used
of the test element. On account of this difficulty more recent by Twyman and Hitchen (19, 20) in the methods of quantita-
investigators (4,11, 18) turned to methods which involve tive analysis developed at the Hilger laboratories.
the comparison of the relative strengths of spectral lines, The principal objections to the use of the logarithmic
but these also proved to be of insufficient reliability, because disk are the assumption of the validity of the reciprocity
of the impossibility of standardizing the conditions of ex- law, the introduction of the intermittency effect ( 2 ) , and the
posure and development of the plate over the general range difficulty in determining the exact lengths of the photographed
of analysis. lines. The reciprocity law is known not to be valid, and
Perhaps the greatest errors in the older methods of analysis errors arising from its failure become very large in certain
came from the comparison of lines in different spectrograms. spectral regions. Lines photographed with a sectored disk
It is practically impossible to make equivalent exposures for have no definite length but gradually fade out a t one end.
the recording of spectra. Realizing this, Gerlach (3) de- Like Twyman and Hitchen, Nitchie and Standen (12)
veloped a method of “internal control” in which the analysis make use of a working curve from which the amount of the
was based upon finding pairs of selected lines of equal strength test element present can be read off directly. A micro-
-one of the element under test and one of the principal photometer is employed to measure the densities of the
or basic element in the material being analyzed. These spectral lines, though, apparently, the blackenings were
pairs of lines are examined in the spectrum of the unknown, actually used in the analysis. No means of calibrating the
and analysis is made by comparison with their relative plate is mentioned, but the authors state that only those
strengths in a prepared series of specimens. If no pairs have spectrograms should be used in which the relative densities
equal strength in the spectrum of the unknown, an estimate fall upon the straight-line part of the characteristic curve
of the concentration of the test element is made by an inter- for the plate. With this assumption, the difference in the
polation between the steps in the standard series. blackening of two selected lines is taken as proportional
to the logarithm of the ratio of the intensities of these lines
PHOTOGRAPHIC SPECTRALPHOTOMETRY and the working curve is determined in a manner similar to
In the foregoing discussion, the term “strength” of a that of Twyman and Hitchen.
spectral line has been used, without definition, to mean the The authors have, for a number of years, been using spec-
relative blackness or opaqueness of the line as it appears in trographic methods for estimating small amounts of metals in
the photographed spectrum. The strengths of spectral lines various alloys, particularly nickel alloys, in connection with
have usually been estimated merely by examining the the development of nickel alloys containing barium by the AC
photographed spectrum, often during the time of measure- Spark Plug Company. Since early in 1931 this method has
ment of the plate with a comparator. Purely arbitrary been used continuously for the quantitative determination of
scales were chosen and the value given any line came from the barium in nickel alloys. During this time numerous checks
exercise of the best judgment of the observer. by means of chemical analyses have shown that the spectro-
I n making use of spectral lines in quantitative analysis, scopic determinations are fully as consistent and accurate as
it is the intensity of the light in a given spectral line (of a those made by chemical methods. This long-continued use
226
July 15, 1933 INDUSTRIAL AND ENGINEERING CHEMISTRY
of a method of accurate quantitative analysis for a particular
purpose clearly demonstrates the reliability and value of
spectroscopic methods of quantitative analysis. The method
has been used with equal success in determining silicon in
steel, chromium in nickel, and magnesium, sodium, and po-
tassium in various solutions, and seems capable of very
general application.
FIGURE1. MICROPHOTOMETER RECORD OF A PORTION
OF Two SPECTROGRAMS
AND OF A STEP-DIAPHRAGMCALI-
BRATION PATTERN,
PHOTOGRAPHED ON THE SAMEPLATE
One satisfactory method of determining the relative in-
tensities of spectral lines from their relative densities or
blackenings is that of Hansen (1, 7, 9, 15). By means of a
calibrated step diaphragm, a set of strips or bands is re-
corded on the photographic plate in which the relative
intensity of the light is known. The measurement of the
relative densities or blackenings of these bands gives a
complete calibration curve for that particular plate, from
which curve the relative intensities of any two spectral lines
can be determined from a measurement of the relative den-
sities or blackenings of the lines. Figure 1 is a reproduction
of a microphotometer record of the relative blackenings of
such a set of strips, together with those of a certain pair of
barium and nickel lines taken from the same plate. It
will be noted that the ratio of the blackenings of the barium
and nickel lines changes markedly from one alloy t o the other.
The exposures were so timed that the blackening of the more
intense line, in this case the barium line, did not exceed the
maximum blackening of the straight-line portion of the
characteristic curve of the plate. It is the experience of the
authors that loss in accuracy of analysis occurs if the blacken-
ing of one of the lines exceeds this amount. The blackening
of the weaker line may be less than the minimum blackening
of the straight-line portion without a serious loss of accuracy.
From such a record as that of Figure 1 the curve of Figure 2
may be drawn. The slit width of the spectrograph should be
sufficient so that the photographed lines are broad enough
completely to intercept the beam of light in the micro-
photometer; otherwise the galvanometer deflections will
not be proportional to the blackenings. Another convenient
method of measuring the relative intensities of spectral lines
is described by Thomson and Duffendack (17).
The choice of lines for a given analysis is determined by a
number of considerations. When possible, lines should be
chosen which lie in a region of slowly varying contrast of
the plate used. When an element occurs in fairly consider-
able amounts in a mixture, one must guard against reversal
of the lines and so lines which are not readily absorbed
should be chosen. Some lines of an element are very strongly
affected by the presence of even small amounts of another
element, while other lines of the same element are not.
These effects are most pronounced where close resonance
exists between excited states of two elements, and so, when a
mixture of several elements is being examined, the lines
chosen should be tested for such effects. With a spark
source, the relative intensities of arc and spark lines fre-
227
quently show wide variations. These variations are not so
great in an arc source, and it is sometimes possible to use
the spark line of one element with the arc line of another,
as was done in the case of the barium and nickel alloys used
as an illustration in this report. Another troublesome matter
is plate background. While corrections can be applied for
background, none of them seems reliable and it is better to
arrange the exposures so that the background is of negligible
amount. The plates for Figure 3 and for all other measure-
ments in this investigation had no appreciable background,
the lines standing out on a clear plate.
OF ARC
STANDARDIZATION
It is possible to make accurate analyses by the use of
an arc source as well as a spark source. Upon investigation
it was found that the relative intensities of a selected pair of
spectral lines vary with the arc current in a manner indicated
by the curve in Figure 3, drawn for the lines barium 4554.0
and nickel 4546.9. A similar relationship was found for
the lines chromium 4289.73 and nickel 4401.55. It may be
noted from Figure 3 that after the current reaches a certain
value, the ratio of the intensities of a pair of lines changes
quite slowly with the arc current. Thus, for currents greater
than a certain amount, the ratio of intensities of selected
lines becomes quite insensitive to such fluctuations of the
current as exist in a normal arc. By working with such arcs,
errors due to variations in the arc current become insignificant,
and accurate a n a l y s e s are
possible.
The arc has the advantage
over the spark in that the
spectral lines, especially arc
lines, are much more uniform
over the length of the arc.
Furthermore, the arc con-
sumes more material of the
electrodes than does a spark
and hence g i v e s a b e t t e r
average c o m p o s i t i o n . A
spark will seek out points on
the cathode where materials
of low work f u n c t i o n may
exist in small i n c l u s i o n s ,
whereas a strong arc strikes
to a considerable fraction of
~ A c K ~ M ~
the surface of the ends of the VELOPED pHoToGRAPHrC
electrodes. PLATEAND EXPOSURE OF
PLATE
METHODOF ANALYSIS
The method of quantitative spectrographic analysis de-
veloped by the authors consists of the following steps:
I. FIXINGOF EXCITATION CONDITIONS. An arc between
electrodes consisting of or containing the material to be tested
is used and the relation between the relative intensities of se-
lected spectral lines and the arc current, as indicated in Figure 3,
is determined. A value of arc current is chosen at which, for a
small change in current, the change in relative intensity of the
lines is small. A spark source may be used and standardized
by the method of Gerlach and Schweitzer (4).
11. DETERMINING OF WORKING CURVE. A series of alloys,
solutions, or mixtures is made in which the percentage compo-
sition of the material t o be estimated is varied over the range
chosen for the analysis. The value of the logarithm of the relative
intensities of the selected lines is determined for each alloy in
the series. In every case the spectrum is excited in the manner
indicated by the experiments under I. These results are plotted
as in Figure 4.
111. ANALYSISOF AN UNKNOWNSPECIMEN. The spectrum
of the unknown specimen is photographed in the manner de-
termined under I and the logarithm of the relative intensities
228 ANALYTICAL EDITION Vol. 5 , No. 4
of the selected lines is determined. The logarithm of the relative determine whether the extent of variation is sufficient to
intensities of the spectral lines is applied to the working curve,
Figure 4,and the percentage composition is read on the scale. affect the working curve of the other test element.

TABLE1. COMPARATIVE ANALYSESOF BARIUM-NICKEL ALLOYS

OF METHOD
DISCUSSION BARIUM
ALLOYNUMBER Chemical analysis Spectroscopicanalysis
The method of analysis outlined above consists essentially % %
in applying to this problem a correct method of photographic 716 0.016 0.015
637 0.021 0.028
photometry. The calibration of each plate by putting upon 821 0.022 0.023
it an intensity pattern is an essential feature of the method. 682
667
0.048
0.052
0.051
0.055
No assumptions are made with respect to the reciprocity law 669 0.058 0.057
736 0.060 0.059
or with respect to whether or not the line intensitie’s fall on 662 0,064 0.064
the straight-line part of the characteristic curve. It was only 692
665
0.078
0.095
0.075
0.091
after the introduction of this step that accurate, repeatable, 678 0.130 0.130
420 0.142 0.145
consistent analyses could be made. 179 0.37 0.38
178 0.52 0.49

I n order to indicate the degree of accuracy that may be

attained in regular production practice, Table I gives some
comparative analyses of the same specimens of barium-
nickel alloys by spectroscopic and by chemical means. These
analyses were chosen a t random from a list of approximately
three hundred. The mean deviations in the determination
of the percentage of barium in a given specimen by spectro-
scopic methods were approximately the same as by chemical

I1 \ I
1 8 Curry? 8 !
FIGURE3. RELATION BETWEEN ARC
CURRENTAND RELATIVE INTENSITY
OF A PAIR OF LINES IN SPECTRUM
OF A BARIUM-NICKEL ALLOY
./d.04 %Ea
.1,6 .20 .24 ,
.Q8 .lZ
One working curve suffices for a considerable range of
percentage compositions of a given kind of material. Steps I FIGURE 4. RELATION BETWEEN PER-
CENTAGE OF BARIUMALLOYS AND
and I1 need be performed but once for a particular kind of LOGARITHM OF RELATIVEINTENSITY
alloy, solution, or mixture. Step I11 alone must be repeated OF A PAIROF LINES IN SPECTRAOF
for each analysis. In this step, only one calibration curve THE ALLOYS
(Figure 2) is necessary for a plate which may bear the spectro- I, barium-nickel alloy. 11, barium-nickel
grams of a number, ten or twelve, of specimens to be analyzed. alloy containing 20 per cent copper
I n the cases of barium-nickel and chromium-nickel alloys,
spectrographic analyses have been found to be as accurate methods and amount to *0.001 per cent in the range from
as chemical analyses and can be carried out in a small fraction 0.01 to 0.150 per cent of barium. Some of the larger dis-
of the time required for chemical analyses. crepancies between the spectroscopic and chemical determi-
If a certain material is analyzed for several different nations may be due to slight nonhomogeneities in the alloy.
elements, it is likely that the lines which will be used for the
several elements will lie in more or less separated spectral ACKNOWLEDQMENT
regions. On account of the change in the contrast of the The alloys used for the analyses mentioned in this paper
plate with wave length, separate calibration curves like that were prepared by the AC Spark Plug Company and the
of Figure 1 will be needed. I n such cases, it is desirable t o chemical analyses were made in the company’s laboratories.
put on the plate an intensity pattern which extends over all The authors wish t o express their appreciation of the friendly
wave lengths. This can be done by means of the Hansen cooperation of D. W. Randolph, director of research a t the
method coupled with the spectrograph and more conveniently AC Spark Plug Company, and his assistants, and to thank the
by the method of Thomson and Duffendack. With such a company for its financial support.
pattern, Calibration curves can be determined for any wave
length desired. LITERATURE
CITED
Figure 4 shows two working curves for the estimation of (1) Cittert, P. H. van, 2. P h y s i k , 73, 249 (1931).
barium, one for a nickel alloy and one for a nickel-copper (2) Davis, R.,Bur. Standards, Sci. P a p e r 528 (1926).
alloy. In both these curves, the same pair of barium and (3) Gerlaoh, W., 2.anorg. allgem. Chem., 142,389 (1925).
nickel lines is used, and it is obvious that the addition of (4) Gerlach and Schweitzer, “Chemische Emissions Spektral-
analyse. Grundlagen und Methoden,” Leopold Voss, Leipzig,
copper to the alloy affects the variation of the relative 1930.
intensities of these lines with the percentage of barium. (5) Gramont, A. de, Compt. rend., 144, 1101 (1906); 159, 6 (1914).
This illustrates a situation which is frequently encountered, (6) Gramont, A. de, and Lecoq de Boisbandrau, F., “Analyse
as was noted by Twyman and Hitchen (19, 20). Account spectrale appliquee aux recherches de chimie minbrale,” J.
Hermann, Paris, 1923.
must be taken of this effect and working curves are necessary (7) Hansen, G., 2. P h y s i k , 29, 356 (1924).
for any given element in each type of alloy. If a second (8) Hartley, W. N., Chem. SOC.J., 33, 210 (1882); Trans. Roy.
element also varies in amount in a given alloy, one must SOC.(London), 175, 11, 49, 325 (1884).
July 15, 1933 INDUSTRIAL AND ENGINEERING CHEMISTRY
Hippel, A. von, Ann. Physilc, 80, 672 (1926).
Lockyer, N., Trans. Roy. SOC.(London), 164,II, 479 (1874).
Meggers, Kiess, and Stimson, Bur. Standards, Sei. Paper 444
(1922).
Nitchie and Standen,IND. ENG.CHEM.,Anal. Ed., 4, 182 (1932).
Pollock and Leonard, Proc. Roy. SOC.Dublin, (2) 11, 217, 229,
267 (1908).
Ross, ‘F. E., ”Physics of Developed Photographic Image,”
Van Nostrand, New York, 1924.
Schachtschabel, Ann. Physik, 81,929 (1926).
Soheibe and NeuhLusser, 2.angew. Chem., 41, 1218 (1928).
Determination of Equivalent Acidity and
Basicity of Fertilizers
W. H. PIERRE,West Virginia Agricultural Experiment Station, Morgantown, W. Va.
URING the last few years there has been a general
recognition of the acid-forming character of certain
nitrogenous fertilizers. The amounts of limestone
required to neutralize the acidity developed by these ferti-
lizers are well established. The amount of acidity developed
by mixed fertilizers, however, is not known; nor has it been
possible to determine it satisfactorily.
It is readily evident from a consideration of the common
nitrogen carriers used that mixed fertilizers are usually acid-
forming. I n an attempt to correct this condition some
progressive fertilizer manufacturers have been using ground
limestone instead of sand as the filler. This practice is sound
economically and should be encouraged. Recent work by
MacIntire et al. (5, 6) shows that no loss of available phos-
phoric acid results from the use of the dolomitic limestone.
Such a practice probably would not become general unless
recognition were given to the value of the added limestone.
Parker (7) has pointed out that if the use of limestone’ is to
be encouraged, a method must be developed that can be used
to determine the limestone in mixed fertilizers and thus
make possible the guarantee and chemical control of this
desirable constituent of fertilizers.
With these considerations in mind this investigation was
started with the purpose of developing a simple, quantitative
laboratory method for determining the equivalent acidity
or basicity of any fertilizer carrier or mixture. (Equivalent
acidity may be defined as the acidity developed by the
fertilizer, measured in terms of calcium carbonate required
for its neutralization. Equivalent bssicity refers to the basic
residue left in the soil by fertilizers, expressed as equivalent
calcium carbonate.)
ON SOIL
ACTIONOF FERTILIZERS ACIDITY
Before discussing the method developed for determining
the equivalent acidity or basicity of fertilizers, it seems
desirable to review briefly the effect of fertilizers on soil
acidity and the principles governing this action. A review
of the investigations regarding the effect of fertilizers on soil
reaction leads to the following conclusions:
1. The common potassium fertilizers such as kainit and
muriate and sulfate of potash have no residual effect on soil
reaction.
2. Phosphate fertilizers in which one of the three hydrogens
of phosphoric acid is replaced by a base will in general have no
permanent effect on soil acidity, whereas those in which more
than one hydrogen is replaced by base will be basic in their
1 In this paper limestone includes both calcium limestone and dolomite.
The latter is preferable in fertilizers, for it reacts less with superphosphate
(MacIntire) and also serves as a source of magnesium.
229
(17) Thomson and Duffendack, J. OpticaE SOC.Am., 23, 101 (1933),
and references quoted in this article.
(18) Twvman.
- . F.. J . SOC.Chem. Ind.. 46. 284. 307 11927).
,---I
(19) Twyman, “Practice of Spectrum AGlysis with Hilger Instru-
ments,” Adam Hilger, London, 1931.
(20) Twyman and Hitchen, Proc. Roll. SOC.(London), A133, 72
,- - - - ,.
(lQ31)
(21) Twyman and Simeon, Trans. Optical SOC.(London), 31, 169
(1930).
RECEIV~D
November 28, 1932.
effect on soil reaction. (For those phosphates in which the base
is ammonia see next point.)
3. Fertilizers containin nitrogen in the form of ammonia
or in a form subject to nitrifcation in the soil will produce acidity
unless sufficient base is present to neutralize this acidity.
4. Nitrogenous fertilizers in which the nitrogen is in the
nitrate form and combined with bases such as sodium or calcium
will upon being utilized by plants result in decreased soil acidity.
The fact that potassium fertilizers do not have a residual
effect on soil reaction is generally recognized. With regard
to phosphate fertilizers, however, there has been some con-
troversy. Most of the results obtained in long-time field
*
experiments show that superphosphate slightly decreases
or has no effect on soil acidity. I n recent studies the writer
has found that the action of phosphates on soil reaction
will depend to some extent on the original acidity of the soil
and that for cultivated soils of the humid region, the p H
of which lie mostly between 5 and 6, superphosphate has no
appreciable effect on soil reaction. Any method for deter-
mining the acidic or basic property of fertilizers should
consequently consider monocalcium phosphate as neutral
and the di- and tricalcium-phosphate as basic.
Nitrogenous fertilizers containing ammonia have been
shown to be acid-forming because of the process of nitrifica-
tion which takes place when the fertilizer is added to the
soil (8). A method for determining the acidity resulting
from ammonium fertilizers should consider the acidity that
is developed when complete nitrification of the ammonia
takes place. Although in some cases there is likely to be
some absorption of the ammonia by the plant before being
nitrified, this amount is probably small under average con-
ditions.
The amount of acidity formed from ammonium sulfate
was found (8)to be lower than that calculated on the assump-
tion that all the nitrogen and sulfate would act as nitric
and sulfuric acid, respectively. I n these experiments con-
ditions were very favorable for nitrification and the low
values obtained were believed to be due to the action of the
plant. Apparently plants absorb from calcium or sodium
nitrate more of the nitrate ion than of the calcium or sodium
ion. The result of such an action is to cause sodium and
calcium nitrate to leave a basic residue in the soil and to
reduce the acidity resulting from the nitrification of ammonia
present in acid-forming fertilizers. From a calculation of
the basic action of calcium and sodium nitrate on soils (8)
it was found that 51 per cent of the base had been effective
in reducing soil acidity, indicating that approximately one-
half as much of the base had been absorbed by the plant as
of the nitrate ion. It follows, therefore, that only one-half