A METHOD OF ANALYSIS BY
MEANS OF THE SPECTROGRAPH

By

MAURICE JEROME DAY

A THESIS
Submitted to the Faculty of
M ichigan State College of Agriculture an d A pplied Science
in Partial Fulfillment of the Requirements for
the Degree of
Doctor of Philosophy

KEDZIE CHEMICAL LABORATORY
East Lansing, Michigan
1937

ProQ uest Number: 10008487

All rights reserved
INFORMATION TO ALL USERS
T he quality of this reproduction is dependent upon the quality of the copy submitted.
In the unlikely even t that the author did not sen d a com plete manuscript
and there are m issing p a g es, th e se will be noted. Also, if material had to be removed,
a note will indicate the deletion.

uest.
ProQuest 10008487
Published by ProQ uest LLC (2016). Copyright of the Dissertation is held by the Author.
All rights reserved.
This work is protected against unauthorized copying under Title 17, United S ta tes C ode
Microform Edition © ProQ uest LLC.

ProQ uest LLC.
789 East Eisenhow er Parkway
P.O. Box 1346
Ann Arbor, Ml 4 8 1 0 6 - 1 3 4 6

A

METHOD OF ANALYSIS BY MEANS OF THE SPECTROGRAPH

A spectrographic m ethod is described herein for the quantitative determ ination of m etals present
a s minor constituents or impurities in alloys, salts, solutions or other m aterials. Im proved m ethods of
m easuring photographic blackenings a re em ployed w hich result in a more accurate evaluation of
relative spectral intensities. This is a n essential a n d fundam ental p art of a n y spectrographic analytical
procedure.
In this investigation the m ethod w as u sed for the determ ination of several m etals present a t very
low percentages in copper alloys. A new procedure of photographic plate control is dem onstrated
incorporating internal control without the use of com puted ratios of spectral intensities or photographic
blackenings. The applicability of the method for the elimination of some undesirable effects of the
variable characteristics of the spectrogram such a s those arising from the processing technique or
photographic m aterials is em phasized.
The dem and for more r a p id . methods of analysis of substances for impurities or minor constituents
h a s b een increasing for some time a n d for certain classes of m aterials this n eed is being m et with the
spectrograph.
Lockyer, (Lockyer, N., Trans. Roy. Soc. (London) 164, II, 479 (1874), a n early investigator in this
work, noted that the distance a spectral frequency could b e observed from the source varied with the
am ount of the elem ent emitting that frequency. Since then a num ber of procedures h av e been in­
troduced for the spectrographic quantitative estimation of elements. These procedures a re b a se d on
the principle that a s the concentration of a n elem ent increases the intensity of the light emitted by .the
spectrum of that elem ent increases. The results of the analysis are obtained b y the com parison of
these spectral frequencies with the corresponding frequencies in sam ples of known concentrations. In
general, all the methods involved the recording of ihe spectrum on a photographic plate, giving a
series of lines which are im ages of the slit. Each line corresponds to a n individual spectral frequency
in the light which entered the slit.
The various methods described in the literature for the spectrographic determ ination of a n elem ent
m a y be differentiated b y variations in the form of the sam ple, m eans of excitation, type of ap p aratu s
used, a n d the procedure for the utilization of the spectrum. G erlach (Gerlach, W« Z. Anorg. Allgem.
Chem. 142, 389 (1925) found that a large error w as due to the com parison of the lines of one spectrum to
the corresponding lines in another spectrum. He reduced this error in his m ethod of "internal control"
b y correlating the density of certain selected lines of the m ajor elem ent of the substance being analyzed
with the blackening of the line or lines of the element being determined. A set of standard specim ens
w as u s e d to p rep are a chart showing the concentration steps at which pairs of lines of the sam e black­
ening were found. The analysis w as carried out by the inspection of a spectrum of the unknow n for
such a pair of lines. If a p air of lines of equal blackening w as found, the chart w as consulted for the
concentration. G erlach used this method in order to reduce the errors encountered in attem pting to re­
produce spectra.
This visual m atching of pairs or the interpolation betw een steps in a stan d ard series of pairs w as
soon replaced b y more flexible methods. Twym an an d Hitchen (Twyxnan and Hitchen, Proc. Roy. Soc.
(London) 31, 169 (1930), Nitchie a n d Standen, (Nitchie and Standen, Ind. Eng. Chem., Anal. Ed., 4, 182
(1932) a n d other workers used m echanical a n d optical m eans of evaluating the relation betw een pho­
tographic blackening a n d the concentration of the element under test. Blackening-concentration curves
w ere constructed'. By standardization of the conditions of exposure an d the photographic procedures
a concentration w as obtained b y the evaluation of the blackening on this curve. The photographic
variables encountered w ere such that for more accurate results a series of stan d ard specim ens of
know n concentrations w as used on each spectrogram for com parison with the specim ens for analysis.
Using this procedure the results w ere obtained b y either visual com parison or optical m easurem ents.
A blackening-concentration curve for the individual plate m ay b e used for the analysis. By com pari­
son or evaluation of photographic blackenings a s functions of the concentrations less accurate results
w ere obtained than b y the com parison of spectral intensities with concentration. This w as attributed
largely to the errors introduced b y the variations in the photographic m aterials a n d the processing of
the spectrogram.
Thomson an d Duffendach (J. Op. Soc. Am., 23, 101 (1933) described a method of m easuring
the relative intensities of spectral lines in which a H ansen (Hansen, G.» Z. Physik, 29, 356 (1924) "stepdiaphragm " w as substituted for the slit of a spectrograph an d a source of continuous light w as used. A
series of continuous spectra are produced b y one exposure in which, for a given frequency, the inten­
sity of the light varies a s the known widths of the openings in the diaphragm . W hen the
blackening valu es of these continuous spectra a re g raphed ag ain st the corresponding
relative intensities a "calibration curve" is produced. This curve m ay b e u sed to find

263062

the intensities corresponding to the blackening values of the line im ages of the spectral frequencies on
that plate. Hurter a n d Driffield {Hurter and Driffield, J. Soc. Chem. 9, 455 (1890) m ad e a series of inves­
tigations of photographic m aterials a n d clearly set forth the characteristics of this calibration curve.
This curve is sometimes referred to as a n H a n d D curve. Such a curve w as m ad e for the calibration of
each spectrogram b y Duffendach, Wolfe, a n d Smith (Duffendach, Wolfe, and Smith, Ind. Eng. Chem.,
Anal. Ed., 5, 226 (1933). From this curve the relative intensities for the spectral frequencies were found.
These intensity values are expected to v ary as functions of the concentrations. Under proper excitation
conditions ratios of these intensities are reproducible, an d a perm anent working curve m ay be con­
structed. The spectral photometry consists of three steps for analysis: first, the calibration of eac h
spectrogram as referred to above; second, the evaluation of the spectral intensities from photographic
blackenings; a n d third, the deduction of the concentrations from the working curve.
The method outlined in this p ap er deals particularly with photographic spectral photometry. Fig.
4 shows the characteristics of a spectrogram through a series of H a n d D curves. The spectrum for
e ac h concentration in a series of standard sam ples w as photographed on this plate. This spectro­
gram w as m ade b y a standardized spectrographic an d photographic technique a n d the results used for
the construction of a working chart.
EX P E R I ME N T A L
Apparatus and Materials.
A quartz spectrograph of medium dispersion w as found suitable for this work. A dispersion of
2100-7000 A required a ten inch plate. Instruments with more dispersion are n ecessary w hen the
spectral frequencies involved are not clearly separated from the frequencies of other elem ents present.
A m edium quartz Bausch a n d Lomb spectrograph w as found satisfactory for the application in­
volved.
The electrode holders an d a rc stand must h av e a m eans of conveniently raising a n d low ering
each electrode vertically an d hav e a lateral adjustm ent for alignm ent with the optical path.
The rotating adjustable sector w as driven b y a motor a n d m ounted on the optical bench, placed
betw een the slit an d the source.
The electrical circuit consisted of a voltmeter an d am m eter with a variable resistance to adjust the
current furnished by a 300 volt 15 am pere D-C motor generator set.
The photographic blackenings w ere determ ined with a Bausch an d Lomb com parator. Some
modifications of this instrument w ere found helpful in obtaining reproducible values for line blacken­
ings. These modifications will be discussed later in the paper.
Eastm an processing m aterials a n d spectrographic plates w ere used.
Arc Electrodes.
Spectrographic graphite rods, 5/16 inch in diam eter w ere used for electrodes. The regular Acheson graphite rods w ere usually em ployed. These rods contain certain impurities which sometimes
cause error but w ere found sufficiently pure for m any applications. Higher quality rods m ay b e pur­
ch ased for special uses. In routine practice the regular rods m ay be given a purification treatm ent.
Satisfactory m ethods of purification h av e been described b y Standen a n d Kovach (Standen, G. W.,
and L. Kovach, Proc. A. S. T. M. 35, II (1935). Such m ethods hav e been found very useful in the pre­
paration of electrodes for other work in this laboratory. O ne end of these rods of convenient length for
arcing must b e recessed sufficiently to receive the sam ple. Rods treated for purification a re more por­
ous than untreated rods a n d do not require a s large a recess. This recess or bore m ay be m ade with
a h an d cutting tool, a s these rods a re pressed graphite. Since this operation, the preparation of elec­
trodes, is too time consuming for routine practice, it w as necessary to design a m achine tool. The
characteristics of, an d the variations, in, the rods presented a problem which w as solved b y m eans of
a motor driven recessing tool.
The draw ing in Figure 1 shows the assem bly cf the tool, which w as m ade of m achine steel. The
cutting tool, a drill designed to m achine a cavity cf the size an d sh ap e desired, is set in a special
chuck. This chuck h as a cutting face for m achining the end of the graphite rod. The .outside d ia ­
m eter of the chuck is stan d ard for the inside bore of the body. It is interchangeable with other chuck
assem blies which m ay be needed for various types a n d sizes of cavities an d for larger or sm aller d ia­
m eter rods. This chuck assem bly is held in the proper position b y m ean s of a set screw in the body of
the tool. The third part of the assem bly is a n ad o p ter bushing which acts a s a b earing a n d a s a
m eans of centering the graphite rod. The outside tap er of this ad ap ter bushing is standard, a s sev eral
such bushings are necessary for various diam eter graphite rods used in spectrographic analysis.
The parts w hen assem bled m ake a simple tool which will stand long u sag e without showing a loss iri
efficiency, since the graphite rod cuts easily a n d is a self-lubricant for the steel bearing.

CARBON OUrLFT-

^ m/LL z n o te’s.

Figure 1

A TiA PTCf? B t/S H tN G
WITH BJPOWA/ <t SH A /?PE TA PS/?.

GRAPHITE £l£CT&Q£E RECEJSEE

This tool c an be m ounted on the shaft of a sm all motor a n d fixed in place with a set screw as
show n in the draw ing. The rods m ay be m achined b y inserting them into the tool: a n operation
sim ilar to that of sharpening a le ad pencil. This tool carefully m ad e of suitable m aterials will serve
its purpose without contam inating the graphite rod with iron sufficiently to ap p ea r in the spectrum of
the sam ple. If the electrodes require a purification treatm ent the drilling is carried out previously.
27

25
24

5?
-S

23

3

22

■8

J2
m

21
20

IS

3

4

S

6

9

a

9

10

li

Am peres
Figure 2—Curves showing spectral blackening vs.
excitation current for the following frequencies.
A —Beryllium A2348.62 A
O —C opper

A2369.91 A

| —Copper

A2303.13 A

• —Copper

A2400.10 A

o

ss

•H

CO
a
3 30
CJ

25

A m peres
Figure 3—Curves for spectral intensity ratios vs..
excitation current.
r Be A 2348.62 A

/

[C u A 2369.91 A

/

r Cu A 2400.10 A
1 Cu A 2400.10 A

Method of Excitation.
The procedure used in the preparation of stan d ard solutions an d sam ples w as essentially that' d e­
scribed by Nitchie an d Standen (Nitchie and Standen, Ind. Eng. Chem., Anal. Ed., 4, 182 (1932). Aliquot
portions of 0.1 ml. of hydrochloric acid solution containing 250 gram s p er liter of the allo y w ere pipetted
into the recess of the graphite electrodes. These electrodes w ere allow ed to dry in a n oven at 100° C.
For thirty minutes. An arc betw een two graphite electrodes, the lower being the positive an d contain­
ing the sam ple, w as excited for sixty seconds at ten am peres with a potential drop of forty volts
across a n electrode g ap of approxim ately 3/16 inch. At the end of the exposure the sam ple show ed
practically complete volatilization from the electrode.
In Figure 2 it m ay be observed that the blackening of the lines of certain spectral frequencies chosen
show in increase in- density with a n increase in the arcing current. A maximum is ap p ro ached at ten
am peres which is below the maximum density that the plate is cap ab le of recording. At this am per­
ag e the emission becam e almost independent of the sm all variations in the current or the tem peratures
of the electrode.
In Figure 3 it m ay be observed that the ratios cf the intensities of certain spectral frequencies of
copper an d beryllium are practically independent cf the current variations a t about ten am peres. These
variations in the current are limited to about two percent from the m ean value. These conditions a n d
results of excitation are reproducible an d represent the b asis of the modified method of "internal con­
trol" (Gerlach, W., Z. Anorg. Allgem. Chem. 142, 389 (1925); Lundegardh, Die Quantitative Spektralanalyse der Elemente, Gustave Fisher, Jena (1929); Gerlach and Schweitzer, Chemische Emissions
Spektralanalyse, Grundlagen und Methoden, Leopold Voss, Leipzig (1930); Twyman and Hitchen, Proc.
Roy. Soc. (London) 31, 169 (1930); Nitchie and Standen, Ind. Eng. Chem., Anal. Ed„ 4, 182 (1932) u sed
in this work. It w as found that these conditions w ere satisfactory for the excitation of the minor con­
stituents being determ ined. The use of internal control frequencies requires that their intensifies b e a r
reproducible ratios with the intensities of the frequency used for the elem ent being determ ined. This
investigation of excitation conditions w as carried out in order to determ ine the arcing, conditions whicht
yielded the most reproducible results for the working chart.
Working Chart.
The w orking chart used in this method of spectrographic analysis fs a g rap h ical m eans of evalu­
ating the concentrations of the elem ents under test from the photographic blackening of their spectral!
lines. This chart consists of a series of curves with blackenings a s ordinates a n d relative intensities'
a s abscissae. Figure 4 is the working chart em ployed in this method for the determ ination of zinc ins.
copper.. The procedure required for the construction of this chart consisted of three steps-

The first step is fee determ ination of fee H a n d D curve c l a spectrographic plate. This requires
a suitable light source a n d a m eans of varying its intensity for successive exposures. In tins w ork a
copper a rc w as em ployed a s fee light source. A rotating adjustable sector w as u sed to v a ry fee
light for fee exposures. After fee p late w as developed, fee blackenings of a suitable copper frequency
w ere m easured from these spectra an d g raphed against fee know n sector openings a s intensity
values. In feis m anner fee H a n d D curve w as established for this plate.
The second step requires feat a num ber of spectra of each concentration of fee stan d ard sam ples
b e photographed a t various sector openings on the plate being calibrated. This plate m ay b e con­
sidered fee stan d ard spectrogram for the working chart being constructed. If fee spectra of more stand­
a rd sam ples are required than can be photographed on feis standard spectrogram, these exposures
m ay b e m ade on a similar plate an d developed in a like m anner. The H a n d D calibration of these
subsequent plates is not necessary. The blackening of fee frequencies selected for control purposes
a n d of fee elem ent being determ ined w ere evaluated on the comparator,. This d a ta is necessary for
fee construction- of fee working chart.
The third step is fee graphical construction of the working chart. This chart consists of a series
of curves w hich are fee sam e H a n d D curve in various positions wife respect to fee Log E axis. This H
a n d D curve is fee curve determ ined from fee copper arc calibration of fee stan d ard spectrogram .
0
Let this curve b e designated the specific curve for one of fee control frequencies, such a s C u 3329.62 A
in fig u re 4. The pro p er position of a similar curve for a second control frequency w as located in
the following m anner: blackening values for a second control frequency from the stan d ard sam ples
w ere plotted ag ain st Log E values of fee first control frequency of these spectra. In a sim ilar m anner
the curves m ay be located for other control frequencies.
Since fee concentration, of fee elem ent under test varies, fee H an d D curve of its frequency will
h a v e a different position for each of fee stan d ard sam ples in fee series. These curves in feedr proper
positions a re called concentration curves. Such curves -are shown in Figure 4 'and represent the
O
frequency Zn 3345.05 A for fee zinc stand ard s. These curves are located in the following m anner. The
Log E values of the control frequencies of each spectrum of a particular concentration are located. The
density values of fee frequency of the elem ent under test w ere then plotted ag ain st these Log E values.
A curve through these points is fee H a n d D curve in the proper position to represent the particular con­
centration wife respect to its control frequencies. In a sim ilar m anner fee position of the H a n d D curve
m ay b e located for each concentration used. It should b e noted here feat this arrangem ent of curves
is such feat fee blackening values of fee frequencies in a spectrum h av e fee sam e Log E value. This
value h a s b een referred to a s fee actual Log E value of feat particular region of fee spectrum. This
procedure w as em ployed to construct fee perm anent working charts used in this method.
31
30
89

23
27

'---I

m

80
.18
17

0

1

2

3

4

6

6

Log. 2 E
Figure 4—W orking chart for the determ ination of
zinc in copper.

This chart also incorporates a satisfactory m e an s of internal control which a t the sam e time com­
pensates for such variations a s speed a n d contrast in photographic plates. W hen a blackening vcpue
»oi the frequency of a n elem ent under test is properly located on the chart, the concentration m a y b e
re a d from a concentration curve or b y the correct interpolation betw een two adjoining curves.
In the construction of the working charts various procedures w ere investigated for the calibration
of the stan d ard spectrogram . A source of continuous light w as used for the calibration spectra. T he
curves w ere in agreem ent with those obtained from the copper arc. The H ansen step-diaphragm
would h a v e been a convenient m eans of varying the light from a single exposure for calibration pur­
poses.
In analytical procedures for the determ ination of more than one unknown, stan d ard solutions
m ay b e prepared containing the proper concentrations of each element. In this m anner one stand­
a rd spectrogram m ay be used to determ ine the control curves for all the elem ents present. The re­
quired num ber of working charts m a y be constructed from this standard spectrogram or from sub­
sequent plates.
Analysis of Samples.
A sam ple of the copper allo y w as dissolved in hydrochloric acid. The volum e w as adjusted to
give a concentration of 250 gram s of the alloy p er liter. Recessed carbon
electrodes w ere im pregnated
with the alloy from 0.1 ml. portions of this solution. This process m ay b e accom plished in routine prac­
tice with a buret of the type described by Standen a n d Fuller (Standen, G. W., and M. L. Fuller, Ind.
Efrg. Chem., Anal. Ed., 6, 299 (1934). The electrodes a s described under A pparatus an d M aterials,
w ere arced at 10 am peres with a potential drop of 40 volts. A rotating sector betw een the slit a n d
the arc w as adjusted to permit the required intensity of the light to enter the slit. The plate w as ex­
posed to the emission from the arc containing the sam ple for the first sixty seconds of the excitation.
The plate w as developed with a standardized processing technique. The tem perature, time, agitation,
an d developer w ere essentially the sam e a s used for the developm ent of the plate required for the
working chart. The blackening values of two suitable control frequencies an d the frequencies of the
elements being estim ated w ere evaluated on the com parator. The control frequency blackenings
located on the curves of the working chart determ ine a straight line. This line a n d the blackening
value of the frequency chosen for fee elem ent under test locate a point on or betw een two concentra­
tion curves. This point represents a concentration of the elem ent under test. The evaluation of this
concentration is m ade a s follows:
w here
log

Cx — elog C x
Cx = concentration sought.
Cx = (log C i—log Cy ( ^ ~ ^ ) -

log C2

Ci a n d C2 are the concentrations of the curves
betw een which the point is located.
^ 2 — dx = horizontal axial distance betw een the
point an d lower curve,
do — di = horizontal axial distance betw een fee
higher a n d lower concentration curves.
The calculation of the Cx, fee unknown concentration, w as simplified b y the tabulation of fee log­
arithm s of fee concentrations of the standard sam ples a n d fee differences betw een the logarithm s cf
successive concentrations.
Copper alloys containing aluminum, tellurium, beryllium, cadm ium an d zinc w ere an aly zed b y
the above procedure. These m etals w ere present a s minor constituents in copper. No difficulty d u e
to the interference of the spectral frequencies w as encountered. The standard sam ples w ere p rep ared
containing various am ounts of each of these m etals a n d copper. A series of copper solutions contain­
ing these five elem ents at different concentrations w as tested spectrographically a n d found to b e satis­
factory for use in fee preparation of fee working charts. In Table I are shown the frequencies used for
fee concentration a n d control curves of fee elem ents under test for fee charts in Figures 4, 5, 6, 7, a n d 8.
The w avelengths of fee frequencies a re given in International Angstrom Units.
Test Element
Frequency of
Test Element
Frequencies of
Copper control
for the test
elem ent

A1

TABLE
Te

I
Be

Cd

Zn

2373.12

2385.78

2494.87

2288.03

3345.05

2369.91
2392.64
2406.68

2354.63
2369.91
2400.10

2492.15
2441.62
2489.64

2303.13
2276.26

3349.26
3329.62
3354.47

31
30

23
27

25

24

19

16
15

0

1

•2

3

4

5

6

Log. 2 E
Figure 5—W orking chart for the determination of
beryllium in copper.
These values used to identify the frequencies were published b y H. Kayser (Kayser, H., Tabelle der
Haupttinien der Linienspektra aller Elemente, Julius Springer, Berlin (1926). The photographic black­
ening values of these frequencies w ere evaluated for each sam ple. Two exposures w ere made, for
eac h sam ple, one being of a higher intensity than the other. This practice g av e a t least one spectrum
with blackening values for each elem ent that w ere in the straight line section of the H a n d D cutve.
The an aly se s carried out w ere within the range of concentrations shown in Table II.

29
28
27

tJl

25

|0)

M

8
m

23

22
21
20
19
18

16
15

0

1

2

3

Log.

4

2

5

6

E

• Figure 6—W orking chart for the determ ination of
alum inum in copper.

z
Log. 2 E
Figure 7—Working chart for the determ ination of
tellurium in copper.
TABLE
AL
Te
Minimum Cone.
0.005%
0.01%
M aximum 'Cone. 2.0 %
2.0 %
The quantitative 'determ ination of these m etals in
show n in Table II.

II
Be
Cd
Zn
0.0025%
0.01%
0.02%
0.10 %
1.0 %
4.0 %
copper is not limited to th e concentrations

so
29
28
27
26

G*

25

d

24

o

23

■I I

8 »
CQ
H
—
♦

20
19

17

1

3

4

5

6

Log. 2 E
Figure 8—W orking chart for the determ ination of
cadm ium in copper.

A m easure of fhe accu racy of the analysis of one hundred successive determ inations m ay b e ob­
served from the results show n in Table III. The m ean error encountered for all trials w as less than
5.0%, including the unfavorable results in brackets. The untreated graphite rods of the quality used
in this Work, together with som e uncertainty of the excitation conditions, m ake such results unavoidable.
The errors in this work are attributed in the greater p art to variables in the excitation a n d m ay b e
minimized through the u se of electrodes of the alloys. Preliminary experiments with copper alloy elec­
trodes a s stan d ard s a n d specim ens indicate that the excitation conditions m a y b e controlled with a
higher degree of accu racy a n d more dependability than for graphite rods a n d solutions. However,
the graphite electrodes permit lower limits of detection of certain elem ents in copper than h av e been
obtained from m etallic electrodes due to the higher tem peratures attained for a given am perage. Thus
the ad v an tag es, previously pointed out, resulting from the use of graphite rods, w ere of sufficient
im portance to w arran t their use in this work.
TABLE in
Spectrographic determ ination of minor constituents of copper.
Amount Found
Amount Added
per Alloy
Element
Percent
>le Number
Percent
Determined
0.0053
0.0050
la
Be
0.0052
0.0050
lb
Be
0.0052
0.0050
Be
1c
0.0074
0.0075
2a
Be
0.0073
0.0075
2b
Be
0.0073
0.0075
Be
2c
0.0107
0.0100
Be
3a
0.0100
0.0100
Be
3b
0.0151
0.0150
Be
4a
0.0153
0.0150
Be
4b
0.0150
0.0150
Be
4c
0.0255
0.0250
Be
5a
0.0247
0.0250
Be
5b
0.0500
0.0500
Be
6a
0.0485
0.0500
Be
6b
0.0113
0.0112
Be
7a
0.0102
0.0112
Be
7b
0.0106
0.0112
Be
7c
0.0154
0.0165
Be
8a
0.0160
0.0165
Be
8b
0.0073
0.0075
Be
9a
0.0071
0.0075
Be
9b
0.0076
0.0075
Be
9c
0.0193
0.0200
Be
10a
(0.0170)
0.0200
Be
10b
(0.0172)
0.0200
Be
10c
0.0093
0.0100
Be
' 11a
0.0100
0.0100
Be
lib
0.0094
0.0100
Be
11c
0.0379
0.0375
Be
12a
(0.0097)
0.0087
Be
13a
0.0092
0.0088
Be
13b
0.0082
0.0088
Be
13c
0.0065
0.0063
Be
-■ 14a
0.0065
0.0063
Be
14b
0.0124
0.0125
Be
15a
0.0124
0.0125
Be
15b
0.0272
0.0263
Be
16a
0.0257
0.0263
Be
16b
0.0233
0.0225
Be
17a
0.0223
0.0225
Be
Tb
0.0227
0.0225
Be
17c
0.0299
0.0288
Be
18a
0.0280
0.0288
Be
18b
0.0298
0.0288
Be
18c
0.0145
0.0137
Be
19a
0.0132
0.0137
Be
19b
0.0147
0.0162
Be
20a

TABLE Iir (Continued}
20b
21a
2 2a
22b
22c
23a
23b
23c
24a
24b'
24c
25a
25b
26a
26a
26a
26b
26b
26b
27a.
27a
27a
28a
28a
28a
28b
28d
28b
28b
28c
28c
28c
28c
29a
29a
29a
29b
29b
29b
29b
30a
30a
30a
30a
31a
31a
31b
31b
31b
31b
31c
31c

Be
Be
Be
Be
Be
Be
Be
Be
Be
Be
Be
Be
Be
AI
Ca
Zn
AI
Te
Zn
AI
Ca
Zni
AI
Ca
Zn
AI
Ca
Te
Zn
AI
Ca
Te
Zn
AI
Ca
Zn
AI
Ca
Te
Zn
AI
Ca
Te
Zn
AI
Zn
AI
Ca
Te
Zn
AI
Te

0.0162
0.0287
0.0275
0.0275
0.0275
0.0080
0.0080
0.0080
0.0043
0.0043
0.0043
0.0075
0.0075
0.625
0.500
1.250
0.625
1.250
1.250
0.550
0.350
1.100
0.175
0.350
0.350
0.175
0.350
0.350
0.350
0.175
0.350
0.350
0.350
0.300
0.600
0.600
0.300
0.600
0.600
0.600
0.150
0.550
0.550
0.550
0.525
1.050
0.525
0.300
1.050
1.050
0.525
1.050

■

(0.0144?
0.0286
0.0260
0.0299
0.0264
(0.0088)
0.0078
0.0085
0.0042
0.0043
0.0043
0.0073
0.0076
0.640
0.546
1.150
0.670
1.300
1.190
0.510
0.320
1.050
0.160
0.330
0.375
0.180
0.360
0.343
0.350
0.185
0.380
0.376
0.380
0.270
0.550
0.570
0.320
0.640
0.630
0.630
0.142
0.520
0.590
0.510
0.560
1.000
0.500
0.280
0.970
1.080
0.550
1.100

Discussion.
A m aterial m ay be h eated fa incandescence b y m eans of a n arc, spark, or flame. The arc m a y b e
used betw een electrodes containing the material. In this work electrodes im pregnated with a solution
of the alloy w ere used for analysis. This operation w as carried out b y a solution technique w hich
offered som e ad v an tag es over the use of metallic electrodes (Nitchie, C. C., Ind. Eng. Chem., Anal. Ed„
1, 1 (1929). By this procedure representative "unknow n" an d stan d ard sam ples w ere conveniently a n d
accurately obtained. A 50 gram sam ple m ay be dissolved a n d only 0.1 ml. portions of this solution
used, obviously representative of the sam ple. The preparation of copper alloy electrodes of know n
concentration is a difficult a n d tedious procedure for the m etallurgist a n d a representative sam ple is

difficult to obtain for analysis. An arc betw een carbon electrodes reaches a very high tem perature
w hich heats the vapors of the sam ple to incandescence a s volatilization from the electrode takes place.
The sam ple w as volatilized from the electrode in a relatively short arcing period. Through careful selec­
tion of the exposure time a n d the arcing current, the rate of vaporization of each minor constituent to
that of the principal m aterial w as found to b e reproducible from electrodes prep ared b y a standard­
ized technique. Some conditions of arcing, such a s w andering of the arc about the electrodes, caused
poor reproducibility of the intensity of the spectrum. It w as found that the intensity of the spectral
frequency of the elem ent under test varied with the intensities of certain spectral frequencies of the
principal elem ent present. This m eans of internal control w as used to com pensate for the variable.
In some applications internal control h a s been effected more satisfactorily through the use of a
frequency or frequencies of a n elem ent ad d ed a s a n internal standard (Gerlach and Schweitzer, Foun­
dations and Methods of Chemical Analysis by the Emission Spectrum, Adam Hilger, Ltd., London
(1930) in a know n concentration to the solutions of the sam ples. Certain frequencies of the principal
elem ent of the alloy w ere found reliable in this investigation. These frequencies a re represented on
the working charts a s H an d D curves. These curves h av e another function in addition to that of in­
ternal control of intensity. This second function is a m eans of control of the speed a n d contrast of the
em ulsion of the spectrogram. It w as found that the single coated emulsion plates w hen exposed for
calibration purposes an d properly developed gave a characteristic H an d D curve which in the range
of certain spectral frequencies w as a straight line function except n ear the maximum a n d minimum
blackenings. The linear section of this curve represented the characteristics of the emulsion, but the
curved sections n ear the maxim um a n d minimum blackenings w ere not reliable. This straight line sec­
tion of the H a n d D curve w as therefore reproducible betw een certain blackening limits.
Two characteristics of a n emulsion, the speed an d the contrast, w ere found to v ary som ew hat from
p late to plate. The standardized processing technique used for developm ent of the emulsion reduced
these variations considerably. The working chart w as so arran g ed that the variations in the speed
of the em ulsion h av e no effect on the results a s long a s the density values of the frequencies a re on
the straight line section of the curves. However, variations in the contrast of the emulsion h a d to b e
treated b y a method of calibration control incorporated on the chart with internal control. The actual
functioning of this m ethod of control involves the use of H an d D curves. These curves must h av e a
straight line section. The flexibility of this m eans of control depends greatly upon, the length of this
section of the curve. The m aterials a n d the photographic developing practices used determ ine to a
g reat extent the nature of the curve.
Satisfactory H a n d D control curves are shown in Figure 9 for three plates calibrated with line
spectra from a copper arc. The curves for Plate I w ere considered the control curves for Plates II
a n d III. These curves w ere constructed in a m anner similar to the control curves for the working chart
used in this method. Plates I a n d II w ere selected from the sam e box of new plates, exposed a n d
developed under standardized conditions for routine practice. The curves for Plates I a n d II in Figure 9'
show very similar characteristics. Plate III w as selected from another type of single coated emulsion.
The curves for this plate show that both the speed a n d contrast of the emulsion a re som ew hat differ­
ent than the sam e .characteristics of Plate I. However, blackening values from Plate III can b e applied
to the control curves of Plate I with a reliability com parable to that of applying these sam e values to
the curves which represent the emulsion of Plate III. In Figure 9 such values from a spectrum of
Plate III are applied to the respective curves for Plate I taken a s a stan d ard spectrogram. The result
is that two of these values determ ine the straight line A 1 which passes through the third curve a t a
blackening, value corresponding to the v alu e recorded for the third frequency on Plate III. Hence such
a point m a y b e located for concentration curves b y using the control blackening values to locate a
line such a s A 1 a n d then ap plying the blackening value of the frequency for the elem ent under test
to this line. This point m ay b e evaluated in terms of the adjoining concentration curves a s con­
structed for the working charts.
The .curves, however, in Figure 9 for Plates I an d II represent results which w ere in general re­
producible. The blackening v alues of Plate II applied to the control curves of Plate I determ ine such
a line a s A which is perpendicular to the intensity axis an d h a s a n intensity value in terms of Plate I.
It is therefore possible to locate this line with one control value, using a second value to check the re­
liability of the practice. If the plates are not of the sam e contrast the second control blackening value
will, obviously determ ine a straight line inclined to the axis. The dependability of using this line
m ay b e checked b y applying a blackening value of a third control frequency to its curve. This point
should also b e on the sam e straight line determ ined b y the first two values. This practice, a s dem ­
onstrated in Figure 9, should com pensate for the variations in contrast a n d sp eed w hen encountered
a n d check the reliability of using the H a n d D curves a s straight lines within the limits of certain black­
ening values. These limits a n d variables encountered depend considerably on the photographic m a­
terials a n d technique used.

51

38

26
ZS
24

"S
i3
G
m

23

22

21
20
18

15

0

1

2

S
6
Log. 2 E
Figure 9—Chart dem onstrating a method of expos­
ure an d photographic plate control.
© —Copper control curves for Plate I.
A —Copper control curves for Plate II.
O —Copper control curves for Plate III.
3

4

In order to select the most suitable m aterials available certain factors m ust be considered. The
most importont characteristics of the photographic em ulsions a re the speed a n d contrast. In actual
practice the intensity of the illumination can be adjusted in accordance with the exposure time utilized
an d the speed obtained from the photographic m aterials. The contrast, therefore, is the im portant
factor for consideration.
Consider the hypothetical characteristic curves shown in Figure 10. N eglecting the speeds of
these two m aterials the im portant difference lies in the contrast, gam m a, the slope of the character­
istic curve. Assuming such physical characteristics of these two emulsions a s graininess, turbidity, a b ­
sorption, a n d resolving pow er to be the same, the higher contrast of em ulsion A m ak es it the more de­
sirable in quantitative analysis for a short ran g e of concentrations w here high acc u racy is necessary.
Emulsion B m ay be more desirable than A w here the sam ples v ary over a long ran g e of concen­
trations. However, the differential in the blackening for a given ch an g e in concentration is necessarily
lower for the lower contrast emulsion.
In general the quantitative applications require a plate with a high contrast em ulsion. It h a s been
pointed out that the contrast changes in accordance with the frequency or frequencies of the light to
which the em ulsion w as exposed. It follows, then, that the characteristic curve of the em ulsion
repre­
sents the
photographic response to light in certain sections of the spectrum. H ence a n em ulsion
should be chosen with the proper contrast in the region of the spectral frequencies used. It is desir­
ab le to select a plate which h as the sam e, or if necessary, a slowly changing contrast over that section
of the spectrum for which it is suitable. It m ay b e found n ecessary in som e applications to use more
than one emulsion, since frequencies in the visible range m ay b e n eed ed for one constituent an d a n ­
other em ulsion for a n elem ent in the ultraviolet region. Some of the plates u sed in reg u lar practice
m ay be m ade sensitive to ultraviolet, visible, a n d red light b y the proper treatment.
The final step in the selection of a n emulsion or its treatm ent is the construction of a n H a n d D cali­
bration curve for the spectral range involved. The plate selected should h a v e the d esired contrast
a n d give a long straight line section curve. The blackening valu es on the straight line section of the
characteristic curves w ere found more reliable than those n ear the limiting v alues. For this reason
the blackenings recorded w ere limited by certain maximum a n d minimum values. * The processing
technique a n d developing m aterials used depended upon the em ulsion characteristics. Since m an y
developers are affected b y tem perature an d agitation, these conditions w ere standardized for the type
of plate used. The developer w as replenished for each processing care being taken to u se precisely the

30

87

86
85

18

16

2

3

4

5

6

7

8

Log. 2 E
Figure 10—Characteristic H a n d D curves of two
types of photographic emulsions for spectrochemical
analysis.
sam e concentration. The straightness an d length of the straight line section of the calibration curve
depends greatly upon the proper digestion of the emulsion. This digestion problem varies consider­
ably with the em ulsion used an d the method of agitation. If a double coated emulsion is used the two
layers m ay not h av e the sam e contrast. This feature m akes it inadvisable to use an y plates except
single coated em ulsions for work w hich depends on a straight calibration curve.
The blackening values used in this work w ere evaluated with a density com parator. The im age to
be m easured w as brought to a focus on a screen. The optimum intensity of the illumination through a
clear portion of the plate w as adjusted to give a predeterm ined deflection of a galvanom eter in a cir­
cuit with a photronic cell installed back of a slit in the screen. An iris diaphragm properly placed in
the projection system will allow this adjustm ent. The deflection of the galvanom eter caused by the
light which p assed through the darkest portion of the im age w as recorded as the blackening value.
A variable resistance in parallel with the cell w as used to com pensate for variations in the cell resist­
ance or sensitivity. Through these modifications an d the proper calibration of the density com parator
as shown by scale readings, it w as found possible to record galvanom eter deflection readings as
blackening values. These values are reproducible a n d
have proved to be a linear function of the
blackness of the im age. Errors caused b y the variations
of the line voltage used forillumination were
minimized by the use of a voltage regulator.
Summary.
I. A method of spectrographic an aly sis of general application h as been outlined. It is exem pli­
fied by the quantitative determ ination of beryllium, cadmium, aluminum, tellurium, an d zinc; present
a s minor constituents in copper alloys of these metals.
II. A new procedure of "photographic plate control" incorporating internal control of intensities
for spectral frequencies is dem onstrated.