NUMBER 3, 1957 115

The Spectrochemical Analysis of White Metals by the Paint to Plane Spark Technique::'

Henry K. Nausester

Spectrochemical Laboratory, The Canada Metal Company, Ltd., Toronto, Ontario, Canadal

Abstract A method :for the spectrochemlcal analysis of white metals by

the point to plane spark techmque :for the determination of mator, minor and impuri ty elements is described Part one of this paper covers the common lead base alloys such as ant~monlal lead, typemetals, babbitts and solders. Part two deals wi th the common tm base alloys, namely babbttts and solders.

Introduction

Spectrochemical analysis of lead and lead base alloys has been carried out in the Canada Metal Company as early as 1944. From this date on and until recently the point to point (pin to pin) spark technique was utilized. Thas technique did not lend itself to the determination of major and/or minor constituents in lead and tin base al- loys, with the exception of a small number of impurity elements, the reason being the nonuniformity of pin type samples due to segregatmn and pipe effects, despite careful sampling. Since spectrographic disk standards of lead and lead alloys became available and also to relieve the chemical laboratory of numerous time-consuming and tedious tasks, a number of spectrochemical procedures employing the point to plane spark technique was established. Although various methods for the spectrochemlcal analysis of lead and lead alloys have been pubhshed so far, mostly originat- ing from the National Lead Co., all of those methods employ a unidirectional discharge (Multlsource) not ob- tainable with our High Voltage Spark Source. Therefore, the apphcation of those published procedures to our pres- ent equipment was rather limited.

The prehmlnary investagations carried out in estab- lishing a more suitable method for the spectrochemical analysis of lead and tin base alloys dealt with the behavior of various alloying constituents during spark excitation on a basis of time-intensity study.

Apparatus Spectrograph 1 ½ meter grating spectrograph, Abney mounting (Ap- plied Research Laboratories), 24,400 lines per inch grating, linear reciprocal dispersion 6.95 A per mm, wavelength range 2150-4350 A (first order ultraviolet), primary slit width 0.050 mm, slit length 3.0 mm.

Spark Source Independent design, producing a hagh frequency condensed spark. The number of discharge trains per second as con- trolled by an auxiliary gap an series wath the analytical gap.

Power 0.45 kva Input 115 volts Output 15000 volts r.m.s. Capacitance 0.006 mfd Inductance 1.5 mh Resistance, secondary residual

"Paper presented at the TMrd Ot tawa Symposium on Apphed Spec- t r o s c o p y - O t t a w a , Ontario---September 12-14, 19 5 6.

j'Present address Electro Metallurgical Company, Wor.ks Spectrographic Laboratory, Nmgara Falls, New York.

Number of discharge trains 120 per second Auxdmry gap 2.00 mm

These setting are used for all procedures outlined m this paper.

Densitometer Nonrecording, projection type (Applied Research Labora- tories).

Procedure Part one

This seclton covers pig lead, calczum lead, sdver lead, anttmomal lead, typemetals, lead base babb#ts and solder~.

STANDARD SAMPLES: A number of lead and lead alloy standards in the form of chdl cast disks was obtained com- mercially. Various standards were also prepared syntheti- cally or previously analyzed samples were investigated for application as standards.

SAMPLING T E C H N I Q U E PREPARATORY TO SPECTRO-

CHEMICAL ANALYSIS. The metal to be analyzed is eather sampled directly from the furnace (ensuring the heat is well alloyed and mixed) or the sample is melted in a clean iron ladle over a gas burner. In the latter case, the surface of the molten metal is covered with pure charcoal, the temperature kept between 700-900 ° F for 15 minutes and occasional mixing employed. This treatment will ensure proper alloying of antimony and/or tin with lead. (Large variations in spectrochemical antimony and tin results have been traced to improper alloying in spite of uniformity of the sample materml; this can be explained on the basis of '~grain boundary enrichment" of antimony and/or tin. Thas effect is not detected by wet chemistry.) The char- coal is now removed by skimming and the molten metal having a temperature of at least 700 ° F is poured into a cast iron book mold to form a chill cast disk 2 ½ " di- ameter and 4/16" thick. After cooling, the disk is re- moved from the mold, the riser is cut off and the surface to be sparked is machined on a lathe to a smooth finish. Electrode system: Petrey stand Lower eleclrode: ~/~" diameter medium grade graphite

electrode, 2" long, one end machined to a 120 ° cone tip Upper electrode: Machined disk sample Analytical gap: 2.00 mm

EXPOSURE CONDITIONS FOR LEADS

Prespar,~: None Sparl~ exposure: 2 X 15 seconds Exposure index: T of Pb 3229.9 ~ 25-35% Line pazrs and range of concentratton: Table I For maximum sensitlvlty of detection, all element lines are to be corrected for background. The llne pairs and ex- posure conditaons listed above are used for the determina- tion of minor and impurity elements in all lead base alloys. Separate working curves for impurity elements in lead base alloys are not required if matrix correction is applied to the intensity ratios of the hne pairs under consideration.

I 16 A P P L I E D SPECTROSCOPY

T A B L E I

Conc. Conc Element Ltne parr, A range % index ~o

Ant imony Sb 231147 /Pb 3229 9 0 001 -0.05 0.015 Sb 3232.50/Pb 3229.9 0.05 -0.50 0 10

T m Sn 30~4 12/Pb 3229.9 0.0006-0 02 0 005 Arsenic As 2349 84/Pb 3229.9 0 001 -0 05 0 009

As 2998.71/Pb 3229.9 0.05 -0.75 0.245 Copper 'Cu 3273 96/Pb 31192 0.0003-0.015 0.001

Cu 2961 16/Pb 3229 9 0 015 -0.20 0 06 Bismuth Bi 2989.03/Pb 32299 0.01 -0.09 0 021 Sdver Ag 2 4 3 7 7 9 / P b 3229.9 0.001 -0 02 0 005 Tellurium Te 2385 76/Pb 3229 9 0002 -0.07 0 012 Nickel Nx 3050.82/Pb 3229.9 0.001 -0 02 0 004 Zinc Zn 3345.02/Pb 3229.9 0 002 -0.02 0.0032 Cadmmrn Cd 3610.51/Pb 3229 9 0 002 -0 02 0 01 Alurmnum A1 3961.53/Pb 32299 00003-0.008 0.0008 Calcmm Ca 2997.31/Pb 3229 9 0.04 -020 0 15

"'50% neutral filter.

EXPOSURE CONDITIONS VOR ANTIMONIAL LEAD

Prespark: None Spark exposure: 2 X 15 seconds Exposure index: T of Pb 322.9 ~ 25-35% Line pazrs and range of concentratwn: Table II

T A B L E I I

Conc. Conc Element Lme parr, A range % index %

Ant imony Sb 272721/Pb 3229.9 0.30- 3 00 1 35 Sb 4033.54/Pb 32299 3 00-12.00 705

T m Sn 2785 03/Pb 3229 9 0.05- 0 80 0.15 Arsemc As 2744.99/Pb 32299 0 02- 0.15 0.04

As 2 8 9 8 7 1 / P b 3229 9 0.05- 0.80 020

EXPOSURE CONDITIONS FOR ELECTRO TYPEMETAL

Prespark: None Spark exposure: 2 X 15 seconds Exposure mdex: T of Pb 3229.9 ~ 25-35% Line patr~ and range of concentration: Table III

T A B L E I I I

Conc. Conc Element Line parr, A range % index %

Ant tmony Sb 2 7 2 7 2 1 / P b 3229.9 0.90-3.50 0 94 Tin Sn 2761.78/Pb 3229.9 090-4.00 1.45 Arsenic As 2 7 4 4 9 9 / P b 32299 002-0 20 0 06 Copper Cu 2961 16/Pb 3229.9 001-0 15 0.06

EXPOSURE C O N D I T I O N S FOR L I N O - , STEREO- , M O N O - ,

AND F O U N D R Y T Y P E M E T A L , BABBITTS

Prespark: None Spark exposure: 2 X 15 seconds Exposure index: T of Pb 2657.11 ~ 20-30% Line pairs and range of concentration: Table IV

T A B L E I V

Conc. Conc. Element Line pair, A range % index %

Ant imony Sb 272721/Pb 2657.11 9 00-28.00 13 15 Tin Sn 3655.78/Pb 2657.11 3.00-2000 8.00 Arsemc As 2 8 6 0 4 5 / P b 2657.11 0.20- 1.00 042 Copper Cu 2961.16/Pb 2657 11 020- 1 50 0.60

Only one set of working curves is needed for the deter- mination of antimony, tin, arsenic and copper in the vari- ous types of typemetals and lead base babbitts. However, it is advisable to establish and employ a separate working curve for antimony and tin in certain types of babbitts where the percentage of tin exceeds the percentage of antimony. These kinds of babbitts behave like solders dur-

ing spark excitation, i.e. a general enhancement of the antimony/lead and tin/lead line pairs has been observed.

EXPOSURE C O N D I T I O N S FOR SOLDERS C O N T A I N I N G 2- 20% TIN

Prespark: None Spark exposure: 2 X 15 seconds Exposure index: T of Pb 3229.9 ~ 25-35% Line pairs and range of concentration" Table V

T A B L E V

Conc Conc Element Ltne parr, A range % ,tdex ~o

Anttmony Sb 2311.47/Pb 3229 9 0 003 - 0.05 0 009 Sb 323250 /Pb 32299 005 - 050 015 Sb 2 7 2 7 2 1 / P b 3229.9 0 50 - 3 00 1 10

T m Sn 3223.56/Pb 3229 9 2,00 -20 00 7 00 Arsenic As 2349 84/Pb 32299 0003 - 005 0009

As 2744.99/Pb 32299 0.05 - 0 20 0 062 Copper Cu 2961 16/Pb 3229 9 0.01 - 0 20 0 066 Bismuth Bx 3 0 7 6 6 6 / P b 3229.9 0 10 - 0.80 0 37 C a d m m m Cd 3612.87/Pb 3229 9 0.01 - 0.10 0.022 Nickel N~ 3050 82/Pb 3229 9 0 0003-0 006 0 00084 Iron Fe 3631.46/Pb 3229 9 0.001 - 0 05 0 014

EXPOSURE C O N D I T I O N S FOR SOLDERS C O N T A I N I N G 2 0 -

6 0 % T I N

Prespark: N o n e

Spark exposure: 2 X 15 s econds Exposure index: T o f P b 2 6 2 8 . 2 6 ~ 2 0 - 3 5 %

Line pairs and range of concentration: T a b l e V I

T A B L E V I

Conc Conc. Element Ltne pair, A range ~o index ~o

Antimony Sb 2598.06/Pb 2628.26 0 30- 1 00 0 50 Tin Sn 3141.81/Pb 2628 26 20.00-60 00 33 00

For routine analysis of minor and impurity elements in solders containing 2-60% tin, a generahzed working curve for each minor + impurity element suffices. The gen- eralized curves are assembled from standards comprising of 15/85, 30/70 and 50/50 solders, containing minor and impurity elements of varying amounts in order to cover the concentration range under consideration. For all other impurity elements and/or concentration ranges not previ- ously covered, the line pairs listed under "leads" are to be employed with matrix correction.

EXPOSURE C O N D I T I O N S FOR SILVER LEAD ANt) T I N

R I C H SOLDERS

Prespark: None Spark exposure: 10 seconds Exhosure index: T of Pb 2446.19 and Pb 2476.38 ~ 15- 20% Line pairs and range of concentrahon: Table VII

T A B L E V I I

Conc Conc Element Lme pair, A range % index %

Sdver Ag 3382 96/Pb 2446.19 0.02-0.55 0 13 Ag 2 4 3 7 7 9 / P b 2476.38 0 50-3 00 1.50 Cu 327', 96/Pb 2476.38 0.10-1 00 0 34

The working curves for silver and copper are to be set up by using silver lead standards. For the determination of silver in tin rich solders, the silver/lead intensity ratio has to be corrected by the matrix factor.

Part two This section covers pig tin, t m bate babbitts and .~olders. STANDARD SAMPLES: A set of pig tin standards in the

NUMBER 3, 1957

form of rods ½ " diameter was obtained commercially. Tin alloys were prepared synthetically or previously ana- lyzed samples were investigated for application as stand- ards. The rod standards were held in a "bridge" during excitation on the Petrey stand.

S A M P L I N G T E C H N I Q U E PREPARATORY TO SPECTRO-

CHEMmAL ANALYSIS. The metal to be analyzed is either sampled directly from the furnace (ensuring the heat xs well alloyed and mixed) or the sample is melted in a clean heat resistant ceramic dish over a gas burner. Af te r thorough mixing, the surface of the molten metal is freed of dross by skimming and the metal having a temperature of about 800 ° F is poured into a cast iron book mold to form a chill cast disk 2 ½ " diameter and ¼ " thick. Af ter cooling, the disk is removed from the mold, the riser is cut off and the surface to be sparked is machined on a lathe to a smooth finish.

Electrode system: Petrey stand Lower electrode: ¼" diameter medium grade graphite elec-

trode, 2" long, one end machined to a 120 ° cone tip. Upper electrode: Machined disk sample. Analytwal gap: 2.00 mm

EXPOSURE C O N D I T I O N S

Prespark: None Spark_ exposure: 15 seconds Exposure mdex: T of Sn 3223.56 ~ 20-30%; T of Sn

3141.81 -~ 15-25% Lme pairs and range of concentration: Table VIII

T A B L E V I I I

Cone. Cone. Element Ltne pmr, A range % index %

Ant imony Sb 2598.06/Sn 3223 56 0.01 - 0.25 0.055 Sb 2574 l l / S n 3223.56 2.00 -1000 475

Copper Cu 3273.96/Sn 3223 56 0.001- 0.01 0.0022 Cu 2618.37/Sn 3223 56 0.01 - 0.20 0 12 Cu 2961.16/Sn 3223.56 0 20 - 100 0.49 Cu 3108 60/Sn 3223.56 1.00 - 9.00 4 0 0

lnd~um "As 2349.84/Sn 3223.56 0.008- 0.20 0 065 Arsenic Pb 2823 19/Sn 3223.56 0 003- 0.10 0.014 Lead Pb 2614 18/Sn 3223.56 0 02 - 0.20 0 035

B~ 3067.72/Sn 3223 56 0.002- 0.05 0.0094 Bismuth Ag 3382.89/Sn 3223.56 0 001- 0.01 0 0015 Silver Ag 3280.68/Sn 3141.81 0.01 - 0 25 0 043

Ag 2413.18/Sn 3141.81 0.25 - 3 50 1 27 Cd 3610 51/Sn 3223.56 0.008- 0.05 0.035

Cadmium N~ 3414.76/Sn 3223 56 0.002- 0.05 00086 Nickel Co 3405.12/Sn 3223.56 0.002- 005 0.012 Cobalt Fe 302049/Sn 3223.56 0.005- 0 10 0.0125 Iron Zn 3345.02/Sn 3223.56 0.002- 005 00105 Zme "In 3256.09/Sn 3223.56 0 002- 0 05 0.015

For maxlrnum sens~txvlty of detection these lines are to be corrected for background. Separate working curves for impuri ty elements in tin base alloys are not reqmred if matrix correction is applied to the

intensity ratxos of the line pairs under consideration.

P H O T O G R A P H I C PROCESSING

Type of emulsion Development

Stop bath Ftmng Washing Drymg

P H O T O M E T R Y

Kodak SA # 1 film Kodak D 19; rocked 3 minutes at 70 °

F Acetic acid 5 % ; 15 seconds Kodafix; 1 minute Flowing water; 2 minutes Blower and heater; 1 ~ minutes

a) Emulswn cahbratton. The SA # 1 emulsion is cali- brated by the 'Two Line" method as outlined by J. R. Churchill. This method has been found to be convenient, since aluminum alloys are also analyzed in our spectror graphic laboratory. But any other reliable and reproducible

117

method of emulsion cahbration util izing the iron spectrum may be used. The emulsion xs cahbrated for th e2150-4350 A wavelength region using the following hne pair Fe 3047.60/Fe 3037.39,r-~ 1.50.

b) Densttometry and matrix correction. The transmit- tance of the analytical and internal standard lines is mea- sured close to the analytical line; no background is mea- sured for the internal standard lines. The transmittance values are reduced to log intensity ratios by the use of the emulsion calibration curve employed in slide rule fash- ion with a standard log scale. Af te r background correc- tion has been made, the log intensity ratio (log I.R.) is now applied to the proper working curve and the per- centage of the element is established.

Considering minor and impuri ty elements in lead and tin alloys, a large number of working curves is needed to cover the more common alloys. This inconvenience is over- come by using a matr ix correction based on the dilution principle: All impuri ty and most of the minor elements ( to be de- termined) are to be contained in "pure lead" and "pure tin". The working curves established for those elements will be referred to as "pure lead" and "pure t in" curves. Exposure conditions and line pairs listed under leads and pig tin are to be employed for the determination of minor and impuri ty elements in lead and tin alloys. The intensity ratios of the various line pairs are to be corrected as follows:

lead alloys: Log I.K. X (% Pb present in the a l l oy ) / 100 z Net I.R.

Tin alloys: Log I.R. X ( % Sn present m the a l l o y ) / 100 -~ Net I.R.

This Net I.R. is now applied to the "pure lead" or "pure t in" curves and the percentage of the minor and/or im- puri ty element is established. The percentage of the pre- dominant (matrix) element such as lead m lead alloys or tin in tin alloys is found by determining all constituents present in amounts greater than 1%; same are to be added up and deducted from 100%.

c) Working curves. The working curves for major, minor and impuri ty elements are based on the values of log intensity ratio versus log percentage of the element. No improvement in precision of the analytical results was ex- perienced by plot t ing % e lement /% matrix element ver- sus log intensity ratio in the case of alloys containing two major elements.

D A T A ON T H E PRECISION OF T H E VARIOUS PROCE-

DURES. Due to lack of sufficient data no a t tempt has been made to present the precision on a basis of standard devi- ation and coefficient of variation. During a short period of application the reproducibili ty of analytical results on major, minor and impuri ty elements has been found to be satisfactory for routine spectrochemical analysis.

ACCURACY OF T H E VARIOUS PROCEDURES. The accu- racy is presented in Tables IX-XV as a comparison between chemical and spectrochemical results obtained from the analysis of identical sample material. All chemical and spectrochemical results presented are the averages of two single determinations.

T A B L E I X . PER C E N T C A IN C A L C I U M LEAD

Method # Cl # C2 # C3 Chermcal 0.104 0.089 0 10 Spectro 0 11 0.095 0 11 Spectro ~" 0 11 0 091 0.10

Analyzed by an independent laboratory using a solutxon spark tech- nique.

118

T A B L E X . A N T I M O N I A L L E A D

Sample No Method Sb, % Sn, % Cu, % As, %

G 101 c h e m t c a l 5 . 2 6 0 . 2 9 0 . 0 1 0 .15

spectro 5 0 9 0 . 2 6 0 . 0 1 0 12 G 102 c h e m i c a l 6 60 0 37 0 0 4 2 0 . 4 4

spectro 6 4 8 0 35 0 . 0 4 1 0 4 0 G 103 c h e m i c a l 10 67 0 25 0 01 0 . 3 0

spectro 1 0 . 5 5 0 2 7 0 01 0 . 2 9

G 1 0 4 c h e m x c a [ 10 7 6 0 4 4 0 0 6 5 0 37 spectro 1 1 . 2 5 0 45 0 067 0 35

T A B L E X I . T Y P E M E T A L S

Sample No. Method Sb, % Sn, % Cu, % As, %

T 1 c h e m l c a | 10 9 0 7 83 0 7 8 0 13

spectro 1 0 . 7 5 8 .05 0 . 9 3 0 15 T 2 c h e m i c a l 1 1 . 4 0 6 93 1 21 0 . 2 0

spectro 1 1 . 5 5 7 . 1 5 1 3 4 0 17 T 3 c h e m i c a l 14 98 4.95 0 12 0 . 0 8

spectro 1 4 . 6 5 4 . 9 0 0 . 1 0 0 05 T 4 c h e m x c a l 16 83 8 7 8 - - 0 . 2 6

spectro 17 0 0 9 . 0 5 - - 0 . 2 3

M - A c h e m i c a l 1 7 . 2 3 8 7 0 0 30 - - spectro 17 50 8 .55 0 2 7 - -

M - 3 c h e m i c a l 17 55 10 25 0 3 6 - - spectro 17 4 0 10 10 0 31 - -

F 1 c h e m i c a l 2 2 . 1 5 1 3 . 7 7 - - - - spectro 23 0 0 14 30 - - - -

T A B L E X I I . L E A D B A S E B A B B I T S

Sample No. Method Sb, % Sn, % Cu, %

1 c h e m i c a l 10 03 6 . 0 3 0 17 spectro 10 25 5 .95 0 . 1 4

2 c h e m i c a l 1 0 . 0 0 5 .05 - - spectro 9 . 8 5 5 .15 - -

3 c h e m i c a l 1 0 . 1 0 1 0 . 0 0 - - spectro 1 0 . 6 5 9 80 - -

a c h e m x c a l 1 3 . 8 3 1 2 . 9 3 - - spectro 14 10 13 4 0 - -

The accuracy of results of constituents present in amounts greater than 1% is greatly improved by sparking a standard sample, corresponding to the alloy under con- slderatlon, along with the sample to be analyzed. In this manner any "shifts" of working curves can be minimized.

N O T E O N T H E P R E P A R A T I O N O F S T A N D A R D S A M P L E S

The number of standard samples available from com- mercial sources did not furnish all elements and/or ranges of concentration usually reqmred for the spectrochemmal analysis of lead and tm alloys.

In order to meet this shortage a number of lead and tin alloy standards was prepared by diluting various master alloys with pure lead and/or pure tin. Carefully analyzed alloys were also used. Standards prepared in this manner are classified as reference standards and employed only in connection with commercial available standards to supplement the ranges of concentration of various ele- ments not previously covered. All standards were cast into chill cast disks. Considerable efforts were directed to in- vestigate segregation effects arising from pouring and cast- ing the &sks in connection w~th the preparation of stand- ards respectively sampling of material subject to spectro- chemical analysis. Segregation stu&es were undertaken partmularly on antlmonial leads, being the most frequently analyzed alloy in our laboratory. The alloy avadable was previously analyzed by chemical methods: Sb 6.73%, Sn 0.36%, Cu < 0 . 0 1 % , As 0.44%.

This material was remelted and treated as outlined under sampling tecbmque. A series of disks were cast at

A P P L I E D S P E C T R O S C O P Y

T A B L E X I I i . S O L D E R S

Sample No Method Sb, % Sn, % Cu, % As, %

1 S c h e m i c a l 0 . 0 5 1 4 . 9 0 0 . 0 1 0 . 0 2 spectro 0 0 4 14 63 0.006 0 0 0 7

4 S c h e m m a l 2 10 2 4 . 0 0 0 03 0 . 0 5 spectro 2 05 23 50 0 0 3 3 0 . 0 2 9

5 S c h e m i c a l 0 4 l 5 0 . 0 0 - - - - spectro 0 30 5 1 . 5 0 - - - -

6 S c h e m i c a l 0 32 50 00 - - - -

spectro 0 29 50 30 - - - - 11 c h e m i c a l 30 00 - - - - - -

spectro 30 50 - - - - - - 12 c h e m x c a l - - 3 4 . 0 0 - - - -

spectro - - - 3 3 . 7 5 - - - -

15 c h e m t c a l - - 39 50 - - - - spectro - - 3 9 . 1 0 - - - -

16 c h m m c a l 4 4 55 - - - - - -

spectro 45 O0 - - - - - - 17 c h e n u c a l - - 4 9 . 1 5 - - - -

spectro - - 4 8 2 0 - - - -

18 c h e m i c a l - - 50 0 0 - - - - spectro - - 4 9 7 0 - - - -

T A B L E X I V . T I N B A S E B A B B I T S

Sample No. Method Sb, % Cu, % Pb, %

H t 1 c h e t m c a l 7 . 0 5 7 2 0 trace spectro 6 85 6 95 0 14

N 2 c h e m i c a l 7 21 3 3 4 trace spectro 7 35 3 45 0 0 3 1

T A B L E X V . T I N B A S E S O L D E R S

Sample No. Method Sb, % As, % Pb, %

95 A c h e m m a l 5 . 0 2 0 . 0 0 8 trace spectro 5 . 1 0 0 0 0 3 0 . 0 7

95 B c h e m m a l 5 13 0 . 0 1 trace spectro 4 95 0 0 0 2 0 . 1 2

pouring temperatures ranging from 650 to 1000 ° F. By employing spectrochemlcal and statistmal methods, only minor variations in results were encountered. Next the surface of a disk (selected at random) was carefully ma- chined on a lathe by removing a layer of metal 3~ mm thick. Two exposures were obtained from each surface. This procedure was repeated tall the original disk had lost ½ of its thickness. The analytical results thus established are given in Table XVI. These statistical values indicate that segregation in anti- monial lead throughout the &sk is well within the hmits of the experimental error arising primarily from the spec- trographic equipment.

T A B L E X V I . S E G R E G A T I O N T E S T

Average Standard Coefftctcnt Element % # of determ devlatton of vartatton

S b 6 82 8 0 18 2 . 6 4 S n 0 3 6 8 0 0 1 8 5 2 0

C u 0 . 0 0 0 4 3 8 0 . 0 0 0 0 1 4 3 . 2 6 A s 0 . 4 0 8 0 0 1 9 5 00 A g 0 . 0 0 4 8 8 0 0 0 0 2 0 4 17

Submxtted March 2 0 , 1 9 5 7 ; a c c e p t e d M a y 12 , 1 9 5 7

Discussion The procedures for the spectrochemlcal analys~s of the

various types of alloys, termed white metals, presented in this paper incorporate noteworthy features: A minimum of sample preparation is required, sufficient sensitivity of detection for impurity elements, precision and accuracy of the analytical results over a wide range of concentration.

NUMBER 3, 1957 1 19

No reference is made to the spectrochemlcal analysis of a number of alloys also classified as white metals such as eutecuc fusible alloys of bismuth, cadmium, lead and tin.

Due to a small number of usable lines emitted by lead, tin, antimony, arsemc, copper, etc., the element lines re- spectively line pairs were selected on account of suitable mtensxty and reproducibility of their intensity ratios. I t became necessary to employ more than one exposure con- ditlon m order to overcome self-reversal effects observed on antimony, tin and arsenic lines in a lead matrix. On lead alloys one exposure is reqmred for recording the minor and impuri ty elements while the second exposure condi- tion is mainly used for major elements. The spark exposure of 15 seconds or multiples thereof was selected to utdlze a maximum period of excitation, being necessary for the detection of impurities and yet to keep the background in- tensity to a minimum.

The matr ix correction as applied to minor and ~m- puri ty elements should not be construed as an ideal solu- txon, but rather as a compromise.

I t is the opinion of the author that most if not all of the procedures can be applied to any other type of grat ing or prism spectrograph of medium dispersion. The selection of circuit constants of the various high voltage spark sources should be hmxted to: a) Low to moderate power, b) high inductance; 0.5-1.5 mH, c) moderate capaci-

tance, d) a low number of discharge trains; 2-4 discharge trains per cycle.

Acknowledgements

The cooperation of the staff of the chemical laboratory by providing the numerous chemical analyses is hereby gratefully acknowledged. The author is indebted to Mr. M. J. Davison, Chief Metallurgist, for his permission to publish this paper and furthermore for helpful suggestions pertaining to the metallurgy of lead and tin alloys.

Literature Cited

1. Churchill, J. R., "Techniques of quantitative spectro- graphic analysis," Ind. and Eng. Chem. Anal. Ed., 16, 653 (1944).

2. Howarth , J. and Nausester, H. K., Abstract , Apphed Spectroscopy 10, 228 (No. 4, 1956).

3. "Methods for the emission spectrochemical analysis." A.S.T.M. Oct. 1953.

4. "Spectrochemical methods", National Lead Co., Re- search Laboratories, no. 7-64, no. 67, no. 017-50, no. 033-51, no. 034-51.

5. Post, C. B., Schoffstall, D. G. and Hurley, G., "Ddu- t~on factor in spectrochemical analysis," The Carpenter Steel Co., Readmg, Pa.

Submitted March 20, 1957; accepted May 12, 1957.

Atomic Collision Processes Occurring in Gas Discharges Manfred A. Biondi

Westinghouse Research Laboratories, Pittsburgh 35, Pa.

Abstract

Some atonuc colhs~on processes wMch influence the ermsslon o:f radiation :from gas discharge sources are discussed bne:fly The processes are dlwded into the following categories ( I ) Exci tat ion of an atom by electron impact, (a) direct ( f rom the ground state) and (b) m- direct (from an excited s ta te) ; (2) Colhslons between excited atoms; (3) Trans:fer of excitation, and (4) Electron-ion recombination The relative importance o:f these reactions ~s discussed and possible conse- quences with regard to spectrochemlcal analyses are considered.

Spectrochemical analysis based upon the study of the radiation emitted from a discharge has proved an extreme- ly valuable tool in the determination of chemical composi- tions. Unfortunately, the consistent achievement of quan- ti tative rather than qualitative analyses has proved difficult in many cases. The purpose of the present paper is to dis- cuss briefly some of the atomic collision processes which play an important role in determining the radiation emit- ted from discharges and to point out their influence on the problem of quantitative spectrochemical analysis. The vari- ous atomic collision processes are discussed in greater de- tall in the references cited.

I. Production of Radiation in Discharge Sources

Under ideal circumstances the intensities of certain spectral lines (or bands) emitted from a discharge source containing the material to be analyzed bear a linear rela- tionship to the concentration of the parent element In the source over a wide range of relative concentrations of that element. In practice this condition is rarely achieved since

~Presented at the Eleventh Annual Meeting of the Society for Apphed Spectroscopy, New York, N Y., November 2, 1956.

the radiation emitted from a source is determined by a large number of competing processes.

In order to determine the amount of radiation of a given wavelength which is emitted, we must consider the modes of formation and destruction of the excited state of the atom (or molecule) which gives rise to the radiation. The xmportant processes are:

(1) Excitation of an atom by electron impact (a) Direct ( f rom the ground state) (b) Indirect (from an excited state)

(2) Collisions between excited atoms (3) Transfer of excitation (4) Electron-Ion Recombination

In the following sections simple gases will be used to il- lustrate these different processes because it has only been found possible to study the individual atomic collision processes m the simple cases. I t is to be expected, however, that one can extend these ideas to other, more complicated, cases.

II. Excitation by Electron Impact

In discussing the relative importance of the various atomic collision processes it may be helpful to keep in mind a reference atomic cross section. This cross section corresponds roughly to a typical atomic sxze: Reference atomic cross-section ~-~ wa02 ~-~ 10 -~6 cm 2, (1) where a0 is the Bohr radius. We shall compare the various colhsxon cross sections to this reference.

The excitation of atoms by electron impact is shown schematically in Figure 1. The energy levels of the atoms are shown, from the ground state, through the various ex- cited states which converge to the iomzation limit. Ener-