Atomic Absorption Spectroscopy with High Temperature Flamesanalytical.com.pe/adminanalytical/img/Atomic.pdf · 2016. 4. 28. · Atomic Absorption Spectroscopy with High Temperature

Atomic Absorption Spectroscopy with High Temperature Flames

J. B. Willis

An account is given of the history of the development of high temperature flames for the atomic absorp-tion measurement of metals forming refractory oxides. The principles governing the design of premixburners for such flames, and the relative merits of different types of nebulizer burner systems are de-scribed. After a brief account of the structure and emission characteristics of the premixed oxygen-acetylene and nitrous oxide-acetylene flames, the scope and limitations of the latter flame in chemicalanalysis are discussed.

1. Introduction

In his original paper drawing attention to the po-tentialities of the atomic absorption technique inchemical analysis, Walsh' discussed the principles of themethod without making any presuppositions about themeans used to produce the atomic vapor. It is clearfrom his paper, however, that he appreciated thesimplicity and convenience of using a flame to atomize*a solution of the analytical sample, and, in the firstpublished description of an atomic absorption spectro-photometer, 2 the sample solution was introduced intothe air supply of a Meker burner operating on air andcoal gas, with the help of a nebulizer similar to thatused in early types of Perkin-Elmer flame photometers.Even at this early stage of development of the subjectit was realized that the problem of atomization wascentral to the whole atomic absorption technique, andsince that time considerable effort has been expended inattempts to devise methods of effectively atomizing thewidest possible range of elements. Although severalother techniques have been tried, such as cathodicsputtering,3 -8 the arc heated graphite crucible,9 theinduction coupled plasma,' 0 and gas stabilized arcdischarges," use of a suitable flame still remains themethod of atomization favored by the overwhelmingmajority of workers in the field. This is hardly sur-prising since convenient and efficient nebulizers andburners have been developed for flame photometry andare commercially available. While the flame hascertain disadvantages, the equipment required is simple,

The author is with the Institute for Atomic Research, IowaState University, Ames, Iowa 50010.

Received 3 January 1968

* The word atomization is used in its exact sense to mean pro-duction of atoms rather than in its colloquial sense meaningproduction of a fine spray. The device used for spraying a solu-tion into a flame is referred to as a nebulizer.

inexpensive, easy to use, and well adapted to rapidmeasurement of a series of different solutions, so that itis unlikely that any other method of atomization willreplace it for the great bulk of analytical work.

II. Equipment

As the development of high temperature flames foruse in atomic absorption spectroscopy has requiredclose attention to the design of nebulizers and burners,it is necessary to discuss briefly the types of equipmentcurrently used to introduce the sample solution into theflame; a more detailed account may be found else-where.' 2 Nebulizer burner equipment can be dividedinto two types which appear in the literature underseveral different names.

A. Turbulent Flow Burners

The characteristic feature of this type of burner isthat the fuel gas and the combustion supporting gas arenot mixed until the point at which they enter the flame,and the solution to be nebulized is also introduced atthis point. Thus, the burner is a combination nebulizerburner; it is also known as a direct injection or totalconsumption burner, since all the liquid aspirated entersthe flame and is converted to a spray at the point ofentry. This type of burner is used in many emissionflame photometers, since it is particularly well suited tosafe operation with gas combinations having high burn-ing velocities, such as oxygen-hydrogen or oxygen-acetylene. It has been used to some extent in atomicabsorption following the early work of Robinson. 3",4

B. Laminar Flow Burners

In this type of system the combustion supporting gasaspirates the liquid sample and produces an aerosolwhich passes into a spray chamber where the largerdroplets (usually 80-90% by weight of the total) settleout. The fuel gas may be added either before or afterthe spray chamber, and the mixture of fuel gas, support

July 1968 / Vol. 7, No. 7 / APPLIED OPTICS 1295

gas, and fine droplets proceeds to a burner where it isburned. This system, also known as a premix burneror indirect nebulizer burner, was popularized by Lun-degtrdh for emission measurements using the air-acetylene flame, 5 and was introduced into atomicabsorption work by Allan'6 and David' 7 ; it is favored bymost workers in the field. The design and performanceof a typical nebulizer spray chamber system have beendiscussed recently by the author.8

A typical burner for a premixed system consists of ablock 9 2 0 or tube2 ' of stainless steel or other chemicallyresistant material containing either an array of holes ora slot at which the flame burns. An array of holesserves well for low burning velocity, low temperatureflames, such as air-propane, 2 2 but the rapid encrusta-tion of the very small holes required for high burningvelocity, high temperature flames has led to almostuniversal use of slot burners for high temperature pre-mixed flames. Water cooling is not necessary exceptwith the very hottest (oxygen-acetylene) flames.23 -25

C. Relative Merits of Turbulent Flowand Laminar Flow Burners

Both types of burner have been extensively used andtheir relative merits widely discussed.

The turbulent flow burner gives a flame which is talland narrow; while suitable for emission work it pro-vides a much shorter light path and hence poorersensitivity in absorption than does a laminar flowburner. The path length can be increased by usingseveral burners in series or by passing the light beamseveral times through the flame.2

The uptake of solution by the turbulent flow burneris usually less than by the laminar flow one, and, sinceall the aspirated solution enters the flame, the produc-tion of atoms should be more efficient. However, thereis evidence that some of the liquid aspirated in the totalconsumption type of burner passes through the flamewithout being vaporized.27 28

No quantitative comparison of the stabilities ofatomic concentrations in the flame at the two types ofburner under similar conditions seems to have beenmade, but one would expect the fluctuation level to beworse for the turbulent flow burner, if only because ofthe presence of large drops in the flame. Even theaddition of a small premixing chamber above theturbulent flow burner improves the signal-to-fluctuationratio in both emission and absorption. 2 9 Changes inthe temperature of the flame brought about by the in-troduction of the liquid sample are very much smallerin the laminar flow burner.' 0

Chemical interferences, which arise at least in partfrom incomplete vaporization of the solid particlesformed from the larger drops entering the flame, areknown to be worse with the turbulent flow than withthe laminar flow burner."

There is no doubt that the turbulent flow burner isinherently safer than the laminar flow type, since thefuel and support gases do not mix until the actual pointof entry into the flame, and thus flashback is impossible.

One drawback of the turbulent flow burner, however, isthe unpleasantly loud audible noise which it emits whilerunning; the laminar flow burner on the other hand isalmost completely silent in operation.

The author feels that for atomic absorption work, atleast, the laminar flow burner is superior, and it issignificant that almost all the commercial instrumentscurrently produced use this type.

D. Design of Laminar Flow Burners

It is not possible to go into detail here about thetheory of flame propagation, on which the whole sub-ject of burner design depends, and reference must bemade to specialized books on the subject.' 2" An ex-cellent analysis of the factors involved in designing apremix burner for high temperature flames has beengiven recently by Fiorino et al.25

The production of a stationary flame requires thatthe stream velocity of the fuel-oxidant mixture throughthe burner port be at least equal to the burning velocity.If this condition is not met, the flame is liable to flashback down the burner port and cause an explosion in thespray chamber. In practice, both to ensure a margin ofsafety and achieve a reasonably stiff flame, i.e., onewhich is not unduly disturbed by air currents in itsneighborhood, the burner is designed so that the streamvelocity is several times the burning velocity. Theflame is prevented from flashing back at the walls of theburner port, where the stream velocity is zero, by thequenching effect of the walls, and since the quenchingeffect extends some distance out from the walls, there isa limiting size of burner port below which the flame of agiven gas mixture will not flash back. Unfortunately,however, this size is usually so small for high tempera-ture flames that it is not practicable to rely on thequenching effect alone to avoid flashback.

The burning velocity of a fuel-oxidant mixture is afunction of its composition, and usually, though notalways, reaches its maximum value near the stoichio-metric ratio. Thus, it can be seen why a moderatelyfuel rich flame may burn satisfactorily at a burner butflash back when it is extinguished by the usual pro-cedure of cutting off the fuel supply. During thisprocess the flame will pass momentarily through thestoichiometric condition, where, if the gas streamvelocity is too low, flashback will occur. If the fuelgas comprises an appreciable part by volume of thetotal gas mixture, the reduction in stream velocity whenthe fuel is shut off may be sufficient to cause flashback.

The problems of designing burners for high tem-perature flames arise principally from the fact that gasmixtures having high flame temperatures usually havehigh burning velocities. This requires limitation of thesize of the burner port, with the attendant disadvantageof easier blockage by deposits of salt or carbon. If alarge burner port is desired, it is necessary to use largeflow rates of gas mixture and either design a nebulizerto operate on the whole of the oxidant or else introducepart of the oxidant as an auxiliary supply direct to thespray chamber. It may be noted also that for some gasmixtures the burning velocity increases with the tem-

1296 APPLIED OPTICS / Vol. 7, No. 7 / July 1968

perature of preheating,33 so that the burner top shouldbe kept sufficiently cool; it must in any case be keptwell below the ignition temperature of the gas mixture.

E. Optical and Electronic Instrumentation

The use of high temperature flames imposes some-what more stringent requirements on the optical andelectronic components of the atomic absorption spec-trometer than does most work with conventional lowtemperature flames.

Many of the metals for which a high temperatureflame is necessary, such as the rare earths and the mem-bers of groups IVB, VB, and VIB, have complicatedspectra, and effective isolation of the resonance lineoften calls for a monochromator with a resolution of1 A or better. Furthermore, the heat generated byhigh temperature flames is such that suitable shieldingmay be necessary to prevent thermal effects causingdrift in the wavelength setting of the monochromator.

Since emission of light by high temperature flames isconsiderable even in the uv, it is essential to use amodulated light source and an ac amplifying system;in many cases it is highly desirable that the amplifierbe tuned fairly closely to the modulation frequency ofthe source. Fortunately, hollow cathode lamps ofmost of the metals requiring use of a high temperatureflame do not show appreciable self-absorption, so thatfrequently it is possible to maintain an adequate signal-to-noise ratio merely by running the lamp at a suffi-ciently high current. For some elements, such assilicon, use of a high intensity lamp3 4 may be bene-ficial.

The signal-to-noise level of the system determines thelimit of detection, which is usually defined as theconcentration of element in solution required to producean absorption signal equivalent to twice the rms back-ground fluctuation in the zero absorption signal. Itdoes not, however, affect the sensitivity, which is con-ventionally defined as the concentration of element insolution required to produce 1 absorption with aparticular nebulizer and burner system.

111. Development of High Temperature Flames

A. Low Temperature Flames

Air-coal gas and air-propane flames satisfactorilyatomize the metals of groups IA, IB, IIB, together withgallium, indium, thallium, lead, tellurium, manganese,nickel, and palladium, but several important metalssuch as magnesium, calcium, iron, and cobalt are in-completely atomized and are subject to chemical inter-ferences, while most of the remaining metals are scarcelyatomized at all.

The introduction of the air-acetylene flame around1958 brought the list of metals showing useful sen-sitivity in absorption to about thirty-five,2 '35 but anumber of these metals were still subject to chemicalinterferences; calcium absorption, for instance, wasdepressed in the presence of phosphorus, and bothcalcium and magnesium absorption were depressed inthe presence of aluminum. That degree of atomization

is not merely a function of flame temperature wasdemonstrated by Gatehouse and Willis20 and byDavid,36 who found that some metals, such as molyb-denum and tin, showed higher absorption in a fuel richair-acetylene flame than in a hotter, more nearly stoi-chiometric one. Later it was discovered2 4 37 33 thatuse of an air-hydrogen mixture, though giving a coolerflame than air-acetylene, led to improved atomizationfor tin, arsenic, and selenium. However, even withflame conditions chosen to give the highest sensitivitychemical interferences may still be serious.

The burning velocities of oxygen-hydrogen mixturesare so high that premixed flames using these gases haveseldom been employed in flame spectroscopy, while aturbulent oxygen-hydrogen flame, of the type widelyused in emission flame photometry, may have an ef-fective temperature of less than 23000 C when aspiratingaqueous solutions.3 9 For this reason, the oxygen-hydrogen flame used by some workers in atomic ab-sorption spectroscopy must be regarded as a low tem-perature flame, and this view is substantiated by theresults published on its range of usefulness' 4 andtendency to show chemical interferences. 3 ,4 0 Inemission, however, the use of this flame with force fedorganic solutions has recently shown promising resultsin the detection of a number of metals forming re-fractory oxides.4'

Table I shows the burning velocities and calculatedand measured flame temperatures for a number of gasmixtures which have been used or considered for use inatomic absorption work.

B. The Oxygen-Cyanogen Flame

The first published attempt to use a high temperatureflame in atomic absorption spectroscopy was made byRobinson,' 4 who used a total consumption burner inwhich the sample was aspirated by premixed oxygen-cyanogen. For several metals he achieved sensitivitiesslightly better than those obtained with a turbulentflow, oxygen-hydrogen flame, but he was unable todetect any absorption for tin, tantalum, tungsten, andaluminum, while vanadium was only detectable withpoor sensitivity. Since Robinson had observed emis-sion signals from these metals at the same wave-lengths in the oxygen-cyanogen flame, it is surprisingthat he failed to detect absorption; this failure may beexplained in several ways; e.g., (1) the atoms of metalsforming refractory oxides probably exist only in highlylocalized regions of this flame, and it is possible that toosmall a portion of the light beam was intercepted bythe optimum part of the flame; (2) the excited atomsdetected in emission may be formed by some chemi-luminescent process rather than by thermal excitation,so that they are not in equilibrium with a ground statepopulation of atoms which can absorb radiation.

Although the oxygen-cyanogen flame, with its lowburning velocity and high temperature, looks promisingfor the atomization of metals forming refractory oxidesit has several drawbacks: (1) cyanogen is a highly toxicand explosive material, so that it is difficult to envisage

July 1968 / Vol. 7, No. 7 / APPLIED OPTICS 1297

Table I. Characteristics of Premixed Flames of Interest in Atomic Absorption Spectroscopy

M1aximum Maximum temperature, C

flame speed Measured byGas mixture (cm sec-) Referencea Calculated Referencea reversal Referencea

Air-coal gas ca. 55 30 1918 42

Air-propane 43 33 1925 30

Air-hydrogen 440 30 2047 44 2045 44320 43 2100 33

Air-acetylene 266 30 2250 33 2325 44170 43 2290 45 2275b 45

50% oxgyen-50% nitrogen-acetylene 640 43 2815 43

Oxygen-coal gas 2720 46

Oxygen-propane 390 43 2835 43

Oxygen-hydrogen 1190 43 2815 431120 47 2810 33

2680 46 2660 46

Oxygen-acetylene 2480 30 3257 46 3140 461130 43 3060 43

3110 48 3100 48

Oxygen-cyanogen (1:1 molar ratio) 140 30 4600 49 ca. 4500e 49270 50 4540 51 4370 51

Nitrous oxide-propane-butane ca. 250 52 ca. 2550 52

Nitrous oxide-hydrogen 390 53 ca. 2660 53 ca. 2550 55380 54 ca. 2640 54

Nitrous oxide-acetylene 160 53 ca. 29 50d 53 ca. 2700 55

Nitric oxide-hydrogen (1:1 molar ratio) 30 53 2840 53 2820 53

Nitric oxide-acetylene 87 53 3090 53 3095 53

Nitrogen dioxide-hydrogen 150 53 2660 53 1550e 53Nitrogen dioxide-acetylene 135 53

a Reference 30 contains figures from the older literature.b The flame was sheathed and the burner top was made of silver and insulated to minimize heat losses.c Temperature measurements made on the vibrational structure of the violet CN band system.d This temperature was read by interpolation from a graph in Ref.Combustion does not go to completion.

its routine use by unskilled operators; (2) owing to thepoor diffusivity of cyanogen into oxygen it is difficultto maintain a steady flame with this combination ofgases; (3) the oxygen-cyanogen flame is cooled by asmuch as 20000C when aspirating even small volumes ofliquid.6

C. The Oxygen-Acetylene Flame

In 1963, several workers published accounts ofatomic absorption measurements with fuel rich tur-bulent oxygen-acetylene flames. Fassel and Mossotti,2 6

using multiple passages of light from a xenon arcthrough the flame, obtained useful absorption at anumber of wavelengths for vanadium, titanium, nio-bium, scandium, yttrium, and rhenium, while Slavinand Manning, 7 using hollow cathode lamps as sources,demonstrated strong absorption for aluminum, vana-dium, titanium, and beryllium. They, too, used ethano-lic or isopropanolic solutions sprayed into a turbulentflame. Fassel and Mossotti's work was later extendedto demonstrate the feasibility of spectral continua asprimary sources in the determination of a large numberof metals, including many of the rare earths.5 8

Dowling et al.'9 extracted aluminum from aqueoussolutions into 2-methyl-4-pentanone and sprayed thisextract into a turbulent oxygen-acetylene flame.

53. It was probably calculated. 9 3

They showed that use of a fuel rich flame was necessaryto obtain the highest concentration of aluminum atoms,that the atoms were highly localized in the flame, andthat replacement of the organic solvent by water re-duced the absorption almost to zero.

The first published atomic absorption measurementson the lanthanide metals were made by Skogerboe andWoodriff,6 0 who used an oxygen-hydrogen flame fedwith a solution of the lanthanide salt as a line source,with an oxygen-acetylene flame as the absorbingmedium. Absorption was demonstrated in this wayfor organic solutions of europium, thulium, and ytter-bium, but the limits of detection were poor on accountof the low intensity and lack of stability of the lightemitted by the source flame.

The improvement in a signal-to-fluctuation ratio inemission achieved by the addition of a small spraychamber to the turbulent flow oxygen-acetyleneburner2 9 led to the use of this type of burner by Cowleyet al.6 ' in a study of the absorption profiles of scandium,chromium, and lanthanum in oxygen-acetylene flamesof different stoichiometry, and pointed the way to thedevelopment of fully premixed oxygen-acetylene flames.

Amos and Willis2 4 carried out a few experiments witha premixed oxygen-acetylene flame burning at a singlerow of twenty-five holes of 0.6-mm diam, but abandoned

1298 APPLIED OPTICS / Vol. 7, No. 7 / July 1968

the use of this flame in favor of the nitrous oxide-acetylene one. Fiorino et al.2 ' have described a burnerwith an 8-cm slot of adjustable width designed for usewith both premixed oxygen-acetylene and nitrousoxide-acetylene flames, and have shown that for arange of metals forming refractory oxides, fairly similarlimits of detection are observed with the two flames.However, even with this carefully designed burner,when using oxygen-acetylene mixtures, care must betaken to avoid flashback by using a narrow slot (0.25mm), by keeping a total gas flow through the burner ofat least 25 liters/min, by always keeping the mixturefuel rich, and by turning off the oxygen first whenextinguishing the flame.

D. The Oxygen-Nitrogen-Acetylene Flame

Amos and Thomas2 3 investigated the absorption ofaluminum in premixed, air-acetylene flames in whichthe air was enriched with oxygen up to a maximum of85%, using a burner consisting of a row of fourteenholes of 1-mm diam in a steel plate 3 mm thick. Ex-plosion hazards forced these workers to turn to mixturescontaining not more than about 50% oxygen burningat a burner consisting of a stainless steel block 1.25cm thick with a 3-cm X 0.45-mm slot. Later workshowed that the same technique could be applied tomany other metals forming refractory oxides.24

However, the oxygen-nitrogen-acetylene flame hassome disadvantages in routine work since it requiresthe provision of either commercial oxygen-nitrogenmixtures or fairly complex gas mixing equipment inorder to prevent the inadvertent use of mixtures toorich in oxygen. Furthermore, the length of the burnerslot seems limited to about 3 cm because of the highburning velocities of gas mixtures sufficiently rich inoxygen to give adequate atomization of metals formingrefractory oxides.

E. The Nitrous Oxide-Acetylene Flame

A more generally useful flame is that of acetyleneburning in nitrous oxide. 2 Here the burning velocityis relatively low, and the high temperature of the flameis due in part to the energy liberated by the decomposi-tion of the nitrous oxide. This flame can be safelyburned at a slot as long as 10 cm, though a 5-cm slot isfound to be optimum for most metals. Amos andWillis2 4 demonstrated that this flame could be safelyused with standard commercial instruments providinga sufficiently solid burner top was fitted and found goodsensitivities for almost all the metals which form re-fractory oxides and had proved difficult or impossibleto atomize with low temperature flames. This flamehas been adopted by the manufacturers of commercialatomic absorption spectrophotometers as the standardmethod of atomizing metals which are not satisfactorilyatomized in the air-acetylene flame, and detailed ac-counts of the performance of such instruments areavailable.63 -6 6 Its usefulness in emission work has alsobeen demonstrated. 9 4-9 6

F. The Nitric Oxide-Acetylene and NitrogenDioxide-Acetylene Flames

The lower burning velocities and higher flame tem-peratures attainable by burning acetylene with nitricoxide or nitrogen dioxide suggest that such flames mightbe valuable,6 2 but the expense and corrosive nature ofthese oxides of nitrogen militate against their use.Slavin and co-workers6 4 have found that the sensitivityfor several metals is slightly greater in the nitric oxide-acetylene than in the nitrous oxide-acetylene flame, butthe stability of the flame is so poor that the detectionlimits are usually worse than in the latter.

G. The Nitrous Oxide-Propane-Butane FlameThough this flame is not sufficiently hot to atomize

all the metals forming refractory oxides, Butler et al. 2

have shown that in the determination of the alkalineearth metals it eliminates phosphorus and aluminuminterference almost as well as does the nitrous oxide-acetylene one.

H. The Nitrous Oxide-Hydrogen Flame

The author has found" that for the alkaline earthmetals this flame gives useful sensitivities and limits ofdetection, which, however, are distinctly poorer thanthose obtained with the nitrous oxide-acetylene flame;molybdenum and aluminum are not atomized to anyappreciable extent. Ionization of calcium, strontium,and barium appears negligible in this flame.

IV. Structure and Emission Characteristics ofOxygen-Acetylene and NitrousOxide-Acetylene Flames

A. The Oxygen-Acetylene Flame

The fuel rich, premixed oxygen-acetylene flame burn-ing at a slot burner shows three distinct regions: (1)the primary reaction zone, which is usually only 1-2mm high and intensely bright; (2) the interconal zone,which is also intensely bright and difficult to dis-tinguish by eye without the use of dark glasses; and(3) the blue outer mantle.

The interconal zone is relatively rich in carboncontaining species, particularly C2, and it seems largelythese species that determine the ability of the oxygen-acetylene flame to reduce refractory oxides to free metalatoms. In conventional, near-stoichiometric flames,metals such as scandium and lanthanum are notdetectable in either emission or absorption, while in thefuel rich flame their atomic species appear only in theinterconal zone and not in the outer mantle.25

B. The Nitrous Oxide-Acetylene FlameThe premixed, nitrous oxide-acetylene flame burning

at a slot burner is also characterized by three regions:(1) the primary reaction zone, which is usually 1-3 mmhigh and white-blue in color; (2) the characteristic redinterconal zone, sometimes referred to as the red

July 1968 / Vol. 7, No. 7 / APPLIED OPTICS 1299

Table II. Sensitivities and Limits of Detection for Metals Determined in the Nitrous Oxide-Acetylene Flame

ApproximateApproximate limit ofsensitivitya detections

Metal Line, X ug/ml ' pg/ml Notes Applications

Aluminum 3092.7 1 0.1

BariumBerylliumBoronCeriumDysprosiumErbiumEuropiumGadoliniumGermaniumHafniumHolmiumLanthanumLutetiumMolybdenum

NeodymiumNiobiumOsmiumPraseodymiumRheniumSamariumScandiumSiliconStrontiumTantalumTerbiumThoriumThuliumTitaniumTungstenUraniumVanadiumYttriumYtterbiumZirconium

5535.52348.62497.752004211.74008.04594.04078.72651.63072.94103.85501.33359.63132.6

4634.23349.12909.14951.43460.54296.73911.82516.14607.32714.04326.53245.83717.93642.72551.43584.93184.04102.43988.03601.2

0.40.02

5083

0.70.90.8

171.5

141.5

42120.4

1024

1131290.62.50.06

117.5

8500.53.55

1201.51.50.2

15

0.10.0036

0.40.10.24 (at 3684 A)1

150.38

25 (at 3343.7 A)

101.550.20.1

62

0.23 (at 4008 A)

30 (at 3514 A)0.020.3 (at 4077 A)0.045

d

e,f

c ,d

cod

dc ,d

f

e.f,h

d.fd

djfh

djff

d

cd

d

Ihd

ad

f ,h,ifcif i

cdc d

h

Alloys,23 bauxites, 68 organo-aluminums, 6 9 titanium alloys, 7

cements,82 wool,9

2 rocks,97 98 polymers9 9

Rocks68

Process liquors,70 environmental materials"0

Biological materials"

Phosphate rock," yttria84Phosphate rock,7 yttria84Phosphate rock,71 yttria 8 4Phosphate rock"

Phosphate rock1

Phosphate rock"

Steels,72 niobium and tantalum,73 indirect determination ofsilicon0 and phosphorus8'

Phosphate rock,7 yttria84

Phosphate rock"

Phosphate rock7l

Bauxite, 6 8 steels,74 cenment,8 ' rocks97Rocks,68 cement 8 '

Phosphate rock,7 yttria84

Steels,87' nickel-base alloys,' 6 bauxite, 6 8 cement, 8 ' polymers98

Oils,6878 steels,78 titanium alloys77

Phosphate rock"Phosphate rock"Thorium oxide samples,79 indirect determination of fluo-

ride's and ammonia," aluminum alloys1"'

a Concentration giving 1% absorption when sprayed as an aqueous solution into a 5-cm premixed flame. Data from Ref. 12.b Concentration which when sprayed as an aqueous solution into a 5-cm premixed flame in a Perkin-Elmer 303 instrument gives an

absorption equal to twice the noise level of the background. Data from Refs. 63 and 65.c Slightly fuel rich flame.d Ionization suppressed with 1000-,ug/ml potassium.I Absorbance-concentration curve flattens markedly at high concentrations.f Fuel rich flame.a Heated spray chamber used.4 Measured in the presence of 2% hydrofluoric acid.

High brightness, hollow cathode lamp used.Absorbance markedly dependent on spectral slit width.

feather, which varies in height from 0 mm to 20 mm or30 mm depending on the acetylene/nitrous oxide ratio;and (3) the blue outer mantle.

Kirkbright et al.'7 have measured the radiation fromvarious parts of the flame and have shown that the redinterconal zone, which is generally used in atomic ab-sorption work, is characterized by strong CN and NHemission. When enough acetylene is added, the redinterconal zone tends to be masked by the yellow

luminosity, extending throughout the flame, whicharises from incandescent carbon particles. Theseauthors suggest that the reactions occurring in thisflame to produce metal atoms from refractory oxides aremore complex than the simple reducing action of atomiccarbon or incandescent carbon particles which has beeninvoked by various authors to explain the production ofmetal atoms in the oxygen-acetylene flame.

1300 APPLIED OPTICS / Vol. 7, No. 7 / July 1968

1.0

0-80

N

0

0

0;= 0-4U

0-2T.2200°C

0-

- Alkali metals2StS2-'.5SAlkali earth metals S,BoronAlumni.m etc.

2P,,,2 ,2F

Total met.1

P.R.1lO6atMa

50 C\

\~

N \

4 5 6 7

B IONIZATION POTENTIAL, VOLTSI I I t I I I ICs Rb K Na Li Sr AICa Mg

Fig. 1. Degree of ionization, calculated by thifor various metals at temperatures of 22000 C

suming a total metal partial pressure of

V. Use of High Temperature Flamesin Chemical Analysis

A. Range of Application of HighTemperature Flames

Of the high temperature flames mentioinitrous oxide-acetylene flame has beenstudied and applied in practical analyticTable II shows the approximate sensitiviof detection for thirtv-five metals for whthis flame is recommended. The table iswork of Amos and Willis24 and of Mvannipapers reference may be made for a detaiof particular metals. The table also incluto published applications of high temperapractical analysis.

B. Comparative Sensitivities of Nitrou!Oxide-Acetylene and Oxygen-Acetyler

Several workers8"-"5 have compared solimits of detection for various metalsflames, the usual procedure being to comixed nitrous oxide-acetylene flame Istandard 5-cm X 0.5-mm slot burner witflow oxygen-acetylene flame at a totalburner. In every case the longer light Iby use of the slot burner leads to the higher sensitivities with the nitrous onflame, though in some instances it is p4prove the sensitivity of the oxygen-acet,match that of the nitrous oxide-acetylusing several burners in series and by replwith organic solutions. In general, tllevel of the turbulent flow burner is appri

;O than that of the premixed system, so that limits of)_JS 2s2 detection are likely to be worse for the former.8 4

13/2 - ISO The enhancement of absorption achieved by replac-zS. ing aqueous solutions with solutions in organic solvents

is usually greater in the turbulent flow than in thelaminar flow flame, though close comparisons haveseldom been made. Ramakrishna et al."5 found a six-fold enhancement of beryllium absorption in a turbulentflow oxygen-acetylene flame, but only a 1.25-fold en-hancement in a laminar flow nitrous oxide-acetyleneflame on replacement of water with a 5-10% solution ofdiethyleneglycol diethyl ether. Such enhancementsare not due to the greater rate of supply of the solutionto the flame, as the rate of aspiration of the organic solu-tion is actually less than of the aqueous one, at least inthe case just cited, but must be due to the finer dropletsize and improved atomization efficiency achieved byusing solutions in organic solvents. The droplet size

8 9 is more critically dependent on solvent properties withI , the total consumption than with the laminar flow

B Be burner, though even with the latter atomization ef-e Saha equation, ficiency is frequently very sensitive to changes inand 2950'C, as aspiration rate, 8

,4 and use of an adjustable nebulizer is

1'0 atm. highly desirable.Fiorino et al."5 have made a more accurate com-

parison of the two flames, both of them premixed andburning at the same burner. The detection limits formost of the twelve refractory oxide forming metalsstudied by these authors were fairly similar in the twoflames, though those for lanthanum and uranium wereappreciably better with oxygen-acetylene.

ned above, the C. Ionization Effects in High Temperature Flamesmost widely Even a low temperature flame such as air-propane

Bal situations. possesses enough energy to ionize an appreciable frac-ties and limits tion of the atoms of the alkali metals, and the degree ofich the use of ionization rises rapidly with increase in flame tempera-based on the ture, so that in the oxygen-acetylene and nitrous oxide-

ng,'3 to whose acetylene flames, not only the alkali metals but manyiled discussion others may be ionized to a considerable extent. The

des re ferences degree of ionization, -which is calculated by the Sahature flames in equation, depends among other things on the concentra-

tion of metal atoms in the flame, and is greatest at lowS concentrations, as discussed elsewhere."s,"s Figure 1ie Flames shows the calculated degrees of ionization for different

metals at two temperatures, assuming a partial pressureensitivities or of metal of 10-' atm.in these two The ionization of an atom Ill to its positive ion M+mpare a pre- and a free electron e- is assumed to behave as a simpleurning at a dissociation process governed by the law of mass action,

a turbulent and it is found that the ionization of a metal in the flameconsumption can be suppressed by the presence of a sufficiently highath achieved concentration of electrons, which may be provided byattainment of the addition of a salt of a readily ionized metal such as-ide-acetylene cesium or potassium. From the change in absorbancessible to im- of the neutral atomic species when this is done, the

4lene flame to degree of ionization of the metal can often be approx-ene flame by imately determined.acing aqueous Table III shows calculated and measured degrees ofLe fluctuation ionization for a number of metals in the nitrous oxide-eciably higher acetylene flame. The agreement between these two

July 1968 / Vol. 7, No. 7 / APPLIED OPTICS 1301

-

Table Ill. Calculated and Measured Degrees ofMetals in the Nitrous Oxide-Acetylene Fl

Degree of ionization

Measured iiCalculated for nitrous-T = 2950'C oxide-

Ionization PP = 10-' acetyleneMetal potential, eV atm flame

Na 5.14 0.82K 4.34 0.9SBe 9.32 0.00 0.00Mg 7.64 0.03 0.06Ca 6.11 0.43 0.43

0.3SSr 5.69 0.71 0.84Ba 5.21 0.92 0.88

0.92Al 5.98 0.17 ca.0.14Eu 5.67 0.61 ca. 0.60Yb 6.2 0.36 0.20

sets of values is surprisingly good, sincetemperature of the flame nor the partialmetal atoms in it is known with certainty; fthe addition of potassium chloride may alsefficiency of vaporization of the metallic flame. The assumed flame temperature ntoo high by 200-300aC and the partial presstoo high by a factor of 10-100, so thatagreement between calculation and expeiarise merely through a compensation of opp(

Ionization interference is overcome wherin atomic absorption work by the additiondards of the same concentration of interferas in the sample solution. Alternatively, aof the interfering element may be added to es

D. Chemical Interferences in HighTemperature Flames

Chemical interferences occur in atomic abin flame emission work, through the chemiction of the element to be determined with otlor compounds present in the flame. Welamples occurring in low temperature flames bination of calcium with phosphorus and tltion of calcium and of magnesium with alumrefractory compounds formed cannot becompletely at the temperature of the flanlead to partial or complete removal of metalthe flame.

The mechanism of chemical interferenceunderstood in all cases, but interference isin hot flames and generally decreases in theof the flame."1 It is also dependent on droileast for the finest sprays, so that direct injiers show higher levels of interference thaisystems."l

One of the pleasing features of the premiLacetylene and nitrous oxide-acetylene flairelative freedom from some of the chemical i

Ionization of found in flames of lower temperature. For instance,lame Amos ad Willis24 found that the presence of a rel-

atively large concentration of phosphoric acid, which inthe air-acetylene flame strongly suppresses calcium ab-sorption, actually enhanced it slightly in the nitrousoxide-acetylene flame. They attributed much of thisenhancement to reduction in the degree of ionization of

Reference calcium brought about by the presence of phosphorus,as when the ionization was suppressed by the addition of

55 excess potassium chloride the enhancement effect all55 but disappeared. However, Manning and Capacho-24 Delgado demonstrated that the absorption at the wave-24 length of the calcium ion was unaffected by the addition24 Z86 of phosphoric acid, so that the enhancement of absorp-24 tion of neutral calcium atoms must be attributed to in-24 creased efficiency in the dissociation of the calcium salt86 in the presence of phosphoric acid." Becker and Fas-55 sel,'7 using the slot burner described by Fiorino et al.2 '55 have been unable to detect either an enhancing or a24 depressing effect of phosphorus on calcium, under the

conditions of concentration used by Amos and Willis, ineither the nitrous oxide-acetylene or oxygen-acetylene

neither the flame.pressure of The almost complete suppression of magnesium ab-urthermore, sorption by low concentrations of aluminum in the air-o affect the acetylene flame is overcome by use of the nitrous oxide-salts in the acetylene flame; in fact, Fleming" has shown that thisiay well be interference can be eliminated in a flame of acetyleneure of metal burning with a 1:3 nitrous oxide-air mixture.

the happy The absorption of calcium, strontium, and barium is'iment may completely suppressed by low concentrations of alumi-)sing errors. num in the air-acetylene flame; this interference is notre necessary entirely removed, but is reduced to manageable levels,to the stan- in the nitrous oxide-acetylene flame."ing elementlarge excess E. Interelement Enhancement Effects inach." High Temperature Flames

Several interesting enhancement effects have beendiscovered recently when using the nitrous oxide-acety-lene flame for chemical analysis, e.g., the appreciable

sorption, as enhancements found for aluminum and beryllium inal combina- the presence of acetic acid8" and for vanadium7 8 andier elements titanium" in the presence of sulfuric acid. ProbablyI known ex- the most striking is the marked enhancement of the ab-are the com- sorption of titanium, tantalum, hafnium, and zirconiumie combina- in the presence of small concentrations of hydrofluoricinum. The acid.2 4 The magnitude of this effect is such that it isdecomposed possible to determine fluoride ion down to a concentra-le and thus tion of 10-3 l by making use of the enhancement withatoms from zirconium." It is probable that the intermediate for-

mation of volatile oxyfluorides such as ZrOF2 increasesis not fully the efficiency of conversion of the very refractory oxide

always least to the metal in these cases. An even stranger effect isupper parts the enhancement of zirconium absorption caused by thep size, being presence of ammonia and of certain amines."0action burn- A few examples of intermetal enhancement have beeni do premix observed. Robinson, West, and co-workers' 3"' found

that the absorption of vanadium in the nitrous oxide-xed oxygen- acetylene flame was increased in the presence of alu-nes is their minum, reaching a constant value at a molar ratio oflnterferences about 1:1. Titanium was found to enhance the absorp-

1302 APPLIED OPTICS / Vol. 7, No. 7 / July 1968

tion of both vanadium and aluminum in much the sameway. These authors suggested that the equilibrium:V + 0 > VO, which exists in the flame, may be shiftedto the left by aluminum or titanium competing withvanadium for combination with the available oxygen.Taken in conjunction with the observation that vana-dium does not enhance the absorption of aluminum,these results suggested that the stabilities of the metaloxides are in the order Ti > Al > V. The fact thattitanium absorption is enhanced in the presence ofiron,8 whose oxide is certainly less stable than that oftitanium, 9 ' suggests that the full explanation of suchenhancements is more complicated than the one givenby Robinson, West, and their co-workers.

VI. Conclusions

The development of high temperature flames for usein atomic absorption and emission work has not onlyprovided a powerful technique for the solution of prob-blems in analytical chemistry but has also led to a morecritical consideration of the factors determining theperformance of flames as atomizing systems. This, inturn, has given rise to a number of fundamental studieson the structure and spectroscopy of flames and to theopening up of a fascinating field of research in high tem-perature chemistry.

This work, which was performed in the Ames Lab-oratory of the U. S. Atomic Energy Commission, rep-resents Contribution No. 2241. Its author is on leavefrom the Division of Chemical Physics, CommonwealthScientific and Industrial Research Organization, Mel-bourne, Australia.

References1. A. Walsh, Spectrochim. Acta 7, 108 (1955).2. B. J. Russell, J. P. Shelton, and A. Walsh, Spectrochim.

Acta 8, 317 (1957).3. B. J. Russell and A. Walsh, Spectrochim. Acta 15, 883

(1959).4. B. M. Gatehouse and A. Walsh, Spectrochim. Acta 16, 602

(1960).5. A. Walsh, Proceedings Tenthi Colloquiun Spectroscopicum

Internationale, E. R. Lippincott and M. Margoshes, Eds.(Spartan Books, Washington, 1963).

6. J. A. Goleb and J. K. Brody, Anal. Chim. Acta 28, 457(1963).

7. J. A. Goleb and Y. Yokoyama, Anal. Chim. Acta 30, 213(1964).

8. J. A. Goleb, Anal. Chim. Acta 34, 135 (1966).9. B. V. L'vov, Spectrochim. Acta 17, 761 (1961).

10. R. H. Wendt and V. A. Fassel, Anal. Chem. 38, 337 (1966).11. K. E. Friend and A. J. Diefenderfer, Anal. Chem. 38, 1763

(1966).12. J. B. Willis, in Analytical Flame Spectroscopy, R. Mavro-

dineanu, Ed. (N. V. Philips, Eindhoven, 1968).13. J. W. Robinson, Anal. Chim. Acta 23, 458 (1960).14. J. W. Robinson, Anal. Chem. 33, 1067, 1226 (1961).15. H. LundegArdh, Die quantitative Spektralanalyse der Ele-

mente (Fischer, Jena) Part I, 1929; Part II, 1934.16. J. E. Allan, Analyst 83, 466 (1958).17. D. J. David, Analyst 83, 655 (1958).18. J. B. Willis, Spectrochim. Acta 23A, 811 (1967).19. 0. E. Clinton, Spectrochim. Acta 16, 985 (1960).

20. B. M. Gatehouse and J. B. Willis, Spectrochim. Acta 17, 710(1961).

21. J. B. Willis, Anal. Chem. 34, 614 (1962).22. V. C. 0. SchUler and A. V. Jansen, J. S. African Inst. Mining

Met. 790 (1962).23. M. D. Amos and P. E. Thomas, Anal. Chim. Acta 32, 139

(1965).24. M. D. Amos and J. B. Willis, Spectrochim. Acta 22, 1325,

2128 (1966).25. J. A. Fiorino, R. N. Kniseley, and V. A. Fassel, Spectro-

chim. Acta, to be published in Part B, 196826. V. A. Fassel and V. G. Mossotti, Anal. Chem. 35, 252 (1963).27. J. D. Winefordner, C. T. Mansfield, and T. J. Vickers,

Anal. Chem. 35, 1607 (1963).28. R. Pschel, L. Simon, and R. Herrmann, Optik 21, 441

(1964).29. R. N. Kniseley, A. P. D'Silva, and V. A. Fassel, Anal.

Chem. 35, 910 (1963); 36, 1287 (1964).30. R. Mavrodineanu and H. Boiteux, Flame Spectroscopy

(John Wiley & Sons, Inc., New York, 1965).31. W. Slavin, Perkin-Elmer Atomic Absorption Newsletter 6,

9 (1967).32. B. Lewis and G. von Elbe, Combustion, Flames, and Explo-

sions of Gases (Academic Press Inc., New York, 1961).33. A. G. Gaydon and H. G. Wolfhard, Flames, Their Structure,

Radiation, and Temperature (Chapman and Hall Ltd., Lon-don, 1960).

34. J. V. Sullivan and A. Walsh, Spectrochim. Acta 21, 721(1965).

35. J. E. Allan, Spectrochim. Acta 18, 259 (1962).36. D. J. David, Nature 187, 1109 (1960); Analyst 86, 730

(1961).37. J. E. Allan, paper presented at Fourth Australian Spectros-

copy Conference, Canberra, 1963; a summary of theConference appeared in Nature 200, 319 (1963).

38. L. Capacho-Delgado and D. C. Manning, Spectrochim.Acta 22, 1505 (1966).

39. W. H. Foster and D. N. Hume, Anal. Chem. 31, 2028, 2033(1959).

40. P. B. Adams and W. 0. Passmore, Anal. Chem. 38, 630(1966).

41. R. K. Skogerboe, A. T. Heybey, and G. H. Morrison, Anal.Chem. 38, 1821 (1966).

42. B. Lewis, H. Seaman, and G. W. Jones, J. Franklin Inst.215, 149 (1933).

43. E. Bartholom6, Z. Elektrochem. 53, 191 (1949); 54, 169(1950).

44. G. W. Jones, B. Lewis, and H. Seaman, J. Amer. Chem. Soc.53, 3992 (1931).

45. W. Snelleman, A Flame as a Standard of Temperature(Bronder-offset, Rotterdam, 1965), (Thesis, Utrecht, 1965).

46. H. H. Lurie and G. W. Sherman, Ind. Eng. Chem. 25, 404(1933).

47. P. Gray, R. Mackinven, and D. B. Smith, Combustion andFlame 11, 109 (1967).

48. F. Henning and C. Tingwaldt, Z. Physik 48, 805 (1928).49. N. Thomas, A. G. Gaydon, and L. Brewer, J. Chem. Phys.

20, 369 (1952).50. J. B. Conway and A. V. Grosse, J. Amer. Chem. Soc. 80,

2972 (1958).51. J. B. Conway, R. H. Wilson, Jr., and A. V. Grosse, J.

Amer. Chem. Soc. 75, 499 (1953).52. L. R. P. Butler, L. H. Ahrens, and A. Strasheim, Proceedings

Thirteenth Colloquim Spectroscopium Internationale (Hilger,London, 1968), and personal communication.

53. W. G. Parker and H. G. Wolfhard, Proceedings FourthInternational Symposium on Combustion (Williams and Wil-

July 1968 / Vol. 7, No. 7 / APPLIED OPTICS 1303

kins, Baltimore, Maryland, 1952), p. 420; ProceedingsFifth International Symposium on Comibustion (ReinholdPublishing Corporation, New York, 1954), p. 718.

54. J. W. Armitage and P. Gray, Combustion and Flame 9, 173(1965).

55. J. B. Willis, unpublished work; J. B. Willis, J. 0. Rasmuson,R. N. Kniseley, and V. A. Fassel, Spectrochim. Acta, to bepublished.

56. M. R. Baker and B. L. Vallee, Anal. Chem. 31, 2036 (1959).57. W. Slavin and D. C. Manning, Anal. Chem. 35, 253 (1963).58. V. A. Fassel, V. G. Mossotti, W. E. L. Grossman, and R. N.

Kniseley, Spectrochim. Acta 22, 347 (1966).59. F. B. Dowling, C. L. Chakrabarti, and G. R. Lyles, Anal.

Chim. Acta 28, 392; 29, 489 (1963).60. R. K. Skogerboe and R. A. Woodriff, Anal. Chem. 35, 1977

(1963).61. T. G. Cowley, V. A. Fassel, and R. N. Kniseley, Institute

for Atomic Research, Iowa State University, unpublishedwork.

62. J. B. Willis, Nature 207, 715 (1965).63. D. C. Manning, Perkin-Elmer Atomic Absorption News-

letter 4, 267 (1965); 5, 63, 127 (1966); 6, 35 (1967).64. W. Slavin, A. Venghiattis, and D. C. Manning, Perkin-

Elmer Atomic Absorption Newsletter 5, 84 (1966).65. W. Slavin, Appl. Spectrosc. 20, 281 (1966).66. J. Ramfrez-Muhoz and ME. Roth, Flame Notes (Beck-

man) 2, 18 (1967).67. G. F. Kirkbright, M. K. Peters, and T. S. West, Talanta 14,

789 (1967).68. J. A. Bowman and J. B. Willis, Anal. Chem. 39, 1210

(1967).69. J. R. Deily, Perkin-Elmer Atomic Absorption Newsletter 5,

119 (1966).70. D. C. Bowditch, Australian Mineral Development Labora-

tories Bulletin, April 1966, p. 19.71. J. Kirinunen and 0. Lindsj6, Chemist-Analyst 56, 25, 76,

(1967).72. G. F. Kirkbright, A. M. Smith, and T. S. West, Analyst

91, 700 (1966).73. G. F. Kirkbright, M. K. Peters, and T. S. West, Analyst 91,

705 (1966).74. J. J. McAuliffe, Perkin-Elmer Atomic Absorption News-

letter 6, 69 (1967).75. J. B. Headridge and D. P. Hubbard, Anal. Chim. Acta 37,

151 (1967).76. R. A. Mostyn and A. F. Cunningham, Perkin-Elmer Atomic

Absorption Newsletter 6, 86 (1967).

77. D. Myers, Perkin-Elmer Atomic Absorption Newsletter 6,89 (1967).

78. L. Capacho-Delgado and D. C. Manning, Perkin-ElmerAtomic Absorption Newsletter 5, 1 (1966).

79. J. B. Tyler, Perkin-Elmer Atomic Absorption Newsletter6, 14 (1967).

80. T. Kumumaru, Y. Otari, and Y. Yamamoto, Bull. Soc.Chem. Japan 40, 429 (1967).

81. G. F. Kirkbright, A. M. Smith, and T. S. West, Analyst 92,411 (1967).

82. L. Capacho-Delgado and D. C. Manning, Analyst 92, 553(1967).

83. S. L. Sachdev, J. W. Robinson, and P. W. West, Anal.Chim. Acta 37, 12, 156 (1967).

84. R. J. Jaworowski, R. P. Weberling, and D. J. Bracco, Anal.Chim. Acta 37, 284 (1967).

85. T. V. Ramakrishna, P. W. West, and J. W. Robinson, Anal.Chim. Acta 39, 81 (1967).

86. D. C. Manning and L. Capacho-Delgado, Anal. Chim. Acta36, 312 (1966).

87. D. A. Becker and V. A. Fassel, Institute for Atomic Re-search, Iowa State University, personal communication.

88. H. D. Fleming, Spectrochim. Acta 23B, 207 (1967).89. A. M. Bond and T. A. O'Donnell, Anal. Chem., 40, 560

(1968).90. A. M1. Bond and J. B. Willis, unpublished work.91. T. L. Cottrell, The Strength of Chemical Bonds (Butter-

worths Scientific Publications Ltd., London, 1958).92. F. R. Hartley and A. S. Inglis, Analyst 92, 622 (1967).93. H. G. Wolfhard, Institute for Defense Analyses, Arlington,

Va., personal communication.94. E. E. Pickett and S. R. Koirtyohann, Spectrochim. Acta

23B, 235 (1968).95. G. F. Kirkbright, A. Semb, and T. S. West, Spectrosc. Lett.

1, 97 (1968).96. V. A. Fassel, Spectrosc. Lett. 1, 95 (1968).

97. A. Katz, Amer. Mineral. 53, 283 (1968).98. J. C. Van Loon, Perkin-Elmer Atomic Absorption News-

letter 7, 3 (1968).99. D. Druckman, Perkin-Elmer Atomic Absorption Newsletter

6, 113 (1967).100. D. L. Bokowski, Perkin-Elmer Atomic Absorption News-

letter 6, 97 (1967).101. H. Bader and H. Brandenberger, Perkin-Elmer Atomic

Absorption Newsletter 7, 1 (1968).102. L. Wilson, Anal. Chim. Acta 40, 503 (1968).

Cover

The illustration is taken from a book published in 1898 by JohnTyndall, [Six Lectures on Light (D. Appleton Co., New York,1898)]. The distinguished British physicist was describing theclassical experiments by Kirchhoff in the 1860's leading toKirchhoff's theory on the nature of solar energy. Light fromthe sun enters the laboratory through a window and is focusedby lens L onto a screen SS. In passing through the prism P,the light is dispersed into the spectrum and the familiar Frauni-hofer spectrum of the sun is observed. Introduction of a sodiumsalt in the Bunsen flame provides an atomic vapor of sodium and

greatly intensifies the sodium absorption lines at D.

1304 APPLIED OPTICS / Vol. 7, No. 7 / July 1968