Comparison of line and continuous sources in atomic absorption spectrophotometry

ANALYTICA

436

COlMI’ARISON OF LINE AND CONTINUOUS SORPTION SI’EC’TROPHOTOMETIZY

SOURCES

IN

CHIMICA

ATOMIC

ACTA

AB-

Several years after tlie introduction of atomic absorption spcctrophotometry GI~OSS~IAN hNI> COox
In order to indicate the similarities and the differences in the use of a continuous source and a line source in atomic absorption spcctrometry, it is assumed that the same basic equipment consisting of a flame cell, a monochromator and an electrometer-reaclout system is used in both cases. Consequently, the instrumental proportionality factor, K, relating rcaclout-voltage to intensity of radiation reaching the entrance slit of the monochromator, will be the same in both cases. A monochromator is used capable of isolating a single line but not capable of resolving the spectral line profile. This means that the spectral bandwidth of the monochromator, s, is considerably larger than the absorption line width, A&. In turn, the absorption line width is assumed to be larger than the width of the line emitted by the line source, AAs.

The photodetector signal due to radiation emitted by the lint source passing through the flame gases with only blank solution being aspirated is given by rOn

lcavc

from University

of Amstcrclnm,

The Ncthcrlnncls.

LINE

AND

CONTINUOUS

SOURCES

IN

A.A.S.

437

(1) whcreJ

is the intensity per unit of wavelength. The signal from the radiation passing through being aspirated is given by r1 = K

s

AAsJp esp( - k&)

the flame with sample solution

d?.

wliere I, is the path length in the flame. In general, the atomic absorption coefficient, firl, is a comples \ViLVC!hlgt~l and is described by the following set of formulae’ : ka = ko fi(n,v) k(J =

function of the (3)

v+rln 2 e”A02Nf nt c2 AAn

2.0 8 IZT In 2 A?,,, = c d --h!!

where e and VIZare charge and mass of the electron, c the velocity of light, A0 the wavelength at the line center, / the oscillator strength, N the concentration of absorbing atoms in the flarnc, T the flame temperature, IZ the gas constant, M the atomic mass, A& the Doppler half-width of the absorption line, a the damping constant (given by a = (In 2)~Ak/ALn), A& the collisional half-width of the..absorption line, and v is equal to (41n 2)* (A - ilo)/A&. The function 6(a,v) describes the variation of kA o&er the entire abso~~tio~t line. However, the integration in eqn. (2) extends over the emission line width, Ajls, only, which is generally much smaller than the absorption line width. This restricts the range of v to an upper limit of about A&/A&. From the data presented in Table I, it is seen that the relative decrease of k~ with increasing value of v is rather small if the damping constant, u, exceeds a value of 0.5 (see below). In that case, cqn. (2) can be approsimated as r1 = Ke-rL

I

aAsJaodl

=

z,Oe-EL

where E = kob(a,fl)

(5)

represents the average absorption coefficient over the source line width, Ails, and r7is an average over the range from zero to ALalA&\. Consequently, the absorbance is given by A 1 = In(lr”/ll)

= 15L,

and for very small concentrations,

(9) when EL G

I,

the fraction of the intensity absorbed .% Anal.

Chim. Acta,

37 (1967) 43G-444

L. ])I‘: GALAN, 1%‘.W. MCGEE, J. I). WINEFORDNER

438

in tlic flame is given by

The minimum dctcctable atomic concentration, N,,,I,,, is defined ;LI; the concentration that produces a signal equal to twice tllc noise. When a line ~uurcc is used in atomic absorption spectrometry, tlie main contribution to the noise is the variation in the source intensity, x. According to ~VrNr3PoIIrjNRR._ANI) VICKERS”, the detection limit is then found from the relation n t(Nn,l,,) = 2x1 I/KY

(11)

whcrc Af is tllc frequency b;lnclwiclth of tile elcctrometcr-readout of eqn. (IO) into ccln. (II) yieltls

system. Substitution

With a line source, the wavelength region of interest is determined by tllc width of the source line, wlicreas the decisive parameter in the case of a continuous source is the spectral bandwidth of the monochromatoi-, S; over this small range, the tlic l~lanlc photodetector intensity of the source is essentially constant. Tlxxefore, signal is given by (13) and for tlic photodetector signal clue to radiation sample solution being aspirated, we have

passing

through

the flame with

(X4) where the integration limits are o and co because the spectral bandwidtll is assumed to be much larger than the width of the absorption line. The fraction of the intensity absorbed in the flame is given by . - csp (-/%Ji.)cll m (I zg - rc s A ‘1 Lxc =

IcO

=O

s

-=-

s

(IS)

where tin is known as the total absorption 7. Curves representing AT as a function of the concentration N (curves of growtll) depend on the value of the damping constant, ~0~10.At very high concentrations, .4~ is proportional to N*; at very low concentrations, where AAL 4 I at any position in the absorption line, eqn. (15) yields

ac

=

s,”

kaL s

which is independent

CIA = ~&AC,” Nfr,

(16)

1nc”s

of the absorption

line profile.

LINE

AND

CONTINUOUS

SOURCES

The general expression A,

= ln(lcO/lc)

=

-

IN A.A.S.

439

for the absorbance, A, is in this case In (x -

AT/~)

(17)

An expression for the minimum detectable concentration is found in the same way as for the line source, if again the variation in source intensity is assumed to constitute the main contribution to the noise n’mln,c =

2 wac”qc

where tile frcqucncy

Wovking

vu (18)

7ce"l.02fL

bandwidtll,

Af, is the same as in eqn.

(12).

cwves

Working curves in absorption spectrometry are usually presented as a plot of absorbance, A, VHS~IS the solution concentration of the element, C. It follows from eqns. (9) and (IO) that if a line source is used in atomic absorption spectrometry, such a plot should produce a straight line as long as the concentration of an element in the flame is proportional to its concentration in solution. When a very narrow line source is used (i.e. Ails 4 AL, G x o), this straight line should extend over an estrcmely large concentration range ; namely, up to the point where the resonance broadening causes the damping constant to change’, wllicli is in the region of molar solutions. In practice, however, the linear range of the working curve is restricted by chemical interferences in the flame (ionization and compound formation of the atom in concern) and by a nonlinear relation between flame gas concentration, N, and solution concentration, C, due to variation in aspiration efficiency 11. Sloping off of the working curve occurs also if the source lint width is not negligible compared to the absorption line width, since in that case the approsimation of eqn. (7) is no longer valid. If a continuous source is used in atomic absorption spectrometry, the absorbance is a complicated function of the concentration of the element in the flame. According to cqn. (17). the absorbance is proportional to the total absorption, A T, only for small values of the ratio A T/S. On the other hand, the fraction of the intensity absorbed in the flame, cyc, is always proportional to the total absorption. Even in this case, however, the linear portion of the working curve is restricted to the linear portion of the curve of growth (A T veYszfs N) 0.10. The examples presented in Fig. I indicate that the working curves slope off when more than 10% of the radiation is absorbed in the flame. At lower values of the relative absorption linear expansion of cqn. (17) is perfectly valid. Therefore, a similar linear range is obtained for a plot of either absorbance or relative absorption VCYSZG concentration. The working curves shown in Fig. I for a line source extend linearly over an additional decade towards higher concentrations, Therefore, curvature of the working curves obtained with a continuous source constitutes no disadvantage, if with proper scale expansion, atomic absorption spectrometry is used for the determination of low concentrations. Comparison of signals If the same low atomic

concentration

is measured

with

a line source and a

Anal. Claim. Ada.

37

(1967)

436444

L.

440

I)E

GALA?,

W.

W.

hlcGEE,

J.

D.

WINEFORDNER

continuous source, respectively, the ratio of the fractions of the intensity (or the ratio of the absorbances) is found from eqns. (IO) and (IG) to be 1<*=---_=

Ocl

ac

qln

I-

absorbed

2 s c(i(a,z?)

3z

(19)

Ah

The decisive factor in this equation is the ratio oftire spectral bandwidth and the Doppler half-width of the absorption line. For an element of atomic weight 50, the Doppler half-width at zooo”I< varies from 0.01 A at 2000 A to 0.04 A at 8000 A. A

0.001

0.01

70

A.a 1

I/

O.OOloo,

I

0.1

I

1 Cin ppm

10

I

100

Fig. I. Working curve in atomic rrbsorption spcctrophotomctry. l Absorbance vcr~zrS con(continuous source) ; ccntration (COnti~lu0119 source) ; 0 fraction ~bsorbctl ve*szls conccntrntion A absorbnnco v~s16s concentration (lint source).

spectral bandwidth of the same order of magnitude can be obtained only with a highresolution monochromator. For example, a large grating monochromator with second order reciprocal linear dispersion of 2 A/mm and a minimum slit width of I~ ,u yields a spectral bandwidth of 0.02 A. In this case, the signal obtained with a continuous source is generally superior to that ‘of ‘the line source, since the factor t3(n,C) is smaller than one (see below). If, on the other hand, the minimum spectral bandwidth is several A, application of a continuous source should not be attempted. Although with appropriate scale A7rnl. Chim. Actn, 37 (1967) 436-444

LINE

AND

CONTINUOUS

SOURCES

IN A.A.S.

441

expansion fairly low concentrations can still be detected, eqn. (xg) clearly shows that a gain of about two orders of magnitude can be obtained with a line source. However, a better resolution than is offered by a low-dispersion monochromator is often desirable in atomic absorption spectrophotometry, The low intensity of many hollow-cathode discharge tubes necessitates the use of fairly large slit widths in order to dctcct the radiation at all (the photodetector signal is proportional to the slit widths). In determining elements with complicated spectra, a spectral bandwidth smaller than I A is often needed for complete isolation of the analysis line. Improved sensitivity and better linearity of the working curve with decreasing spectral bandwidth have been reported in the analysis of leadi”-, which element possesses a relatively simple spectrum, For this reason, it is unfortunate that low-dispersion monochromators are often used in commercial atomic absorption instruments. A medium-dispersion monochromator, such as the 0.5-m Jarrcll-Ash grating monochromator, is capable ‘of a minimum spectral bandwidth of 0.2 A. This is sufficient to isolate even close lying lines. Now, the advantage of using a line source is reduced to about a factor of IO, even if &a,r7) has its maximum value of unity. The actual value of b(a,v’) is smaller than unity because of the finite width of the source line and because of collisional broadening of the absorption line. Measurelines is of the ments made by YASUUA 13 show that the half-width of hollow-cathode order of 0.01 Bi. which is comparable to the Doppler half-width of many absorption lines. The data in Table I show that for a range of v from zero to unity, the decrease of cS(a,v) is quite large if the damping constant is small. For a damping constant equal to unity, the relative decrease of 6(a,v) due to the finite width of the source line amounts to about IO~/~.

VALUES

FOR

--V

b(a,v)

AFTER

-_.

__-.-_

___._-_________---

0.5

1.0

2.0

5.0

0.01

0.42 0.42 0.40 0.38 0.34 0.30

0.26

0.

0.25 0.25

0.1 I 0. I I

0

I

0.2

0.96

0.00

0.4

0.85 0.70 0.52 0.37

0.5G 0.50 0.43 0.35

0.8 I .o

.-..-___

-.--_.

u 0

0.6

YOUNG’”

.oo

OS?4

10.0

I I

0.11 I I

0.23

0.

0.22

0.

I I

0.057 0.057 0.057 0.057 0.050 0.056

A damping constant equal to zero indicates that the absorption line profile is determined by Doppler broadening only. According to eqn. (16), the integral of the absorption coefficient over the complete absorption line is independent of broadening parameters. Consequently, any additional broadening of the absorption line must result in a decrease of the peak value of the absorption coefficient. This decrease of the absorption coefficient with increasing value of a is evident from the data in Table I. Unfortunately, the value of the damping constant in a particular case is by no means certain. Experimental data show a wide divergence9 and theoretical calculaAnal. Chtna. Acta.

37 (1902)

436-444

L. ME GALAN,

442

W.

W.

hICG&E,

J.

D.

WINEFORDNEIZ

tions can be given to an order of magnitude only 14. The most probable range of a in value of 0.5 flames is, between 0.5 and 2.0 14.15, In combination with an approximate for 8, this yields G(a,C)-values ranging from 0.5 to 0.2 Substitution into eqn. (19) yields a corresponding reduction of the gain, RA, of the line source over the continuous source. This is confirmed by experimentally measured values of the gain, RA, presented in Table II. For the resonance lines of a few elements, the absorption produced by the same low solution concentration in an acetylene/air flame (chamber-type aspiratorhurrier, Perkin-Elmer Corp., Norwalk, Conn.) was measured with both a continuous source (150 W Xenon arc) and with narrow line sources (hollow-cathode discharge

, COMPARISCIN

01’

LINE

SOUIKIS

AND

CONTINUOUS

Blcnzenl rend lhc

SOURCE

_.._--

__---.-..

(RA),&

All>

__ ______._ -__-__-

(Rl\)ox,,

(R~.)ux,,

~

.._ __.. --.-

(40

(Al

Nn

5890

0.047

‘} . 0

2.3

I<

++O‘)

0.020

9.4

3.4 7.0

0.2

‘I.2

I

Mg 2852 Cn 422G Sr 4607 ml 3535 Ccl 2288 %n 2139 _--_

-._- -...

d(r4,0)‘,

0.018

10.2

0.022

8.7 I I.0 12.‘) 27.

o.orG 0.01~ o.ooGg 0.0085

2;.

-^-

I 2

0.6

0.3

0.35 0.7 0.5

0.

O.‘&

1.0

I.2 I

0.6

4.7 4.

I -

0.3

1.5

2.

-

0.08

7. I.5

7.

-

0.3

_-_-_-

11 Cdcukrtccl

from cqn. (19) assuming rS(fc.77) = r nnd s = 0.2 A. 1) Cnlculatccl ;LYthe ralio (~ZA)~~,,/(RA),~~X. (1 ‘I’nkcn from ‘l’;~blc 1 from calculatccl value of ij(fc,77) assuming 5 = 0.5.

tubes and Osram lamps run at the lowest possible currents). The minimum spectral bandwidth of a 0.5-m grating monochromator (Jarrcll-Ash Co., Waltham, Mass.) was measured to be 0.20 k. From eqn, (Ig), the maximum possible gain was calculated by I. The. esperimental values of the gain factor arc significantly lower taking 6(a,7j) = than these calculated maximum values, indicating the importance of the factor f.?(a,G).

It should be evident that a simple and quick method of measuring the aparameter in flames could be based upon a comparison between the absorption signals obtained with a continuous source and a line source, rcspeqtively. A value for S(a,fl) can be derived f,ronl the ratio of the experimentally measurecl gain and the calculated maximum gain ; this then yields a value for a. To the tentative data in Table II, no better accuracy can be attributed than a factor two, but with some refinements (the width of the source line should be known approsimately) accuracies of IoO/~ could be reached. It is to be noted that this method is not based on the emission of the element in the flame and thus can be applied to elements that do not emit in flames. Therefore, it provides a valuable alternative to the methods based upon the curve of growthlo*l~. Coqbarison of detection limits From eqns. (I?) and (IS), the ratio of the minimum for a line source and a &ntinuous source is found to be Anal.

Chim

Acta,

37

(1967)

43G-444

detectable

concentrations

LINE

AND

CONTINUOUS

SOURCES

d-E__. 4ln

IN A.A.S.

A?.D

2 s b(n,fi)

443 ~1 $

--

RN

(20)

I-Z,\

which is equal to the ratio of the noise ratio, RN, and the signal ratio, RA. In general, I
Expressions are given for the absorption signal (absorbance or ‘fraction absorbed) in atomic absorption spectrophotometry with a line source and’a contin;uous source, respectively. A theoretical and experimental comparison is made between the shape of the working curves, the magnitude of the signals, and the limits of detection in both cases. The significance of the spectral bandwidth of the monochromator and the absorption line profile is discussed. With a good medium-dispersion monochromator, a continuous source offers several distinct advantages and yields detection limits which are approximately the same as those obtained with a hollow-cathode discharge tube. Avaal.

Claim. Acta,

37 (1967)

436-444

L.

444

I>E GALAN,

W.

W.

MCGEE,

J.

D.

WINEFORDNER

RfiSUd

Des expressions sont don&es pour le signal d’absorption (absorption ou fraction absorbce) en spectrophotometrie par absorption atomique, avec une source Une comparaison theorique et de ligne et une source continue, t-espcctivcment. unc source. continue expc!rimentale a dtd faitc. Avec un bon monochromateur, presente de nombreux avantages et donne des limites de detection qui sont approximativcment les mt?mes que celles obtenues aver un tube de &charge B cathode creuse. ZUSAMMENFASSUNC Es werden Gleichungen fur die Absorption bei der Flammenahsorptionsphotometrie angegeben und zwar fur cinen Linienstrahlcr bziv. einen Kontinuumsstrahler. Fur beide IXlle wircl ein tbeorctischer und experimenteller Vergleich zwischen der Gestalt der Eichkurven, der Gr6ssc cles Signals und der Nachweisgrenze vorgenommen. Die Redeutung der spcktralen Bandbreite des Monochromators und des Absorptionslinienprofils wird diskutiert. Mit einem guten Monochromator mittlerer Dispersion bietet ein Kontinuumsstrahler bestimmte Vorteile und ergibt Nacbweisgrenzen, welche annaherncl gleich denen sind, die mit der Hohlkathodenlampe erhalten werden,

REITRBNCKS

8 9 IO

A. WALSH. Speclroclbi,,l. rlcta, 7 (1955) 1087. J. H. CI~SON, W. E. L. GROSSMAN AND W. I). COOKE, A)@. Spccluy., 16 (1gG2) 47. V. A. FASSEL AND V. G. MOSSOTTI, Awl. Circw , 35 (1963) 252. N. P. LVANOV AND N. A. KOZYREVA, J. Aml. Chewa. USSR, rg (rgG4) I 178. W. W. McGmz AND J. D. WINIZFORUNER, Anal. Clrinz. Acta, 37 (rgG7) 429. S. R. I
(1v?i)

8x3.

35 (rgG3) I 607. J, D. WINEFOR~NKR, C. T. MANSFXELI) AND ‘LY.J. Vxcrcr~ns , .d JJd. Chv?J.. I2 W. SLAVIN AND D. C,. MANNING, Appi. Spcclry., rg (1gG5) 65. 38 (1966) 592. *3 I<. YASUDA, Amal. CIJCPJJ., +ND J. D. WINKDORDNER, Appr. Speclvy., 20 (1gGG) 223. I 4 M. L. PARSONS, W. J. MCCARTHY I-I. ICOHN,J. Opt.‘Soc.~Zm.,.~7(rg57) 156. 15 EHINNOVAND 16 C. YOUNG, Tnbhs fey C&rrlnLing rhc I’oigl Profile. Univcrsjty of Michignn, 1965, ORA-05863. II

Awai.

Claim .4&n, 37 (1967)

436-444