A Background Emission Correction System for Atomic Absorption Spectrometry

FREDERIC G. DEWALT, JOHN R. AMEND, and RAY W O O D R I F F Department of Chemistry, Montana State University, Bozeman, Montana 59717

A simple system which corrects for both broad band absorption and broad band emission in thermally atomized atomic absorp- tion spectroscopy is presented in this article.

Index Headings: Atomic absorption spectrometry; instrumenta- tion.

INTRODUCTION

Atomic absorption spectrometry has, for a number of years, become a well used and reliable technique for trace metal analysis. Absorbance, defined as the logarithm of the no-sample signal divided by the sample signal, log Io/ /, is directly proportional to the chemical concentration of the species being measured under most conditions.

Two major problems are encountered during the mea- surement of a sample's absorbance in an atomic absorp- tion spectrophotometer. One is the problem of broad band absorption by gaseous molecular species in the optical path. The other is broad band emission by the sample atomization chamber when operated at high tem- peratures. The problem of broad band absorption by gaseous molecular species has been handled in a variety of ways, including the Hitachi Zeeman effect, 1-~ various wavelength modulation techniques, 6-~1 and dual channel techniques. 12-~4 The technique most commonly used and which has been subsequently adopted by most all man- ufacturers of atomic absorption instruments for the cor- rection of broad band absorption was first described by Koirtyahann and Pickett. 15' 16 In this technique the prob- lem of broad band absorption is handled by directing a broad band reference signal (H2 light) through the sam- ple. If the spectral bandpass of the spectrophotometer is large with respect to the band width of the atomic absorption line, absorption will be principally broad band. Absorption of the atomic species plus the gaseous molecular species is determined by passing atomic light from an appropriate hollow cathode lamp through the sample. The background corrected absorbance is calcu- lated by subtracting the logarithm of the hollow cathode lamp signal from the logarithm of the broadi band refer- ence lamp signal.

Siemer ~7 has pointed out that errors can arise if the unmodulated atomic, black-body, and molecular emis- sion signals are not corrected for prior to logarithmic conversion. In electrothermal atomizers this ibackground emission signal can be very large at higher atomizing temperatures (particularly at longer wavelengths) and must be eliminated before the absorbance computation.

An integrating background correction system was de- scribed by Donnelly et al.lS This system was based on a four-phase operational cycle with data acquisition occur-

Received 9 August 1980; revision received 13 October 1980.

176 Volume 35, Number 2, 1981

ring during two of the four phases and analog computa- tion during the other two phases. Donnelly's system compensated for sample chamber emission by using ac coupling at the input to the photomultiplier preamplifier.

The background emission correction system (BECS) described in this paper is based on work earlier reported by Amend. 19' 20 It is also based on a four-phase operation cycle but measures the background emission of the hot sample container in the absence of all other signals during one phase of its cycle. This signal is then subtracted from reference and atomic line signals acquired during the two subsequent phases of the cycle, before converting these values to logarithms for the absorbance computation. Analog computation takes place while data are being acquired, thus increasing the total time available for data acquisition.

I. CIRCUIT DESCRIPTION

The circuit diagram for the BECS unit is presented in Figs. 1 and 2. Fig. 1 comprises the timing section while Fig. 2 presents the computation section' of the BECS. Fig. 3 illustrates the various timing outputs of the timing section which drive the electronic switches in both the lamp power supplies and the computation section of the BECS. Fig. 4 illustrates the activity of the integrators in the computation section of the BECS.

Operation of the system is based on a four-phase cycle derived from a TTL SN74155 decoder driven by a 7473 dual flip-flop and an external crystal clock pulse. The 74155 decoder produces four equally spaced negative logic signals. The first three are inverted to positive logic by NAND gates (TTL 7400). These three positive true signals provide the logic necessary to close the appropri- ate switches in both the HC and H2 power supplies and the computation section of the BECS. The fourth signal is used to generate short "data transfer" and "reset" pulses by triggering monostable multivibrators, A, B, and C. Monostable multivibrator A generates the "data trans- fer" pulse. Monostable multivibrator B provides a 30/zs delay which assures that all data are transferred before the integrators are reset by monostable C. Note that a preset counter precedes monostables A, B, and C. This feature allows 1 to 15 separate measurements to be integrated by integrators INT A, INT B, and INT C before the data are transferred through the BECS (see Fig. 4).

In the first phase of the four-phase cycle, both the hydrogen reference lamp (H2) and the atomic line sample lamp (HC) are allowed to "idle" at a very low light output since both H2 and HC lamp drives are in a "0" logic state. During this first phase only the "furnace acquire" output is active causing switches SW 1 and SW 2 to be closed. An incoming furnace (or sample atomization chamber)

APPLIED SPECTROSCOPY

H 2acquire HC acquire

@ H 2 drive _ ~ +

5 +5

CLOCK (~ 2) 9

+5

+

74

FURNACE DATA acquire TRANSFER

i +5 '[PRESET COUN'TERI

1 '1-5

-2 - + 14 J ~ =l.~--~-.J~+ I ~ I ~31 4 :-~.~._1~7420~'~'1741211

8 7 10 ,,I + S.._ +-' +l'-z ~ I ,o~+ - - ~ 7 11 I L~.;'u,

RESET

T +5 +5

Fro. 1. The circuit diagram for the timing section of the BECS. Of the four equally spaced logic signals produced from the 54155, three are used to close switches in the H2 and hollow cathode lamp power supplies and in the computation section of the BECS. The last logic signal is used to trigger a series of monostable multivibrators which generate successive "data transfer" and "reset" pulses.

-12 +12

IC 1

SW3

P H 2 ~q~i,e

150KI r ",,.-+1-2 ~ I i ,klo ~ I I FURNACE +~ 7 ~ t _ ~ acqu,re

HC acquire

RESET

' - -~ I---'~ "DATA I .i..!2 I H2 a c q u i r r , , , . + 1 2 1 TRAN=SFER i i,,,,,,,,,!.~, i _:_~__: e

INTA TO.luf RESET ~ SH 1

" I N T B ~O'sUl~ 2 IO M E G ~

TO LOG AMP

HC acquire RESET

,c ~..'~-... I sw 8 F - ~ l _ . 4,-1'.7..~/ ~.oo, u, I~m~l l IL1.',+E~uh DATA +'~ LOGTssPA., ~ _ . [ = j ~ . L ~ ~ _ r m,~. ,, , ~ . . L ~ o ; T I I V-'~ , I 1 - ~ sw6 I f-:--I,~'~ TRANS.~R I~L= I °,°.A'

-12(---I I Y '--lea 1"~,,~...I .=acquire I I / : ' ~ " ~ " - " "- l - J • ~ __] + + z ~ - - "+ I N T C _L 0-1uf " _ . ~ IO3 " " ~ ~-. SH 3

Fro. 2. The circuit diagram for the computation section of the BEGS. An amplified P-M tube signal is presented at the "signal in," while the absorbance signal is produced at "signal out" and may be fed to a recorder or a V-F converter and scaler for digital integration.

background emission signal is routed through IC2 and is integrated by integrators INT A and INT B. Note that this signal is inverted in sign by IC2 before application to the integrators. During the second phase of the cycle the "H2 acquire" goes to logic "1" which turns on the hydro- gen reference lamp and closes switch SW 3. A signal equal to reference plus background emission is then accumulated by integrator INT A. Since the inverted integrated background emission signal remained in the integrator, at the end of this phase INT A content will

be equal to [(reference Signal + emission background) - (emission background)] = reference signal. During the third phase of the cycle the "HC acquire" goes to logic "1" turning on the hollow cathode sample lamp, closing switch SW 4 and a similar accumulation of the atomic line sample signal takes place. At the end of this phase INT B content will be equal to [(sample signal + emission background) - (emission background)] = sample signal. During the fourth and final phase of the cycle these signals are transferred to sample hold circuits SH1 and

APPLIED SPECTROSCOPY 177

HHH +

v

o °I L

TO

1 A

G o

E FI 0

0 | ATOMIZER EMISSION [

I( 1 CYCLE TIME

CLOCK PULSE IN from crystal clock

FURNACE acquire Ifirst phasel

H 2 acquire Isecond phasel

HC acquire Ithlrd phasel

DATA TRANSFER and RESET

Ifourth phasel

A TYPICAL ABSORPTION SIGNAL

Fro. 3. The various outputs of the timing section of the BECS with respect to the crystal clock time marks.

gain of amplifier IC3 equal to -1.00. The output voltage of the BECS will then be zero when both reference and sample signals are equal. (e) The system may be conven- iently set up using an oscilloscope to monitor the lamp and furnace emission signals at the input of the BECS.

II. EXPERIMENTAL RESULTS AND DISCUSSION

A block diagram of the system is presented in Fig. 5. A Woodriff furnace electrothermal atomizer was used for the sample atomization chamber. The photomultiplier (P-M) tube was a RCA IP106 powered by a Fluke 412 high voltage power supply. A Beckman DU type mon- ochrometer with a 0.10 mm slit was also utilized. The circuit of the input amplifier is illustrated in Fig. 6 and was operated in an inverting current follower mode. The circuit of the voltage to frequency (V-F) converter is illustrated in Fig. 7. The crystal clock and hollow cathode lamp power supply were identical to the ones described by Dewalt et al. 21 The reference lamp power supply (which drove a Beckman type 96280 H2 lamp as described and utilized by Donnelly e t a l . ) was identical to the hollow cathode lamp power supply except that the sensing resistor (R sense} was changed from 100 to 10 ~2 and a 25 W light bulb was used as a load resistor (RL). (A 2.5 V transformer was also added to power the lamp heater contained inside the Beckman H2 lamp.)

SH2, the integrators are reset and the data acquisition cycle is repeated.

During the second phase of the next acquisition cycle, switches SW 5 and SW 6 close, transferring the emission corrected reference signal through the 755P logarithmic amplifier and the inverting amplifier IC3. This signal is integrated by integrator INT C. During the third phase of this next acquisition cycle, switches SW 7 and SW 8 close, transferring the emission corrected atomic line sample signal through the 755P logarithmic amplifier and the noninverting following amplifier IC4. This signal is integrated by integrator INT C. At the end of this phase the output of INT C will be equal to [(log reference signal - log sample signal)] = log I o / I = absorbance. During the fourth phase of this next acquisition cycle this ab- sorbance signal is transferred to sample hold SH3 and the integrator is reset. The output of the BECS at this point is proportional to the absorbance of the sample data received in the previous cycle. At reasonably fast clock rates this one cycle lag time in the signal received and the absorbance signal produced by the BECS presents no problems in practical analysis.

In order to assure proper operation of the BECS sev- eral criteria must be met: (a) All input signals to the BECS must be of positive polarity. (b) Potentiometer P~ must be adjusted so the magnitude of the input signal (at the output of IC1) matches the signal at the output of ICe (M0 but is opposite in polarity. (c} Potentiometer P2 must be adjusted to assure that the RC time constant of integrator INT B matches that of INT A. This can be accomplished by monitoring both integrators at M2 and M3 and adjusting P~ until equal integration is achieved when both reference and sample signals are equal. (d) Finally, potentiometer P3 must be adjusted, to make the

l ~ ~ A TYPICAL

L ABSORPTION .- - SIGNAL

o,,J atomizer emlssi6n

O ~ a l ~ a t k g r o u n d wait, bee

+H2 I - i \ OUTPUT OF INTA data is transferred V walt ~.~sde Integtat°rs / [reference channel)

,-I:':::;u°':'n'd }\l ::','it:;::::;:"" \ | O U T P U T OFINTB -- I accumulate / M~ '''~aet ~1 Isample channelJ T [ background +HC

A ~c I~Ugm~ ' 2 a ~ # ~ °u~ ~1 ~ t e ~ OUTPUT OF INTC / / i|sb r°atnbs'l ne ig v ? X [log reference -

G ~..I_~.L / I wa,t , ..... , / L ,o9 samp,e, /

E ~ 1 CYCLE +' ' ~ EXAMPLE OF O" -T \~__._---7-- "::o%','o':°d..~ , ~ MULT,PLE

,a ..... ,o.ooo,

\walt ~ 4data acquisition

1 CYCLE TIME

FIe. 4. The outputs of the three integrators in the computation section of the BECS. A typical absorption signal (presented at the top of this figure) is integrated as shown by the three different integrators. The graph at the bottom shows the effect of setting the preset counter in the timing section of the BECS to four. Four cycles of data were accumulated in the integrator before the data are transferred and the integrators reset.

178 Volume 35, Number 2, 1981

__ +I~P-M TUBE J~H 2 LAMP

,+s II

P-M TUBE POWER SUPPLY

SIGNAL OUT

] CRYSTAL CLOCK

HC LAMP DRIVE

,c POWE SOPPLY

L H 2 POWER SUPPLY

RECORDER V-F CONVERT!

SCALER

5. Overafl system diagram.

SW

1

2

3

4

lOOK s w 2 + 1 2 , ~ , - - > - 1 2

]OqKt _"t":-I T IOFFSETI sw, ~ 'OOK~o-,ooK

,N / - . . _ , , L ,20E, , T

( . - - ~ ~ . _I ~ I ,~ T-~'+~+-I ' ~ i

VOLTAGE CURRENT SW 3 W 4 INPUT iNPUT ' T _

AMP GAIN AMP GAIN =11 =1 OUT

INVERT iLION INVERT

~NON INVERT INVERT

- ~, I MEG ~ 1 2 ~ . . ~

6. Circuit diagram for the input amplifier. The LED is a level indicator which is set to flicker below the BECS integrator overload point.

In order to properly evaluate the stability of the BECS, a test was made to show the system's ability to measure a constant simulated absorbance signal over a reasonable time period. After the reference and sample lamps were balanced to equal intensities, an absorbance was simu- lated by decreasing the current flowing through the sam- ple lamp. The resulting output from the BECS was digitized by the V-F converter and counted on a scaler. A programmable crystal-controlled timer attached to the scaler allowed this output signal to be counted for a fixed time period. The output of the BECS was counted and recorded at 40-s intervals. Results of this test are plotted in Fig. 8. Three hundred and sixty samples were taken;

72 points were plotted by averaging groups of five sam- ples. This test shows a larger than expected drift of the BECS output over a 4~h time period (2.45% relative standard deviation). The reference and sample lamps were suspected to be the major cause of this drift and similar stability tests on ~heir output, as seen by the P-M tube, were made. For these tests the P-M tube signal was fed through the input amplifier to the V-F converter and was likewise counted by the scaler with the attached prograramable timer. The lamps were allowed only a 3- min warm up prior to sampling their output intensities. Results of these data are also plotted on Fig. 8. Obser- vations of this figure reveal, even after a 1-h warm up,

APPLIED SPECTROSCOPY 179

lOOK R E G.- 5 ~ - ~ V V V v ~ - ~ R E G. + 5

'iOFFSETI

~IOOK lOOK lOOK

"~ ~ J I,"'--12

452J

V-F

. ~ CONVERTER

7. Circuit diagram for the V-F converter.

,---)+12

--.~-12

2

+5 OU T

L - - ~ G A T E

58-

56-

54-

52-

w a r m ~ ' ,

°p :: v BE OU,PUT 2.45 % relative ttandard ~eviatian

O

03

46-

44-

42-

H 2 LAMP INTENSITY

• HC LAMP L INTENSITY

40-

0 lamps on

\

I I 2 3 4 1 5

TIME (HOURS)

Fro. 8. Data for the system's ability to measure a constant simulated absorbance and data for lamp intensity drifts.

the intensities of both lamps drift while operating under constant current regulation. Differences in the rates of drift for both lamps partially account for the behavior of the BECS output. To prove this statement, nondrifting lamp intensity signals were needed. These signals were simulated electronically by the summing amplifier dia- grammed in Fig. 9. The output of this amplifier was fed directly to the input of the BECS. The BECS output was digitized, counted, and recorded as before. Results of these data showed a relative standard deviation of 0.59% over a 4-h period. Apparently the weakest link of the system is long-term lamp drift. We are currently explor-

ing addition of a reference beam system which will mon- itor and correct for long-term lamp drift.

The system was next tested for its broad band emission correction abilities. Unfortunately, at the time of this experiment, the atomizer had recently been rebuilt and was contaminated with elements which require higher atomization temperatures (such as Cu, A1, Fe, and Ni). However, the atomizer was free of Pb, which requires a lower atomization temperature (1800 to 1900°C). This lower temperature produces an emission signal of only 1 to 2% of the sample signal (hollow cathode lamp inten- sity). A larger atomizer emission was simulated by in- creasing the idle current of the reference lamp. This had the same effect at the P-M tube as increasing background due to higher atomization temperatures. Lead standards were analyzed and the results of this test showed good reproducibility (maximum relative standard deviation of 5.6%) and good linearity at very low concentrations (1 to 3 × 10 -l° g of Pb) with an effective atomizer emission of 12% of the sample signal.

The system's ability to correct for broad band absorp- tion was tested at three different magnitudes of broad band absorption. KBr was used for matrix material to decrease the intensity of the reference lamp. This mate- rial caused nonconsistent decreases of the reference lamp intensity at equal concentrations (deviating as much as +_40%). Approximately 104 to l0 t times more KBr than analyte was used. In each of these three tests graphs were made containing two linear plots calculated with a linear regression program on a Hewlett-Packard 55 cal- culator. One plot consisted of samples containing only analyte (Pb) while the other was a plot of samples containing both Pb and KBr. The difference between the two linear plots on each of the three graphs is a measure of the system's ability to correct for broad band absorp- tion. The three different magnitudes of broad band ab- sorption were approximately 12, 55, and 63% decreases in reference lamp intensities. The maximum deviations of the Pb and KBr plot from the Pb only plot were found to be 2.4, 0.9, and 2.2%, respectively. The results of the last of these tests are plotted in Fig. 10.

III. CONCLUSION

The background emission correction system (BECS) has been shown to be quite stable (0.59% relative stand- ard deviation over a 4-h time period) and able to correct for both broad band emission and absorption efficiently.

180 Volume 35, Number 2, 1981

H 2 DRIVE

-12~ 1" ~.12 5 6 K

R E G.+ S '~--~-~,L~---e'4'~- J, ~ ' ~ * ~ . . . (SIMUtA+ED.2 S,dNAL)I4~-L?!%I I ~ look

.12 lOg +12

REG +5 lOOK TO ' " i I~ I~ i i I-TA~ ~.- - -&~'~,~k- i ~ I SIGNAL ( S I M U L A T E D H C S I G N A L ) L ~ J I

. . . . . . . . . - -

(POT FOR~SIMULATED AiOMIZER EMISSION) ~

. . ~ )REGULATED +5

9. Circuit used to generate simulated sample container emission and lamp intensities.

• A b s . u n i t s

O o 30 -

X

I -

Z 250

o

Z 200- g3

O

< ISO-

- - J - ~ - Pb only

; - - - Pb+KBr

Approx. 63 % decrease of H 2 lamp intensity with Pb+KBr

@

1o0-1 $

CONC. OF Pb (x 10-1°g)

FIG. 10. Data for the system's ability to correct for broad band absorp- tion using 90 × 10 -G g of KBr.

ACKNOWLEDGMENTS The authors would like to give credit for financial support of this work to

Environmental Protection Agency (Grant R-805935 and partial support by Na- tional Science Foundation Grant CHE-7415060 and Energy Research and Devel- opment Administration Grant ET-78-C-01-3087.

Special thanks are due to Frank Schaner and Charles Wirtz for their help in atomizer construction and operation.

1. H. Koizumi and K. Yasnda, Anal. Chem. 49, 1679 (1975). 2. H. Koizumi and K. Yasuda, Anal. Chem. 48, 1178 (1976). 3. H. Koizumi and K. Yasuda, Speqtrochim. Acta 31B, 237 (1976/. 4. H. Koizumi and K. Yasuda, Spe¢trochim. Acta 31B, 523 (1976). 5. H. Koizumi, K. Yasuda, and M. Katayama, Anal. Chem. 49, 1106 (1977). 6. W. Snelleman, Spectrochim. Acta 23B, 403 (1968}. 7. W. Snelleman, T. C. Rains, K. W. Yee, H. D. Cook, and O. Menis, Anal.

Chem. 42, 394 (1970). 8. M. S. Epstein and T. C. O'Haver, Spectrochim. Acta 3OB, 135 (1975). 9. G. M. ttieftje and R. J. Sydor, Appl. Spectrosc. 26, 624 (1972).

10. R. W. Spillman and H. V. Malmstadt, Anal. Chem. 48, 303 (1976}. II. R. J. Sydor and G. M. Hieftje, &nal. Chem. 48, 535 (1976). 12. R. L. Sellers, G. W. Lowry, and R. W. Kane, Am. Lab. 5, 61, Mar. (1973). 13. D. W. Brinkman and R. D. Sacks, Anal. Chem. 47, 1723 (1975). 14. R. Woodriff and D. Shrader, Appl. Spectrosc. 27, 181 (1973). 15. S. R. Koirtyahann and E. E. Pickett, Anal. Chem. 37, 601 (1965). 16. S. R. Koirtyahann and E. E. Pickett, Anal. Chem. 28, 585 (1965). 17. D. D. Siemer, Appl. Spectrosc. 32, 245 (1978). 18. T. H. Donnelly, A. J. Eccleston, and R. L. Gully, Appl. Spectrosc. 29, 149

(1975). 19. J. Amend, "An integrating three-phase background corrected data acquisition

system for atomic ahsorption spectroscopy," in Proceedings of the Stanford Workshop on Diagnostics for Combustion MHD (Stanford University, Palo Alto, CA, 1977}.

20. J. Amend, "A three-phase background correcting data acquisition system for atomic absorption spectroscopy/' MHD Power Generation: Research, De. velopment and Engineering, 1, Annual Report (Montana Energy Reseach and Development Institute, Butte, Oct. 1977).

21. F. G. Dewalt, J. Amend, and R. Woodrifl~ Appl. Spectrosc. 33, 460 (1979).

Application of a Search System and Vapor.Phase Library to Spectral Identification Problems

MITCHELL D. ERICKSON Analytical Sciences Division, Chemistry and Life Sciences Group, Research Triangle Institute, Research Triangle Park, North Carolina 27709

A r e c e n t m a j o r a d v a n c e in t he field o f g a s c h r o m a t o g r a p h y / F o u r i e r t r a n s f o r m i n f r a r e d s p e c t r o m e t r y is the ava i l ab i l i ty o f a s e a r c h s y s t e m a n d digi t ized v a p o r p h a s e i n f r a r e d l i b r a ry . Th i s

Received 21 August 1980; revision received 4 October 1980.

Volume 35, Number 2, 1981

p a p e r p r e s e n t s t he r e s u l t s o f t he a p p l i c a t i o n o f th i s s e a r c h system to a r e a l - w o r l d sample . E x a m p l e s a r e s h o w n of c o r r e c t m a t c h e s a n d a n i n c o r r e c t m a t c h . Overa l l , t he s e a r c h p e r f o r m s v e r y wel l w i t h good qua l i t y u n k n o w n spec t r a , b u t c a n n o t dif- f e r e n t i a t e b e t w e e n a s p e c t r u m a n d b a c k g r o u n d . T h e s e r e s u l t s

APPLIED SPECTROSCOPY 181