Instrumentation and Principles of Flame Spectrometry Automatic

An adaptation of a multichannel flame spectrometer to correct automatically for background radiation is described. Background intensity is measured to...
0 downloads 0 Views 512KB Size
Instrumentation and Principles of Flame Spectrometry Automatic Background Correction for Multichannel Flame Spectrometer MARVIN MARGOSHES and BERT L. VALLEE Biophysics Research Laboratory, Department o f Medicine, Harvard Medical School, and Peter Bent Brigham Hospital, Boston, Mass.

An adaptation of a multichannel flame spectrometer to correct automatically for background radiation is described. Background intensity is measured to one or both sides of each line or band and is subtracted automatically from the line-plus-background intensity measured at the peak of the line or band. The performance of the instrument with this modification is discussed. Very high concentrations-above about 1 gram per liter-of some salts cause a depression of emission intensity in the flame. This depression can be eliminated by addition of ethyl alcohol to the solutions. A possible cause of this depression in emission intensity is discussed.

T

HE development of instruments for direct-reading analysis with a flame source has been largely restricted t o those permitting the determination of only one element at a time-photometers or spectrophotometers. Instruments for the simultaneous determination of several elements in a sample using arc or spark sources-spectrometers-have been developed. Unfortunately, they are not well suited for use with flame sources. Earlier papers (10, 11) described a multichannel flame spectrometer for the simultaneous determination of several elements in one sample. This instrument was designed t o take advantage of the stability of the flame source and the high intensity of emission by some elements in the hydrogen-oxygen flame. At present the instrument is being employed for the determination of sodium, potassium, magnesium, calcium, and strontium, and facilities for the determination of other elements can be added readily.

that of calcium (6, 9). However, studies with the multichannel flame spectrometer (8, 1 0 ) indicate that the chief effect of extraneous cations is the production of heterochromatic background radiation. The effect on the monochromatic emission intensity of a given species measured above background is apparently negligible. The background produced by each one of the extraneous elements in the sample is proportional to its concentration. Furthermore, all light intensities at a particular wave length, whether due to background or an emission line, are additive. It is possible, therefore, to estimate general flame background plus that due to extraneous ions and to subtract both from the line-plusbackground reading, thus obtaining the line intensity. The sequence of calculations for such indirect corrections has been described and the precision of the method has been tested (8) In the determination of the alkaline earths in biological fluids, the presence of relatively large amounts of sodium and potassium contributes background intensities which may be larger than the line emission intensities of the alkaline earths. Under these conditions the estimation of background due to sodium and potassium introduces an excessive error into the determination Direct measurement of background intensity has been found to give more precise results than the indirect estimation of background described above and previously. In addition, it is relatively easy to arrange the electrical connections in the multichannel flame spectrometer so that background is automatically subtracted from the line-plus-background readings. The automatic background correctlon is more rapid and more precise than the indirect method.

DESCRIPTION OF INSTRUMENT Bockqround Photomultiplier

-

,

Figure 1 shows a block diagram of the multichannel flame spectrometer modified for automatic background correction. The components-source, monochromator, power supplies, etc.have been described in detail (11) and the analytical wave lengths are identical with those used previously (8, 10, 11). The instrument was modified by the addition of auxiliary exit slits and detectors to measure background, and pairs of amplifiers were connected to subtract background automatically.

I

Monoc h romalor

Line Piwiomultiplier

I I I +

7 I

Figure 1. Block diagram of multichannel flame spectrometer modifled for automatic background correction

When several elements in a single sample are to be determined simultaneously, it is necessary t o consider any interactions between those elements which might affect their emission i:itensities in the flame. Several workers have reported that one cation affects the emission of light by another in the flame. For example, sodium and potassium have been variously reported to increase (19) and decrease ( I ) their respective emissions and

An auxiliary exit slit of the same width as the main exit slitLe., 0.4 mm.- was placed inch t o one side of each line t o be measured. As the linear reciprocal dispersion of the monochromator is 10.8 A. per mm., background is measured approximately 70 A. from the line. Scanning experiments (8, 10) have shown that the light intensity 1/1 inch from the line is almost entirely due t o background. The calcium oxide band head a t 5498 A., used for the determination of calcium, is close to the strong sodium doublet a t 5890 and 5896 A. The background intensity from sodium is higher on the long wave-length side of the calcium oxide band head than on the short wave-length side ( 8 ) , while the background due t o potassium is the same on either side of the band head. For this reason, the calcium channel is provided with two auxiliary exit slits, on either side of the band head. By an arrangement decribed below. the light uassing through both of these exit slits is combined and the average int‘knsity 6 measured. Light passing through the main exit slit in each channel-line plus-background-is received by a photomultiplier with its center placed 21/s inches directly behind the exit slit. An auxiliary

1066

V O L U M E 2 8 , NO. 7, J U L Y 1 9 5 6 photomultiplier is located alongside each of t h e line-plus-background detectors. As the diameter of a photomultiplier tube is 1 inch, the auxiliary photomultiplier is not directly behind the auxiliary exit slit. A front surface mirror, 1/2 inch behind the auxiliary exit slit, reflects the light slightly to one side, where it is received by the auxiliary photomultiplier. There is a mirror behind each of the avuiliary slits of the calcium channel, reflecting the light onto one photomultiplier tube, The signal from each photomultiplier is sent to a sepaiate direct current amplifier. Two amplifiers are used for each channel, one for line-plus-background and one for background alone. Each amplifier chassis also includes a high voltage power supply providing a regulated direct current voltage, variable from 700 to 900 volts, for the photomultiplier. The amplifier outputs are connected together in pairs. T h e “low” sides of the line-plusbackground and background amplifiers are connected together; the “high” sides of the two amplifiers are connected across the microammeter. The meter thus indicates the difference between the two signals: the line intensity. Figlire 2 shows the instrument as adapted for automatic background correction. The inset shows the details of the arrangement of exit slits and photomultipliers. The pairs of photomultipliers for measurement of line-plusbackground and background are not critically matched. The photocells are chosen to have similar responses to the same light level; a difference of 10 to l5Y0is considered acceptable. Final matching of the photomultipliers is accomplished by a reduction of the dynode voltage on the more sensitive photomultiplier. For this purpose, background radiation is provided by the atomization into the flame of a solution containing one of the extraneous elements. With both amplifiers in the channel a t their most sensitive settings, the dynode voltage on the more sensitive photomultiplier is reduced until the sensitivity of both detectors is equalized exactly. When this point is reached, closing of the entranre slit of the monochroniator does not

1067 change the ammeter reading. I n practice, the adjustment is made a t the start of work and has remained stable throughout the day. T h e setting may be checked a t any time by flushing a solution containing only extraneous cations through the burner. T h e background photomultiplier on the calcium channel receives light from two exit slits, so that the total light intensity received by it is about twice the background intensity reaching the corrpsponding line-plus-background photomultiplier. When the dynode voltage to the background photomultiplier is adjusted in the usual way, the voltage is reduced so that the sensitivity of the detector is about one half of normal. I n effect, the photocell averages the two light intensities. EXPERI\IENTA L

Concentrated stock solutions were prepared from reagent grade chemicals. These reagents had been examined previously by readings taken a t the peaks of lines or bands and a t n-ave lengths to either side ( 8 ) , and found t o be sufficiently free of impurities for this !vorB. For example, reagent grade sodium chloride did not contain detectable amounts of calcium or magne$um. Metal chlorides were dried overnight or longer a t 130 C., weighed on an analytical balance, and dissolved in water purified by passage through a mixed-bed ion exchange column. Appropriate aliquots of these stock solutions were diluted to prepare the solutions for the experiments described in the following section. 811 solutions to be compared directly were prepared a t the same time. The method of operation of the instrument has been described (8,11). Only slight changes in procedure need be made when the automatic background correction is used. Matching of photocells by adjustment of the dynode voltages has been described above. Thereafter, readings may be obtained in the usual fashion, except that care must be taken to ensure that both the line-

Mirror

Figure 2.

llIultichanne1 flame spectrometer modified for automatic background correction

Inset shows details of exit slit-photomultiplier arrangement A . Beckman atomizer burner G. Photomultiplier mount B . Plane front surface mirror H . Photomultiplier tube CI. Collimating mirror I. Bank of amplifiers C. Gratinr J . Low voltage power supply 6. Detes’or mount K. High voltage power supplies and ampliE. Aperture in detector mount fiers F. Adjustable holder for front surface mirror L. Microammeters

ANALYTICAL CHEMISTRY

1068 plus-background and background amplifiers of a particular channel are a t the same sensitivity setting a t all times. RESULTS

'

The performance of the background correction has been tested by the comparison of readings obtained with solutions t h a t contain the cation to be measured alone and in the presence of considerably larger concentrations of extraneous cations. Many of these experiments have been performed on the calcium channel. The proximit,y of the strong sodium doublet to the calcium oxide band head introduces complications not present at the other analytical wave lengths, thus providing a test under the most difficult conditions. Table I summarizes a set of data on the calcium channel for solutions containing various concentrations of calcium alone and in the presence of 1000 p.p.ni. of sodium. The meter readings are averages of 17 consecutive determinations and the standard deviation of each set of determinations is given in microamperes and as a percentage of the average meter reading. The background from 1000 p.p.ni. of sodium was 380 pa. at this wave length, approximately eight times the signal from 1 p.p.m. of calcium. The photomultipliers were matched by adjusting the dynode voltages when a solution containing 300 p . p m of sodium was flushed through the burner. The b:wkground from 300 p.p.m. of sodium was 125 pa. Table I1 shows similar results for calcium in the presence of 1500 p,p.m. of sodium. The results given in this and all succeeding tables represent single determinations on each solution. The background from 1500 p.p,m. of sodium was 470 pa. At this level of background intensity, noise arising in the photomultiplier becomes a serious factor; the ammeter needle quivers rapidly, so that it is difficult to obtain readings.

Table I. Ca, P.P. 11.

Effect of 1000 P.P.M. of Sodium on Calibration Curve for Calcium Sa. P.P.I\f

0

1

3

10 1 3 10 a

b

1000 1000

Standard Deviation ____pa.

ro

1.5

3.1 1.1 1 7

1 6 8.8 1.4

2 9

2 6

1 9

7.3

1 5

Average of 17 consecutive determinations. Background from 1000 p.p,m. of sodium = 380 pa

Table 11. CE, P.P.hl.

"

0 0 iOOOh

Meter Reading,a pa. 48.3 147.7 511 16.9 139 3 494

Effect of 1500 P.P.\I. of Sodium on Calibration Curve for Calcium Na. P.P.M.

Meter Reading, pa.

Ca, P P.M.

Na, P.P.M.

Meter Reading, pa.

Background from 1500 p,p,m. of sodium = 470 pa.

Table I11 gives the data obtained on the strontium channel for solutions containing 3 p.p.m. of strontium and various concentrations of sodium, potassium, and calcium. For this experiment, calcium chloride was prepared from Spec-Pure calcium carbonate (Johnson, Matthey and Co., Ltd., London, England). Reagent grade calcium chloride was found to contain a significant quantity of strontium, detectable with the flame spectrometer. Strontium was also found in the sample by spectrographic analysis with a spark source. Table I V shows readings obtained on the calcium channel with

solutions containing various concentrations of calcium alone and in the presence of 1000 p.p,m. of potassium. The solutions containing excess potassium gave definitely lower readings than those containing calcium only. The largest discrepancy was observed for the solutions containing 30 p.p.m, of calcium; the reading for the solution with excess potassium is lower by 19% than the reading for the solution containing calcium alone. .4 similar tendency may be noticed in Table 11, where a n excess concentration of sodium chloride caused a slight depression of the readings. The low readings for calcium obtained in the presence of 1000 p p.ni of potassium could be caused by either failure of the backgiound correction or a decrease in emission intensity in the flame.

Table 111. Sr

P.P.Al.

Effect of Extraneous Ions on Readings of Strontium Channel Sa, P.P.hI.

10 100

10005 0 0 0 0 0 0

Ca, P .P.hZ. 0

I< P.P.M. I

0 0

10 100 1000 b

0 0 10 3 3 0 100 3 0 IOOOC 0 Background from 1000 p.p.m. of sodium = 180 pa. t Background from 1000 p.p.m. of calcium = 90 wa. c Background from 1000 p.p.m. of pota.ssillin = 1.55 pa.

Meter Reading, pa.

43 44 48 45 45 45 46 47 47

In order to distinguish between these two possible causes, other metals were substituted for potassium in the solutions; 1000 p,p.m. of lithium, as the chloride, depressed the readings from 1 to 100 p.p.m. of calcium by a n average of 12?& The readings from the same concentrations of calcium were decreased by an average of lOyc when 2.5 grams per liter of cadmium chloride were added t o the solutions. The background a t the calcium oxide band head from this concentration of cadmium chloride was only 1 pa. Zinc chloride, a t a concentration of 2.5 grams per lit'er, reduced the emission of 1 to 100 p.p.m. of calcium by rlmmonium chloride, at concentrations a a n average of 167,. high as 2.6 grams per liter, had no detectable effect on the readings obtained on the calcium channel. The depression in emission intensity caused by the presence of such large amounts of some salts can be alleviated by the addition of ethyl alcohol to the solutions. -1s an example, Table V shows the readings obtained for 1 to 100 p.p.m. of calcium alone and in the presence of 2000 p.p.m. of potassium. All of the solutions contained 2570 of ethyl alcohol by volume. I n spite of the large amount of extraneous salt added, good agreement was obtained between the two sets of solutions Table VI lists the readings obtained n-ith solutions containing various concentrations of stront'ium alone and with 1000 and 2000 p.p.m. of sodium added. The readings obtained for the solutions containing 2000 p.p.m. of sodium show the depression in emission intensity caused by the presence of a large amount of extraneous salt. Similar results have been obtained for 1 to 100 p.p.m. of strontium alone and with 1000 p.p.m. of potassium added. For example, 1 p,p.m. of strontium alone gave a reading of 25 Ha. and 1 p,p.m. of strontium with 1000 p.p.m. of potassium gave a reading of 23 pa. DISCUSSION

The results obtained with the automatic background correction were much more precise than those obtained by corrections based on indirect estimation of background. (See Figure 6 of 8.) I n Table 11, for example, the background intensity from 1500

V O L U M E 28, NO. 7, J U L Y 1 9 5 6

1069

p.p.m. of sodium (470 pa.) was more than ten times the line Table V. Effect of 2000 P.P.M. of Potassium on intensit,y from 1 p.p,ni, of calcium. Even under these extreme Calibration Curve for Calcium conditions, the difference in ammeter readings for the two (.411 solutions contained 25% rthyl alcohol b y volume) solutions containing 1 p.p.m. of calcium was only 2 pa, This Metrr Meter represents an error in the determination of 5% of the amount Reading, Ca. K, Reading, Ca, K, 1'. P. h l . P. P.AI, w.. I',p,ir, P.P.M. ua. present, while Figure 6 of ( 8 ) indicates that errors as large as 1 0 81 1 2000 80 100yc may be expected for this ratio of background and line 7 0 173. 1 2000 1SI) ~. ~.~~ .10 0 560 10 2000 630 intensities when the indirect method of background correction 1630 30 0 3n 2000 1610 is employed. 1no 0 J200 100 moo ainn The ratios of background :tnd line iut(>nsitics are not as large in the other rcwiit,s shown; hut most involve hackEffect of 1000 and 2000 P.P.31. of Podium nn calibration Table \ . I . Curve for Strontiuni gi,c~riiid intensities considerably higher t h m the line intensities. Cndei, these Meter Meter Meter Sr, Sa. Reading. Sr, Sa, Rrading, Sr, Na, Reading, conditions, :I comparatively small error in P.P.31, P.P.31. pa. 1'. P,11. P. P. BI. pa. P.P.M. P.P.N. pa. the background correction would he re2 1ooo'L 40 2 2OOOb ::i 2 0 39 flected in a much larger difference iii the li 1000 13 2 ri 2000 I I!) 6 0 128 410 20 2000 420 20 ioon $00 20 0 ammeter readings. I n t,he example cited !I80 AO won 990 980 no 1000 RO 0 above, a difference of 2 pa. in the Background from 1000 p . p . ~ n .of sodium = 127 pa. xmmeter reading represents a 3:b error b Brtckground from 2000 p.p.m. of sodium = 260 p a . iii the determination, but orily a 0.UFc erroi. in the background correction. I n general, the background correction api t i this paper should be iwsilhllode1 DU spectrophotometcxr ivith the flame att,achment i)y orate during their passage through the flame can bring about a direct estimation (6-6). T h c h:ic.kground is read to one side of decrease in emission intensity. It is considered likely that the t,he line or band. This value is subtracted from the line-pluspresence of high concentrations of certain salts in the flame background intensity measnred a t the peak of the line or hand. c