Absorption-emission intensity ratio measurements ... - ACS Publications

Absorption-emission intensity ratio measurements in flame spectrophotometry. Rodney K. Skogerboe, Rosemary. Todd, and George Harold. Morrison. Anal...
2 downloads 0 Views 421KB Size
Absorption-Emission Intensity Ratio Measurements in Flame Spectrophotometry R. K. Skogerboe, Rosemary Todd, and G . H. Morrison Department of Chemistry, Cornell University, Ithaca, N . Y. 14850

The phenomenon of self-absorption is analytically useful in flame spectrophotometry. The method, which involves measuring the intensity ratios of multiplet pairs, actually employs an element as its own internal standard. Among the advantages demonstrated for the systems studied are prominent reductions in the dependence on flame conditions, ionization equilibria, and solvent effects; and elimination or reduction of solute vaporization interferences.

THEPHENOMENON OF SELF-ABSORPTION, well known in the various types of emission spectrometry has been a deterrent to quantitative analytical methods and normally attempts are made to avoid it. However, because the degree of self-absorption is concentration dependent, Dennen (I), Canney (2), van Calker (3),and others (4) have studied its application to quantitative arc emission spectrography with moderate success. It was generally concluded that the accuracy of this method was limited by the ability to control the behavior of the absorbing vapor medium-a control not easily attained in arc spectrography. Of all the different excitation media used in emission spectrometry, the flame offers the most easily regulated source and, at the same time, is perhaps most subject to self-absorption problems. Accordingly, the present study involves an investigation of the role of self-absorption in flame spectrophotometry, utilizing the internal standard principle in the form of absorption-emission intensity ratios. The intensity of a spectral line can be described in the form of the Einstein-Boltzmann equation

where N is the density of the emitting species and u is its partition function; h is Planck’s constant; k is Boltzmann’s constant; g is the statistical weight and E the energy of the upper level; A is Einstein’s transition probability; y is the frequency and l i s the intensity of the line; and Tis the absolute temperature of the excitation medium (5). If two lines which are members of the same multiplet are considered, the terms N , u, and E are equal for both lines and Equation 1 can be rearranged to show that the intensity ratio of the lines, in the absence of self-absorption, is given by

where the subscripts refer to the respective lines. Thus the multiplet intensity ratio is constant in the absence of self-absorption. In actual practice, the constant value of the ratio may deviate from that predicted by Equation 2 but the agree(1) W. H. Dennen, Ph.D. Thesis, Dept. of Geology, MIT, Cambridge, Mass., 1949. (2) F. C. Canney, Bull. Geol. SOC.Am., 63,1238 (1954). (3) J. van Calker, Spectrochim. Acta, 1,403 (1941). (4) F. E. Roach and T. J. Rollins, J . Opt. SOC.Am., 37,lO (1947). ( 5 ) C. H. Corliss and W. R. Bozrnan, “Transition Probabilities,” Natl. Bur. Std. Monograph 53, 1962. 2

ANALYTICAL CHEMISTRY

Table I.

Element ca Cr Cs K

Li Mn Na Ni Rb

Intensity Ratios and Range of Linearity for Elements Studied MultiplFt Range of linearity, ratio, A CrgId 393414227 2-100 357913605 500-2000 852118943 4-10 766517699 3-10 404414047 2000->10,000 670816104 20-200 403114034 200-10,000 279512801 20-500 589015896 iwo 352513493 100-2000 352513515 100-2000 1800/1948 2-10

ment is typically within reason. When the concentration of the emitting element is increased a level is reached where the most intense component line of the multiplet begins to selfabsorb. Thereafter, the multiplet intensity ratio changes continuously with concentration up to a level at which the second component line begins to self-absorb. The concentrations at which self-absorption begins depends on the oscillator strengths of the two lines which thus define the range over which the ratio change occurs. Application of this concept actually involves the use of an element as its own internal standard and should offer a nearly ideal satisfaction of criteria used in the selection of an internal standard. For instance, the excitation potentials of the two lines are matched, thereby reducing the temperature variation effects; the volatilization and ionization behaviors of both lines are the same; and the emission lines are mutually syrnpathetic to matrix or compositional changes. From another viewpoint, this approach essentially combines flame emission and atomic absorption and permits the utilization of salient features of both methods. It is shown herein that this approach is applicable to flame spectrophotometric analyses for a number of elements, and that many of the anticipated advantages for this idealized application of the internal standard principle are realized. A reduction in the required degree of regulation of analytical conditions and the elimination or reduction of several prominent interelement effects are two prime realizations. EXPERIMENTAL

Apparatus. The instrumentation used has been described (6). For measurements in the near infrared wavelength region, a 1200 line/mm grating blazed for 6000 A was used in conjunction with an RCA 7102 photomultiplier. Oxyhydrogen flames were used throughout. Solutions. High purity metals, oxides, or halides were dissolved in a minimum volume of the appropriate AR grade acid and diluted to volume with distilled deionized water. (6) R. K. Skogerboe, Ann T.Heybey, and G. H. Morrison, ANAL.

CHEM., 38,1821 (1966).

h-11

I

I

'

-

10

L

100

1000

10,000

4

Concentration, rg/rnl

Cs 8521

Na5890 'Na6'

Mn 2795

I

I

8

10

I

I

~

F

I

IqJJ 9

12

I

I

I1

13

I

I

J

13

15

I7

I P ~ I

l;'ll

Hydrogen Flow R o t e , L l r n l n

Figure 1. Typical absorption-emission intensity ratio calibration curves '349 '-8

I

6

Figure 2. Effects of hydrogen flow rate on intensity ratios and line intensities (0)

(0)

Mn4031

'm

Stock solutions of 1.O metal concentration prepared in this manner were subsequently diluted to prepare the required concentration series. Selection of Analysis Conditions. The elements having ground state multiplets were determined from atomic absorption and transition probability data (5). Resonance multiplets do not exist for Ca and Li, requiring that lines which were not members of the same multiplet be used. Because the resonance lines for these two elements were much more intense than the other lines selected, nylon attenuation screens were placed in the optical path when measuring the intensities of Ca and Li to permit making the measurements at the same instrumental sensitivities. The wavelengths utilized are listed in Table I. The intensity ratios of the respective multiplets were surveyed as a function of concentration under standard flame conditions. The gas flow rates, the flame region viewed by the monochromator, and the sample feed rate were subsequently varied to determine their respective effects on the ratios at a concentration level for which self-absorption of the more intense line was evident. The settings of the respective variables which minimized the intensity ratio (absorbing line/nonabsorbing line) were selected for all following measurements. Because only the ratios of intensities were of interest, the photomultiplier voltage and/or amplifier gain adjustments were made to keep the line signals on scale-Le., these settings are not as critical as they would be for usual flame intensity measurements. Ratios were determined by scanning the lines consecutively at 5 A/minute. RESULTS AND DISCUSSION

Figure 1 presents typical curves which verify the applicability of the absorption-emission intensity ratio method to quantitative analysis. As previously mentioned, the onset of self-absorption is coincident with the lower and higher concentration inflection points of the curves for the more intense and weaker multiplet lines, respectively. The quantitatively useful linear portions of the curves depend on the magnitudes of the oscillator strengths for the two lines. When the concentration range of interest is not coincident with the linear region of the curve, it is frequently possible to choose another multiplet pair(s) offering linearity at that level--e.g., the Mn pairs in Figure 1 . The systems investigated are listed with the range of linearity in Table I.

c.

z::E,

0

~

0 Cr 3605

B

'? &-a).--

.. .

I

I

I

I

10

20

30

40

P

''

P

15

~

I

I

25

35

I

P

~

4 S J J IO

I

20

I

I

30

40

Flams Region Viewed, mm A bow Burner Tlp

Figure 3. Effect of flame region viewed on intensity ratios and line intensities Key is the same as for Figure 2

In view of the fact that the ratio approach closely approximates an ideal application of the internal standard principle, it might be logically argued that whatever affects one component line must affect the other to the same degree. This argument is further strengthened by the observation that, in general, factors which change the degree of atomic excitation produce a corresponding change in the degree of atomic absorption. On these bases, it would be anticipated that analysis conditions (flame and/or instrumental) and interference effects would be reduced to minimal, if not negligible levels. These possibilities were examined for a number of representative elements. Effects of Flame Conditions. Figures 2 to 4 present typical data demonstrating that the ratio measurements are instrumental in reducing (or eliminating) the effects of the flame variables in comparison to their effects on the line intensities alone. These data were obtained at concentration levels for which self-absorption for only the strongest component line was apparent. For all elements studied, changing the VOL 40, NO. 1, JANUARY 1968

3

IC N N

0

2.4

S

8

s

n m n 0 0 H

c N

0 0

0

5 0.8

.-

H

-., Q

K

I 1

0

1

1

1

20 10 Anion -to-Calcium

1

1

1

30 Concentration

1

1

40

1

50

1

1

60

Ratio

Figure 4. Effect of sample feed rate on intensity ratios and line intensity

Figure 6. Anion effects on calcium (10pg/ml calcium present)

Key is the same as for Figure 2

A . Intensity ratio B . Intensity 0 Sod-2

hydrogen or oxygen flow rates produced significant changes in the average intensity ratios (greater than *2% change) only for those elements with a tendency to form stable oxides in the flame--e.g., Cr. Varying the flame region viewed resulted in significant ratio changes for only the elements having unusually low ionization potentials-.g., Cs. Variation of the sample feed rate does produce reduced but significant ratio changes for all the elements studied as might be anticipated. Forcing more sample into the flame must increase the density of neutral (absorbing) atoms and should also have a cooling effect for aqueous solutions. The cooling effect should increase the neutral-to-excited atom ratio with a consequent increase in self-absorption. The intensity maxima observed for increasing feed rates are consistent with this possibility. The spectral band pass of the monochromator might also be expected to affect the intensity ratios. In general, the ratios increased with the band pass but the relative changes were typically less than 5 % for extreme slit widths of 5 and 50 microns. This instrumental variable has been shown to be highly critical for instruments of higher resolution, however ( I , 2).

1.8

1.0

I

I

I

1

1

I

I

I

I

”+

-

-

1

0

1

200

1

1

I

I

I

400 600 K Conerntration.ug/ml

I

800

,

I

1000

Figure 5. Ionization repression effect of potassium on cesium (5 &ml Cs present) 0-

4

s

,:r

ANALYTICAL CHEMISTRY

0 Cs 8943

0

A c104-

For all the flame variable-element combinations studied, the relative intensities of the nonabsorbing lines changed by factors of 2 to 30 or more depending on the element and the flame variable studied. On a comparative basis, it is obvious that the above variables exert considerably less pronounced effects on the ratio measurement and can frequently be changed over a wider experimental range than is possible when line intensity alone is used as the measurement parameter. When an effect was observed, the levels of the variables which minimized the intensity ratios were selected for subsequent studies. Interference Effects. The effects included in this general category present some of the most serious deterrents to analyses by both the flame and atomic absorption methods because of the inaccuracies which accrue when an existing interference is not recognized by the analyst. The role of the ratio technique in eliminating several prominent interferences was thus examined. Any spectral interference must affect the intensity ratio unless the interference intensity contribution is the same at both multiplet wavelengths. The latter situation is conceivable for continuous background radiation or possibly a molecular band. Measurement of multiplet intensity ratios in the absence of self-absorption, however, can readily be used to verify the existence of a spectral interference on one of the component lines. On the other hand, arguments for and against the possible elimination of physio-chemical interferences by using the ratio method can be made. The validity of these arguments, however, depend on the type of interference under consideration. Mutual element effects which produce prominent changes in the ionization equilibria in the flame occur for elements of low ionization potential (7). As an example, Figure 5 presents the effect of adding potassium (as KHP) to a solution containing 5 pg/ml cesium. Although the reduction in the degree of cesium ionization resulting from the presence of a 200-fold excess of potassium causes a 70 increase in the intensity of (7) John A. Dean, “Flame Photometry,” McGraw-Hill, New York, 1960.

the neutral atom line, the multiplet ratio is reduced less than 8%. These changes would be equivalent to 60 and 1 8 z increases in the measured concentration for the intensity and the intensity ratio methods, respectively. Similar results were obtained for investigation of several other alkali metal combinations. In general, ionization suppression affected both the intensity and the intensity ratios but the magnitude of the effect was strikingly reduced by the intensity ratio measurement for all combinations studied. Solute vaporization interferences caused by the formation of stable molecular species are common in both flame and atomic absorption analyses. The classical anion effects on calcium presented in Figure 6 were studied in this evaluation by adding the anions as the acid. The data distinctly indicate that the adverse effects on the line intensity are canceled by the ratio measurement. Similar results were obtained for three other calcium concentrations within the range of linearity (Table I). The elimination of these effects results from the fact that both calcium lines were affected to the same degree by the respective anions. Note that the reduction of the ratio at high sulfate levels originates from calcium impurity in the acid added, as evidenced by the relative intensity curve. Solvent vaporization effects were also compared for the two methods. For a 10 pg/ml sodium solution the change in the absorption-emission intensity ratios corresponded to an 18 % increase in the apparent sodium concentration when ethanol and aqueous solutions were compared at the same analysis conditions. In contrast, a 45% increase in the apparent Na concentration resulted from the change to ethanol when direct intensity measurements were used. A similar comparison for 50 pg/ml of manganese resulted in a 10% increase in the apparent Mn concentration for the ratio method and a 460% increase for the intensity method. In view of these results, it is apparent that the combined emission-absorption measurements are instrumental in either eliminating or significantly reducing a number of the well known physio-chemical interferences common to flame and atomic absorption spectrometry, In essence, improvements in the analytical accuracy can be realized through the use of the ratio method for analysis systems subject to biasing effects of the general types discussed above. This advantage is further enhanced by the fact that attempts do not have to be made to find a suitable internal standard-the analysis element itself serves this function. Precision. Although the effects discussed above are of prime importance in determining the accuracy of the method,

the measurement precision limits the accuracy attained in the absence of such effects. For the systems studied herein, the relative standard deviation of measurement, as determined from four ratio measurements, was consistently less than =k3 %. Two factors contribute to the magnitude of this estimate. First, the necessity for making sequential measurements on the two lines with the single-beam monochromator used did not allow the cancellation of flame and/or electronic variations as would be possible with a dual-monochromator, ratio-readout system. Second, the relative standard deviation (RSD) of a ratio measurement is determined by: RSD,~,

=

+

~ R S D , ~R S D , ~

(1)

where the subscripts x and y indicate the respective line intensities. Accordingly, the ratio precision estimate must always be greater than the estimate for the intensity of one line although the difference becomes very small for highly precise measurements-eg, RSD < 1 %. It should be noted that lines which are not members of the same multiplet can also be utilized. The calcium data, for example, are ion-to-resonance line ratios. In addition to self-absorption, the ion line intensity per unit concentration decreases with concentration because of the ionization repression effect caused by increasing calcium concentration. A flattening of the ratio curve is the net result. CONCLUSIONS

In view of the results reported in this survey, absorptionemission intensity ratio measurements offer analytical potential in flame spectrophotometry. A brief examination of multiplet tables considering the current capabilities of flame excitation indicate that over 30 elements might be determined in this manner. It is apparent that the best ultimate employment of this approach would require the utilization of dual-beam, ratio-readout instrumentation. With such a system, precision better than il should be attainable. In the absence of systematic errors, this capability would satisfy certain analytical requirements that cannot at present be readily achieved by flame analyses. RECEIVED for review July 27,1967. Accepted October 2,1967. Work supported by the Advanced Research Projects Agency through the use of the Central Facilities of the Materials Science Center, Cornel1 University.

VOL 40, NO. 1, JANUARY 1968

0

5