contains about 7 % CaO, 2% A1203,and 4% MgO, and the in-house lime glass contains approximately 15% CaO, 6 % AlZO3,and 1 % MgO. Collectively, these materials span a range of compositions sufficient to permit evaluation of the applicability of the method to many technical materials. Both citrate and acetylacetone mask the interferences in these samples, but the end-point breaks are significantly less sharp than with the synthetic samples. This is not unexpected, and indicates the need for caution in projecting results from synthetic samples to technical materials. As with the synthetic samples, acetylacetone gives sharper end points than does citrate. Results using a 0.02M concentration of either masking agent are reported in Table I1 together with the accepted values for the NBS standards and for the in-house standard as determined by visual titration as described in the Experimental Section. For all reference materials, the potentiometric determination agrees with the accepted value to within 1-3 ppt. The precision of the determination, as reflected in standard deviations (where applicable) is likewise of the order of 3 ppt.
The NBS SRM-620 soda-lime glass is of particular interest in that the new EGTA potentiometric method for calcium provided the first reliable evidence in our laboratories that the nominal figure of 7.5% CaO for this glass, as transmitted to us initially, was incorrect. The “accepted” value for CaO here is the average of results reported by ten laboratories, in which calcium was determined gravimetrically by ignition to the oxide after a double oxalate precipitation. ACKNOWLEDGMENT
Helpful discussions with D. E. Campbell and Y.-s. Su are gratefully acknowledged. RECEIVED for review February 18, 1972. Accepted April 11, 1972. Presented in part by I. E. Lichtenstein at the 162nd National Meeting, American Chemical Society, Washington, D.C., September 13, 1971. Supported in part by NSF grant GP9557.
~~
“Color Temperatu re” Atomic FIuorescence Method for Flame Temperature Measurement Nicolo Omenetto,’ Richard Browner, and James Winefordnerz Department of Chemistry, University of Florida, Gainesville, Fla. 32601 Guglielmo Rossi and Pietro Benettia C C R Euratom, Chemistry Division, Zspra, Ztaly
IT HAS BEEN RECENTLY demonstrated ( I ) that atomic fluorescence spectrometry can provide several possible ways for determining the electronic excitation temperature of analytical flames in the range 2000-3000°K. In this work, the following basic relationship, derived by Alkemade (2), was applied
“20 Fz1
= -
(k)5
exp (- Vl/kT,)
Exo?
where : number of secondary photons emitted per sec at X0z assuming radiational excitation via 1 * 2 only (anti-Stokes fluorescence) Fzl = number of secondary photons emitted per sec at X12 assuming radiational excitation via 0 * 2 only (direct line fluorescence) Ex,, = spectral irradiance of the continuum excitation nm-I) source at wavelength XI*(erg sec-l Exoz= spectral irradiance of the continuum excitation source at wavelength Xo2 (erg sec-’ c m 2 nm-I)
Fzo
=
1 On leave from Institute of Inorganic and General Chemistry, University of Pavia, Pavia, Italy. 2 Author t o whom reprint requests should be sent. a Institute of Inorganic and General Chemistry, University of Pavia, Pavia, Italy.
X
=
VI
=
T,
k
central wavelength of the absorption transition (nm) excitation energy of the level above the ground state (erg) = flame temperature (OK) = Boltzmann constant (erg/”K)
The above relationship holds only for a continuum source of excitation and for low atomic concentrations of the analyte. With the introduction of thallium solutions into different flames, the application of the above formula to the determination of the flame temperature ( I ) gave results in excellent agreement with those calculated by the line reversal method and the two-line method. From the theoretical expressions given by Alkemade (2), it was also suggested that a simple, straight-forward procedure for calculating flame temperatures could be devised if the color temperature of the excitation source were known ( I ) . The color temperature can be deo z )the T ratio of fined in the following way (3); let ( E ~ l z / E ~ be the spectral irradiances of the source at wavelengths Xl2, h02, and at temperature T. The color temperature of the source Tc (Al2, XOZ)can be identified with the temperature of a blackbody having, at the same wavelengths Xlz and Xoz, a spectral irradiance ratio equal to that of the source. Therefore (Exl?l~xoz)s = (EA1?iEXOJT,bb
(2)
where superscripts s and bb refer to the excitation source and a blackbody, respectively.
(1) N. Omenetto, P. Benetti, and G. Rossi, Spectrochim. Acta B
in press. (2) C. Th. J. Alkemade, Pure Appl. Chem., 23,73 (1970).
(3) G.Ribaud, ‘‘Trait6 de Pyrometrie,” Ed. l’optique, Paris, 1931. ANALYTICAL CHEMISTRY, VOL. 44, NO. 9, AUGUST 1972
1683
(COLOR TEMPERATURE) (Tila IO‘) LAMP VOLTAGE ( V I
Figure 1. Source spectral irradiances ratio 535.0 and 377.6 nm as a function of the voltage applied to the lamp If the ratio in Equation 2 is measured experimentally, the color temperature can be calculated by the use of Wien’s law. Furthermore, if this ratio is substituted into Equation 1, the following relationship results (I, 2) F20/F21 = exp [VdTC-’ - Tf--l)/kI
(3)
The above formula offers two possibilities for determining the flame temperature: (i) The ratio F20/F21for a given flame temperature can be lower or greater than unity, depending on the color of the source. Therefore, the color temperature of the source can be continuously varied until a “reversal” point occurs, i.e., the ratio F2,/F21 is unity. In this case, the color temperature will be equal to the flame temperature. (ii) A straight line will be obtained by plotting log ( F ~ o / FZl)us. TC-l. The extrapolation of this line to log (FgO/ F21)= 0 will give the flame temperature. No experimental data were presented in the paper by Omenetto, Benetti, and Rossi (I). Previous results obtained in this laboratory ( 4 ) by an atomic absorption method and with a simple quartz-iodine tungsten source led us to investigate the practical feasibility of the “color temperature’’ method. EXPERIMENTAL
The filament of a quartz-iodine lamp (General Electric, DYP 120V, 600W) was focussed slightly behind the flame by the use of a condensing, heat absorbing, glass lens and reflected back through the flame by a concave spherical mirror. The lamp voltage was adjusted to any desired value by the use of a simple Variac control. The light was mod(4) R. F. Browner and J. D. Winefordner, ANAL.CHEM., 44, 267
(1972). 1684
Figure 2. Source color temperature as a function of the spectral irradiances ratio at wavelengths 535.0 and 377.6 nm ulated at 666 Hz by a mechanical chopper (Princeton Applied Research, Princeton, N.J., Model 125). The emitted fluorescence radiation was focussed onto the entrance slit (1 mm wide X 10 mm high) of a compact grating monochromator (Bausch and Lomb, Rochester, N.Y.) equipped with a 1350 grooves/mm grating and a EM1 9781B photomultiplier tube. The resulting signal. was fed into a low-noise pre-amplifier (Princeton Applied Research, Model CR4) and lock-in amplifier (Princeton Applied Research, Model JB-4) set for 666 Hz, and displayed on a 0-25 mV potentiometric recorder (Model SR, E. H . Sargent Co., Chicago, Ill.). Two glass color filters (Corning glass filters No. 3385 and 5850) were used in order to isolate the thallium fluorescence transitions at 535.0 and 377.6 nm. The experimental results were corrected for the spectral response characteristics of the monochromator and photomultiplier with standard tungsten ribbon lamp (NBS calibrated EPUV 1068-35A, the Eppley Laboratory, Inc., Newport, R.I.) Whenever necessary, neutral density filters (No. 26-5793 and 26-5801, Ealing Corp., Cambridge, Mass.) were used. Two low-background premixed, laminar, flame shielded Ar-02-H2flames, burning on a Meker type burner with 20-mm path length, were investigated. The flame was shielded by a metal-free outer flame with the same gas flow rates as the inner one. The gas settings were exactly the same as those described by Browner and Winefordner (4). RESULTS AND DISCUSSION
The excitation source was placed on the optical axis of the spectrometric system, and the experimental spectral irradiance ratio measured at different applied voltages. A calibration of color temperature us. applied voltage was made (see Figures 1 and 2). From the data in Figures 1 and 2, an experimental knowledge of the voltage corresponding to F20 = F21should give the flame temperature. However, it was found experi-
ANALYTICAL CHEMISTRY, VOL. 44, NO. 9, AUGUST 1972
Figure 3. Anti-Stokes to direct line fluorescence ratios as a function of the reciprocal of the color temperature of the excitation source Lower curve: Ar-02-H2 flame: line reversal temperature 2098 K ( 4 ) ; correlation coefficient: 0.9985 Upper curve: Ar.-O2-H2flame: line reversal temperature 2250 K ( 4 ) ; correlationcoefficient: 0.9943
mentally that no reliable fluorescence measurements could be obtained when the lamp was operated at a voltage less than 70 V which corresponded to a color temperature of less than 2405 K. Therefore, with theparticular lamp used in this work, the extrapolation method was preferred and repetitive fluorescence measurements were carried out in the range 70-100 V with a thallium concentration varying from 200 to 500 ppm. The experimentally observed variation (drift) of the spectral output of the excitation lamp with time was taken into account with a simple modification of the described set-up. The source filament was focussed onto the entrance slit of a Jarrell-Ash 0.5-m Ebert monochromator (calibrated with the NBS calibrated tungsten filament source described above) positioned in front of the source and at 90" from the compact monochromator (Bausch and Lomb). In this way, the spectral irradiance ratio was monitored after each fluorescence measurement by simply removing the spherical mirror behind the flame. The log of the ratio of anti-Stokes to direct line fluorescence, corrected for the variation of the spectral response of the measuring system, was plotted against the reciprocal of the corresponding source color temperature; these results, shown in Figure 3, are in excellent agreement with the previously reported values (4) of 2098 and 2250°K (line reversal temperatures). The overall precision and accuracy of the method are related to different independent sources of error, namely : (i) the error in calibration of the spectral system (a systematic error) (ii) the error in measuring the spectral irradiance ratio and consequently the error in the color temperature values (a random error) (iii) the error in measuring the fluorescence ratio (a random error) By expressing T f explicitly in Equation 3, (4)
where F designates F20JF21 and includes the instrumental correction factor. The error in the flame temperature clearly depends on the errors in the source color temperature and in the fluorescence ratio. Therefore,
By solving the derivatives and rearranging,
e Tf2
=
(TC-l
- -k Vl In F)-'[T,-'
nTc2
+
(5)(k)
~ , 2 ]
and by use of Equation 4
Equation 7 can be used to estimate the overall precision of flame temperature measurement. The error in the source color temperature is related to the error in the spectral irradiance ratio, and it can be shown by expressing Tc by use of Wien's formula that =
TC
(5)' =
C'(
0*434)2 5040 V I
($)'
(8)
where E means Eh,,/Ex0,and includes the instrumental correction factor. In the case of T1 (Vl = 0.966 eV), this formula becomes
From these considerations, it can be shown that at the temperatures of 2100 and 2250°K, the relative error of the color temperature will be less than 1 even assuming a 5 error in measuring the spectral irradiance ratio and in calibrating the set-up. A typical experimental error of 2 % in the spectral irradiance ratio has been found at most of the applied voltages. Repetitive measurements of FzO and FZ1 were found to be reproducible within 3 % for F ~ and o 1 for FZl. Therefore, the total random error affecting the calculated flame temperature can be predicted by the use of Equation 7 to be about 1 - 2 z . This prediction was clearly borne out experimentally and temperatures of 2103 + 26°K and 2240 f 35 "K were measured.
z
z
z
CONCLUSIONS
The application of the extrapolation procedure described above results in a simple and accurate method of measuring the flame electronic excitation temperature. Moreover, the ANALYTICAL CHEMISTRY, VOL. 44,
NO. 9,AUGUST 1972
1685
excitation temperature of a low level (VI = 0.966 eV) is measured, this level being much less liable to deviations from Boltzmann equilibrium than the higher excitation levels. On the other hand, it should be mentioned that the typical advantages of a null method [i.e., measurements at log (F20/ FZ1)= 01 have not been realized, However, according to Snelleman and Alkemade (3, this goal can be achieved by
means of an additional calibrated attenuation filter placed in front of the excitation source. This filter, by suitably modifying the ratio E A ~ ~ / E would A ~ ~permit , the realization of low color temperatures with high real source temperatures,
(5) W. Snelleman and C . Th. J. Alkemade, Utrecht University, Utrecht, The Netherlands, personal communication, 1972.
for review January 3 1 3 1972. Accepted 1972. Research supported by AF-AFOSR-70-1880C.
18,
~
Combined Ion Exchange-Solvent Extraction (CIESE) Studies of Metal Ions on Ion Exchange Papers Synergistic Effects and Chromatographic Separations Ani1 K. De and Chitta R. Bhattacharyya Department of Chemistry, Visva-Bharati, Santiniketan, West Bengal, India
THECONCEPT OF combined ion exchange-solvent extraction (CIESE) was introduced by Korkisch and collaborators (1). Literature reports on combined ion exchange-solvent extraction (CIESE) studies of metal ions using mixed aqueous organic solvents have pointed out greatly increased selectivities (2-18) of the technique. In our previous papers (8, 9), we have described the use of 2-thenoyltrifluoroacetone, tri-n-butyl phosphate, acetyl acetone, methyl ethyl ketone, and methyl isobutyl ketone for the CIESE separation of many metal ion mixtures generally associated with ores and minerals. In this paper, SA-2 and SB-2 ion exchange papers have been used to describe several selected chromatographic systems. No such work has yet been reported. EXPERIMENTAL Amberlite SA-2 (H+ form) and SB-2 (Cl- form) ion exchange paper strips (25 cm X 2.5 cm.) (H. Reeve Angel Co.,
Clifton, N. J.) have been used. The former contains Amberlite IR-120 sulfonic acid cation exchange resin and the latter contains IRA-400 quaternary ammonium anion exchange resin. 2-Thenoyltrifluoroacetone (TTA) (Columbia Organic Chemicals, Columbia, S. C.,) and tri-n-butyl phosphate (TBP) (Matheson Coleman & Bell Company, Rutherford, N. J.) have been used throughout the work. Initial zones of the solutions (containing 1 mg of each metal per milliliter of the solution) were spotted with fine glass capillaries and the chromatographic runs were carried out in 30-cm X 5-cm glass jars by ascending paper chromatography technique. The developed zones were quite well defined, development being carried out by spraying with suitable reagents as follows: (1) Trisodium pentacyano aminoferrate rubeanic acid : Mn (light blue), Fe(II1) (deep blue), Co (yellowish brown), Ni(I1) (blue), Cu(1I) (apple green), Zn (red); (2) Alizarin Red S : La(II1) (brown), Ce(IV) (violet), Zr(IV) (red), Th(1V) (reddish violet), In(II1) (violet), A1 (reddish violet) ; (3) KI SnClz: Pt(I1) (yellow t o brownish yellow), Pd(I1) CHsCOOH: (pink t o dark purple); (4) K4Fe(CN)G U(V1) (light brown), Mo(V1) (deep brown), V(V) (yellow) ; ( 5 ) K I (aqueous): TI(1) (yellow); (6) Oxine: Nb(V) (pink), Ta(V) (brown); (7) NH4SCN SnClz: Cr(II1) (red), W(V1) (red); (8) Rhodamine B: Ga(II1) (red); (9) Dithizone: Zn (red), C d (purple), Hg(I1) (pink), As(II1) (yellow), Sb(II1) (red), Bi(II1) (purple). Before spraying with reagents Numbers 1 and 9, the papers were exposed t o ammonia vapor. The paper after spraying with reagent number 2 was exposed to ammonia vapor.
+
+
(1) J. Korkisch and S . S . Ahluwalia, ANAL.CHEM.,38,497 (1966). (2) J. Korkisch, Progr. Nucl. Energy Ser. I X , 6, 1 (1966). (3) K. Korkisch, Separ. Sci., 1, 159 (1966). (4) J. Sherma, Talanta, 11, 1371 (1964). (5) J. Korkisch, Separ. Sci.,1, 154 (1966). (6) J. Sherma, Chemist-Analyst, 55, 86 (1966). (7) J. Sherma and K. M. Rich, J. Chromatogr., 26 327 (1967). (8) A. K. De, S . K. Sarkar, and C. R. Bhattacharya, Ind. J . Chem., in press. (9) A. K. De and C. R. Bhattacharya, Anal. Chim.(Warsaw)(communicated). (10) J. Korkisch and S. S. Ahluwalia, Anal. Chim. Acta, 34, 308 ( 1966). (11) J. Korkisch, Nature, 210, 626 (1966). (12) J. Korkisch and K. A. Orlandini, ANAL.CHEM.,40, 1127 (1968). (13) J. Musich, K. A. Orlandini, and J. Korkisch, U.S. At. Energy Comm. Rept. ANL, January 1968. (14) K. A. Orlandini and J. Korkisch, Separ. Sci., 3, 255 (1968). (15) J. Korkisch and K. A. Orlandini, Talanta, 16,45 (1969). (16) J. Korkisch and K. A. Orlandini, ANAL.CHEM.,40, 1952 (1968). (17) K. A. Orlandini and J. Korkisch, Anal. Chim. Acta, 43, 459 (1968). (18) W. Wahlgren, K. A. Orlandini, and J. Korkisch, ibid., 52, 551 (1970) 1686
+
+
RESULTS AND DISCUSSION Cation Exchange Systems. SEPARATION OF METALIONS. The separations of several metal ion mixtures are summarized in Table I, the metal ions being arranged in order of increasing atomic numbers. A combination of 0.10M solution of TTA in acetone with 6 N hydrochloric acid has been used as the developing solvent. A variation in TT4 concentration (90 to 70%) and hydrochloric acid (10 to 30%) has been found sufficient for the effective separations. A 0.1M TTA solution in acetone mixed with 6 N aqueous hydrochloric acid
ANALYTICAL CHEMISTRY, VOL. 44, NO. 9,AUGUST 1972
