event, this anomaly, in which the sulfuric acid concentration exceeds the total sulfate, occurs in less than 2x of the more than 100 samples analyzed. Stability. Some field samples on filters of glass fiber were reanalyzed after 43 days. The results indicated a decay rate between 0.6 and 1 . 6 6 z per day. Because these samples were several weeks old when first analyzed, it would be difficult to extrapolate to zero time. We do not know whether the decay rate is a linear function. It may be possible to stabilize the sample by treating it with gaseous ammonia after collection. In any event, fresh samples should be analyzed to eliminate this phenomenon as a source of error. In conclusion, the recommended method is more sensitive, more reproducible, and more specific than the methods in general use for this toxic pollutant. It can provide accurate and specific measurements for particle size distribution of
these aerosols in a short time interval and can differentiate between sulfuric acid and inert sulfates. ACKNOWLEDGMENT
The authors express their thanks to George Morgan of the Air Quality and Emission Data Program, Standards and Criteria Development, National Air Pollution Control Administration, for supplying us with Hi-Vol Filters and for analyses of the total sulfates; and to J. P. Bell for assistance in the installation of the flame photometric instrument, RECEIVED for review December 17,1968. Accepted February 17, 1968. Presented in part before the Division of Water, Air and Waste Chemistry, ACS San Francisco, April 1968. Mention of commercial products does not constitute endorsement by the Public Health Service.
A Comparative Study of Premixed and Turbulent Air-Hydrogen Flames in Atomic Fluorescence Spectrometry M. P. Bratzel, Jr., R. M. Dagnall,' and J. D. Winefordner Department of Chemistry, University of Florida, Gainesville, FIa. 32601 Turbulent diffusion and laminar premixed air-hydrogen flames are critically studied as atomizers for several elements measured by atomic fluorescence flame spectrometry with respect to the effect of aspirating aqueous and aqueous-organic mixture solvents, the effect of nebulizing gas on the quenching of fluorescence, the effect of exciting radiation scatter as a result of incomplete solvent and solute evaporation, and the effect of flame height and background. The turbulent flame system results generally in greater or comparable sensitivities and limits of detection to the premixed flame system for most elements. Because the turbulent flame does not cause an abnormally large scatter signal and is not very sensitive to choice of nebulizing gas and because total consumption nebulizer burners are simple and safer to use, the turbulent air-hydrogen flame produced is recommended for atomic fluorescence studies of elements which are appreciably atomized in such a flame-e.g., Cd, Ga, Fe, Pb, TI, and Sn.
THEOPTIMUM ATOM reservoir and nebulizer burner for atomic absorption spectrometric measurements has been found to be the long-path premixed flame and chamber nebulizer burner. Recent developments in this area have been directed toward the use of new gas mixtures, the development of more efficient nebulization techniques, and the design of the burner head to increase the number of elements which can be determined. On the other hand, turbulent flames with totalconsumption burners have been used with great success for atomic emission flame spectrometry. The optimum flame shape for atomic fluorescence spectrometry is a circular flame of relatively small diameter, but the most suitable means by which such a flame is obtained is not clear ( I , 2). The chamber-type nebulizer-burner system
has a low solution uptake rate in comparison with most total-consumption nebulizer burners, which may result in a loss of sensitivity and is often susceptible to explosive flashbacks. On the other hand, total-consumption nebulizer burners produce relatively large solvent droplets (especially in the lower flame regions) which may cause a background signal and noise due to random scattering of incident radiation from droplets in which solvent evaporation is incomplete or from the resulting solute particles when solute vaporization is incomplete (3). Also, the entrainment of considerable quantities of oxygen from the atmosphere result in flames which are not extremely reducing and may result in greater quenching effects than in premixed flames. Finally, the flame flicker noise associated with a turbulent flame is greater than with comparable premixed flames and is often the major noise which influences detection limits in atomic fluorescence spectrometry. The extreme simplicity and greater solution uptake rates of total consumption nebulizer burners are sufficient to warrant a more detailed examination of its capabilities with respect to atomic air-hydrogen flame fluorescence spectrometry. In this communication, it is shown that flames produced by a total consumption nebulizer burner are generally equivalent to or superior to air-hydrogen premixed flames produced by a chamber-type nebulizer burner. Measurements are made of the atomic fluorescence characteristics of a broad range of elements in both types of flames with special reference to absolute sensitivity, solution uptake rates, effect of aqueous and mixed organic solvents, effect of viscosity and surface tension, position of measurement in the flame, effect of quenching, incident light scattering, and magnitude and 1
(1) J. D. Winefordner and T. J. Vickers, ANAL.CHEM.,36, 161 (1964). (2) R. M. Dagnall, K. C. Thompson, and T. S. West, Anal. Chirn. Acta, 36, 269 (1966).
On leave from Imperial College, London, S.W. 7, U. K.
(3) M. P. Parsons and J. D. Winefordner, ANAL.CHEM., 38, 1595
(1966). VOL. 41,NO. 6, MAY 1969
713
Table I. Sensitivity Comparison of a Premixed and Turbulent Flame in Atomic Fluorescence Spectrometry
Element Cd ( 5 ppm) Ga (100 ppm) Pb (100 ppm) Fe (100 ppm) Sn (1000 ppm) TI (10 ppm)’ TI (10 ppm)’
Wavelength of measuremeat, A 2288 4172 4058 2483 3034 3776 5350
Flame conditions TCa Airb H2 Airc H2 (l/min) (limin) 4.2 3.2 9.2 14.2 15.0 9.4 15.0 4.2 4.4 12.2 15.0 4.2 4.4 7.8 14.6 4.2 15.0 4.2 15.0 8.3 15.0 4.2 8.4 10.0 15.0 4.2 8.4 10.0 C.T.“
Burner htd C.T.“ T.C.” 3.5 10.0 3.0 5.5 3.5 10.5 3.5 4.0 3.5 5.5 3.5 5.5 3.5 6.5
-Atomic emission
Atomic fluorescence C.T.a,C T.C.a.6 C.T.a~6 T.C.” (Relative response) (Relative response) 0.00 0.00 0.50 1.oo 0.00 0.57 0.32 1 .oo 0.17 0.22 0.28 1.oo 0.00 0.82 0.73 1 .oo 0.00 0.16 0.39 1.00 0.00 0.46 0.46 1.00 0.02 0.29 0.29 1 .oo
C.T. = chamber-type nebulizer burner; T.C. = total-consumption nebulizer burner. Corresponds to pressure of 20 psi. c Corresponds to pressure of 11-20 psi. d Height of measured flame region was from top of burner head to center of monochromator slit, cm. e Relative to Zeiss burner, error f0.02. f Different photomultiplier voltage used. a
b
flicker of flame background. The elements studied include: cadmium, gallium, lead, iron, thallium, and tin. The most sensitive fluorescence lines of these ejements covero a fairly wide spectral range (from Cd-2288 A to Tl-5350 A). Several elements (Cd, Ga, Pb, and Ti) are easily atomized, whereas others (Fe and Sn) are relatively difficult to atomize. Some of these elements (such as Fe) are normally measured within the lower regions of a turbulent flame, while others (such as Pb) are measured in the higher flame regions. In fact, Cd and Pb can actually be measured above the luminous flame tip. The atomic fluorescence emission of several elements is cia a resonance-fluorescence process (Cd, Fe, TI) while that of others (Ga and Pb) is cia a direct-line fluorescence process. Thallium exhibits both resonance and direct-line fluorescence as well as a considerable amount of atomic thermal emission. Tin is included because its atomic fluorescence characteristics have not been previously observed. Gallium gives rise to a thermally assisted direct-line fluorescence signal. The atomic fluorescence for Cd and F e occurs in regions of low-flame background, while for Sn it occurs in regions of very high flame background. EXPERIMENTAL
The apparatus and procedure used were exactly the same as those described in a previous communication ( 4 ) . The Zeiss total-consumption nebulizer burner (Carl Zeiss, Inc., New York, N. Y.) was selected to produce turbulent flames because this burner was previously found to be superior for hydrogen-supported flames to other total-consumption nebulizer burners for atomic fluorescence spectrometry (5). This burner was compared with a chamber-type nebulizer burner to produce laminar flames (when using air-hydrogen flames, the nebulizer-burner assembly used with the PerkinElmer atomic absorption spectrophotometers was less susceptible to capillary clogging and also had a larger solution uptake rate than several other commercial chamber-type nebulizer-burner systems). The normal long-path atomic absorption burner head was replaced in our experiments by a circular burner head (3.5 cm in height, 2.0 cm in diameter) which contained 13 holes (1.0 mm in diameter), each sym(4) K. E. Zacha, M. P. Bratzel, Jr., J. D. Winefordner, and J. M. Mansfield, Ir., ibid., 40, 1733 (1968). (5) M. P. Bratzel, Jr. and J. D. Winefordner, Anal. Lerrers, 1 (l), 43 (1967). 714
0
ANALYTICAL CHEMISTRY
metrically arranged in a square pattern (1.0 X 1.0 cm). This head was connected to the chamber assembly uiu a rubber gasket. The aspirant and fuel gases were monitored by pressure gauges and rotameters. Slight variation in the air-flow rate with the chamber-type nebulizer burner and in the aspirant and gas-flow rates in the total consumption burner caused negligible change in the solution transport rate, RESULTS AND DISCUSSION
Relative Response in Water. I n Table I, relative atomic emission and atomic fluorescence signals are given for seven elements in air-hydrogen flames optimized for atomic fluorescence and obtained with turbulent flames (Zeiss burner) and with laminar premixed flames (Perkin-Elmer burner). No systematic errors resulted in the intensity ratios owing to change-over of burner systems. For each element, it is necessary to correct for flame background and thermal emission in accordance with the experimental system used. The amplifier gain, the photomultiplier voltage, and the slit-width settings are held constant for each element and only the flame gas mixture and the height of measurement in the flame are adjusted. The atomic fluorescence signals (see Table I) obtained with the total-consumption nebulizer burner are 2-3 times greater than those obtained with the chamber-type nebulizer burner, except for iron, where the signals are comparable. For iron, the measurements were taken low in the turbulent flame where there is a high background and incomplete evaporation of solvent droplets. It appears that the position of measurement is quite critical with turbulent flames but remarkably insensirice with premixed flames. Turbulent flames not only produced greater atomic fluorescence but also greater flame emission. This emission could be either thermal or chemiluminescent and in either case must be subtracted from the “apparent” atomic fluorescence emission when using a dc-measurement system. The increase in sensitivity with the Zeiss burner shown in Table I can be accounted for by the greater solution transport rate (about 6 ml/min) compared to the effective transport rate for the Perkin-Elmer burner (sample solution transport rate 9 ml/min times chamber aspiration yield = 0.33 mllmin). From the values in Table I, the aspiration efficiency of nebulized solutions and the atomization efficiency of the atomic species of interest are considerably less than 100% when the total-
60
w 40
2
1 W
a20 0
TI -37 76
Y
0
20 40 60 PERCENT ALCOHOL
100
80
0
Figure 1. Vanation in atomic fluorescence intensity with isopropyl alcohol concentration in the premixed flame 2288 A with air 4.2 l./min, HP 2.5 l./min, burner TI: 3776 with argon 4.2 I./min, HP 6.8 I./min, burner TI: 5350 with argon 4.2 I./min, Hz 6.8 I./min, burner Pb: 4058 A with air 4.2 I./min, HP 3.5 I./&, burner Cd:
0
4 4
ht ht ht ht
2.5 3.5 3.5 3.5
cm cm cm cm
consumption nebulizer burner is used with turbulent flames (particularly in the lower flame regions). In the higher flame regions, it becomes increasingly more difficult to illuminate the entire flame efficiently and to collect the fluorescence radiation. A further important consideration in the turbulent flame is the very high rise velocity which influences the time for solvent and solute evaporation. In the airhydrogen flame at large solution uptake rates, incomplete solvent and solute evaporation are expected. Smaller rise velocities in premixed laminar flames allow more time for efficient solvent and solute evaporation to occur. Relative Response in Organic Solvents. Prior atomic fluorescence measurements with premixed flames and chambertype nebulizer burners have shown that organic solvents result in an increased sensitivity, because of the increased solution uptake rate (2, 6). However, the effect of organic solvents with turbulent flames and total consumption nebulizer burners in atomic fluorescence spectrometry is not known. (6) R. M. Dagnall, T. S . West, and P. Young, Tulantu, 13, 803 (1966).
20 40 60 PERCENT ALCOHOL
80
100
Figure 2. Variation in atomic fluorescence intensity with isopropyl alcohol in the turbulent diffusion flame Cd: 2288 nA with air 9.1 I./min, H? 14.2 I./min, burner ht 10 cm TI: 3776 4 with argon 8.2 l./min, HZ14.4 I./min, burner ht 6.5 cm TI: 5380 9 with argon 8.2 I./min, HP 14.4 I./min, burner ht 6.5 cm Pb: 4058 4 with air 12.2 I./min, HP 15.0 I./min, burner ht 10.5 cm Sn: 3034 A with air 8.3 I./min, HP 15.0 I./min, burner ht 5.5 cm
In this study, experiments were carried out with cadmium, thallium, lead, and tin in various mixtures of iso-propyl alcohol and distilled water and in ethyl alcohol and distilled water. In Figures 1 and 2, the variation in atomic fluorescence signal with increasing concentration of iso-propyl alcohol in distilled water is given for the premixed flame (the chambertype nebulizer burner) and for the turbulent flame (the totalconsumption nebulizer burner), respectively. Tin was not included in Figure 1 because it gave a relatively low atomic fluorescence signal which was further obscured by the high OH-band background emission. In Figure 3, the atomic fluorescence intensity variation for lead with ethyl alcohol and distilled water mixtures in the premixed flame is given, and in Figure 4, a similar study for cadmium and lead in the turbulent flame is illustrated. With the chamber-type nebulizer burner and laminar flames, an increased atomic fluorescence signal (compared with
Pb
c ij=,6 0 Z W
w 40
2
t-
a
1
0' 0
i
20 40 60 PERCENT ALCOHOL
80
100
Figure 3. Variation in atomic fluorescence intensity with ethyl alcohol concentration in the premixed flame Pb:
4054
A with air
4.2 I./min, Hz 3.5 I./&,
burner ht 3.5 cm
0 ' 0