Optimum Parameters for Flame Emission Spectrometry with the Nitrous Oxide- Acetylene FIa me Gary D. Christian Department of Chemistry, University of Kentucky, Lexington, K y . 40506
Fredric J. Feldman Beckman Instruments, Inc., Clinical Instruments Operations, 2500 Harbor Boulevard, Fullerton, Calif. 92634
IN 1966, AMOSAND WILLIS( I ) introduced the nitrous oxideacetylene flame for the atomization of refractory elements and their atomic absorption determination. Its advantages for atomization accrue from the highly reducing atmosphere maintained in the red feather of the flame (2) or the lack of oxygen containing species ( 3 ) in addition to its relatively high temperature. These same factors, combined with the high excitation energy of the flame, have led to an interest in employing it for flame emission spectrometry, and it indeed appears at present to be the flame of choice for the flame emission analyses of most elements in terms of sensitivity and convenience (4). In 1967, Amos (5) reported that the flame was superior for the emission of the rare earth elements and aluminum. In 1968, Pickett and Koirtyohann (6) described the general emission characteristics for 34 elements in aqueous solution, using the premixed flame with a slot burner. In the same year Christian (7) demonstrated the emission of 52 elements in this flame; in this preliminary study, 183 emission lines and bands were investigated for possible analytical use, using both aqueous and aqueousacetone solutions. Conditions were not optimized, and relative detection limits were reported using an instrument without filtering of the photomultiplier signals. In spite of these restrictions, only selenium did not give a measurable emission in aqueous solution. Pickett and Koirtyohann have more recently reported on the detailed studies for lithium and the alkaline earth metals (8) and for the elements aluminum, gallium, indium, thallium, germanium, and tin (9). Knisely et a/. (10)recently reported detection limits for the rare earth elements using the nitrous oxide-acetylene flame. Based on the previous preliminary investigation (3, the present study was undertaken to determine the conditions for maximum signal/noise ratio for the flame emission spectrometry determination of all possible elements using the nitrous oxide-acetylene flame and a slot burner. The present report summarizes the results for those elements for which optimum conditions have not been previously reported. (1) M. D. Amos and J. B. Willis, Spectrochim. Acta, 22, 1325 (1966). (2) G.F. Kirkbright, M. K. Peters, and T. S. West, Talarzta, 14, 789 (1967). (3) J. D.Winefordner and T. J. Vickers, ANAL.CHEM.,42, 206R (1970). (4) E. E. Pickett and S . R. Koirtyohann, ibid., 41 (14), 28A (1969). (5) M. D. Amos, The Element, No. 17 (1967). (6) E. E. Pickett and S . R. Koirtyohann, Spectrochim. Acta, 23B, 235 (1968). (7) G. D.Christian, Anal. Lett.. 1. 845 (1968). (8) S . R. Koirtyohann and E. E. Pickett, Spectrochim. Acta, 23B, 673 (1968). (9) E. E. Pickett and S . R. Koirtyohann, ibid., 24B,325 (1969). (10)R. N. Knisely, C. C. Butler, and V. A. Fassel, ANAL.CHEM., 41, 1494 (1969).
Conditions reported include wavelength, slitwidth, photomultiplier voltage, flame stoichiometry, and height of measurement in the flame. Detection limits are given for aqueous solutions of each of the 22 elements reported. This report is an extension of previously reported results using conditions for maximum signal/noise ratio (11, 12). EXPERIMENTAL
Solutions were prepared from reagent grade or high purity salts or oxides of the elements, or from high purity metals as previously described (7). Flame emission measurements were made with an Instrumentation Laboratory Model 153 Atomic Absorption/Emission Spectrophotometer. The mode of operation was similar to that previously described (7) except that the demodulation section was bypassed to allow the signal from the photomultiplier to reach the lock-in amplifier. This allowed the emission signal to be amplified, integrated, and read directly on the digital counter. The design of the instrument has been discussed (23). The spectrtl bandpass for different $it settings was as follows: 40 p, 2 A ; 8 0 p , 4 A ; 160p, 8 A . A 1P28 photomultiplier was used as the detector unless otherwise specified. Elements that emit at long wavelengths were determined using a redsensitive 1P21 photomultiplier tube. Theje include potassium (7664.9 A) and rubidium (7800.2 A). A red filter should be used with these elements to eliminate second-order interference from the flame background. The burner used for emission measurements was the Instrumentation Laboratory 24036 nitrous oxide burner head. This has a stainless steel head with a single slot (0.02-in. X 5 cm) along the optic axis. The gas flow as adjusted at the inlet gauges to give the desired fuel/oxidant ratio (ca. 10 1. NzO/min and CzH2 adjusted at around 6 l./min). Solution aspiration rate was set at 4 ml/min for all measurements. Emission spectra of all the elements were recorded to ascertain their line emission. o
RESULTS AND DISCUSSION
Optimum conditions were determined for maximum emission sensitivity. The more sensitive lines from the previous report were studied in detail (7). Table I summarizes the results along with the detection limits determined for aqueous solutions of the elements. Signals were integrated for 10-sec periods for detection limit studies. The detection limit was defined as the concentration, in parts per million, that resulted in a signal equal to twice the standard deviation of the background fluctuations. The standard (11) G. D. Christian and F. J. Feldman, 157th National Meeting, American Chemical Society, Minneapolis, Minn., April 13-18, 1969. (12) G. D. Christian and F. J. Feldman, 21st Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 1-6, 1970. (13) F. J. Feldman, Res./Del;elop.,20 (IO), 22 (1969). ANALYTICAL CHEMISTRY, VOL. 43, NO. 4, APRIL 1971
611
~~~
~~
Table I. Optimum Conditions for Flame Emission Spectrometry with Nitrous Oxide-Acetylene Flame Photomultiplier Red feather Height of Detection Element Wavelength, A Slitwidth,p voltage height, mm observation, m m limits, ppm As 1937.0 160 lo00 3 10 10 B 51800 80 900 1 8 0.05 Be 2348.6 80 900 15 10 1 Bi 3067.7 40 800 8 10 20 cs 4555.3 160 800 5 5 0.6 Hf 5311.6(II)b 80 800 40 10 20 2536.5 160 800 8 9 10 Hg 2 9 3 Ir 3800.1 80 800 55005 160 800 5 6 0.4 K 7664.9 80 800 5 5 O.ooOo5 Na 5890.0 80 800 6 6 0.0005 os 4420.5 160 800 15 5 2 Rb 7800.2 160 800 5 8 0.008 1947.6 160 800 5 8 3 Rh 3434.9 80 800 5 7 0.03 Ru 3728.0 160 800 5 5 0.3 Sb 2528.5 160 800 15 8 0.6 Se 1960.3 160 lo00 20 8 100 Si 2516.1 160 800 14 8 3 Ta 4740.2 160 800 11 6 4 Te 4866.2 160 800 6 7 2 3 20 800 35 8 10 Th 4919,8(II) U 54480 160 800 10 6 5 Zn 2138.6 160 900 5 7 10 Band emission. Ion line. (1
deviation of the background fluctuations was calculated from ten integrated readings. The concentration of the test element was within a factor of ten of its detection limit and the average of ten integrated readings was taken for the deflection. In addition to the elements reported in the table, phosphorus (5259 A) and sulfur (3837 A) band emission was studied. The detectability of these two elements was very poor in the nitrous oxide-acetylene flame and an argon-hydrogenentrained air flame is recommended for these elements. The flame/oxidant ratio is important, primarily for three reasons. First the intensity of background band emission from the flame changes markedly with flame stoichiometry. The flame emission spectrum of the red feather (7) is observed except in fuel lean flames where the quieter background of the blue zone (7) is observed. The NH, CH, CP bands decreas? as the flame is made leaner while the OH band (3064 A) increases (6). Second, ionization of the atoms is somewhat more prominent in the lean flame but this can generally be minimized by adding a high concentration of an ionization suppressant, such as potassium. Third, the formation of refractory oxides by certain elements is dependent on flame conditions, and a carbon-rich flame is generally preferred for these elements. Because of the quieter flame background as well as the enhanced excitation energy, a noncarbon-rich flame is generally preferred when possible. Thus, iridium, for example, exhibits maximum detectability in a lean flame. The slit width used is a compromise between resolution and response. A narrow slit is desired when possible to minimize spectral interference. But for those elements whose emission intensities are weak, a wide slit is required in order to permit the maximum amount of light to fall on the detector. Arsenic and zinc are examples. An increased photomultiplier voltage is also used to enhance the signal output, being limited by the increase in the electronic noise. When a 1P28 photomultiplier is used at very long or very 612
ANALYTICAL CHEMISTRY, VOL. 43, NO. 4, APRIL 1971
A;
A,
short wavelengths (e.g., Li, 6707.8 As, 1937.0 Se, 1960.3 A) where its sensitivity to the radiation is low, then increased slitwidth and/or photomultiplier voltage is required. Boron and uranium were determined using band emission because of the poor sensitivities of their line emission. A lean flame is used for maximum formation and excitation of their oxides coupled with minimum flame background. Determinations using band emission are, of course, less specific than by using line emission. Uranium exhibits several strong emission lines (2635.5,2865.7, 3812.0,4090.1 A) but they fall on strong flame bands. Thus, a lean flame is required to eliminate background interference with these, but it results in uranium oxide formation and cannot be used. All detection limits were determined in the absence of ionization suppressants. They are often enhanced twoor threefold, however, in the presence of an ionization suppressant. An ionization suppressant should generally be added with those elements having ionization potentials less than 7.5 eV (4, including the alkali, alkaline earth, and rare earth metals, and aluminum, gallium, indium, thallium, tin, lead, scandium, titanium, vanadium, chromium, manganese, yttrium, zirconium, niobium, molybdenum, and tantalum. High concentrations of sodium exhibit an appreciable NaOH continuum in the nitrous oxide-acetylene flame and a correction for this added background emission should be made (14). Potassium is a better ionization suppressant than sodium. Background continuum from the sample matrix should always be considered a possible interference. This can be corrected for either by scanning the spectrum in the region of the emission line or by taking a background reading at a wavelength about twice the spectral slitwidth removed from the line, while aspirating the sample. In the latter case, the average of the background readings on both sides of the line should be taken for correction. (14) C. E. Matkovich and G. D. Christian, unpublished data.
The 16-A bandpass used for the measurement of thorium would allow the passage of a number of relatively strong Thrr lines as well as a few moderately strong Th’ lines. The sensitive hafnium line at 5311.6 is an ion line. Some elements exhibit maximum sensitivity using band emission, but this is limited in its selectivity. Thus, for example, marked improvement in sensitivity is found with band emission from iridium, in addition to boron and uranium. Beryllium also shows fivefold higher sensitivity at the Be0 4708.6 band. Most of the detection limits reported here can undoubtedly be improved by using organic solvents (7), although there is increased danger of chemical interference (4). Where
A,
A
applicable, flame emission spectrometry with the nitrous oxide-acetylene flame offers decreased chance of chemical interference for those elements (somewhat over half) that are normally determined in cooler flames by atomic absorption spectrometry. RECEIVEDfor review June 1, 1970. Accepted December 18, 1970. The work on which this report is based was supported in part by the Office of Water Resources Research, Department of the Interior, under provisions of Public Law 88-379, as Project Number A-01 30-KY. Experimental work by FJF done at Instrumentation Laboratory, Inc., 113 Hartwell Ave., Lexington, Mass. 02173
Method for Rapid Determination of Adsorption Properties of Molecular Sieves George R. Landolt Research Department, Mobil Research and Development Corporation, Paulsboro, N . J. 08066
ADSORPTION PROPERTIES of molecular sieves are important criteria for their many and specific applications. Zeolitic properties such as pore-channel structure, surface area, rate of crystallization, changes produced by ion exchange, thermal stability, and others may be determined by adsorption. Adsorption properties have been investigated by McBain ( I ) , using a quartz spring balance. The McBain balance and the Cahn electrobalance, although more sensitive, both suffer the disadvantage that only one sample can be tested at a time. We have designed and constructed a simple apparatus to rapidly and semiautomatically determine zeolite sorption properties (2). The apparatus is unique in that many zeolite samples may be treated simultaneously.
ZlMMERLl GAUGE MANOSTAT I
DSORPTION CHAMBER
0
VACUUM PUMP
2 9
.
EXPERIMENTAL
Apparatus. The apparatus is shown schematically in Figure 1. It consists of a glass adsorption chamber (holding up to 11 zeolite samples in 15-ml weighing bottles) sealed by a viton O-ring recessed in an aluminum cap. The adsorption chamber is connected by means of a common manifold and solenoid valve to three separate adsorbate reservoirs. Water, n-hexane, and cyclohexane are used, representing adsorbates of different molecular size or volume. The adsorbate pressure is normally controlled by a mercury manostat with preset contact points (12 and 20 mm), which operates the solenoid valve by means of a relay. The complete range of adsorbate pressures can be set manually by means of the Zimmerli gauge. Kontes high vacuum O-ring seal stopcocks with Teflon (Du Pont) plugs are used to prevent contamination of the adsorbates. The complete system is evacuated by a rotary vacuum oil pump equipped with a dry ice-acetone trap. Air or nitrogen can be introduced to the sorption chamber through an attached drier containing activated Linde 13X pellets and Drierite. All of the component parts are readily available, and the entire apparatus may be assembled for approximately $500. (1) James W. McBain, “The Sorption of Gases and Vapors by Solids,” G. Routledge & Sons, Ltd., London, 1932. (2) G. R. Landolt, Abstracts, 158th National Meeting of the American Chemical Society, New York, N. Y . , September 1969.
Figure 1. Adsorption apparatus
Procedure. Equilibrium sorption capacities are determined at ambient temperatures (21-27 “C) with humidity control at 30-50z. The zeolite samples are preactivated in air at a prescribed temperature, normally about 350 “C, for 2-16 hours. Meanwhile, the desired sorbate reservoir and manifold are evacuated. The samples are then cooled to room temperature in a desiccator containing activated Linde 5A sieve. Approximately 1-gram samples are rapidly transferred to tared 15-ml weighing bottles, stoppered, and reweighed. The bottles are then placed in the sorption chamber and the stoppers rapidly removed prior to closing the chamber to minimize exposure of the samples to air. The system is evacuated to lova Torr, and the desired sorbate is then admitted and its pressure maintained at the preset value (the pressure is chosen to give a relatively low partial pressure so that the uptake is due mainly to intracrystalline adsorption). Normally, the time required to reach equilibrium is approximately 1 hour for C,-Clo hydrocarbons, and 3 hours for water. The samples are then removed, re-weighed, and the amount ANALYTICAL CHEMISTRY, VOL. 43, NO. 4, APRIL 1971
0
613
