erivation of Expressions for Calculation of Limiting Detectable Atomic Colncentration in Atomic Fluorescence Flame Spectrometry J. D. Winefordner, M. L. Parsons,’ J. M. Mansfield, and W. J. McCarthy Department of Chemistry, University of Florida, Gainescille, Fla. 32601 An expression for the limiting detectable atomic concentration in atomic fluorescence flame spectrometry is derived using signal-to-noise ratio theory. The variation of the limiting detectable atomic concentration is considered with respect to the type of source of excitation (narrow line or continuous), the source intensity and flicker, the flame background intensity and flicker, the power efficiency of the fluorescence to absorption transitions, the monochromator slit width, the frequency response band width of the amplifierreadout system, and the amplifier type (d.c. or a.c.). Values of the limiting detectable atomic concentration are given for Cd-2288 A, Cu-3248 A, and Na-5890 A as a function of the above parameters for a practical and useful experimental system. Correlation of the limiting detectable atomic concentration to the limiting detectable solution concentration is also considered.
Sample calculations are given for three elements (Cd, Cu, and Na) with widely different resonance line wavelengths, THEORY
The limiting detectable signal in atomic fluorescence flame spectrometry is defined (1, 2) by
1 Present address, Phillips Petroleum Co., Research and Development Department, Bartlesville, Okla. 74004
where iF(,,,)is the minimum detectable photoanodic detector signal, in amperes, due to atomic fluorescence, 2, is the rootmean-square noise, in amperes, a t the photoanode caused by all noise signals, and n is the number of signal-background pair readings. Equation 1 is valid from a statistical standpoint. A signal-to-noise ratio of 3dZ/d;results in a minimum detectable signal with about a 99.5x confidence (8). For all calculations performed in this manuscript, parameter n was chosen to be 5, which yields 99.5% confidence a t a signal-to-noise ratio of 2 ( I ,2). The signal-tonoise ratio in any given experiment can be estimated by measuring the detector signal, in amperes or volts, and dividing by about one fifth of the peak-to-peak noise signal, in amperes or volts (9, 10). ‘The same signal-to-noise ratio results whether photodetector anodic current, voltage drop across the phototube load resistor, or voltage at the readout system is measured. For the equations developed below for the fluorescence signal and noise terms, it is necessary only to make the following assumptions: that a flame sample cell (or other cell capable of producing species in the atomic state) is positioned in front of the entrance slit of a monochromator such that the complete cell width is observed [however, if more than one source image is focused on the flame gases, this can be treated by the approach used by Winefordner and Vickers (1) in the atomic emission paper]; that the complete flame cell width is excited by either a narrow line or a continuous source; similarly if more than one image of the flame gases being measured is focused on the entrance slit of the monochromator, a suitable factor is introduced as described below; and that the atomic fluorescence signal results from a single, spectrally isolated transition. The general case in which there is more than one line that can be excited to produce a fluorescence signal is a simple extension of the expressions given here (11, 12). Also it is assumed that a monochromator with equal entrance and exit slits is used. This is generally
(1) J. D. Winefordner and T. J. Vickers, ANAL.CHEM., 36, 1939 (1964). ( 2 ) Ibid., p. 1947. (3) J . D. Winefordner and C. Veillon, Ibid.,37,416 (1965).
