Influence of Electrometer Noise on Limit of Detectability in Atomic

noise, thermal agitation noise, flicker noise, power supply hum, and micro- phonic hum (1). Each of these types of noise have been considered by Canno...
0 downloads 0 Views 278KB Size
Influence of Electrometer Noise on Limit of Detectability in Atomic Absorption and Atomic Emission Flame Spectrometry SIR: In papers on limit of detectability in atomic emission (4) and atomic absorption ( 5 ) flame spectrometry, the noise signal was defined as the root-mean-square fluctuation in the background anodic current of the phototube. However, in some cases the electrometer noise must be considered, as well as the noises previously (3-5) considered. The electrometer noise is primarily caused by five sources: shot noise, thermal agitation noise, flicker noise, power supply hum, and microphonic hum ( I ) . Each of these types of noise have been considered by Cannon ( 1 ) . In literature concerning commercial electrometers ( 2 ) and in measurements of electrometer noise in these laboratories, electrometer noise is most conveniently listed as a certain fraction of the input signal of the electrometer. In this manuscript, it is also more convenient to refer electrometer noise to the input ( d ) . The total noise signal and the signal caused by the sample are also conveniently referred to the input of the electrometer. The influence of electrometer noise on the limit of detectability and optimum slit widths in atomic emission and atomic absorption flame spectrometery will be considered. In atomic emission flame spectrometry, the total noise signal introduced into the electrometer input is given by

-

where Ai, is the root-mean-square fluctuation current caused by noise in the phototube, 2 is the root-meansquare fluctuation current caused by flicker in the flame source, and is the root-mean-square fluctuation current per second-1/2 in the electrometer. Because these noises are independent, they will add quadratically as indicated above. The electrometer noise is assumed to be white as previously indicated for other types of noise ( 4 ) , and thus Affl the frequency response band width in seconds-', of the detectorreadout y st e m must be included. The value of Ai, ran be simply measured, and so no espression for it is necessary. The value of N,, the minimum detectable number of atoms per cubic centimeter of flame gases, differs from the previous (4)only in that a n 416

ANALYTICAL CHEMISTRY

-

-

additional noise term, AiG2Afmust also be included within the square root.

=

3.8 X B(T)eEdkT X v,g,illTJTi~t'H(;l/F2)

{2e,B?dAf[id + yT,WH(S/F') ICs]+ [rTlWH(A/F2)El,sl2Af4- Z e 2 A f ) " ' (2)

,111 terms have been previously defined (4). The exact equation for W,, the optimum - slit width, differs only by a term Aie2-i.e.,

+

x o w if sc