Analysis of the limiting noise and identification of some factors that dictate the detection limits in a low-power inductively coupled argon plasma system

Abstract

The background signal defined as the output signal ( x B) of the photomultiplier and the relative standard deviation (RSD) of x B at the wavelengths of 36 prominent analysis lines in the 190–320 nm region were measured under ICP compromise conditions for simultaneous multi-element analysis using a monochromator with an experimental spectral bandwidth of 0.015 nm. The RSD was found to vary systematically with x B according to the theoretically expected relationship RSD = (α B 2 + gβ/ x B) 1 2 where α B(≈0.5%) is the constant source flicker noise contribution, gβ/ x B the shot-noise contribution and gβ a coefficient containing the photomultiplier (PMT) gain g. Dark current detector noise was negligible, at least in part, owing to the use of lock-in amplification. The validity of relationship (0) was tested for various types of PMT and for both pure aqueous solutions and solutions with 1 to 3% w/v amounts of either calcium chloride, sodium chloride, or a mixture of nickel and cobalt nitrates. Only in the case of the nickel-cobalt matrix were some departures found and these were attributable to line coincidences. Relationship (0) was found to apply also to net line signals, the flicker noise term, α s, then being about 1% instead of 0.5% for the present ICP system. Detection limits for 36 prominent lines of the elements As, Au, B, Be, Bi, Ge, In, Mg, Mn, Ni, P, Pb, Sb, Se, Sn and Zn were computed on the basis of relationship (0) and measured signal-to-background ratios (SBR). The results permitted an assessment of the separate effects that SBR, source flicker noise and shot noise have on the detection limits and this facilitated a detailed comparison of the detection limits obtained in this work with those reported by Winge, Peterson and Fassel [ Appl. Spectrosc. 33, 106 (1979)]. The measurements of detection limits were extended to solutions with matrices (calcium chloride, sodium chloride, and a mixture of nickel and cobalt nitrates) to test the validity of the conclusions drawn from experiments with pure aqueous solutions under less idealized analysis conditions and to provide some results obtained with a low-power argon ICP in the scope of the “ICP Detection Limits Program”. These experiments included the measurement of the various quantities that may undergo an influence from the matrix, viz. background, net signal, SBR, and RSD of background signal. In this way it was possible not only to state the gross effect of the matrices on the detection limits, but also to give a quantitative account of the various sub-effects that were responsible for the gross effects. Since the calcium chloride matrix was included to assess the system for its stray light rejection characteristics, some experiments involving stray light elimination by a band rejection filter or a solar blind PMT were performed. A basic conclusion of this work is that the full exploitation of the important low ultraviolet wavelength region (190–250 nm) in ICP trace analysis requires the use of spectrometers with a high optical conductance, efficient entrance optics, and photomultipliers with high spectral sensitivity and low dark current noise. Only under such conditions can the “ideal” RSD of the background signal of 1% be realized, or at least approached, down to the lower end of the wavelength range, if one requires in addition that a reasonably high spectral resolution (e.g. 0.015 nm) be used to maximize the signal-to-background ratios and minimize spectral interferences.