2.2 Components of a Flame Atomic Absorption/Emission Spectrometer System
2.2.10 Instrumental Noise in Detectors
Chemists have little difficulty when they are measuring high concentrations of an analyte, but this situation is relatively rare. Most analysts attempt to determine if an analyte is present or not, and in doing so must measure the smallest amount of an analyte in a sample near the detection limit of the instrument. In many areas of science, such as environmental, forensics, and biological applications, the analyst is always struggling to lower the detection limit. This is referred to as “chasing the detection limit” since one of the goals of instrumental manufacture is to improve instruments to measure smaller and smaller quantities of analytes. The detection limit is basically the minimum concentration that one can measure above the “noise” in an instrument. To fully understand this concept, different types of noise must be discussed.
Environmental noise consists of factors in the immediate lab environment that will affect an instrument or sample. Obviously if an instrument is sensitive to vibrations, such as an NMR, one would not place this instrument in an area of high vibration, such as next to an elevator. Similarly an analyst would not want to make delicate ppm-, ppb-, and ppt-level concentration measurements in an environment where the analyte of interest is present in high concentrations (such as in the air near a metal smelter). Most environmental noise with respect to metal contamination can be avoided by locating the sample preparation area away from the instruments or locating each in a clean room or HEPA hood.
Chemical noise tends to be unavoidable and specific to each analyte. One of the simplest types of chemical noise, that is completely unavoidable, is the electronic and vibrational fluctuations present in an atom or compound due to the Maxwell distribution of energies. These result in broad UV-visible absorption and emission spectra in aqueous samples but are of little consequence in gas phase transitions experienced in AA and ICP. Other forms of chemical noise include slight temperature and pressure fluctuations that affect measurements.
Instrumental noise is common and in many cases can be avoidable or managed. Instrumental noise is separated into three categories: thermal, shot, and flicker noise. Thermal noise (also referred to as Johnson noise) results from the thermal agitation of electrons in resistors, capacitors, and detectors. Most of the thermal noise can be overcome by cooling specific components of the instrument such as is done in advanced detectors (i.e. charge injection devices). Shot noise results from a current being generated by the premature movement of electrons across a junction. The best example is an electron being emitted from a photoemissive material, such as in photocells or electron multipliers that are used to measure the intensity of visible and UV light. There are a few electronic ways of minimizing shot noise, but these are beyond the scope of our discussion here. Flicker noise results from random fluctuations in current and is inversely related to frequency. Flicker noise is overcome by electronically modulating the detector output signal to a higher frequency where less noise is present (i.e. from 102 Hertz to 104 Hertz).
In practice, the analyst is concerned with distinguishing between a real signal and instrumental noise, quantified as the signal to noise ratio. Signal to noise is mathematically defined as
where S is the mean of approximately 20 blank measurements and N is the standard deviation of the these measurements. From a statistical standpoint, it should be noted that S/N is equal to the reciprocal of the relative standard deviation (RSD).
Two approaches are used to minimize noise: hardware and software. Common hardware approaches that are used to decrease noise are (1) grounding and shielding components and detectors, (2) using separate amplifiers for different signals, (3) placing frequency or wavelength filters “up line” from the detector, (4) modulating the instrument signal to a “clean” frequency, and (5) chopping the signal to obtain a reference reading of the background that can be subtracted from the sample signal. For spectroscopy, one of the most common software noise reduction techniques is to take as many readings as reasonably possible. The observed signal to noise ratio (S/N) is a function of the number of readings (n) taken as shown by
where Sx and Nx are the signal and noise readings for a specific setting. Note that as one takes more readings the S/N decreases by the square root of the number of measurements. By taking two measurements, one can increase the S/N by a factor of 1.4; or by taking four measurements the analyst can half the noise. For the topics covered in this Etextbook, FAAS, ICP-AES, and ICP-MS instruments generally allow for multiple measurements over 5 to 10 seconds to be taken and averaged. And of course, the analyst can always analyze a sample multiple times given the common presence of automatic samplers in the modern laboratory.
| Frank's Homepage |
©Dunnivant & Ginsbach, 2008