Problem: As it is commonly available in most laboratories and does not require tedious sample preparation, most instrument manufacturers use the Raman emission spectrum from ultra-pure water to determine the sensitivity of their spectrofluorometer. To eliminate variations created by differing instrument parameters, the signal-to-noise ratio (SNR) of the intensity of the Raman peak compared to the noise present in a signal-free region is often used as a metric for reporting sensitivity, and thus instrument performance. However, there are different methods and techniques used to evaluate the SNR of an instrument.

Theoretically, the SNR can be calculated exclusively from the Raman peak signal fluctuations using the variation in the peak intensity over time or a time scan analysis. However, the dark noise is not the major cause of signal variation, therefore making this method unreliable for determining SNR in low signal measurements.

Solution: The most accurate method of measuring SNR is to ratio the peak Raman intensity to the average intensity of a signal free “baseline” region. The value of the intensity of the signal can be evaluated either by a single intensity data point extracted from scanning data or by averaging the peak intensity using a time scan.

There are also two methods for measuring the noise value of the SNR. The first method is based on measuring the fluctuations in a spectral region known to be free from signal. These random variations in the intensity values form the basis for calculating either the peak-topeak noise (maxmin) or the root mean square (RMS) noise (standard deviation) of the spectral response over a defined wavelength region. It is important with either measurement to have a statistically significant number of points to ensure a representative noise value. One problem with this measurement is the uncertainty about whether residual signal remains in the region that is assumed to be signal free. A baseline correction is frequently applied to remove any residual signal originating from broadband fluorescence or scattering.

Alternatively, the noise value can be determined from a time scan analysis. In this method, the noise is measured as a function of time at a single wavelength. As discussed above, either the peak-to-peak variation or the RMS variation is calculated using a large number of points from the resulting time scan.

While much of the sensitivity of a fluorescence spectrometer is directly related to the molar absorptivity and quantum yields of the molecules being analyzed, instrument performance is also critical to detecting low concentrations of fluorophores or small changes in larger signals. For measuring the SNR ratio, the Thermo Scientific Lumina fluorescence spectrometer, for example, uses a hybrid method where the value of the signal is determined by a scan of the Raman line. A time scan is used to evaluate the noise floor and to determine the noise value. The SNR is then determined from these two measurements.

For more information, visit www.thermoscientific.com.