Unfortunately, there are a number of different ways of handling the data, all of which are valid but which will give quite different numbers. Therefore, it is important not only to know how the water Raman S/N values is measured, but also how the data were treated.
In general, the water Raman S/N test method combines a value for system sensitivity (in the presence of a signal) with a value for system noise (in the absence of signal) to show the overall performance of the instrument.
At HORIBA Jobin Yvon we define the S/N ratio as the difference of peak and background signal, divided by the square root of the Background signal. The peak signal is measured at the water Raman peak (397 nm for 350 nm excitation) and the noise in a region (450 nm) where no Raman signal is present, and an "ideal" system would give a signal value of zero.
Another commonly used method is to divide the difference (Peak signal - Background signal) by the rms value of the noise on the background signal. This second method is used by a few other manufacturers.
Some actual data from our FluoroLog FL3-11 system (this was a typical system, a few years old) will serve to show the difference between the two methods.
The experimental conditions were as follows:
Excitation 350 nm with 5 nm bandpass
Emission 360 - 450 nm with 5nm bandpass
Interval 1nm
Integration 1s
No smoothing of data points
Standard room temperature, red sensitive, detector (Note:
make sure the test is caried out with the actual detector
you will be using. All SPEX® systems are specified with a
R928P PMT at room temperature).
The
measurements provided the following data:
Peak signal (at 397nm) = 501,500 cps
Background (at 450 nm) = 10,500 cps
Peak to peak noise of background (at 450 nm) = 223c (measured with a separate kinetic scan), which gives an rms noise of the background signal of 223/5 = 44.6
Therefore, the HORIBA Jobin Yvon method gives a water Raman S/N of (501500-10500)/ (10500)½ = 4791
The second method similarly gives a water Raman S/N of (501500-10500)/ 44.6 = 11008
HORIBA Jobin Yvon feels that the first method is correct although it gives a lower number. The second method only takes into account the detector noise and the shot noise of the electronics.
On the other hand, by using the background total intensity as a measure of noise, the HORIBA Jobin Yvon method is more representative of a real "live" experiment where noise is also influenced by factors like the quality of the optics and scattered light in the system. These additional factors will influence the ability to measure a very low signal from a sample and should not be left out.
