There are various terms to express how sensitive a spectrometer system is, such as efficiency, throughput. The typical question is: I have x amount of photons, how many of them are ending up on the detector and what signal is generated by these. Finally, the signal/noise value is important. In the following we consider how individual components are influencing the flow of light. Key is that all these “transfer” functions are wavelength dependent.
It all starts with a light source, which emits x amount of photons, often from a wider area over a wide range of emitting angles. Only a fraction of that light is captured. This depends on the geometry of the optical path respectively what components are used.
A typical setup is to use a lens to collect and then focus the light into an optical fiber. The source area and the magnification of the lens setup is one defining equation, the other the acceptance angle. Alternatively, light might be picked up by one or various fibers in direct coupling. It is advantageous if the distance emitting area to fiber is short.
In some cases, the light source can be used directly for illumination, e.g. small halogen bulbs are available with integrated lens. All in all less than 40% of the optical energy is typically collected, depending on the source it could be a lot worse.
For a broad-range incl. UV the demand for the optics is rising, it has to be UV transmissive plus achromatism becomes more important. Back to top▲
A light guide accepts light from an acceptance cone given by the fiber design. The acceptance angle is maintained within a fiber. At the light source end, the acceptance cone cannot be large enough, but the limiting factor is the other end, for a sample setup or finally the spectrometer cannot accept light of bigger larger angles.
The fiber or bundle diameter should be as large as possible, again the end to the sample is often the limiting factor.
At the interface of air-to-fiber so-called Fresnel losses reduce the intensity entering an leaving the light guide by about 4% (quartz fibers). A bundle reduces the amount of light transferred by fill-factor losses (> 30%). Within the fibers damping reduces the light level, although in the VIS and NIR and at the typical length of less than 10m the absorption is almost negligible. Of course, it is important to select the right fiber quality. If necessary, bundles with two different fiber qualities can be made.
The efficiency of measuring heads and probes are varying significantly. While a fiber-optic immersion probe has efficiencies in the range of 30 to 40%, diffuse reflection heads have a lot lower. Back to top▲
The spectrometer is the component which is in conjunction of efficiency most heavily influenced by design. It all starts with a good coupling of the light coming from the probe to the entrance slit. In case a round fiber is connected to a slit, then the slit slices a fraction out of the fiber emitting area, thus throwing away most of the light. If the slit has a high aspect ratio, cross-section converters can improve the throughput by an order of magnitude.
The entrance slit are is directly proportional the overall light throughput, but only, of all this light can be effectively handled by all the other components. Therefore, an effective imaging from the entrance slit to the exit slit / detector pixels is key.
A small entrance slit also limits the light throughput and thus the sensitivity. Typical considerations: to find the best trade-off between resolution and sensitivity! To be kept in mind: the better the resolution the lower the sensitivity!
Key element influencing the overall sensitivity of a spectrometer is the grating. To achieve high diffraction efficiencies, gratings have to be blazed, a special technique to give the grating groove faces a predominant direction. Even a blazed grating has no high efficiency over the full spectral range. After a steep rise from the shorter wavelength to the maximum follows a moderate decrease towards the long wavelength. Back to top▲
For effective transportation of the light from entrance slit to detector either a lenses or mirrors are required or a so-called imaging grating has to be used. The latter has the advantage that no additional element/surface is introduced into the light path. If the imaging grating is also flat-field corrected then most of the light is imaged effectively on the detector area and the focal plane is close to the detector plane. If this imaging is not done with good quality (either by lenses, mirrors or gratings) the light is not ending up on the detector this light is lost. Furthermore, if the focal position of the image does not match to the detector plane the spectrum gets blurred lowering the resolving power.
The final element in the overall light path is the detector. Typically, the larger the detector respectively the pixels, the more light can be collected. It is important that detectors have a high quantum efficiency (QE), the percentage of how many photons are transferred to electrons, and a large well depth, i.e. many electrons contribute to the signal. The more pixels have to be illuminated the less light ends up at the individual pixels, so again: the better the resolution the lower the sensitivity! Back to top▲
At last, a remark about electronics: each detector signal can be amplified almost limitless, so huge electronic signals could be generated. The problem is that the noise or the uncertainty of the signal is amplified in a similar way. Therefore, the real statement about a signal quality is the signal-to-noise ratio. Back to top▲