Spectrometers

Many spectrometers exploit the wavelength-dependent diffraction on one or several diffraction gratings, or the wavelength-dependent refraction at one or several prisms. Typically, an incident beam is collimated (made parallel) before being sent to a grating or prism. After that dispersive element, different wavelength components propagate in slightly different directions. The light may then go through some additional optics and finally get to a photodetector.

For obtaining a spectrograph (a non-scanning spectrometer), the photodetector may be a photodiode array, CCD array or similar, recording intensities over some spatial range, which correspond to a certain spectral interval. One does directly obtain a spectrum by mapping the detector pixels to wavelengths. The data acquisition is then relatively fast, because all wavelength components can be measured simultaneously. The resolution is then typically limited by the density of detector pixels, or possibly by the optical setup. Spectral peak positions may be determined with an accuracy better than according to the pixel spacing by using interpolation.

In a scanning spectrometer, the detector can be a single photodiode or a photomultiplier, placed after a narrow optical slit, so that only one narrow interval of wavelengths can get to the detector at one time. The slit position or the angular orientation of the grating or prism may then be moved such that a certain wavelength range can be scanned, assuming that the PSD of the input remains constant during that time. The setup then functions as a tunable monochromator. Figure 1 shows the common design of the Czerny–Turner monochromator. The acquisition time for a full spectrum can be long if a wide spectral range is scanned with high resolution and if the detector cannot be very fast, e.g. because the low optical power enforces a substantial average time. Long acquisition times may not only be inconvenient, but a real problem if the properties of the light source are not constant.

Some grating spectrometers are extremely compact, with a width of only a few centimeters. However, the highest performance – in particular in terms of resolution and sensitivity – is obtained with much larger instruments. Typical wavelength resolutions reached with such spectrometers are about 0.01 nm to 0.1 nm.

Depending on the type of spectrometer used (e.g. a grating spectrometer), various details have to be observed:

• The input light often has to be sent onto an entrance slit of variable width. For highest spectral resolution, the slit should be made narrow, but that reduces the transmitted power and may therefore lead to increased noise or longer acquisition times, particularly for sources with low brightness. Some spectrometers have a fiber input, either with multimode fiber or single-mode fiber. Multimode fibers make it easier to collect light, while single-mode fibers allow for highest spectrometer performance.
• Diffraction gratings are often used with the first diffraction order, but sometimes with higher diffraction orders in order to obtain a better performance. In any case, there can be problems with artifacts related to light being diffracted on other orders. If one finds spectral features which are hard to explain, one may check whether they could be such artifacts.
• The response of a spectrometer may depend on the polarization, because the diffraction efficiency of a grating or the reflection loss on a prism arrangement is polarization-dependent.
• Some spectrometers have to be calibrated by the user. For wavelength calibration, one may use certain discharge lamps, emitting line spectra with precisely defined wavelength components. Calibration of the responsivity for the whole wavelength interval is often more difficult. One possibility is to use an incandescent lamp with a known filament temperature or calibrated spectrum.

A high spectral resolution, but only in a very limited spectral range, is obtained with various types of interferometric spectrometers:

• Some instruments are based on a Fabry–Pérot interferometer, where the mirror spacing is mechanically scanned, e.g. via a piezo actuator, while the transmitted optical power is registered. The usable spectral interval is the so-called free spectral range, determined by the mirror distance; this is typically of the order of 0.1 GHz to 10 GHz, i.e., very small in terms of nanometers. The resolution bandwidth is the free spectral range divided by the finesse, the latter being determined essentially by the mirror reflectivities. A large distances between the mirrors allows for a higher performance, but also leads to a narrow free spectral range.
• Fourier-transform spectrometers can contain a Michelson interferometer, where one arm length can be mechanically scanned over a long distance (millimeters, centimeters or even more). The detector signal versus time, recorded during a full arm length scan, has to be Fourier-transformed to obtain the optical spectrum. A simplified version is the wavemeter, specialized on precisely measuring only the wavelength of a laser source, rather than recording full spectra.
• Arrayed waveguide gratings are sometimes used for very compact spectrometers. They are based on interference effects in a small waveguide device.

A totally different principle of operation can be realized for the spectral analysis of broadband ultrashort pulses. One may simply send such pulses through a long fiber, which introduces substantial chromatic dispersion. The latter then results in substantially different arrival times of different spectral components after the fiber: a pulse with an original pulse duration of far below 100 fs may then be spread over several nanoseconds, for example. By analyzing that light with a fast photodiode and oscilloscope, for example, one can obtain information on the spectrum. Of course, one should make sure that optical nonlinearities are not invalidating the results; in fibers, this strongly limits the allowable peak power.

A very attractive property of that method is the high speed with which a spectrum can be recorded. A single ultrashort pulse is sufficient for that purpose, whereas a conventional scanning spectrometer, for example, may require many seconds for that and would average the spectrum over many pulses of a pulse train.

Traditional spectrometers work in the visible spectral range, with infrared or possibly with ultraviolet light. However, there are also spectrometers for operation in extreme spectral regions such as the extreme ultraviolet (EUV) or even the X-ray region, with wavelengths of only a few nanometers. Such devices may be based on diffraction gratings with very small line spacings, or in the X-ray region even on single crystals, exploiting the periodicity on the atomic scale. As photodetectors, one may use special X-ray CCD cameras or multichannel plate (MCP) detectors (→ photomultipliers).

Here you can submit questions and comments. As far as they get accepted by the author, they will appear above this paragraph together with the author’s answer. The author will decide on acceptance based on certain criteria. Essentially, the issue must be of sufficiently broad interest.

Please do not enter personal data here; we would otherwise delete it soon. (See also our privacy declaration.) If you wish to receive personal feedback or consultancy from the author, please contact him e.g. via e-mail.