Optical detectors used in spectroscopic instruments are often classified as either single-channel detectors (SCDs) or multichannel detectors (MCDs).
Single-channel detectors have one active sensing element that acts as single transducer. Photons reaching the detector, within its operating wavelength range, are absorbed by the active material of the detector and encoded as an electrical signal. The output signals produced by the detector vary according to the detector specifications, but generally include analog (voltage or current) and digital (pulse-counting) domains.
A spectrometer with its PMT is operated by moving the diffraction grating to present different wavelengths to a focal point at the exit slit, where the PMT will sequentially record the signal, one wavelength at a time. In contrast, multichannel detectors have multiple active sensing areas which collect many wavelengths simultaneously at the focal plane of a spectrograph. No exit slit is used in this arrangement. Each arrangement has advantages and disadvantages, and it is the application requirement that defines which configuration is most suitable.
Categories of single-channel detectors: photoemissive devices, internal photoelectric devices, and thermal detectors.
Probably the best known detector is still the classic PMT, which offers good performance in detecting photons across a relatively wide wavelength range at modest cost. However, many types of single-channel detectors are available on the market today. They are classified according to the method of optical to electrical conversion. First, a photon is absorbed by the material creating an electron-hole pair. If the generated photoelectron is further emitted from the material, becoming available for collection or multiplication, the device is called a photoemissive device, or one based on the external photoelectric effect. PMTs are the most common example of this type of detector.
If instead no emission takes place but the photogenerated electron-hole pair is available for the current circulation in an external circuit, we call this an internal photoelectric device. Solid state detectors fall in this category, but may be subdividedinto photovoltaic (photodiodes, for the visible-NIR range) and photoconductive (or photoresistive, for the MIR). Although photoemissive devices frequently have gain, providing higher sensitivity, they have the drawbacks of vacuum tube technology, requirement for high-bias voltage and limited spectral range of operation.
A third category of photodetectors is thermal detectors. These operate according to a two-step process: first, radiation is dissipated in an absorbing material; then the resulting small increase in temperature is measured by a change in the electrical properties (such as resistance) of the material. Although these detectors often cover a broad spectral range (visible - MIR), they have much lower sensitivity compared to the other categories of photodetectors and are not often used in spectroscopy except in limited cases in MIR systems. Table Table 7 summarizes typical characteristics of these devices.
Single-channel detectors (SCDs) have one active sensing element that acts as single transducer. Photons reaching the detector, within its operating wavelength range, are absorbed by the active material of the detector and encoded as an electrical signal.
While they are excellent photon transducers, optical detectors exhibit some intrinsic background, even when no optical signal is present. Most detectors used for spectroscopy benefit from cooling to reduce this background signal. As the detector is cooled below room temperature, often to very low operating temperatures, the background signal and its associated noise component is reduced. Common cooling techniques involve thermoelectric as well as cryogenic cooling using liquid nitrogen. To eliminate condensation and the potential for corrosion, the sensors themselves are often mounted in a vacuum housing and sealed using hermetic methods, ensuring years of maintenance-free use.
The first SCDs selection criterion is the wavelength range to be measured. The next consideration is whether the measurements require temporal analysis.
Detectors are selected based on experimental requirements.
The first selection criterion is the wavelength range to be measured. In many cases this will narrow the choice down to one or two detectors. For example, if the wavelength range is 7 - 15 μm, an HgCdTe detector will be the only option. However, if the wavelength range is 400 - 700 nm, there are several options with the choice of optimum detector depending on other experimental requirements. For steadystate measurements with high light levels and a sample that is not easily damaged by exposure to light, either an uncooled PMT; or uncooled or TE cooled silicon photodiode (especially if future research plans involve NIR measurements) would be a good choice. For extremely low light levels, a PMT with photon counting will be the best choice.
Table 8 shows the spectral response for the most commonly used solid state detectors. These detectors are most commonly used for steady-state measurements, but in certain cases, when coupled to a fast amplifier, may also be used for time-resolved measurements. When two detectors cover a similar wavelength range, such as InGaAs and Ge, the choice is based on gain, noise factors and response at a specific wavelength of interest. Ease of operation, based on not needing liquid nitrogen may also be a factor.
The next consideration is whether the measurements require temporal analysis. For example, fluorescence lifetime measurements provide a much more detailed observation of the molecular processes that occur in biology and biophysics, materials science, and chemistry than do steady-state measurements. The fluorescence lifetime τ of these materials generally ranges from a few picoseconds to hundreds of nanoseconds. Fig. 43 shows an example of a fluorescence lifetime measurement on KCl:Eu2 , obtained using Time Correlated Single Photon Counting (TCSPC).
A PMT can be optimized to perform picosecond lifetime measurements. With a carefully tuned fast amplifier-discriminator circuit, PMTs can be used for TCSPC measurements to obtain lifetimes of biological, nano materials and other samples.
There are 2 basic techniques for measuring fluorescence lifetimes: frequency domain measurements (such as in our MF2 instrument) and time-domain measurements (including TCSPC and transient decay methods).
By careful design of the electronics, a relatively inexpensive PMT can be optimized to perform picosecond lifetime measurements. With a carefully tuned fast amplifier-discriminator circuit, these PMTs can be used for TCSPC measurements to obtain lifetimes of biological, nano materials and other samples. The most popular models use a multi-alkali photocathode and cover the spectral range 185 - 850 nm, with some optimized for a particular subsection of this range. Other PMTs with different photocathode composition are optimized to extend the spectral response up to 1 micron. Variants that extend beyond 1 micron are available, however, these are costly and require careful handling as they can be easily damaged by excess light, rendering them less effective in the red and possibly destroying the tube. Generally, due to cooling requirements imposed by the higher thermionic emission, these models extend the usable range to approximately 1.6 microns. Photoluminescence (PL) and Electroluminescence (EL) lifetimes may be measured using the same experimental methods.
Another single-channel detector used in spectroscopy is the avalanche photodiode (APD).This is a small device made of silicon or InGaAs, which may be operated in single photoncounting mode (Geiger Mode), sometimes abbreviated “SPAD”, which has very high gain and can be used to measure very low light levels. The disadvantage of this device is that it has very high dark current and associated shot noise, even for a detector with a diameter 80 microns or less, and must be cooled. Another new type of singlechannel detector with potential use in spectroscopy is the discrete amplification photon detector (DAPD), which uses a novel approach to reduce the excess noise factor inherent in Geiger-mode APDs from approximately 1.3 to less than 1.05 with gain of 105 with nanosecond rise times. These devices may be produced from silicon or indium-gallium arsenide materials, offering a potential alternative to the PMT.
The most popular multichannel detectors for spectroscopy are silicon based charge-coupled devices (CCDs) with several thousand elements, or pixels, arranged in a rectangle. Scientific-grade CCDs exhibit high responsivity from the near-ultraviolet to the near-infrared (NIR) region of the spectrum — 200 nanometers to 1.1 microns.
The most popular multichannel detectors for spectroscopy are silicon based charge-coupled devices (CCDs) with several thousand elements, or pixels, arranged in a rectangle. Inexpensive linear CCDs (and photodiode arrays) are available on the market, but in most scientific applications, 2D CCDs are used. Scientific-grade CCDs exhibit high responsivity from the near-ultraviolet to the near-infrared (NIR) region of the spectrum — 200 nanometers to 1.1 microns. At longer wavelengths, the photon energy is lower than the silicon bandgap and the silicon becomes transparent to the incident photons. However, III-V semiconductor materials such as indium gallium arsenide (InGaAs) have a lower energy bandgap and can absorb the NIR photons (see Fig. 44). For this reason, InGaAs arrays are the array detector of choice between 0.9 and 1.7 μm, and are now available up to 2 μm. Other multichannel detectors such as HgCdTe and InSb MIR arrays are available, but are not used as frequently, mainly due to their high cost.
In spectroscopy mode, a CCD works by first summing the electrical charges of the selected pixels in a column at the bottom of that column of the array into a “super-pixel”. This charge-shifting is called a parallel shift. The combined charge from each of these superpixels is transferred serially by the readout register to the output node amplifier. Here, these individual charges are read out and transferred to the ADC (analog-to-digital converter) and processed by the electronics. In a full frame (FF) CCD, the entire array is exposed to light. In a frame transfer (FT) CCD, the top half of the array is exposed to light and the bottom half is used to store the electrical charges before readout. This format allows faster spectral acquisition rate, and may sometimes be used for kinetics measurements.
Silicon CCDs are available with various options to optimize performance. A basic, front-illuminated CCD may be coated with a phosphor to enhance its UV response. The phosphor absorbs the UV photons and re-emits green photons in the spectral range where the CCD is most sensitive. Modifications also can be made to the gate structure in order to improve the detection and raise the effective QE. This variation is called an “Open Electrode” (or “Open Poly”) CCD. In such a device, approximately one third of the gate is removed in the center of the pixel, allowing more light to reach the silicon, which improves detection.
In a back-thinned CCD, the entire device is thinned so that it can be illuminated from the rear of the CCD which eliminates the problem of absorption from the polysilicon gates on the front. Because of the reduced silicon thickness, however, constructive and destructive interference may occur, resulting in an etalon effect that produces a pattern on the signal sensitivity, which can be difficult to correct mathematically.
In another modification, CCD devices are made from special high resistivity material, creating deeper depletion layers in the silicon, resulting in a thicker active layer in which long-wavelength photons are more likely to be absorbed, enhancing the QE in the red region of the spectrum.
Because CCDs are used for many applications, new types are appearing all the time. Pixel sizes are decreasing, which can improve spectral resolution, but at the same time decrease apparent sensitivity and dynamic range, since the charge capacity of each pixel decreases with size. Higher pixel densities also require longer readout times after exposure. Manufacturers of analytical spectroscopic instruments are constantly making tradeoffs in order to supply high sensitivity and speed while maintaining affordability.
Other types of CCD detectors are available that augment the capabilities of standard detectors by adding gain and temporal resolution/gating. These include the Electron Multiplying CCD (EMCCD) and the Intensified CCD (ICCD) detectors.
An EMCCD may be thought of as a standard CCD with optional gain. They are often used for low-light applications with fragile samples, including biological applications and single-molecule work, where the option of integrating the signal for longer periods of time does not exist due to the transient nature of the sample.
EMCCDs are often used for imaging applications, and are commonly available in square formats intended to be used in microscopy, but are also available in other aspect ratios, making them more useful for spectroscopy. EMCCDs have on-chip gain, which is derived from a cascade register arranged between the serial register and a second output node. As the charge from each pixel is clocked out, the magnitude of this charge is increased as it steps through the gain register. Typically, the gain can be adjusted over a range of values, reaching as high as 103. While useful, there are some caveats in using this gain mode for quantitative work, and to alleviate this concern, many EMCCD sensors have both a ‘standard’ as well as a ‘multiplier’ output, giving the researcher the option of using the most appropriate mode for their particular experiment.
Other silicon-based multichannel detectors are now on the market. Most of these are read out by each individual pixel (rather than by shifting charge to the readout register as with a CCD), and the spectrum is reconstructed by the host software during post-processing. Some CMOS array detectors may be read nondestructively, so it becomes possible to use various integration times for each pixel without having to read out the entire sensor. This may be exploited to increase the dynamic range of a measurement. The scientific CMOS detector (sCMOS), a relatively new technology, is used for imaging — including machine vision such as inspection of defects in semiconductors, and is beginning to be used in microscopy. Infrared multichannel detectors, often called “focal plane arrays” are available in InGaAs, InSb and HgCdTe materials.
One interesting application using a 2D multichannel detector is called multi-track spectroscopy. In this type of measurement, a linear bundle of fibers is presented at the spectrometer entrance, with each of the fibers collecting light from a different analyte or part of an extended sample. The multichannel detector is divided into a number of horizontal strips, corresponding to the output from each fiber. In this manner, multiple spectra can be measured simultaneously.
A multitude of detector options have been presented in this article. Here is a general set of questions to serve as guidelines to help the user narrow down the range of choices and select the most appropriate detector for the application.
The first seven questions require clear definition and analysis of the application. Specif ication of the wavelength range is fairly straightforward. In certain cases when multiple samples may be studied, the user may need to initially accept a lower spectral range, with the intention of extending the range in the future.
Definition of the required spectral resolution is fairly straightforward, but may determine whether or not it is possible to use a CCD or multichannel detector. For applications with relatively low spectral resolution requirements, greater than several angstroms, in the spectral range below 2 microns, a multichannel detector will probably be a good option. But, for higher resolution, careful calculations must be performed to ascertain whether pixel size will limit the spectral resolution. It is not always possible to achieve the required resolution of a MCD system by increasing the dispersion (by increasing the grating groove density, increasing the focal length of the spectrometer, or both), especially when considering the required spectral range. Thus, effective slit widths that are smaller than typical pixel sizes are required. In cases where additional dispersion can not compensate for the size of the pixels, a SCD behind a narrow slit will often yield the required resolution.
As for SCDs versus MCDs, as a general rule, when multiple wavelengths bands need to measured, MCDs offer faster data acquisition or superior signal to noise compared with SCDs due to the Felgett multichannel advantage - namely that instead of measuring one wavelength band at a time, you can measure multiple bands for the same amount of time. Their tradeoff is cost and complexity. For single or a small number of wavelength bands, SCDs can offer optimum performance at a lower cost. Also, for certain temporal requirements, such as sub 100s of picoseconds, an SCD is the only option. Similarly, the choice of array detectors for wavelengths greater than 2 microns is very restricted and
includes few, if any, practical options.
The answer to the fourth question should also be straightforward: if the application involves time-resolved measurements certain detector choices may be eliminated, although the specific technique must still be defined.
The fifth and sixth questions can often be answered by doing preliminary literature investigation of the sample to be studied. Low light level measurements require a detector with high sensitivity and gain, and the capability for long exposure times. The background signal level in the entire spectroscopy system must also be considered. A fragile sample cannot be studied using a detector requiring a long exposure time, so either some type of CCD or PMT with photon counting may be a good choice.
The seventh question requires a good deal of thought. Since the image of the light at the entrance of the spectrometer is projected onto the monochromator exit, the size and shape are important to consider. If a 6 mm tall linear fiber bundle (6 fibers) is mounted at the entrance slit, the detector at the exit should be tall enough to capture all of the photons. An avalanche photodiode with 50 micron diameter mounted directly at the exit slit will not work well, although a PMT with an 8 mm diameter photocathode will be fine. A 2 mm diameter photodiode mounted in a housing containing an elliptical mirror (6x de-magnification) will work well. If the application requires the capture of 6 spectra (one from each
fiber) simultaneously, a 2D CCD is required.
The remaining questions require the user to consider the robustness of the detector, whether the housing and mounting hardware will allow additional detectors to be added or substituted, and the amount of daily upkeep and care that is required.
Do you have any questions or requests? Use this form to contact our specialists.