then the signal at the pixel in question will include contributions from both wavelengths. If, on the other hand, there is no deep UV light, then only 400nm will be detected. The second order light is usually removed by using filters. These filters have a transmission band and a blocking band to restrict radiation to a certain wavelength region. These filters are installed permanently inside the optical bench. Alternatively, linear variable filters (Figure 3) can be designed to match the dispersed spectra and provide the right blocking at each pixel in an array. The trade-off is that these linear variable filters are designed and fabricated for a particular bench, grating and starting wavelength so the options for adjusting spectra range are lost.
DETECTORS
with a bias voltage, and photons that are absorbed by the silicon generate a current. The sampling circuitry generally features a sample and hold scheme and the ability to strobe through the row of detectors to acquire the voltage signal from each pixel. The stream of analogue signals is amplified and converted to digital information by external circuitry.
CCD DETECTORS Figure 4: Silicon CCD Detector
Detectors are chosen for several design features. Semiconductors are used to capture photons through absorption and convert them into current. The absorption band of silicon detectors (Figure 4) allows for good sensitivity over the UV, VIS and short wave NIR region (from as low as 160nm to as high as 1100nm). InGaAs detectors (Figure 5) generally work well from 900nm and higher. The upper wavelength range is 1700nm, but specially doped InGaAs can be used to extend that range to 2100 or 2500nm. The material of choice for mid IR range is Mercury Cadmium Telluride (HbCdTe).
PHOTODIODE DETECTORS
The effective range for detectors also depends on architecture. The simplest design is called a photodiode array. Each diode in the array is connected to readout circuitry that occupies a position next to the array. The diodes are supplied
The interest in imaging detectors led to the invention of the CCD or charge coupled detector. Here the photodiode is covered by a transparent capacitor that accumulates the signal during a length of time called an integration period. The advantage in this architecture is that there are no dead spots occupied by readout circuitry and images can be captured. A linear CCD array is the same architecture but consists of a single line of CCD devices instead of the bare photodiodes. CCDs have a great advantage over photodiodes in that they have very low levels of readout noise. Their main disadvantage is that their polysilicon gates or capacitors absorb UV light, and so CCDs generally do not respond to light much below 350nm. There are two remedies to this problem. The CCD array can be coated with a phosphor, which absorbs UV and emits visible light. This renders a detector with adequate UV response for many applications. A more expensive solution is to make the device very thin and to turn it around so it is illuminated from the back side. This exposes the photodiodes to the UV light and the resulting device is significantly more sensitive to UV that the phosphor coated types.
SPEED OF ELECTRONICS, WELL DEPTH AND SIGNAL TO NOISE
Figure 5: InGaAs Detector
Detectors are also characterised by the speed of their electronics, their well depth, sensitivity and signal to noise. Generally, detectors designed for imaging applications are optimised for speed. The ILX511 detector is a 2048-pixel linear CCD detector used in the USB2000+ (Figure 6) and can capture and transfer full spectra (2048 wavelengths) in 1 millisecond over a USB 2.0 port. The S7031 detector is a FFT-CCD detector used in the QE65000 (Figure 7) and is 16x slower than the ILX511, taking 8 milliseconds to acquire and transfer full spectra (1024 wavelengths). The well depth is the number of electrons that can be stored in the capacitor. The CCD capacitors are fully charged at time zero (equivalent to dark). If the quantum efficiency is 100%, each photon striking the surface depletes 1 electron from the well. When the well is fully depleted it stops responding to photons. The arrival of photons on the detector is a random process. This random shot noise ultimately limits the signal to noise of the measurement. The total noise or random variation is the sum of the shot noise, equal to the square root of the number of photons,
combined with the dark noise and readout noise. When signals are high, as when the spectrometer is being used to measure light, the signal to noise is essentially equal to the shot noise. When the signal is low, as when a dark spectrum is being recorded, the noise is equal to the noise on the dark signal + the readout noise (coming from the electronics). The ratio of the maximum signal to the dark noise is called the dynamic range. A deep well detector will have a greater dynamic range than a shallow well. For example the dynamic range of the ILX511 is about 1300:1. The dynamic range of the S7031 is about 25000:1. This also means that it takes more photons to get a full-scale signal with the S7031 than with the ILX511. In fact, well depth and dynamic range are inversely proportional to sensitivity. In most applications it is desirable to maximize signal to noise, and the best detector depends on the type of light being sampled. If a transient event is being measured - for example, an arc from a xenon flash - then a high dynamic range detector is desirable. If the light is continuous, then in a given time frame the signal averaging of the fast shallow well detector will equal the performance of the slower deep well device. If the signal is very low, then the comparison of the signal to the readout noise must be evaluated. Generally, the shallow well device will be best.
COOLED DETECTORS FOR LOW LIGHT LEVEL APPLICATIONS
For low light levels, cooling the detector will lower the dark current and also the noise on the dark current. Signal to noise will be improved, and the ability to integrate signals for longer time periods is also made possible. Thus, the ILX511 shallow well detector will outperform the S7031 detector operating at room temperature when looking at low light levels. However, the S7031 features a built-in Peltier cooler and if it is used near 0°C, then it will have better performance than the ILX511. The trade- off here is the cost, power consumption and size of the Peltier cooling circuitry.
CONCLUSION
There are many things to consider when you configure a spectrometer and selecting design criteria involves accepting tradeoffs. The optimal configuration depends entirely on the application so before you can make good choices about hardware, you must understand your needs. The key is to learn system requirements early in the design process and rank the list: is performance more important than cost, or is a balanced approach preferred? As the Rolling Stones say, you cannot always get what you want. But when it comes to spectrometers, if you understand your requirements and consider the design tradeoffs, you can get what you need.
Figure 6: USB2000+ Spectrometer
Figure 7: QE65000 Spectrometer
Spectroscopy Focus
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32
