NONIONIZING ELECTROMAGNETIC RADIATION · 253 is sequentially generated and detected by a single detector. In a diode array spectrometer, multiple detectors are used to accomplish the simultaneous measurement of a large number of spectral detec- tion bands. Each pixel in the array acts as an exit slit for a particular wavelength. Figure 5.8 shows the schematic of a spectroradiometer based on a diode array spectrometer for measuring spectral irradi- ance. The earliest systems were restricted to silicon detector arrays covering the 300 nm to 1000 nm wavelength range. Wavelength range was extended to 1.7 µm with the development of germanium and InGaAs detector arrays. More recently, arrays based on extended InGaAs detectors and also PbS detec- tors have extended the long wavelength limit to 2.5 µm. Arrays based on InSb and HgCdTe detectors, whose response extends to wavelengths longer than 10 µm, are available, but they are currently pro- hibitively expensive and require cooling to cryogenic temperatures (Rogalski, 1995). There are obvious advantages to be gained by using a detector array spectrometer, including speed (all wavelengths are recorded simultaneously), compactness, light weight, and more efficient use of the available photons. However, diode array spectrometers suffer from a number of draw- backs compared to double grating monochromators, the main one being the much higher levels of stray light encountered. In reporting the stray light characteristics of a spectrometer, several qualifying instrument parameters have to be considered (Arthurs, Drummond, & Kremer, 1995). The ability to suppress stray light depends on factors such as the wavelength being observed, the spectrum of the source being analyzed, the quality of the components, and the type of photodetection system used. A single grating monochromator can usually suppress stray light down to 1 part in 10 4 , whereas a double grating monochromator can suppress stray light to less than 1 part in 10 8 . Compare these values with stray light levels encountered in a diode array spectrometer, which are typically 1 part in 10 3 , but can be as high as 1 part in 10 (Zong et al., 2006). A detector array spectrometer is inherently restricted to a single spectrometer. The higher stray light levels encountered in diode array spectrometers are to be expected. In a diode array spectrometer, the detector array is "immersed" in the spectrometer so that it receives stray light over a full hemisphere (2 sr). The stray light depends on radiation scattered from optical components, but in a diode array spectrometer, the radiation reflected by the detector array itself has been found to make a significant contribution to the stray light (Arthurs et al., 1995). This is not an issue with a monochromator-based system because the detector is placed outside the instrument. Furthermore, one of the main attractions of diode array spectrometers is their small size, but this makes it difficult to suppress stray light by, for example, adding baffles. The spectroradiometer arrangement shown in Figure 5.8 is a particularly poor example because radiation exits the integrating sphere over a 2 solid angle, but only the radia- tion within the solid angle supported by the grating aperture is utilized (typically in an F/4 cone). The remaining radiation contributes to stray light. In some instruments, radiation is guided to the diode array spectrometer using optical fibers. This can benefit stray light, but the use of optical fibers can introduce a host of other problems. Stray light problems are particularly serious in applications requiring the measurement of low-UV spectral irradiance levels in the presence of high irradiance of longer wavelengths. This is the type of situ- ation encountered in the measurement of the spectral irradiance of a tungsten lamp in the UV spectral region. The problem is made worse by the fact that the detector arrays being used are based on silicon, which has a considerably higher responsivity at long wavelengths compared to UV wavelengths. Methods have been proposed and demonstrated that quantify the stray light contribution in the out- put of diode array spectrometers (Brown et al., 2003; Zong et al., 2006). This implies that the stray light contribution in diode array spectrometers can potentially be taken into account. However, measurement of the stray light component in diode array spectrometers is not straightforward because it requires a suite