Anti-reflection coatings may be used to enhance responsivity to about 25 per cent at the required wavelength, but this leads to reduced efficiency at other wavelengths that these reflect. Package window also plays an important role in shaping the spectral response. The standard glass window absorbs wavelengths shorter than 300nm. For UV detection, a fused silica or UV transmitting glass window is used. Various filter windows are also available to tailor the spectral response for an application. Optical filters can also be added to change the spectral response. A common example is the use of a specific filter to modify the normal silicon response to approximate the spectral response of the human eye.
Due to improved charge-collection efficiency in photodiodes, responsivity increases slightly with applied reverse bias. It also exhibits dependence on temperature variations due to variation in band-gap energy. The band-gap energy varies inversely with the change in temperature.
The term ‘responsivity’ should not be confused with ‘sensitivity’. Sensitivity is the lowest detectable light level that is typically determined by detection noise. It is also significantly influenced by the required detection bandwidth. Also, a photosensor should ideally be operated in a spectral region where its responsivity is not far below the highest possible value, because this leads to the lowest possible detection noise and thus to a high signal-to-noise ratio and high sensitivity.
Noise equivalent power (NEP) is the input power to a sensor that generates an output signal current equal to the total internal noise current of the device, which implies signal-to-noise ratio of 1 (one). In other words, it is the minimum detectable radiation level of the sensor. Noise power and thus noise-equivalent power depend on the assumed detection bandwidth.
If one were to use the full detection bandwidth of a device to compute NEP, the NEP would not allow a fair comparison of sensors with different bandwidths. Therefore it is a common practice to assume a bandwidth of 1Hz, which is usually far below the detection bandwidth. NEP is usually specified in units of W/Hz rather than watts.
Detectivity of a sensor is the reciprocal of its NEP. Sensors with higher value of detectivity are more sensitive. Detectivity, like NEP, depends upon noise bandwidth and sensor area. To eliminate these factors, a normalised figure of detectivity, referred to as D* (Dee-star), is used. It is defined as the detectivity normalised to an area of 1cm2 and noise bandwidth of 1Hz.
Quantum efficiency is the ratio of the number of photoelectrons released to the number of incident light photons absorbed. It is the percentage of input radiation power converted into photo current. It is expressed as rise/fall time parameter in photoelectric sensors and as time constant parameter in thermal sensors. Rise and fall times are the time durations for the output to change from 10 per cent to 90 per cent, and 90 per cent to 10 per cent of the final response, respectively. Rise/fall time parameter determines the highest signal frequency to which a sensor can respond. Time constant is defined as the time required by the output to reach 63 per cent of the final response from zero initial value.
Noise is the most critical factor in designing sensitive radiation detection systems. Noise in these systems is generated in photosensors, radiation sources and post-detection circuitry. Photosensor noise mainly comprises Johnson noise, shot noise, generation-recombination noise and flicker noise.
Johnson noise, also known as Nyquist noise or thermal noise, is caused by thermal motion of charged particles in a resistive element. The RMS value of noise voltage depends on the resistance value, temperature and system bandwidth.
Shot noise in a photosensor is caused by the discrete nature of photoelectrons generated. It is related to the statistical fluctuation of both the dark current and the photo current. It depends on the average current through the photosensor and system bandwidth. It is the dominant source of noise in the case of photodiodes operating in photoconductive mode.
Generation-recombination noise is caused by fluctuation in current generation and recombination rates in a photosensor. This type of noise is predominant in photoconductive sensors operating at infrared wavelengths.
Flicker noise, or 1/f noise, occurs in all conductors where the conducting medium is not a metal and all semiconductor devices that require bias current for their operation. Its amplitude is inversely proportional to the frequency. Flicker noise is usually predominant at frequencies below 100Hz.