Conventionally nuclear structure experiments have used analogue electronics to shape and filter the preamplifier pulse from a semiconductor gamma ray detector to produce a pulse whose height is proportional to the energy deposited in the detector by the gamma ray. The pulse has then been converted using an accurate analogue to digital converter (ADC) and stored, along with successive pulses, in a memory. The stored pulses are used for spectroscopy by making a spectrum which plots the number of counts for each of the possible ADC output values, so with a 12 bit ADC we have a 4000 channel spectrum where the channel number is proportional to energy. The peaks in the spectrum correspond to particular deposited energies and their width is determined by the charge collection process, the quality of the detector, the quality of the preamplifier and the processing electronics. Some additional broadening occurs due to Doppler shift if the gamma rays originate from a moving nucleus.
For many years engineers have known that the pulse filtering can be performed using digital signal processing techniques but until recently it has not been possible to apply such techniques to signals from semiconductor detectors. The problem has been that for digital pulse processing we must sample the preamplifier output pulse many times with an accurate fast ADC rather than sampling an analogue filter output once with a much slower (but still very accurate) ADC typically taking 5μs to perform the conversion. For digital pulse processing the ADC must convert approximately every 50ns, that means it must be 100 times faster but still have at least 12 bits range. ADCs like this have become available recently, along with very fast Digital Signal Processors (DSP) and powerful programmable arrays of logic gates called Field Programmable Gate Arrays (FPGA). The DSP and FPGA devices are used to perform digital filtering on the sampled ADC data in real time, that is within 5 or 10μs before the next gamma ray arrives.
Some filter types are difficult to achieve using analogue filtering, for example a trapezoidal filter function can only be approximated using analogue filtering. However in digital signal processing a true trapezoidal filter can be applied. For large semiconductor detectors a trapezoidal filter is very useful because it allows time for all the charge released by the gamma ray to be collected, even the charge released in areas where the electric field is low or released a long way from the electrodes and which therefore takes longer to travel to the electrodes. So digital signal processing allows us to improve energy resolution for detectors where charge collection times show a wide variation either because the detector is large or because of neutron damage to the detector (or both).
Importantly, digitising the preamplifier signal gives us new information too. We can analyse not just the final amplitude of the pulse to extract the energy, but also the pulse shape as it collects the charge. Examination of the pulse shape shows us what is happening while the charge is being collected from which we can deduce where the charge was released inside the detector. This positional information has two main uses. Firstly we can apply a Doppler broadening correction to overcome the photopeak broadening which dominates the resolution when Ge detectors are used in-beam especially when the Ge subtends a large opening angle and/or the beam velocity is high. Secondly we can track gamma rays inside a detector and therefore remove the need to put escape suppression shields around Ge detectors to determine whether or not the gamma ray scattered out of the Ge. These two ideas are explained in following sections.
Both Doppler correction and tracking rely on good position determination. We can make the job of position determination easier for the electronics if we also subdivide the detector's charge collection electrode into many smaller electrodes. This is known as segmentation and is another idea which has been around for many years, but until recently the key requirements of small inter-segment gaps, good isolation and reliability have not been achievable in large coaxial germanium gamma ray detectors. Segmented detectors are now available from 2 commercial detector manufacturers.
So by combining digital pulse processing techniques with segmented detectors we can take advantage of recent advances in technology to improve energy resolution by Doppler correction and to track gamma rays interactions within a single detector and potentially across multiple detectors.