Recaudación de Impuestos:

It has been shown that the ODLTCS technique can successfully be applied to the characterisation of deep-level defects of highly irradiated silicon. The defect concentration in this material was considerably larger than 0.1 of the background impurity concentration. Such material cannot be characterised using the conventional capacitance based DLTS.

It should be noted that the marginal oscillator detector conductance mode does not require high electric fields, as is often the case in conventional DLTS. Therefore the possibility of complicating Poole-Frenkel (field-sensitive emission) effects on the measurement of trap activation energy is diminished. Furthermore, the use of high-frequency oscillators (~ 40 MHz compared with 1 MHz in the common commercial capacitance bridges used for DLTCS) enables the measurement of deep level defects at higher scanning temperatures (higher emission

70 90 110 130 150 170 190 210 230 250 270 290 Temperature (K) 0 50 100 150 200 250 300

rates) thus avoiding freeze-out of the shallow background levels at low temperatures, an effect which limits the usefulness of DLTS [227].

The lack of an applied voltage eliminates the possibility of electric field disturbance of the defect state behaviour. Additionally, the high reverse current usually present in such highly irradiated detectors is absent. In many junctions techniques where capacitance or current transients are to be measured a high sample reverse current can cause considerable problems.

A disadvantage of ODLTCS is the inability to provide a direct measurement of trap concentration. This arises because of the absence of knowledge of the primary photocarrier concentration, diffusion kinetics of the carriers, non-linear generation of carriers with depth because of light absorption, and degree of partial traps filling. In addition, the amplitude of the conductance signal is also determined by the carrier recombination lifetime, a further unknown quantity. However, despite this drawback the qualitative comparison of relative trap concentration can be useful.

Although trap concentrations may not be measurable, trapping cross sections can be estimated by measuring the prefactor in Equation 3.30 (intercept on emission axis for T -1 as it approaches zero).

It would be advantageous if the circuit could be modified to accommodate the reverse biasing and pulse injection to the sample so as to provide a means of defining the sensitive volume and hence allow determination of trap concentrations. Although a variety of alterations to the existing circuit where tried, no viable change could be obtained. The marginal oscillator was quenched in all attempts.

Further development of the marginal oscillator detector circuit could be directed towards the use of higher frequency oscillators using high Q tank circuits, possibly of the tuned cavity type. Le Cleach [228] has reported the use of a contactless microwave photoconductive technique which may demonstrate some of the advantages of a higher frequency system. An amplitude limited oscillator reported by Robinson [257], has the advantages of lower noise and

greater insensitivity to microphonics than the simpler oscillator reported here. They should be investigated together, using an electronically variable conductance calibrator for sensitivity determination.

This technique is now available for the study of radiation induced defects in highly damaged detector grade silicon.

3.7 Conclusion

The preliminary study of this thesis was mostly directed to developing measurement techniques for the characterisation of radiation detector materials. As part of this program it was shown that the presence of a MOS capacitor will lead to an under estimation of Neff as a result of an additional capacitance contribution to the junction depletion capacitance.

It was also shown that for an accurate determination of Neff in a square junction device based on high resistivity silicon that the Copeland peripheral capacitance correction is not adequate. A numerical solution was used to obtain an improved correction factor for detectors with square junction areas from 0.09 cm2 to 1 cm2. It may be possible to obtain confirmation of this result by performing a computer simulation of the electric field profile about the corner and perimeter regions of the junction. From this the capacitance contribution could be determined. Semiconductor device simulation tools such as DESSIS of the TCAD program suite should be capable of solving this problem.

In terms of the post irradiation measurements, similar results were obtained as those reported previously in the literature (as discussed in Chapter 2). Following neutron irradiation C-V measurements for profiling the effective impurity concentration and junction built in bias were not possible. This result was shown to be caused by the frequency dependence of the measured capacitance in the presence of a deep level acceptor impurity located about the mid bandgap of the n-type bulk.

Measurement of the reverse current following irradiation allowed a determination of the reverse current damage constant. Agreement of this parameter with published literature was only possible after determination of the equivalent 1 MeV neutron fluence in silicon experienced by each detector. This was a time consuming process and only possible due to the relatively simple kinematics of the monoenergetic neutron production reaction employed in the irradiation. Although reasonable agreement was observed following correction, a remaining variability in α between different detectors irradiated at different angle from the beam line axis was seen. This may have been due to a failure to consider the variation in the neutron yield of the Li7(p,n)Be7 reaction as a function of angle. The experimental neutron flux was monitored in the 0° angle only and the inverse square law used to determine the flux at different radial distances from the target. A consideration of the effect of the device under test (including mounting components) in terms of neutron beam attenuation and neutron energy moderation was not done. Nor was the effect of room scattered neutrons taken into account. Such experimental factors are extremely difficult to quantify from a theoretical perspective. From these points of view it would be advantageous to have a simple and reliable monitor capable of responding directing in terms of the equivalent 1 MeV neutron fluence in silicon. Such a monitor could be used to characterise the neutron field prior to insertion of the device to be irradiated. Or alternatively the monitor could be mounted close to the device under test during the irradiation to provide an on-line assessment of the equivalent 1 MeV neutron fluence in silicon. The development of such a monitor was a second focus of work completed for this thesis.

Experimental observation of the radiation induced deep level defects using the DLTS technique was not possible. This technique becomes invalid for the case of high defect concentrations. To counter this limitation the alternative technique called Optical Deep Level Transient Conductance Spectroscopy was employed. While this technique had been developed for the characterisation of deep level defects in semi insulating materials it was shown to operate successfully for the case of highly irradiated silicon based detector test structures. The A-centre defect was successfully identified in a detector which had been irradiated by a 1 MeV

equivalent neutron fluence of 1013 n⋅cm-2. This neutron fluence is two orders of magnitude in excess of the neutron fluence at which the conventional DLTS technique fails.