interrelated; if a sufficient field cannot be applied to the device, charge carriers may be trapped before being collected. Problems manifested by crystals that exhibit IRR spectrum types (3) or (4) seem to be related to either the lithium-oxygen interaction or high dislocation density, respectively. With regard to carrier trapping, identified with spectrum types (2) and (5), the problem is more complex. The origin of spectrum type (5) has not been identified. Spectrum type (2) is characteristic of the neutron-irradiated reference diode, but it is not possible at present to link the defects known to be produced in germanium by neutron irradiation directly with the defects that produce IRR spectra of this type in the test crystals. Concurrent with the IRR study of germanium, IRR work on silicon nuclear radiation detectors was carried out [6,12]. The results of these measurements, mainly from radiation-damage experiments, were used to support the interpretation of IRR results on germanium. Figure 10 shows IRR spectra obtained for a lithium-drifted silicon. detector before and after irradiation with fast neutrons through the p-contact to a fluence of 5 × 109 cm-2 [12]. As might be expected for such a fluence the two spectra are nearly the same. However, the peaks at 0.86 and 0.77 eV are somewhat enhanced, and the feature at 0.93 eV is much enhanced after irradiation. The two lower energy peaks are associated with levels located 0.40 eV below the conduction band (0.77eV) and 0.31 eV above the valence band (0.86 eV). Such levels have been previously reported from observations of photoconductivity on silicon specimens irradiated with neutrons at fluences from 1 × 1016 to 5 x 1019 cm-2 and are attributed to divacancies [17]. Figure 11 shows IRR spectra obtained from three commercial lithiumdrifted silicon detectors presumed to be fabricated from specimens of the same silicon crystal. Diode NBS-3S was not irradiated, diode NBS-4S was irradiated with 1.9-MeV protons to a fluence of 1 x 1014 cm-2 incident on the p-contact, and diode NBS-5S was irradiated with 1.5-MeV electrons to a fluence of approximately 3 x 1013 cm 2 incident on the p-contact. The spectra of diodes NBS-3S and NBS-4S are similar as might be expected since damage caused by 1.9-MeV protons should be localized in the contact region and not in the compensated region of the device. As in the case of NBS-6S (figure 10), the main features are observed at 0.77 and 0.86 eV, and 1.03 eV. The IRR spectrum of the electron-irradiated specimen, diode NBS-5S, shows features at 0.99, and 0.90 eV as well as those at 0.93, 0.86, and 0.77 eV observed in the other silicon diodes, and attributed to divancies after electron irradiation [18]. In the literature, a level located 0.18 eV below the conduction band (0.99 eV) arising from the vacancy-oxygen complex has been reported after neutronand gamma-irradiation [19,20], while a level 0.27 eV above the valence (0.90 eV) may be associated with lithium precipitates [21]. Since the commercial lithium-drifted silicon detectors used in this study had gold p-contacts, approximately 15 to 20 nm thick, through which the infrared radiation must pass, it was necessary to determine what effects, if any, this thin contact had on the measurement of IRR. The infrared transmission of a 0.14-mm thick silicon filter, the thickness used in the measurement of the IRR of silicon detectors, was measured using a thermocouple detector both before and after evaporation of approximately 15 nm of gold onto one side of the filter. No effect was observed in the spectral distribution of the radiation; however, the overall transmission was reduced by a factor of about three as compared to the uncoated filter. Most In connection with a comprehensive study of methods for the evaluation of germanium suitable for use in Ge (Li) detectors, 85 germanium specimens were collected between 1967 and 1973. of these had been rejected for use in the fabrication of high-quality Ge (Li) detectors. Infrared response spectra from 55 of these specimens were identified as to spectrum type; the remaining specimens were of insufficient thickness to permit the fabrication of diodes, were damaged during processing, or yielded IRR spectra whose type could not be identified due to excess noise. In early IRR measurements, good agreement had been found between energy levels determined from the energies of IRR spectral features and energy levels arising from radiation damage experiments in germanium [3]. However, at that time no attempt was made to distinguish whether features seen in IRR measurements were associated with levels in the upper or in the lower half of the forbidden energy gap. If the various levels observed in IRR spectra could be uniquely identified based on data in the technical literature, and thus linked to a specific crystalline defect, then it might be possible for the crystal grower to identify the source of the defect in the crystal growth procedure. If corrective action could be taken, the quality of the crystal could be upgraded based on the IRR data. 6.1 ENERGY LEVELS IN GERMANIUM After a review of the literature concerning energy levels resulting from radiation or thermal damage in germanium as measured either by photoconductivity or Hall effect [13,23-46], an attempt was made to link features in the IRR spectra specifically with energy levels observed in the literature [22]. The results are shown in figure 12. From the data published on fast neutron, proton, gamma ray, fast electron, and thermal damage, etc., the summary of energy levels shown in column B was obtained. The shaded bars indicate the energy range of levels for which there were some discrepancy in reported energy but which appeared to arise from the same center. In column A are shown energy levels resulting from IRR measurements of diode NBS 83-3. In this case each energy level appears only once in the scheme; the level was placed at the appropriate energy either below the conduction band edge or above the valence band edge using the levels shown in column B as the basis for comparison. In the majority of cases, a feature observed in the IRR spectrum can be linked uniquely on the basis of energy with a level or band of levels obtained from the summary of the results reported in the literature. The energy level located 0.18 eV above the valence band (observed at an energy of about 0.54 eV in IRR spectra) is reported to arise specifically from the divacancy-lithium (VVLi) complex and thus should not be observed in lithium-free germanium [38,46]. Comparison of the IRR spectra obtained from a lithium-drifted diode, NBS 83-3, and from a diode fabricated from high-purity germanium (without lithium compensation), NBS-112, confirms this expectation. The spectrum of NBS 83-3 exhibits a feature at 0.54 eV, but the spectrum of NBS-112 does not. It is possible to make some specific statements, on the basis of IRR measurements, with regard to the nature of trapping centers in Ge (Li) detectors. Table 4 summarizes the data on the two IRR spectrum types covering detectors that exhibit hole or electron trapping. Table 4 Energy Levels in Ge (Li) Detectors Exhibiting Trapping Work on growing high purity germanium indicates that residual oxygen creates charge trapping centers in this material [54]. Thus the assignment of Lio+ as the donor species in those levels thought to arise from WD complexes (Ey+0.11 eV and Ec- 0.21 eV [46]) seems reasonable. That the main electron trap in Ge (Li) detectors must lie deeper than 0.175 eV [55], appears to point to the Ec- 0.21 eV level which is not observed in hole-trapping crystals. 6.2 ENERGY LEVELS IN SILICON The energy level scheme in figure 13 is a summary of the state of the IRR measurements on radiation-damaged, lithium-drifted silicon nuclear radiation detectors. As in the case of germanium, energy levels reported in the literature [17,18,20,47,50-53] are listed in column B, while energy levels observed in IRR measurements are listed in column A. Tentative identification of IRR-detected levels with corresponding levels reported in the literature has been made previously [6]. It has been demonstrated that the IRR technique can be used to qualitatively detect and identify impurities and defects in germanium and silicon with a high degree of sensitivity. In the case of lithium-drifted nuclear radiation detectors, the technique can be used as a predictor of crystal suitability for detector use early in the fabrication process. Similarly, the results of such measurements regarding the presence of certain types of defects in germanium and silicon single crystals should be useful to the crystal grower to better control the growing process. Given suitable sources of optical radiation, the IRR technique should be applicable to a wide range of semiconductor materials from which diodes can be fabricated. In the case of discrete microelectronic devices (transistors and diodes) the technique might be extended to studies of impurities and defects. This would be useful in analyzing device failure and, in particular, radiation damage effects. 8. 1. 2. 3. 4. 5. 6. 7. 8. 9. REFERENCES 10. 11. 12. 13. 14. Armantrout, G. A., Infrared Evaluation Techniques for Ge (Li) Detectors, IEEE Trans. Nucl. Sci. NS-17, No. 1, 16-23 (1970). 15. 16. 17. Sher, A. H., and Keery, W. J., Improved Infrared-Response Technique for Determining Impurity and Defect Levels in Semiconductors, Appl. Phys. Lett. 20, No. 3, 120-122 (1972). 18. Methods of Measurement for Semiconductor Materials, Process Control, and Devices, W. M. Bullis, ed., NBS Tech. Note 571, pp. 16-18 (April 1971). Neuberger, M., Electronic Properties Information Center, Report DS-143 (1965); available from NTIS, Springfield, VA., Accession Number AD 610828. 19. Miles, M. H., Extrinsic Photoconductivity from Edge Dislocations in Germanium, Sher, A. H., Liu, Y. M., and Keery, W. J., Infrared Response Measurements on Radiation-Damaged Si(Li) Detectors, IEEE Trans. Nucl. Sci. NS-19, No. 3, 312-317 (1972). Ryvkin, S. M., Photoelectric Effects in Semiconductors, pp. 369-373 (Consultants Bureau, New York, 1964). Berkovskii, F. M., and Ryvkin, S. M., Sensitivity of Germanium and Silicon Photocells in the Impurity Excitation Region, Sov. Phys. -Solid State 4, 263-269 (1962). Sher, A. H., Keery, W. J., and Dyson, H. E., Improved Infrared Response Measurements in Semiconductor Nuclear Radiation Detectors, IEEE Trans. Nucl. Sci. NS-19, 341-344 (1972). Diodes supplied by R. D. Baertsch, General Electric Company, Schenectady, NY., 12301. Crystal Supplied by I. L. Fowler, AECL, Chalk River Nuclear Laboratories, Chalk River, Ontario, Canada. Methods of Measurement for Semiconductor Materials, Process Control, and Devices, W. M. Bullis, ed., NBS Tech. Note 727, pp. 16-21 (June 1972). Konopleva, R. F., Novikov, S. R., and Ryvkin, S. M., High-Temperature Annealing of Defects Produced by Fast Neutrons in Germanium, Sov. Phys. -Solid State 6, 2610-2612 (1965). Methods of Measurement for Semiconductor Materials, Process Control, and Devices, W. M. Bullis, ed., NBS Tech. Note 733, pp. 17-23 (September 1972). Logan, R. A., Thermally Induced Acceptors in Germanium, Phys. Rev. 101, 1455-1459 (September 1972). Methods of Measurement for Semiconductor Materials, Process Control, and Devices, W. M. Bullis, ed., NBS Tech. Note. 743, pp. 17-22 (December 1972). Lappo. M. T., and Tkachev, V. D., Divacancies in Silicon Irradiated with Fast Young, R. C., and Corelli, J. C., Infrared Photoconductivity of Electron-Irradiated Daly D. F., and Noffke, H. E.. An EPR Study of Fast Neutron Radiation Damage in |