RADIATION HARDNESS OF MOS STRUCTURES EXPOSED TO HIGH-ENERGY IONS

Recently there has been a growing interest in the use of highenergy heavy ion irradiation for electronic materials modification [1]. The issues involved in high-energy ion implantation are open both on the fundamental [2] as well as on the applied aspects of the process [3]: the understanding of defect creation and its kinetics on the one hand, and the application of deep implantation on the other hand.


Introduction
Recently there has been a growing interest in the use of highenergy heavy ion irradiation for electronic materials modification [1]. The issues involved in high-energy ion implantation are open both on the fundamental [2] as well as on the applied aspects of the process [3]: the understanding of defect creation and its kinetics on the one hand, and the application of deep implantation on the other hand.
This study is also very useful for knowledge of the radiation hardness of electronic devices in a high radiation environment in a reactor laboratory or in the outer space.
The scope of this paper is to investigate the effect of the 305 MeV Kr and 710 MeV Bi ion irradiation on the electrical properties of MOS structures.
We used the capacitance-voltage C-V method to determine the flat-band voltage U FB and the free carriers concentration profile n(x) in the structures before and after irradiation. The kinetics of the generation process that takes place in the space charge region was diagnosed by the non-equilibrium C-t method [4]. Parameters obtained from this method, the relaxation time t r , the generation lifetime of minority charge carriers g and the surface generation velocity S g characterize the electrical behaviour of defects in the MOS structure.
The energy distribution of the insulator-semiconductor interface trap density D it was obtained from the low frequency C-V curve. We used the quasistatic charge-voltage method [5].
In particular, deep level transient spectroscopy (DLTS) [6] proved to be highly applicable because of its good sensitivity, and specific defect centers with well-defined signatures can be moni-tored to concentrations of ϳ10 Ϫ5 below the dopant concentration. We used the standard capacitance DLTS method with electrical excitation.

Experiment
An n-type Czochralski-grown antimony doped (100)-oriented homogeneous wafer of silicon with resistivity 2-5 ⍀cm and thickness 300 m was used as a substrate of the MOS structure. The gate SiO 2 layers were prepared by thermal oxidation in the atmosphere of dry oxygen at 1050 °C for 90 minutes. The thickness of the SiO 2 layer was about 100 nm. All gates were vapour deposited and patterned photolithographically. After manufacturing the MOS structures, the sample was annealed in N 2 ϩ H 2 at 460 °C for 20 minutes. The ohmic contact from the backside of the wafer was prepared by vapour deposition of Al.
Afterwards, the MOS structures were exposed to 305 MeV Kr ion irradiation with a fluence of 10 9 cm Ϫ2 (samples D1) and a fluence of 10 10 cm Ϫ2 (samples D2) and to 710 MeV Bi ion irradiation with a fluence of 10 9 cm Ϫ2 (samples D8) and with a fluence of 10 10 cm Ϫ2 (samples D6) at the Joint Institute for Nuclear Research in Dubna, Russia [7].
The non-irradiated MOS structure (D0) and the irradiated MOS structures (D1, D2, D6, D8) were characterized by C-V, low-frequency C-V and C-t methods. High frequency capacitancevoltage C-V and non-equilibrium capacitance-time C-t measurements were performed using the 4280 1 MHz C Meter/C-V Plotter Hewlett-Packard [8]. Quasistatic low frequency C-V measurements were performed using the Keithley 595 Quasistatic CV Meter. The MOS structures (D0) and irradiated (D6, D8) were also investigated by a standard DLTS method. This method is a high-frequency transient capacitance thermal scanning method, where the capac- itance transient effect in the potential barrier of a semiconductor is caused by an electrical excitation pulse. DLTS measurements were performed using a BioRad boxcar detection system.

Results and Discussion
The MOS structures (sample D0) used in this experiment were of high quality, having a low density of defects. This presumption was verified by many measured DLTS spectra under different measuring conditions (Fig. 1). We found only one deep level DL1 (0.61 eV) with a very low concentration of 2 ϫ 10 13 cm Ϫ3 . This deep level corresponds to the presence of Au (0.59 eV) [9].
In C-V measured curves (Fig. 2), we observed a rise in the positive defect charge in the oxide and at the interface with the silicon substrate, which resulted in an increased flat band voltage (Tab. 1).
The antimony doped MOS structures had the same or only slightly lower concentration of impurities, N D , after irradiation with Kr and Bi ions. Significant changes in the parameters of the MOS structure were observed in the non-equilibrium state. In C-t measurements we observed a marked decrease of the relaxation time, tr, see Fig. 3.   Parameters extracted from C-V, low-frequency C-V and C-t measurements on MOS structures. Tab. 1 The values of D it were determined for the level 0.22 eV above the midgap (Fig. 6). The rise in the density of radiation defects is expressed quantitatively through parameters g and S g (Tab. 1). The highest density of electrically active defects was found in the case of Bi ions with higher energies than those of Kr. Remarkable differences in the generation parameters were observed also in dependence on the fluence of irradiation. In Fig. 4 one can see the shortening of the time of relaxation after irradiation by Kr ions with a higher fluence.
In the case of irradiating with ions of Bi with a fluence of 10 10 cm Ϫ2 a significant increase in the conductivity of the insulating layer was observed (Fig. 5).
C-V measurements and the calculation of D it (Fig. 6) reveal that the growing dose and energy of implanted ions increase the density, D it . A similar trend can be observed also in the change of the surface generation velocity, S g . One can deduce that the quality of the Si -SiO 2 deteriorates. In spite of that the irradiated MOS structures still remain their capacitance properties and functionality.
The measured DLTS spectra of the MOS structures after Bi ion irradiation with fluences 10 9 cm Ϫ2 and 10 10 cm Ϫ2 (D6 and D8) exhibit a strong deviation from an exponential dependence (Fig. 7).  cross section for electrons n were determined from the Arrhenius diagram using the known equations [6] (they are listed in Tab. 2).
In principle it was shown, see Fig. 7, that the deep energy deep DL1 was found in all the three investigated samples D0, D6, D8, while the deep energy level DL2 can be detected only in the structures irradiated by Bi ions. We can suppose that it was a radiation defect with an activation enthalpy of 0.52 eV, which probably corresponds to an E -centre (vacancy-phosphorous) in irradiated Si [9]. This deep level dominates in structure D6. DL3, DL4, DL5 and DL6 peaks were detected only in D8 structures irradiated by Bi ions with fluency 10 9 cm Ϫ2 . These are radiation defects. We suppose that DL3 (0.27 eV) and DL4 (0.25 eV) deep level are divacancies V Ϫ2 in Si [10] created by the implantation process and DL5 (0.17 eV) and DL6 (0.14 eV) correspond probably to the A -centre (vacancy-oxygen) in irradiated Si [11].

Conclusion
From the capacitance C-V and C-t measurements we found that MOS structures irradiated by high energies and doses of Kr and Bi ions were functional in spite of a high density of radiation defects. In the case of Bi ions with a higher energy than that of Kr ions, the conductivity of the insulating layer increased. We did not observe a decrease in the concentration of shallow impurities in irradiated MOS structures. The radiation defects created at the Si-SiO 2 interface increased the trap density and their electrical activity.
We detected six deep levels in MOS structures irradiated by Bi ions with energy 710 MeV and fluency 10 9 and 10 10 cm Ϫ2 . Five of them, DL2-DL6, were produced by radiation. DL3-DL6 levels cannot be detected under higher fluency (10 10 cm Ϫ2 ). Our interpretation of this fact is following: irradiation with a fluency of 10 10 cm Ϫ2 caused gettering of DL3-DL6 defects. This fluency also causes great oxide layer damage. At the same time, the concentration of DL2 defect in structure D6 rose strongly.