Solid State Phenomena Vols. 131-133 (2008) pp. 125-130 online at htlp://www.scientific.net Q (2008) Trans Tech Publications, Switzerland
Vacancy Clusters in Germanium
A.R. ~ e a k e r ' ~V. ' , P. ~ a r k e v i c h ' J. ~ ~slotte3. ~ ~ , K. ~ u i t u n e nF. ~ ,~ u o m i s t oA. ~, satta4, E. simoen4, I. capan5, B. pivac5 and R. ~ a ~ i m o v i c ~ .University of Manchester, School of Electrical and Electronic Engineering, Manchester M60 IQD, UK 'Joint Institute of Solid State and Semiconductor Physics, 220072 Minsk, Belarus 3~elsinkiUniversity of Technology, FIN-02015 HUT, Finland 41MEC, Kapeldreef 75, 8-3001 Leuven, Belgium ' ~ u d j e rBoskovic Institute,lO 000 Zagreb, Croatia '~ozefStefan Institute, SI-1000 Ljubljana, Slovenia
[email protected],
[email protected] Keywords: germanium, neutron irradiation, ion implantation, vacancy complexes
Abstract. Fast neutron irradiation of germanium has been used to study vacancy reactions and vacancy clustering in germanium as a model system to understand ion implantation and the vacancy reactions which are responsible for the apparently low n-type doping ceiling in implanted germanium. It is found that at low neutron doses (-10"cm-~) the damage produced is very similar the to that resulting from electron or gamma irradiation whereas at higher doses (> 10'~cm'~) damage is similar to that resulting from ion implantation as observed in the region near the peak of a doping implant. Electrical measurements including CV profiling, spreading resistance, DeepLevel Transient-Spectroscopy and high resolution Laplace Deep-Level Transient-Spectroscopy have been used in conjunction with positron annihilation and annealing studies. In germanium most radiation and implantation defects are acceptor like and in n-type material the vacancy is negatively charged. In consequence the coulombic repulsion between two vacancies and between vacancies and other radiation-induced defects mitigates against the formation of complexes so that simple defects such as the vacancy donor pair predominate. However in the case of ion implantation and neutron irradiation it is postulated that localized high concentrations of acceptor like defects produce regions of type inversion in which the vacancy is neutral and can combine with itself or with other radiation induced acceptor like defects. In this paper the progression from simple damage to complex damage with increasing neutron dose is examined. Introduction
Germanium is being considered as a future channel material for sub 25nm CMOS because of its high carrier mobilities. However a major problem with Ge devices is the process of ion implantation. Recent attempts at implantation doping of Ge have revealed severe difficulties in terms of maximum achievable doping levels, activation of n-type dopants and the removal of defects [I, 21. A key issue in extremely scaled devices is the problem of enhanced diffusivity of the dopants. In Si the problem is well researched. Interstitials produced during implantation form stable clusters which subsequently dissolve during annealing releasing the interstitials. Because the diffusion of dopants in Si, with the exception of antimony, is mediated by interstitials, this provides enhancement of the diffusivity. The case of boron is technologically the most problematic with transient enhancements of more than an order of magnitude being observed [3].
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttanet. (ID: 130.88.1 18.88-17/09/07,10:49:18)
126
Gettering and Defect Engineering in Semiconductor Technology XI1
In Ge the energy of formation of the vacancy is rather low so most diffusion, with the exception of boron [4], is vacancy mediated [5]. In consequence and in contrast to Si, vacancy clusters in Ge are likely to be of most importance technologically. The ease of formation of the vacancy is also thought to be responsible for the difficulties experienced in achieving high levels of active donors in germanium [ 1,2,6]. Neutron irradiation provides a more manageable methodology for the study of damage and particularly clusters of intrinsic defects. This is because ion implantation into any semiconductor is difficult to study systematically as the damage produced is extremely inhomogeneous with depth. Neutrons produce damage clusters somewhat similar to ions but the clusters are uniformly distributed through the measurement volume of the material. Electron and gamma radiation, which is well documented and understood, does not produce damage clusters. In consequence neutron irradiations provide a model system to study cluster formation and the dissolution of clusters in a systematic way. Experimental
We have studied (100) n-type Ge slices in the resistivity range 0.1 to 5 R-cm grown by the Czochralski process and doped with Sb. Oxygen and carbon concentrations in these ingots are extremely low and undetectable by optical absorption techniques putting an upper limit on the concentration of 1015 ~ m - ~ . The germanium has been irradiated with fast neutrons using the 250 kW TRIGA Mark I1 reactor at the Jozef Stefan Institute in Ljubljana, the samples were contained in a cadmium box to filter out thermal neutrons. During irradiation the sample temperature was not controlled but ~ estimated to be 30°C. The accumulated fluences of neutrons ranged from 5x10" to 1 x 1 0 ' cmq2. The evolution of the induced defects has been studied after 30 minute isochronal annealing stages in the temperature range 100-40O0C. Gold Schottky barriers have been deposited by thermal evaporation with a sample at room temperature and electrical measurements made using a backside Ohmic-contact. CV measurements at 1 MHz have been used to determine carrier profiles and Deep-Level Transient-Spectroscopy (DLTS) to study electrically active defects in both the upper and lower halves of the gap. This is because Au-Ge Schottky barriers have a barrier height close to the band gap. Such a high barrier results in the appearance of an inversion layer with a high concentration of holes near the semiconductor surface resulting under forward bias of a flux of holes from the inversion layer to the semiconductor bulk [7,8]. Some defects have been examined using Laplace DLTS which provides an order of magnitude higher energy resolution [9]. We have used Positron Annihilation Spectroscopy (PAS) to study vacancies and vacancy clusters in the more highly damaged samples with a conventional fast-fast coincidence system and a 2 2 ~source. a ~ The ~ source was placed between the sample and a Ge-reference sample in which no positron trapping was observed. The reference (bulk) lifetime in the measurements was 224 ps. The sample temperature during the measurements was controlled by a closed-cycle He cryostat. The lifetime spectra were analyzed in terms of exponential decay components de-convoluted with the Gaussian resolution function of the spectrometer. The positron diffusion length before annihilation with an electron varies from several hundred nanometers to less than ten nanometers depending on the type of defects and their concentration in the material [lo]. Due to the missing positively charged nucleus the diffusing positrons are trapped by open volume defects such as vacancies and vacancy clusters. This increases the positron lifetime. Additional information can be obtained from Doppler broadening spectroscopy where the momentum of the annihilating electron positron pair is detected as a broadening of the 51 1 keV annihilation peak. The annihilation spectrum is described using the conventional S and W shape parameters [ l I]. From these techniques it is possible to determine information about the trapping centres provided they are present in sufficiently high concentration (> 1014cm-3under our experimental conditions ... much higher than the concentration required for DLTS experiments).
Solid State Phenomena Vols. 131-133
127
Results and Discussion
-
I
.. (1)Ge:Sb (e) 190K -----(2)Ge:SbNl 190K 1
1 1 1 1
:
.....--.....--
I
4--
10
Temperature (K)
Fig. 1. DLTS spectra for neutron-irradiated Ge samples recorded with (upper lines) "normal electron filling" (en = 80 s-', Ub = 5.0, Up = -0.5 V; t,, = 1 ms) and (1ow:r spectra) "injection, hole filling7' (en = 80 s- , Ub = -3.0, Up = +2.0 V; tp = 1 ms) DLTS settings. Initial resistivities of the samples studied were (Nl) 5 0-cm, (N2) 2 R-cm, and (N3) 0.2 R-cm. The neutron fluences were (Nl) 5x10" ~ m - (N2) ~ , 2.5x10I2 cm-2 and (N3) 3x10'~~ m - ~ .
100
1000
Emission, s-'
Fig. 2. Normalised Laplace DLTS spectra measured at 190K, which compare the position and linewidth of signals due to the Sb-V centre in electron irradiated Ge (1 solid line) with the same centre in low dose neutron irradiated material (2 dashed line). The dotted line (3) shows the LDLTS spectrum of the 160K peak in the higher dose N3 sample. This broader line is attributed to a cluster of vacancies.
Figure 1 shows conventional DLTS scans of 3 samples subjected to fluencies of 5x10", 2.5x10I2 and 3x10I3 neutrons per cm2. The spectra in the top half of the diagram represents electron emission from states in the upper half of the gap, while the lower set of spectra represent hole emission from traps in the lower half of the gap. These are filled by the application of an injection pulse (i.e. a forward bias pulse). The spectra have been normalized in magnitude to the value of the largest peak. Considering first the upper curves of electron emission the highest temperature peak with the maximum at about 191 K corresponds to the E-centre, which is formed when a donor atom, in this case antimony, combines with a difhsing vacancy. The E-centre is a double acceptor in Ge and emission characteristics for its second acceptor level derived from Arrhenius plots of the Laplace DLTS data are identical to those for the E-centres produced by electron or gamma irradiation [8,9,12]. The annealing behaviour (temperature dependence) is also very similar to that we have observed for V-Sb [12]. Further confirmation that the highest concentration defect in the low dose neutron samples spectra is Sb-V is provided in Fig. 2. This shows Laplace DLTS spectra measured at 190 K with a dominant peak due to emission from the E-centre in two samples one irradiated with a low dose of neutrons the other with 4 MeV electrons. The two spectra are almost identical. It is presumed that the slight shift in emission rate is due to a small temperature error (
