No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Carbon-related centres in irradiated SiGe alloys
Abstract
We report measurements on the Ci–Oi ‘C-centre’ in Czochralski-grown Si1−xGex (x⩽0.06), using the complementary techniques of photoluminescence (PL) and deep-level transient spectroscopy (DLTS). Both techniques show that the donor level of the C-centre in SiGe alloys shifts towards the valence band with increasing x. Unexpectedly, the shift rate of d(ΔHn)/dx=+550meV detected using DLTS is found to be 1.6 times greater than that from PL measurements. Alloy broadening of the PL line and the DLTS signal are similar, and suggest that the C-centre is preferentially found in a Ge-rich environment.
... This alloy usually becomes polycrystalline for the other range of Ge content [6,7]. The radiation damage of the Si 1−x Ge x devices related with introduction of radiation defects [3,[8][9][10][11] can be healed by the atomic reconfiguration of the crystal structure during material anneal [12][13][14]. The particle detector degradation [15][16][17][18] usually appears through decrease of carrier recombination lifetime. ...
The particle detector degradation mainly appears through decrease of carrier recombination lifetime and manifestation of carrier trapping effects related to introduction of carrier capture and emission centers. In this work, the carrier trap spectroscopy in Si1−xGex structures, containing either 1 or 5% of Ge, has been performed by combining the microwave probed photoconductivity, pulsed barrier capacitance transients and spectra of steady-state photo-ionization. These characteristics were examined in pristine, 5.5 MeV electron and 1.6 MeV proton irradiated Si and SiGe diodes with n+p structure.
... The influence of the Ge content on the properties of a few point defects known for pure Si has been understood [33][34][35][36][37][38][39]. An example is given in Fig. 2. The conclusions drawn from these studies with respect to the defect level position in the gap are as follows [40,41]: (i) The observed level displacement towards the valence band with increasing Ge is characteristic of (almost) all defect levels. ...
The article is dedicated to the review and analysis of the effects and processes occurring in Si-Ge quantum size semiconductor structures upon particle irradiation including ion implantation. Comparisons to bulk materials are drawn. The reasons of the enhanced radiation hardness of superlattices and quantum dots are elucidated. Some technological applications of the radiation treatment are reviewed.
In this work, electrically active defects of pristine and 5.5 MeV electron irradiated p-type silicon–germanium (Si1−xGex)-based diodes were examined by combining regular capacitance deep-level transient spectroscopy (C-DLTS) and Laplace DLTS (L-DLTS) techniques. The p-type SiGe alloys with slightly different Ge contents were examined. It was deduced from C-DLTS and L-DLTS spectra that the carbon/oxygen-associated complexes prevailed in the pristine Si0.949Ge0.051 alloys. Irradiation with 5.5 MeV electrons led to a considerable change in the DLT spectrum containing up to seven spectral peaks due to the introduction of radiation defects. These defects were identified using activation energy values reported in the literature. The double interstitial and oxygen complexes and the vacancy, di-vacancy and tri-vacancy ascribed traps were revealed in the irradiated samples. The interstitial carbon and the metastable as well as stable forms of carbon–oxygen (CiOi* and CiOi) complexes were also identified for the electron-irradiated SiGe alloys. It was found that the unstable form of the carbon–oxygen complex became a stable complex in the irradiated and the subsequently annealed (at 125 °C) SiGe samples. The activation energy shifts in the radiation-induced deep traps to lower values were defined when increasing Ge content in the SiGe alloy.
The equilibrium geometries, phonon spectra, electronic structures and optical properties of the CiOi defect in bulk Si and Si1−xGex systems are calculated using the ab initio plane wave density-functional method. We find that in a Ge-doped Si crystal it is more energetically favourable for the defect to stabilize in a configuration with no Ge atoms in the first sphere. Our calculations show that the vibrational properties of the defect in the Si1−xGex alloy are similar to those in the pure Si bulk crystal and only one local vibrational mode is sensitive to the presence of the Ge substitutions. The effect of C, O, Si and Ge isotopes on the phonon spectra are also investigated and found to be in good agreement with available experimental data. It also follows from our calculations of the singlet–triplet splitting and a position of the gap state of the defect with respect to the top of the valence band that the optical energies of the CiOi defect are expected to increase due to Ge doping, which is in agreement with available experimental data.
Density-functional supercell calculations are employed to follow the location of the donor levels arising from interstitial carbon and interstitial carbon–oxygen complexes in SiGe alloys. We show that these complexes interact weakly with neighboring Ge atoms, and that energetics rules out the existence of defect-Ge complexes involving direct GeC or GeO bonds. The CiOi defect is predicted to produce a hole trap that varies as E(0/+) −Ev = 0.41 − 0.76xeV, implying its disappearance for Ge fractions x greater than 0.5.
We show that using a density-functional supercell method we are able to follow the location of defect gap levels in SiGe alloys for different alloying compositions. The method is tested for several properties of the alloys, with special emphasis in the local bond-length relaxations. Our results indicate a predominant Pauling character for the alloys, with a topological rigidity parameter a** lying between 0.7 and 0.8. A comparative study between the electrical properties of the interstitial carbon C i and carbon-oxygen C i O i centers in the alloys, shows that these complexes interact weakly with Ge atoms. The C i O i defect is predicted to produce a hole trap that varies as E0/ + = 0.41− 0.76x eV, implying its disappearance for Ge fractions x greater than 0.5.
We present studies of muonium behaviour in bulk, Czochralski-grown Si1−xGex alloy material, focusing in particular on the hyperfine parameter of the tetrahedral muonium species. In contrast to the bond-centred species, the hyperfine parameter of the tetrahedral-site muonium centre (MuT) appears to vary non-linearly with alloy composition. The temperature dependence of the MuT hyperfine parameter observed in low-Ge alloy material is compared with that seen in pure Si, and previous models of the MuT behaviour in Si are discussed in the light of results from Si1−xGex alloys.
Much of our knowledge of the charge states, lattice site and behaviour of hydrogen in bulk semiconductors comes from observation of its muonium analogue. Here we present studies of muonium behaviour across the composition range in bulk, Czochralski-grown Si1−xGex alloy material, focusing in particular on the muonium hyperfine parameters. For the bond-centred muonium species, a broad distribution of parameters is observed, consistent with a variety of bonding environments. The average value of the isotropic component of the bond-centred hyperfine parameter shows a linear variation with alloy composition, which might be expected based on the linear variation with composition of alloy bond lengths. In contrast, the hyperfine parameter of the tetrahedral-site muonium species (MuT) appears to vary non-linearly with alloy composition, and an explanation of this in terms of MuT mobility is provided. The temperature dependence of the MuT hyperfine parameter observed in several alloy compositions is compared with that seen in pure Si. Previous descriptions of the low-temperature behaviour of the MuT parameter in Si are discussed in the light of results from Si1−xGex material.
A brief report is made on the carbon and oxygen isotope structures of the local vibrational modes of the C-line and G-line luminescence systems created by radiation damage of Czochralski silicon with a high carbon concentration, together with a first time investigation of the local mode structures of the F-line and H-line luminescence systems created by subsequent annealing at 350 and 450°C. A preliminary analysis shows that there is no carbon-carbon or carbon-oxygen bonding in any of these centres, and only very minor dynamic interaction between the different carbon and oxygen atoms exists. The presence of similar and, in some cases, almost identical local modes for the different centres suggests that parts of defect structures responsible for G- and C-line, F- and H-line and some other luminescence systems have similar atomic configurations. It is suggested that the F-line centre is created by the capture of a mobile version of the G-line centre by the C-line centre, and that the subsequent capturing of extra oxygen atom gives rise to the H-line centre.
The microscopic behavior of iron in relaxed Si1-xGex alloy is addressed in the present work where various new aspects are highlighted. In p-type materials two types of defects involving iron may coexist under equilibrium; the isolated form, Fei, and the iron-acceptor pair, Fei-As. The latter complex is favored over the former because it is thermodynamically more stable. In each case the iron atom stabilizes at the interstitial tetrahedral site. When boron is the acceptor impurity, both the isolated and the paired forms introduce donor-like levels, distant from each other by 0.28 eV. In the relaxed Si1-xGex bulk alloy, these levels are shown to remain separated by the same amount. However, they shift toward the valence band much faster than the shrinkage of the band gap when the Ge content is increased. The consequence is that the pair-related donor level merges with the valence band at a fairly low alloy composition (x>=7%) while the iron donor level is predicted to disappear from the gap for x>=25%. We also show that neither the entropy nor the enthalpy of migration of free iron, whose experimental determination requires one to take into account the above-mentioned shift, are affected by alloying. Therefore, the fast diffusing character, attributed to iron in silicon, still holds in the alloy. The origin of spectral broadening, related to the chemical disorder, is discussed. Finally, the major technological implication emerging from our new findings is addressed. In particular, we show that both the gettering by segregation, routinely used in silicon, and the field-induced outdiffusion, established in n-type silicon ten years ago, are totally inefficient in the Si1-xGex alloy.
The interstitial carbon-oxygen defect is a prominent defect formed in e-irradiated Cz-Si containing carbon. Previous stress alignment investigations have shown that the oxygen atom weakly perturb the carbon interstitial but the lack of a high-frequency oxygen mode has been taken to imply that the oxygen atom is severely affected and becomes overcoordinated. Local vibrational mode spectroscopy and ab initio modeling are used to investigate the defect. We find new modes whose oxygen isotopic shifts give further evidence for oxygen overcoordination. Moreover, we find that the calculated stress-energy tensor and energy levels are in good agreement with experimental values. The complexes formed by adding both single (CiOiH) and a pair of H atoms (CiOiH2), as well as the addition of a second oxygen atom, are considered theoretically. It is shown that the first is bistable with a shallow donor and deep acceptor level, while the second is passive. The properties of CiOiH and CiO2iH are strikingly similar to the first two members of a family of shallow thermal donors that contain hydrogen.
A quantitative improvement in deep‐level transient spectroscopy (DLTS) resolution has been demonstrated by using Laplace transform method for the emission rate analysis. Numerous tests performed on the software used for the calculations as well as on the experimental setup clearly demonstrated that in this way the resolution of the method can be increased by more than an order of magnitude. Considerable confidence in this approach was gained through measurements of a selection of well‐characterized point defects in various semiconductors. The results for platinum in silicon and EL2 in GaAs are presented. For each of these cases conventional DLTS give broad featureless lines, while Laplace DLTS reveals a fine structure in the emission process producing the spectra.
Deep energy levels caused by high‐energy low‐dose proton irradiation of both n‐ and p‐type silicon have been investigated. Energy positions in the band gap, capture coefficients, and their temperature dependences for majority and minority carrier capture and entropy factors have been measured by deep level transient spectroscopy. Computer simulations have been employed to obtain the correct numbers of injected charge carriers needed for the evaluation of minority carrier capture data. From these measurements, it is possible to deduce the charge carrier lifetime profiles in proton irradiated n‐type silicon for different injection concentrations and temperatures. At room temperature and for low injection, it is found that the singly negative divacancy level with a band‐gap enthalpy of H C -H T =0.421 eV has the largest influence on the lifetime. At high injection, the vacancy–oxygen center, H C -H T =0.164 eV, is mostly responsible for the lifetime reduction. © 1996 American Institute of Physics.
The low-temperature near-band-gap photoluminescence of Si1-xGex is studied over the whole composition range 0≤x≤1. We identify free- and bound-exciton processes and determine the properties of momentum-conserving phonons. From our results we determine the low-temperature band gap of the alloys. Analytical expressions are derived for the X and L bands: EgxX(x) =1.155-0.43x+0.206x2 eV; EgxL(x)=2.010-1.270x eV. The intensity and the linewidth of the various excitonic transitions are found to depend only on the statistical alloy fluctuations. No preferential clustering of Si and Ge atoms is detected.
In 1967 Stoneham showed how the width of a zero phonon line may be determined from the strain field of point defects in the crystal. The direct application of the theory is difficult because of the need to evaluate numerically a two-dimensional integral. It is shown here how the integral may be trivially evaluated to within a few per cent for any zero phonon transition between non-degenerate electronic states. The method exploits the change in linewidth which results from the defects being forbidden to approach closer than some minimum distance to the optical centres. This change is found to be approximately the same for all transitions between non-degenerate levels.
The interstial-carbon (Ci) defect in molecular-beam epitaxy grown, strain relaxed n-or p-type Si1-xGex for 0<~x<~0.50 has been created by 2-MeV proton or electron irradiations, and studied by deep-level transient spectroscopy on p+n- and n+p-mesa diodes. The energy difference between the shallow acceptor and donor levels of the Ci defect remains at a constant value of 0.8 eV as the Ge content is varied. The migration enthalpy of Ci is independent of composition in the composition range 0<~x<~0.15. The observed increased stability of the Ci defect with increasing x is the result of a decrease in the entropy of the process.
The interstial-carbon (Ci) defect in molecular-beam epitaxy grown, strain relaxed n-or p-type Si1-xGex for 0<~x<~0.50 has been created by 2-MeV proton or electron irradiations, and studied by deep-level transient spectroscopy on p+n- and n+p-mesa diodes. The energy difference between the shallow acceptor and donor levels of the Ci defect remains at a constant value of 0.8 eV as the Ge content is varied. The migration enthalpy of Ci is independent of composition in the composition range 0<~x<~0.15. The observed increased stability of the Ci defect with increasing x is the result of a decrease in the entropy of the process.
High-resolution Laplace deep-level transient spectroscopy spectra for gold- or platinum-diffused SiGe samples show an alloy splitting that is associated with the alloy fluctuations in the proximity of the defect. For the case of the platinum acceptor state, the effect of the level splitting caused by alloying in the first and also in the second shell of surrounding atoms can be distinguished. For the case of the gold acceptor, only the effect from the first shell of atoms is observable but the manifestation of alloying in the second-nearest shell can be seen as line broadening. We have found that the electronic energy level is affected by alloying in the first-nearest neighborhood by a factor of 2–3 more than by alloying in the second-nearest shell. The absolute values of the energy differences obtained from the Arrhenius plots for different defect configurations agree with those inferred from the peak separations observed in the spectra. A clear preference for gold and platinum to enter substitutional Si sites adjacent to Ge has been revealed. This may be interpreted in terms of an enthalpy lowering as a result of the fact that both metals are able to replace the host silicon atom more easily than the germanium atom in the substitutional position.
Using transient capacitance spectroscopy, we studied defect energy levels and their annealing behavior in boron-doped silicon of various resistivities irradiated with 1-MeV electrons at room temperature. Three levels located at Ev+0.23, Ev+0.38, and Ec-0.27 eV consistently appear in various samples, showing they are characteristic defects in boron-doped silicon. Many properties of the Ev+0.23-eV level and the divacancy are the same, according to the present study and others. We correlated the Ev+0.38-eV level to the vacancy-oxygen-carbon complex recently identified by Lee and Corbett using the EPR technique. The Ec-0.27-eV level could arise from an interstitial defect of oxygen and boron; and a new level at Ev+0.30 eV arising upon its disappearance could be a vacancy defect trapping an oxygen atom and a boron. Several additional defect levels are reported.
Ci and CiCs defects, created by proton irradiation of n-type, strain-relaxed, epitaxial Si1−xGex of 0.005 ⩽ x ⩽ 0.5, have been studied using deep-level transient spectroscopy (DLTS). The ionization enthalpies of the two defects relative to the conduction band edge, ΔH, are found to increase linearly with increasing Ge content. It is shown that the corresponding levels are not pinned to any of the band edges. Furthermore, it is shown that, for both defects, the slopes, δΔH/δx, as well as the full width at half maximum (FWHM) of the corresponding DLTS peaks, are similar. These observations are in agreement with conclusions deduced from previous electron-paramagnetic resonance (EPR) measurements in pure silicon, stating that, for both defects, the trapped electron is preferentially located at the Ci atom because of its larger electronegativity as compared to those of silicon and germanium. The anneal temperature of the Ci defect, and correspondingly the in-growth temperature of the CiCs complex, increase with increasing Ge content. This is equivalent to an increasing retardation of the diffusion of Ci in Si1−xGex with increasing Ge content. © 1999 American Institute of Physics.
Electronic properties of the vacancy–oxygen complex in unstrained Si1−xGex crystals (0<x ⩽ 0.055) grown by the Czochralski method were studied by means of capacitance transient techniques. The enthalpy of electron ionization for the single acceptor level of the defect relative to the conduction band edge, ΔHn, was found to increase from 0.16 to 0.19 eV with the increase in Ge content. The change of the lattice parameter in Si1−xGex alloys is argued to be one of the main reasons of the observed ΔHn change. © 2003 American Institute of Physics.
Room-temperature irradiation of silicon with 2 MeV electrons can create many defects that give rise to optical absorption lines. The authors describe a simple procedure for calculating the strengths of the more important absorption features and the concentrations of the corresponding defect centres as functions of the radiation dose and of the carbon and oxygen concentrations in the silicon. The following absorption features and centres are considered: the 969 meV line (two-carbon-atom centre), the 865 cm-1 line (C+O centre), the 1020 cm-1 line (C+O+self-interstitial), the 835 cm-1 line (vacancy+O'A centre'), the 1.7 mu m band (di-vacancy) and the 488 meV line (C+O+vacancy centre).
A deep trap in the fundamental gap of a semiconductor has a sharp (delta function) character. In a semiconductor alloy the presence of disorder introduces a distribution of trap activation energies. The authors have undertaken a detailed analytical and numerical exercise to examine the effect of such broadening on the capacitance transient and on deep-level transient spectroscopy (DLTS) analysis of traps. In general the standard DLTS analysis introduces negligible error except in cases of severe broadening where it overestimates the activation energy. They illustrate their analysis by considering both simulated and actual experimental situations.
Low-dose proton- and helium-implanted silicon was studied by deep-level transient spectroscopy. By comparing the spectra as well as the defect level concentration profiles, five electron traps and one hole trap (after proton implantation at room temperature) and one dominant electron trap (after proton implantation at 80K) were identified to be hydrogen-related. Two of the hydrogen-related defect levels produced at room temperature represent different charge states of the same defect with a structure probably containing two hydrogen atoms. The electron trap produced by proton implantation at 80K is a donor level located at about Ec-0.2 eV. The defect is tentatively identified as a vacancy-hydrogen complex or a hydrogen atom in a single interstitial site and anneals out before reaching room temperature.
The paper is concerned with carbon interstitial (Ci), Ev+0.28 eV, and a carbon-related centre, Ev+0.37 eV, growing almost simultaneously with the removal of the Ci atoms in electron-irradiated p-type Cz-grown Si. An observed inverse annealing stage of the former defect at approximately 275 K most likely indicates the existence of internal processes taking place below room temperature which liberate Ci as a final product. These processes are in accord with other experimental findings showing that the concentration of the Ev+0.37 eV defect is larger than that of the Ci. The identity of the two centres found to contribute to the creation of the Ev+0.37 eV defect state is discussed.
Highly excited states of the 0.79 eV luminescent defect observed in photoluminescence excitation (PLE) measurements are interpreted as effective-mass (EMT) states of a pseudo-donor with Ei=38.3 meV. The interpretation implies that the 1s ground state of the donor electron is fivefold split in the C1h symmetric strain field of the defect. Modelling of the internal deformation around the defect by a compressive uniaxial field of 80 MPa along (001) allows the authors to explain the ground state splitting quantitatively and gives rise to a re-interpretation of published uniaxial stress data. Transient decay data suggest that the lowest excited defect state is a split singlet-triplet bound exciton, the 0.79 eV line being emitted by the higher-energy singlet.
The activation energy of phosphorus substituted into Si1−xGex alloys (0 ≦ x ≦ 0.27) is investigated by IR absorption. A weak shift of the activation energy with germanium content is found: E(A1) = (45.5 − 26x) meV. This can be explained by multivalley effective mass theory with a phenomenological central cell potential which does not depend on alloy composition within the investigated range. The experiments show that P atoms are distributed at random in the alloy and that their positions are uncorrelated to those of the Ge atoms.
Die Aktivierungsenergie des Phosphorgrundzustandes in Si1−xGex-Mischkristallen im Bereich 0 ≦ x ≦ 0,27 wird mit IR-Absorptionsmessungen untersucht. Mit zunehmendem Germaniumgehalt verringert sich diese Energie um 26x meV. Dies kann mit der Effektiv-Massen-Näherung erklärt werden, wenn ein im untersuchten Bereich von der Zusammensetzung des Kristalls unabhängiges „Zentral-Zellen-Potential” angenommen wird. P- und Ge-Atome sind statistisch im Mischkristall eingebaut.
The photoluminescence spectra of crystalline silicon samples are measured for temperatures below 1000 K. The optical transitions are analyzed in terms of excitonic and band‐to‐band transitions. From the modeling of the line shape we are able to determine the fundamental indirect band gap for temperatures up to 750 K. The temperature dependence follows the Varshni equation with E g (0)=1.1692 eV, α=(4.9±0.2)×10<sup>-4</sup> eV/K and β=(655±40) K. © 1996 American Institute of Physics.
Over one hundred independent photoluminescence transitions are now known in crystalline silicon. This paper begins by outlining those properties of silicon which are relevant to understanding the photoluminescence. The advantages of, and problems with, the different optical techniques are outlined. Topics discussed in detail include the quantitative understanding of the optical effects of radiation damage, and the vibroni c bandshapes of the photoluminescence, including local and resonant modes, the vibronic coupling of nearly degenerate excited states and isotope effects. The link is described between the excited states of some of the centres and the band states of the silicon host lattice. Relationships within groups of similar optical centres, and the widths of the zero-phonon transitions are also discussed. Tables include a list of published photoluminescence transitions indicating their key properties.
Isoelectronic substitutional carbon atoms, present as an impurity in crystalline silicon, are shown to split the no-phonon luminescence line produced by the decay of excitons bound to phosphorus donors. The centroid of the doublet moves to lower photon energy in proportion to the carbon concentration. The splitting and shift can be explained quantitatively as a perturbation of the transition by the strain produced by the carbon. © 1996 The American Physical Society.
- P M Mooney
P.M. Mooney, et al., Phys. Rev. B 15 (1977) 3836.
- V Alex
V. Alex, et al., J. Appl. Phys. 79 (1996) 6943.
[20] R. Kr.
ussmann, et al., Phys. Stat. Sol. (b) 118 (1983)
275.
- K Goscinski
K. Goscinski, et al., Phys. Rev. B 63 (2001) 235309.
- K Irmsher
K. Irmsher, et al., J Phys. C: Solid State Phys. 17 (1984)
6317.
- K Thonke
K. Thonke, et al., J. Phys. C: Solid State Phys. 18 (1985)
L795.
- A Mesli
A. Mesli, et al., Phys. Rev. B 66 (2002) 045206.
- A N Safonov
A.N. Safonov, et al., Phys. Rev. B 54 (1996) 4409.
- A Das
A. Das, et al., Semicond. Sci. Technol. 3 (1988) 1177.
- J Coutinho
J. Coutinho, et al., Phys. Rev. B 65 (2002) 014109.
- T P L Pedersen
T.P.L. Pedersen, et al., Appl. Phys. Lett. 75 (1999) 4085.
- G Davies
G. Davies, Physics Reports 176 (1989) 83 (Review Section
of Physics Letters).
- G Davies
G. Davies, et al., Semicond. Sci. Technol. 2 (1987) 524.
- C A Londos
C.A. Londos, Semicond. Sci. Technol. 5 (1990) 645.
- G Davies
G. Davies, Physics Reports 176 (1989) 83 (Review Section
of Physics Letters).
- E C Lightowlers
- A N Safonov
E.C. Lightowlers, A.N. Safonov, Mater. Sci. Forum. 258-263 (1997) 617.
- G Davies
G. Davies, J. Phys. D: Appl. Phys. 4 (1971) 1340.
- G Davies
G. Davies, Semicond. Semimetals. 51B (1998) 1.
- J Weber
- M I Alonso
J. Weber, M.I. Alonso, Phys. Rev. B 40 (1989) 5683.
- R Kr
- Ussmann
R. Kr.
ussmann, et al., Phys. Stat. Sol. (b) 118 (1983)
275.