Float Zone Silicon Research Articles

Hydrogen-Defect Shallow Donors in Si

The influence of different original Si crystals and neutron fluence on the formation of hydrogen-defect shallow donors in neutron-irradiated floating-zone silicon grown in hydrogen atmosphere (FZ Si:H2) is studied. The annealing behavior of neutron-irradiated floating-zone silicon grown in argon atmosphere (FZ Si:Ar), neutron-irradiated FZ Si:H2 kept for three years at room temperature (RT) and only fast-neutron-irradiated FZ Si:H2 reveals that shallow donors are directly related to hydrogen and defects, especially to thermal-neutron-radiation point defects. The maximum concentration of the shallow donors approaches to a stable value with an increase in neutron fluence. Some characteristics of the shallow donors are discussed.

Hydrogen-related defect centers in float-zone and epitaxial n-type proton implanted silicon

Hydrogen-related defects in float zone (Fz) and epitaxial (Epi) n-type silicon crystals have been studied by means of deep level transient spectroscopy. These defects, as well as the characteristic vacancy-oxygen (VO) and divacancy (V 2) centers were introduced by proton implantation (1.3 MeV) using a dose of 1×10 10/cm 2. A hydrogen-related defect level located at 0.45 eV below the conduction band edge ( E c) appears in both kind of samples. Another hydrogen-related defect appears predominantly in the Fz samples with a level at E c−0.32 eV. Depth profiling as well as annealing studies strongly suggest that the level at E c−0.45 eV is due to a complex involving hydrogen and V 2. The level at E c−0.32 eV is strongly suppressed in the high purity Epi samples and the same holds for VO center. These results together with annealing data provide substantial evidence that the E c−0.32 eV level originates from a VO-center partly saturated with hydrogen (a VOH complex). Finally, in the Epi samples a new level at ∼ E c−0.31 eV is resolved, which exhibits a concentration versus depth profile strongly confined to the damage peak region. The origin of this level is not known but the extremely narrow depth profile may indicate a higher-order defect of either vacancy or interstitial type.

Room Temperature Annealing Behavior of Copper-Related Deep Levels in P-Type Floating Zone Silicon Wafers

Deep level transient spectroscopy analysis of copper-doped p-type floating-zone silicon (FZ-Si) wafer was performed. Five deep energy level states observed in copper-doped silicon (Ev+0.1 eV, Ev+0.15 eV, Ev+0.22 eV, Ev+0.26 eV and Ev+0.43 eV) with concentrations of 1×1011 to 1×1012 cm-3 were detected. We observed that the amplitude of only one peak (deep level at Ev+0.26 eV) dramatically decreased with time during storage at room temperature, but stabilized at a concentration of about 1×1012 cm-3 after 2 days. The other deep level concentrations did not change after two months storage at room temperature. Therefore, it was concluded that room-temperature annealing resulted in a decrease in the deep level at Ev+0.26 eV due to the formation of copper-precipitate related defects and out-diffusion to the silicon surface.

Oxygen and hydrogen accumulation at buried implantation-damage layers in hydrogen- and helium-implanted Czochralski silicon

Oxygen and hydrogen accumulations at buried implantation-damage layers were studied after post-implant-ation annealing of hydrogen- and helium-implanted Czochralski (Cz) silicon. Hydrogen implantation was carried out at energies E=180 keV and doses D=2.7×1016 cm-2, and helium implantation at E=300 keV and D=1016 cm-2. For comparison hydrogen implantation was also done into float-zone (Fz) silicon wafers. Post-implantation annealing at 1000 °C was done either in H2 or N2 atmosphere. Hydrogen and oxygen concentration profiles were measured by secondary ion mass spectroscopy (SIMS). It is shown that the ambient during annealing plays a significant role for the gettering of oxygen at buried implantation-damage layers in Cz Si. For both hydrogen and helium implantations, the buried defect layers act as rather effective getter centers for oxygen and hydrogen at appropriate conditions. The more efficient gettering of oxygen during post-implantation annealing in a hydrogen ambient can be attributed to a hydrogen-enhanced diffusion of oxygen towards the buried implantation-damage layers, where a fast oxygen accumulation occurs. Oxygen concentrations well above 1019 cm-3 can be obtained. From the comparison of measurements on hydrogen-implanted Cz Si and Fz Si one can conclude that at the buried defect layers hydrogen is most probably trapped by voids and/or may be stable as immobile molecular hydrogen species. Therefore hydrogen accumulated at the defect layers, and is preserved even after high-temperature annealing at 1000 °C.

Thermoelectromagnetic convection in floating zone silicon growth with a nonaxisymmetric temperature and a strong axial magnetic field

Thermoelectromagnetic convection (TEMC) is the melt motion driven by the interaction of an externally applied magnetic field and the thermoelectric current due to variations of the absolute thermoelectric power and of the temperature. This paper presents numerical solutions for the three-dimensional TEMC in a simplified model of the floating zone process with a nonaxisymmetric temperature variation over the crystal–melt interface. The axial magnetic field is sufficiently strong that inertial effects are negligible, so that each azimuthal Fourier component of the TEMC can be treated independently. Results are presented for the cos θ and cos(2 θ) components of the TEMC. The results show that a very small deviation from axisymmetry in the temperature produces a very large deviation from axisymmetry in the radial and axial components of the TEMC velocity.

The deformation regimes of the yield point of silicon

The upper yield point of undoped floating-zone silicon has been investigated. The measurements were performed on dislocation-free single crystals deformed at temperatures between 800 and 1300°C and a strain-rate range extending over three orders of magnitude. The upper yield point emerges up to 1250°C. In principle, the data reflect the behaviour of the lower yield point, thus confirming the existence of a novel deformation regime of silicon at high temperatures and small strain rates. There are, however, features which resemble the behaviour of highly doped material to some extent. This surprising similarity raises questions concerning the interpretation of the different deformation regimes observed in undoped and doped silicon.

Easy-to-fabricate 20% efficient large-area silicon solar cells

Recently, an innovative silicon solar cell structure has been developed at ISFH which is capable of achieving very high cell efficiencies on industrial-size wafers with a simple photolithography-free processing sequence. As the corresponding solar cells essentially rely on the application of obliquely evaporated contacts they are denoted as OECO cells. In this paper the successful up-scaling of the novel OECO process from 21% efficient 4 cm 2 laboratory devices to the fabrication of large-area (⩾100 cm 2) silicon solar cells is described, and independently confirmed total area efficiencies of 20% are reported for 10×10 cm 2 OECO-type solar cells fabricated on p-type float-zone silicon.

Spectroscopic study and structural characterization of a Li-related photoluminescence center in neutron-irradiated Si

We report on a new Li-related photoluminescence center with zero-phonon line at 879.3 meV. The center is created at 550--600 \ifmmode^\circ\else\textdegree\fi{}C in the final stages of annealing out radiation-induced point defects in float-zone silicon. Isotope and chemical correlation data establish that the center contains Li and C atoms. The isotope shift from ${}^{6}\mathrm{Li}$ to ${}^{7}\mathrm{Li},$ $\ensuremath{\Delta}{E=E(}^{7}\mathrm{Li})\ensuremath{-}{E(}^{6}\mathrm{Li})=0.18\mathrm{meV},$ is similar, per Li atom, to that observed for other Li-related centers in silicon. Uniaxial stress measurements establish the symmetry as monoclinic I, with only small departures from trigonal symmetry. A simple method for the transition is introduced to fit simultaneously the energies, polarizations, and relative intensities of the stress-split components. The transition's dipole is shown to be close to a bonding direction in the plane perpendicular to the characteristic 〈110〉 axis of the monoclinic I center. The vibronic sideband is produced by coupling to modes of 16, 31, and 36 meV, with a Huang-Rhys factor $S=1.1.$ This value can be predicted simply from the uniaxial-stress data. The temperature dependence of the zero-phonon line can be fitted precisely using the spectrum of coupled phonons derived from the vibronic band shape, plus the approximation that the differences in frequencies of the phonons in the ground and excited electronic states are proportional to the phonon frequency. The luminescence from the center is reversibly quenched with increasing temperature, with an activation energy ${E}_{a}=32\ifmmode\pm\else\textpm\fi{}5\mathrm{meV},$ however, we show that this result does not arise from the excited state containing a shallow particle. Although the center is created at 600 \ifmmode^\circ\else\textdegree\fi{}C, it is rapidly destroyed at room temperature through passivation by the capture of one mobile Li atom.

Electron paramagnetic resonance study of S2 defects in hydrogen-implanted silicon

Electron paramagnetic resonance spectra of S2 paramagnetic center in Float zone silicon (FZ-Si) and Czochralski silicon (CZ-Si) produced by H + 2 ion-implantation and by subsequent annealing have been studied. At room temperature the spectrum exhibits an isotropic line g=2.0066±0.0002, which, then broadens and transforms in a complex spectrum at 120 K. This complex spectrum can be decomposed into isotropic and anisotropic component. The relative concentration of anisotropic component is larger in CZ-Si and smaller in FZ-Si than the respective isotropic component. The g tensor evaluated from the anisotropic component at 120 K shows that the spectrum originated from the center with trigonal symmetry. The involvement of hydrogen and oxygen atoms in the center has been discussed. It is suggested that the S2 center is developed in the multi-vacancy defect containing oxygen and hydrogen atoms. From reversible thermal effect of the linewidth narrowing the motional properties of the dangling bond in the S2 center is described by the activation energy of 0.03 eV.

Photoluminescence of erbium-doped silicon: Excitation power and temperature dependence

Photoluminescence measurements have been made on float-zone and Czochralski-grown silicon samples which were doped with erbium by ion implantation. The characteristic luminescence spectra in the wavelength range between 1.5 and 1.6 μm were observed. Differences in the multiple line structure of the spectra indicated that the active luminescent centers have different symmetries and atomic structure. The dependence of the photoluminescence intensity on the laser excitation power and on the temperature was measured. Results are discussed on the basis of a physical model which includes the formation of free excitons, the binding of excitons to erbium ions, the excitation of 4f inner-shell electrons of the erbium ions, and their subsequent decay by light emission. To obtain a quantitative agreement between model analysis and experimental data the consideration of Auger processes by which erbium-bound excitons and erbium ions in excited state can decay by dissipating energy to conduction band electrons appears to be required. From the temperature dependence two activation energies are derived which are associated with the exciton binding energies and with an energy transfer process from excited erbium ions back to erbium-bound excitons, respectively. A good quantitative agreement can be obtained for suitable values of the model parameters. The luminescent properties of the samples of the different types of crystalline silicon are remarkably similar.

Dimer of substitutional carbon in silicon studied by EPR andab initiomethods

An EPR signal observed in carbon-doped float-zone silicon after irradiation with 2-MeV electrons at room temperature has been investigated. It represents a defect with S=(1/2), an apparently isotropic g factor (=2.0030), and a complicated hyperfine structure from {sup 29}Si nuclei in five shells that are consistent with an overall trigonal symmetry. Subtle asymmetries of the hyperfine pattern indicate the presence of a small trigonal component of the g tensor as well. An additional pair of satellite lines is identified by the relative intensity (1%) as arising from {sup 13}C in natural abundance, occupying two equivalent sites on the trigonal axis. Several defect structures that contain two equivalent carbon atoms on a trigonal axis were investigated by ab initio Hartree-Fock calculations. Only the negative charge state of a dicarbon center C{sub s}-C{sub s}, in which the carbon atoms occupy adjacent substitutional sites, was found to be consistent with the EPR data. (c) 2000 The American Physical Society.

Deep levels in high-energy proton-irradiated tin-doped n-type Czochralski silicon

A deep level transient spectroscopy study of defects created by 61 MeV proton irradiation of tin-doped n-type Czochralski silicon is reported. A comparison is made with the deep levels observed in irradiated p–n junction diodes fabricated in n-type float-zone silicon, without tin doping. The main conclusions are that in Sn-doped material, at least two additional deep radiation centers are introduced at 0.29±0.01 and 0.61±0.02 eV below the conduction band. From annealing experiments, it is concluded that these electron traps dissociate below 120 °C, which is lower than observed before for Sn–V related levels. It is demonstrated that the introduction rates of the well-known radiation defects are significantly smaller in Sn-doped material.

Defects and transport in a-Si:H/c-Si heterojunctions

We have investigated amorphous silicon/crystalline silicon np and pn heterojunction devices with transport and capacitance measurement techniques. The devices consist of doped n-type or p-type hydrogenated amorphous silicon deposited on p-type or n-type float-zone crystalline silicon wafers. Low temperature (100 K) photocurrent measurements in conjunction with numerical simulations indicate a large conduction band offset at the interface between the two materials. We show experimentally and with numerical simulation that defects at the heterojunction interface affect the low frequency capacitance and the electronic transport in the devices.

Silicon wafer oxygenation from SiO2 layers for radiation hard detectors

Radiation produces lattice damage in silicon by displacing the atoms from their original positions and thereby generating the corresponding defects. As a result, new states are created in the semiconductorforbidden band gap, negatively affecting the electrical performance of the devices. Endurance to radiation can be improved by having a high oxygen concentration in the silicon. For detector fabrication, high resistivity silicon is also needed, thus float zone wafers are preferred; however, this kind of material exhibits a low oxygen concentration. Although different ways to incorporate oxygen in float zone silicon have been proposed, all of them imply modifications during the ingot growth. Thermal diffusion from SiO2 layers on polished wafers is an interesting alternative to improve their oxygen content. Different thermal processes aimed at obtaining oxygen enriched silicon for the fabrication of radiation hard detectors have been tested. Attention has also been paid to carbon introduction during processing since, high concentrations of this element has been proved deleterious.

Divacancy-tin complexes in electron-irradiated silicon studied by EPR

$n$- and p-type float-zone silicon containing ${10}^{18}\ensuremath{-}{\mathrm{cm}}^{\ensuremath{-}3}$ tin were irradiated with 2 MeV electrons to a dose of ${10}^{18}{\mathrm{cm}}^{\ensuremath{-}2}$ and subsequently examined by electron paramagnetic resonance (EPR). The p-type material yields only the well-known Si-G29 signal due to the tin-vacancy complex ${\mathrm{SnV}}^{0},$ whereas the as-irradiated n-type material in addition displays the Si-G7 signal $({\mathrm{V}}_{2}^{\ensuremath{-}}),$ DK4, recently assigned to ${\mathrm{SnV}}^{\ensuremath{-}}$ in a set of slightly inequivalent configurations, and a new signal DK1, from a defects with $S=1/2$ containing one tin nucleus. DK1, which we assign to $(\mathrm{SnV}\ensuremath{-}\mathrm{V}{)}^{\ensuremath{-}},$ undergoes a reversible triclinic-monoclinic transformation at $\ensuremath{\approx}15\mathrm{K}.$ Annealing at 428 K removes Si-G29 and DK4 and produces a six-fold increase of Si-G7 and DK1, the kinetics indicating that about 50% of SnV is transformed into ${\mathrm{V}}_{2}$ and (SnV-V). Annealing at 503 K destroys Si-G7 and DK1, the decay of DK1 being linked to the emergence of two new signals DK2 and DK3 from defects with $S=1/2,$ monoclinic-$I$ symmetry, and two equivalent tin nuclei each, which we identify as $(\mathrm{SnV}\ensuremath{-}\mathrm{VSn}{)}^{\ensuremath{-}}$ and $({\mathrm{Sn}}_{2}\mathrm{V}\ensuremath{-}\mathrm{V}{)}^{\ensuremath{-}}.$ The structures of the tin-divacancy complexes are discussed in terms of modifications imposed on the basic divacancy structure by the larger size and lower ionization potential of the tin atom as compared to silicon. A model is proposed for the migration of (SnV-V) in the lattice at 500 K, indicated by the process (SnV-V)+Sn$\ensuremath{\rightarrow}$(SnV-VSn).

Study of Vacancy and Impurity Complexes in Si Solid-Phase Epitaxial Crystallization with Positron Annihilation Spectroscopy

AbstractWe investigate the residual vacancy defect species after crystallization of amorphous Si (a-Si) by solid phase epitaxy (SPE). To this end, we correlate the total and electronically-active doping concentrations measured by secondary mass spectrometry and spreading resistance analysis, and data from positron annihilation spectroscopy (PAS), which is sensitive to openvolume defects. Float-zone silicon substrates were implanted with boron, phosphorus and both phosphorus and boron ions to create nonuniform doping profiles at degenerate doping levels, after an amorphization step by 29Si+ ions. Samples were vacuum annealed at 600°C to induce SPE, and the SPE rate was measured by time-resolved reflectivity. PAS was used for identification of the impurity-defect complexes. Momentum-resolved PAS measurements enable the detection of phosphorus-vacancy (P-V) and oxygen-vacancy (O-V) complexes.

Impact of the divacancy (?) on the generation-recombination properties of 10 MeV proton irradiated Float-Zone silicon diodes

In this paper, the correlation between the microscopic damage parameters and the macroscopic device performance degradation is investigated for 10 MeV proton-irradiated Float-Zone (FZ) silicon diodes, with n- and p-type doping density in the 10 14 cm −3 range. It is shown that the radiation-induced deep-level concentrations, obtained from deep-level transient spectroscopy show a proportional increase with the proton fluence, for the dominant traps and this both in n- and p-type substrates. The same applies for the increase in the reverse diode current and the reduction of the inverse carrier recombination lifetime. The obtained spectroscopic information is substituted in the Shockley–Read–Hall model for the diode reverse current and recombination lifetime. From the comparison between calculated and measured parameters, it is concluded that in proton-irradiated diodes fabricated in FZ p- and n-type substrates, the divacancy related deep-level centres at E c −0.42 eV play a dominant role in both the generation and the recombination lifetime.

Structural and electrical properties of silicon epitaxial layers grown by LPE on highly resistive monocrystalline substrates

The present paper reports on the properties of liquid-phase epitaxy (LPE) silicon layers grown on monocrystalline highly resistive (1 0 0)-oriented float zone (FZ) silicon substrates. All layers were grown in an indium/gallium solution saturated with Si at temperatures near 920°C. For the optimisation of layer properties for solar cell applications, parameters such as growth rate and Ga concentration of the In solution were varied. Hall measurements, X-ray spectroscopy, mechanical surface profiling (Dektak) and a lifetime analyser were used to characterise the 100–200 μm thick LPE layers.

Metal contamination analysis of the epitaxial starting material for scientific CCDs

Traces of unintentionally introduced titanium into a 3 k Ω cm float-zone epitaxial silicon at ultra low levels of 10 10 cm −3 were found to be the origin of charge transfer loss in pn-CCDs. We identified and backtracked the titanium impurity. The full-depletion design of the pn-CCD, a thin entrance window at the back of the CCD and the low oxygen content of the float-zone material allow no common gettering step. Titanium is introduced into the wafer during the epitaxy process. This is independent of the reactor and the producer of the epitaxial silicon and seems to be a common epitaxy-related problem. To identify the impurity, electron emission rates of traps were measured by means of the CCD. The data were compared with emission rates of identical material obtained by standard-DLTS. The data agree well with literature data of the titanium acceptor level and the titanium donor level. An analysis of the capture cross section by means of the CCD gives a high electron capture cross section. This explains the strong effect of titanium even in very little concentrations in the pn-CCD. A two-dimensional distribution of the trap concentration was visualized on some wafers.

Annealing of the photoluminescence W-center in proton-irradiated silicon

We study radiation defects generated by 55MeV protons in float-zone silicon using photoluminescence (PL) and deep-level transient spectroscopy (DLTS). The investigated material is weakly phosphorus-doped and contains varying amounts of interstitial oxygen (<2×1014…1.7×1017cm−3). Special attention is paid to the as of yet unidentified luminescence W-center. Its isochronal annealing behavior (40–160°C) is discussed in conjunction with the evolution of the cluster damage-related features in the DLTS spectrum. We find the annealing rate of the W-center to be independent of the oxygen content and thus exclude oxygen as a constituent of the center. There is no evident correlation between the main peaks in the DLTS spectrum and the W-center. However, a weak signal at EC-0.075eV is found to be a good candidate for a DLTS signal of the center. Further studies are needed to relate the formation of the W-center to the interstitials and/or vacancies released from damage clusters during annealing, which might help in identifying the constituents of the complex defect.