Liquid Encapsulated Czochralski Method Research Articles

Absorption spectrum of EL2 defect in p-type GaAs

Absorption of p-type GaAs crystals grown by horizontal Bridgman, liquid-encapsulated Czochralski, and liquid-phase electroepitaxy methods were measured at 4 K. The spectra of melt-grown crystals show a wide absorption band extending from 0.6 eV to the fundamental absorption edge. The comparison with the photocapacitance spectrum of EL2 defect and deep-level transient spectroscopy measurements allowed to interpret this absorption as due to transitions between the valence band and the doubly ionized EL2 level. Absorption can be bleached by intense illumination at temperatures below 60 K with the dark spectrum recovering upon heating with an activation energy of 55 meV.

Deep Ti donor in GaAs

Ti-doped GaAs crystals grown by the liquid-encapsulated Czochralski method were investigated by using the observation of direct capacitance transients. The charge transfer transitions to the deep Ti donor level close to midgap are unambiguously identified in both n-type and p-type material. The energy position of Ti3+/Ti4+ is determined with high precision of Ec−(0.87±0.01) eV at 300 K and the respective cross sections are σn∞=(7±3)×10−15 cm2 for electron capture and σp∞=(10±5)×10−16 cm2 for hole capture. The position of the Ti2+/Ti3+ acceptor is confirmed to be at Ec−(0.19±0.01) eV. Its cross section for electron capture is σn∞=(3±1)×10−16 cm2.

The plasticity of GaAs between 415 and 730°C

Abstract GaAs single crystals grown by the liquid-encapsulated Czochralski method are compressed along ‘123’ under a protective atmosphere of flowing Ar. The resulting stress-strain curves, which are rather similar to those observed in other III-V compounds, are analysed according to the beginning of plastic deformation and to dynamical recovery. The lower yield point is characterized by an activation energy of l·24eV and a stress exponent of 3·5. Both values compare favourably with those obtained from measurements of the dislocation velocity. From the strain-rate and temperature dependence of the stress τIII at the onset of the first recovery stage, an activation energy of approximately 2eV is derived, which may be regarded as a lower bound for the activation energy of self-diffusion of the slowest-moving species. Comparing these results with those obtained for other semiconductors corroborates a view increasingly supported in the literature, namely that moving dislocations emit or absorb point effects.

Photoreflectance measurements of indium content in indium-alloyed semi-insulating GaAs substrates

Room-temperature photoreflectance measurements were used to determine the radial and axial distribution of low levels of indium in 3-in.-diameter semi-insulating bulk GaAs materials grown by the liquid-encapsulated Czochralski method. These results were compared with 4.2-K photoluminescence data and found to be accurate and more convenient for this application. Room-temperature photoreflectance allows an accurate determination of the indium content in the range of mole fraction 0.1%–2.0% with standard deviation of 0.03%. Two types of radial inhomogeneity were found in commercially available GaAs wafers. This is discussed in terms of indium segregation and shape of solid and liquid interface during the crystal growth.

A study on dislocation lines in an In-doped LEC GaAs crystal by IR light scattering and transmission microscopy

Defects, especially dislocations in an In-doped GaAs crystal grown along [001] by liquid encapsulated Czochralski (LEC) method, were microscopically visualized by infrared light scattering tomography to study growth history and mechanism of the crystal. Here, three types of dislocations were found: (i) Grown-in dislocations, in which helical and bent dislocations were included. (ii) Dislocations introduced within liquid-solid interfaces. Some ends of the dislocations penetrated into the just-grown ingot and slipped on the (100) planes due to thermal stresses, because the ingot was still growing with a large temperature gradient from its molten phase. The opposite end terminals remained within the interfaces at their initial stages but some of them reacted between themselves to diminish their density or to make a linkage with the terminals. The terminals and some other dislocation ends acted as origins of the grown-in dislocations. (iii) A lot of dislocations were introduced into the ingot from its outer-side surface by thermal stresses during its cooling stage after the growth.

Characterisation of deep electron states in LEC grown GaAs material

Deep electron states in undoped GaAs grown by the liquid-encapsulated Czochralski (LEC) method using a pyrolitic boron nitride (PBN) crucible were investigated by DLTS and Hall measurements. Six electron traps with activation energies ranging from 0.14 to 0.82 eV were detected. One of these traps, with an energy of 0.27 eV (EL6), was reduced in concentrations by more than an order of magnitude by an 800 degrees C heat treatment for 1 h. Temperature-dependent Hall measurement on as-grown material indicated that the 0.27 eV state plays an important role in the decrease of the resistivity of undoped crystals that have a low concentration of carbon impurity. The effect of gamma irradiation on the annealing properties of the traps was also demonstrated.

On the character of defects in GaAs

Positron lifetime measurements on GaAs are presented and discussed and former measurements are reviewed. The limitations and appropriate criteria for adequate spectrum analyses are considered in detail. It is shown that exceptionally good statistical accuracy is necessary for a reliable and consequential decomposition, and that source corrections are very important. Results for liquid phase electro-epitaxially grown GaAs conclusively show that the bulk lifetime for GaAs is 220+or-1 ps. All other GaAs samples (grown by either the liquid encapsulated Czochralski method or by the horizontal Bridgman method) show variable bulk lifetimes up to 232 ps. Shallow traps are concluded to be the cause for this. The binding energy of positrons in such traps is estimated to be about 23 meV, and the trapping rate into deep traps from such shallow traps is found to be approximately five times smaller than from the bulk state. The shallow trap is likely to be substitutional boron or nitrogen. The shallow traps anneal mainly around 700 K. Deep traps are found in all samples, yielding lifetimes tau DI approximately=260 ps and 290 ps<or= tau DII<or=315 ps. These lifetimes occur in n-type, p-type and semi-insulating materials. The results support the authors' earlier contention that the 290-315 ps components are due to divacancies and the 260 ps lifetime is due to monovacancies, in complex form with some impurity. The divacancy anneals around 620 K while the mono-vacancy-impurity complexes generally anneal around 800 K.

Electrical activation of silicon coimplanted with nitrogen, phosphorus, or arsenic in semi-insulating GaAs substrates grown by the liquid encapsulated Czochralski method

We fit some existing data on the electrical activation of Si ions coimplanted with either N, P, or As ions in semi-insulating GaAs substrates grown by the liquid encapsulated Czochralski method. For one data set the activation anneal was done with proximity capped samples in a flowing H2 ambient, and for the other data sets anneals were done with dielectric capped samples. We are able to obtain good fits to all data with simple choices for the depth profiles of implanted ions if we adopt the following assumptions: (1) With no group V coimplants, the Si donors formed during a dielectric cap anneal, possibly SiGaVAs, are 100% ionized, while some of those formed during a proximity cap anneal in H2, possibly SiGaHAs, are less than 100% ionized and also serve to eliminate some arsenic vacancies. (2) All Si-related acceptors appear as BGaSiAs defects. (3) The observed enhancement of Si activation with coimplanted P is due to a decreased concentration of BGaSiAs acceptors whose formation is inhibited by P. (4) Coimplanted N or As does not inhibit the formation of BGaSiAs acceptors but, like coimplanted P at high doses, forms deep acceptors.

Deep acceptorlike recombination centers in bulk liquid encapsulated Czochralski GaP, studied with optically detected magnetic resonance

The optically detected magnetic resonance technique is applied to determine the nature of broad, featureless emission bands observed in the near-infrared region in bulk GaP crystals grown by the liquid encapsulated Czochralski method. A broad emission band with maximum at ≊8000 Å (≊1.55 eV) was consistently observed in undoped and donor-doped (S, Te, Se, Ge) crystals, and is shown to be due to donor-acceptor pair (DAP) recombination. The analysis of the optically detected magnetic resonance experiments indicates that shallow donors and deep acceptors are active in the 8000-Å recombination. The identity of the ≊0.7-eV deep acceptor center active in this DAP transition could not be determined from the experimental results, but a single acceptor complex consisting of a Ga vacancy and two adjacent donors is proposed as a tentative model.

Stoichiometric structures of defects in high-purity GaAs grown by the liquid encapsulated Czochralski method

We analyze some existing data obtained on a GaAs sample grown by the liquid encapsulated Czochralski (LEC) method from a near-stoichiometric melt after the sample was cycled through various thermal processes. By using the constraint of constant deviation from stoichiometry we are led to suggest that the defects observed or inferred to exist in the sample have the following properties: (1) the acceptor associated with the 1.45-eV photoluminescence signal has the stoichiometric structure of GaAs if doubly charged or of VGaGaAs if singly charged; (2) the (presumed) donor at Ec−0.134 eV has the stoichiometric structure of VAs; (3) another (inferred) acceptor has the stoichiometric structure of VGa; and (4) the very shallow donor at Ec−0.003 eV is the precursor of EL2 and becomes EL2 upon reaction with VGa or its stoichiometric equivalent.

ChemInform Abstract: Growth and Characterization of Large Size, High Quality, InP Single Crystals.

AbstractSingle‐crystal InP wafers grown by both the vertical gradient freeze (VGF) technique (described in detail) and the liquid encapsulated Czochralski (LEC) method are studied by transmission X‐ray topography, transmission cathodoluminescence, and defect revealing etching.

Deep levels in semiconducting In-alloyed bulk n-GaAs and its resistivity conversions by thermal treatments

Thermal conversions of resistivities have been studied for In-alloyed semiconducting (100–106 Ω cm) n-GaAs grown by the liquid-encapsulated Czochralski method. These dislocation-free as-grown crystals are converted into semi-insulating (&amp;gt;107 Ω cm) crystals by annealing at 950 °C for 2 h followed by a fast cooling. Such semi-insulating crystals can be converted further into crystals with lower resistivities (∼106 Ω cm) by treating them at 470 °C for 100 h. In the analysis of the as-grown samples by the temperature-dependent Hall measurements, four levels have been found with activation energies 0.13, 0.20, 0.42, and 0.50 eV. It has been shown that these resistivity conversions are induced by concentration changes of the deep states other than the midgap donor EL2, some of them being the levels found in this study.

An anomalous feature on the DLTS spectrum of silicon : C. A. Londos. Solid St. Commun.62, 719 (1987)

Photoluminescence spectroscopy, secondary ion mass spectrometry and Hall profiling measurements are used to study the effects of encapsulation and annealing on semi-insulating (SI) InP. Control of both indium and phosphorus loss from the InP surface during annealing is necessary to prevent anomalous activation of shallow, low dose silicon implants in SI InP. Phosphorus vacancy-related complexes are detected for encapsulant deposition temperatures in excess of 300°C and their concentration increases with increased deposition temperatures. Indium out-diffusion is detected for furnace anneals at 750°C for 30 min of samples capped with oxygen-free silicon nitride (Si3N4) or phosphosilicate glass (PSG), but substantially more indium is detected in the PSG film than in the Si3N4. Silicon indiffusion is greatest for PSG-capped InP grown by the liquid-encapsulated Czochralski method and least for Si3N4-capped material grown by liquid phase epitaxy. Conducting layers with sheet electron concentrations approaching 2×1012 cm-2 are always formed in PSG-capped samples, but only occasionally in Si3N4-capped samples. Net doping is strongly correlated with indium out-diffusion and silicon in-diffusion, suggesting that under the anneal conditions studied here the supply of indium vacancies limits silicon in-diffusion from the cap and also plays a significant role in determining the amount of doping produced from the in-diffused silicon. Iron was observed to be gettered to the encapsulant-substrate interfacee and, in some Si3N4-capped samples, to be depleted from the near-surface region. These results point to the need for low temperature dielectric deposition processes and rapid thermal annealing for controllable low dose implant activation in SI InP.

Microcontamination reduction in PECVD systems—Part 2 : D. G. Hemmes. Semiconductor int., 98 (June 1987)

Photoluminescence spectroscopy, secondary ion mass spectrometry and Hall profiling measurements are used to study the effects of encapsulation and annealing on semi-insulating (SI) InP. Control of both indium and phosphorus loss from the InP surface during annealing is necessary to prevent anomalous activation of shallow, low dose silicon implants in SI InP. Phosphorus vacancy-related complexes are detected for encapsulant deposition temperatures in excess of 300°C and their concentration increases with increased deposition temperatures. Indium out-diffusion is detected for furnace anneals at 750°C for 30 min of samples capped with oxygen-free silicon nitride (Si3N4) or phosphosilicate glass (PSG), but substantially more indium is detected in the PSG film than in the Si3N4. Silicon indiffusion is greatest for PSG-capped InP grown by the liquid-encapsulated Czochralski method and least for Si3N4-capped material grown by liquid phase epitaxy. Conducting layers with sheet electron concentrations approaching 2×1012 cm-2 are always formed in PSG-capped samples, but only occasionally in Si3N4-capped samples. Net doping is strongly correlated with indium out-diffusion and silicon in-diffusion, suggesting that under the anneal conditions studied here the supply of indium vacancies limits silicon in-diffusion from the cap and also plays a significant role in determining the amount of doping produced from the in-diffused silicon. Iron was observed to be gettered to the encapsulant-substrate interfacee and, in some Si3N4-capped samples, to be depleted from the near-surface region. These results point to the need for low temperature dielectric deposition processes and rapid thermal annealing for controllable low dose implant activation in SI InP.

Electrical activation of silicon implanted in semi-insulating GaAs: Role of boron and the midgap electron trap (EL2)

A model based on the constraint of stoichiometry is developed to account for some existing data on the donor activation of silicon ions implanted in semi-insulating, boron-doped GaAs grown by the liquid encapsulated Czochralski method. As a function of increasing implant dose (up to 1013 ions/cm2), three stages of activation are pictured: facilitation by EL2, inhibition by boron, and finally, 100% differential activation when boron is depleted. Possible atomic structures are suggested for the boron-related defects and silicon-related donors. A proposal for using implant measurements to help establish the atomic structure of EL2 is advanced.

Distribution coefficient of carbon in melt-grown GaAs

The results of a detailed investigation of the distribution coefficient of carbon in GaAs are reported for a series of crystals grown at different pulling rates by the liquid-encapsulated Czochralski method. From localized vibration mode absorption measurements, it was found that the concentration of carbon always decreases along the length of the ingots and that the effective distribution coefficient increases strongly with a decrease in pulling rate. The equilibrium distribution coefficient determined from present data, K0=3.5, is higher than previously reported values for the effective distribution coefficient; however, it is consistent with chemical trends for group IV elements in GaAs.

Growth and characterization of low defect GaAs by vertical gradient freeze

As the diameter of GaAs substrates has increased, so has the difficulty and cost of growing GaAs by the liquid encapsulated Czochralski (LEC) method. One alternative crystal growth method with offers promise is growth by vertical gradient freeze (VGF). This technique avoids the problems associated with diameter control during low gradient LEC growth. In this paper we discuss the growth of 2 inch diameter undoped and In-alloyed GaAs by VGF. Through the use of low gradients, undoped material with a dislocation density of (2–6) x 10 3 cm -2 has been obtained. With the addition of 0.6 at% InAs to the melt, this density is further reduced to 0–1000 cm -2. In contrast to standard LEC, GaAs grown by VGF shows low concentrations of the EL2 deep level and of the C shallow acceptor level. The material is semi-insulating ρ ⩾ 10 7 ω cm, and is thus suitable for device applications. From temperature dependent Hall measurements, the semi-insulating (SI) behavior is believed to be caused by Fermi level pinning at a defect with an energy level of ∼ 0.5 eV.

Growth and characterization of GaAs alloyed with In and P

Indium- and phosphorus-alloyed GaAs crystals were grown by the liquid encapsulated Czochralski (LEC) method. Phosphorus was found to complement indium, which has relatively low concentration at the seed end, in providing the lattice hardening. the dislocation density was low even at the seed end of the crystals. Lattice-matched wafers exhibited near zero band gap shift in the crystal where the ratio of phosphorus to indium ranged from 1 to 1.3. The distribution coefficient of phosphorus was determined to be 3.0±0.3 for the two crystals grown at 2 and 4 mm/h. The crystals exhibited high room-temperature electron mobility of 000 cm 2/V·s and resistivities ranging from low 10 7 to low 10 8 ω cm. Temperature-dependent Hall measurements revealed no extraneous levels attributable to indium or phosphorus in the band gap.

Dislocation bundle formation during liquid encapsulated Czochralski growth of GaAs crystals

Residual dislocations in a P-alloyed and an undoped GaAs crystal grown by the liquid encapsulated Czochralski method were investigated by transmission electron microscopy. In the undoped GaAs, dislocation bundles constructing cell walls and/or lineages consist of at least three different Burgers vectors. In the P-alloyed GaAs with low dislocation density, individual dislocation reactions were observed. Sessile dislocations were found to be formed by an interaction between two kinds of dislocations. The sessile dislocation formation may be a first step for the complicated dislocation structure in the undoped crystal.

Growth of large diameter dislocation-free indium phosphide ingots

InP single crystals have been grown in a Metal Research MSR6R puller by the liquid encapsulation Czochralski method. A home made automatic diameter control system has been developed. Up to 1.3 kg in weight, 52 mm in diameter ingots are pulled with a diameter control accuracy better than ±4% over the useable length. Dislocation-free (DF) sulphur doped substrates for optoelectronics applications have been obtained in two ways: (i) using a sulphur doping level higher than 6 x 10 18 cm -3; (ii) using a lower sulphur doping level (as low as 2 × 10 18cm -3), modified growth conditions and a necking procedure, either in the 〈001〉 or 〈111〉 pulling direction. Overall assessment of DF crystals includes X-ray transmission topography and multiple X-ray diffraction measurements leading to the measurement of the lattice parameter a in absolute value. A calibration curve for Δ a/a versus sulphur concentration has been made. The Δ a/a sensitivity is 5 x 10 -6, which gives an [S] sensitivity ≌10 18cm -3. The best dislocation free material thus exhibits a Δ a/a value of the order of 10 -5. Crystals pulled along the 〈111〉 direction exhibit a facetted growth core. From X-ray topograph transmission and calibrated IR absorption measurements, a sulphur doping variation in the facet and off the facet region has been carefully measured and precise data on this effect will be given.