Defect-formation processes in silicon doped with manganese and germanium

Deep-level transient spectroscopy has been used to study the effect of Ge atoms on the behavior of Mn in Si. It is shown that Ge atoms introduced into Si during growth manifest no electrical activity, even though their concentration is rather high: 10¹⁶–10¹⁹ cm⁻³. It is established that the presence of Ge atoms in the Si lattice enhances the efficiency of the formation of the deep levels E _c -0.42 eV and E _c -0.54 eV, which are associated with Mn in the Si lattice: the concentration of these deep levels in Si〈Ge, Mn〉 samples is a factor of 3–4 greater than in Si〈Mn〉. It is found that the presence of Ge atoms stabilizes the properties of the Mn levels in Si: They anneal more slowly than in Si〈Mn〉 by a factor of 5–6. It is assumed that the detected effects are associated with the features of the defect structure of Si doped with Ge and Mn. Fiz. Tekh. Poluprovodn. 32, 676–678 (June 1998)     

The properties of Si/Si1-xGex films grown on Si substrates by chemical vapor deposition

A growth parameter study was made to determine the proper of a SiGe superlattice-type configuration grown on Si substrates by chemical vapor deposition (CVD). The study included such variables as growth temperature, layer composition, layer thickness, total film thickness, doping concentrations, and film orientation. Si and SiGe layers were grown using SiH₄ as the Si source and GeH₄ as the Ge source. When intentional doping was desired, diluted diborane for p-type films and phosphine for n-type films were used. The study led to films grown at ∼1000°C with mobilities from ∼20 to 40 percent higher than that of epitaxial Si layers and ∼100 percent higher than that of epitaxial SiGe layers grown on (100) Si in the same deposition system for net carrier concentrations of ∼8x10¹⁵ cm^-3 to ∼2x10¹⁷ cm^-3. Enhanced mobilities were found in multilayer (100)-oriented Si/Si_1-xGe_x films for layer thicknesses ≥400A, for film thicknesses >2μm, and for layers with x = 0.15. No enhanced mobility was found for (111)-oriented films and for B-doped multilayered (100)-orlented films. Supported in part by NASA-Langley Research Center, Hampton, VA, Contract NAS1-16102 (R. Stermer & A. Fripp, Contr. Mon.)     

Au/Ge/Ni ohmic contacts to n-Type InP

The relationship between the electrical properties and microstructure for annealed Au/Ge/Ni contacts to n-type InP, with an initial doping level of 10¹⁷ cm^-3, have been studied. Metal layers were deposited by electron beam evaporation in the following sequence: 25 nm Ni, 50 nm Ge, and 40 nm Au. Annealing was done in a nitrogen atmosphere at 250-400‡C. The onset of ohmic behavior at 325‡C corresponded to the decomposition of a ternary Ni-In-P phase at the InP surface and the subsequent formation of Ni₂P plus Au₁₀In₃, producing a lower barrier height at the InP interface. This reaction was driven by the inward diffusion of Au and outward diffusion of In. Further annealing, up to 400‡C, resulted in a decrease in contact resistance, which corresponded to the formation of NiP and Au₉ln₄ from Ni₂P and Au₁₀In₃,respectively, with some Ge doping of InP also likely. A minimum contact resistance of 10^-7 Ω-cm² was achieved with a 10 s anneal at 400‡C.     

Cavity formation and impurity gettering in He-implanted Si

Cavity microstructures formed in Si after ion implantation of He (30 or 130 keV) and annealing at 700°C or above are examined with cross-section transmission electron microscopy. A threshold concentration of 1.6 at.% He is identified as required to form cavities that survive such anneals. The cavities coarsen with a constant volume corresponding to ∼0.75 lattice sites per implanted He atom and have surface areas 3-7 times that of the wafer area for fluences of 1 × 10¹⁷ He/ cm². Transition metal atoms (Cu, Ni, Co, Fe, Au) are shown to be strongly trapped (1.5–2.2 eV) on the cavity walls by chemisorption. Whereas Cu, Au, and Ni are bound more strongly to the cavity sites than to their respective precipitated phases, Co and Fe are more strongly bound to their silicides; nonetheless, appreciable trapping of Co and Fe does occur in equilibrium with the silicides. Cavity trapping appears to be an effective gettering mechanism at low impurity levels, as needed to meet future microelectronics device requirements.     

P- and N-type doping of GaN and AlGaN epitaxial layers grown by metalorganic chemical vapor deposition

Mg- and Si-doped GaN and AlGaN films were grown by metalorganic chemical vapor deposition and characterized by room-temperature photoluminescence and Hall-effect measurements. We show that the p-type carrier concentration resulting from Mg incorporation in GaN:Mg films exhibits a nonlinear dependence both on growth temperature and growth pressure. For GaN and AlGaN, n-type doping due to Si incorporation was found to be a linear function of the silane molar flow. Mg-doped GaN layers with 300K hole concentrations p ∼2×10¹⁸ cm⁻³ and Si-doped GaN films with electron concentrations n∼1×10¹⁹ cm⁻³ have been grown. N-type Al_0.10Ga_0.90N:Si films with resistivities as low as p ∼6.6×10⁻³ Ω-cm have been measured.     

Recent status on high temperature superconducting Bi2Sr2CaCu2O8+x wire development at NYSIS: 1-90 meter length Jcs and 3 meter diameter ring furnace design

The status of long length, Bi₂Sr₂CaCu₂O_8+x (Bi-2212) wire development at the New York State Institute on Superconductivity (NYSIS) is reviewed and updated. Transport J_cs (4.2K, 0 T) of Bi-2212/Ag oxide powder-in-tube singlefilamentary tapes have reached 70,000-80,000, 50,000-60,000, and 30,000–40,000 A/cm² for 1, 4–15, and 40–90 meter length tapes, respectively. The decrease in J_c as the tape length was increased from 15 to 90 meters was attributed to the (measured) sensitivity of J_c to temperature nonuniformities (±3‡C) in the box-type furnace used for annealing. To reduce this problem, a ringtype high-temperature furnace (∼3 meter diameter) was designed and constructed which provides a large-volume (∼13^w × 10^h × 1000¹ cm) processing zone with expected excellent temperature uniformity (±0.5‡C). The advantages of the ring-type furnace for processing of kilometer-length conductors are described.     

Germanium as a deep level in alxga1− xas grown by organometallic vapor phase epitaxy

Optoelectronic devices require materials which exhibit extremely low trap concentrations. The Al_xGa_1−xAs system has been used extensively for optoelectronic applications despite trap concentrations in the Al_xGa_1−xAs which limit the efficiency of the resulting devices. Deep level transient spectroscopy (DLTS) performed on Al_0.2Ga_0.8As layers grown by organometallic vapor phase epitaxy (OMVPE) has revealed three traps with concentrations >10¹³ cm⁻³ -E _c-E_t = 0.3, 0.5 and 0.7 eV. The dominant source of the 0.3 eV trap has proven to be a Ge impurity in arsine. SIMS analysis of Al_0.2Ga_0.8As samples show Ge as the only candidate for the impurity responsible for the 0.3 eV trap. DLTS and SIMS analysis performed on Al_0.2Ga_0.8As samples intentionally doped with Ge displayed a proportional increase in the 0.3 eV trap concentration with the Ge concentration and establishes that Ge is indeed the source of the 0.3 eV trap in Al_xGa_1−xAs. Comparison of C-V, SIMS and DLTS measurements performed on Al_xGa_1-xAs:Ge indicate that approximately 30% of elemental Ge incorporated created the 0.3 eV trap, DX_Ge.     

The lattice position of diffused Zn in InP

We have measured the substitutional fraction (f_s) for Zn atoms diffused into InP crystals using the proton-induced x-ray excitation (PIXE) technique. Diffusion times ranged from 15–60 min at 425–650° C. For several samples with diffusion depths in the range 0.75-3.7 μm (as determined by SIMS analysis), we find that the Zn impurity atoms reside almost totally on lattice sites: f_s = 0.9 ± 0.1. Contrary to results of an earlier study, we find no evidence for precipitates in the diffused layers. However, only ∼10^-3-10^-1 of the Zn is electrically active, consistent with Tuck and Hooper’s suggestion of neutral V_p Zn_In V_p complexes.     

Dislocation–impurity interaction in Si

The dynamic interaction between dislocations and impurities B, P and Ge in Si with concentrations up to 2.5×10²⁰ cm⁻³ is investigated by the etch-pit technique, in comparison with that of O impurity in Si. Dislocation generation from a surface scratch is strongly suppressed when the concentration of B and P impurities exceeds 1×10¹⁹ cm⁻³, originating from the immobilization by preferential impurity segregation. Dislocation velocity in motion enhances on increasing the concentration in B and P impurities. Neutral impurity Ge has weak effect on dislocation generation and velocity enhancement.     

Doping and impurity compensation by ion implantation in a-SiGe films

This paper discusses the electrical properties of a-SiGe films (N _Ge∼2.2 at. %) prepared by co-evaporation of Si and Ge from separate sources and doped by ion implantation of substitutional impurities (B⁺ and P⁺), as well as the results of controlled impurity compensation by ion-beam doping. It was found that B⁺ and P⁺ implantation into a-SiGe films in the dose range 1.3×10¹⁴–1.3×10¹⁷ cm⁻², followed by annealing at 350 °C, increased the conductivity of these films from 10⁻⁹ to 10⁻⁴ and to 10⁻⁵ S/cm for B⁺ and P⁺, respectively. The position of the Fermi level could be varied from (E _v+0.27) to (E _c−0.19) eV. These investigations indicate that compensation of pre-doped a-SiGe films by ion implantation is feasible and reproducible. It is also found that higher doping efficiency of a-SiGe films is obtained by using boron than by using phosphorus. Fiz. Tekh. Poluprovodn. 32, 1260–1262 (October 1998)     

Specific contact resistivity of indium contacts to n-type CdTe

Indium alloyed to n-type CdTe of about 10¹⁶ cm^-3 electron concentration provides a contact resistivity of about 7 x 10^-3 ohm cm². This is achieved by alloying for 10 minutes at 150-450‡C in a sealed ampoule with an overpressure of cadmium. If the alloying is done in an open tube H₂ flow without a Cd vapor overpressure, alloying temperatures above 250‡C cause the contact resistance to rise as cadmium vacancies increase the compensation in the CdTe. Further improvement of the contact resistivity to 1 x 10^-3 ohm cm is obtained by a 900‡C diffusion of In into the n-CdTe (electron concentration 10¹⁶ cm^-3 before the diffusion).     

Reversible electrical properties of LEC GaAs

Undoped, low-pressure, liquid-encapsulated Czochralski GaAs can be reversibly changed from conducting (ρ ∼ 1Ω-cm) to semi-insulating (ρ ∼ 10⁷Ω-cm) by either slow or fast cooling, respectively, after a 5 hr, 950° C soak in an evacuated quartz ampoule. The semi-insulating wafers are very uniform and lead to tight threshold-voltage control in direct-implant MESFET’s. We have studied crystals in both states by temperature-dependent Hall effect, photoluminescence, IR absorption, mass spectroscopy, and DLTS. It is shown that donor and acceptor concentrations are typically more than an order of magnitude greater than the C and Si concentrations, which are both less than 3 × 10¹⁴ cm⁻³. The EL2 concentration remains relatively constant at about 1.0 × 10¹⁶ cm⁻³. Thus, the normal EL2-Si-C compensation model does not apply. The most likely explanation for the reversibility involves a delicate balance between native-defect donors and acceptors in equilibrium at 950° C, but with the donors dominating after a slow cool, and the acceptors after a fast cool. A consistent model includes a dominant donor at E_c 0.13eV, probably V_As – As_Ga, and a dominant acceptor at E_v + 0.07eV, probably V_Ga Ga_As. In this model, vacancy motion is very important during the slow cool. Such processes must be strongly considered in the growth of bulk, high-purity GaAs.     

Solubility,diffusion and electrical activity of Na in bulk Ge crystals

By studying the drift of Na⁺ ions in the firstly grown Na-doped bulk Ge crystals as well as by analyzing optical and some other characteristics of this material, the following conclusions are made, many of which are different from the commonly accepted statements: (1) Ge can be uniformly doped with Na during the bulk Ge crystals growth from the melt; (2) maximum solubility at room temperature and distribution coefficient of Na in Ge are (0.3–1)×10¹⁵ cm⁻³ and (0.7–2.3)×10⁻⁷, respectively; (3) Na is a donor impurity in bulk Ge, and Na atoms introduced during the crystal growth are predominantly electrically active; (4) the evaluated values of diffusion parameters of Na in Ge are as follows: the diffusion coefficient D=3.6×10⁻⁷ cm²/s, pre-exponential factor D₀=0.13 cm²/s, the activation energy for diffusion Q=0.33 eV; (5) Na is an interstitial impurity in Ge and rather rapidly drifts in an electric field, most likely, via interstitial sites; (6) the resistance distribution along the crystal length may be changed by DC electric field application and remain stable at the long-term crystal storage. The stability in the Ge:Na properties opens the possibility for using Ge:Na crystals not only for creating passive optical elements of infrared imaging technique, as we are doing now, but also for the electrical appliances, in particular for the substitution of the thermally unstable Li for Na in germanium detectors of γ-radiation.     

Structures and physical properties of sputtered amorphous SiC films

Optical and electrical properties have been measured for amorphous SiC films prepared by rf sputtering in a pure Ar atmosphere with a sintered 6H-SiC target. The absorption edge E₀ determined from the relation of αhΝ = B(hΝ-E₀)² ranged from 1.45 to 1.80 eV depending on the film thickness and the substrate temperature. The room temperature electrical conductivity is in the range of 5.4×10⁻¹¹ and 1.4×10⁻⁵ Ω⁻¹cm⁻¹. The absorption edge decreases and the conductivity increases with increasing film thickness. The absorption edge shifts to shorter wavelengths (blue shift) and the conductivity decreases during annealing below 400‡C for 60 min, whereas the absorption edge shifts to the longer wavelength side (red shift) and the conductivity increases during annealing at 800‡C It is proposed that the two annealing processes cause structural changes in amorphous SiC films, one of which involves removal of defects or voids while the other involves rearrangement or rebonding of the component atoms.     

Electrically active defects in shallow pre-amorphisedp + n junctions in silicon

An examination of shallow pre-amorphisedp ⁺ n junctions in silicon has revealed three distinct defect related phenomena determined largely by the annealing temperature and relative location of the junction and the amorphous-crystalline (α-c) boundary. For temperatures below 800‡ C all samples displayed leakage currents of ∼10⁻³ A/cm² irrespective of the amorphising atom (Si⁺, Ge⁺ or Sn⁺). The generation centres responsible were identified to be near mid-gap deep level donors lying beyond the α-c interface. For samples annealed above 800‡ C, the leakage current was determined by the interstitial dislocation loops at the α-c boundary. If these were deeper than the junction, a leakage current density of ∼10⁻⁵ A/cm² resulted. From the growth of these loops during furnace annealing it was concluded that the growth was supported by the influx of recoil implanted silicon interstitials initially positioned beyond the α-c boundary. In the case where the as-implanted junction was deeper than the α-c boundary, annealing above 800° C resulted in a transient enhancement in the boron diffusion coefficient. As with the dislocation loop growth, this was attributed to the presence of the recoil implanted silicon interstitials.     

Short-wavelength photoluminescence of SiO2 layers implanted with high doses of Si+, Ge+, and Ar+ ions

The short-wavelength (400–700 nm) photoluminescence (PL) spectra of SiO₂ layers implanted with Si⁺, Ge⁺, and Ar⁺ ions in the dose range 3.2×10¹⁶–1.2×10¹⁷ cm⁻² are compared. After Ar⁺ implantation an extremely weak luminescence, which vanishes completely after annealing for 30 min at 400 °C or 20 ms at 1050 °C, was observed. After implantation of group-IV elements the luminescence intensities were 1 to 2 orders of magnitude higher, and the luminescence remained not only with annealings but it could also increase. The dose and heating dependences of the luminescence show that it is due to the formation of impurity clusters and this process is more likely to be of a percolation than a diffusion character. For both group-IV impurities an intense blue band and a weaker band in the orange part of the spectrum were observed immediately after implantation. The ratio of the excitation and emission energies of the blue luminescence is characteristic of oxygen vacancies in SiO₂; its properties are determined by the direct interaction of group-IV atoms. On this basis it is believed that the centers of blue PL are chains of Si (or Ge) atoms embedded in SiO₂. The orange luminescence remained after annealings only in the case of Si⁺ implantation. This is attributed directly to the nonphase precipitates of Si in the form of strongly developed nanometer-size clusters. Fiz. Tekh. Poluprovodn. 32, 439–444 (April 1998)     

Zinc-doping of InP during chemical beam epitaxy using diethylzinc

Diethylzinc was used as ap-type dopant source during InP growth by chemical beam epitaxy. In InP, electrically activated Zn saturated at a concentration of ∼2.0 × 10¹⁸ cm⁻³ for epilayers grown at 540‡ C. Higher role concentrations were obtained by lowering the growth temperature. However, measurements with SIMS indicated that very serious Zn diffusion occurred when the Zn concentration appeared to reduce the pyrolysis efficiency of trimethylindium. This caused a reduction in the InP growth rate and InAs mole fraction in InGaAs epilayers. No Zn “memory effect≓ was detected in our system. Undoped InP epilayers maintained an n-type background of ∼5 × 10¹⁵ cm⁻³.     

Light-emitting Si nanostructures formed in SiO<Subscript>2</Subscript> on irradiation with swift heavy ions

SiO₂ layers containing implanted excess Si are irradiated with Xe ions with an energy of 130 MeV and doses of 3 × 10¹²–10¹⁴ cm⁻². In the samples irradiated with a dose of 3 × 10¹² cm⁻², ∼10¹² cm⁻² segregated clusters 3–4 nm in dimension are detected by transmission electron microscopy. With increasing dose, the dimensions and number of these clusters increase. In the photoluminescence spectrum, a 660- to 680-nm band is observed, with the intensity dependent on the dose. After passivation of the sample with hydrogen at 500°C, the band disappears, but a new ∼780-nm band typical of Si nanocrystals becomes evident. On the basis of the entire set of data, it is concluded that the 660- to 680-nm band is associated with imperfect Si nanocrystals grown in the tracks of Xe ions due to high ionization losses. The nonmonotonic dependence of the photoluminescence intensity on the dose is attributed to the difference between the diameters of tracks and the diameters of the displacements’ cascades responsible for defect formation.     

Effect of ion dose and annealing mode on photoluminescence from SiO2 implanted with Si ions

This paper discusses the photoluminescence spectra of 500-nm-thick layers of SiO₂ implanted with Si ions at doses of 1.6×10¹⁶, 4×10¹⁶, and 1.6×10¹⁷ cm⁻² and then annealed in the steady-state region (30 min) and pulsed regime (1 s and 20 ms). Structural changes were monitored by high-resolution electron microscopy and Raman scattering. It was found that when the ion dose was decreased from 4×10¹⁶ cm⁻² to 1.6×10¹⁶ cm⁻², generation of centers that luminesce weakly in the visible ceased. Moreover, subsequent anneals no longer led to the formation of silicon nanocrystallites or centers that luminesce strongly in the infrared. Annealing after heavy ion doses affected the photoluminescence spectrum in the following ways, depending on the anneal temperature: growth (up to ∼700 °C), quenching (at 800–900 °C), and the appearance of a very intense photoluminescence band near 820 nm (at >900 °C). The last stage corresponds to the appearance of Si nanocrystallites. The dose dependence is explained by a loss of stability brought on by segregation of Si from SiO₂ and interactions between the excess Si atoms, which form percolation clusters. At low heating levels, the distinctive features of the anneals originate predominantly with the percolation Si clusters; above ∼700 °C these clusters are converted into amorphous Si-phase nanoprecipitates, which emit no photoluminescence. At temperatures above 900 °C the Si nanocrystallites that form emit in a strong luminescence band because of the quantum-well effect. The difference between the rates of percolation and conversion of the clusters into nanoprecipitates allows the precipitation of Si to be controlled by combinations of these annealings. Fiz. Tekh. Poluprovodn. 32, 1371–1377 (November 1998)     

Dependence of electrical properties of polycrystalline cvd si films on grain size and impurity doping concentrationa,b

The electrical properties of sets of simultaneously grown p-type polycrystalline Si films, deposited by SiH₄ pyrolysis on polycrystalline high-purity alumina substrates and B-doped during growth, were determined by Hall-effect measurements in the temperature range 77-420K as functions both of impurity doping concentration N (~10^l5 to ~10²⁰cm⁻³) and average grain size (≈1 to ≈125μm) in the film. Room temperature data showed rapidly increasing resistivities and rapidly decreasing free-carrier concentrations for doping below a critical concentration N_m and distinct mobility minima at that concentration, with the value of N_m being larger the smaller the average grain size. Measurements as a function of sample temperature showed the intergrain barrier height E_b, decreasing from a maximum value of ~0.4eV at the critical concentration to very small values (~0.01eV) for concentrations above 10¹⁹cm⁻³, with a functional dependence close to E_b ∝l/N^1/2 and E_b for any given concentration being larger the smaller the average grain size. Results are interpreted in terms of the grain-boundary trapping model. Trapped carrier densities in the grain boundaries were calculated to range from ~5×l0¹¹cm⁻² at N≈N_m to ~5×l0¹²cm⁻² for N>10¹⁹cm⁻³, the density being higher the smaller the grain size, and evidence was found for an energy distribution of traps in the Si bandgap, rather than a fixed density at a single discrete energy level. The observed relationship between N_m and average grain size nearly coincides with that of the model for films with ~lμm grain size but sharply departs from it for larger grain sizes, indicating probable applicability of the model for grain sizes up to that range. ^aThis work was supported by the U.S. Department of Energythrough its San Francisco Operations Office under Contract DE-AC03-79ET23045 and monitored by the Solar Energy Research Institute, Golden, CO. bThese results were first described at the 22nd Electronic Materials Conference, Ithaca, NY, June 21–27, 1980, Paper No. M4.