Showing papers in "Solid-state Electronics in 1980"

A single-frequency approximation for interface-state density determination

Abstract: Fast interface state densities in the SiSi 2 system can be determined by measurements of the MIS capacitor admittance. Traditional detailed analysis require elaborate frequency dependent techniques. The more commonly used approximation techniques are difficult to interpret for interface state densities less than 1 × 10 11 eV −1 cm −2 . We present here a new single frequency technique as an approximation method which provides quantitative criteria on the quality of such interfaces. The data required are a single high frequency capacitance vs voltage measurement and a corresponding conductance vs voltage measurement. The validity of this technique is best demonstrated in a three-dimensional plot of conductance, frequency and voltage. This also gives added insight into the relationship between the temporal and thermodynamic properties of interface electronic states. Comparison of results using this approximation to more detailed treatments demonstrates the validity of this new method for low surface state density determination in a range from 7 × 10 9 eV −1 cm −2 to 8 × 10 11 eV −1 cm −2 .

Specific contact resistance using a circular transmission line model

Abstract: The measurement of the specific contact resistance of ohmic contacts to semiconductors can be made in a number of ways. One of the methods uses a transmission line model of an ohmic contact on a semiconductor and this paper describes the application of the transmission line model to a contact test pattern of circular symmetry. By using a circular test pattern, the mesa etch step necessary for the standard rectangular test pattern may be omitted, thus simplifying test pattern fabrication.

Determination of existing stress in silicon films on sapphire substrate using Raman spectroscopy

Abstract: Raman spectroscopy is used to determine built-in stresses in silicon on sapphire (SOS) devices. The method is direct, nondestructive and can be applied at various temperatures. For epitaxial silicon films on sapphire substrates a built-in stress of 7.0±0.3 kbar at room temperature and 8.7±0.3 kbar at liquid nitrogen temperature was measured.

Electron transport properties in GaAs at high electric fields

Abstract: Electron distribution function, drift velocity, mean energy, valley population fractions and diffusion coefficient were calculated at high, up to 100 kV/cm, electric fields. Calculations were carried out by the Monte Carlo method. The three-valley model of GaAs conduction band was used. The obtained results were compared with the experimental data.

Admittance spectroscopy: A powerful characterization technique for semiconductor crystals—Application to ZnTe

Abstract: The admittance spectroscopy technique is shown to be a very convenient tool for analyzing majority carrier traps: energy level, capture cross section and concentrations are easily obtained without complicated mathematical treatment. The series resistance in the material underlying the Schottky or pn junction is also detected when the free carriers are freezing out. It allows to get the shallowest level energy and its compensation ratio. The method has been applied to ZnTe material analysis and the effect on the admittance of six different acceptors is demonstrated. When comparing the electrically determined ionization energy of a given impurity with its optical value the former appears as systematically lower but most of the difference can be ascribed to Poole Frenkel or impurity concentration effects.

Effect of titanium, copper and iron on silicon solar cells

Abstract: The effect of Ti, Cu and Fe on silicon solar cells has been investigated. Ti severely degrades cell performance above a concentration of 10 11 cm −3 . The presence of 2 × 10 14 cm −3 Ti results in a 63% loss in cell performance and more than an order of magnitude reduction in carrier lifetime. Ti gives rise to two deep levels in Si at Ev + 0.30 eV and Ec − 0.27 eV. Copper, at concentrations below 10 16 cm −3 , has negligible effect on cell performance and carrier lifetime. Above 10 16 cm −3 copper occasionally produces a 10–15% loss in cell performance with a noticeable increase in junction excess current. No recombination centers were found due to Cu, instead considerable precipitation in the starting material was observed. Fe begins to hurt the cell performance above a concentration of 2 × 10 14 cm −3 . Iron at 1.7 × 10 15 cm −3 results in 46% loss in cell efficiency and about an order of magnitude reduction in lifetime. Fe induces a deep level in silicon at Ev + 0.4 eV. The active center density, for both Ti and Fe, is only a very small fraction of the total impurity content in the starting silicon wafer.

On the physics and modeling of small semiconductor devices—I

Abstract: Current LSI technology has progressed rapidly and is pushing toward fabrication of sub-micron dimensioned devices. Several authors have previously used static characteristics, power dissipation, noise, and packing density to look at limiting properties of small devices, although the actual device physics was not considered in detail. As devices become smaller, we expect that the temporal and spatial scales in these devices become sufficiently small that the semi-classical approach to transport theory, as expressed by the Boltzmann transport equation, becomes of questionable validity. In this paper, we address the question of whether our physical understanding of devices and their operation can be extrapolated to small space and time scales, and to what extent the essential quantum electronics prevents a down-scaling. We attempt to lay here a conceptual framework for an ultimate physics of small devices and the modeling necessary to characterize these devices. In this first paper, we work with a dimensional scale of l ∼ 2500 A , the medium small device, leaving a smaller scale to a subsequent work. Although this scale is marginally in a region where the semi-classical approach is valid, extensive modifications must be made to incorporate several new physical effects, including: intra-collision field effect, retarded spatial and temporal non-local effects, two-dimensional quantization, memory effects in the transport parameters, nonlinear screening/descreening, and multiple scattering effects.

Velocity of surface carriers in inversion layers on silicon

Abstract: Velocity-field curves for surface free-carriers in silicon are determined from measurements on resistivegate IGFETs. The measurements were performed on n-channel devices fabricated on both (100) and (111) substrates and on p-channel devices fabricated on (100) substrates. The channel length of the devices is ∼8 μm and the impurity concentration of the substrates is ∼ 1015 cm−3. The dependence of velocity on the field strength along the channel is found to be well approximated by an empirical relationship involving three parameters: low-field mobility μ0, a critical field E cy signalling the onset of velocity saturation, and a parameter α that determines the curvature between the constant-mobility and constant-velocity branches of the curve. The curve-fitting parameters are given in tabular form for the two n-channel and one p-channel systems studied. The dependence of the velocity-field curves on temperatures in the range 100–350K is also reported.

Model for 1/f; noise in MOS transistors biased in the linear region

Abstract: 1/ f ; Noise calculations and experiments are presented for conductance fluctuations in inversion layers. The layers are biased in the ohmic region at very low drain-source voltages. The model makes use of an experimental fact that competing scattering mechanisms other than lattice scattering lead to a reduction of 1/ f ; noise, but does not consider trapping of charge carriers in surface states as the source of 1/ f ; noise. The free charge carrier distribution and a mobility profile play an important part in the model. The model describes the measured results well. A reduction of the effective mobility with increasing gate voltage is accompanied by a strong reduction of the noise.

Ohmic contacts in GaAs

Abstract: Ohmic contacts of n -type GaAs can be reproducibly made to exhibit specific contact resistivities less than 1 × 10 −6 Ω - cm −2 . To do this requires an understanding of the physics involved, a knowledge of the history of previous treatment of the GaAs wafer surface, and processing techniques which are compatible with precisely controlled donor impurity site location determination. The present paper correlates the electrical effects observed in several significant recent developments with theory and interface chemistry to provide workers in the field with a physical understanding of what is essential for reproducible, effective, and reliable ohmic contacts.

Temperature dependence of threshold current for a quantum-well heterostructure laser

Abstract: The threshold current density, Jth, of a quantum-well laser diode is calculated taking into account the quasi-two-dimensional nature of the heterostructure. The calculated value of Jth(T) for a quantum-well laser diode is found, in agreement with experiment, to be less temperature sensitive than that of a conventional double heterojunction laser. The step-like densities of states and the perturbed (hot) carrier distribution of a quasi-two-dimensional structure are responsible for the weaker temperature dependence. Supporting data on quantum-well AlxGa1-xAsGaAs heterostructure laser diodes grown by MO-CVD are presented showing that in the conventional expression Jth(T) = Jth(0) exp (T/T0), T0 can be as high as ∼ 437°C.

Application of Monte Carlo techniques to hot carrier diffusion noise calculation in unipolar semiconducting components

Abstract: It is shown in this paper that Monte Carlo techniques are available for a complete study of unipolar semiconducting components for both diffusion noise properties and static characteristics. This necessitates taking into account space-charge reaction by integration of Poisson's equation. Two possible methods are proposed and carried out in a unidimensional treatment. Emphasis is given to the various problems which are encountered, especially those concerning the reliability of the solutions. In the first of the two methods (single carrier) the validity of classical electrokinetics equations is investigated, showing the influence of possible relaxation effects in a component. In the second method (multicarrier), it is shown that diffusion noise properties are approachable by observation and spectral analysis of current or voltage fluctuations at the ends of the component.

Electrical, Rutherford backscattering and transmission electron microscopy studies of furnace annealed zinc implanted GaAs

Abstract: Electrical, Rutherford backscattering and transmission electron microscopy measurements have been carried out on GaAs samples implanted with 150 keV, 1.10 15 zinc ions/cm 2 and furnace annealed in the temperature range from room temperature to 900°C. A correlation between three types of measurement technique was established and four distinct annealing stages have been identified. For perfect recrystallization and maximum electrical activation an annealing temperature of 900°C is required. The maximum peak hole concentration was in the range 1–2.10 19 holes/cm 3 .

On the mobility of polycrystalline semiconductors

Abstract: A simple phenomenological model based on tunnel and thermionic emission across grain boundary barrier has been developed. The present model has been applied to polycrystalline silicon assuming that the dangling bonds at the grain boundaries behave as electron traps. The calculations have been carried out in two different ways; one, assuming that the interface states density, N is (cm −2 eV −1 ), is constant across the energy gap; and second with the boundary states, N T (cm −2 ), localized around a very narrow energy range at E T . In the first case no differences in the mobility reduction have been found between n and p type polysilicon, but for the assumption of states localized at an energy E T in the upper half of the gap , the barrier height is larger in n -type than in p -type material and consequently the calculated mobility of n -type polysilicon becomes lower than the p -type mobility. In general, the mobility increases with the dopant concentration approaching the monocrystalline behaviour for very large dopings, in qualitative agreement with other approaches and with available experiments.

On the physics and modeling of small semiconductor devices—II: The very small device

Abstract: In a previous paper, we attempted to lay a conceptual framework for an ultimate physics of small semiconductor devices and concentrated on the medium small device. Here we treat the very small device (VSD), characterized by an effective channel length of 250 A. We demonstrate how such a device could conceivably be fabricated using two side processing of the wafer. In treating the transport, however, it is found that the time and distance scales are such that the Boltzmann transport equation is completely invalidated. Here we develop the appropriate quantum transport equations based upon the density matrix for the entire system, device plus boundaries plus environment. It is found that the boundaries and environment can lead to renormalization of the energy spectrum as well as long range dissipative interactions. Two special cases of the transport equations are treated. If the transport is dominantly stochastic, an exact Langevin equation is found for the various transport parameters. In a second case, a parameterized density matrix is used in analogy to the displaced Maxwellian. In this latter case, a hierarchy of moment equations can be developed to yield, e.g. energy and momentum balance equations.

1/f; noise model for MOSTs biased in nonohmic region

Abstract: Relations are derived for the 1/f; noise current and for the equivalent input noise voltage of a MOST biased in the nonlinear region. The experimentally obtained results are in agreement with the calculations. In addition the value of power spectrum of the noise current is related to that in the ohmic region. In the last region the 1/f; noise appears to be caused by mobility fluctuations of the charge carriers. In the nonohmic region, the noise consists of two contributions: (i) mobility fluctuations and (ii) number fluctuations in the charge carriers due to fluctuations of the effective gate voltage induced by mobility fluctuations.

Metallurgical and electrical properties of chromium silicon interfaces

Abstract: The effect of annealing treatment up to 500°C on CrSi contacts was studied from physico-chemical and electrical view points. The solid-solid reactions between a 1000 A thick Cr layer and a 〈111〉N single crystal of silicon, were studied by the He+ ion backscattering method, X-ray diffraction and transmission electron microscopy. We first observed a growth in the Cr grains and then the nucleation and growth of the disilicide CrSi2. For annealing temperatures greater than 415°C, the growth is linear in time with an activation energy of 1.5±0.1 eV and for lower temperatures it becomes superlinear. The growth mechanism is discussed in terms of growth limiting phenomena. The variations of the electrical parameters (ideality factor n and barrier height 0Bn) as a function of 15 min anneals between 300 and 500°C were correlated to the physico chemical observations. We establish, firstly, an optimal temperature annealing range in order to obtain good Schottky CrSi diodes and, secondly, a low limit of Cr thickness which must be deposited to obtain acceptable Schottky CrSi2  Si diodes after annealing.

Impact ionization by electrons and holes in InP

Abstract: Ionization coefficients of holes in InP exceed those of electrons by a factor of approximately four over the electric field range from 2.5 to 4.5 × 105 V/cm and have been measured in Pt on p-type 〈100〉 orientation InP Schottky diodes. The shape of the baseline for the hole induced multiplication has been fitted by using one adjustable parameter, namely: the mean free path for optical phonon scattering. The model includes effects of quantum mechanical transmission, barrier height lowering and phonon scattering before the barrier maximum. The shape of the baseline for electron multiplication was likewise fitted by one adjustable parameter: the minority carrier diffusion length. Both models produce an excellent fit to low field characteristics that have appreciable curvature. The average phonon scattering mean free path for holes is ∼ 35 A roughly consistent with that required to characterize the hole ionization coefficient. The electron diffusion length in p-type InP was found to be ∼ 18 μ for all diodes. Ionization coefficients were obtained for diodes in the doping range from 5.0 × 1015 to 7.8 × 1015 cm−3. The impact ionization curves were fitted by Baraff-Chwang curves with slightly field dependent average energies for pair production. The threshold energies have been calculated from a consideration of energy and momentum conservation in a theoretical band structure. The deduced mean free paths for phonon scattering and breakdown voltages calculated from the ionization coefficient measurements are in excellent agreement with other independent experiments.

The effect of implant temperature on the electrical characteristics of ion implanted indium phosphide

Abstract: The sample temperature during ion implantation in InP has a pronounced effect on the electrical characteristics of the resulting layers. For the heavy ions, Se (donor) and Cd (acceptor) implant temperatures ≥ 150°C are necessary to minimize n -type residual damage and to achieve maximum activation of the implanted ions. Results for the intermediate mass n -type impurity Si are similar to those for the heavy ions, whereas for the intermediate mass acceptor Mg the implant temperature effects appear to be strongly dose dependent. With the light acceptor ion Be, room temperature implants are as good as or better than those done at elevated temperatures. Results for the light ion C yield n -type layers with very low electrical activation. Higher activation is generally achieved with n -type than with p -type impurities and electron concentrations in excess of 10 19 cm −3 are readily attainable.

High-temperature carrier transport in n-type epitaxial GaAs

Abstract: Resistivity and Hall coefficient have been measured in the temperature range 300-1250 K on samples of n -type epitaxial GaAs from which the substrates have been removed to eliminate substrate conduction at high temperatures. These data are useful for determining energy band parameters as well as for modeling epitaxial growth and other high temperature phenomena. The mobilities of electrons in the Γ 6 c minimum and L 6 c minima and the mobility of holes in the Γ 8 v maxima were calculated for the temperature range 300-970 K. From these calculations the mobility in the L 6 c minima was found to decrease faster with increasing temperature than the mobility in the Γ 6 c minimum. Using the temperature dependence of the mobilities, the Γ 6 c − L 6 c − X 6 c ordering of the conduction band minima, and a calculated intrinsic carrier concentration, a self-consistent model for four-band carrier transport was obtained.

A simple model of the threshold voltage of short and narrow channel MOSFETs

Abstract: Simple but reasonably accurate equations are proposed which describe the behavior of threshold voltage for short and narrow-channel MOSFETs, for low drain-source voltages. It will be shown that good agreement is obtained between the model, experiment and two dimensional calculations, for channel lengths and widths as small as 1 ∼ 2 μm. Moreover, by careful analysis of the model results, some new properties of the threshold voltage of small size devices can be derived.

Determination of interface-state parameters in a MOS capacitor by DLTS

Abstract: A modification of the DLTS (Deep Level Transient Spectroscopy) technique for interface-state measurement is described in which the surface potential is used to determine the energy of the interface states contributing to the emission signal. This technique allows an accurate and unambiguous determination of interface-state energies and cross sections. Expressions are determined for interface-state emission as a function of surface potential. Measurements of interface-state density and majority-carrier cross sections as functions of energy for n- and p-type MOS samples are presented.

Real-space electron transfer by thermionic emission in GaAsAlxGa1−xAs heterostructures: Analytical model for large layer widths

Abstract: Calculations are presented for negative differential resistance (NDR) and switching in layered GaAsAlxGa1−xAs heterostructures with a high electric field parallel to the interface. The mechanism is based on thermionic emission of hot electrons from the GaAs layers into the AlxGa1−xAs layers. An analytical model is obtained in the limit of relatively large layer widths (400 A or wider). The method of moments is employed to solve the Boltzmann equation, assuming a position-dependent electron temperature and Quasi-Fermi level in the AlxGa1−xAs layers, and a position-independent electron temperature and Quasi-Fermi level in the narrower GaAs layer. Thermal conduction of hot electrons from the GaAs layer into the AlxGa1−xAs layers is taken into account. The results of the calculations show that the threshold electric field for the onset of NDR and the peak-to-valley ratio can be controlled to a large extent by adjusting the mobility of the AlxGa1−xAs layer, the layer dimensions, and the potential barrier (Al mole fraction in the AlxGa1−xAs).

Aluminum-silicon ohmic contact on “shallow” n+p junctions☆

Abstract: In this work various technologies for contacts of Al on n+Si have been experimentally investigated, particularly in view of their suitability to very shallow np junctions. Special test-patterns have been used to measure the contact resistivity, while diodes reverse current density has been checked to evaluate the junction leakage induced by the aluminum-silicon interaction during sintering. Best results are obtained by depositing a thin polysilicon layer on the front surface before the doping process and the Al evaporation. In this case both the requirements of low contact resistivity (< 10−4 ohm · cm2) and low junction leakage current are satisfied. Comparison with the conventional Al/Si and AgTi/Si ohmic contacts has been performed.

The impact of molybdenum on silicon and silicon solar cell performance

Abstract: Deep level transient spectroscopy coupled with dark and lighted I–V measurements were used to study the electrical properties of silicon crystals and solar cells purposely contaminated with controlled amounts of molybdenum. Mo severely degrades minority carrier lifetime, and hence solar cell performance, by inducing a recombination center at EV + 0.30 eV. Neither HCl nor POCl3 gettering at temperatures as high as 1100°C and times up to five hours mitigate the effects of Mo. Because the Mo segregation coefficient is small, 4.5 × 10−8, impurity contamination of silicon during crystal growth can be kept below the levels for which electrical properties are affected.

Pd/Ge contacts to n-type GaAs

Abstract: Sintered metal-semiconductor contacts, formed by thin, evaporated layers of Pd and Ge on n -type GaAs, were studied using Auger electron spectroscopy, X-ray diffraction, X-ray photoelectron spectroscopy, secondary ion mass spectroscopy, current-voltage measurements, and capacitance-voltage measurements. Prior to sintering, the as-deposited Pd/Ge/GaAs contacts were rectifying and exhibited a reproducible Schottky barrier energy φ Bn of 0.67±0.02 eV. Auger analysis indicated the initial behavior of the contact structure, upon sintering, to be an interdiffusion and reaction of Pd and Ge on a non-reacting GaAs substrate. Two germanide phases, Pd 2 Ge and PdGe, were identified using X-ray diffraction and Auger analysis. The intervening Ge layer prevented the reaction of Pd with the GaAs substrate at low temperatures. Because of the PdGe reaction, φ Bn increased to approximately 0.85 eV. Sintering at higher temperatures (i.e. between 300 and 400°C) produced additional reactions between Pd and the GaAs substrate. The electrical properties of the contact remained rectifying and φ Bn exhibited little change from the value of 0.85 eV with the interdiffusion of Pd, Ga, and As. Sintering above 400°C resulted in the formation of ohmic contacts. The diffusion of Ge to the GaAs interface was found to correlate with the onset of ohmic behavior. Current conduction in the contact was best described by thermionic-field emission theory, and a specific contact resistance of 3.5 × 10 −4 Ω-cm 2 was obtained after sintering above 550°C, independent of the initial impurity concentration in the substrate. Over the entire range of sintering temperatures (i.e. at or below 600°C), the interaction between the thin-film layers appeared to be governed by diffusion-controlled, solid-phase processes with no evidence of the formation of a liquid phase. As a result, the surface of the contact structure remained smooth and uniform during sintering.

Simulation of semiconductor transport using coupled and decoupled solution techniques

Abstract: Two dimensional mobile carrier transport in a semiconductor is simulated using the finite element method. Two algorithms are compared for computational efficiency and accuracy in modeling an IGFET transistor. With the first, the non-linear partial differential equations are decoupled and solved sequentially using a minor and a major iterative cycle. In the second, Newton's method is used to linearize the equations, and they are solved simultaneously using only a major loop. Results show the decoupled approach to be faster and require less storage than the coupled Newton approach for simulation of device operation where the mobile carrier densities are equal to or less than the fixed charge density. The coupled approach is more efficient for highly non-linear problems where the mobile carrier densities exceed the fixed charge density but requires far more storage (2.5x) and set up time (100x).

On the physics and modeling of small semiconductor devices—III: Transient response in the finite collision-duration regime

Abstract: The role of the finite, non-zero collision duration in high electric fields is examined for its effect on transient and over-shoot response of the carrier velocity and energy. The finite collision duration introduces a temporal retardation effect on the collisional relaxation mechanisms for energy and momentum. As a consequence, the effective temperature also undergoes an overshoot behavior, which leads to a general quickening of the total transient response. Calculations were performed for steady, homogeneous fields utilizing a displaced Maxwellian approach. These calculations were performed for GaAs and Si and have significance for sub-micron devices in these materials. The generally faster response leads to the prospect of improved high frequency properties over what is normally expected.

The Shockley-like equations for the carrier densities and the current flows in materials with a nonuniform composition

Abstract: The basic equations for the current and carrier densities in materials with a nonuniform band structure are reviewed. These equations are then recast into forms containing the effective intrinsic carrier density nie and the intrinsic level Ei. The carrier densities formally attain the standard forms as given by Shockley; however, these forms must be interpreted with care. In particular, nie is not a constant and the intrinsic level Ei does not have the profile of the electrostatic potential energy t-eφ. The current densities are also recast into Shockley-like forms; extra terms arise over and above the conventional results.

Temperature effects in silicon solar cells

Abstract: The temperature variation of the relative spectral response RSR, short circuit current Isc and open circuit voltage Voc is measured and the results are theoretically discussed. The RSR at wavelengths larger than the peak wavelengths always increases with temperature. The observed increase is found to agree with theory when temperature variation of the absorption coefficient of light and of Ln the carrier diffusion length in the base is taken into account properly. The temperature variation of short wavelength RSR depends on S, the surface recombination velocity and L p d , L p is the carrier diffusion length in the diffused layer and d is the junction depth. If S is small and L p d is large, the RSR is practically independent of temperature. On the other hand if S is large and/or L p d is small, the RSR decreases as the temperature increases. A new relation between Ln, d and the peak position of RSR is derived. The observed temperature shift in the peak position agrees well with that predicted by this relation. The temperature increase of Isc and decrease of Voc are different for AM0 and AM1 illuminations. The observed differences can be explained at least qualitatively on the basis of the results obtained in this paper.