Showing papers in "Solid-state Electronics in 1985"
IBM1
TL;DR: In this paper, the authors investigated the effect of particle diffusion and heat flux on the velocity vs distance curve in MOSFETs and found that diffusion, usually neglected in previous studies, plays a major role and considerably modifies the features of the velocity versus distance curve, leading to an increase of the carrier drift velocity in the low-field region.
Abstract: Electron dynamics in silicon is investigated by means of improved momentum- and energy-balance equations including particle diffusion and heat flux. The resulting system of partial differential equations is numerically solved in a variety of field configurations including strong discontinuities, in order to enhance velocity overshoot effects. It is found that diffusion, usually neglected in previous studies, plays a major role, and considerably modifies the features of the velocity vs distance curve, leading to an increase of the carrier drift velocity in the low-field region, i.e. before experiencing the effect of the strong field. In addition, it is found that, in order to take full advantage of velocity overshoot effects in MOSFET's, a structure must be designed having the strongest possible field at the source-end of the channel, where carrier density is controlled by the gate.
361 citations
TL;DR: In this paper, the Moll and Ross integral relations for the current flow through the base region of a bipolar transistor and for the base transit time were generalized to the case of a heterostructure bipolar transistor with a nonuniform energy gap.
Abstract: The two integral relations by Moll and Ross for the current flow through the base region of a bipolar transistor, and for the base transit time, are generalized to the case of a heterostructure bipolar transistor with a nonuniform energy gap in the base region.
247 citations
TL;DR: In this paper, two Schottky diodes were fabricated by evaporation of nickel on to an n-type CdF2:YF3 semiconductor and the characteristics of the interface states (energy position, density, time constant and capture cross-section) were obtained for the values of the forward bias in the range 0.0 V ≦V ≦ 0.2 V.
Abstract: Two Schottky diodes were fabricated by evaporation of nickel on to an n-type CdF2:YF3 semiconductor. Diode A was prepared on a slightly etched polished surface and diode B on an unpolished strongly etched surface. The current-voltage (I-V), capacitance-voltage (C-V), and conductance-voltage (G-V) characteristics were determined at room temperature. Both diodes showed non-ideal I-V behaviour with ideality factors 1.5 and 2.0, respectively and are thought to have a metal-interface layer-semiconductor configuration. Under forward bias, the admittance showed large frequency dispersion possibly caused by the interface states in thermal equilibrium with the semiconductor. Analysis of the C-V data in terms of Lehovec's model of an interface state continuum required the supposition of two time constants differing by 2 to 3 orders of magnitude. The characteristic parameters of the interface states (energy position, density, time constant and capture cross-section) were obtained for the values of the forward bias in the range 0.0 V ≦ V ≦ 0.2 V. The diode B is found to have the interface state densities about two orders of magnitude higher than the diode A which may be attributed to the different surface treatments. The C-V measurements at 100 kHz also indicated the presence of a deep donor trap about 0.6 eV below the conduction band edge in diode A.
191 citations
TL;DR: In this paper, it was shown that in heavily doped regions, only two independent parameters control the transport and recombination of minority carriers, and the extraction of the minority carrier lifetime, diffusion coefficient, or bandgap narrowing is therefore impossible from DC measurements only.
Abstract: From a fundamental transport formulation it is demonstrated that in heavily doped regions, only two independent parameters control the transport and recombination of minority carriers. The extraction of the minority carrier lifetime, diffusion coefficient, or bandgap narrowing is therefore impossible from DC measurements only. At least one additional AC measurement is necessary. The reported experimental data in the literature are critically revisited. When the published data are interpreted in terms of the original measurements a comprehensive picture appears, with surprising agreement among authors. Modeling of heavily doped regions in devices is shown to be possible, and good predictions of the emitter saturation current are demonstrated.
142 citations
TL;DR: In this paper, the currentvoltage characteristics of photovoltaic solar cells can be described by implicit equations related to a single exponential model (SEM) or to a double exponential model.
Abstract: The current-voltage characteristics of photovoltaic solar cells can be described by implicit equations related to a single exponential model (SEM) or to a double exponential model (DEM). Different sets of computer parameters can be found from numerical procedures for one single experimental characteristic measured under specified illumination and temperature conditions. A criterion of effectiveness is defined to evaluate the performance of a set of computer parameters and helps to derive the best-descriptive set of parameters. These are related to the physical parameters and their variations are established from dark to light concentration. It is concluded that the model can produce a good description of the experimental characteristic in a restricted bias domain only.
113 citations
TL;DR: The state-of-the-art of contact formation on n and p-type cadmium telluride is discussed in this article, where the main surface properties, which influence the contact behavior, and various technologies, which have been used to fabricate ohmic and rectifying contacts on CdTe are discussed.
Abstract: The state-of-the-art of contact formation on n and p-type cadmium telluride is discussed in this paper. We summarize the main surface properties, which influence the contact behavior, and the various technologies, which have been used to fabricate ohmic and rectifying contacts on CdTe. In this way MS and MIS structures, p-n junctions, and heterojunctions have been considered. In all cases we indicate the advantages and disadvantages of the electrical behavior of the devices.
107 citations
TL;DR: In this paper, power drift step recovery diodes for shaping 1 kV voltage steps with nanosecond fronts have been designed and investigated, and a maximum voltage rise rate of 2·10 12 v/s has been estimated.
Abstract: Reverse recovery processes in diodes, leading to a high voltage rise rate, have been considered. A maximum voltage rise rate for silicon diodes as high as 2·10 12 v/s has been estimated. Power drift step recovery diodes for shaping 1 kV voltage steps with nanosecond fronts have been designed and investigated.
106 citations
TL;DR: In this article, the principal theoretical developments in the theory of band tails in heavily doped semiconductors are reviewed and applications are briefly discussed, but no accurate comparison of theory and experiment has yet been possible.
Abstract: The principal theoretical developments in the theory of band tails in heavily doped semiconductors are reviewed. Applications are briefly discussed. It appears that no accurate comparison of theory and experiment has yet been possible.
90 citations
TL;DR: In this paper, the integration of Poisson's equation proceeds from the interior towards the surface, instead of from the surface inwards as is usually the case, so that a considerable saving in time is achieved and no stability problems arise.
Abstract: A simple method of modelling the C–V profiles of high-low junctions and heterojunctions is described in which the integration of Poisson's equation proceeds from the interior towards the surface, instead of from the surface inwards as is usually the case. No iteration is necessary, so that a considerable saving in time is achieved and no stability problems arise. A typical program takes about 30 min to run on a desk-top computer.
74 citations
TL;DR: In this article, a Monte Carlo simulation program for computing time-dependent carrier diffusion effects in a single-photon avalanche diode (SPAD) has been developed, suitable also for small minicomputers (e.g. PDP 11 23 ).
Abstract: A single-photon avalanche diode (SPAD), in which single carrier events can be measured, is used as a probe for accurate measurements of time-dependent carrier diffusion effects in silicon. Basic criteria are discussed for a Monte Carlo simulation program for computing such diffusion effects in various device geometries. A simple program, suitable also for small minicomputers ( e.g. PDP 11 23 ) has been developed. Its accuracy has been checked by comparison with the experimental data obtained from the available SPAD devices, with pulsed illumination at various wavelengths from laser sources. Results are discussed showing that the simulation program can be reliably employed as a tool for analysis and design of new devices.
63 citations
TL;DR: In this article, the authors measured threshold values in oil bths at the temperature range 100-280°C, approaching operation temperature of integrated circuits, and found that threshold does not depend on annealing atmospheres, such as N2, O2 and N2/H2.
Abstract: Electromigration threshold was measured in oil bths at the temperature range 100–280°C, approaching operation temperature of integrated circuits. For uncovered unannealed films, threshold was constant in the temperature range below 200°C, yielding a low threshold value of about 420 A/cm (42 μm × 105 A/cm2) for the product of current density and stripe length. Annealing does not affect temperature behavior, but threshold values grow with increasing annealing temperatures. Threshold does not depend on annealing atmospheres, such as N2, O2 and N2/H2. Different film compositions like Al, Al/Si and Al/Si/1% Cu are equivalent concerning threshold. Elevated, slightly temperature-dependent threshold values were measured using covering SiO2 films. The threshold of covered films depends on test conditions. Reduced current densities and stress times result in increasing values which may be beneficial for electromigration reliability. Moreover, electromigration test results can be influenced by threshold, which has to be considered. The discrepancy to former threshold results is assumed to be due to N2/H2 test ambients, reducing electromigration mass flow. The reason for threshold is suggested to proceed from different bond energies for the transition from grain boundaries to surface.
TL;DR: In this article, the principal parameters for short-channel MOSFETs were derived from the experimental curves ID(VG) and ID(VD) curves. But the main assumption is that the devices fabricated on the same silicon chip have the same technological reduction of transistor channel length and the same series resistance of source and drain.
Abstract: In this paper we present a new method of determining the principal parameters for short-channel MOSFET modelling: VT, μ0, θ, ΔL and RSD. They are deduced from the experimental curves ID(VG) (for small drain voltages) and ID(VD) (for relatively large gate voltages) curves. The main assumption is that the devices fabricated on the same silicon chip have the same technological reduction of transistor channel length and the same series resistance of source and drain. Our method takes advantageously into account (a) a new accurate determination of the threshold voltage and (b) the variation of the low-field mobility with channel length. Results obtained by applying this method to short-channel devices are given and discussed.
TL;DR: In this article, large-grained Al lines of different lengths and widths were interconnected by varying TiN auxiliary layers, and test current densities lay between 4 × 10 5 A/cm 2 and 6 × 10 6 A/ cm 2 at 200°C.
Abstract: Special test structures were used for investigating electromigration mechanisms. Large-grained Al lines of different lengths and widths were interconnected by varying TiN auxiliary layers. Test current densities lay between 4 × 10 5 A/cm 2 and 6 × 10 5 A/cm 2 at 200°C. Considering electromigration threshold, grain boundary electromigration was eliminated and interface electromigration appeared, affecting the conductive Al/TiN interface. Interface electromigration clearly contributes to the mass flow of Al lines, and thus can be detrimental for the reliability of metallization. The interface diffusion activation energy is comparable to the grain boundary activation energy. Contrary to a conductive interface, the technical Al surface does not contribute to mass flow. The elimination of interface effects finally brings out homogeneous bulk electromigration. The drift velocity was directly measured after a stress period of 8300 hours at 200°C. For a current density of 4 × 10 5 A/cm 2 bulk drift velocity was 7 × 10 −12 cm/s, while grain boundary electromigration surpassed this value by a factor of 300. Electromigration threshold was ascertained for grain boundary as well as for interface and bulk diffusion.
TL;DR: The Insulated Gate Transistor (IGT) as mentioned in this paper is a new power semiconductor device with the high input impedance features of the power MOSFET and the ability to operate at high current densities even exceeding that of power bipolar transistors.
Abstract: The Insulated Gate Transistor (IGT) is a new power semiconductor device with the high input impedance features of the power MOSFET and the ability to operate at high current densities even exceeding that of power bipolar transistors. The high temperature operating characteristics of the device are discussed here. Unlike the power MOSFET whose operating current density decreases by over a factor of 2 when the ambient temperature is raised to 150°C, the IGT is found to maintain its high operating current density at elevated temperatures. The temperature coefficient of the output current is found to be positive at forward drops below 1.5 V and negative at forward drops above 1.5 V. These characteristics make the IGT suitable for applications with high ambient temperatures. The results also indicate that these devices can be paralleled without current hogging problems if the forward conduction occurs at forward voltage drops in excess of 1.5 V.
TL;DR: In this article, the authors considered the lifetime of an electron vacancy in the conduction band of a single parabolic band at non-zero temperature and showed that the effect can be observed.
Abstract: In doped and in highly excited semiconductors many-body effects produce a shrinkage of the band gap However, the Coulomb interactions between the conduction band electrons also cause a spreading of the conduction band edge and similar effects occur at the valence band edge This effect is estimated here by considering the lifetime of an electron vacancy in the conduction band Owing to the Coulomb interaction induced electron collisions (“Auger effect”) this lifetime is finite and broaders the state considered The broadening is estimated theoretically for the simple model of a single parabolic band at non-zero temperature, and experimental evidence is presented from related studies showing that the effect can be observed The spreading out of levels considered here corresponds to the imaginary part of the self energy (the real part is normally calculated as the main energy band narrowing) The results imply that under high excitation conditions the imaginary part can be a significant fraction of the real part
TL;DR: In this paper, a double-crystal X-ray diffractometry (DCD) was used to analyze silicon wafers implanted at room temperature and liquid nitrogen temperatures and annealed at 750°C.
Abstract: Silicon wafers implanted at room temperature and liquid nitrogen temperatures and annealed at 750°C are analyzed by transmission electron microscopy (TEM) and double-crystal X-ray diffractometry (DCD). TEM evidences a buried layer of perfect dislocation loops in the room-temperature—implanted sample and a perfect crystal for the case of liquid-nitrogen—temperature implant. DCD reveals a buried strained layer and a thin deformed surface region for both cases. The deep strain distributions are related to residual damage beneath the original amorphous layer, whereas the surface defects are due to an impurity coming from outside.
IBM1
TL;DR: In this article, it was shown that the observed reduction in emitter injection efficiency in bipolar transistors is not strictly due to a gap reduction, as generally believed, but to three very different effects, namely electron-electron and electron-impurity interactions and the effect of disorder in the impurity distribution.
Abstract: The authors review briefly the existing theoretical treatments of the various effects that contribute to the reduction of the energy gap in heavily doped Si, namely electron-electron and electron-impurity interactions and the effect of disorder in the impurity distribution. They then turn to the longstanding question why energy-gap reductions extracted from three different types of experiments have persistently produced values with substantial discrepancies, making it impossible to compare with theoretical values. First, they demonstrate that a meaningful comparison between theory and experiment can indeed be made if theoretical calculations are carried out for actual quantities that experiments measure, e.g. luminescence spectra, as recently done by Selloni and Pantelides. Then, they demonstrate that, independent of any theoretical calculations, the optical absorption spectra are fully consistent with the luminescence spectra and that the discrepancies in the energy-gap reductions extracted from the two sets of spectra are caused entirely by the curve-fitting procedures used in analyzing optical-absorption data. Finally, they show explicitly that, as already believed by many authors, energy-gap reductions extracted from electrical measurements on transistors do not correspond to true gap reductions. They identify two corrections that must be added to the values extracted from the electrical data in order to arrive at the true gap reductions and show that the resulting values are in good overall agreement with luminescence and absorption data. They, therefore, demonstrate that the observed reduction in emitter injection efficiency in bipolar transistors is not strictly due to a gap reduction, as generally believed, but to three very different effects.
TL;DR: In this paper, the Fowler-Nordheim (FN) tunneling of electrons into thermal SiO 2 at high levels of injection has been studied and a modified I-V measurement technique has been developed to correct for charge trapping phenomena, which are unavoidable under the high field and current conditions.
Abstract: The Fowler-Nordheim (FN) tunneling of electrons into thermal SiO 2 at high levels of injection has been studied A modified I–V measurement technique has been developed to correct for charge trapping phenomena, which are unavoidable under the high field and current conditions It is shown that the results fit a universal FN type current-field dependence which is insensitive to oxide thickness and free of the commonly observed hysteresis phenomena
TL;DR: In this paper, a series of measurements in the region of the metal-insulator transition in amorphous Nb x Si 1−x, where the authors measured the tunneling density of states and the electrical conductivity.
Abstract: In this paper we discuss a series of measurements in the region of the metal-insulator transition in amorphous Nb x Si 1− x . We have measured the tunneling density of states and the electrical conductivity. We have observed modifications to the tunneling density of states on an energy scale Δ which goes to zero at the metal insulator transition. Our samples were prepared in a manner which allows control of the Nb concentration, allowing one to step smoothly through the metal-insulator transition.
TL;DR: In this article, the authors proposed a curve-fitted approximation of the light generation rate, which is represented by a series of three to five exponential terms resulting in an analytical solution of the continuity equation.
Abstract: In order to shorten the calculation time for solar cell properties, the light generation rate usually expressed as a sum of individual contributions over the whole solar spectrum is replaced by a curve-fitted approximation. This approximation is represented by a series of three to five exponential terms resulting in an analytical solution of the continuity equation which has the same form as for the actual generation rate. Using the proposed approximation the calculated contribution of the base region to the short-circuit current fits closely the result obtained with the actual generation rate.
TL;DR: In this paper, the authors developed the theory of capacitance transient of a junction in the case where the usual approximation, namely that the defect concentration N T is not negligible compared to the free carrier concentration, is not fulfilled.
Abstract: We have developed the theory of the capacitance transient of a junction in the case where the usual approximation, namely that the defect concentration N T is not negligible compared to the free carrier concentration, is not fulfilled. We show that the correct analysis of this transient by the so-called Deep Level Transient Spectroscopy technique must take into account the fact that, during the transient, the width of the space charge region varies. The validity of the expressions obtained for the shift of the signature (variation of the emission rate vs temperature) and for the concentration N t is verified by comparison with experimental results obtained for electron induced defects in n -GaAs.
TL;DR: In this article, the 1/f noise in MOS transistors has been investigated and is shown to correlate with charge transfer inefficiency experiments on surface-channel CCDs.
Abstract: The 1/f noise in MOS transistors has been investigated and is shown to correlate with charge transfer inefficiency experiments on surface-channel CCDs. Both independent phenomena can be quantitatively explained by the same interface state model. The oxide trap density turns out to vary by more than a factor 10. The 1/f noise is compared with McWhorter's number fluctuation model and with the mobility fluctuation model. The oxide trap density is calculated from the charge transfer inefficiency in surface CCDs. Both the quantitative agreement between oxide trap density and 1/f noise and the observed dependence of 1/f noise on gate voltage here give strong arguments in favour of the McWhorter model. The investigated MOS transistors fall into a category that cannot be explained by the present mobility fluctuation model.
TL;DR: In this paper, the Bethe condition for thermionic emission in Schottky diodes is analyzed in terms of the transport of carriers near the top of an absorbing barrier, under forward bias.
Abstract: The Bethe condition for thermionic emission in Schottky diodes is analysed in terms of the transport of carriers near the top of an absorbing barrier, under forward bias. It is shown that the usually accepted values of the drift velocity vs and of the carrier density n(0) at the barrier maximum are not correct, vs being underestimated by a factor of 2, and n(0) being overestimated by about the same factor. Fortunately, these discrepancies compensate in the product n(0)vs, which yields the correct forward current. It is shown also by means of examples that the Bethe condition is satisfied only over a relatively limited range.
TL;DR: In this paper, the authors used numerical integration to evaluate the enhanced emission rate and found that the field dependence of the emission rate could be approximated by a simple analytic expression, which may be a consequence of interaction between traps, which results in a higher emission rate than predicted by the simple theory.
Abstract: Large electric fields present in polysilicon pn junctions result in field enhancement of the emission and capture rates of traps in the depletion region. We used numerical integration to evaluate the enhanced emission rate and found that the field dependence of the emission rate could be approximated by a simple analytic expression. Calculation of the junction currents in forward and reverse bias were in qualitative agreement with measurements. However, the predicted magnitudes of the currents were considerably lower than those measured at low temperatures. This may be a consequence of interaction between traps, which results in a higher emission rate than predicted by the simple theory.
TL;DR: In this paper, the effect of gate area modulation on charge-pumping current has been analyzed on the basis of the Shockley-Read-Hall theory of trapping, and a model describing the effect for any trapezoidal gate waveform and any reverse biasing source voltage is derived.
Abstract: Interface-trap charge-pumping effect is analysed on the basis of Shockley-Read-Hall theory of trapping, and a model describing the effect for any trapezoidal gate waveform and any reverse biasing source voltage is derived. A simplified version of the model which is valid for identical rise and fall times is also presented and experimentally verified. Experimental results indicate that the spatial variation of surface potential and the modulation of effective gate area by source voltage may strongly influence the charge-pumping current. It is also shown that the effect of gate-area modulation can be characterized directly from charge-pumping measurements.
TL;DR: In this article, the carrier velocity saturation concept was used to calculate the I/V and small-signal ac characteristics of AlGaAs/GaAs modulation-doped FETs.
Abstract: dc and small-signal ac characteristics of AlGaAs/GaAs modulation-doped FET's have been calculated by combining the conventional charge-control model with the carrier velocity saturation concept in realistic submicron gate structures. For the desired threshold voltage and the maximum gate voltage swing, the designer basically selects the thickness of the n -AlGaAs layer, for the preferred doping concentration, from the design graphs provided. Representative theoretical I/V characteristics and saturation current and transconductance versus gate voltage graphs are presented for the 0.25 μm and 0.8 μm gate-length structures with and without the presence of source and drain series resistances.
TL;DR: In this article, an analysis is given of the diffusion length in semiconductors from the induced current profiles obtained by scanning an electron beam with normal incidence on a Schottky diode, assuming that the carrier recombination velocity at the free semiconductor surface is vs = 0.
Abstract: An analysis is given of the determination of bulk diffusion lengths in semiconductors from the induced current profiles that are obtained by scanning an electron beam with normal incidence on a Schottky diode. The discussion assumes that the carrier recombination velocity at the free semiconductor surface is vs = 0. In this case the mixed boundary conditions of the diffusion problem for excess minority carriers can be converted into normal ones by using polar coordinates, and an explicit expression for the induced current profile can be given. This expression is compared to that already known for the opposite case vs = ∞, to establish the influence of the surface recombination velocity on a number of profile properties, such as symmetry, asymptotic decay, or low-order moments of the derivative. It is shown that by evaluating the variance of the profile derivative at two beam energies the diffusion length can be determined independently of the knowledge of the value of vs.
TL;DR: In this paper, the ideal low and high-frequency capacitance-voltage curves of a semiconductor(2)-insulator-semiconductor(1) (SIS) structure were first calculated with the insulator thickness, conductivity types and doping concentrations in semiconductor (1) and (2) as parameters.
Abstract: The ideal low- and high-frequency capacitance-voltage curves of a semiconductor(2)-insulator-semiconductor(1) (SIS) structure were first calculated with the insulator thickness, conductivity types and doping concentrations in semiconductor(1) and semiconductor(2) as parameters. The effects of fixed oxide charge and interface trap charge on the low and high frequency capacitance-voltage curves were also calculated. It was found that the fixed oxide sheet charge density with its centroid and the order estimation of the interface trap charge density with its effective type in addition to the insulator thickness, conductivity types and doping concentrations in semiconductor(1) and semiconductor(2) could be estimated from measured low and high frequency capacitance-voltage curves of an SIS structure.
TL;DR: In this paper, the authors measured the leakage current characteristics of n + p drain diodes of a p -well C-MOS process and compared the measured values with calculated values owing to thermal generation according to the Poole-Frenkel effect and phonon-assisted tunneling.
Abstract: Soft leakage current characteristics of n + p drain diodes of a p -well C-MOS process have been measured and analyzed. A field dependent component of the leakage current could be determined after subtraction of the diffusion current. The values of the field-dependent generation current found were compared with calculated values owing to thermal generation according to the Poole-Frenkel effect and phonon-assisted tunneling. The measured values for different diode structures were found to agree well with the theoretical prediction of phonon-assisted tunneling through a barrier lowered by the presence of an electric field.
TL;DR: In this article, a set of physical constants for In 0.53 Ga 0.47 As as required for transport calculations is obtained by reviewing the literature, and velocities for fields up to 100 kV/cm, calculated by the Monte Carlo method using these constants, are presented for the temperatures of 95 and 300 K.
Abstract: A set of physical constants for In 0.53 Ga 0.47 As as required for transport calculations is obtained by reviewing the literature. Velocities for fields up to 100 kV/cm, calculated by the Monte Carlo method using these constants, are presented for the temperatures of 95 and 300 K. The calculated values are found to be in good agreement with the available experimental results.