Showing papers in "Solid-state Electronics in 1970"

Measurement of the ionization rates in diffused silicon p-n junctions

Abstract: An improved method is presented for calculating the ionization rates αn and αp from charge multiplication measurements on diffused silicon p-n junctions. The main features of this method are: The real impurity profile is approximated by an exponential function whose parameters are calculated from capacitance measurements; the ratio γ = αp/αn as a function of the electric field is calculated from multiplication measurements; the ionization rates are solved from the ionization integral for pure electron injection, taking the influence of the threshold energy into account. Measurements on narrow junctions agree with measurements on wide junctions by assuming a threshold energy of 1.8 eV for electrons, in agreement with the results of M oll and van O verstraeten .(1) The ionization rates differ from those of M oll and van O verstraeten (1) and of L ee , L ogan et al.(2) mainly because these authors neglect the influence of the threshold energy. The electron and hole data satisfy Chynoweth's law α(E) = α ∞ exp (−b/¦E¦), cm −1 with: for electrons α∞ = 7.03 × 105 cm−1 b = 1.231 × 106 V cm−1 for 1.75 × 105 ⩽ E ⩽ 6.0 × 105 V cm−1 for holes α∞ = 1.582 × 106 cm−1b = 2.036 × 106 V cm−1 for 1.75 × 105 ⩽ E ⩽ 4.0 × 105 V cm−1 and α∞ = 6.71 × 105 cm−1b = 1.693 × 106 V cm−1 for 4.0 × 105 ⩽ E ⩽ 6.0 × 105 V cm−1 Breakdown voltages are computed for high voltage p-n and p-i-n diodes. These are in good agreement with experiments, indicating the reliability of the ionization rates.

A quasi-static technique for MOS C-V and surface state measurements

Abstract: A quasi-static technique is discussed for obtaining the ‘low frequency’ thermal equilibrium MOS capacitance-voltage characteristics The method is based on a measurement of the MOS charging current in response to a linear voltage ramp, so that the charging current is directly proportional to the incremental MOS capacitance With this technique, surface potential and the surface state density can be obtained relatively simply and over a large part of the energy gap on a single sample, while also providing a direct test for the presence of gross nonuniformities in MOS structures This method has been used to determine the surface state distribution at the interface of a bias grown steam oxide and 10 ω-cm n -type silicon, and the results are compared with composite measurements using the conductance technique for a similar interface The sensitivity for surface state density measurements is estimated to be of the order of 10 10 states per cm 2 eV near mid-gap for 10 ω-cm silicon and improves with decreasing doping density Some applications and limitations are also briefly discussed

Electron tunneling and contact resistance of metal-silicon contact barriers

Abstract: The contact resistance of Al and Pt on n -type Si over a wide range of doping concentrations (10 18 → 2 × 10 20 cm −3 ) has been measured at both room temperature and liquid nitrogen temperature. These experimental results are compared with theoretical calculations based on a model with electron tunneling through the potential barrier at the interface as the dominant mechanism of current flow. Good agreement is found. It is hoped that this physical model can be used as a guideline in developing ohmic contacts for various semiconductor devices.

Thermal and optical emission and capture rates and cross sections of electrons and holes at imperfection centers in semiconductors from photo and dark junction current and capacitance experiments

Abstract: Photo and dark current and capacitance transient experiments are described which can provide highly accurate and unique data of the electronic properties of impurity centers in semiconductors such as energy level, multiplicity of charge state, thermal and optical emission rates, thermal capture rates, thermal and optical cross sections and the dependences of the rates and cross sections on sample temperature, static electric field and photon energy. These experiments are grouped into four categories in the discussion according to the temperature (below and above freeze-out or negligible and finite thermal rates) and presence or absence of optical excitation at the impurity center. Each category is subdivided into a number of cases according to the method used to set the initial charge state of the impurity center prior to the measurements. Methods of setting the initial state include electrical switching of the bias voltage and optical excitation using either interband or band to impurity monochromatic light. Experimental examples are given for many of the experiment methods discussed.

Reverse current-voltage characteristics of metal-silicide Schottky diodes

Abstract: The soft behavior of reverse biased Schottky barrier diodes has often been difficult to interpret quantitatively. The development of metal-silicide devices with diffused guard rings has made it possible to verify experimentally an advanced theoretical model. Reverse characteristics can now be accurately predicted over wide ranges of current, voltage, barrier height and temperature. The theoretical description accounts for anisotropy of effective masses, scattering by optical phonons, and quantum mechanical reflection and tunneling at the metal-semiconductor interface. These considerations yield practical Richardson constants equal to 112 for electrons and 32 for holes in silicon. Absence of true saturation in the reverse characteristic is caused by an electric field dependence of the effective barrier height. In addition to the usual image-force correction, the barrier height is lowered by a newly recognized effect attributed to an electrostatic dipole layer at the metal-semiconductor interface. Experimental devices have been fabricated using RhSi, ZrSi2, and PtSi contacts, forming barriers in both n- and p-type silicon. The resulting structures have been found to be extremely stable and uniform; furthermore, the metal-semiconductor interface, produced by solid-solid chemical reaction, is believed to be free from intervening layers of oxide and other contaminants. When necessary to eliminate field-enhancement at the electrode periphery, diffused guard rings have been incorporated into the structures. Agreement between experimental data and theory is obtained over nearly five orders of magnitude in reverse bias and eleven orders of magnitude in reverse current density, usually with an rms deviation of less than 10 per cent.

Carrier transport across metal-semiconductor barriers

Abstract: Carrier transport across metal-semiconductor barriers has been studied theoretically and experimentally to give a generalized and quantitative presentation. The thermionic and tunneling processes have been analyzed in terms of accurate quantum transmission coefficients. The effects of image-force lowering, temperature, and two-dimensional statistical variation of impurity concentration have also been incorporated in the theory. Theoretical results give a description of the current transport, due to combined effect of tunneling and thermionic emission over a temperature range from essentially absolute zero to the highest practical temperatures, and over doping densities from 1014 cm−3 to complete degeneracy. An interesting result of the analysis is the existence of a minimum in the saturation current density Js near 1016 cm−3; the current density rises slightly at lower dopings because of enhanced transmission coefficient for thermionic emission and increases drastically at higher dopings because of tunneling. For example for PtSiSi system at 300°K with a barrier height of 0.85 eV, Js is 80 nA/cm2 at 1014 cm−3, reaches a minimum of 60 nA/cm2 at 1016 cm−3, then rapidly increases to 103 A/cm2 at 1020 cm−3. In the high doping range the average saturation current density is considerably increased by the effect of two-dimensional impurity variation. The room-temperature transition doping for breakdown in metal-silicon systems occurs at 8×1017 cm−3; for lower dopings the breakdown is due to avalanche multiplication, and for higher dopings it is due to tunneling of carriers from the metal Fermi level to semiconductor bands. The metal-silicon diodes were fabricated by planar technology with guard-ring structures to eliminate edge effects. Extensive experimental studies, including current-voltage, capacitance-voltage, and photoelectric measurements covering the doping range from 1014 to 1020 cm−3 and the temperature range from 77°K to 373°K, gave good agreement with theoretical predictions.

Intermetallic formation in gold-aluminum systems

Abstract: Intermetallic formation in gold-aluminum systems has been studied at 200°C–460°C using butt-welded diffusion couples. The predominant intermetallic phase was found to be Au5Al2, but all five equilibrium phases of the system were present in the couples after annealing at long times. The kinetics of the process were measured and the activation energy for overall intermetallic formation was found to be 15.9 kcal/mol. The effect of silicon on intermetallic formation was also investigated. It was found that while silicon had little effect on the growth rate of AuAl, Au2Al, Au5Al2, and Au4Al, it substantially increased the amount of purple phase in the couples. An electron microprobe traverse indicated that this purple phase contained 6 wt.% Si. The relative tensile strengths of the intermetallics were determined indirectly by measuring the tensile properties of the diffused couples. It was found that all the intermetallics were stronger than gold or aluminum. However, appreciable voiding occurred in Au5Al2 after long times above 300°C and in AuAl2 after long times above 400°C weakening these phases catastrophically. The voiding was greatly accelerated by intermittent aging.

Theory of tunneling into interface states

Abstract: A theory of electron tunneling from the metal through the insulator and into the semiconductor of a MIS-structure is developed. Using a square well model, the interface states are represented by a distribution of δ-function potentials, and their wave functions calculated. Following the method of Bardeen, the tunneling matrix element and current are found and used with an extension of Shockley-Read recombination theory to determine a tunnelling time constant, τ To , similar to the surface recombination time constant, τ. Small signal solutions are calculated and presented in the form of equivalent circuits. For a single level state the equivalent circuit consists of the thick insulator, non-tunneling MIS-capacitor plus a tunneling current, j 1 , injected from the metal into the midpoint of the recombination RC circuit. j 1 is made up of three terms caused by modulation of the effective barrier height, and by modulation of the metal and trap occupancy functions. For τ To ⪡ τ , the case of an ideal Schottky barrier, the tunneling current is controlled by the recombination time, τ. If τ To > τ , the tunneling limited case, j 1 is controlled by τ To . This case is discussed in detail. The model's current and conductance characteristics are numerically calculated and graphically presented as a function of interface state density and distribution, interface potential, insulator thickness, and a.c. frequency.

Voltage-current characteristics of GaAs J-FET's in the hot electron range☆

Abstract: This paper presents a theoretical analysis of the voltage-current characteristics of GaAs J-FET's at drift velocity saturation, i.e. in the hot electron range. Approximate values for cut-off frequency and maximum frequency of oscillations are derived by lumped network analysis in order to assess the microwave frequency performance in relation to material properties and geometrical design parameters. Characteristics of experimental devices employing the single-gate, epitaxial device configuration are correlated with the theory.

Silicon gate technology

Abstract: This paper describes the technology and characteristics of insulated-gate field-effect transistor integrated circuits using deposited polycrystalline silicon as the gate electrode. After a brief outline of the characteristics of the silicon gate technology, some of the basic properties of the silicon-silicon dioxide-silicon system, the processing steps for the fabrication of silicon-gate devices, and the electrical characteristics of the devices obtained will be reviewed. A comparison between silicon gate technology and standard technology will be carried out, using the 3705, an eight-channel multiplexer switch with decoding logic. Design considerations for silicon gate technology and some design examples will be given.

On the separation of bulk and surface components of lifetime using the pulsed MOS capacitor

Abstract: A significant component of reverse current can occur due to lateral surface depletion effects in pulsed MOS capacitor experiments. This component of surface current is approximately proportional to the space-charge region width and can therefore appear as an apparent degradation of the bulk lifetime. A simple model, incorporating this lateral effect, permits direct separation of surface-controlled from bulk-controlled generation components.

Physical model for burst noise in semiconductor devices

Abstract: A physical model for burst noise in p−n junction devices is presented. It is proposed that burst noise results when the current through a defect is modulated by a change in the charge state of a single recombination-generation center located adjacent to the defect. The burst noise amplitude and pulse widths are related to the basic properties of the recombination-generation center and the defect. The model leads to a simple interpretation of the equivalent circuit for diodes which exhibit this type of noise.

Effects of image force and tunneling on current transport in metal-semiconductor (Schottky barrier) contacts

Abstract: Image force lowering of the potential energy barrier is included in a theoretical calculation of current transport in metal-semiconductor (Schottky barrier) contacts. Thermionic and thermionic-field (tunnel) emission are analyzed in a normalized formulation to yield the current (I) vs. voltage (V) relationship. Quantum-mechanical reflection of carriers near the top of the image force rounded barrier is included in the theory by the use of Kemble's transmission probability which incorporates the one-dimensional WKB-type tunneling approximation into a transmission probability applicable both above and below the top of the barrier. Carrier distributions in the semiconductor and in the metal are described by Maxwell-Boltzmann statistics. For any given combination of three dimensionless input parameters E b kT , kT E 00 , and E 00 E 11 , which correspond to bias, temperature and donor concentration respectively, two dimensionless output parameters I f I m (current) and the diode n value (inverse slope of the semilog I vs. V relationship) are determined. Computer solutions are presented in both graphical and tabular form. The results permit a straightforward calculation of the barrier height and the semiconductor donor concentration from experimental I−V data. In comparison with the predictions of current transport models that neglect image force lowering, the present work shows that inclusion of image force leads to a significant increase in the predicted magnitude of the current density and to minor changes in the magnitude of the diode n value. Corrections to the predictions of models that neglect image force arise primarily from enhanced thermionic emission over the image force lowered barrier rather than from enhanced tunnel emission through the image force narrowed barrier. The Kemble transmission probability may be defined in terms of a characteristic transmission energy, Et, which is useful when thermionic emission dominates the conduction process to the extent that quantum-mechanical tunneling and reflection may be considered as a perturbation on thermionic emission. When this occurs Et can be used to estimate the magnitude of the perturbation.

The equivalent circuit model in solid-state electronics—III: Conduction and displacement currents

Abstract: The equilibrium and nonequilibrium equivalent circuits of a single and a multiple energy level recombination centers were developed in parts I and II. This paper extends the analysis to include the nonequilibrium case arising from a d.c. steady state conduction current in the semiconductor structure. Application is made to minority carrier small signal transport in both homogeneous and heterogeneous semiconductors at low steady state levels ( P « N in n- type sample ) . Samples containing p-n junctions are not included but the effect of static, built-in electric field on the signal propagation is considered. In an extrinsic sample, it is shown that the nonequilibrium equivalent circuit is identical to the equilibrium case for considering carrier trapping, recombination and generation at the defect or impurity centers if the recombination conductances and capacitances are defined in terms of the steady-state carrier concentrations. The conditions on the properties of the imperfection centers and the frequency range in which trapping, recombination or generation event dominates at the centers are discussed in detail. It is shown that the effective small-signal minority carrier lifetime, τp in n-type semiconductor, is a complex variable due to the charge storage or trapping effect at the centers. In addition, τp is positional dependent due to the spatial variation of the steady state carrier concentrations. The small-signal lifetime, τp, is substantially different from the steady state lifetime τpSS which is commonly used in high-frequency, lumped-model device analysis. For strongly extrinsic samples with low concentration of recombination centers, the usual approximation, τ p ⋍ τ pSS ⋍ τ p 0 = 1/c p N TT , is valid. The lump model approximation to a specimen of length W is rigorously derived for minority carrier transport. It is shown that the commonly used stored-charge lump model is valid only at low frequency and zero recombination when the sample is geometrically divided into two or more lumps. For finite recombination, the effective length of each lump is shorter than at zero recombination and the sum of the length of each lump is less than the physical length of the sample.

Theory of low frequency noise in Si MOST's

Abstract: The paper deals with the theory of low frequency noise in MOST's due to traps with a wide spread of relaxation times. Particular attention is paid to the case when the drain voltage is small. Distributions of relaxation times leading to a 1/ƒα noise spectrum are discussed (1 ⩽ α < 2). It is shown that for any given noise spectrum, the effect of traps on noise amplitude depends on their ‘surface state efficiency’, a parameter which is directly accessible to measurement. The validity of this trapping model of noise can therefore be tested using equations which relate noise with other measurable parameters of the device.

Characteristics of aluminum-silicon schottky barrier diode

Abstract: Aluminum n -type silicon Schottky barrier diodes with near-ideal characteristics have recently been developed. In this paper the characteristics of such a Schottky barrier are discussed. The I–V characteristics agree well with the theoretical thermionic emission model. The barrier height is determined from the saturation current, temperature dependence of forward current, and photoemission to be0.69±0.01eV. The switching measurements show no minority carrier storage, as expected. The low-frequency noise is very low and is comparable to the best p-n junction and guard-ring Schottky barrier. These desirable features, coupled with the simple process of the Al- n Si Schottky barrier, make them attractive in a variety of applications.

Quantum mechanical calculation of the carrier distribution and the thickness of the inversion layer of a MOS field-effect transistor

Abstract: The carrier distribution in the inversion layer of a MOS-structure at room temperature was calculated assuming a quantization of the allowed energy levels at the surface and a linear electrostatic potential. It was found that for strong inversion the carrier distribution deviates considerably from the one found by using classical statistics but approaches the classical limit for weak inversion when many electric subbands are occupied. A new definition for the channel thickness was introduced based on the integrated charge in the channel and compared to other definitions. Channel thicknesses so defined range from 30 to 400 A for practical devices depending on surface potential. The integrated charge in the channel deviated little from the one found using classical statistics.

Technique used in Hall effect analysis of ion implanted Si and Ge

Abstract: Hall effect and sheet resistivity measurements are commonly used to evaluate ion implanted layers. We describe the general principles, sources of error, and factors influencing the measurements. Implantation conditions, sample preparation and measurement technique used for van der Pauw geometry are discussed. Considerations important in annealing, layer removal and measurement temperature are also presented.

Calculated efficiencies of practical GaAs and Si solar cells including the effect of built-in electric fields

Abstract: The performance of GaAs solar cells has been calculated as a function of the doping levels, using practical values for the transport parameters Calculations show that surface recombination is a more probable cause than recombination in the junction region of the poor efficiencies obtained in practice Electric fields built into the cell by doping gradations may be used to reduce surface losses and produce an efficiency exceeding 20 per cent for a surface recombination velocity of 10 6 cm/sec This figure allows for the finite resistance of the surface layer, the effect of which is considered in detail for several cases Results for Si cells are also presented These are well in accord with the values obtained in practice For both materials consideration is given to the degradation brought about by particle bombardment

Geometrical magnetoresistance and hall mobility in gunn effect devices

Abstract: Geometrical magnetoresistance is developed as a technique for the determination of the Hall mobility of carriers in bulk effect GaAs microwave diodes at low electric fields. Spurious magnetoresistance effects due to inhomogeneities are analyzed and are shown to be negligible to first order for conductivity gradients oriented along the direction of the impressed electric field. For epitaxially grown GaAs, this is the predominant type of inhomogeneity and is due to diffusion of impurities out of the substrate or to a change in the doping during the growing process. The effect of contact resistance on the measurements of mobility is discussed as are a number of other practical aspects of the method, including the effect of misorientation of the magnetic field, the errors caused by finite aspect ratios of the diodes, heating by the measuring current, and the effect of magnetic shielding by parts of some commonly used microwave packages. The relationship between the Hall mobility and the magnetoresistance mobility is determined experimentally for GaAs at room temperature. The ratio is found to beμm/μH = 1.03±0.07.

Electrode effects and bistable switching of amorphous Nb2O5 diodes

Abstract: Nb-Nb 2 O 5 -metal diodes, after breakdown, exhibit switching between two stable conduction states and are potentially usable in an electrically alterable memory. Detailed procedures for oxide formation and electrical breakdown to produce Nb-Nb 2 O 5 -Bi diodes with good switching characteristics are described. The ratio of conductivities in the two stable states varies between 30 and 100. Switching from low to high conductivity can be done with 0.5–1.0 μsec pulses; switching from a high- to low-conductivity state requires 1.0–10 μsec pulses. Switching voltages are less than 1 V, switching powers are less than 100 μW, and the direction of switching is controlled by the polarity of the applied voltage. The dependence of device stability on temperature, storage, and bias are reported. When low-work function metals such as In, Cd, Zn, or Pb are used as counter-elelctrodes, a voltage-controlled negative resistance inflection occurs in the high-conductivity state. Diodes switch at 4.2°K. Switching appears to be an electronic phenomenon, occurring at a small number of regions in the oxide.

Surface state related noise in MOS transistors

Abstract: The theory of surface state effects on 1 ƒ noise in p−n junctions is extended to explain low-frequency excess noise in MOS transistors. Experimental results are critically compared to theory. The results show that low-frequency excess noise in MOS transistors is due to the fluctuation of charge density in the conduction channel caused by the surface potential fluctuation. The fluctuation of the surface potential is introduced by the random charge occupancy of surface states. The low-frequency excess noise in MOS transistors is found to be proportional to the surface state density and the square of the transconductance of the device, and inversely proportional to the gate area and the square of the unit area gate-insulator capacitance. It is also shown that the surface state density when the surface is strongly inverted can be obtained from noise measurements. Finally it is shown that by proper heat treatment it is possible to reduce the low-frequency excess noise of MOS transistors.

Titanium-silicon Schottky barrier diodes

Abstract: Schottky barriers have been formed by vacuum evaporation of titanium onto chemically cleaned n- and p-type silicon. The barrier heights of the contacts were found to 0.50 and 0.61 eV for n- and p-type barriers, respectively. The barrier heights were determined from measurements of the 1 C 2 vs . V characteristics, the reverse saturation current density, and the activation energy of the reverse current. The effects of fabrication technique on diode properties are discussed. It is found that p-type diodes can be fabricated using standard oxide passivation techniques, without severe degradation of most diode properties; n-type diodes are severely degraded by the presence of silicon dioxide at the periphery of the diode, but this problem can be completely eliminated by the use of a diffused p-type ‘guard-ring’. Noise measurements are also presented for the p-type oxide-passivated diodes and for the n-type ‘guard-ring’ diodes; these diodes are found to have essentially ideal noise behavior above a few kHz.

Surface state related 1f noise in p-n junctions

Abstract: It has been shown experimentally by Hsu , Fitzgerald and Grove that the 1 f noise in p-n junctions can be increased by a particular type of surface state introduced by the application of an electric field across the oxide at elevated temperatures This paper extends their work both theoretically and experimentally, leading to an understanding of the mechanism by which surface states affect p-n junction noise The results show that for gate-controlled diodes the noise power is proportional to the product of the density of these surface states and the square of the transconductance, defined as the incremental change in forward current with gate bias For non-gate-controlled diodes an equivalent transconductance can be defined

Properties of Si diodes prepared by alloying Al into n-type Si with heat pulses from a Nd:YAG laser

Abstract: p − n junctions have been prepared by locally alloying a vacuum-evaporated Al film into n -type Si with a focused heat pulse derived form a Nd:YAG laser. The preparation and the gross features of such junctions are described and electrical measurements of the forward bias and reverse bias current, reverse bias capacitance and transient response are given for room temperature. The forward and reverse bias current measurement show that charge recombination and generation within the depletion region controls the flow of current as would be expected for a Si junction. The reverse bias capacitance shows that the change in the doping profile in these junctions is abrupt. From the transient response the reverse recovery time is typically 0.5 × 10 −9 sec.

Noise sources in transport equations associated with ambipolar diffusion and Shockley-Read recombination

Abstract: A discussion is presented of the noise sources which enter into the stochastic transport equations which govern carrier flow in SCL diodes and ordinary junction devices. The essential sources associated with diffusion (or microscopically intraband scattering) and generation-recombination (or microscopically interband transitions) are presented. The connection with thermal noise is discussed. The paper is further devoted to the manifestation of these sources in diffusion, ambipolar flow, band-band generation and recombination and Shockley-Read-Hall recombination. Both low and high injection are treated.

Selfheating and thermal runaway phenomena in semiconductor devices

Abstract: A unitary treatment of the stationary and transient selfheating of semiconductor devices is presented which allows an analytic approach to various problems related to the thermal runaway phenomenon.

The prediction of tunnel diode voltage-current characteristics

Abstract: This paper presents a simplified method of predicting the shape and magnitude of tunnel diode voltage-current characteristics when the substrate material and doping concentrations are specified. The method consists of first describing the tunnel diode V-I characteristics by an algebraic equation, the variables of which are then expressed as functions of the impurity densities. Previous theory is used to express the current variables, however the peak voltage expression is obtained by using Fermi-Dirac statistics and aligning the maximum of the density of electronic states with that of hole states. This leads to a verification and extension of Kane's predictions relating the peak voltage to the doping degeneracies. A comparison of the theory with the actual measured characteristic yields excellent agreement.

Heterojunction solar cell calculations

Abstract: Collection efficiency and spectral response calculations for semiconductor heterojunction solar cells

X-ray measurement of elastic strain and annealing in semiconductors

Abstract: Elastic strain in single crystal slices of Si and GaAs has been measured by use of an X-ray double crystal spectrometer. Strain as large as 1 × 10ȡ4 was introduced by lapping one face of a slice, and was measured by determining the resulting radius of curvature of the other, polished face. The residual strain after annealing at various temperatures was also measured on the polished face. Slices lapped on one face showed a continuous decrease in radius of curvature from very large radii, for large thickness, to less than 4 m for a slice of 0.22 mm thickness. The strain was slowly removed by successive anneals at increasing temperatures. At 1000°C a large reduction of strain was observed for both materials. In GaAs this was accompanied by the generation of a high density of defects which appeared on the polished face. The measurement technique makes use of a double crystal spectrometer in the unusual 'double beam' mode, where both theKα1 andKα2 lines are resolved, and passed through the spectrometer.