Showing papers in "Solid-state Electronics in 1981"

Carrier mobilities in silicon semi-empirically related to temperature, doping and injection level

Abstract: From a review of different publications on the carrier mobilities in silicon, the authors propose an approximated calculation procedure which permits a quick and accurate evaluation of these mobilities over a large range of temperatures, doping concentrations and injection-levels. The proposed relations are well adapted to semiconductor device simulation becuase they allow short computation times.

A new dielectric isolation method using porous silicon

Abstract: A new dielectric isolation technology is proposed. In the new structure, single crystalline Si islands are separated from the silicon substrate by oxidized porous silicon. It is based on the following characteristics of the porous silicon oxide formation: (1) p -type Si is more easily changed to porous silicon than n -type Si; (2) porous silicon is formed along the anodic reaction current flow line; (3) the change in volume of porous silicon after oxidation is relatively small; (4) thick porous silicon films (10 μm) can be obtained easily. In this method, a p -type isolated layer is obtained by proton implantation used for an n -type layer formation. Lateral p - n junctions fabricated in such isolated silicon layers show lower leakage current than those reported in SOS technology.

Tunneling hot electron transfer amplifiers (theta): Amplifiers operating up to the infrared

Abstract: A family of novel three-terminal devices which relies on the transfer of a quasi-monoenergetic hot electron beam through a thin base is described. The devices are similar in principle to the proposed tunneling amplifier by Mead in the early sixties (“Cold Cathode” or “Metal Base” amplifiers). Results are reviewed and the probable reasons for the poor performances are pointed out. It is predicted that, with a proper choice of parameters, metal-base amplifiers can operate as switches, negative resistance devices and continuous amplifiers in the subpicosecond range. Two subclasses are described: The tunneling emitter (THETA), in the major part of the work, and the nontunneling emitter (BHETA) amplifiers. In the THETA family the metal-oxide-metal-oxide-metal (MOMOM), the MOM-semiconductor (MOMS), and the heterojunctions devices are described. Members of the BHETA family generate quasi-monoenergetic electron beams by injecting electrons by an n + n − or a metal- n − junctions, and include a variety of metals and semiconductor combinations. Very thin films are required in these devices (oxides ∼15 A, metals ∼100 A, semiconductors ∼100 A). The molecular beam epitaxy technique and lattice matching considerations are required for pinhole free semiconductors and metal films with minimum interface states. Sputter-oxidation methods are needed for thin oxide growth. Systems which combine these features with availability of microfabrication make these devices feasible today.

Operating limits of Al-alloyed high-low junctions for BSF solar cells

Abstract: Experimental estimations of the effective surface recombination velocity of the high-low junction (Seff) and of the base diffusion length are carried out for Al-alloyed n+pp+ bifacial cells and the results are presented in form of histograms. These results agree with calculated values of Seff when the characteristics of the recrystallized Si layer and heavy doping effects are taken into account. It is concluded that thick Al layers and high alloying temperatures (over 800°C) are necessary to obtain low values of Seff. This conclusion agrees with experimental results of other authors. Recomendations to avoid diffusion length degradation are given and the operating limits of the Al alloying technology are discussed.

Low resistance ohmic contacts to n- and p-InP

Abstract: The contact properties of various metal combinations, deposited by vacuum evaporation on InP, were studied. Among these metal combinations, Au/Ge + Ni and Au/Zn proved to be most suitable. The former on n-InP (n = 8 × 1017/cm3) and the latter on p-InP (p = 9 × 1017/cm3) exhibited specific contact resistances as low as 1.2 × 10−6 and 1.1 × 10−4 Ωcm2, respectively. The specific contact resistances were analyzed using a four-point method which also accounts for the spreading resistance. Furthermore, the resistances of metal contacts to InP were calculated as a function of doping concentration and were compared with the experimental results. The described contacting technique was successfully applied to the preparation of quaternary lasers.

Activation energies for the different electromigration mechanisms in aluminum

Abstract: The activation energies concerning all electromigration contributions in aluminum are theoretically determined starting from the activation energy for self-diffusion. For grain boundary electromigration the activation energy was found to be Eg = 0.4–0.5 ± 0.04 eV, whereas all other contributions except the surface electromigration are described by higher activation energies. For comparing with experimental results the grain boundary activation energy was determined by direct measurement of aluminum drift velocities. It is shown that the known measuring method is valid only in combination with some modifications. The measured activation energies agree well with the estimated ones.

Silicon photovoltaic cells

Abstract: The principles governing the performance of silicon solar cells are reviewed with emphasis on clarifying the essential concepts. Principal attention is devoted to the planar p−n junction cell and recent developments that have contributed to increased efficiency. Other solar cell structures are also reviewed and their relative advantages and shortcomings are discussed. Important areas where our present understanding is inadequate are pointed out and some common misconceptions are corrected.

Transient distortion and nth order filtering in deep level transient spectroscopy (DnLTS)

Abstract: Problems associated with intrinsic sensitivity and time constant selectivity of capacitive DLTS systems are discussed as well as intrinsic limitations in predominantly deep level doped semiconductors. A previously published figure of merit for the optimized exponential correlator for capacitive DLTS studies is shown to be too high because of improper consideration of effects of level restoration: when correctly compared, a filter with a 12% lower figure of merit can be constructed based purely on gating and a weighted phase inversion before the integrator and a phase sensitive detector has a 20% lower figure of merit. The exponential DLTS correlator is also inadequate for analysis of continuous spectra because of its slow drop off in response (∝ TS for TS shorter than the peak response time constant and ∝ TS−1 for longer times). Blanking is necessary to achieve more selectivity relative to short time constants. When performed on-line, D2LTS gives a response ∝ TS−2 for longer times. Still more selective filters of order n, or DnLTS, are considered based on weighted averages over time intervals in geometric progression that are suitable for DLTS and a system with ±1 weighting and suitably chosen time intervals for use with DDLTS. For these filters there is no penalty in figure of merit associated with choice of DDLTS which also appears to be easier to achieve than DLTS. On-line filters with long time constant responses ∝ TS−3 or higher order are shown to exact a large penalty in figure of merit. Equivalent filters can, however, be synthesized with much better figure of merit by a software compensation of multichannel data. The channels are then selected so that the responses of successive channels peak a factor of 2 away in TS and the number of pulses used is decreased by a factor of two. Relative to a multipoint averager, the software compensated analyzer requires a factor of 100 less in measurement time for comparable accuracy when a spectrum must cover a range of 1000:1 in relaxation times. There are also comparable improvements in the holding time specifications and the number of A/D conversions if the system is to be coupled to a computer. When the deep level concentration is comparable to the shallow doping concentration, the peak responses of both DLTS and DDLTS are broadened, the sensitivity increases, the response peaks are shifted to larger relaxation times, and the system becomes essentially nonlinear in response to the individual deep level constituents. This distortion can, in principle, be corrected by going to a constant capacitance mode of operation both during the initial driving pulse and during the recovery transient.

Transport equations for the analysis of heavily doped semiconductor devices

Abstract: Transport equations for use in analyzing heavily doped semiconductor devices are considered. These transport equations describe the effects of the nonuniform band structure and the influence of Fermi-Dirac statistics, which are important in heavily doped semiconductors. Previous workers [1, 2] have derived transport equations in terms of the nonuniform band structure. These equations, however, are not convenient for use in semiconductor device analysis because the band structure of heavily doped semiconductors is not well known. In this paper, the transport equations of Marshak and van Vliet [1, 3] are recast into a simple, Boltzmann-like form in which the effects associated with the nonuniform band structure and degenerate carrier concentrations are described by two parameters, the effective gap shrinkage, Δ G , and the effective asymmetry factor, γ. The experimental determination of both of these parameters is also discussed. Finally, Adler's contention [4], that some important features of semiconductor device operation can be modeled accurately by using an electrically measured Δ G with an arbitrarily chosen γ, is considered. The validity of this procedure, under certain simplifying assumptions, is established.

Near ballistic electron transport in GaAs devices at 77°K

Abstract: The effect of collisions on near ballistic electron transport in short GaAs terminal devices at 77°K is analyzed in the frame of the model based on the equations of momentum and energy balance where momentum and energy relaxation times are fit to agree with the results of Monte Carlo calculation for static constant electric fields. Solving the equations by the iteration technique yields the criterion of the ballistic transport (L(μm) ⪡ 0.44 μ (m2/V.s) V12(v) for GaAs at low voltages). The computer solution is used to obtain current-voltage characteristics, field, energy, velocity and voltage distributions for GaAs devices at 77°K. The results of the calculation show that the ballistic effects are dominant at 77°K even at relatively high doping levels (such as 1016 cm−3) for short devices (∼ 0.2 μm long). These effects lead to a higher electron velocity at low voltages and could be utilized in building high speed GaAs integrated circuits.

Stress-sensitive properties of silicon-gate MOS devices

Abstract: Stress-sensitive properties were measured on both p- and n-channel silicongate MOS devices fabricated on (100) Si at room temperature. Stress-induced variations in drain currents for both enhancement- and depletion-mode MOS transistors with various channel-dopings were measured over a wide range of gate biases. In addition to piezoresistance effect, remarkable drain-current variations were observed at weak-inversion and explained theoretically in terms of changes in minority carrier densities due to energy band shifts by stresses. Elastoresistance shear-constants for polycrystalline-silicon gate layers were also obtained and compared with coefficients for source-drain diffused layers. Further, the elastoresistance of p-type polycrystalline-silicon films was investigated on doping-concentration dependences. A theoretical model for polycrystalline-silicon elastoresistance was developed based on the barrier model for conductivity in polycrystalline-silicon. Results obtained from the model were compared with the experimental results and found to be in good agreement at higher doping-concentrations than trap density.

Theoretical analysis of hall factor and hall mobility in p-type silicon☆

Abstract: Using a two-band (i.e. heavy-hole and light-hole band) model and the relaxation time approximation, the Hall factor was calculated for the case of silicon doped with boron. Contributions from scattering by acoustical and optical phonons and by ionized and neutral impurities were considered. In addition, the effects of hole-hole scattering, as well as valence band nonparabolicity and anistropy were also taken into account. The scattering and anistropy factors were separately evaluated to emphasize their individual contributions to the Hall factor. Theoretical values of the Hall factor at 300 K vary between 0.882 and 0.714 over the dopant density ranges 1014 ≦ NA ≦ 3 × 1018 cm−3. Hall mobilities for p-type silicon were calculated and compared with published data 100 ≤ T ≤ 400 K and 1014 ≦ NA ≦ 3 × 1018 cm−3. The present model is limited to the case of uncompensated material and for the weak field in which μHB ⪡ 1.

On the theory of Debye averaging in the C-V profiling of semiconductors

Abstract: We discuss the nature of the Debye averaging that takes place in the C-V profiling of highly non-uniform electron distributions in semiconductors. It is shown that the averaging process preserves the moment of the electron distribution. This property can be used to extract certain quantitative details about the electron distribution that appear to have been obliterated by the averaging, without a need to reconstruct the entire true electron distribution. As an example, the extraction of the steepness of a diffusion gradient is discussed. In the limit of a weak electron concentration gradient, the width of the Debye averaging is shown to be such that the RMS averaging distance is (2)L D , where L D is the Debye length.

The diffusion of silicon in germanium

Abstract: The diffusion coefficients of silicon in germanium as a function of temperature have been measured for the first time. The measurements were performed for lightly doped n-type and p-type germanium; because of the very slight difference observed in diffusion between these cases, the values 2.9 eV for the activation energy and 0.24 cm2/s for the frequency factor can be considered to correspond to those of intrinsic germanium. The 30Si tracer element was implanted into the germanium samples and was detected using the (p, γ) resonance broadening method. The annealing temperatures ranged from 650 to 900°C.

An analytical approximation for the Fermi-Dirac integral F32(η)

Abstract: A simple analytical expression for the Fermi-Dirac integral of 3 2 and 1 2 orders is proposed. This expression is valid for − ∞ (j = 3 2 ) and 0.53% (j = 1 2 ) .

The doped Si/SiO2 interface

Abstract: A study is reported of the influence of dopant atoms on the SiSiO2 interface states of thermally oxidized silicon. It was found that acceptor or donor atoms induce interface states and oxide charges. The effect is largest in the case of acceptor dopants and is independent of the doping process. The influence of the dopant atoms on oxide charge is probably related to the different segregation coefficients of acceptors and donors.

Silicon nitride for the improvement of silicon inversion layer solar cells

Abstract: It is demonstrated that CVD Si-nitride films as transparent dielectric satisfy all requirements for achieving MIS/IL solar cells with high efficiency and long term stability. Deposited on silicon by the SiH4/NH3 reaction at temperatures lower than usual (between 600° and 650°C) fixed positive interface charge densities QN/q up to 7 × 1012 cm−2 with excellent stability have been obtained. Utilizing the Si-nitride charge storage effect, the highest known QN/q values (> 1013cm−2) combined with low values of Nit have been achieved. The charge distribution is discussed and an energy band diagram modified according to new analytical results is presented. MIS/IL solar cells with AM1 efficiencies of 15% (active area) and high UV sensitivity have been obtained.

Electromigration measuring techniques for grain boundary diffusion activation energy in aluminum

Abstract: The activation energy E g for electromigration along aluminum grain boundaries has been examined intensively by many authors and different methods. However, the common techniques yield very contradictory results, with E g ranging between 0.3 and 1.2 eV. For clarifying these contradictions, the present work examines the validity of the different test methods and their basic assumptions. It is shown that for most of these techniques the test quantity being measured is given by time dependent divergences which cause local stripe variations from scarcely detectable voids up to stripe interruptions. This means that constant conditions are not maintained during the test period, thus yielding more or less incorrect activation energies. A well defined measurement of the mass transport and the activation energy is practicable only if the divergence of the electromigration is both constant for a fixed test condition and proportional to the flux for different test conditions. Additionally, all other diffusion terms except electromigration must be negligible, especially the diffusion term based on pressure gradients. These preconditions are satisfied for the direct measurement of the drift velocity of an aluminum front on a TiN stripe, proper test conditions provided. That is why this technique is applied here extensively for clarifying several contradictions. Other techniques were performed for comparison yielding different results. The here obtained valid values of E g and their variation between 0.43 and 0.47 eV are smaller than generally reported, even for extremely different evaporation and sputtering conditions. In contrast to former results obtained by improper test methods, the activation energy is temperature independent in the examined temperature range between 140 and 280°C, and E g is not dependent on grain size even for most different fabrication conditions, yielding grain sizes from 0.3 up to 3 μm.

High-low junctions for solar-cell applications

Abstract: A new theoretical model to calculate the effective surface recombination velocity ($a) of a high-low junction with an arbitrary impurity distribution is presented. The model is applied to erfc-diffused pp+ junctions using experimental data of bandgap narrowing, lifetime and mobility. Bandgap narrowing is shown to degrade the minority carrier reflecting properties of the high-low junction. Computer results are applied for the design of BSF solar cells and to study other solar cell structures based on high-low junctions. NOTATION diffusivity of minority carriers diffusivity of electrons in a p+ region minority carrier current electron current in a p+ region

A comparison of Pd Schottky contacts on InP, GaAs and Si

Abstract: The Schottky-barrier energy φB of Pd contacts on InP, GaAs and Si were measured and the metallurgical behavior of the contact structures were studied using Auger electron spectroscopy. A carefully processed set of samples were used to show conclusively that φB is greater on p-InP that on n-InP, unlike the behavior of GaAs and Si Schottky diodes fabricated at the same time with similar processing steps.

Observation of electron traps in electrochemically deposited CdTe films

Abstract: The I–V and C-V data of Schottky devices formed on electrodeposited n-CdTe films are interpreted to determine the principle trap energy and density. The observed trap is an electron trap located at 0.55 eV below the conduction band with a density of ∼7 × 1015/cm3. This correlates well with the values reported for CdTe prepared by different methods. Nickel is found to be an injecting contact to electrodeposited CdTe films. Au/n-CdTe barrier height is determined to be 0.75–0.85 eV for Schottky devices.

Measurement of Richardson constant of GaAs Schottky barriers

Abstract: The Richardson constants for near-ideal AuGaAs and AlGaAs Schottky diodes have been determined from an analysis of forward current-voltage characteristics. Measurement of capacitance-voltage characteristics at different temperatures shows that the barrier heights of the diodes have very similar temperature dependence for both Au and Al contacts. On taking into account the temperature dependence of the barrier height, the corrected value of the Richardson constant for AuGaAs diode is found to be very close to the predicted theoretical value. The corresponding value for AlGaAs diode is about a factor of five smaller which is explained on the basis of a thin interfacial layer between Al and GaAs.

Depletion layer effects in the open-circuit- voltage-decay lifetime measurement

Abstract: There is a renewal of interest in open circuit voltage decay as a technique for determining the base region minority carrier lifetime in semiconductor diodes. Although the existing theory of open circuit voltage decay provides a substantial foundation for interpreting the experimental data, major features of the decay curves of real silicon diodes cannot be satisfactorily explained unless depletion layer effects are taken into account. Theoretical decay curves are calculated to show the effects of depletion layer capacitance and recombination current on the otherwise ideal open circuit voltage decay. From these and from experimental decay curves, it is shown that each of these depletion layer effects is significant only below a threshold junction voltage that depends upon material parameters of the device.

Resistivity changes of heavily-boron-doped CVD-prepared polycrystalline silicon caused by thermal annealing

Abstract: Heavily-boron-doped polycrystalline Si films were deposited at 600°C on thermally grown SiO2 by the thermal decomposition of SiH4-BCl3-H2 mixture. Resistivity changes with isochronal or sequential annealing were systematically examined. Temperature dependence of equilibrium saturation carrier concentration was determined at 800 ∼ 1100°C. Since as-deposited polycrystalline Si is in the super-saturated state, carrier concentration decreases from the super-saturated to equilibrium saturation value by annealings over 700°C for poly Si doped with over 2 × 1020 cm−3 resulting in anomalous resistivity change. Carrier concentration changes reversibly between saturation values with sequential annealing and is determined by the last annealing temperature when the annealing time is long enough. Mobility increases with annealing temperature, however, less increase is found for heavily doped poly Si, which is attributed to the suppression of grain growth caused by electrically inactive Si-B compounds.

Growth and characterization of large diameter undoped semi-insulating GaAs for direct ion implanted FET technology

Abstract: The growth of large diameter, semi-insulating GaAs crystals of improved purity by Liquid Encapsulated Czochralski (LEC) pulling from pyrolytic boron nitride (PBN) crucibles and characterization of this material for direct ion implantation technology, is described. Three-inch diameter, 〈100〉-oriented GaAs crystals have been grown in a high pressure Melbourn crystal puller using 3 kg starting charges synthesized in-situ from 6/9s purity elemental gallium and arsenic. Undoped and Cr-doped LEC GaAs crystals pulled from PBN crucibles exhibit bulk resistivities in the 10 7 and 10 8 Ω cm range, respectively. High sensitivity secondary ion mass spectrometry (SIMS) demonstrates that GaAs crystals grown from PBN crucibles contain residual silicon concentrations in the mid 10 14 cm −3 range, compared to concentrations up to the 10 16 cm −3 range for growths in fused silica containers. The residual chromium content in undoped LEC grown GaAs crystals is below the SIMS detection limit for Cr (4 × 10 14 cm −3 ). The achievement of direct ion implanted channel layers of near-theoretical mobilities is further evidence of the improved purity of undoped, semi-insulating GaAs prepared by LEC/PBN crucible techniques. Direct implant FET channels with (1–1.5) × 10 17 cm −3 peak donor concentrations exhibit channel mobilities of 4,800–5,000 cm 2 /V sec in undoped, semi-insulating GaAs substrates, compared with mobilities ranging from 3,700 to 4,500 cm 2 /V sec for various Cr-doped GaAs substrates. The concentration of compensating acceptor impurities in semi-insulating GaAs/PBN substrates is estimated to be 1 × 10 16 cm −3 or less, and permits the implantation of 2 × 10 16 cm −3 channels which exhibit mobilities of 5,700 and 12,000 cm 2 /V sec at 298K and 77K, respectively. Discrete power FET's which exhibit 0.7 watts/mm output and 8 dB associated gain at 8 GHz have been fabricated using these directly implanted semi-insulating GaAs substrates.

Effect of emitter recombinations on the open circuit voltage decay of a junction diode

Abstract: It is shown that when the effect of the carriers stored in the emitter of a p - n junction diode is important due to the bandgap narrowing in the emitter, the continuity differential equations in the emitter and the base defining the Open Circuit Voltage Decay (OCVD) form a coupled system. The solutions of the coupled equations are derived and shapes of the minority-carrier profiles are calculated and discussed. It is found that the quasistatic approximation is valid to a high degree of accuracy in the emitter except for very small values of time. However, this approximation is not valid even roughly in the base. The OCVD plots are approximately parallel straight lines for t ⪢ τ B ( τ B is the lifetime of the minority-carriers in the base) and the slope of the plots is ∼1/ τ B . A method is suggested to obtain more accurate values of τ B from the observed OCVD. For small values of time (0 t τ B ), the OCVD plot is sharply curved. By fitting the observed OCVD with the theory in this range, the values of bandgap narrowing and emitter lifetime can also be derived. If ( τ B / τ E ) > 20, ( τ E is the lifetime of minority-carriers in the emitter) the value of τ E derived in this manner is however not accurate. A typical experimental plot is found to be consistent with the theory. The theory based on charge control method and quasistatic approximation does not agree with our results or with the experimental observations discussed in the paper.

Ionization coefficient measurement in GaAs by using multiplication noise characteristics

Abstract: Multiplication noise measurements for p+n type (100) GaAs avalanche photodiodes with various n-layer dopings ranging from 6 × 1015 to 9 × 1016 cm−3 confirmed that the ionization coefficient of electrons α is about two times larger than that of holes β in the electric field range from 24 × 105 to 56 × 105 V/cm When pure electrons were injected into the avalanche region, the multiplication noise power was proportional to the 27th power of the multiplication factor and the ionization coefficient ratio k = βα was constant, where k = 05 in the above electric field range The result was consistent with the multiplication factor dependence on light wavelength Using the constant ionization coefficient ratio k and the multiplication factor dependence on applied bias voltage, ionization coefficients α and β for electrons and holes were estimated

Punch-through currents in P+NP+ and N+PN+ sandwich structures—I: Introduction and basic calculations

Abstract: A qualitative description and a quantitative approximation of the current/voltage characteristic of the punch-through effect, based on drift- and diffusion-theory, is presented. An exact definition of the punch-through voltage is given. For small currents the current/voltage characteristic of the punch-through effect is an exponential curve i = I 0 exp [ q ( V − V PT )/ m ( i ) kT ] where m ( i ) is a non-ideality factor which is equal or larger than 2 and which increases with increasing current. At larger currents a deviation of the exponential curve is found due to space-charge limiting effects. A more general theory for small currents and some experimental verifications are described in Part II[48].

A time-dependent numerical model of the insulated-gate field-effect transistor

Abstract: A new method is described for computing the flow of mobile carriers in a semiconductor device, in two space dimensions plus time. In this method, each variable is updated separately, and in each step only linear systems of equations are solved. The method is stable, independently of the time step. Various current components are resolved by stream functions, and current balance is maintained. A sample computation is included, involving the switching off of an n-channel IGFET.