Base Transit Time Research Articles

Factors Influencing Prediction of Transistor Current Gain in Neutron Radiation

Prediction of transistor performance in neutron radiation is subject to many variables. Several of these were investigated in order to improve prediction accuracy. Experimental data are discussed for the base transit time parameter, variation in damage with different bias conditions during irradiation, changes in damage observed after room-temperature storage and repeated high current measurements, and junction capacitance as a function of measurement frequency before and after irradiation.

Transit Time of Junction Transistors at High Bias Currents†

A study is made of the behaviour of the small-signal base transit time of a transistor at high bias current levels, where increase of transit time occurs due to collector depletion layer contraction in combination with high level injection effects in the base. Computation of theoretical current dependence of transit time is carried out for devices of single-diffused base grading with alloyed emitter and collector junctions. A physical model of the devices, involving exponential base impurity density grading, is used as a basis of analysis. The physical parameters of this model are determined specifically for the transistor samples under study by interpretation of measured terminal properties under low-level injection conditions. Very close agreement between measured and computed dependence of base transit time on d. c. bias current is obtained, subject to appropriate allowance, in analysis, for variation of the operating temperature of the device with d. c. bias condition.

Effect of Operating Conditions and Transistor Parameters on Gain Degradation

Data for predicting transistor gain degradation in a neutron radiation environment were obtained from experimental studies of the variation of the radiation damage constant KT as a function of temperature and current during both measurement and irradiation. Relatively small spreads in the values of KT were obtained when the individual base transit times were measured and the radiation exposures were precisely determined. Radiation damage effect data from 20 different n-p-n silicon transistor types were obtained and normalized to the same minority carrier concentration in the base region, using the transit time and the VBE-Ic characteristics of the specific transistors. The resulting relative dispersions of the damage constant for minority carrier densities of 1015/cm3 and 1016/cm3 were typically 15 to 20 percent. These dispersions indicate that electrical measurements of transistor physical parameters can provide reasonably accurate predictions of the gain degradation for a wide variety of transistors without additional radiation testing of specific transistor types.

Computers in the front lines: micromodules make it possible : A. S. Rettig, Electronics, p. 77 (Jan. 4, 1963)

In this paper the total switch time for a transistor in a Direct Coupled Transistor Logic (DCTL) circuit is simulated by using Laplace transformations of the Ebers-Moll equations. The influence of doping gradients and germanium gradients in the base is investigated and their relative importance and their limitations are established. In a well designed bipolar transistor only a minor enhancement of the total switch time is obtained with the use of a doping gradient in the base. However, for bipolar transistors with base thickness over 500 Å, an improperly selected doping profile could be devastating for the total switch time. For a bipolar transistor the improvement of the total switch time due to a linear germanium gradient in the base could be up to about 30% compared with an ordinary silicon bipolar transistor. Still, a too high germanium gradient forces the normal transistor current gain (αN) to grow and the total switch time is thereby increased. Further enhancement could be achieved by the use of a second degree polynomial germanium profile in the base. Also in this case, care must be taken not to enlarge the germanium gradient too much as the total switch time then starts to increase. In all cases the betterment of the base transit time that is introduced by the electric field will not be directly used to reduce the base transit time. Instead the improvement is mostly used to lower the emitter transition charging time. However, the most important parameter to control is the normal transistor current gain (αN) that has to be kept within a narrow range to keep the total switch time low.

Semiconductor surface phenomena, their causes and effects : E. Bartels, Internat. Eltekon. Rundschau17, Nr.7, S. pp. 345–348 (1963)

In this paper the total switch time for a transistor in a Direct Coupled Transistor Logic (DCTL) circuit is simulated by using Laplace transformations of the Ebers-Moll equations. The influence of doping gradients and germanium gradients in the base is investigated and their relative importance and their limitations are established. In a well designed bipolar transistor only a minor enhancement of the total switch time is obtained with the use of a doping gradient in the base. However, for bipolar transistors with base thickness over 500 Å, an improperly selected doping profile could be devastating for the total switch time. For a bipolar transistor the improvement of the total switch time due to a linear germanium gradient in the base could be up to about 30% compared with an ordinary silicon bipolar transistor. Still, a too high germanium gradient forces the normal transistor current gain (αN) to grow and the total switch time is thereby increased. Further enhancement could be achieved by the use of a second degree polynomial germanium profile in the base. Also in this case, care must be taken not to enlarge the germanium gradient too much as the total switch time then starts to increase. In all cases the betterment of the base transit time that is introduced by the electric field will not be directly used to reduce the base transit time. Instead the improvement is mostly used to lower the emitter transition charging time. However, the most important parameter to control is the normal transistor current gain (αN) that has to be kept within a narrow range to keep the total switch time low.

Determination of the physical parameters of transistors of single- and double-diffused structure

This paper is concerned with the estimation of physical parameters of the internal device structure of transistors in which the collector-base junction is of gradual (diffused) transition. The parameters evaluated are the dimensional quantities of base width, depletion layer width, and junction area, together with those that characterize the impurity density distribution in the base and other regions according to some representative functional expression. Previous determinations of such parameters have been restricted to devices of step (alloy) junction type, but here single- and double-diffused structures, in which the collector junction is graded, are considered. The estimation of physical parameters is carried out in terms of relatively simply measured properties of the transistor--in particular, the dependence of base transit time and junction capacitance on operating bias conditions. The theoretical basis of the determination is developed on the assumption of a simplified model of the device structure in which the major part of the base region and the collector are represented by an exponential distribution of impurity density, while a possible opposing field region near the emitter is considered to be of linear grading. In the analysis of this model, the position-dependence of the diffusion and mobility coefficients is taken into account. Experimental results show that consistent evaluation of the physical parameters of the device (viz, of its model representation) is possible for a range of structural types.

Stored charge and the Ebers-Moll ON voltage

Usual analyses of the current-saturated condition for the junction triode transistor have related storage time and the emitter-to-collector ON voltage to current density and the forward and inverse alpha. It is shown here that in the current-saturated condition the stored charge is proportional to current and to base transit time, and is reciprocal to the emitter-to-collector ON voltage. This reciprocity is linear when the collector-to base junction is significantly forward biased. As current injected at the collector becomes small in comparison to that which it receives from the emitter, the stored charge approaches a minimum for the given emitter-to-collector current and collector-to-emitter voltage rises without appreciable decrease of stored charge. These results can also be obtained by regarding the conductivity modulated base layer as a resistor whose resistance is reciprocal to the stored charge. An obvious conclusion is that extremely low Ebers-Moll electronic ON voltage is undesirable in fast switching transistors designed for maximum speed. An optimum value of the ON voltage appears to be around 0.06 volts for homogeneous base transistors and perhaps 0.12 volts for diffused base structures.

Indium antimonide transistors

The design, fabrication, and electrical characteristics of an n-p-n indium antimonide transistor which operates at 77°K are discussed. An analysis of the expected high-frequency performance is presented and a comparison made to a p-n-p germanium transistor. Because of its high electron mobility, the indium antimonide transistor should have a base transit time 1 26 that of a similar germanium transistor. Although these ultimate capabilities have not been approached, transistors with f t as high as 300 Mc/s and β's of 500 have been observed. Switching speeds approaching those of the best state-of-the-art germanium transistors have been measured.

Impurity-density distribution in the base region of drift transistors

The nature of the impurity grading in the base of drift transistors is studied by measuring the dependence of the base transit time on collector voltage. Provided that modulation of collector-depletion layer width caused by change of collector voltage occurs only in the base material, as in the case of an alloyed collector, it is possible to deduce the base-region field parameter m=\Delta V/(kT/q) from such transit-time measurements. By this means, the validity of assumed distributions of impurity density may be verified; in particular, it may be established whether the impurity grading approximates an exponential or a complementary error function (erfc) form. Results are given for a number of drift-transistor samples, most of which are believed to have undergone impurity diffusion into the base material from a constant surface concentration during fabrication. In all cases, however, interpretation of measured data indicates a base impurity-density distribution approximating exponential rather than erfc form to be present.

Microwave Operation of Drift Transistors in Transit-Time Mode

The transit-time transistor and its operation was first proposed by Weinreich <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">^1</tex> in relation to a diffusion transistor. The similarity to the diffusion-delay diode and the existence of a negative resistance has been pointed out in accord with the original theoretical work by Shockley <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">^2</tex> . In this paper the drift transistor is analyzed in respect to transit-time effects of minority carriers across the base region, which appear at frequencies above the normal cutoff frequency and are more favorable than in a diffusion transistor due to larger excess phase shifts in the base transport factor. Amplification and power-generation properties at frequencies properly related to the base transit time are theoretically predicted and lie within frequency bands above the cutoff frequency. Experimental vertification of gain and power have been obtained with some commercial and developmental high-frequency mesa-type drift transistors. The results are reported, together with the special microwave circuitry utilized in the experiments. Although realization of gain and power in the microwave region with present transistor encapsulations encounters difficulties in the form of parasitic circuit elements, e.g., lead inductances, and interelectrode capacities, a promising application of suitably designed and packaged drift transistors in the microwave-frequency region will be demonstrated. The operation of the drift transit-time transistor is similar to its vacuum-tube counterpart. Microwave performance to date is very encouraging, and certain advantages over its vacuum-tube equivalent are indicated.

Determination of physical parameters of diffusion and drift transistors

The dynamic properties of drift and diffussion transistors are studied under both low- and high-level injection conditions in terms of the excess carrier charge in the base region. Subject to the assumptions of effectively one-dimensional device geometry with exponential impurity grading and collector conductivity much greater than that of the base, methods of determination of the main physical and high-frequency equivalent-circuit parameters of transistors are presented, utilizing the bias dependence of excess carrier charge in the base and space charge in the collector-depletion region. It is shown that measurement of the base transit time and the junction capacitances under low-level injection conditions enables the following physical parameters to be determined: base field parameter <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m = \DeltaV/(kT/q)</tex> , base width, base impurity distribution, emitter area, collector-depletion layer width. All the measurements are carried out at relatively low frequencies. Solutions for the base charge distribution, and hence charge-defined transit time, have been derived for the cases of exponential and erfc impurity grading under high-level injection conditions, assuming one-dimensional device geometry, constant base cross-sectional area, and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\micro</tex> independent of injection level. Experimental results have been found to differ greatly from the theoretical expectations based on these assumptions, the drift transistor showing greater and the diffusion transistor less than the predicted dependence on injection level. Explanations for the observed effects have been put forward in terms of reduction of emitter area and of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\micro</tex> at high injection levels.