Index of content:
Volume 35, Issue 7, 01 July 1964
High Photovoltages in Silicon and Silicon Carbide Films and Their Origin from a Trap‐Induced Space Charge35(1964); http://dx.doi.org/10.1063/1.1702783View Description Hide Description
Photovoltages were observed in thin films of silicon and siliconcarbide which were not uniform in thickness. In some cases, the photovoltages exceeded the band gap. The photovoltages were retained by the film after removal of the light and decayed exponentially with time constants of the order of tens of seconds. The films were capable of acquiring persistent space charges upon application of an external voltage. The space charges decayed exponentially after removal of the applied voltage with time constants similar to that of the decay of the photovoltages. The photovoltages seemed related to the nonuniformity of the film, increasing with increasing nonuniformity. These results were interpreted to mean that the photovoltage is due to a space charge produced by a nonuniform distribution of trapped minority carriers, and a mathematical formulation was given to this concept. The temperature, light intensity, and carrier lifetime dependence of the photovoltage were in qualitative agreement with this concept. The temperature dependence of the decay constants of the photovoltage could also be interpreted satisfactorily in terms of this theory. A remarkable correlation between the photovoltage and the apparent resistance of the films was noted. It is suggested that the high photovoltages observed in films of other materials may have an origin similar to that discussed here.
35(1964); http://dx.doi.org/10.1063/1.1702784View Description Hide Description
The spontaneous fluctuations of polarization in polar dielectrics are considered by using the Einstein thermodynamic method, a general dielectrictheory method, and an equivalent circuit method. Mean‐square fluctuation of the polarization field is obtained by the first two methods and the frequency spectrum by use of the dispersionproperties. The equivalent circuit method gives frequency spectra directly and the two lines of analysis are shown to give consistent results. Noise magnitudes are exhibited and noise is shown to be a consequence of dielectricdispersion in all matter.
Line Strengths for Noble‐Gas Maser Transitions; Calculations of Gain/Inversion at Various Wavelengths35(1964); http://dx.doi.org/10.1063/1.1702785View Description Hide Description
Relative line strengths for s‐p, p‐d, and d‐f transitions of Ne, A, Kr, and Xe are derived by the method of Koster and Statz, under the assumption of the j‐l coupling scheme of Racah. When the relative strengths are given a common denominator, a set of rules for strong lines becomes apparent, similar to rules which have been noted for L‐S coupling. For comparison with experiment, we consider several sets of Ne lines, each set having one specific initial, and one final configuration. For the higher l values (p‐d and d‐f) the lines found to give oscillation are almost exclusively the lines with large relative strengths. The absolute line strengths S are then calculated in the Coulomb approximation of Bates and Damgaard; we consider only transitions between two excited states. There is the following simple relation between (gain constant/volume density of inversion) and S:α/(N 2/g 2−N 1/ g 1)=1.76×10−13 (mass number)1/2 S. The units of α are cm−1. N 1 and N 2 are in cm−3. S is in atomic units, a 0 2 e 2. The linewidth is taken to be determined by Doppler broadening at 400°K. For several lines upon which measurements of α have been made, we give the corresponding values of (N 2/g 2−N 1/g 1). This inversion quantity is a population difference between elementary quantum states.
35(1964); http://dx.doi.org/10.1063/1.1702786View Description Hide Description
A technique for the measurement of ionization rate and collision frequency in gases uses high‐powered focused microwaves. By using short pulses and focusing the energy into a small spot within a large vacuum chamber,diffusion losses are made negligible. Under these conditions the breakdown equation can be simply solved for the ionization rate in terms of easily measured parameters—the pressure, pulse width, and electric field. Experimental results with this technique are compared with published results. The agreement between the two is sufficiently good to confirm the usefulness of this technique.
Due to the focusing of energy into a spot remote from walls, the walls do not enter into the ionization process. In this way a possible source of error in the measurement of ionization rate is avoided.
Since the technique arranges conditions so that diffusion losses are negligible, it is not necessary to know the diffusion coefficient in order to determine the ionization rate.
35(1964); http://dx.doi.org/10.1063/1.1702787View Description Hide Description
Electron beams of approximately 2 kW power and acceleration voltages in the range about 100 kV are focused on metal surfaces in spots of 15 to 250 μ so that maximum power densities of more than 1013 W/m2 can be achieved.
Such electrons penetrate several orders of magnitude deeper into the solid or liquid materials than the classical penetration depth of 100 keV electrons would allow for. A high‐speed movie (2500 frames per second) of the process shows that the surface is closed by a liquid film most of the time and that it only ruptures during very short intervals. The temperatures at the surface and in the interior of the welds have been determined during the process, experimentally and theoretically, respectively. Also, the dislocation of the metal within the weld has been studied with an electron‐probe x‐ray microanalyzer and with radio‐active tracing. Measurements of mass density show that it decreases after the material has been affected by the electron beam.
In the discussion, a tentative explanation of the effect is offered.
35(1964); http://dx.doi.org/10.1063/1.1702788View Description Hide Description
The slow waves that propagate in longitudinally magnetized ferrite rods enclosed in cylindrical wave‐guides are investigated theoretically. The backward, azimuthally symmetric volume waves that propagate in a lossy ferrite rod are studied with the quasistatic analysis and the lossy permeability tensor derived from the Bloch‐Bloembergen phenomenological formulation. The attenuation constant is proportional to the linewidth, and for polycrystallineyttriumirongarnet (YIG) (ΔH ≈ 50 Oe), the loss is approximately 30 dB/cm. The interaction of these waves with an electron stream which passes longitudinally through a hole in the center of the ferrite rod is studied; for normal values of beam perveance (∼10−6 A/V3/2), backward‐wave oscillations should occur only for rods having a linewidth less than several oersteds and a length greater than a few centimeters. When the ratio of rod radius to beam hole radius is 1.5, the interaction parameter C is reduced by about 25% of the value when this ratio approaches infinity.
Brillouin diagrams are shown for the azimuthally dependent volume and surface waves in a lossless ferrite. Also, the results of a dynamic analysis are shown demonstrating the relationship between the magnetostatic modes and the waveguide modes. The quasistatic analysis is shown to give a close approximation to the correct dispersion relation even in the limit of fast waves, provided the radius is sufficiently small.
35(1964); http://dx.doi.org/10.1063/1.1702789View Description Hide Description
A theory is presented which investigates the emission fluctuations arising from random arrival of barium at the surface of the grains followed by donor migration through the effective emitting part of the grains. The basic causes for the noise are diffusion of donor centers and electrolysis in the coating, processes which are known to be vital in cathode operation. The computed spectra are characterized by a diffusionrelaxation time, a diffusion‐conduction time constant, and a parameter involving the electric field in the coating surface layer. Three different types of spectra result from this.
Experimental data are reported, mainly on 2×2 diodes which support various aspects of the theory. The dependence of the turnover frequencies on temperature, activation, and current is in good agreement with the predicted results. The magnitude of the noise is close to Poissonian, contrary to previous claims. Recommendations for attaining low flicker noise are given.
35(1964); http://dx.doi.org/10.1063/1.1702790View Description Hide Description
The photoelectric current due to the absorption of two photons has been calculated. Its magnitude is dependent upon the square of the incident radiation intensity, whereas, for the photoelectric effect due to the absorption of one photon, the magnitude of the current is linearly dependent upon the intensity.
The magnitude of the photoelectric current due to the absorption of two photons is small. However, under intense radiation, such as is obtainable from a ruby laser, a photoelectric current due to two‐photon absorption should be observed.
35(1964); http://dx.doi.org/10.1063/1.1702791View Description Hide Description
Single‐crystal whiskers of iron and iron‐cobalt alloys with diameters from 250 to 4000 Å were prepared by evaporation of the metal in a vacuum. Larger diameter whiskers, from 0.2 to 200 μ, were made by hydrogen reduction of iron bromide. Electron microscopic examination of the vacuum‐grown small whiskers showed them to be perfect appearing rods of uniform diameter. Electron diffraction indicated that they were all single crystals with a thin oxide layer. The whisker axes of the small whiskers were all along the 〈100〉 direction, but some of the large whiskers had a 〈111〉 orientation. Magnetic and chemical analysis indicated that the oxide thickness increased with increasing diameter from about 30 Å for 300‐Å whiskers to about 100 Å for 1000‐Å whiskers. New calculations were carried out for the remanence curves, the intrinsic coercive force and the remanent coercive force of arrays of infinite cylinders with random orientation in either two or three dimensions reversing their magnetization by curling, by fanning, or coherently. The dependence of coercive force,remanence, and rotational hysteresis of the randomly oriented whiskers as a function of diameter correlates well with the curling mode of magnetization reversal for diameters less than about 1000 Å. These results provide the first clear experimental evidence for magnetization curling. This is in sharp contrast to the size‐independent behavior of bumpy, irregular, electrodeposited ESD particles, whose properties are described by a fanning mode. The results for alloywhiskers, with up to 60% cobalt and with various crystallographic orientations, were the same as for pure iron. Since the only difference between the alloy and the pure ironwhiskers is the crystal anisotropy energy, it is concluded that the crystal anisotropy contribution is negligible. The results suggest a gradual transition from the uniform curling mode below about 1000 Å to some sort of heterogeneous reversal process propagating from one or more nucleation centers in each whisker with increasing diameter.
35(1964); http://dx.doi.org/10.1063/1.1702792View Description Hide Description
Superconducting thin film strips of indium and tin from 3000 to 10 000 Å thick, having sharp, magnetically induced resistive transitions between their superconducting and normal states, have been produced. Usually, the width of such transitions may be as much as 50% to 100% of the critical field because of the relatively high fields required to switch the thinner, tapered edges of the films. This tendency has been counteracted by preferentially depressing the critical temperature near the film edges through the use of an overlapping film of copper or gold, up to several hundred angstroms thick. Transition sharpening equivalent to that observed when the edges of the superconducting films are mechanically removed has thus been achieved. Behavior predicted by analyses of a simplified model has been found to be consistent with experimental observations.
35(1964); http://dx.doi.org/10.1063/1.1702793View Description Hide Description
The steady‐state thermal dissipation from heat‐generating thin‐film strips, immersed in liquid helium, has been examined by measuring the thermal hysteresis of the magnetically induced resistive transition of superconducting thin films of tin, deposited onto glass and aluminum substrates. The dissipated heat was found to depend linearly on the temperature difference between the film strip and the helium bath. It was thus possible to characterize the heat transfer by an effective coefficient of heat transfer that was approximately ten times as great for films on aluminum substrates as for films on glass substrates. For a 0.12‐×2.3‐ mm2film strip, its value was approximately 8×10−3 W/°K for the aluminum substrate, approximately 9×10−4 W/°K for the glass substrate. The coefficient was also dependent on the nature of other thin filmsdeposited beneath or above the film strip. Analysis of the data indicated that the thermal conductance between the film strip and the helium bath was about 0.1 W/cm2 °K, and the thermal conductance between the film strip and the aluminum about 1.7 W/cm2 °K.
35(1964); http://dx.doi.org/10.1063/1.1702794View Description Hide Description
Calculations are made of certain averages for particles from a plane, unidirectional source being slowed down by elastic collisions in an infinite medium of randomly located target atoms. Calculations are based on the moments form of the Boltzmann transport equation and follow procedures frequently used in the theory of neutron moderation. Assuming isotropic scattering in center‐of‐mass coordinates, averages are determined for powers of the distance from the source and for Legendre polynomials which give information on the direction of motion. When the mean free path varies as a power of the energy a number of averages are determined at all energies for equal masses of fast particles and target atoms, and for the first collision interval for unequal masses. Expressions for spatial averages at zero energy are obtained by taking the Laplace transform of the moments equation. For equal masses and λ proportional to Ek , it is verified that the distribution of stopped particles at moderately deep penetrations is of the form zp exp[‐z/λ(E 0)]. For k>1, p is approximately 2/(2k+1). A few comparisons are made with moments calculated by other methods and comments are included on the place of such calculations in the theory of the stopping of kilovolt ions. An appendix reports some preliminary results for anisotropic scattering.
35(1964); http://dx.doi.org/10.1063/1.1702795View Description Hide Description
Electrical resistivity and Seebeck coefficient values were measured on compact polycrystalline samples from the MoSe2−WSe2 and MoTe2−WTe2 systems. In the telluride system the resistivity versus temperature data indicated a number of activation energies from room temperature to the dissociation temperature. In the selenide system there was little divergence from the semiconducting properties of the isomorphic binary compounds. Only one activation energy was found for each of the samples from this system.
Determination of Unloading Behavior of Uniaxially Strained 6061‐T6 Aluminum from Residual Strain Measurements35(1964); http://dx.doi.org/10.1063/1.1702796View Description Hide Description
The zero stress point on the one‐dimensional strain, dynamic stress—strain unloading curve of shock loaded 6061‐T6 aluminum has been determined by measuring the residual strain in the material after shock loading. As any postulated release path from a given stress should pass through this point, it is possible to deduce certain aspects of the manner in which the material unloads from a shocked state. A comparison of the results of these experiments with the strains predicted on the basis of the simplified elastic—plastic theory establishes the inability of the simplified theory to accurately predict the release path of 6061‐T6 aluminum from a stressed state. Two major additions to the simple theory are required to explain the permanent strain data. These are a consideration of the Bauschinger effect, and the incorporation of an increased flow stress greater than that predicted from quasistatic data; the latter effect especially is quite possibly strain‐rate dependent. Another somewhat surprising result of the permanent strain experiment is the demonstration of the large variations in permanent strain resulting from seemingly minor changes in material composition.
35(1964); http://dx.doi.org/10.1063/1.1702797View Description Hide Description
Some properties of oxidizedPermalloy films with well defined rotatable initial susceptibility (RIS) have been measured throughout the composition range 70% to 90% nickel and over the thickness range 500 to 7000 Å. These properties include initial suceptibility, coercive force,saturation magnetization, rotational hysteresis and the rotational hysteresis integral. Interesting correlations were found between these properties and the thickness and composition. In particular, the rotational hysteresis integral has the characteristic value 1.5 for type 1 RIS and a different value 3.5 for type 2 RIS, independent of thickness or composition in the range studied. It was concluded that there is no significant correlation between the RIS property and macroscopic magnetostriction of the Permalloy used.
35(1964); http://dx.doi.org/10.1063/1.1702798View Description Hide Description
In estimating the nucleation field of an infinite rectangular bar, Aharoni used a variational method to calculate a lower bound to the magnetostatic energy. He constrained the variable potential to a specified form inside an ellipse circumscribed about the rectangle. In the present article, the method is improved by imposing the constraint only inside the rectangle itself. The resulting potential problem can be reduced to that of a bar of infinite permeability in an applied field of specified form; the final results can be expressed in terms of the demagnetizing factors Dx ∞ and Dy ∞ of such a bar. Similarly, an upper bound to the energy can be expressed in terms of the demagnetizing factors Dx 0 and Dy 0 of a bar of zero permeability. For rotation in quasiunison, the new lower bound to the nucleation field is considerably better than the old, but the new upper bound is not as good as that already obtained by constraining the magnetization to uniformity. New tables of , and as functions of dimension ratio p are included; the demagnetizing factors at infinite and at zero permeabilities satisfy the relations Dx ∞+Dy 0= Dx 0+Dy ∞=1.
35(1964); http://dx.doi.org/10.1063/1.1702799View Description Hide Description
Some aspects of the radiation of plane elastic waves from a depletion layer in a piezoelectricsemiconductor are examined. A formula, relating sonic output flux to frequency, drive amplitude, and junction characteristics, is developed for the case of small electromechanical coupling. The production of elastic energy by the depletion layer transducer is nonlinear. Harmonic production and suppression of selected harmonics is discussed. A half‐power bandwidth of 73% is possible for fundamental operation.
35(1964); http://dx.doi.org/10.1063/1.1702800View Description Hide Description
Cesium antimonide photocathodes having various compositions can be prepared. The changes in the photoemissive yield and the resistance correspond reversibly to the changes in the amount of Cs. The maximum yield is attained in a composition Cs3−δSb, which is a p‐type material (δ is a positive fraction). If the formation of a Cs−Sb photocathode is carried out under a Cs pressure about 10−5 Torr at 170°C, the photocathode maintains the most suitable composition for photoemissive yield. By flowing a Cs ion current through the Cs−Sb film, a p‐n junction is formed in it. The mobility of Cs ions is responsible for the change in photoemissive yield during the operation.
35(1964); http://dx.doi.org/10.1063/1.1702801View Description Hide Description
The conductivity of Al‐Al2O3‐metal diodes that show low‐frequency negative resistance in their current—voltage characteristics depends on impurities in the oxide and on the metal used as counterelectrode. For heavily doped Al2O3, development of diode conductivity by application of voltages occurs at ∼4 V, independent of oxide thickness. For oxide films that are not deliberately doped, the field in the insulator is more important than voltage in developing conductivity. Al‐Al2O3‐metal diodes have been constructed with Ag,Au,Cu, Co, Sn, In, Bi, Pb, Al, and Mg counterelectrodes. The current—voltage characteristics which develop depend on the metal and on polarity of the diode voltage during development of conductivity. With Ag as counterelectrode, most diodes were initially shorted; with Mg as counterelectrode, no diode conductivity could be developed. Other metals fall in between and give peak currents in the current—voltage characteristics in the sequence Au,Cu, Co, Pb, Sn, Bi, In, Al. There is no correlation between Al‐Al2O3‐metal diode conductivity and metal radius or work function.
35(1964); http://dx.doi.org/10.1063/1.1702802View Description Hide Description
A theoretical model is developed for the effect of mechanical stress on the electrical characteristics of Ge and Si p‐n junction devices. This model is based upon the stress‐induced variations in energy band structure and their effect on minority carrier densities. The changes in minority carrier densities are shown to depend upon the type of stress applied, with anisotropic stresses causing larger changes than hydrostatic stresses. From the calculated dependence of minority carrier densities upon stress, the equations are developed for the current—voltage characteristics of diodes and transistors under stress.
Following the general analysis several examples are given in which stress is applied to a small section of a junction in a diode or transistor. The results show that at stress levels greater than 1010 dyn/cm2, the device currents can change by several orders of magnitude when stress levels are changed by a factor of 2. The theory is compared with published experimental data and found to be in good agreement.