Report of Project MICHIGAN Infrared (u) Contract No. DA -36-039- SC-52654 1 March 1955 to 1 September 1955 2144-59-P The University of Michigan Engineering Research Institute Willow Run Laboratories Willow Run Airport Ypsilanti, Michigan

UNIVERSITY OF MICHIGAN 2144-59-P III CURRENT ACTIVITIES 3.1 DETECTOR RESEARCH (U) The research on infrared photodetectors has as its primary objective the development of new photodetector materials. Special emphasis is being placed on exploiting the semi-conductor nature of the III-V intermetallic compounds, so designated because they are comprised of elements which fall in the third and fifth columns of the periodic table. Samples of these compounds are being prepared, their optical and electrical properties are being measured, and associated theoretical studies are being carried out. 3.1.1 Preparation and Purification of Materials During the previous period, the zone purification of indium metal was carried out in a hydrogen atmosphere. It was found that a thick oxide coating formed on the ingot during the purification process. Inasmuch as the purification process takes almost a month of continuous heating of the metal, even the slightest traces of oxygen in the zone-melt atmosphere will eventually accumulate on the surface of the metal. For this reason, the zone purification of indium was tried in a vacuum; in this way, the oxide formation was greatly decreased. Spectroscopic analysis of one such purified indium ingot revealed that large concentrations of lead which existed in the original material were not moved appreciably by the purification process, while smaller concentrations of copper and tin were largely removed from the front of the ingot, although detectable traces still remained. Traces of seven impurity elements in the original material remained in the purified ingot. Recently, a source of 99. 9995 per cent pure indium metal was brought to Project MICHIGAN's attention. The high purity is obtained by the use of certain organic solvents, although the details of the process are not yet available. The spectroscopic analysis of two zone-purified antimony ingots show that of the 15 trace elements in the original material, 5 elements 5

UNIVERS ITY OF MICHIGAN 2144-59-P were reduced in concentration by detectable amounts, but further purification would be worthwhile before combining the antimony into III-V compounds. The results of the zone purification of antimony indicated that prepurification by vacuum distillation may be advantageous. A system for the slow vacuum distillation of antimony was built and operated (Fig. 3. 1-1). The spectroscopic analysis of the resulting distilled antimony ingots (Fig. 3.1-2) has not yet been carried out. Zone purification on the portion of distilled antimony which is expected to be most pure is in process. Arsenic was also distilled successfully. The oxides of arsenic were first driven off to the end of the distillation system. The temperature applied to the arsenic was then increased until the arsenic itself was distilled. While the results of this distillation process are being examined, commercially distilled arsenic is being used for the preparation of the arsenides. 3.1. 2 Preparation of Indium Arsenide Crystals The direct combination of indium and arsenic in an uncontrolled exothermic reaction will usually raise the vapor pressure of the uncombined arsenic to explosive magnitudes. For this reason, methods for mixing the two elements at a controlled rate are being investigated. The first procedure tried made use of two connecting evacuated sealed bulbs. One contained the arsenic at the distillation temperature and the other contained the indium at the melting point of indium. The chemical reaction then occurred at the surface of the molten indium. As the reaction proceeded, the temperature of the mixture of indium and indium arsenide was gradually raised. The process failed to give stoichiometric mixing because some of the arsenic condensed on an exposed (cooler) portion of the indium bulb. A second procedure, being tried at present, makes use of a single bulb in which the arsenic is placed around a carbon boat containing the inrdium (Fig. 3.1-3). The temperature of the entire bulb is then gradually raised to the melting point of the indium. The vapor pressure of the arsenic at this temperature should be safely below the tolerance of the bulb. Evaporation of the arsenic is continued for several days until it has completely disappeared. 6

UNIVERSITY OF MICHIGAN 2144-59-P ~ I l r ic.a furnace FIG. 3.1 -1 VACUUM DISTILLATION EQUIPMENT

UNIVERSITY OF MICHIGAN 2144-59-P j I &?:0:000~~~~~~~ Ic ~~~~~~~~~~~~~~~~~~~~:.::l:: i::i::: i::! ii i:i i:ii:::: iii~':X. FIG. 3.1 -2 DISTILLED ANTIMONY 8,,::::-:ii(~: ~~..~-:~.: ~ iii:::11~i!:i:},::'i~~~~~~~~~~~~~~~~~~~~~~~~~~~~p"~~i-ii i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i9.:'~~~~~~~~~~~~~~~~~~~~~~~: Y!::::....:i i: ~i:-:::::ii jii::~ F:IG ~:;-:i:::.l' 3:ii:'i::.1 -2 D~ ISTi IL L E D ANTMON

UNIVERS ITY OF M I CH IGAN 2144-59-P Vycor Tube i21 Arsenic Carbon Boat 1 Inch i Indium Metal FIG. 3.1-3 COMBINATION OF INDIUM AND ARSENIC 9

UN I VE RS ITY OF M I CHIGAN 2144-59-P 3.1.3 Preparation of Indium Antimony Crystals The preparation of indium antimony seed crystals was attempted by successively growing and shrinking a polycrystalline boule while pulling it from the melt in a hydrogen atmosphere. Large crystallites of indium antimony were obtained but indium metal appeared as a second phase in the boule. Some loss of antimony was noted in the pulling process which may have been due to the effects of water vapor or oxygen in the atmos - phere. Passing the hydrogen gas through a drying system and tightening the seals on the gas chamber should correct this difficulty. A new method of growing single crystals was tried: the crystal material was inserted into an evacuated glass tube and sealed off. The glass was then heated over a flame and pulled to make a capillary tube. The molten metal inside the capillary was pulled along with the glass to produce a fine wire of metal. Drawn samples of indium antimony produced in this way are shown in Figure 3. 1-4. Under favorable conditions, appreciable lengths of single crystal metal should be obtained. Because the semi-conducting materials under investigation expand upon freezing, a glass which softens at a temperature below the melting point of the metal must be used. This procedure allows the metal to crystallize with a minimum of constraint. There are several advantages to growing crystals this way: the geometry of the sample is that which is desired for most photoconducting detectors without further treatment; the rapid growth rate assures greater homogeneity of the sample; and the glass tube collapses around the metal before the pulling of the capillary to form a tight capsule which prevents gross dissociation of volatile constituents of the metal. On the other hand, certain disadvantages are encountered: the rapid freezing can produce greater concentrations of crystal defects, although annealing may reduce this concentration somewhat; potential probes constructed from the same crystal cannot be used for test purposes; and the orientation of the crystal axis is not well controlled. 3.1.4 Specimen Preparation Previous work on the shaping of small and fragile single crystal test samples (Ref. 1) made use of a dust cutter mounted on a lathe tool mount, 10

UNIVERSITY OF MICHIGAN 2144 -5 9 -P 1 Inch FIG. 3.1 -4 DRAWN SAMPLES OF INDIUM ANTIMONY 11

U NI VE RS ITY OF M I CH I GAN 2144-59 -P the desired shape being obtained by programmed travel of the mount. An improved method which makes use of a wire mask and plastic frame was developed during the reporting period. In the new procedure, the sample blank is first placed in the plastic frame and appropriate wire leads are soldered to it. The wires are fastened to the plastic frame so that the sample blank is suspended in the frame by the wire leads. The leads are then covered with polystyrene to protect them from the action of the etch or polish bath. The wire mask is placed in front of the sample blank and the exposed portion of the sample blank is eroded away. Greater uniformity can be obtained by this means as can be seen from Figure 3.1-5. After being shaped, the sample can be placed in the etch or polish bath without removing the frame. For low-temperature measurements, the sample is placed, frame and all, into a lowtemperature cell, as shown in Figure 3.1-6. This procedure for shaping samples is essential for work with indium antimony, which is many times more prone to fracture than is germanium. 3. 1.5 Electrical Measurements The lifetime of the minority carriers in germanium was measured by electrical injection. In this method, the minority carriers are injected at the current contacts by a pulsed current, and the time rate of change of the resistance of the sample is measured immediately following the pulse. The circuit used is a modification of that described in Reference 2. The apparatus is shown in Figure 3. 1-7. The lifetime of minority carriers in 30 ohm-cm undoped germanium was found to be about 50 microseconds. No special precautions were taken to protect the etched surface. Lifetime measurements of 2 ohm-cm undoped germanium, with surface treatment similar to the 30 ohm-cm sample, yielded values of about 5 microseconds. A sample of p-type nickel-doped germanium showed a minority carrier lifetime of about 1 microsecond. 3.1.6 Low-Temperature Measurements It is frequently desirable in semi-conductor work to make physical measurements of such things as Hall effect (Sec. 3. 1. 7), resistivity 12

UNIVERS ITY OF MICHIGAN 2144-59-P E U- - r uw I

UNIVERSITY OF MICHIGAN 2144-59-P::::Plastic Sample Frame FIG. 3.1-6 SAMPLE MOUNTED ON LIQUID AIR CELL 14

UNIVER SITY OF MICHIGAN 2144-59-P;:t w LU U I15

UNIVE RS ITY OF MI CH IGAN 2144-59-P (Sec. 3. 1.7), and noise (Sec. 3. 1.8) at extremely low temperatures. To facilitate these measurements, two low-temperature cells were constructed, one designed to use liquid helium as the cooling medium, and the other liquid nitrogen. The basic principles of construction of these low-temperature cells are explained in detail in Reference 1. As the first step in setting up the low temperature systems, the liquid helium and liquid nitrogen cells were tested for vacuum tightness both at room temperature and at liquid nitrogen temperature. A helium leak detector (made available through the courtesy of Dr. R. W. Pidd of The University of Michigan Physics Department) was used for the tests. The sample holder designed for use during low-temperature measurements was shown in Figure 3.1-6. It is made out of a cylindrical red copper bar which is held in thermal contact with the liquid coolant container by four machine screws (not shown). The mating surfaces were well polished, and thermal contact was assured by introducing a thin layer of vacuum grease, colloidal silver, or indium foil. In mounting, a sample is first suspended by its leads in a rigid plastic frame; the frame is then cemented to the sample holder, and one end of the sample is soldered with indium to the sample-holder surface. Finally, a radiation shield is placed over the sample holder to protect the sample from thermal radiation. Heat losses through the electrical connections are minimized by cooling a considerable length of the leads adjacent to the sample. Certain measurements - e. g. noise studies - are made with the sample insulated electrically from the sample holder. In this case, a thin insulating layer of krylon, instead of indium solder, is used to attach the sample to the sample holder. With this imperfect contact, the sample acquired temperatures two or three degrees higher than that of the sample holder. Temperatures in the liquid nitrogen region were measured with a copper-constantan thermocouple. A carbon resistor was used as a temperature indicator in the liquid helium cell, since its resistance 16

U NI VE RS ITY OF M I CH I GAN 2144-59-P changes linearly with temperature in this region. A second such resistor, immersed in the liquid helium, serves as a level-indicating device, since an abrupt resistance change occurs when the level falls sufficiently to expose the resistor. In order to carry out measurements of the photoelectromagnetic (PEM) effect (Ref. 8) at reduced temperatures, a liquid nitrogen cell was equipped with a transparent window, as shown in Figure 3.1-8. 3. 1. 7 Hall Effect and Resistivity Measurements The Hall effect is one of the chief sources of information about the conduction properties of semi-conductors. The mobility, the carrier concentration, and the type of material (n or p) can be determined from the Hall constant and the resistivity measurements. Information on energy gaps can be determined by making these measurements over a wide temperature range. To facilitate the Hall effect measurements, a 6-inch magnet from Varian Associates of California was purchased, together with a regulated power supply, and installed in the laboratory. The magnet was tested and then calibrated for one specific gap setting. The magnet, together with the mounting frame and the liquid helium cell, are shown in Figure 3.1-9. All Hall effect measurements were performed using d-c sample current and d-c magnetic field. A total of four measurements were taken for each of the four possible combinations of the relative directions of the sample current and the magnetic field. By considering these four measurements simultaneously, it was possible to obtain the corresponding value of the Hall voltage, free from all the associated effects but the Ettingshausen voltage; the latter cannot be separated from the Hall voltage if the d-c method is used. The Ettingshausen voltage usually constitutes about 5 per cent of the total measured value; hence the calculated Hall voltage will be accurate to 5 per cent even though the measuring equipment is inherently capable of much greater accuracy. To make the measurements more accurate, an a-c sample current may be used while still retaining the d-c magnetic field. 17

UNIVERSITY OF MICHIGAN 2144-59-P i i'......................... To Vacuum Pump Vacuum Seals for Electrical Leads 10 Inches Sample Mount Optical Window "i~~~~~~~~~~~~~~~~~~~~~~i~~~~~~~~~~~~~~iiiiiiiiiiliiiii.;:if':~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~?,,ii!:i,,''~,iii: FIG. 3.1 -8 LIQUID AI1R CELL 18

UNIVE RS ITY OF MICH I GAN 2144-59-P a act-, 10 Inches i: i Liquid Helium Cell Varian Magnet.... FIG. 3.1-9 LIQUID HELIUM CELL MOUNTED IN MAGNET 19

U NI VE RS ITY OF MICHIGAN 2144 -59 -P The resistivity of the samples was measured conventionally by simultaneous observation of applied voltage and current flowing. Four semi-conductor samples (commercial grade germanium, indium-doped germanium, and two specimens of nickel-doped germanium) were investigated at liquid air temperatures; to aid in sample preparation their resistivities and Hall coefficients were measured, and their carrier mobilities were computed. The electrical measurements program is being extended to include study of the electrical properties of semi-conductors in the microwave region. The phenomenon of microwave resonance has proven itself to be a powerful and dependable tool in the hands of the solid state physicist. However, a systematic exploration of microwave resonance as an aid to solid state physics research, particularly in the field of semi-conductors, has never been attempted. As the first step in the magnetic resonance work, the nature of impurities and crystal imperfections in the III-V and II-VI intermetallic compounds, at room temperatures, will be studied and correlation of the results obtained with carriers in these compounds will be attempted. Initially, the measuring apparatus will consist of a klystron and power supply, an attenuator, a high-Q cavity (operating in the TE01 1 mode), a frequency meter, a crystal detector, and an oscilloscope for the observation of the differential power absorption. The equipment will be operated at a frequency of 16, 000 Mc. Later, a phase-sensitive detector will be added to increase sensitivity. It is hoped to develop a system eventually which will detect paramagnetic impurities of concentrations of less than 1 part in 106. 3.1.8 Noise Studies As pointed out in Reference 1, noise spectra measurements of semiconductors cannot lead unambiguously to the determination of the microscopic process which produces the noise. One approach is to hypothesize a possible microscopic process and then test its validity experimentally. Such a hypothesis and test were presented recently by A. L. McWhorter 20

U NI VE RS ITY OF M I CH I GAN 2144-59 -P (Ref. 3). He suggests that surface traps exist at the surface of the material in the form of a contamination layer. The trapping of minority carriers occurs by means of tunneling through the surface to the traps. The spatial distribution of the traps provides (along with the tunneling effect) the proper distribution of lifetime for the microscopic process to give a 1/f noise spectrum. His experiments showed that various surface treatments caused corresponding modifications in the noise spectra of the bulk material. This work implies that almost any type of barrier layer through which minority carriers can tunnel can likewise produce a 1/f noise spectrum. Project MICHIGAN's experience has been that it is the exceptional sample which shows less than 5 per cent variations in resistivity over a length of 1 cm at room temperature. At liquid air temperature, resistance inhomogeneities tend to become greatly magnified. Hence, the present approach to the study of noise is to examine the possibility that resistance inhomogeneities in single crystal material may also be a source of the 1/f type of noise. Noise spectra of six samples of semi-conducting material were measured, and the degree of correlation of the noise arising from seriesconnected segments of two of these samples was studied following the procedure of Montgomery (Ref. 9). A monocrystalline sample of 30 ohm-cm germanium showed a considerable amount of correlation. A technique has been devised by means of which the signal-to-noise ratio of a detector can be increased in cases where appreciable noise correlation exists. In this procedure (Fig. 3.1-10), one segment (between A-B) of the crystal receives the optical signal to be detected, and the electrical connections are so arranged that the signal from this segment is subtracted from the (appropriately amplified) signal arising in the series-connected unilluminated segment (between B-C), thus removing the correlated noise. An example of the improvement obtained in the case of a germanium crystal illuminated with a sinusoidallymodulated neon light is shown in Figure 3.1-11. Here the upper oscilloscope trace is that obtained directly from the illuminated segment, 21

UNIVERSITY OF MI CHIGAN 2144-59-P Neon Bulb'~~~~~~ V C........-........ ~~~/' 1 mm FIG. 3.1-10 ELECTRICAL CONNECTIONS TO CRYSTAL FOR NOISE CORRELATION 22

U N I VE R S I TY OF M I C H I GAN 2144-59-P, " FIG. 3.1 -11 IMPROVED SIGNAL-TO- NOISE RATIO DUE TO CORRELATION OF NOISE 23

UNIVERSITY OF MICH I GAN 2144-59-P while the lower trace was obtained by removing the correlated noise. Improvements in the signal-to-noise ratio by a factor of as high as seven have been obtained in this way. The source of correlated noise may usually be traced to a contact. Since contact noise is a continual source of trouble (especially with detectors whose temperature is successively changed over rather wide limits), its removal by the correlation technique described above may be helpful in both laboratory and field equipments. A 1/f noise spectrum was found in a sample of indium antimony; however, further work must be done to determine if this noise was simply contact noise or noise generated in the bulk material. 3.1.9 One-Dimensional Lattice Calculations The calculation of the distribution in energy of the quantum states of a disordered one-dimensional crystal, begun during the last reporting period, was completed. Most of the value of these one-dimensional calculations lies in the qualitative insight obtained into the more complicated three-dimensional case of real crystals. Little attempt had been made by previous investigators to choose realistic values of parameters in these calculations. The present calculations attempted to improve this somewhat by choosing physically reasonable parameters which give the correct value of the energy gap for indium antimony for an infinite and ideal crystal. The effect of disorder and non-stoichiometry on the distribution of states in energy was calculated. Figure 3.1-12 shows a typical set of energy levels found by this method. The solid curve represents the distribution of states for a finite (3000 atom) onedimensional crystal. The points represent the results of calculations on MIDAC (the MIchigan Digital Automatic Computer) of the distribution of states, assuming a 90 per cent chance of short-range order. These calculations are in qualitative agreement with the observed variation of energy gap with type of carrier, and also indicate a high electron mobility compared to the mobility of holes, as has been observed in indium antimony. It was shown that where an isolated impurity level lies very close to the valence or conduction band, strong interaction will occur between impurities separated by several hundred atoms in the crystal. 24

40 30 20 - (/) Lu uii/ >- 10 0 Z 90C") 80 0/ I I~~~~,vr 6000 ENERGY (Electron Volts) FIG. 3.1 -12 DISTRIBUTION OF STATES FOR 3000-ATOM ONE-DIMENSIONAL CRYSTAL r ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~,o 80~~~~~~~~~~ 5970 v z 39 39. 125 39.250 39.375 39.500 39.625 ENERGY (Electron Volts) FIG. 3.1 -12 DISTRIBUTION OF STATES FOR 3000- ATOM ONE- DIMENSIONAL CRYSTA~L

UNIVERSITY OF MICHIGAN 311111111II I 9 1111115111 0liii352 3 9015 03483 5267