US2788298A - Methods of growing crystals and making electrical translators - Google Patents

Description

A nl 9, 1957 E. N. CLARKE 2,738,298

METHODS OF GRQWING CRYSTALS AND ELECTRICAL TRANSLATQRS Filed Nov. 2, 1951 4 E ti m lslwl Z8 5 .& y g s 3'40 E INVENTOR' 50 EDWARD N.CLARKE BY w 5 4 3 -z 0 ma ATTORNEY www- United States Patent METHODS OF GROWING CRYSTALS AND MAKING ELECTRECAL TRANSLATORS Edward N. Clarke, .Levittown, N. Y., assignor to Sylvania Electric Products 1126., a corporation of Massachusetts Application November 2, 1951, Serial No. 254,478

2 Claims. ((31. 148-15) The present invention relates to semiconductor trans lators, particularly to electrical devices of this class used for rectification and for amplification purposes.

The possibility of producing amplification and related translating functions by means of a thin layer of semiconductor of one type sandwiched between two bodies of the opposite type of semiconductor is known to the art. These may be termed area-junction translators. The object of the present invention is to provide a novel, effective method for forming in integral assembly a single body of a semiconductor having a thin layer of one type of semiconductor interposed between the opposite types of semiconductor. The thin interposed layer may be either N-type or P-type, and correspondingly the end portions of the element will be P-type or N-type, respec tively. A further object is to produce area-junction translators by a novel, effective method.

As will be seen from the illustrative disclosure de scribed below, in which the method is practiced using germanium containing appropriate impurity constituents, it is possible to produce a thin layer of one type of semiconductor when growing a crystal from a melt where the bulk of the crystal produced is of the other type merely by adjusting the pulling rate and the temperature in such fashion that the crystal being grown breaks away from the melt. This provides a barrier useful in making area junction rectifiers. The pulled crystal thus broken can again be dipped into the molten material for renewed growth. Where the molten material is such as to produce an N-type crystal, the break and renewed growth results in a P-type layer of extreme thinness. Where the melt is of such character as normally results in production of a P-type crystal, the layer produced as a result of breaking the smooth crystal-growing process followed by renewed dipping and further smooth growth is an N-type layer. Generally, and viewed apart from the designations of the particular semiconductor types, the invention is concerned with the formation of an electrical barrier and with forming multiple barriers close to each other in a unitary semiconductor body, and with the manufacture of semiconductor translators with such bodies. Further, the invention is concerned with introducing distinctive inhomogeneities in a layer of a crystal grown out of a melt.

The thickness and electrical properties of the layer are influenced somewhat by numerous factors which may be varied by the practitioner of the invention, including the time that the broken crystal is allowed to remain suspended above the melt, the speed of withdrawing the redipped end of the crystal, the extents to which the broken crystal is separated from the melt and re-dipped into the melt, and by such additional factors as temperature, temperature gradient, the extent and types of impurities in the melt, whether a gas is enclosed or the furnace is evacuated, and the time between the break and the redipping of the crystal for renewed growth of the crystal.

' The'na'ture of the invention and its further aspects of novelty will be apparent from the following detailed description of an illustrative embodiment thereof shown in the accompanying drawings.

in the drawing, Fig. l is the schematic view of a furnace, in longitudinal cross-section, suitable for carrying out the novel process. Figs. 2, 3., and 4 are somewhat on (.3 views which illustrate the sequence of steps in the 11 ustrative embodiment of the invention. Fig. 5 is an illustrative translator incorporating the distinctive layer, and Fig. 6 is a family of performance curves of the device in Fig. 5.

Referring to Fig. 1 there is shown a hard glass container to enclosing the crucible 12, which crucible is heated by resistance wire 14 having leads 16 extending outside the chamber 10. The resistance wire is wound with such distribution, and is properly subdivided, to give control over the temperature gradient in the melt. Ceramic walls 18 support the heating wire and prevent excess transfer of heat to the exterior. Rod 20 extends to a gland or vacuum seal 22 and can be manipulated for rotation and for axial travel whereby it is possible to lower and elevate a seed 24 as of crystalline germanium into the melt 23 of suitably doped germanium. A crystal 30 is shown being grown by gradually raising rod 20 after seed 2% has been immersed in the molten bath 28. The temperature of the molten bath is maintained critically, as a matter of judgment, in relation to the speed of with drawal of rod ill. The crucible is exposed to a helium atmosphere in growing both N-type and P-type germanium crystals. The atmosphere is continuously changed by flowing pure, cool gas into and out of chamber 10, through tubes 32.

in a typical process of growing a crystal 30, germanium is loaded in powder form or in pieces into crucible 12, together with any doping constituent that may be desired, and the furnace is heated by energizing resistance wire 14 after the chamber has been flushed with helium. After the material that is to form the semiconductor has become molten and stabilized at a desired temperature the seed 24 of approximately the same material but in solid crystalline form (being secured at the bottom of rod 2t?) is brought into contact with the surface of the melt and thereafter is slowly withdrawn vertically. The molten germanium in contact with the seed (and there after with the growing crystal) gradually solidifies and adds to the crystal as it is raised. Suitably the rate of withdrawal may be 6 inches per hour and the germanium may be near its melting point, slightly below the melting point so as to be supercooled at the solid-liquid interface.

At the outset the germanium used is of a high order of purity and it may contain in mixture or in alloy form a small amount of doping constituents and predominantly of a doping constituent such as to form an N-type or a P-type semiconductor. It appears probable that slight amounts of both donor and acceptor impurities are inherently present in the germanium used (although below spectroscopically detected amounts) apart from the doping constituent deliberately added.

As shown in Figs. 2, 3, and 4 a barrier of the opposite type of semiconductor can be formed in a crystal of a given type of semiconductor by the abrupt withdrawal of the growing crystal from the surface of the melt and, after a pause, reimmersion of that growing crystal. The crystal should remain suspended until the molten germanium at its bottom face has become solidified. Purely as a matter of illustration the germanium when withdrawn should be raised from the surface of the melt by no more than and, for a growing crystal of /8 diameter in a flowing helium atmosphere, no more than 5 seconds are required for solidification of the molten germanium at the bottom of the growing crystal to be completed, after which the crystal can again be brought into contact with the germanium melt and further growing of the crystal can be resumed.

In Fig. 3 the surface of the growing crystal 30 that has just been withdrawn from the melt 28 is designated lrtla. This is molten initially and solidifies rather rapidly in the ga atmosphere of the furnace. Zone 30b in Fig. 4

represents the physical appearance of the crystal that has been grown as in Fig. 2, broken as in Fig. 3, and where renewed crystal growth has taken place. The transverse dimension of the approximately cylindrical crystal 3% may be of the order of a as" or it can be much smaller or larger; and the region where the irregular appearance 30b is noticed may be about measured along the crystal axis. Irrespective of the axial extent of this break in the crystal growth, the layer of a contrasting type of semiconductor that is formed in the crystal produced may be no more than a thousandth of an inch or possibly as high as five thousandths of an inch, depending upon all conditions connected with the melt, the interruption of crystal growth, and the re-dipping. The junction of the grown crystal 30 with the seed 24 shows a continuity of the crystalline structure as evidenced by visual examination after etching the surface with a ferric chloride hydrochloric acid etch; and this continuous crystalline structure is evident despite the irregularity of contour at zone 30b, in most junctions formed.

It is possible to so cool the crystal after raising it from the melt, and to dip it and promptly start raising it in resumed crystal growing, that the germanium of the melt merely adheres to the dipped crystal. It is also possible to immerse the end of the re-dipped crystal too deep and-for too long a time, so as to obliterate the junction entirely. However, it is possible by adjusting the technique properly to obtain an integral continuity of crystal growth with a distinctive barrier layer of a contrasting semiconductor type in a large crystal of the opposite semiconductor type. This is the desired result. The technique of interrupting the growth of the crystal by removing it to a spaced position above the melt and again dipping it can be repeated so as to provide a sequence of thin contrasting layers extending all the way across the crystal being grown at spaced positions along its axis.

In forming semiconductor translators a crystal such as that shown in Fig. 4 is sub-divided by a transverse cut above and below zone 3% and thereafter by perpendicular families of cuts through the slice so formed. Before sub-dividing the slice, its opposite faces are advantageously electroplated as with copper to provide ohmic terminal connections. This provides individual semiconductor translator elements with two ohmic contacts 40 and 42 at the ends of the element and a third ohmic contact 44 is made to the interposed contrasting layer. It is desirable to polish and etch the lateral surfaces for removing foreign material and deformed mechanically worked parts of the crystal in the barrier region, and to facilitate establishing contact 44 properly. Portions 30' (Fig. may be of N-type and in that event portion 30b is of P-type germanium; and on the other hand, if the melt is predominantly of such character as to produce P-type portions 30' then portion 39b is of N-type semiconductor.

The portion 30b can be located simply by probing. One effective way of doing this is to rest the etched crystal against a metal base extending from terminal 40 to terminal 42 and to connect one of the terminals 44 or 42 to a galvanometer circuit with the opposite terminal of the galvanometer circuit constituted of a soldering iron suitably heated, having a sharp contact point at its end. At the instant that the probe touches the semiconductor on a lateral face the galvanometer will kick. The direction that it kicks, whether positive or negative, is an indication of the semiconductor in that immediate region, regardless of how thick or thin. Other probing t ch iques are well known to those skilled in the art.

Another way of locating the layer is by selective etching with an acetic acid, hydrofluoric acid, nitric acid etching solution containing bromine.

After the portion 30b has been located, contact 44 is established that should ideally be ohmic. Phosphor bronze is desirable for this purpose, the connection of the phosphor bronze wire to the layer becoming efitective after brief operation and/or electrical pulsing if it is not actually suitable at the outset. The dimensions of the abutting end of contact 44 against layer 39b are fixed by the thickness of the layer formed and the contact 44. will naturally be of a very sharp-ended wire where the layer is no more than a thousandth of an inch. Several contacts may be made about the circumference of layer 3% which contacts should be interconnected, but for most purposes a single connection is suitable. Any suitable mechanical support may be used to hold contact 44 in position.

The unit in Fig. 5 may be employed in a variety of circuits but with input and bias connections between terminals 42 and 44, static output (collector) characteristics of the circuit between terminals 40 and 44 may be obtained as shown in Fig. 6. Ie is the input-circuit current.

I have no theory with which I am thoroughly satisfied as to the mechanism of accomplishing the foregoing effect, and the following is offered by way of explanation only. It is Well known that when a long melt of germanium is gradually allowed to solidify progressively from one end of the melt to the other the ingot formed very often has portions of semiconductor of different characteristics at its ends. This is said to be due to a migration of impurities from the portion being solidified to the portion still molten. As the germanium that is molten at region 30a gradually solidifies, it may be that certain impurities in that limited molten volume migrate towards the surface not yet solidified, that is, the surface directly opposite melt 28, and exposed to the heat of the melt. The solidification of the limited quantity results in the formation of a thin layer of a contrasting type of semiconductor. Such layer is effective as an area rectifier or a photodetector when properly formed terminals are applied, without any re-dipping. When the bulk is N-type, the skin formed is P-type. After redipping, the layer formed is fused to the further mass of crystal that is grown thereafter. The crystalline structure that is first formed in the region 30 develops with a certain crystalline orientaiton relative to the solidliquid interface; and this same crystal orientaiton is realized in renewed growth after dipping, judging from the appearance of an etched grown crystal that has been broken and re-dipped.

The foregoing disclosure of methods for producing a thin layer of semiconductor in or on a larger body of a different form of semiconductor, and the conclusion of semiconductor translator using such composite semiconductor, will be recognized by those skilled in the art as subject to a latitude of variations. Those skilled with the art of semiconductors and with the art of growing crystals will recognize a variety of application of the various aspects of the invention represented in this disclosure; and accordingly the appended claims should be allowed that broad scope of interpretation consistent with the spirit and scope of the invention.

What is claimed is:

1. The method of making an electrical translator, including the steps of growing a crystal out of a germanium melt containing traces of conductivity-type determining impurities, during which operation impurities present in the melt tend to separate from the germanium that is in the process of solidifying on the growing end of the crystal, such impurities thereby separating preferentially between the growing crystal and the melt, removing the growing crystal from the melt with a molten layer of germanium on its end containing entrapped impurities, causing such molten layer to become solidified thereby to produce a layer of germanium of conductivity type opposite to that of said crystal, and applying electrical terminals to the solidified layer and to another portion of the grown crystal.

2. The method of making an electrical translator, including the steps of growing a crystal out of a germanium melt containing conductivity-type determining impurities, during which operation impurities present in the melt tend to separate from the germanium that is in the process of solidifying on the growing end of the crystal, such impurities thereby remaining preponderantly in the melt, removing the growing crystal from the melt with a molten layer of germanium on its end containing entrapped impurities, causing such molten layer to become solidified thereby to produce a layer of germanium of conductivity type opposite to that of said crystal, engaging such solidified layer with the melt in a way to enable resumed growth of the crystal and resuming growing of the crystal,

and thereafter applying terminals to the solidified layer and to the portions of the crystal on opposite sides thereof.

References Cited in the file of this patent UNITED STATES PATENTS 2,402,582 Scait June 25, 1946 2,402,661 Ohl June 25, 1946 2,565,338 Amico .Aug. 21, 1951 2,576,267 Scatf et a1 Nov. 27, 1951 2,583,008 Olsen Jan. 22, 1952 2,623,102 Shockley Dec. 23, 1952 2,623,103 Kircher Dec. 23, 1952 2,631,356 Sparks et a1 Mar. 17, 1953 OTHER REFERENCES Preparation of Metal Single Crystals," by A. N. Holden. ASM 1949 Rreprint No. 35, also published in Transactions, American Society for Metals, vol. 42 (1950).

Claims (1)

1. THE METHOD OF MAKING AN ELECTRICAL TRANSLATOR, IN CLUDING THE STEPS OF GROWING A CRYSTAL OUT OF A GERMANIUM MELT CONTAINING TRACES O CONDUCTIVITY-TYPE DETERMINING IMPURITIES, DURING WHICH OPERATION IMPURITIES PRESENT IN THE MELT TEND TO SEPARATE FROM THE GERMANIUM THAT IS IN THE PROCESS OF SOLIDIFYING ON THE GROWING END OF THE CRYSTAL, SUCH IMPURITIES THEREBY SEPARATING PREFERENTIALLY BETWEEN THE GROWING CRYSTAL AND THE MELT, REMOVING THE GROWING CRYSTAL FROM THE MELT WITH A MOLTEN LAYER OF GERMANIUM ON ITS END CONTAINING ENTRAPPED IMPURITIES, CAUSING SUCH MOLTEN LAYER TO BECOME SOLIDIFIED THEREBY TO PRODUCE A LAYER OF ERMANIUM OF CONDUCTIVITY TYPE OPPOSITE TO THAT OF SAID CRYSTAL, AND APPLYING ELECTRICAL TERMINALS TO THE SOLIDIFIED LAYER AND TO ANOTHER PORTION OF THE GROWN CRYSTAL.

Images