Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.


  1. Advanced Patent Search
Publication numberUS2768914 A
Publication typeGrant
Publication date30 Oct 1956
Filing date29 Jun 1951
Priority date29 Jun 1951
Publication numberUS 2768914 A, US 2768914A, US-A-2768914, US2768914 A, US2768914A
InventorsBuehler Ernest, Gordon K Teal
Original AssigneeBell Telephone Labor Inc
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Process for producing semiconductive crystals of uniform resistivity
US 2768914 A
Abstract  available in
Previous page
Next page
Claims  available in
Description  (OCR text may contain errors)

1956 E. BUEHLER ET AL 2,768,914


G. K. TEAL MOSQW INVENTORS: HL ER Oct. 30, 1956 E. BUEHLER ET AL 7 2,768,914



CRYSTALS OF UNIFORM RESISTIVITY Fiied June 29, 1951 4 Sheets-Sheet 5 o/sr/wca -/NC/-/ES DISTANCE -/NC/1'E.5



By ax. TEAL low 8.0m



Gordon K. Teal, Summit, N. J., assignors to Bell Telephone Laboratories, Incorporated, New York, N. Y., a corporation of New York Application June 29, 1951, Serial No. 234,408

Claims. (Cl. 1481.6)

This invention relates to methods of forming semiconductive materials usable, for example, in transistor varying cross-sections, may be multicrystalline or may consist of but a single crystal, may be of constant resistivity and excess impurity concentration at any desired level over substantial portions of their length, may contain P-N boundaries at any desired location or locations longitudinally or laterally and of any desired electrical characteristics, and may contain N-P-N junctions at any desired location or locations, and of tailored electrical specifications.

In short, the process of the present invention consists of the drawing of a crystal or crystals from a molten body of a semiconductive material by dipping a seed crystal into the melt, allowing the interface of the crystal and the melt to arrive at thermal equilibrium and pulling the crystal at such a rate that the molten material crystallizes out on the seed. The forming crystalline body is rotated during the pulling procedure so as to result in a mass with a symmetrical cross-section and'the pulling rate and melt temperature are constantly varied so as to control the level of excess impurity present at any point in the forming mass. Also described are means of controlling resistivity and conductivity type additions of significant impurities and by heat treatment all during the drawing procedure. 7

It has previously been suggested that crystals of germanium could be grown from the melt using acons'tant pulling rate. It has been observed, however, that with a predominance of most significant impurities in germanium, resistivity in the solid decreases progressively in such a process as the molten mass crystallizes, thus indicating a I of less than 1 for these impurities (I=impurity concentration in the solid over impurity concentration in the liquid). That is to say, the forming crystalline matter approximates the normal freezing curve of the alloy of germanium and whatever impurity is being used. An example of such a freezing curve for antimony in germanium may be found in The Physical Review,

vol.*-77, pages 809m 813, March 15, 1950,;lg5arson',

Struthers, Theuerer, Fig. 3.

By thefprocess of the present invention, the'normal freezing curve is avoided by varying'the' rate of pull. The generalprinciple, 'whereit is desiredf to produce a crystal having a constant resistivity zone and'where I is lessthan 1- (as it is in all impurities thus far.testecl for which I has been computed); is to decrease the rate of pull gradually as the crystal is formed so as to entrap less and less impurity and so as to negate" the'naturaI tendency of the forming crystal to follow its freezing curve. i .w

Were rate of pull to be the only variable in the process, the crystal would increase draw'nifrom the. m t duet t dr ppinsirate pf. pull. lTolcounteract ,7

this effect and to maintain a constant cross-section over 2,768,914 Patented Oct. 30, 1956 "ice 2 the constant resistivity Zone, the decreasing pulling rates are balanced by an increasing melt temperature. The latter decreases the rate of crystallization and is monitored so as to be exactly equal and opposite in efiect to the increase in cross-section that would result from the decreasing pulling rate. 4 v I p p v The invention can be better understood by reference to the accompanying drawings in which: p V

Fig. 1 is a front elevation of a piece of apparatus on which the process of this invention can be carried out;

Fig. 2 is a sectional view of the vibrating member which beats against the wire supporting the crystal in the apparatus of Fig. 1, thereby having a stirring effect; I

Fig. 3 is a front elevation of the crucible assembly embodied in the apparatus of Fig. l;

Fig. 4 is a front elevation, partly in section, of an alternate form of crucible assembly showing doping means whereby the resistivity and conductivity type of the forming crystal may be influenced by the addition of controlled amounts of significant impurities;

Fig. 5 is a perspective view of still another form of crucible;

Figs. 6A and 6B are graphical representations of the variation of resistivity along the length of specimens which may be produced under different conditions of operation by the process of this invention;

Figs. 7A and 7B are graphical representations of the variation of impurity concentration along the length of specimens which may be produced under different conditions by the process of this invention;

Fig. 8 is a graphical representation of resistivity variation with length under still other conditions of operation.

Referring to Fig. 1, the apparatus is put into use as follows:

An ingot of the semiconductive material is placed in carbon crucible 1, a seed 2 of the same material 1s placed in chuck. 3; crucible assembly 4 on which is mounted crucible 1 is then inserted in the bottom of quartz envelope 5. The equipment is flushed by passing. nitrogen into gas inlet 6, through envelope 5 and gas outlet 7. After the system has been flushed with nitrogen, hydrogen or some other gas which will have a minimum effect on the composition of the forming crystal is then allowed to, circulate by the same route and is continued throughout the process. A high frequency generator, not shown, is then turned on and a current is passed through induction coil 8 in order to heat carbon crucible 1. After the ingot is completely molten, spindle 9 to which seed 2 is attached is then lowered until the seed just touches the melt. Vibratorlt) and rotator 11 are then turned on.

After the desired waiting period, motor 12 is turned on If what is desired is a single crystal of germanium,-the

process is allowed to.-proceed until all the melt has-been withdrawn with appropriate variation in pulling rateand in temperature 'of the melt as later described. If it is desired to form aP-N boundary bygas doping at the appropriate. point in the process, valves 16 and 17 are opened and valve 118is closed thus allowing hydrogen to pass through inlet 19'through the reservoir containing the desired doping substance in liquid form 20 and gaseous form 21, thus sweeping doping gas 21 through flow meter 22 into atmosphere of quartz envelope 5. This doping gas could alternatively be passed directly into the melt through base 4, through a hole in the crucible not shown, or through doping tube 23. i I

v In certain cases it is desirable to; dope by meansvof solid pellet. Where desired, the desired number and sequence of pellets may be mounted in magazine 24.

At the desired time, motor 25 is turned on. This motor is directly coupled to dispenser 26 containing an opening which is brought in line with one chamber of magazine 24 thereby allowing a pellet to flow through tube 23 into the melt.

Throughout the process, cooling water is passed through water inlet 4A into a water jacket surrounding crucible 1 and is discharged through water outlet 4B. The upper portion of quartz tube 5 is also cooled by passing water through 40 and out 4D. Switches A, B, and C control, respectively, rotation, vibration and solid doping. Valve D controls gas inlet 6.

Fig. 2 is a detail drawing of vibrating member 19. Eccentric 27 mounted on the shaft of motor 28 causes .shoe 29 to vibrate cable 14.

Fig. 3 is a detail view of the crucible assembly. From this figure, one may get a clearer picture of base 4 and crucible 1. The temperature induced in the melt and crucible 1. by coil 8 is controlled by means of an electrical circuit, not shown, monitored by thermocouple 30.

Fig. 4 illustrates an alternate crucible assembly construction to permit solid pellet doping. By this process the pellet 31 is placed in hole 32 and is supported by quartz rod 33 which can be raised above the surface of the bottom of the crucible by actuating steel rod 34. By this alternate process, pellet 31 is melted along with the melt in the crucible so that the doping material enters the melt not as a solid pellet but, rather, as a molten doping alloy.

Fig. 5 illustrates a crucible assembly for yet another alternate method of doping. By this process the pellet 31 is placed in an indentation on the rim of crucible 1. After this pellet has become molten and at the desired time, actuating member 32 which may be made of quartz or carbon is rotated so as to push the molten doping substance over the rim of the crucible 1 and into the melt.

Figs. 6A and 6B are plots of resistivity against distance measured by two point probes along single undoped crystals of germanium produced by this process. As may be seen from curve 6A, the resistivity of the initial portion of the crystal is dropped from point 35 to 36 while the cross-section of the crystal is allowed to build up. The portion of the curve between points 36 and 37 represents the desired constant resistivity zone which is produced by monitoring the drawing rate and the temperature as will be later described. After the drawing rate has come to a virtual standstill at point 37 so that no more crystal of resistivity 36-37 can be drawn, the resistivity is allowed to drop to a valuerepresented bypoint 38 by a sudden increase in drawing rate. The drawing rate and temperatures are again monitored so as to result in flatportion 38-39. After the drawing rate has again come to a virtual standstill the rate of drawing is allowed to remain constant so as to draw up the remaining portion of the ingot resulting in normal freezing curve Sid-4t Fig. 6B is" similar to Fig. 6A but represents a crystal having only one'flat resistivity zone. Increasing the pulling rateand decreasing the temperature to maintain the cross-section resulted in gradient 41-42. Zone 4-2-43 resulted from a normal monitoring of pulling and temperature rates as above described. Portion 43-4 represents the normal freezing curve in which the remainder of the ingot was drawn at any constant rate. I

1 Figs. 7A and 7B are graphical representations of crystals which have been doped. The coordinates are log of excess impurity concentration expressed in atoms per cubic centimeter against length expressed in inches. The

I log of excess impurity concentration in the negative or downward direction represents a p conductivity type of decreasing resistivity, while the positive portion of the vertical represents n conductivity type. The resistivity values may be determined from the excess impurity concentration values by use of the equation neg.

where p represents resistivity in ohm-centimeters, n equals excess impurity concentration expressed in atoms per cubic centimeter, e is the charge on the electron, while ,u is the mobility of the electrons or holes expressed in centimeters squared per volt second. Curve A of Fig. 7A represents the crystal which has been doped either with one large pellet or one burst of gas as later described. Zone 45-46 of this figure represents the constant resistivity section produced by monitoring the pulling rate and temperature. The pelletis dropped into the melt at point 46 resulting in sudden change in conductivity type 46-47 while section 47-43 represents that part of the crystal drawn from the remaining portion of the ingot. Although this portion of the curve appears to be flatter than the corresponding portions of Fig. 6A or Fig. 6B, both plotted in terms of resistivity against distance, it would appear similar to the normal freezing curve 39-40 if expressed on those coordinates.

Curve B in Fig. 7A represents a crystal with a P-N junction, produced by doping as in curve A except that by means of a controllable doping process using either multiple pellet or gradual gas doping the transition range is spread out from point 51 to point 52. Zone 49-50 again represents theconstant resistivity portion produced by the-normal monitoring process as will be described. The addition of a first pellet or burst of gas results in zone 50-51; zone 52-53 results from a similar addition of a pellet or burst of gas of opposite impurity type, while 53-54 is the normal freezing portion of the crystal resulting from drawing the remaining portion of the melt. Curve C of this same figure represents a crystal formed either through constant gas doping or multiple pellet doping resulting in a constant slope of curve 55-56.

Fig. 7B represents three illustrative crystals which have been doped during the drawing process so as to result in three different types of N-P-N junctions. Portion 57-58 of curve A represents the constant resistivity zone brought about as otherwise described, while 58-59 is produced by doping with a single pellet. Zone 60-61 represents doping with .a pellet of opposite semiconductivity type so as to bring the crystal back into the 11 region. ioint 61 need not be higher than zone 57-58 but in this specimen the crystal was brought back to a higher It level at this point so as to produce a higher excess of n-type impurity which is desirable in a good emitter. Section 61-62 could again be a normal freezing portion of the crystal. The breadth d of the p region is determined by the time lapse between doping steps 58-59 and 60-61.

Curve B of Fig. 7B is similar to curve B of Fig. 7A except that after the crystal has been brought into the p region it is brought back into the n region by a reversed doping procedure using a pellet or gas of opposite semiconductivity type. Therefore, section 63-64 represents the constant resistivity zone, while 64-65 represents doping with a first pellet or burst of gas. Section 65-66 represents controlled doping either withpellets or with controlled bursts. of gas so as to produce the transition region. Section 66-67 represents doping with a large pellet-or with a burst of gas. Afteran interval represented by d which may be controlled by varying pulling rate or time sequence the melt is doped with an impurity of opposite semiconductivity type'either by pellet or burst of gas to bring the crystal back to point 69. Point 69-70 represents controlled doping either by pellet orby a burst of gas. Point 70-71 represents doping in this example,

of equal but opposite semiconductivity type to that of,

zone 64-65, while 71-72 could again be :normal freezing zone.

Curve C of Fig. 7B represents controlled gas dopingor multiple pellet doping whereby the crystal is brought from point 73-74 either by constant gas doping or mul tiple pellet doping. Point 74-75 represents the region in which no impurity is added; This zone may -'or' may not be of constant resistivity type as desired. Zone 7576 is the reverse of 73-74 and represents doping of an opposite semiconductivity type brought about either by gradual gas doping or multiple pellet doping.

Fig. 8 is a plot of resistivity against distance of a single crystal of germanium containing two constant resistivity zones 7879 and 80-81 similar tothose of the crystal of Fig. 6A but containing two N-P-N junctions 7980 and 81-82 which were produced thermally and without the addition of any significant impurities by doping.

Although in theory, simply varying the pulling rates and varying the temperature as above described should result in a smooth, circular rod of controlled electrical characteristics, in fact, this was proved not to be the case due to a thermal gradient across the surface of the melt and consequent varying rates of crystallization on different portions of the forming rod. Not only were the surfaces of the forming rod caused to form in irregular fashion, but the electrical characteristics of the germanium were not kept constant even in cross-section. It was found that this could be overcome by rotating the crystal as it formed. Rotation rates of from 50 to 5,000 revolutions per minute are satisfactory and a rate of several hundred, as for instance 200 to 500, is to be preferred.

Although it was found that so rotating a crystal had sufiicient stirring action to result in a rod of circular cross-section, at first glance, it was noted that ring-like irregularities were forming on the surface of the rod, especially at the lower speeds of rotation. This indicated that the temperature gradient, although diminished, was still in evidence.

It was found that this irregularity could be eliminated by the addition of a pumping action. This was done by adding a vibrating member placed in such a position so as to alternately stretch and contract the cable to which was connected the forming crystal. Vibration rates of from beats per second at an amplitude of about 10 mils to about 500 beats per second at-an amplitude of about of a mil have been found to be satisfactory, although here as with the stated rotation rates, the values cited are only suggestive and do not represent absolute limits. There is no apparent reason why they cannot be exceeded in either direction. Both rates are limited on the low side by ineifectiveness and on the high side pri-' marily by equipment capability.

The crystals of the present invention may be formed from germanium ingots such as described in the copending application of J. H. Scaff and H. C. Theuerer, Serial No. 638,351, filed December 29, 1945, except that in order to minimize impurity content, the last portion ofthe ingot formed is cropped once and is remelted, refrozen and recropped as often as is required depending on the purity desired. Varying this initial charge will, of course, vary the electrical characteristics of the fina crystal.

. Pulling rates of from .0001 up to .006 inch per second are, habitually used. At rates higher than .006 inch per second, and, under some circumstances, even at this rate, a torsional strain results in the crystal causing twinning, which is generally considered'undesirable in single crystals. Fo'rzthis reason, it. ispreferred not to exceed about .003 inch per second. The lower value represents only a'pr'actical limit. As will be seenfrom Example 9 below, it is sometimes desirable .to reduce the rate of pull to a complete standstill. a

As above set forth, when varying the pulling ratepit is'necessary also to vary the temperature of the melt where it is desired to. maintain the cross-sectional diameterv of the forming rod constant. A typical schedule follows:

6 Although this monitoring may be done by hand, it has been found desirable to use an automatic programming control, several types of which are well known in the commercial chemical production field. Such a device is adjusted so as to vary pulling rates and temperature in accordance with the schedule above set forth. It should be noted that the tabulated temperatures represent measurements made at the wall of the crucible. Some time lag from the wall of the crucible to the melt is to be expected. 7

In order to maintain the lifetimes of the holes and conductivity electrons in the forming rod at the highest possible level, it is desirable to cool the rod as soon after it emerges as possible. This is done by means of water cooling coils and a constant fiow of cooling gas through the envelope. Any good thermally conducting, non-oxidizing and otherwise non-reactive gas will be satisfactory for this purpose. Hydrogen, helium and nitrogen have been found to be satisfactory in this case.

One of the most important features of the process of this invention is the flexibility with which semiconductive transition zones may be produced. This may be done in two ways. The most important consists of doping either with a solid or with a gas containing the desired significant impurity in amount sufficient to carry the material over to the opposite semiconductivity type. Solid doping of crystals formed from charges of about 50 grams utilizes pellets of from 1 milligram up to 50 milligrams of alloys of germanium together with the desired impurity, or of any compound of the impurity which will have the effect of injecting the desired impurity into the melt.

The most satisfactory doping elements are gallium and boron for N to P conversion and arsenic and antimony where P to N conversion is desired. For solid doping the elements may be added in elemental form or in the form of any alloy or compound which will result in the addition of the impurity to the melt. Examples are theoxides and germanium alloys. The trichlorides have been found to be satisfactory in gas doping. Again the amounts of impurity used either in solid doping or in gas doping will vary with the size and purity of the ini' tial charge, the amount of charge left in the crucible and the results desired. As has been seen, pellet doping may be carried out with the pellets in solid form or in molten form, and P-N or N-P-N junctions may be tailored to have any desired characteristics by varying in solid doping by the size of the pellets, composition of pellets, and time sequence of addition of said pellets; and in gas doping by the rate at which the gas is allowed to come in contact with the melt, and on whether or not it is passed in in bursts or gradually. With either type of doping, the junctions may be varied by, controlling pulling rates and the temperature of the melt.

Another method of producing P-N and N-P-N junctions without addition of impurities is by the simultaneous variation of temperature and pulling rate. Here it is necessary to choose a starting material such that heat treatment at about 980 C. will produce a swing of semiconductivity type from N to P. An example of such a material is a germanium of a resistivity of at least 10 ohm-centimeters.

It may be noted here that a twin boundary maybe formed eitherby using two seed crystals side by side, or by starting with a seed containing a twin boundary;

Where it is desired to produce a maximum amount of .constant resistivity material, and after, through monitoring, the rate of pull has been reduced to a virtual standstill so as to make possible only a negligible increase increasing the pullingrate to some maximum value suddenly and then monitoring its decrease as wasgdonein' the first zone. Z Since thenormal freezing curve on co ordinates of resistivity against length is generally changing slope in this zone, it has been found preferable to decelerate the rate of pull more rapidly during the formation of this second portion of the crystal.

Although crystals may be grown using seeds of any crystalline orientation, it has been found preferable to either place the seed in such a position or to grind it in such a manner as to approximate the [100] or the [111] orientation.

A general description of a typical process for the production of a single crystal of germanium containing two constant resistivity zones follows: The seed crystal, cut from a rod produced by this process, is first cleaned and mounted in the chuck of the spindle. The charge, a one hundred gram germanium ingot such as produced by the process of copending application Serial No. 638.351. filed December 29, 1945, is loaded in the crucible. and the crucible assembly is mounted in position. The induction coil is turned on and the charge is melted at a temperature of about 980 C. The seed is immersed in a melt to a depth of about mils and is left in this position sutliciently long to result in thermal equilibrium of the interface. A period of about five minutes has been found to be satisfactory. With the seed crystal being pulled from the melt at an initial rate of about .003 inch per second, the rotating mechanism is turned on and the vibrator is set into play. The temperature is dropped to about 935 C. and the pulling rate is kept constant for a period of three minutes. During this three-minute interval, the diameter of the crystal has increased to about M; of an inch. Thereafter, where it is desirable to form a single crystal of uniform diameter, the rate of pull and the temperature are programmed as follows:

Temperature Increased to (Degrees Centigrade) Pulling Rate Decreased to (Inches per Second) The process thus far will result in approximately a 1 /2 inch constant-resistivity zone of about ohm-centimeters. When the pulling rate has been reduced to a virtual standstill, the resistivity is dropped down to about half the level of the first zone by increasing the pulling rate to .003 inch per second suddenly. During the formation of this second Zone, the pulling rates and temperatures are programmed exactly as above set forth, except at approximately double the deceleration rate. The rate of pull is then increased to any value and allowed to remain constant until the remainder of the melt has been drawn off.

If, at any time during the formation of the germanium crystal, it is desirable tof orm P-N or N-P-N boundaries, this may be done by pellet or gas doping by any one of the methods heretofore described. Typical examples of P-N and N-P-N boundaries that may be tailcred to specification by varying pellet size and/or frequency, by varying gas doping rates, and/or by heat treatment, may be seen in Examples 3 through 9.

Examples of how the above processes may be modified to produce nine difierent types of crystals corresponding to the curves of Figs. 6A, 63, 7A and 7B and 8 follow: Example 1 003 inch per second. melt temperature was dropped from an initial value of 980 C. to a value of about 935 (3., the temperature change taking place in .a period of about ten seconds. The pulling rate was maintained constant at the rate of about .003 'inch per second for about three minutes during which time the diameter of the initial portion of the crystal increased to about A; of an inch. The pulling rate and temperature were then programmed as set forth in the general description of the process above set forth. After the pulling rate had been reduced to a virtual standstill, the pulling rate was then increased suddenly to about .003 inch per second and once again the pulling rates and temperatures were programmed except at twice the deceleration rate. The remaining portion of the melt was then allowed to be drawn off at a rate of .003 inch per second. A crystal so produced is here referred to as a two-step crystal. This particular specimen contained two constant resistivity zones, one of about 10 ohm-centimeters running for a length of about an inch and a half, while the second zone had a resistivity of about 5 ohm-centimeters and extended for about A of an inch. Where it is desirable to produce semiconductive material of constant resistivity, and where it is possible to use materials of both resistivity levels, it is this modification of the invention which produces the highest degree of efficiency.

Example 2 Temperature Drawing Rate (Inch-es per Second) (Degrees Centigrade) In the example shown in Fig. 6B, the pulling rate was varied as shown over a period of from two to three minutes. The pulling rates were then decreased and the temperature was then increased, as scheduled in Example 1. This monitoring sequence took about fifteen minutes and resulted in a constant resistivity zone of about 3 /2 ohm-centimeters over a length of about 1% inches. The remainder of the melt was then run ed at any desired speed resulting in the normal freezing portion 43-44 of Fig. 6B. The rate of pulling used in this final portion has not been specified since to date this portion of the crystal has not found use in any transistor or rectifier device and is simply reused as part of the load of another run.

Example 3 The seed was mounted, the charge was loaded and the equipment was set into operation, as above described. The pulling rate was allowed to drop to about .0001 inch per second, while monitoring the temperature and pulling rates as above scheduled. The pulling rate was allowed to remain constant at this value for about seven or eight minutes and all the desired impurity was added at this point in the form of. a pellet. In order to obtain a crystal of electrical characteristics such as indicated by curve A of Fig. 7A, a lump sum of impurity which would result in a total change in the crystal of about 10 atoms per cubic centimeter was added. For an ingot of about 50 grams, this amounted to a pellet of approximately 10 milligrams of gallium dioxide. An equivalent amount of boron in the form of a germanium alloy could have been.

used. This. could. alsohavebeen done with one blast of a doping gas such as boron trichloride or gallium trichloride, containing approximately twice as much of the desired impurity, since only about half of the gas goes into solution with the melt. The pulling rate was then raised to about .001 inch per second and the remainder of the melt was drawn out. The materials produced by such a doping process are usable in voltage regulating devices due to their low back voltage.

Example 4 Multiple pellet doping or gradual gas doping was resorted to in order to obtain a crystal with characteristics corresponding to curve B of Fig. 7A. The process was put into operation exactly as set forth in Example 3, and after the pulling rate had been allowed to continue at .0001 inch per second for several minutes, one pellet consisting of about 2 milligrams of a gallium-germanium alloy containing about .35 percent gallium was dropped into the melt by any of the doping methods before described. This was done in order to bring the excess concentration of impurity to about 10 atoms per cubic centimeters. Three 2-milligram pellets of a gallium-germanium alloy containing about .05 percent gallium, spaced about one second apart, were dropped into the melt. Finally, a second S-milligram pellet of galliumgermanium alloy, containing about 1.4 percent gallium, was dropped into the melt. The remainder of the melt was then drawn out at a constant rate of about .001 inch per second. It should be noted that the breadth and electrical characteristics, such as the back voltage of the P-N boundary so produced, may be tailored as desired by varying any one of the following: the size of the pellet, the composition of the pellet, the number of pellets used, the time sequence and the drawing rate. It is also possible to produce a crystal corresponding to this curve by using gas doping. The first doping step could be by means of a burst of about 10 cubic centimeters of hydrogen gas containing in the neighborhood of X10 atoms per cubic centimeter of gallium trichloride or any other gas containing the desired impurity of sufficient vapor pressure. The second doping step would consist of passing into the melt a doping gas at a constant rate over about one minute and containing suflicient impurity such that the level of the forming crystal is varied a total of 5x10 atoms per cubic centimeter. In order to bring the curve down to its final resistivity level, another short burst of about cubic centimeters of hydrogen containing about the same amount of impurity is allowed to come in contact with the melt. Crystals so produced contain excellent P-N junctions and may be tailored so as to meet the exacting specifications of any desired rectifier or transistor structure.

Example 5 In order to produce a crystal corresponding with curve C of Fig. 7A, the process was set into motion as above described, except that the initial pulling rate was .001 inch per second. The diameter of the crystal was built up to about M; of an inch by dropping the temperature to about 935 C. whilethe pulling rate was maintained constant for from seven'to eight minutes. Hydrogen containing suflicient impurity to vary the excess impurity level of the forming crystal about 5 X10 atoms per cubic centimeter of the desirable impurity in any utilizable gaseous form was then added at a constant rate over a period of from three to four minutes always keeping the pulling rate constant at about .001 inch per second. The only requirement for the gaseous compound here used other than that it be of sufiicient vapor pressure is that it should not be one which will impair the semiconductive properties of the crystal form and that it be non-corrosive to the apparatus used. As above set forth, the trichlorides of gallium, and boron are satisfactory. It is conceivable that such a crystal could be produced using a very large number of very small pellets, although as Curve A of Fig. 7B represents a doping process similar to that of curve A of Fig. 7A, except that after the semiconductive type was brought into the p region a pellet containing donor impurity was added. Such a run follows. After the pulling rate was maintained constant at .0001 for several minutes, and again with a SO-gram starting charge, a S-milligram pellet of a gallium-germanium alloy containing about 1.4 percent gallium was added to the melt. After a time lapse of from five to ten seconds, a l0-milligram pellet of arsenic trioxide or pure arsenic was added (the reason that an arsenic-germanium alloy was not used here is that the solubility of arsenic in germanium is so slight as would make the alloy pellet prohibitive in size). After the second doping step, the drawing rate of about .0001 inch. per second was maintained for about eight minutes. After this period the drawing rate was increased to about .001 inch per second and the remainder of the melt was drawn out. Materials formed by this process have been successfully used in N-P-N transistor structures such as those described in patent application of W. Shockley, Serial No. 34,423, filed June 26, 1948. Varying the breadth of the key region represented by d on the curve will affect the frequency cut-off of the final transistor structure. The smaller at, the higher the frequency cut-off. The breadth of this region may be varied by varying the time interval between the two pellets and/or by varying the drawing rate. The specimen of curve A of Fig. 7B was brought to a higher resistivity level after the key region, since this specimen was to be used as a transistor and since this zone was to be used as an emitter. An increased number of excess impurity atoms are desirable in that zone of a transistor which is to be used as an emitter.

Example 7 The initial portion of the crystal represented by curve B of Fig. 7B was produced by the same doping amount and sequence as crystal represented by curve B of Fig. 7A. After the last dopingstep, however, the rate of pull was maintained constant at .0001 inch per second for about five minutes, whereupon the same doping steps were repeated in inverse order using donor rather than acceptor impurity. Such materials also find use in N-P-N transistor structures.

Example 8 The crystal represented by curve C of Fig. 7B was produced by constant gas doping or, alternately, multipellet doping at twice the rate and half the time used in Example 5. After a time lapse of about fifteen minutes, the same gradual doping process was repeated, using impurity of opposite semiconductive type.

Example 9 To form a crystal corresponding to the curve of Fig. 8, it was necessary to start with the initial portion of the drawn rod at a resistivity level of at least 10 ohm-centimeters and preferably at least 20 ohm-centimeters. After the process had been set into motion as described in Example 1, and after the first constant resistivity zone had been formed at a level of about 20 ohm-centimeters, the rate of pull was reduced to zero, while the temperature was maintained at about 980 C. for a period of several minutes, whereupon a second constant resistivity zone was formed exactly as in Example 1, the level of the second zone, in this instance, being in the neighborhood of 10 ohm-centimeters. The pulling was again stopped, and the temperature was allowed to remain constant at about 980 C. for several minutes, thereafter the remainder of the melt was drawn out at some constant rate.

Zone 7778 represents that portion of the crystal drawn at a constant rate of .003 inch per second, during which time the diameter was allowed to build up to some desired value (in this specimen 'V; of an inch). Zone '78-'79 represents a constant resistivity zone produced by decreasing the rate of pull from .003 inch per second down to 0 inch per second, and a simultaneous increase of the temperature of the melt of from 935 C. to 980 C. Zone 7980, which in this specimen was about of an inch in thickness, resulted from keeping that portion of the crystal in contact with the melt for a period of about five to ten minutes, resulting in thermal conversion from N to P type. Zone Sit-81 represents a second constant resistivity zone formed in the same manner within the same monitoring limits used in the formation of the first constant resistivity zone but at double the deceleration rate. 'P-zone 81-82 was produced by a second thermal conversion in the same manner and for the same time as thatused in zone 7980, while 8233 represents that portion of the crystal which followed its normal freezing curve, resulting from drawing out the remainder of the melt at a constant pull rate of .001 inch per second. Material containing such N-P-N transition zones is produced without the use of any doping mechanism. There is, of course, no reason why this thermal conversion step may not be combined with any one of the doping methods above described so as to give any desired flexibility in the formation of the N-P-N transition areas. Crystals pro duced by the thermal conversion process above outlined have electrical characteristics making them suitable for use in devices such as those disclosed in the application of W. Shockley, Serial No. 34,423, filed June 26, 1948, now abandoned.

Although most of the description is in terms of germanium it is to be understood that the described process works equally well with other semiconductive materials. Silicon, for example, has been used and the crystals produced from this material have been found to possess excellent properties.

What is claimed is:

1. The method of producing a crystal of normally solid semiconductive material containing a significant impurity having a l of less than 1 comprising inserting a seed into a melt of the said semiconductive material containing the said significant impurity, and withdrawing the said seed at a decreasing withdrawal rate such as to substantially compensate for the resulting increasing concentration of the said significant impurity in the melt and such as to produce a crystal manifesting substantially uniform resistivity over that portion of the crystal produced While decreasing the withdrawal rate, the rate of withdrawal being at all times such as to maintain a solid-liquid interface between the crystal and the melt, 1" being defined as the ratio of impurity concentration in the solid to the impurity concentration in the liquid with the two phases at equilibrium.

2. The process of claim 1 in which the temperature of the melt is progressivel increased as the withdrawal rate is decreased in such manner as to maintain the crosssection of the crystal substantially constant.

3. The process of claim 1 in which the crystal is 1'0- tated during withdrawal.

4-. The process of claim 3 in which the normally solid semiconductive material is germanium.

5. The process of claim 2 in which the normally .solid semiconducti'v'e material is silicon.

References Cited in the file of this patent UNITED STATES PATENTS 1,353,571 Dreibrodt Sept. 21, 1920 1,531,784 Hazelett Mar. 31, 1925 1,921,934 Lewis Aug. 8, 1933 2,091,903 Baggett et al. Aug. 31, 1937 2,188,771 W'elch Jan. 30, 1940 2,514,879 Lark-Horowitz ct al. July 11, 1950 2,530,110 Wcodyard Nov. 14, 1950 2,683,676 Little July 13, 1954 FOREIGN PATENTS 706,858 Great Britain Apr. 7, 1954 OTHER REFERENCES Journal of Applied Physics, vol. 6, 1935, pages 111- 116.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US1353571 *27 Jun 191421 Sep 1920Elektrochemische Werke GmbhMethod of and apparatus for forming large crystals
US1531784 *13 Dec 192131 Mar 1925Cleveland Trust CoSheet metal
US1921934 *30 Aug 19308 Aug 1933Blair Strip Steel CompanyArt of annealing
US2091903 *4 Nov 193531 Aug 1937Hughes Tool CoMethod of treating tungsten carbide
US2188771 *7 Oct 193630 Jan 1940Firestone Tire & Rubber CoElectrode
US2514879 *13 Jul 194511 Jul 1950Purdue Research FoundationAlloys and rectifiers made thereof
US2530110 *2 Jun 194414 Nov 1950Sperry CorpNonlinear circuit device utilizing germanium
US2683676 *13 Jan 195013 Jul 1954Bell Telephone Labor IncProduction of germanium rods having longitudinal crystal boundaries
GB706858A * Title not available
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US2822308 *29 Mar 19554 Feb 1958Gen ElectricSemiconductor p-n junction units and method of making the same
US2835614 *30 Nov 195520 May 1958Raulaud CorpMethod of manufacturing crystalline material
US2852420 *28 Jun 195616 Sep 1958Rauland CorpMethod of manufacturing semiconductor crystals
US2876147 *21 May 19563 Mar 1959Siemens AgMethod of and apparatus for producing semiconductor material
US2893847 *18 Feb 19557 Jul 1959Siemens AgApparatus for preparing rod-shaped, crystalline bodies, particularly semiconductor bodies
US2935385 *25 Mar 19553 May 1960Texas Instruments IncApparatus for manufacturing semiconductor materials
US2935478 *30 Aug 19563 May 1960Gen Electric Co LtdProduction of semi-conductor bodies
US2961475 *29 May 195722 Nov 1960Rca CorpSolid-state charge carrier valve
US2975036 *5 Oct 195614 Mar 1961Motorola IncCrystal pulling apparatus
US2979386 *2 Aug 195611 Apr 1961William ShockleyCrystal growing apparatus
US2981687 *26 Mar 195925 Apr 1961British Thomson Houston Co LtdProduction of mono-crystal semiconductor bodies
US2984549 *21 Jun 195716 May 1961Clevite CorpSemiconductor product and method
US2998335 *31 Jan 195729 Aug 1961Telefunken GmbhMethod and apparatusfor growing single crystals from molten bodies
US2999776 *9 Jan 195612 Sep 1961Siemens AgMethod of producing differentiated doping zones in semiconductor crystals
US3025191 *10 Feb 196013 Mar 1962Raytheon CoCrystal-growing apparatus and methods
US3074785 *22 Aug 195622 Jan 1963Siemens AgApparatus for pulling crystals from molten compounds
US3093456 *2 Sep 195811 Jun 1963Texas Instruments IncMethod for recovery and reuse of quartz containers
US3102061 *5 Jan 196027 Aug 1963Texas Instruments IncMethod for thermally etching silicon surfaces
US3108073 *26 Oct 196122 Oct 1963Grace W R & CoProcess for doping semiconductive bodies
US3119778 *20 Jan 195928 Jan 1964Clevite CorpMethod and apparatus for crystal growth
US3134700 *20 Apr 196026 May 1964Siemens AgDislocation removal by a last pass starting at a location displaced from the original seed into the grown crystal
US3226268 *2 Mar 196428 Dec 1965Maurice G BernardSemiconductor structures for microwave parametric amplifiers
US3259468 *2 May 19635 Jul 1966Monsanto CoSlim crystalline rod pullers with centering means
US3359077 *25 May 196419 Dec 1967Globe Union IncMethod of growing a crystal
US4270972 *31 Mar 19802 Jun 1981Rockwell International CorporationMethod for controlled doping semiconductor material with highly volatile dopant
US638717731 Jul 200014 May 2002Forschunginstitut Fur Mineralische Und Metallische Werkstoffe Edelsteine/Edelmetalle GmbhMethod for manufacturing a segmented crystal
DE1188555B *10 May 196011 Mar 1965Wacker Chemie GmbhVerfahren zur Herstellung hochreiner kristalliner Koerper aus Nitriden, Phosphiden oder Arseniden der III. Hauptgruppe des Periodensystems
DE1211592B *19 Apr 19603 Mar 1966Westinghouse Electric CorpVerfahren zur Herstellung eng toleriert dotierter Siliciumstaebe
DE1211593B *27 Apr 19603 Mar 1966Wacker Chemie GmbhVerfahren zur tiegelfreien Herstellung hochreiner, elektrisch halbleitender, kristalliner Verbindungen
DE1227874B *10 Apr 19593 Nov 1966Itt Ind GmbhVerfahren zum Herstellen von n-dotierten Siliciumeinkristallen
U.S. Classification117/22, 117/21, 252/62.30E, 117/936, 438/925, 23/301, 117/932
Cooperative ClassificationY10S438/925
European ClassificationC30B/