US3198929A

US3198929A - Electric control apparatus for zone melting of semiconductor rods

Info

Publication number: US3198929A
Application number: US243973A
Authority: US
Inventors: Stut Hans
Original assignee: Siemens AG
Current assignee: Siemens Schuckertwerke AG; Siemens AG
Priority date: 1961-12-07
Filing date: 1962-12-06
Publication date: 1965-08-03
Anticipated expiration: 1982-08-03
Also published as: CH407958A; GB1032217A; DE1209551B

Description

Aug. 3, 1965 H. sTuT 3,198,929

ELECTRIC CONTROL APPARATUS FOR ZONE MELTING OF SEMICONDUCTOR RODS Filed Dec. 6, 1962 5 Sheets-Sheet 1 FIG. 1

Aug. 3, 1965 s u 3,198,929

ELECTRIC CONTROL APPARATUS FOR ZONE MELTING OF SEMICONDUCTOR RODS Filed Dec. 6, 1962 5 Sheets-Sheet 2 A E l] E F B N B FIG. 2

03 L E F E H B Aug. 3, 1965 sT 3,198,929

ELECTRIC CONTROL APPARATUS FOR ZONE MELTING OF SEMICONDUCTOR RODS Filed Dec. 6, 1962 5 Sheets-Sheet 5 FIG. 4

Aug. 3, 1965 H. STUT 3,198,929

ELECTRIC CONTROL APPARATUS FOR ZONE MELTING OF SEMICONDUCTOR RODS Filed Dec. 6, 1962 5 Sheets-Sheet 4 E w QE '9 a I ll fifi v. a 1 E v N z 1 5 E; J v. E O m m I h u 2 u E a .W mm ,H. v. E t m M I! q 7- h V. IE m w s; 1 N e, u 1 E m m u n q 5 l.|| x .5 I a, H m m -25 x H m w INF WK )2' N z fiv 5 1 a: r b n NT E N sl 1 a H wwmH lg N3: [HI m= u N: u u 2 5 r. 2N mm H m A Y- N N a; 1 z a; 5;

Aug. 3, 1965 H. STUT 3,198,929

ELECTRIC CONTROL APPARATUS FOR ZONE MELTING OF SEMICONDUCTOR RODS Filed Dec. 6, 1962 5 Sheets-Sheet 5 United States Patent (Ilaims. of. 219-10.

My invention relates to electric apparatus for controlling the zone melting of rod material, particularly semiconductor rods to be employed, for example, as an intermediate product in the manufacture of semiconductor junction diodes, transistors, semiconductor switching devices of the p-n-p-n type and others, Such semiconductor rods amenable to zone melting may consist of germanium, silicon, gallium arsenide, indium antimonide or various other intermetallic compounds.

Zone melting is employed for various purposes. it may serve for purifying a semiconductor rod of undesired impurities, for uniformly distributing a dopant in the rod (zone levelling), or for converting a polycrystalline rod to a monocrystal with the aid of a crystal seed fused to one end of the rod. One and the same zone-melting operation may simultaneously perform two or more of such functions.

Some zone-melting operations require a considerable amount of attention and skill on the part of attendant personnel. This is the case particularly when converting a polycrystalline rod into a monocrystal with the aid of a monocrystalline seed. As a rule, the seed has a smaller diameter than the rod to which the seed is fused, because a seed of smaller diameter can be more easily and more reliably produced in perfect monocrystalline constitution and with fewer crystal-latticed dislocations than are apt to occur in larger crystals, However, the use of relatively small crystal seeds may render it difiicult to reliably perform the zone-melting operation at the locality where the small cross-sectional area of the seed merges with the larger area of the rod proper.

It has been necessary, therefore, to provide for corresponding manual control of the zone-melting equipment in order to attain a satisfactory transfer of the molten zone from the thin seed to the thicker rod during a zonemelting pass. This requires varying the amount of heating power supplied from the induction heater to the molten zone by correspondingly controlling the electric generator that furnishes current to the primary winding of the induction heater producing the molten zone. Simultaneously, the necessary change and adaptation of the diameter or cross section is obtained by shifting one of the two rod holders in which the respective ends of the rod-seed assembly are mounted, toward or away from the other holder for compressing the molten zone to increase its diameter, or axially stretching the zone to reduce its diameter. This must be done in accordance with the particular shape of the rod-seed assembly being used, thus requiring a control in accordance with pre-existing diameter values. On the other hand, it is also necessary to vary the supply of heating energy to the molten zone in order to maintain the zone at the desired temperature regardless of the change in zone diameter along the rod. Such maintenance of given temperature limits in the molten zone is desirable for proper coordination of the weight to the diameter and the temperature-dependent surface tension of the molten material, because it is the surface tension which prevents the molten zone from flowing apart.

These various requirements are difiicult to satisfy dur- 3,198,929 Patented Aug. 3, 1965 ing zone-melting operations and require attendance of considerable training and skill.

It is an object of my invention to make the performance of such zone-melting operations independent of particular attention on the part of personnel during the course of the melting operations, particularly their most critical stages. More specifically, it is an object of my invention to more reliably alrord attainment of predetermined cross sections or mergers between such cross sections during performance of zone-melting operations in rods, while simultaneously maintaining a'predetermined melting temperature in the molten zone at any stage of operation.

According to my invention, relating to apparatus for controlling the zone-melting operation axially along a semiconductor rod, I apply a digital control to the heating energy imparted to the melting zone and also to the cross section or diameter of the rod, in dependence upon the travel of the zone-melting heater along the rod or rodseed assembly. heating energy and the digital control of the rod cross section in accordance with respective programs so that the increase or decrease of the heating energy and the conjoint increase or decrease in cross-sectional diameter take place in increments depending upon passage of the zoneheating device through one or a given number of steps along its travel relative to the semiconductor rod.

According to another feature of my invention, a threaded feed spindle or other device for displacing the zone-heating device along the rod is employed for controlling the issuance of digital counting pulses, for example to a binary chain of counters. When using the transporting spindle of the zone heater in this manner, it is preferable to connect with the spindle a rotating member to perform a given number of control actions per rotation, thus issuing a corresponding sequence of control pulses to the input circuit of a binary counter chain. Since one rotation of the threaded spindle corresponds to a given screw pitch or axial travel of the heater device, the number of pulses transmitted to the binary counters is indicative of respective consecutive steps of heater travel.

According to another feature of my invention, the above-mentioned chain of binary counters is composed of a first portion comprising a number of counting stages, and a subsequent, second portion which also comprises a number of sequential counting stages. The first counting stage of the second portion in the chain feeds pulses .to a conversion unit which converts the incoming digital coding of the travel points into a continuous travel image in accordance with the natural number sequence and thereby puts into effect the particular program storer or memory with which the conversion unit is connected by a corresponding number of leads. Thev program storer or memory system consists, for example, of a cross-bar distributor or coordinate-type selector which possesses two systems of mutually crossing groups of contact rails, although other suitable mechanical memory devices, for example punch-card, punch-tape or magnetic memory devices, are also applicable. Connecting contacts can be plugged or screwed into the cross bars at the intersection points of each distributor, thus interconnecting the corresponding two rails in each memory system in accordance with the preselected program.

According to a preferred embodiment of my invention, the above-mentioned cross-bar distributor is used for issuing the travel-dependent signal through connecting leads to the input circuit of AND-gates. Simultaneously these AND-gates receive through respective connecting leads the corresponding counting pulses from the counting stages in the first portion of the counter chain. Each of these AND-gates issues a different number of pulses per rotation of the heater transport spindle depending I further preset the digital control of the o upon the programmed datum value preset in the cross-bar distributor, and each of these issuing pulses represents or controls a unit step with respect to the change in heating energy or change in rod diameter. The output pulses of the AND-gates are supplied through a common output lead to a digital-analog converter of the heating-energy control system and to a corresponding digital-analog converter of the diameter control system. Consequently, the output circuit of the particular digital-analog converter provides an electrical control magnitude which corresponds to the desired absolute value of heating power to be supplied to the molten zone or to the absolute value of diameter or cross section to be adjusted in the molten zone of the rod.

For further describing and explaining the invention, reference will be made to the embodiment of a floatingzone melting apparatus and control system according to the invention illustrated by way of example on the accompanying drawings in which:

FIG. 1 shows schematically the essential features of the zone-melting apparatus together with a diagram of the electrical system.

FIGS. 2 and 3 are graphs explanatory of the system performance.

FIG. 4 is an electric circuit diagram of one of the various counting stages according to FIG. 1; FIG. 5 is a circuit dagram of details adjacent to one of the counting stages according to FIG. 4 in the system of FIG. 1; and FIGS. 6, 7. and 8 are circuit diagrams of other components in the system of FIG. 1.

According to FIG. 1, a semiconductor rod to be zonemelted has a crystal seed 2 fused to one end. The two ends of the rod-seed structure are clamped in axially spaced holders in the conventional manner. One of the holders can be placed in rotation about the common axis relative to the other for uniform distribution of the molten material in the molten zone la. The two holders and hence the ends of the semiconductor rod are displaceable axially one relative to the other by means of a control device 74 in order to stretch or compress the molten zone 1a in order to thereby maintain a constant diameter of the rod portion solidifying out of the molten zone. Details of mechanical devices for thus mounting the rod, for maintaining one rod end in rotation relative to the other, and for axially displacing one rod end toward and away from the other are known as such, for example from U.S. Patents 2,972,525 and 3,030,194 of R. Emeis, assigned to the assignee of the present invention.

The molten zone 1a in the semiconductor rod is produced by means of an inductive heater coil schematically shown at 3 whose design and mounting may also be in accordance with details known from the patents just mentioned. The heater coil 3 constitutes the primary winding of an electric induction system operating with a current whose frequency is sufficiently high for good regulation. Suitable, for example, is a frequency of 40 megacycles per second. The heater winding 3 is carried by a transporting device that comprises a vertically mounted, rotatable screw spindle 5 with a traveling nut 4 which is prevented from rotation so that it becomes displaced upwardly or downwardly by revolution of the spindle 5, thus moving the heater winding 3 axially along the rod 1.

The revolution of spindle 5 is simultaneously used for driving an electric contact device serving as a digital pulse transmitter. This device comprises a cam disc 6 with a number of lobes 7 for controlling a switch lever 8 cooperating with a fixed contact 9. The lobes '7, for example sixteen, are uniformly distributed over the cam periphery, only eight lobes being shown in FIG. 1 for simplicity. For each revolution of spindle 5, therefore, the contact is between lever S and fixed contact 9 made and broken sixteen times, and corresponding pulses are transmitted from a voltage source 8 to the input circuit of a chain 10 of binary counters whose individual, sequential stages are denoted by 11 to 1?. The internal circuitry of each of the stages will be described in a later place with reference to FIG. 4.

The counting stages 11 to 13 according to FIG. 1 constitute a first portion of the chain it). The active input lead of each

counting stage

11 and 12, as well as the corresponding active input leads of other digital counting stages mentioned hereinfater, are denoted by 121, whereas the identifying numeral 122 is applied to the output lead of the respective counting stages. The stages 11 to 13 in the first portion of chain 10 have their output leads 122 connected through respective leads 21 to 23 with respective AND-gates 25E to 27E for the heating-energy control system, and also with respective AND- gates 25D to 27D of the rod-diameter control system. Two AND-gates 24E and 24D are directly connected to the switch contact 9.

The counting stages 14 to 19, as well as any additional, not illustrated stages in chain 10, constitute together the second position of the counting chain 10. The respective output leads of these counting stages are connected to a conversion unit 28 composed of switching and gate circuits for converting the binary coding of the travel-step points into a continuous travel-path image or signal according to the natural numerical sequence. The output leads of the converter 28 are connected to respective vertical bars 29 to 45 of a double cross-bar distributor. The distributor further comprises a set of horizontal bars 47E to 50E of a program storer system for heating-energy control, and another set of horizontal bars 47D to 50D forming a program control system for diameter control. Each of the horizontal distributor bars 47E to 50E is connected to an input lead of respective AND-gates 2413 to 27E. Each of the horizontal distributor bars 47]) to StiD is analogously connected to an input lead of one of the respective AND gates 24D to 27D. The vertical bars 29 to 45, on the one hand, and the bars 47E to 59B and 47D to 50D, on the other hand, can be selectively connected with each other at the respective intersections by means of plugs or screws S in order to preset a travel-responsive program for power control as well as a travel-responsive program for diameter control.

Selected connections of this kind are indicated in FIG. 1 by two concentric circles surrounding at each locality S the intersection point between a vertical bar and one of the horizontal cross bars in each of the two programming systems. The cross-bar selector 45 in FIG. 1 thus exemplifies two preset programs for heat control and diameter control respectively, both programs being dependent upon the digital pulses issued in response to the travel of the heater along the rod.

The AND-gates 24E to 27E have a common output lead 51E connected to a digital-analog converter 52E. Analogously, the AND-gates 24D to 27D have a common output lead 51D connected to another digital-analog converter 52D. Each of converters 52B and 52D comprises counting stages a to f and a direct-voltage supply denoted by a source 68B and 68D respectively. Each of converters 52E, 52D further comprises a series-connected resistor 53E, 53D and a number of transverse resistors 54E to 59E and 54D to 59D respectively. The latter resistors are connected to the respective counting stages a to f in each converter. By virtue of these circuit components, the

output terminals

60E and 61E of converter 52E furnish an electric voltage whose magnitude is determined by the voltage division to which the direct voltage between input terminals 62B and 63E is subjected by the series connection of the resistor 53E with the particular transverse resistor or resistors 54E to 59E that are switched into operation at a time. Analogously, the output voltage between the output terminal 60D and 61D of the digital analog converter 52D is determined by the division to which the direct voltage impressed upon input terminal 62D and 63D is subjected by the series connection of the resistor 53D with the one or more of resistors 54D to 59D switched into activity at a time.

The respective output voltages thus furnished between terminals 60E61E and 6tlD-61D are taken as datum values and are compared with the actual value of the heating energy supplied to the semiconductor rod, and with the actual diameter of the rod, respectively. This comparison is made in

respective regulating devices

64 and 70 which take care of correcting the energy and diameter values to the datum value of the respective programs in the event the actual values depart from the programmed datum values.

In order to effect this regulating operation for the automatic stepwise control of the electric heating energy supplied by the inductive heater to the molten zone of the semiconductor rod, the regulator 64 is supplied through a connecting lead 65 with a converted measuring value indicative of the current which a high-frequency generator 67 supplies through a measuring instrument 66 to the heater winding 3. The lead 65 is shown connected to the current-measuring instrument 66 to indicate that the signal supplied to the regulator 64 is proportional to the heating current. Another lead 69 connects the regulator 64 with the high-frequency generator 67 and supplies a control magnitude corresponding to the departure of the pilot (actual current) magnitude supplied through lead 65 from the datum value set by the programming system and supplied from terminal 60E, so that the generator 67 is controlled to increase or decrease the heating current as is needed to have the actual amount of heating energy correspond to the program value.

For regulating the rod diameter to the programmed datum value, the latter value issues from the output terminals 60D, MD of the converter 52D to the regulator '76. The regulator 70 further receives through an input lead 72 a signal indicative of the actual diameter of the molten zone. This diameter pilot signal is produced in the illustrated embodiment by means of an inductive loop 71 which surrounds the semiconductor rod 1 at the location where the molten zone commences to freeze. The inductive loop 71 is connected to a transformer 72 or converter which issues a corresponding signal through the lead 72 to the regulator 70.

In regulator 70, the datum value from the program system is compared with the actual diameter value as manifested by the pilot signal, and a corresponding difference voltage is impressed through an output lead 73 upon the displacement control system 74 of the zonemelting apparatus. The system 74, in response to the regulating signal, displaces the lower holder of the rod upwardly or downwardly to cause a compression or axial expansion of the molten zone, thus regulating the diameter of the rod where it freezes out of the molten zone, to the value preset in the program system.

Before dealing with further details, an explanation of the technological advantages of the invention and a further description of its essential functiong will be helpful.

As mentioned, when a polycrystalline constitution by zone melting, the crystal seed used for this purpose and fused to one end of the rod is generally given a much smaller cross section than the rod itself. One or more zone-melting passes must be commenced at the seed crystal near its fusion junction with the semiconductor rod proper. Hence it is necessary to vary the heating energy supplied to the rod zone in order to melt it and heat it up to a given temperature, so that the supply of heat is adapted to change in cross section occurring along the rod-seed assembly. The shape of the assembly components prior to zone melting, as Well as the desired shape after zone melting, is predetermined. Hence these given shapes determine the program to be preset in the programming system 46 in dependence upon the travel of the heater relative to the rod.

In principle, a travel-responsive program could be 6. stored in such a manner that the control of the change in heat supply and the control of axial displacement for obtaining the desired diameter, take place continuously. This, however, would require an excessive amount of equipment. It is considerably more advantageous, according to the invention, to provide separate programming systems of digital performance which control the stretching-compressing motion of the seciconductor rod and the heat supply to the molten Zone, so that, depending upon the steepness of the transition from seed to rod, the absolute value of the rod diameter and of the heat supply to the molten zone are changed by a correspondingly greater or smaller number of individual, uniform steps.

This incremental change in heating energy supplied to the molten zone, and the change in diameter by unit steps will be further elucidated by the graphs shown in FIGS. 2 and 3. FIG. 2 indicates on the ordinate the amount of heating energy, and on the abscissa a number of successive, equal steps of travel traversed by the inductive heating device relative to the rod in the axial direction of the latter. FIG. 3 shows along the ordinate the diameter values of the molten zone, also versus incremental steps of heater travel indicated along the abscissa.

According to FIG. 3, a program is assumed to be desired according to which the diameter of the rod-seed assembly is to increase from the value D to a larger value D corresponding to the conditions existing, for example, at the junction between the seed and the rod proper. Since the volume to be converted to molten condition increases with increasing diameter, it is necessary to accordingly increase the electric heating energy supplied by the induction heater to the molten zone.

The transfer from the diameter D to the larger diameter D is performed in unit steps as indicated by the stepped curve train in FIG. 3. The control in these unit steps is dependent upon the steps of heater travel between the points A and B on the abscissa. Consequently, during each unit of travel, for example A-C, or C-D, or DE, traversed during continuous travel of the heater, a preselected number of individual steps in the direction of increasing diameter is added to the preceding absolute magnitude of the diameter.

As will be seen from FIG. 3, the merger curve exemplified in the graph possesses differen degrees of steepness in different portions. The larger the steepness is between each two steps of heater travel, the greater is the number of the individual steps of increase in diameter. Conversely, if .a decrease in diameter is needed, the corresponding performance would subtract for each step of heater travel a number of unit changes in diameter as correspond to the steepness of the curve for that particular travel step. For example, the travel steps AC and DE are of equal size, but the steepness of the merger curve in the range of step AC is considerably smaller than in the range of step D-E. Consequently, for satisfying the slight gradient of step A-C, a smaller number of individual unit steps is needed for correspondingly changing the diameter than in the range of step DE. Accordingly, the step A-C is represented in FIG. 3 to comprise only two unit steps of diametrical change, whereas the step DE comprises four such unit steps. The particular number of these unit steps is predetermined over the entire zone-melting travel of the heater along the rod by the setting chosen in the programming system 46 (FIG. 1).

As mentioned, the change in diameter must be accompanied by a travel-responsive change in supply of heating energy. Thus, according to FIG. 2, the heating energy is changed from the value E to the higher value E of the above-mentioned travel distance from point A to B. This is done incrementally in such a manner that whenever the heater coil passes through one unit step of its travel, a corresponding, preselected number of unit changes in heating energy is effected, depending upon the steepness of the transition between the absolute energy value at the beginning and the value at the end of tie travel step, thus correspondingly increasing, or reducing as the case may be, the tot-a1 amount of heat supplied to the molten zone. This control performance will again be exemplified by comparing the travel steps A-C and DE with each other. The program curve shown in FIG. 2 shows that during the travel step A-C of the inductance heater the characteristic curve has a relatively slight steepness in comparison with the steepness during travel step D-E. Therefore, as exemplified, four individual unit steps of change in heating energy are applied during travel step AC, but eight unit steps are effective during travel step DE. In this manner, a proper adaptation of the change in diameter and the correlated change in electric heating is obtained as required for maintaining the desired median temperature value in the molten zone of the semiconductor rod.

in the circuit system exemplified by FIG. 1, four successive counting stages 11 to 14, operating on a binary code, are employed for response to the switching of the pulse transmitter contacts 8, 9, in dependence upon each individual travel step of the inductance heater. Consequently, after each occurrence of 24: 16 pulses, which are registered in the stages 11 to 14, a pulse is effective in the second portion of the counting chain 11? to act upon the counting stages 15 and so forth. Since the sixteen pulses counted by the stages 11 to 14 occur within a single travel step to be processed in the programming device 4-6, apparatus according to the invention as exemplified by the illustrated embodiment aiiords issuing sixteen individual steps upon the energy controlling chain 52E and upon the diameter controlling chain 52D. That is, the program curves as exemplified in FIGS. 2 and 3 may comprise up to 16 steps of heat or diameter change for an individual step of Y heater travel.

FIG. 4 shows by way of example the circuit of a bistable flip-flop stage applicable in each of the counting stages 11, 12, 13, 14, 15 etc. as well as in the counting stages 52Ea to f and 52Da to f. In FIG. 4 the

terminals

111, 112, 121, 122 correspond to the equally denoted leads of the counting stage 525a, for example; and the lead 107 in FIG. 4 corresponds to lead 17 in FIG. 1. Terminals 1%9 and 1113 indicate the positive and negative terminal or lead of a direct-current supply of constant voltage, not shown in FIG. 1.

The flip-flop stage according to FIG. 4 comprises two

transistors

101 and 102 for example of the p-n-p type. An ohmic resistor 103, 164 is connected in series with the collector of each transistor 1111, 192. The collector of transistor 1111 is further connected through an ohmic resistor 105 with the base of transistor 1112. The collector of transistor 102 is analogously connected through a resistor 196 with the base of transistor 1tl1.

Capacitors

124 and 125 lie across respective resistors 1&5 and 1116. The common emitter lead 1117 of the transistors is attached to the positive terminal 1119 of the direct-voltage source. The negative terminal 110 of the same directvoltage source is connected to the common collector lead 108. The signal terminal 111 is attached to the base of transistor 1&1, and the signal terminal 112 to the base of transistor 102. A capacitor 113, two opposingly poled

diodes

114 and 115, as well as a capacitor 115 lie all in series between the respective bases of transistors 1 31 and 1112. The lead 117 joining capacitor 113 with diode .114 is connected through an ohmic resistor 118 with the collector of transistor 1112. The corresponding lead 119 between capacitor 116 and diode 115 is in connection with the collector of transistor 1121 through an ohmic resistor 124). The signal terminal 121 is connected between the two diodes 114 and 1115. Leads and

terminals

122, 123, constitute the output mmebers of the flip-flop circuit.

The operation of the bistable circuit is as follows.

Assume that at the moment under observation the transistor 1131 is conductive (turned on) and simultaneously the transistor 1112 is blocked (turned cit). Then a control current flows from the pulse pole 109 of the voltage source through the emitter-base path of transistor 1111, resistor 1%, resistor 104 and lead 103 to the minus pole 110. Simultaneously, a working current flows through the emitter-collector path of the same transistor 1111 and the resistor 1113, to lead 108 and minus pole 116. In this condition, the resistance of the emitter-collector path is virtually zero so that the output lead 122 with terminal A has substantially the same potential as the emitter lead 1%7. Since at this time the emitter-collector path of transistor 1112 is blocked, the other output lead 123 and its terminal A now have a potential determined by the potential of lead 107 minus the voltage drop produced in resistor 1% by the control current flowing through the emitter-base path of transistor 1191. The resistance ratio of resistors 1% and 104, as well as that of resistors and 1113, is very large. Hence lead 123 has virtually the potential of the minus pole 110.

When now a signal is applied to the signal terminal 112, this signal constituting a negative potential relative to the plus pole 109 of the voltage source, a control current will flow through the emitter-base path of transistor 192. As a result, the transistor 102 starts conducting in its emitter-collector path. This raises the potential of the output lead 123 virtually to the potential of lead 197, and the negligible voltage now obtaining between emitter and collector of transistor 102 can no longer drive suiticicnt current through the emitter-base path of transistor 191 and resistor 1116 to keep the transistor 1191 turned on. In other words, as soon as transistor 1112 is turned on by a signal at terminal 112, the transistor 192 is turned 011. As a result, the output lead 122 assumes the potential of the minus pole 111 in the same manner as was previously the case for the output lead 123 when transistor 1132 was turned off. In this condition, the lead 123 has zero potential, and the lead 122 has a finite potential relative to the plus pole 1119, determined by the voltage of the source connected between terminals 189 and 111 The circuit can be switched to its opposite stable condition by applying a corresponding negative signal to the signal terminal 1111. Thereafter the output lead 123 furnishes an output signal of a predetermined voltage value relative to plus pole 1119, whereas the output lead 122 again assumes zero potential relative to the same pole 10?.

In summary, according to the mode of operation so far described, the triggering of the bistable flip-flop circult from one to the other stable condition is effected by appyling a negative potential to the base of the one transistor that is turned ofr at a time so that in this manher a current will flow through its emittter-base path. This particular mode of operation is employed in the counting stages 52Ea to f and 52Da to f in conjunction with additional circuit components described hereinafter with reference to FIG. 5.

However, the same bistable switching circuit can also be operated in a different mode according to which sequential pulses are applied to one and the same control lead with the eflect that the circuit will trigger from its stable condition then occupied to the other stable condition. This particular mode of operation is employed in the counting stages of the chain 10 according to FIG. 1. It requires using the control terminal 121 and involves the further circuit components described above, namely the

capacitors

113, 116 and the

diodes

114, 115.

For explaining the latter mode of operation, again assume that the transistor 161 is turned on and the transistor 102 is turned oif. As soon as this condition wa initiated, current flowed from the plus pole 109 through emitter and base of transistor 101, capacitor 113, resistor 11%, resistor 1114, and lead 108 to the minus pole 110, but this current flow ceased when the capacitor 113 was fully charged. In charged condition the capacitor 9 has a positive potential on its left electrode and a corresponding negative potential on its right electrode. Hence a negative potential is normally applied to terminal 121.

If now a signal is to be impressed upon terminal 121, this terminal is abruptly placed at the potential of the plus pole 109. In the counting stage 11 as shown in FIG. 1, for example, such as abrupt change in potential is effected by the closing of contacts 8, 9 in the earn-operated switch responsive to the steps of travel performed by the inductance heater.

As a result of this change in potential at lead (terminal) 121, the capacitor 113 (FIG. 4) discharges through the still conductive emitter-base of transistor 161 in te direction from the base to the emitter. In other words, the emitter-base path is connected to the polarity of the capacitor 113 acting as a voltage source. Thus the base received a positive potential and the emitter a negative potential. This causes the flow of current in the control path of transistor 101 to cease, and the transistor is turned off. As a result, the potential of the collector of transistor 1111 becomes more negative. Now a current flows from the plus terminal 169 through the emitter-base path of transistor 102, the parallel connection of capacitor 124 and resistor 105, and thence through resistor 1%, lead 108 to the minus pole of the voltage source. This flow of current commences immediately at full intensity because the current can pass through the unchanged capacitor 124 and declines only gradually to a lower value determined by the resistor 105. The flow of control current through the emitter-base path of transistor 102 renders its emitter-collector path conducting. However, now the capacitor 116 is charged in the manner explained above with reference to capacitor 113. When the potential of terminal 121 is reduced to minus and thereafter abruptly to the plus value of terminal 109, the now charged capacitor 116 is switched upon the emitter-base part of transistor 102. so that the fiow of control current is blocked and the circuit is triggered in the manner already described, As a result, the transistor 1412 is turned 011 and the transistor 1111 is again turned on.

The above-described circuit, operating in the mode last referred to, is employed in the counting stages 11, 12 etc. of chain 10 by issuing sequential control signals to the input lead 121 and employing each time the output value resulting between the reference potential at the terminal 109 and the output lead 122. This output value is issued to the input lead 121 of the next following counting stage in the chain 10.

It will be noted that in FIG. 1 the

stages

11 and 12 in counting chain 16 are provided with

reference numerals

121, 122 indicating the respective signal leads and output leads in the same manner as these numerals are employed for the corresponding signal and output terminals in FIG. 4.

A bistable circuit according to FIG. 4 is used in each of the counting stages of the two digital analog converters 52E and 52D together with the supplement shown in FIG. 5. It is assumed in FIG. 5 that the supplemental components relate to the first stage a in the counting chain 52E. Accordingly, there is shown in FIG. 5 the series-connected resistor 53E preceding the counting chain, as well as the first transverse resistor 54E, corresponding to the two equally denoted resistors in FIG. 1. Connected in series with resistor 54E is the emitter-collector path of a transistor 126 whose base is connected through an ohmic resistor 127 to the output lead 123 according to FIG. 4. When the output lead 122. (FIG. 4) furnishes a finite output voltage, so that transistor 162 is turned on and no voltage obtains between the emitter and collector of transistor 1112, no current will flow through the emitter-base path of transistor 126 (FIG. 5 Consequently, the emitter-collector path of transistor 126 is blocked; that is, the transistor 126 is turned off. Under these conditions the series resistor 53E does not produce a voltage drop as would occur if a transverse current flowed through resistor 54E and transistor 126. Hence, at the right end of resistor 53E there exists a higher potential than when the transverse connection 54E/126 is switched on. The latter occurs when the bistable circuit according to FIG. 4 triggers to the other stable condition in which the transistor 1611 is turned on and the transistor 102 is turned off. Thus, the output voltage furnished by each of the counting chains 52E and 52D is a function of the transverse connections that are switched on at any time through the resistor 54B and 54D to 59D respecti-vely.

For indicating the insertion of the circuitry according to FIG. 4 and FIG. 5 in the counting chains 52E and 52D, there are indicated in the block diagram of FIG, 1 the leads and connections of the counting stages a to f by the same reference characters as used in FIGS. 4 and 5. In this case, the signal terminals and leads 111, 112 are also employed. The leads 112 are used for presetting signals into the counting stages by means of switches l to q. The leads 11 1 serve for issuing clearing signals to the same counting stages of the counting chains.

If desired, the length of the travel steps during wln'ch the diameter and energy control is performed in unit steps during zone-melting operation, can also be varied, for example by correspondingly varying the number of control pulses issuing per revolution of the transport spindle 5 and displacing the heating device.

This can be effected by exchanging the cam disc 6 for a cam having a different number of peripherally distributed lobes. If desired, a group of coaxial cam discs may be provided which can be selectively placed into mechanical connection with the switch 8 to be driven. With a smaller number of cam lobes, the travel steps are larger than with a greater number of lobes. In lieu of the abovementioned mechanical changes, the number of the counting stages in the first portion of the two-part counting chain 10' can be varied. By increasing or decreasing the number of these stages, the size of the travel step is increased or decreased accordingly.

For performing a zone melting operation with the aid of control apparatus according to the invention, the base value of the rod diameter and the base value of the heating energy are first preset into the counting chains 52D and 52B by closing the corresponding switches l to q. The switches l to q are only signal contacts, such as keys or pushbutton contacts, which after entering the signal or counting value return to the illustrated open position. Instead of employing individual switches l to q, a mechanically coupled mechanism can be used operating as a selector switch which, after being adjusted to the selected setting, is used for supplying a corresponding voltage pulse through a common supply lead.

More in detail, relating to the system as illustrated in FIG. 1, the presetting of base values for rod diameter and heating energy is effected as follows. As explained, the purpose of the digital-analog converters 52B and 52D is to convert the signal voltages received from the preceding portion of the system into analog voltages and to supply them to the respective regulating devies 64 and which in turn control the supply of heating energy and the rod diameter respectively. Before starting the operation, each of the converters 52B and 52D must be set for a desired initial output voltage to be impressed upon the regulating

device

64 or 70. This is done by closing a selected number of the switches l, m, n, 0, p, q of each converter 52E, 52D in accordance with a binary code combination. Assume, for example, that the code requires closing the switches l, 0 and p. After these three switches are closed, a voltage source (not illustrated) is connected with the common supply lead of the switches to supply a voltage pulse. This pulse correspondingly sets the flip-flop circuits in respective stages a, d and e of converter 52E or 52D. Starting from the voltage value thus pre-adjusted, each of the grasses 1 1 digital-analog converters 52E and 52]) is switched forward in the sense of a binary counting system.

For ultimately clearing the counting stages a to f in converter 5213, 521), a voltage pulse is applied to the respective clearing leads L (111). This triggers the stages back to the original condition.

The above-described entering of presetting and clearing signals into the respective binary counter stages a to f in converters 52B and 52D will be further understood by reverting to the circuit diagram of such a stage shown in FIG. 4 and generally described above. The signal pulse for presetting the stage is applied to the signal input terminal E (112). Due to the signal, the transistor 1&2 is turned on and the transistor 161 is turned oif so that the signal appears at output terminal A (lead 122.). The clearing signal is applied to the input terminal E (111). This signal switches the bistable network to its other stable condition in which the transistor N52 is turned off Whereas the transistor 191 is turned on.

Assume now that each of the converters 52E and 52D has been set, by closing of a selected binary code combination of switches l to q to furnish a proper output voltage to the respective regulating devices 64 and 7% so that the subsequent counting operation of the system commences from these entered base values for the diameter of the semiconductor rod and for the electric heating energy to be supplied to the melting Zone. The heater winding must now be placed to the zero position of its travel relative to the rod. This can be done in response to a signal and in conjunction with a limit switch or stop which termniates the displacement to the zero position after that position is reached.

When now the apparatus is put into operation so that the transport spindle for moving the heater Winding commences revolving, the travelling nut commences to shift upwardly. Assume, for example, that the spindle thread has a pitch of 1 mm. height and that the cam 6 produces eight pulses for each revolution of the spindle, which pulses are issued to the counter chain It and also to one of the input leads for the AND-gates 24E to 27E for the energy counting chain 52E, and to one of the input leads of the AND-gates 24D to 27D for the diameter counting chain 52D. The first portion of the counter chain it) possesses four counting stages 11 to 14- for binary coding. Consequently, after each two revolutions of the transport spindle 5 and the issuance of a total of sixteen pulses, the second portion of the counter chain 16 receives a signal at its first stage 15. This means that the individual travel step in the example here being discussed amounts to 2 mm. Within each travel step, the change in energy supply and in diameter of the rod 1 is performed in accordance with the number of unit steps pre-adjusted at the cross-bar distributor 46. For example, if for the travel step under observation four unit steps are to be performed for energy change and also for change in diameter, then the cross-bar distributor must release for this travel step the AND-gate 26E as well as the AND-gate 26D. These two gates then pass the output values to the input stages of the respective digital-analog converters constituted by the counting chains 52B and 52D.

While in the foregoing, the over-all design and performance of the system has been described, a more detailed description and explanation will now be presented with respect to the binary-digital converter 28 (FIG. 1) separately shown in FIG. 6, the heat-regulating components separately shown in FIG. 7, and the diameterregulating components separately shown in FIG. 9.

As mentioned above, the converter 28 translates the binary coding of the travel points into analog voltage values continuously representing the travel in accordance with a natural sequence of numbers. For this purpose, the converter 28 may be given the electrical design exemplified in FIG. 6. FIG 6 shows the same counting stages 14-, 15, 16 etc., as illustrated in FIG. 1. In

the converter 28, each of these counting stages is equipped with two amplifier units a and b for respectively amplifying the yes signals and no signals which each of the counting stages 14 etc. furnishes at the respective output terminals A and A of its bistable lip-flop network according to FIG. 4. All components which, according to FIG. 6, form part of the converter are shown within a dot-and-dash enclosure for more clearly indicating the relation of FIG. 6 to FIG. 1.

The converter 28 further comprises diodes designated by D with a numerical subscript, and ohmic resistors analogously denoted by W. At the left side of the illustrated confines, there issues from the converter 28 a multiplicity of individual leads which, for simplicity of illustration, are shown horizontally in FIG. 6, whereas in FIG. 1 they extend vertically from the converter unit 28. In further distinction from FIG. 1, FIG. 6 shows a smaller number of counting stages 14 etc. but a larger number of output leads 261-232 corresponding to those denoted by 29 to 45 in FIG. 1, which issue from the converter 28 for connection to the vertical bars of the crossbar distributor 46.

For explaining the performance of the converter 28 as shown in FIG. 6, assume that the counting system, comprising the counting stages 14 etc., is set to zero. In this condition, the output terminals of the amplifiers b are conductively connected with the zero potential of the voltage source feeding the system. The source terminal that constitutes this zero potential in the system is designated by (-1-). Now the amplifiers a in each of the respective counting stages are not connected to the voltage source and hence cannot apply a potential to any other component of the system.

In this condition, current flows from the positive terminal of the voltage source to the minus terminal through each of amplifiers b, and a combination of the diodes D and one of the resistors W. For example, one of the current paths now established is the following: 15b, D204, W29 i, Another current path is as follows: 17b, diodes D4117, D313, D225, resistor W225, The current can flow through each of the respective resistors W261 to W332. This has the result that the potential obtaining at the connection points between the right-hand ends of resistors W201 to W232 and the diode network D2611 to D232, D301 to D316, D491 to D468, and D591 to D51l4, is virtually equal to zero with respect to the positive pole of the volt-age source. As long as this situation prevails, the output leads 201 to 232, corresponding to those denoted by 29 to 45 in FIG. 1, are on Zero potential.

In the assumption, made above, that all counting stages 14 etc. of converter 28 are in zero position, the only resistor not traversed by current is the resistor W201. Consequently, in distributor 46 the bar 261 connected to resistor W291, corresponding to bar 29 in FIG. 1, has the full negative potential. This negative potential on distributor bar 2111 (or 2?) thus designates the travel point zero for the travel of the heater coil 3 along the semiconductor rod 1 according to FIG. 1, in accordance with the particular control program then in effect.

With the aid of this distinguishing potential on distributor bar 261 (or 29), corresponding signals are given onto the AND-gates in accordance with the selected distributor connections exemplified by the above-described contact plugs or screw bolts S (FIG. 1).

When new a signal appears at the input terminal of counting stage 14, arriving through the chain of counters 11 to 13, the heater coil has reached the travel point 1 and the pulse contact 9 has been closed eight times. The signal in stage 14 switches the flip-flop circuit (FlG. 4) to its other stable condition so that the output terminals of the flip-flop circuit reverse their respective functions. Consequently, the output lead of amplifier 14a is now connected wit-h the plus pole of the voltage source, whereas the output lead 14:) is no longer connected with a pole of the source. Now the currents flowing in the diode network of converter 28 are different in the sense that the next following distributor bar 30 according to FIG. 1, or 202 according to FIG. 6, assumes the negative potential relative to the plus pole of the voltage source. This designates the travel point 1 up to which the heater coil 3 has travelled. The further performance of the system of networks in converter 28 is analogous to the one just described.

FIG. 7 .shows more in detail an example of a regulating circuit applicable for controlling the power supply to the heater coil 3 according to FIG. 1. Represented in FIG. 7 and identified by dot-and-dash enclosures are the same regulating device 64 and high-frequency generator 67 as shown in FIG. 1, as well as the sensing device 66 to which the

components

64 and 67 are connected.

As exemplified in FIG. 7, the high-frequency generator .67 comprises a tube circuit 67a with an output transformer 67b whose inductive coupling between primary and secondary winding is adjustable. The coupling is controlled by an electric motor 670, depending upon the running direction of the motor, the connection of motor 670 with the coupling control device of the transformer being indicated by a broken line 67d.

The measuring device 66 comprises a current transformer 66a to which an ohmic resistor 66!; is connected in parallel. A series-connection of a diode 66c and a capacitor 66d lies in parallel to the resistor 66b. This circuit serves for measuring the peak value of rectified voltage by means of an indicating instrument 66a. The peak value occurring at capacitor 66d is simultaneously impressed between two leads 65 connected to the regulating device 64.

The regulator 64 is equipped with a polarized relay whose winding 64a controls two interconnected movable contacts 64!) and 640 cooperating with respective pairs of fixed contacts which are connected by leads 69 with the motor 67c. The relay contacts 64b and 64c operate as a reversing switch for connecting the motor 670 to a voltage source with one or the other voltage polarity, so that the motor runs in one or the other direction, depending upon occurrence and direction of current flow through the relay 64a.

The regulator 64 i further equipped with two ohmic resistors 64k and MI. The relay winding 64a is connected to the zero potential of the voltage source (D). The right-hand end of resistor 64l is connected to the output lead 69E of the digital-analog converter 52E (FIG. 1). The

components

66d, 64k, 641, connected to the relay winding 64a and across the voltage source (60E-O), constitute a Wheatstone bridge network. Two branches of the bridge network contain respective voltage ources 66:! and (6EO), whereas the relay winding 64a lies in the output diagonal. Consequently, the bridge network compares the voltages furnished from 66d and 60E-O respectively. Depending upon which of the voltages is preponderant, a corresponding difference voltage is impressed upon the relay winding 64a. The current in the relay thus depends as to magnitude and direction upon the magnitude and polarity of the voltage difference. The relay contacts 64b, 640 are displaced accordingly and cause the motor 67c in the generator assembly 67 to correspondingly vary the degree of coupling between the two windings of the transformer 6712. This reduces or increases the heating power supplied to the inductive heater coil 3, depending upon the error voltage formed in the regulating device 64.

FIG. 8 illustrates by way of example an embodiment of circuitry applicable for the

control components

72 and 74 shown in FIG. 1. These components serve for controlling the rod diameter in response to a diameter measurement in the region of the freezing front of the molten zone 1a. This diameter is sensed, according to FIGS. 1 and 8, by means of a winding 71 surrounding the rod 1 at the freezing front. The change in rod diameter modifies the inductance of the winding. The end of the winding are connected to the device 72 identified by dot-anddash confines in FIG. 8. The device comprises a highfrequency generator 72a of about 20 megacycles per second which is connected in the input diagonal of a bridge network whose branches comprise respectively two

ohmic resistors

72b, 72c and an inductance coil 72d. The inductance or coil 72d constitutes, in effect, the datum value in the network, and as soon as the reactive impedances of

components

71 and 72d are equal, the bridge network is in an electrical condition at which a given output value of voltage appears in its output diagonal comprising the primary Winding 7 2c of a transformer 72k.

When the inductance of sensing coil 71 changes in dependence upon changes in rod diameter, the condition of the bridge is modified, and a correspondingly increased or reduced voltage appears in the output diagonal. This output voltage is applied by the secondary winding 72f of transformer 72k through an amplifier 72g to a series connection composed of a diode 72h and a capacitor 72i. The rectified peak value of voltage appearing across the capacitor 721', is applied through lead 75 to the regulator 70.

The regulator 70 may be given a design and performance as described above with reference to the regulator 64 shown in FIG. 7. Accordingly, there is connected to the regulator 70 the pole O of the voltage source and a lead coming from the terminal 60D of the digital-analog converter 52D, as is also apparent from FIG. 1. As explained with reference to FIG. 7, the regulator contains a bridge network whose output diagonal comprises a polarized relay which, in turn, energizes the reversible motor 74a in control device 74 to run in one or the other direction, depending upon the direction in which the rod diameter departs from the programmed datum value. The motor 74a acts through a transmission gear 74 upon the lower holder 2 of the semiconductor rod 1 for axially displacing the holder in one or the other direction, thus subjecting the molten zone to compression or stretching as required for maintaining a constant diameter at the freezing front of the molten zone.

I claim:

1. Electric control apparatus for floating-zone melting of semiconductor rods, comprising two coaxial rod-end holders jointly defining a rod axis, an electric heating device movable along said axis for producing a molten zone in the rod being processed, holder displacing means for axially displacing one of said holders relative to the other to vary the cross section of the molten zone, electric power supply means connected to said device and being controllable for varying the heating energy transmitted to the rod being processed, a digital control system connected to said displacing means and said power supply means for conjointly controlling said cross section and said heating power in dependence upon the travel of said heating device along the rod axis, said control system having counting-pulse transmitter means in operative connection with said heating device for issuing a given number of pulses for given travel steps of said device, and said system having two programming means for presetting a travel-responsive power program and a travel-responsive cross-section program respectively, whereby the heating power supplied to the zone and the zone cross section are controlled in unit increments of change corresponding to the steepness of the respective programs.

2. In electric control apparatus for zone melting according to claim 1, said heating device comprising an inductance heater winding extending about said axis, and a transport mechanism for moving said heater along the rod axis, said pulse transmitter means of said digital control system being connected with said transport mechanism for issuing said given number of counting pulses for each of equal steps of travel of said heater.

3. Electric control apparatus for floating-zone melting of semiconductor rods, comprising two coaxial rod-end holders jointly defining a rod axis, an electric heating device movable along said axis for producing a molten zone in the rod being processed, holder displacing means for 1 axially displacing one of said holders relative to the other to vary the cross section of the molten zone, electric power supply means connected to said device and being controllable for varying the heating energy transmitted to the rod being processed, a digital control system connected to said displacing means and said power supply means for conjointly controlling said cross section and said heating power in dependence upon the travel of said heating de vice along the rod axis, said control system having counting-pulse transmitter means in operative connection with said heating device for issuing a given number of pulses for given travel steps of said device, and said system having two programming means for presetting a travel-responsive power program and a travel-responsive cross section program respectively, whereby the heating power supplied to the zone and the zone cross section are controlled in unit increments of change corresponding to the steepness of the respective programs, said digital control system comprising a pulse-counting chain of successive counter stages whose first stage is connected to said travel-step responsive pule transmitter means and whose respective subsequent stages are inputwise connected to the preceding stage and outputwise to the next following stage whereby each subsequent stage receives a pulse when a given number of pulses are counted by the preceding stage, two groups of AND-gates of which each has two input leads and an output circuit, said counting chain having a first portion which comprises a plurality of said counter stages commencing with said first stage, and said chain having a second portion which comprises a sequence of said stages succeeding said first portion, one of said two input leads of said respective AND-gates being connected to the respective counter stages in said first portion of said chain, said counter stages in said second portion being connected through said two programming means with the respective other input leads of said respective AND-gates in said respective two gate groups, and two digital-analog converters interposed between the two gate output circuits and said rod-holder displacing means and said power regulating means respectively.

4. In an electric control apparatu for zone melting according to claim 3, each of said digital-analog converters comprising two voltage supply buses, a resistor, a voltage source connected to said buses in series with said resistor, a plurality of counter stages and resistors, said latter counter stages being connected across said two buses in series with one of said latter resistors respectively to jointly form a chain network of counter stages, whereby a travel-responsive and program-controlled output voltage appears between said buses at the end of said chain network remote from said source.

5. In electric control apparatus for Zone melting according to claim 3, said programming means comprising a cros bar selector having groups of members selectively interconnectable at respective intersection points for setting the desired control program.

References Cited by the Examiner UNITED STATES PATENTS 2,992,311 7/61 Keller 219-l0.77 3,002,115 9/61 Johnson et al 318-162 3,046,379 7/62 Keller et al 2l9-10.77

RICHARD M. WOOD, Primary Examiner.

Claims

1. ELECTRIC CONTROL APPARATUS FOR FLOATING-ZONE MELTING OF SEMICONDUCTOR RODS, COMPRISING TWO COAXIAL ROD-END HOLDERS JOINTLY DEFINING A ROD AXIS, AN ELECTRIC HEATING DEVICE MOVABLE ALONG SAID AXIS FOR PRODUCING A MOLTEN ZONE IN THE ROD BEING PROCESSED, HOLDER DISPLACING MEANS FOR AXIALLY DISPLACING ONE OF SAID HOLDERS RELATIVE TO THE OTHER TO VARY THE CROSS SECTION OF THE MOLTEN ZONE, ELECTRIC POWER SUPPLY MEANS CONNECTED TO SAID DEVICE AND BEING CONTROLLABLE FOR VARYING THE HEATING ENERGY TRANSMITTED TO THE ROD BEING PROCESSED, A DIGITAL CONTROL SYSTEM CONNECTED TO SAID DISPLACING MEANS AND SAID POWER SUPPLY MEANS FOR CONJOINTLY CONTROLLING SAID CROSS SECTION AND SAID HEATING POWER IN DEPENDENCE UPON THE TRAVEL OF SAID HEATING DEVICE ALONG THE ROD AXIS, SAID CONTROL SYSTEM HAVING COUNTING-PULSE TRANSMITTER MEANS IN OPERATIVE CONNECTION WITH SAID HEATING DEVICE FOR ISSUING A GIVEN NUMBER OF PULSES FOR GIVEN TRAVEL STEPS OF SAID DEVICE, AND SAID