US3205101A

US3205101A - Vacuum cleaning and vapor deposition of solvent material prior to effecting traveling solvent process

Info

Publication number: US3205101A
Application number: US287561A
Authority: US
Inventors: Abraham I Mlavsky; Leonard B Griffiths
Original assignee: Tyco Laboratories Inc
Current assignee: Tyco International Ltd
Priority date: 1963-06-13
Filing date: 1963-06-13
Publication date: 1965-09-07
Anticipated expiration: 1982-09-07
Also published as: BE649024A; NL6406340A; FR1398471A

Description

Sept. 7, 1965 VACUUM CLEANING AND VAPOR DEPOSITION OF SOLVENT MATERIAL PRIOR TO EFFEGTING TRAVELING SOLVENT PROCESS Flled June 13, 1963 A l. MLAVSKY ET AL 2 Sheets-Sheet 1 I (In Or FILM A C. SUPPLY Q C. (2 SUPPLY 3 /6 \J 2 F G. I

Si C F l G. 2

-SiC

INVENTORS ABRAHAM l. MLAVSKY IEEONARD B. GRIFFITHS /WMM ATTORNEYS Sept. 7, 1965 A. MLAVSKY ETAL. 3,205,101 VACUUM CLEANING AND VAPOR DEPOSITION OF SOLVENT MATERIAL PRIOR TO EFFECTING TRAVELING SOLVENT PROCESS Filed June 13, 1963 2 Sheets-Sheet 2 E a 46 Qz FIG. 3 l 44 h I /SiC 5Z F l G. 4

INVENTORS ABRAHAM l. MLAVSKY EFONARD B. GRIFFITHS WW/ W W ATTOR N EYS United States Patent Oflice 3,205,101 Patented Sept. 7, 1965 3,205,101 VACUUM CLEANING AND VAPOR DEPOSITION OF SOLVENT MATERIAL PRIOR TO EFFECTING TRAVELING SOLVENT PROCESS Abraham I. Mlavsky, Lexington, and Leonard B. Griffiths, Northhoro, Mass., assignors to Tyco Laboratories, Inc., Waltham, Mass., a corporation of Massachusetts Filed June 13, 1963, Ser. No. 287,561 20 Claims. (Cl. 1'48171) The present invention relates to a method of growing silicon carbide crystals and silicon carbide P-N junctions.

The utility of silicon carbide as a component in semiconductor devices is well known and considerable effort has been expended in attempting to grow large single crystals of silicon carbide and to fabricate silicon carbide electronic devices with good electrical characteristics. Various methods have been. considered for growing silicon carbide crystals and for fabricating silicon carbide bodies with P-N junctions. In the main these methods follow one of three different approaches for growing crystals of semiconducting compounds. One approach is to grow the crystals from the vapor phase. Another approach is to grow them from the melt. The third approach is to grow the crystals from solution. The first approach is unsatisfactory because of the high operating temperature which necessitates the use of costly apparatus and makes it difficult to obtain good electrical characteristics on a reproducible basis. Growing silicon carbide crystals from the melt is complicated because it has no liquid phase at normal pressures. To grow silicon carbide from the melt it is necessary to resort to pressurization of the growth chamber to a high level, e.g., approximately 300 atmospheres. It is believed to be obvious that the pressurization requirement renders this approach impractical. The present invention is directed to the third approach, namely, growth from solution, using what we call the travelling solvent concept which has appeared to offer the most promise since itlends itself not only to the growth of single crystals but also to preparation of P-N, P-N-P, and N-P-N junctions.

The travelling solvent concept involves (1) formation of a sandwich consisting of two slices of a crystalline material and a solvent formed as a molten zone between these two slices, and (2) establishment of a temperature gradient across the sandwich with the average temperature of the sandwich kept at a level' sufiicient to maintain the solvent in a molten condition. Due to the temperature gradient the temperature at the interface of the molten zone and one slice of material will be higher than at the interface with the other slice. Since its solubility will be higher at the higher temperature interface, the crystalline material will dissolve at that interface. The solvent zone will travel through the slice having the hotter interface, dissolving that slice as it moves; The dissolved portion will deposit out at the coolerinterface, undergoing crystal regrowth on the other slice of crystalline material. The molten solvent zone will travel through the dissolving slice at a rate which is a function of the average temperature of the sandwich, the temperature gradient between the two interfaces, the solubility of the crystalline material in the solvent, and the liquid diffusion coefficient of the crystalline material in the molten zone at the average temperature of the sandwich. Examples of the knowledge and use of the travelling solvent concept prior to the present inventors are provided by W. G.

single crystals of silicon carbide using chromium as the solvent, persons skilled in the art of growing crystals have been unable to obtain good silicon carbide crystal growth in reproducible fashion using the information and techniques available prior to. the present invention. Moreover, while it has been possible to produce P-N junctions in a silicon carbide body, by means of the travelling solvent. concept, the results have been sporadic with poor and often no yield from a given run. When assembled and used as diodes, silicon carbide junctions obtained prior to the present invention demonstrated unsuitable and generally greatly varying electrical characteristics. A major problem with silicon carbide diodes heretofore obtainable has been their. forward voltage drop. With diodes made according to the prior art, it has not been possible to achieve reasonable current densities at; voltages less than 3 volts at room temperature and frequently some diodes would require a forward voltage dropof as much as 6 volts to draw a reasonable current density. From'a purely theoretical point of view an a silicon carbide. P-N- junction would be expected typically to draw a reasonable current with a forward voltage drop of the order of 1.4 to 1.7 volts and in any, event no larger than its energy gap of about 2.75 electron volts.

The primary object of the present invention is to provide an improvement in the travelling solvent method of growing crystals which makes it possible to achieve good crystal growth.

Another object of this invention is to provide an improvement in the method of growing crystals by use of a travelling solvent zone whichmakes it possible to grow silicon carbide crystals with physical and electrical characteristics notably superior to what has been possible previously, and of growing suchsuperior crystals. in reproducible fashion with relatively high yield.

A more specific objectof the. invention is to provide a reliable method'of making silicon carbide P-N junctions having good rectifying characteristics and in particular having a forward-voltage drop in the order of 1.4 to. 1.7 volts.

Other objects and many of the attendant. advantages of the invention will be readily apparent from the following detailed specification when considered together with. the accompanying'drawings wherein:

FIG. 1 shows apparatus for cleaning and plating specimens of siliconcarbide according to'the present invention;

FIG. 2 is an exaggerated cross-sectional view of a sandwich prepared in accordance with. the present invention;

FIG. 3 illustrates an arrangement for growing crystals by the travelling solvent method; and

FIG. 4 is a cross-sectional view of a silicon carbide body illustratingthe. improvement effected. by the present invention.

We have discovered that a major cause of the poor results and unreliability of the travelling solventmethod as previously established was failure of the solvent zone to move evenly through one of the outer layers of the sandwich. Investigation revealed that. with the method as practiced before this invention, difficulty was encountered in obtaining good interfacial wetting between the chromium and silicon carbide. While some wetting occurred, it was non-uniform and did not occur in reproducible "fashion. With interfacial dissolution taken as a criterion of wetting, the existence of difficulties in this area was unexpected since silicon carbide is soluble in chromium when introduced into molten. chromium in an arc furnace. In the lightof the discovery that with the travelling solvent method as previously known, chromium and silicon carbide exhibit. poor interfacial' wetting, we conceived that better diffusion of the chromium zone through the silicon carbide could be achieved by improving the interfacial wetting. This line of attack was predicated on the hypothesis that some physical barrier preventing wetting exists at the silicon carbide-chromium interface and that the molten solvent metal is unable to penetrate the barrier. In this connection it is to be observed that single crystal silicon carbide is obtainable commercially in reasonable quantities only in pieces of irregular shapes and sizes so that cutting and grinding operations are necessary for utilization in the travelling solvent methodof crystal growth. This preliminary sizing operation is usually followed by polishing with diamond powder and etching in molten sodium hydroxide at 700 C. as finishing operations. Careful observation of the surfaces of specimens prepared in the foregoing manner indicated, by the existence of specular surface colorations, the presence of a layer or layers of oxide or adsorbed gas of appreciable thickness, even though the etching step is designed to clean the surfaces of the specimen. Experimentation in controlling the etching parameters such as temperature and time proved unsuccessful in eliminating the various surface colorations and we hypothesized that it was necessary to provide a surface treatment capable of promoting good and uniform interfacial Wetting in order to achieve excellent solvent zone movement with uniform crystal growth. The obvious step of cleaning the surface by treatment with HF solution did not materially improve the wetting characteristics. Accordingly, we concluded that another surface treatment would have to be devised for removal of the physical barrier to wetting which appears to be present when practicing the travelling solvent method as heretofore known. While the exact nature of the physical barrier is not known with certainty, it is believed to take the form of a thin tanacious oxide film which may be monomolecular and/ or a layer of adsorbed gas. The surface treatment provided by the present invention not only achieves its intended purpose but also makes it possible to fabricate silicon carbide P-N junctions having electrical characteristics superior to those obtainable with the travelling solvent method as heretofore practiced.

Briefly described, the new surface preparation involves heating the silicon carbide slices in a high vacuum to remove surface oxides and adsorbed gases, and evaporating a film of chromium onto the cleaned surfaces before the gases are readsorbed and before new oxides are formed thereon. Preferably, but not necessarily, the heating is accomplished by electron bombardment and the vacuum is of the order of to 10 millimeters of mercury, with the pressure preferably being less rather than greater than 10- mm. of mercury. The immediate visible result of the high vacuum heating is that of greatly enhanced reflectivity. After the chromium films have been deposited thereon, the silicon carbide slices are put together in pairs with their chromium coatings in confronting relation with each other.. The result is a sandwich consisting of two outer layers of silicon carbide and, as a practical matter, one inner layer of chromium. Preferably, but not necessarily, an extra amount of chromium metal is placed between the two slices of silicon carbide so as to form a somewhat thicker sandwich. The latter is heated by suitable means to melt the chromium and to create a temperature gradient across the silicon carbide-chromium interfaces suitable to cause the chromium zone to pass through one of the two slices of silicon carbide. The latter dissolves and regrows from the chromium solution onto the other slice of silicon carbide with excellent results.

Referring to FIG. 1 the present invention will now be described in more detail. FIG. 1 shows apparatus for cleaning silicon carbide specimens and evaporating thereon a thin layer of chromium. The apparatus shown in FIG. 1 comprises a glass vacuum chamber 2 having therein a tungsten pedestal 4 on which specimens 6 of silicon carbide may be disposed for cleaning and coating with chromium. Also disposed within the vacuum chamber 2 is an electron-emitting filament 8 and a tungsten heater coil 10. Filament 8 is connected by means of a variable transformer 12 to a suitable A.C. power supply 14. Filament 8 and pedestal 4 are also connected to a high voltage D.C. supply 16, pedestal 4 being disposed at a positive potential relative to filament 8. Although not shown it is to be understood that the tungsten heater coil 10 is connected to a suitable power supply through a suitable control unit. The chamber 2 is adapted to be evacuated through a pair of liquid nitrogen traps 26 and 28 by means of a pair of mercury diffusion pumps 20 backed up by a rotary pump 24. The nitrogen traps condense all condensable gases, such as nitrogen, oxygen and water vapor, and simultaneously prevent backflow of mercury vapor into chamber 2 from diffusion pumps 2!). The magnitude of the vacuum in the chamber 20 is indicated by an ionization gauge 30. Also forming part of the apparatus is a hood 32 whose walls are provided with an electric heating element 34. The hood can be lowered to the position shown in FIG. 1 where it surrounds and can heat up chamber 2. When not in use, hood 32 is raised above chamber 2 by means not shown.

In accordance with the present invention specimens of silicon carbide are cut and ground to proper dimensions and polished to a smooth surface finish. Then they are etched in molten sodium hydroxide at a temperature of approximately 700 C., followed by immersion in HF solution and washing and drying. Thereafter the specimens are placed within vacuum chamber 2 on the pedestal 4. Although only one specimen is illustrated in FIG. 2, it is to be understood that the pedestal 4 may be large enough to accommodate a relatively large number of specimens at the same time. A suitable supply of chromium having been placed previously onto the heater coil 10, the chamber is sealed off and the vacuum pumps started. The chamber is evacuated and the hood 32 is lowered to the position shown in FIG. 1. The heating element 34 is energized to heat up chamber 2 to a temperature of about 450 C. The chamber is baked at that temperature for about 10-12 hours or longer to outgas it, with the vacuum pumps operating .to establish and maintain a vacuum of 10- to 10- mm. 'of mercury. At the end of the baking step, the hood is removed, the liquid nitrogen traps are filled, and power is supplied to the filament 8 to initiate heating of the specimens by electron bombardment. This operation is facilitated by making the pedestal 4 highly positive with respect to the filament 8. The specimens are heated by electron bombardment to a temperature at least as high as 1200 C. (preferably in the range of 1300 to 1400 C.). This temperature is kept steady for approximately 4 to 10 minutes. During this interval greatly enhanced reflectivity becomes evident at the exposed surface of the specimens. Then the electron bombardment is discontinued and energy is supplied to heater coil 10. Because of the high vacuum environment it is not necessary to heat .the chromium to its melting point vof approximately 1900 C. in order to evaporate a sufficient quantity for surface coating purposes. The chromium on the tungsten coil 10 can be heated to any suitable temperature below 1900 C. at which evapora tion of the chromium will occur. Of course, the closer the temperature is brought to 1900 C. the faster will be the evaporation of chromium. The chromium can also be heated above 1900 C. Evaporation in a high vacuum chamber greatly assists in purifying the chromium, particularly with respect to gas content. Hence, any possibility that impurities in the chromium affect the degree tend to return to the specimens and also that there may be a tendency for an oxide layer to form again on the specimens after the electron bombardment heating is stopped. It is reasonable to assume also that the extent to which this will occur will depend on the concentrations of such impurities, the extent of the cooling of the specimens, and in particular the time lag between the electron bombardment heating and the evaporation of the chromium. Therefore, care is taken to start evaporating the chromium with as little delay as possible. The best practice is to do this while at as high a vacuum as possible since this will minimize the rate of readsorption of impurities by the specimens. The thickness of the chromium films is not critical and may vary considerably. The important thing is that it be thick enough to protect the specimen surfaces against readsorption of impurities and oxidation. In practice it is preferred that the chromium film have a thickness in the order of 1,000 to 10,000 Angstroms. The chromium deposits in smooth layers and adheres strongly to the silicon carbide substrates. Thus, if the specimens are reheated to 1300 C. immediately following deposition of the chromium, they exhibit no evidence of peeling or blistering of the chromium films.

After the silicon carbide specimens have been coated with the chromium, they are arranged in pairs to form sandwiches. Depending upon the thicknesses of the deposited chromium metal coatings, an extra slice of chromium metal may or may not be included in each sandwich. The greater the thickness of chromium applied to the silicon carbide by evaporation, the less the thickness of the extra slice of chromium metal which is inserted in the sandwich. It is to be understood that the chromium platings may be made sufiiciently thick to eliminate the need for an extra piece of chromium metal. However, it is preferred that the amount of chromium applied by evaporation be kept in the order of ten thousand Angstroms and that an extra piece of chromium metal be inserted between the two slices of silicon carbide. This preferred form of sandwich is illustrated generally at 36 in FIG. 2. The sandwich consists of flat upper and lower pieces of silicon carbide, the upper piece having a thin film of chromium on its underside and the bottom piece having a thin film of chromium on its upper side. Sandwiched between the two pieces of silicon carbide is an extra piece of chromium metal whose thickness is substantially greater than that of the chromium films.

The sandwich of FIG. 2 is heated in accordance with the requirements of the travelling solvent concept using suitable apparatus such as the one shown in FIG. 3. The illustrated apparatus comprises an elongated quartz tube 40 surrounded by an RF heating coil 42. Mounted within the tube by means of a carbon support rod 44 is a carbon block 46. The rod 44 is mounted on a suitable base 48. A protective inert atmosphere such as argon or helium is fed into the top of the tube by an inlet conduit 50 and removed at the bottom end by another conduit 51. The sandwich 36 is positioned upon the carbon block 46. The size of the sandwich is exaggerated in FIG. 3 for convenience of illustration. When the RF coil 42 is energized, the carbon block heats up rapidly and a temperature gradient is established across the sandwich with the highest temperature at the bottom. Cooling from the upper surface of the sandwich occurs by radiation. The graphite block 46 is heated to a temperature in the range of 1700 to 1900 C. preferably in the order of between 1800 and 1900 C. It is preferred that the average sandwich temperature be maintained below the melting point of elemental chromium which is approximately 1900" C. This procedure is preferred in order to minimize loss of chromium from the sandwich by evaporation. It is possible to practice the travelling solvent method at a temperature below the melting point of elemental chromium since a eutectic exists in the chromiumsilicon carbide alloy system having a melting point of about 1600 C. In this connection it is to be understood that the graphite block can be heated to a temperature higher than 1900 C. since there is a temperature drop between the graphite block and the sandwich. At a temperature of about 1870 C. on the block, there will be a temperature gradient of approximately 100 C. across a sandwich of total thickness of -100 mils. The temperature gradient which is established across the sandwich should be sufficient to cause growth, i.e., zone movement, to occur at a reasonable rate. A differential of C. is suitable, providing a zone movement of about 30 mils per hour. Preferably, the graphite block is held at its temperature for about two hours which is sutficient time for the chromium to melt and form a liquid zone which can dissolve and travel through the bottom slice of silicon carbide. The chromium zone will pass completely through the bottom slice of silicon carbide. As the chromium zone moves down through it, the bottom slice will dissolve into the chromium at the interface and will regrow onto the adjacent face of the top slice of silicon carbide. When the process is complete the resulting body comprises a monocrystal of silicon carbide containing no voids, occlusions and inclusions. Confirmation that good crystal regrowth occurs is obtainable by standard Laue X-ray techniques.

The importance of our surface preparation can be seen by a simple experiment. The experiment consists of taking a first slice of silicon carbide having a de posited layer of chromium on one surface, scratching away some of the deposited chromium from a limited area of the silicon carbide slice, leaving that slice of silicon carbide in the atmosphere for approximately twenty minutes, and then bringing it together with a second chromium-coated slice and an additional quantity of chromium to make a sandwich as previously described. When this sandwich is processed in the manner illustrated in FIG. 3 with the aforesaid first slice on top, a void will occur where the deposited chromium was removed, thereby demonstrating that wetting did not occur over that particular area of the first slice of silicon carbide. FIG. 4 is an enlarged cross-section of a silicon carbide body 52 produced from a sandwich of two separate slices of single crystal silicon carbide treated according to the foregoing experiment. The sandwich was treated long enough for the chromium zone to pass completely through the bottom piece of silicon carbide. The chromium layer is seen at 54. For reference purposes the junction of the two originally separate slices is indicated by the dashed horizontal line 56. As can be seen, a void 58 occurs in the plane of the junction. This void is in line with the region where chromium was removed from the original top slice of silicon carbide. However, where the deposited chromium film was not disturbed, regrowth of the crystal occurs in a uniform manner Without any voids.

A specific example of how the improved method is executed is presented hereinafter in connection with the fabrication of P-N junctions where the influence of the surface preparation aspect of the invention on the electrical characteristics of the semi-conducting material is most evident. In this specific example, the apparatus of FIGS.1 and 3 is employed. A slice of aluminum-doped P-type single crystal silicon carbide and a slice of nitrogendoped N-type single crystal silicon carbide each with a thickness of approximately 40 mils are polished with diamond powder, etched in molten sodium hydroxide at approximately 700 C. for ten minutes, immersed in hydrofiuoric acid for five minutes, washed, dried and then placed in the apparatus of 'FIG. 1. The chamber 2 is evacuated to approximately 10- mm. of mercury. Then with the vacuum being maintained by the pumps 20 and 24, the hood 32 is lowered and its heating element 34 is energized to heat chamber 2 to a temperature of 450 C. After twelve hours the hood is raised enough to expose nitrogen trap 28 which then is filled with liquid nitrogen. Two hours later the hood is raised above chamber 2 and the filament S and coil 10 are pulsed to heat them up to a temperature sufficient to outgas them. Then the ionization gauge is outgassed. Finally the upper trap 26 is filled with liquid'nitrogen. With the outgassing thus completed, filament 8 is energized to heat the two specimens to a temperature of l300-1400 C. for approximately five minutes. Immediately thereafter the heater 10 (laden with chromium) is energized. Chromium evaporates from the heater coil and deposits on the two pieces of silicon carbide. Evaporation is discontinned after a visible film (thickness approximately 10,000 Angstroms) is attained. Then the two pieces of silicon carbide are placed together with an intermediate thin sheet of chromium measuring approximately mils in thickness so as to form asandwich. The sandwich is placed in the apparatus of FIG. 3 on the carbon block 46 in the manner previously describedand heated for a period of two hours with the carbon block temperature at approximately 1870 C. Argon is pumped through the quartz tube all the while that heating operation is in progress. The sandwich is disposed so that the P-type silicon carbide is on top and the N-type silicon carbide is on the bottom. During the two hour heating period the chromium content, including both the plated chromium layers and the extra sheet of chromium, will melt to form a molten zone and this molten zone will move down through the N-type silicon carbide. As the chromium zone travels downwards the N-type silicon carbide will dissolve and then redeposit out onto the P-type silicon carbide so as to form a unitary void-free crystalline body with a uniform P-N junction formed at the initial plane of growth. The chromium ends up as a single layer at the bottom of the sandwich and is sliced off. The remaining mass is then diced to form a plurality of small slices (each with a P-N junction) which can be made into diodes. Whenmade into diodes, they will show substantially uniform electrical characteristics. The significant thing is that with forward voltages in the region of 1.4 to 1.7 they will exhibit current densities heretofore attainable with other silicon carbide diodes only at forward voltages in the order of at least 3 volts or more. Current densities in the order of 3x10 micro amps per square centimeter are obtained with a forward voltage of 1 volt and approximately 1.7)( micro amps per square centimeter are obtained with a forward voltage of 1.5 volts.

Although in the foregoing examples the solvent zone passes through N-type silicon carbide, it also may be passed through P-type silicon carbide to form a P-N junction. Silicon carbide diodes made by passing the chromium zone through P-type material to form a P-N junction exhibit substantially the same electrical characteristics as those made by passing the solvent zone through N-type silicon carbide material. Thus it is believed to be apparent that the improved electrical characteristics which are obtained are due essentially to the surface preparation prescribed by this invention and are not determined by the chromium passing through a particular type (P or N) silicon carbide.

As indicated previously, the need for introducing an extra amount of chromium between the two pieces of chromium-plated silicon carbide is determined by the thickness of the chromium films on the pieces of silicon carbide. If the thicknesses of the chromium platings are sufiiciently great to provide a uniform thickness of molten chromium between the two slices despite any voids or depressions which may exist in either or both of the confronting surfaces of silicon carbide, then no additional sheet of chromium need be added to the sandwish. However, 'in practice it is preferred to add a thin sheet of chromium having a thickness of the order of 1-5 mils to the sandwich in order to assure that good results will be obtained. Of course, this additional chromium should be as pure as possible to avoid contamination of the silicon carbide which might alter its electrical characteristics or adversely affect crystal growth. A relatively larger amount of chromium does not affect the process except that the thickness of the chromium zone should not be so great as to fully consume the silicon carbide slices.

An important consideration of the present invention is that the regrowth of one pure crystal of silicon carbide onto the second crystal of silicon carbide is epitaxial, with the second crystal functioning as a seed. In principle both pieces of silicon carbide material in a given sandwich need not be single crystals at the outset to produce a single crystal end product. It is to be understood that the piece which is to be on the cooler side of the sandwich, i.e., the one on which growth is to occur, may be a single crystal while the other piece may be a poly-crystalline body; and that when the process is carried out in the manner previously described, the poly-crystalline body will dissolve on one side of the solvent zone and redeposit out on the other ide, growing epitaxially onto the original single crystal piece which functions as a seed.

Another interesting aspect of the invention is that the crystal regrowth is carried out at a temperature below the temperature at which the oc-fi transition occurs. Normally the 3 form of silicon carbide occurs below 2000 C. and the a form occurs above that temperature. Hence in the present invention, even though the initial material is on silicon carbide, it would be reasonable to assume that the material which undergoes crystal regrowth will assume the ,8 form. However, in practice the regrown material is a silicon carbide.

Although the invention as herein described is based on the use of a silicon carbide, it is contemplated that the invention is applicable also to the growth of B silicon carbide crystals and P-N junctions.

While the invention has been described in connection with the growth of silicon carbide crystals using chromium as a solvent, it is contemplated that the invention may use other solvent materials and may be applicable to the growth of crystals of different chemical composition. Thus in the case of silicon carbide it may be possible to use pure silicon or platinum, silicon-chromium, siliconplatinum and platinum-chromium alloys as a solvent in place of chromium. Also it is reasonable to expect that gallium arsenide or gallium phosphide could be grown with gallium as a solvent, so that a relatively large area P-N junction may be made by regrowing gallium arsenide or gallium phosphide onto a slice of single crystal gallium arsenide or gallium phosphide.

It is to be understood also that the process is not limited to the apparatus shown in the drawings or to the precise conditions and materials set forth in the foregoing specification. Thus, for example, the specimens need not be heated by electron bombardment but may be heated by some other means, such as RF energy, prior .to deposition of the chromium film. Similarly some means other than the apparatus shown in FIG. 3 may be used to heat the sandwich to get the necessary temperature gradient.

Obviously many other modifications and variations of the present invention are possible in view of the above teachings. lt is to be understood, therefore, that the invention is not limited in its application to the details and arrangements specifically described or illustrated, and that within the scope of the appended claims, it may be practiced otherwise than as specifically described or illustrated.

What is claimed is:

1. A method of growing a crystalline material comprising the steps of providingg two separate bodies of said crystalline material, heating each body in a vacuum to remove surface impurities and depositing a thin adherent film of a solvent for said material on a surface of each body while in said vacuum, said solvent being a solid which melts below the melting point of said material, arranging a sandwich with said bodies as outer layers and said solvent as an intermediate layer, establishing a temperature gradient across said sandwich suificient to melt said intermediate layer and form a migrating liquid zone which dissolves the body on .the hotter side of said zone and allows it to reform on the cooler side as an integral part of the body on said cooler side.

2. A method as defined by claim 1 wherein the vacuum is in the order of mm. of mercury.

3. A method as defined by claim 1 wherein an additional mass of said solvent is inserted in the sandwich between said bodies prior to establishing said temperature gradient.

4. A method as defined by claim 1 wherein said sand wich is heated in an inert atmosphere.

5. A method as defined by claim 1 wherein said bodies are single crystals.

6. A method as defined by claim 1 wherein one of said bodies is an N-type semiconductor and the other is a P-type semiconductor.

7. A method of growing silicon carbide comprising the steps of providing two separate pieces of crystalline silicon carbide, heating each body in a high vacuum to remove surface impurities and depositing a thin adherent film of a solvent for silicon carbide on a surface of each piece while in said high vacuum, said solvent being a solid which melts below the melting point of silicon carbide, arranging a sandwich of said pieces with said films forming the inner layer of said sandwich, and establishing a temperature gradient across said sandwich suflicient to convert said inner layer to a migrating liquid zone which dissolves one of said pieces and allows it to reform as an integral part of the other piece.

8. A method of growing a crystalline material comprising the steps of providing first and second bodies of said material each having a fiat surface, subjecting said surfaces to electron bombardment under a high vacuum to remove surface impurities, terminating said bombardment and coating said surfaces with a thin layer of a metal solvent for said material while still under said vacuum, arranging said bodies to form a sandwich with said metal solvent as the inner layer of said sandwich, establishing a temperature gradient across the sandwich with (1) the temperature of said first body lower than the temperature of said second body and (2) the temperature of said metal solvent material at a level sufiicient for said solvent material to melt and form a travelling molten zone which dissolves said second body -on one side thereof and reforms it on the other side thereof as an integral portion of said first body.

9. The method of claim 8 wherein said vacuum is at least of the order of 10- to 10- mm. of mercury.

10. The method of claim 8 wherein said bodies are silicon carbide.

11. The method of claim 10 wherein said solvent is chromium.

12. A method of growing silicon carbide crystals comprising the steps of providing two separate slices of single crystal silicon carbide, removing oxygen from the surfaces of said slices by heating in a vacuum, depositing a thin adherent film of chromium on an oxygen-free surface of each of said slices, arranging a sandwich of three layers of material with the two outer layers comprising said two slices of silicon carbide and the intermediate layer comprising said chromium films, establishing a temperature gradient across the sandwich sufficient to melt the intermediate layer to form a migrating liquid zone which dissolves the silicon carbide slice on the hotter side of said zone and allows it to reform as an integral crystalline part of the other slice of silicon carbide.

13. A method as defined by claim 12 wherein one slice is P-type silicon carbide and the other is N-type silicon carbide.

14. A method of growing a silicon carbide P-N junction comprising the steps of providing a body of P-type and a body of N-type silicon carbide each having a flat surface, heating said bodies to an elevated temperature by electron bombardment under a high vacuum, evaporating a thin film of chromium into said surfaces terminating said bombardment, arranging said bodies to form by electron bombardment under a high vacuum, terminating said bombardment, evaporating a thin film of chrominum onto said surfaces while under said high vacuum, arranging said bodies to form a sandwich with the thin films of chromium confronting each other and comprising an intermediiate inner layer, establishing a temperature gradient across the sandwich with the temperature of the intermediate layer at a temperature sufiicient to melt said intermediate layer, whereby said intermediate layer forms a travelling liquid zone which dissolves the hotter silicon carbide on one side thereof and reforms it as an integral part of the cooler silicon carbide on the other side thereof, with a P-N junction in the region of the original boundary between said two bodies of silicon carbide.

15. A method as defined by claim 14 wherein said vacuum is of the order of 10* to 10- mm. of mercury.

16. A method as defined by claim 14 wherein an additional piece of chromium is inserted in the sandwich between said thin films prior to establishing said temperature gradient.

17. The method of producing a silicon carbide P-N junction comprising the steps of: placing a piece of P- type silicon carbide in a chamber, evacuating said chamber and heating said P-type piece to remove volatile surface impurities, terminating said heating and evaporating a thin film of chromium onto a surface of said P-type piece while said piece is in said evacuated chamber; evaporating a thin film of chromium onto a piece of N-type silicon carbide according to the foregoing steps; thereafter forming a sandwich of said P-type and N-type pieces with the chromium films confronting each other as an intermediate inner layer of said sandwich, and establishing a temperature gradient across said sandwich with the temperature of said intermediate layer at a level sufficient for said chromium to melt and form a traveling liquid zone which dissolves the hotter silicon carbide on one side thereof and reforms it as an integral part of the cooler silicon carbide on the other side thereof, with a P-N junction formed in the region of the original boundary between said P-type and N-type pieces of silicon carbide.

18. A method of growing a crystalline material comprising the steps of providing two separate bodies of said crystalline material, heating each body in a vacuum to remove surface impurities and evaporating a thin adherent film of a solvent for said material on a surface of each body while in said vacuum, said solvent being a solid which melts below the melting point of said material, arranging a sandwich with said bodies as outer layers and said solvent as an intermediate layer, and establishing a temperature gradient across said sandwich sufiicient to melt said intermediate layer and form a migrating liquid zone which dissolves the body on the hotter side of said zone and allows it to reform on the cooler side as an integral part of the body on said cooler side.

19. A method of growing silicon carbide comprising the steps of providing two separate bodies of crystalline silicon carbide, heating each body in a vacuum to remove surface impurities and evaporating a thin adherent film of a solvent for said material on a surface of each body while in said vacuum, said solvent being a solid which melts below the melting point of silicon carbide, arranging the sandwich with said bodies as outer layers and said solvent as an intermediate layer, and establishing a temperature gradient across said sandwich sufiicient to melt said intermediate layer and form a migrating liquid zone which dissolves the body on the hotter side of said zone and allows it to reform on the cooler side as an integral part of the body on said cooler side.

20. A method of growing a crystalline material comprising the steps of providing two separate bodies of said crystalline material, heating at least one of said bodies in 1 1 1 2 a vacuum to remove surface impurities and depositing References Cited by the Examiner an adherent film of a solvent for said crystalline material UNITED STATES PATENTS on a surface thereof While in said vacuum, said solven hav- I ing a melting point below that of said crystalline mate- 2)789'06,8 4/ 57 Maserllan rial, arranging a sandwich with said bodies as outer lay- 5 3211552 51 5 2? ers and said solvent as an intermediate layer, establishmg 3,030,704 4/62 n n 148 185 a temperature gradient across said sandwich sufiicient to melt said intermediate layer and form a migrating liquid zone which dissolves the body on the hotter side of said DAVID RECK Pnmary Bummer zone and allows it to reform on the cooler side as an 10 integral part of the body on said cold side.

UNITED STATES PATENT OFFICE CERTIFICATE OF CORRECTION Patent No 3,205,101 September 7, 1965 Abraham In Mlavsky et al.

It is hereby certified that error appears in the above numbered patent requiring correction and that the said Letters Patent should read as corrected below.

Column 8, line 69, for "providingg" read providing column 10, lines 1 to 4, strike out "evaporating a thin film of chromium into said surfaces terminating said bombardment, arranging said bodies to form by electron bombardment under a high vacuum,"; line 9, for "intermediiate" read intermediate column 11, line ,3, for "solven" read solvent Signed and sealed this 31st day of May 1966.,

(SEAL) Attest:

ERNEST w. SWIDER EDWARD J. BRENNER Attesting Officer Commissioner of Patents

Claims

1. A METHOD OF GROWING A CRYSTALLINE MATERIAL COMPRISING THE STEPS OF PROVIDING TWO SEPARATE BODIES OF SAID CRYSTALLINE MATERIAL, HEATING EACH BODY IN A VACUUM TO REMOVE SURFACE IMPURITIES AND DEPOSITING A THIN ADHERENT FILM OF A SOLVENT FOR SAID MATERIAL ON A SURFACE OF EACH BODY WHILE IN SAID VACUUM, SAID SOLVENT BEING A SOLID WHICH MELTS BELOW THE MELTING POINT OF SAID MATERIAL, ARRANGING A SANDWICH WITH SAID BODIES AS OUTER LAYRS AND SAID SOLVENT AS AN INTERMEDIATE LAYER, ESTABLISHING A TEMPERATURE GRADIENT ACROSS SAID SANDWICH SUFFICIENT TO MELT SAID INTERMEDIATE LAYER AND FORM A MIGRATING LIQUID ZONE WHICH DISSOLVES THE BODY ON THE HOTTER SIDE OF SAID ZONE AND ALLOWS IT TO REFORM ON THE COOLER SIDE AS AN INTEGRAL PART OF THE BODY ON SAID COOLER SIDE.