Semiconductor Device With Conductive Rectifying Rods

The subject matter of this application is related to application Ser. No. 940,371 and application Ser No. 940,700 filed concurrently herewith, respectively filed Dec. 11, 1986 and Feb. 9, 1988. The latter is now U.S. Pat. No. 4,724,223.

This invention relates to semiconductor devices. More particularly, it is concerned with devices employing composites of semiconductor material and conductive materials that form a eutectic with the semiconductor material.

In conventional planar semiconductor device technology the flow of current between the source and drain electrodes of a field effect transistor (FET) is controlled by applying a gate voltage to its gate electrode. The gate voltage increases or decreases the width of the depletion layer of the gate region between the source and the drain, and thus alters the effective resistance of the device. That is, an FET is basically a voltage-controlled resistor. In order to reduce significantly the current passing between the source and the drain, the gate voltage should be able to increase the width of the depletion layer to a dimension comparable to the total width of the current path. Thus, for practical purposes, the current channel of an FET fabricated by conventional planar technology is limited to about 1 to 10 micrometers. These physical limitations of the device elements limit the amount of current which can be controlled.

An improved semiconductor device in accordance with the present invention comprises a body including a matrix of semiconductor material having disposed therein an array of individual rods of conductive material which form rectifying barriers at the interfaces of the rods and the semiconductor material A first conductive contact member is in ohmic contact with a plurality of the rods of the array at one end of each of the rods of the plurality. A second conductive contact member is in ohmic contact with the semiconductor material of the matrix.

In the drawings:

FIG. 1 is a schematic representation of a perspective view taken in cross-section of a fragment of a body which includes a matrix of semiconductor material having conductive rods disposed therein;

FIGS. 2A-2E are elevational views in cross-section illustrating various stages in the process of fabricating a semiconductor device in accordance with the present invention;

FIG. 3 is a schematic representation in perspective of a diode in accordance with the present invention; and

FIG. 4 is a schematic representation in perspective of an FET in accordance with the present invention.

In the figures the various elements are not drawn to scale. Certain dimensions are exaggerated in relation to other dimensions in order to present a clearer understanding of the invention.

For a better understanding of the present invention, together with other and further objects, advantages, and capabilities thereof, reference is made to the following disclosure and appended claims in connection with the above-described drawings.

In fabricating devices in accordance with the present invention, a charge containing the constituents of a eutectic composition of a semiconductor material and a conductive material is prepared Any of several eutectic systems may be utilized in which the semiconductor phase is the matrix phase and the conductive material is the minor phase. These systems may be, for example, of the form Ge--MGe.sub.2, Si--MSi.sub.2, and GaAs--MAs, where M is a metal. A charge containing the constituents in proper proportions to the eutectic composition of the semiconductor material and the conductive compound is melted in a suitable crucible and solidified unidirectionally.

More specifically, a Czochralski crystal growth technique is employed in which a single crystal seed of the semiconductor material is lowered into contact with the molten surface of the charge, and the seed is pulled upward. A boule consisting of a composite of the semiconductor material and the conductive material that forms a eutectic with the semiconductor material is produced in which the semiconductor material is in the form of a single crystal matrix with rods of the conductive material disposed in the matrix. The matrix of semiconductor material is of the same crystal orientation as the seed crystal and the rods extend generally parallel to the direction of pulling.

FIG. 1 illustrates a fragment of a section or a body 10 cut from a boule grown as above. The body consists essentially of a matrix of semiconductor material 11 and an array of rods 12 of the conductive material which forms the eutectic composition with the semiconductor material. As illustrated in FIG. 1, the upper and lower major surfaces of the body 10 are a result of cutting the boule transverse to the direction of pulling of the seed crystal. The side surface are formed by cuts parallel to the rods 12. Each rod 12 extends perpendicularly from the upper major surfaces to the lower major surface. The rods 12 are not necessarily of perfect circular cross-section nor are they necessarily perfectly parallel. The rods 12 are, however, each individual elements and do not interconnect, and the semiconductor matrix 11 is entirely interconnected and surrounds each of the individual rods.

The semiconductor material of the melt is appropriately doped with conductivity type imparting material of either N or P-type so that Schottky barriers are formed between the conductive rods 12 and the semiconductor matrix 11. The carrier concentration in the semiconductor material is from 10.sup.14 to 10.sup.16 cm.sup.-3. The volume fraction of the rods 12 in the body 10 should be between 0.5% and 35%. The inter-rod spacing of a composite formed by eutectic solidification is given by .lambda..sup.2 v=constant, where .lambda. is the average inter-rod spacing, v is the growth rate, and the constant is a characteristic of the particular eutectic system. The rod diameter (d) is related to the inter-rod spacing (.lambda.) by the volume fraction of the rods as determined by the eutectic phase diagram of the particular eutectic system. For suitable device structures the inter-rod spacing (.lambda.) should be between 1 and 50 micrometers, and the rod diameter (d) should be between 0.1 and 15 micrometers.

FIGS. 2A-2E illustrate stages in the fabrication of a device from a body 10 as illustrated in FIG. 1 of a matrix 11 of semiconductor material having disposed therein an array of conductive rods 12. For purposes of discussion the semiconductor matrix 11 is of single crystal silicon with conductive rods 12 of tantalum silicide (TaSi.sub.2) grown from a Si--TaSi.sub.2 eutectic composition. Examples of other suitable Si-silicide eutectic systems are Si-tungsten silicide, Si-niobium silicide, Si-zirconium silicide, and Si-chromium silicide

After suitable polishing of the surfaces of the body 10 to remove surface damage, the body is treated in wet oxygen to form an adherent layer of silicon oxide 15 on the upper major surface as illustrated in FIG. 2A. A standard photoresist material 16 is deposited on the silicon oxide coating 15. By employing known photolithographic masking and etching techniques an opening is formed in the photoresist layer 16 and then an opening 17 is etched in the oxide layer 15. The photoresist material is then removed As illustrated in FIG. 2B the end surfaces of several of the conductive rods 12 encircled by the silicon matrix 11 are exposed at the uncovered area of the upper major surface.

A film 19 of a suitable metal which forms a conductive compound with silicon, specifically cobalt, is deposited over the layer of silicon oxide 15 and also over the surface of the body at the opening 17 as illustrated in FIG. 2C. The body is then subjected to a rapid thermal annealing treatment by heating at a temperature of about 800.degree. C. for about 12 seconds to cause the cobalt in contact with silicon to react and form a conductive member 20 of cobalt disilicide as illustrated in FIG. 2D. The cobalt over the silicon oxide does not react and remains as cobalt. This unreacted cobalt is removed by a suitable chemical etch which attacks the cobalt but does not significantly affect the cobalt disilicide.

In the resulting structure as illustrated in FIG. 2E the cobalt disilicide contact member 20 forms an ohmic contact with the ends of several of the TaSi.sub.2 rods 12. The interface of the cobalt disilicide contact member 20 with the silicon of the matrix 11 is a Schottky rectifying barrier The rectifying barrier formed between the contact member 20 and the semiconductor matrix 11 should have a barrier potential which is equal to or higher than the barrier potential of the rectifying barrier between the conductive rods 12 and the semiconductor matrix 11. Platinum silicide has a high potential barrier with N-type silicon and yttrium silicide has a high potential barrier with P-type silicide. Other silicide-forming metals which may be used include nickel, titanium, tantalum, tungsten, and molybdenum.

FIG. 3 illustrates a two electrode, or diode, device fabricated from the structure as illustrated in FIG. 2E. Silicon oxide 15 is removed from a portion of the major surface of the body 10 and an ohmic contact 25, for example of gold doped with antimony for N-type silicon, is applied.

FIG. 4 illustrates an FET fabricated from the structure as illustrated in FIG. 2E. The contact member 20 in ohmic contact with several rods serves as the gate contact member. Ohmic contacts 30 and 31 are made to opposite side surfaces of the body 10 to serve as source and drain contact members, respectively Current from the source contact 30 to the drain contact 31 flows through the channel region of the silicon matrix which underlies the gate contact member 20. The channel region is interspersed with the conductive rods which contact the gate contact 20. Since the interfaces between the rods and the silicon matrix are Schottky rectifying junctions, a depletion zone is produced in the silicon adjacent to each of the rods by biasing potential applied to the gate contact 20. Each depletion zone extends from the interface of the rod with the matrix laterally outward into the matrix. Due to the lack of mobile current carriers in these zones, they have a very high resistance relative to the resistance of the normally-doped silicon of the matrix. Thus, the effective channel for current flow is determined by the spacing between the depletion zones.

The size of the depletion zones is determined primarily by the carrier concentration in the silicon and the magnitude of the bias voltage applied to the gate contact 20. The carrier concentration of the silicon of the matrix is designed such that without a bias voltage applied to the conductive rods, the depletion zones are small, consuming a sufficiently small volume of the channel region so as to have little or no effect on the resistivity of the composite and thus of current flow from the source to the drain. The growth rate (v) at which the composite is pulled from the molten charge is chosen to produce inter-rod spacing (.lambda.) such that the application of a suitable reverse bias voltage to the rods is sufficient to enlarge the depletion zones around the rods such that the depletion zones extend across the channel. The channel is thereby closed, effectively "pinching off" current flow between the source and drain and switching the device off.

A significant distinction between an FET as described and a conventional planar FET is the magnitude of the channel. In the eutectic composites as discussed hereinabove the rods are continuous from the top surface to the bottom surface of the body. Thus, the rectifying junctions extend through the entire thickness of the body producing a three-dimensional volume of semiconductor material to be controlled rather than, in effect, a two-dimensional channel as in a conventional planar FET. Since the thickness of the composite body may be orders of magnitude greater than the channel of a conventional planar FET, it is capable of switching currents which are orders of magnitude greater.

As an example of devices produced in accordance with the invention, Si--TaSi.sub.2 eutectic composites were grown directly from a melt in a Czochralski crystal growth furnace in accordance with Czochralski crystal growth techniques. A charge of silicon and tantalum was placed in a quartz crucible with a graphite susceptor. The tantalum was 5.5% by weight of the charge, providing the mole ratio of silicon to tantalum in the Si--TaSi.sub.2 composition. The charge was doped with phosphorus. After rf heating the charge to above the eutectic temperature and obtaining charge homogenization, a (111) orientation silicon seed was lowered onto the melt surface. Composite boules were pulled from the melt at a growth rate (v) of 20 cm/h in a flowing argon atmosphere. The resulting boule was approximately 2.5 cm in diameter and 12 cm in length. The silicon matrix phase was of single crystal (111) orientation and was free of grain boundaries. The boule was 2% by volume of conductive, metallic, TaSi.sub.2 rods. At a transverse-cut major surface the composite contained 1.6.times.10.sup.6 TaSi.sub.2 rods/cm.sup.2 with an average rod diameter (d) of 1.2 micrometers. The average inter-rod spacing (.lambda.) was 7.9 micrometers. The carrier concentration in the silicon matrix as determined by Hall measurements varied from 1.1.times.10.sup.15 cm.sup.-3 at one end of the boule to 1.8.times.10.sup.15 cm.sup.-3 at the opposite end. The electron mobility in the matrix was approximately 925 cm.sup.2 /V-sec.

Diodes were fabricated from 500 micrometer thick slices of the boule. The slices were polished using colloidal silica to remove damaged surface material. A 0.2 micrometer thick silicon oxide coating was grown on the surfaces. The silicon oxide coating was covered with a photoresist layer, and using standard photolithographic and etching techniques 127 micrometer diameter openings were made in the oxide coating. The photoresist was removed, and a layer of cobalt 700 angstroms thick was deposited over the silicon oxide layer and the exposed surfaces of the slice at the openings in the silicon oxide layer. The slice was treated by rapid thermal annealing at a temperature of 800.degree. C. for 12 seconds. As explained hereinabove, the cobalt reacted with the exposed silicon at the openings in the silicon oxide coating to form cobalt disilicide contacts. The unreacted cobalt overlying the silicon oxide was removed by etchings in an HNO.sub.3 solution. The cobalt disilicide provided ohmic metallic contacts to the exposed ends of the TaSi.sub.2 rods while forming Schottky barriers with the silicon of the matrix. The Schottky barriers between the cobalt disilicide and the silicon have a higher barrier potential than the Schottky barriers between the TaSi.sub.2 rods and silicon. An ohmic contact was formed adjacent to the edge of the body of each device by etching away silicon oxide and then depositing and alloying a gold-antimony film.

Based on the area of the cobalt disilicide contact member and the rod density, the effective region of a diode device contains 190 rods. For a 500 micrometer thick device the total junction area is 3.6.times.10.sup.-3 cm.sup.2. This area is approximately thirty times the area of contact of the cobalt disilicide member to the silicon matrix. Since cobalt disilicide provides a slightly higher Schottky barrier to silicon than does tantalum disilicide (0.64 eV versus 0.59 eV) the contribution of the cobalt disilicide-silicon junction to the current-voltage and capacitance-voltage characteristics is negligible.

The current-voltage and capacitance-voltage characteristics of the diodes were analyzed The diodes were ideally low in leakage and the breakdown voltage exceeded 10 volt. Breakdown was "soft" and occurred at about 30 volts. Analysis of the current-voltage characteristics based on the Si--TaSi.sub.2 junction area indicated a Schottky barrier potential of 0.62 eV. The ideality factor of the device was nearly ideal and comparable to conventional planar diodes. The capacitance-voltage analysis indicated that the junctions operated like cylindrical junctions, as opposed to planar junctions. The capacitance values verify that the entire junction area was contributing to the capacitance.

Electron beam induced current (EBIC) techniques were also used to study the junctions and the eutectic composite. The results indicated that with a 10 volt reverse bias gate voltage, a value much less than the breakdown voltage, the depletion zone around each rod can be enlarged from 0.8 micrometers to 2.5 micrometers. At 2.5 micrometers the depletion zone volume fraction of the composite is 45%, nearly enough to cause "pinch off." Resistivity measurements indicate that increasing the volume fraction of depleted material from 5% to 10% is sufficient to increase the resistivity of the device by a factor of 7. The EBIC measurement technique also verified that the rods extend completely through slices 1270 micrometers thick.

Thus, devices can be fabricated from eutectic composites produced by employing Czochralski techniques to directionally solidify a semiconductor material and a conductive compound of the semiconductor material and a metal which form a suitable eutectic composition. A three-dimensional or bulk structure is thus provided for the fabrication of an FET in which the gate structure is an array of Schottky barriers with the bulk semiconductor material. Such a device is capable of controlling large amounts of current.

While there has been shown and described what are considered preferred embodiments of the present invention, it will be obvious to those skilled in the art that various changes and modifications may be made therein without departing from the invention as defined by the appended claims.