METHOD OF MAKING A PLANAR III-V BIPOLAR TRANSISTOR BY SELECTIVE ION IMPLANTATION AND A DEVICE MADE THEREWITH
BACKGROUND OF THE INVENTION 1. Field of the Invention
This invention relates to the formation of devices comprising at least two p-n junctions in III-V materials, employing sequential ion implantation through openings in masks of dense material. This invention also relates to
III-V bipolar transistors formed by sequential ion implantation into a substrate of III-V material, employing the foregoing process; in particular, this invention relates to III-V bipolar transistors having a planar configuration 2. Description of the Prior Art
Early work in CaAs transistors focused on bipolar configurations, which have base, emitter and collector regions. However, due to early discouraging experimental results and poor reproducibility, emphasis shifted to GaAs field effect transistors (FETs), which have source, gate and drain regions. The CaAs FET technology has matured; such devices have now advanced to a level where they exhibit useful microwave performance at frequencies above 2C GHz. Success in this maturing technology has been due, in part, to ion implantation.
Attempts are under way to integrate the GaAs MESFET (metal-semiconductor FET) technology with CaAs bipolar junction and transistor devices to form functional GaAs integrated circuits (ICs) . Accordingly, emphasis is shifting back to bipolar transistors, in an effort to improve device properties. An example of
such a development is described in Vol. 16, Electronics Letters, pp. 637-638 (1980). As disclosed therein, GaAs n-p-n bipolar transistors are fabricated by sequentially implanting beryllium and selenium ions into bulk n-type substrate material. Devices with mesa collectors exhibit a DC current gain hFE of up to 8 and a reverse bias leakage current of less than 10 nA. While the device may represent a breakthrough in GaAs bipolar transistors, it should be noted that the device is of mesa configuration, not planar, and thus suffers the disadvantages characteristic of mesa-type devices. Further, integration with GaAs FETS on the same substrate to form planar ICs cannot be done with the above approach. Finally, the process for fabricating the device requires two anneals at different temperatures, the second anneal being done at 700°C, which is not adequate to anneal out implant-induced damage sufficiently to obtain good minority carrier properties in the base. This in turn requires use of an overly thin base region to achieve adequate current gain, with consequent poor saturation in the transistor.
Fabrication of planar III-V bipolar transistors requires the formation of at least two p-n junctions and reproducible control of the distance between the two junctions, namely, the base-width. Such techniques have not heretofore been available for III-V bipolar transistors.
SUMMARY OF THE INVENTION In accordance with the invention, sequentially ion-implanted p-n junctions in III-V materials are provided. The process for forming at least two p-n junctions in a III-V material includes:
(a) forming a passivating layer to protect the III-V material;
(b) forming a first mask of dense material with appropriate openings therein;
(c) implanting through the openings ions of a first species to form a first implanted region having a first conductivity type;
(d) removing the first implant mask;
(e) forming a second mask of dense material with appropriate openings therein;
(f) implanting through the openings ions of a second species to form a second implanted region within the first implanted region, the second implanted region having a conductivity type opposite to that of the first implanted region; and
(g) removing the second implant mask. The III-V material may be of a conductivity type opposite to that of the first implant. In such a case, two p-n junctions are formed. Alternatively, employing a material either of the same conductivity as the first implant or semi-insulating would require a third implantation to be done, within the second implanted region, using a third heavy metal mask provided with appropriate openings. Of course, more than two p-n junctions may be formed by additional implants within preceding implanted regions. The sequential ion implantation process provided herein is especially useful for fabricating planar III-V bipolar transistors. The bipolar transistor, which comprises emitter, base and collector, is formed by a process which includes: (a) providing a substrate of III-V material;
(b) forming on one portion of the substrate a first region, a portion of which comprises the collector, having a first conductivity type;
(c) forming a passivating layer on at least the first regio ;
(d) forming in at least one portion of the first region at least one second region, a portion of which comprises the base, said second region being formed by implanting an ionic species through appropriate openings in a first mask of dense material supported on the passivating layer to provide the second region with a conductivity type opposite to that of the first region;
(e) forming in at least one portion of the second region at least one third region, a portion of which comprises the emitter, said third region being formed by implanting an ionic species through appropriate openings in a second mask of dense material supported on the passivating layer to provide the third region with a conductivity type the same as that of the first region; (f) annealing to activate the implanted ions; and
(g) forming ohmic contacts to the emitter, base and collector regions.
The first region may be epitaxially deposited on the substrate, followed by formation of the passivating layer thereon,. as set forth above. Alternatively, the passivating layer may be formed directly on the substrate, and the collector region formed in a portion of the substrate, employing a mask of dense material in conjunction with an ion implantation, such as described above.
The masks of dense material permit ion implantation depths of 1 to 1-5 μm and high pattern resolution (e.g., 1.5 urn linewidths and less). Annealing to activate the impurities need only be done once and at a high enough temperature to obtain good minority carrier properties. Use of at least two sequential ion implantations permits formation of planar structures not heretofore successfully and reproducibiy realized. In particular, planar
III-V bipolar transistors may be fabricated. Advantageously, such planar bipolar transistors permit full integration of both bipolar and MESFET devices on one substrate.
BRIEF DESCRIPTION OF THE DRAWING
FIG. 1 depicts, in cross-section, a portion of the bipolar transistor of the invention during one stage of processing;
FIG. 2 depicts, in cross-section, the bipolar transistor of the invention at a subsequent stage in processing;
FIG. 3 depicts, partly in perspective and partly in cross-section, a fabricated bipolar transistor of the invention;
FIG. 4 depicts, in cross-section, a combination of a planar bipolar transistor of the invention and a MESFET on one substrate; and
FIG. 5, on coordinates of current in amps and voltage in volts, depicts the current/voltage characteristics of one embodiment of the bipolar transistor of the invention, employing GaAs.
DETAILED DESCRIPTION OF THE INVENTION The description which follows below is given in terms of a particular III-V material, GaAs. However, it will be understood that the planar bipolar transistor of the invention may comprise any III-V compound semiconductor, such as InP, or mixed compound semiconductor, such as (In,Ga)As or ( In,Ga) (As,P), having the requisite properties, such as high electron mobility. Further, the device described below is of n-p-n configuration. However, the corresponding complementary p-n-p configuration is also contemplated within the scope of the
invention, making suitable substitution of p-type for n-type and vice versa. Further, the sequential ion implantation process employing masks of dense material may also be employed to form other devices comprising at least two p-n junctions, such as certain junction gate charge coupled devices.
The high electron mobilities and wide bandgap of GaAs make it an attractive material for high speed, high temperature device applications. Integrated circuits employing MESFET devices have been fabricated. The speed, packing density and drive capability advantages associated with bipolar devices are attractive, and devices fabricated from GaAs are desirable. ln particular, if both bipolar and MESFET devices can be integrated into a planar IC process, the material properties of GaAs can be more fully exploited.
In order to apply ion implantation to the fabrication of GaAs bipolar transistors, however, the implanted regions will range up to about 1 to 1.5 μm in depth (for ion-implanted collector). use of a photoresist layer to mask against ion implantation requires a photoresist thickness of about 7 to 10 urn. Such thick photoresist layers are not capable of high pattern resolution (i.e., line widths of about 1.5 urn and less). Further, the doping profile of implanted regions must be controlled to about 0.1 μm. Without such control, considerable overlap of n and p regions may occur, with concomitant poor device operation.
The planar GaAs bipolar transistor described herein may be formed in a variety of ways. For example, the base and emitter regions may be implanted into n-type epitaxial GaAs, which acts as the collector. In such a case, the epitaxial layer is supported either on a highly
doped substrate of the same conductivity or on a semiinsulating substrate. Following fabrication of an array of bipolar transistors in epitaxial layers, the individual devices can be isolated by ion bombardment in the regions between devices, using ions such as protons, deuterons, tritons, boron ions or oxygen ions.
Alternatively, a semi-insulating substrate alone is employed, and the collector, base and emitter regions are implanted therein. No isolation implant is required, and parasitic capacitance is reduced.
In either case, sequential ion implantation is performed to form the two p-n junctions.
Implantations of base and emitter regions (and, where appropriate, collector regions) are performed using a thin layer of dense material deposited on the substrate to serve as an implant mask. Selective p-type base regions are formed by ion implantation with an acceptor. Ion implantation with a donor is then done to form the n-type emitter region, again employing a thin mask of dense material. The devices are annealed to electrically activate the impurities, following which a passivating dielectric layer is deposited on the surface of the epitaxial layer. Ohmic contacts to the collector and emitter regions are formed with Au:Ge/Ni, while ohmic contact to the base region is made with Ag:Mn or Au:Zn.
The use of masks of dense materials permits ion implantation depths of 1 to 1.5 uπ and high pattern resolution. Materials of high density, resulting from high average atomic number and high atomic density, can prevent the unwanted implantation of ions into the substrate using much thinner layers than would be required using photoresist. For example, a 2.0 μm thick layer of germanium will stop ions that would penetrate 1.5 μm
into GaAs, while allowing an ample safety margin. Even thinner layers of gold, tungsten, tantalum or lead would be required. The use of such thin layers permits improved lithographic resolution. The density of the mask material should be at least equal to, and preferably greater than, the density of the layer being implanted and may comprise an elemental metal or an alloy having the requisite properties.
Annealing to activate the impurities need only be done once, and at a high enough temperature to obtain good minority carrier properties. Further, the combination of the sequential ion implantation, the use of thin masks of dense material and the one-time anneal at the high temperature permit control over the implanted profiles to about 0.1 μm.
The device and process for making the device (not to scale) are more fully illustrated in FIGS. 1-3. In FIG. 1 is shown a substrate 10, here of n-type GaAs, which is a portion of a wafer having been sliced from a single crystal boule and appropriately processed. The n-type GaAs substrate may be highly doped, say to about I018 cm-3. The highly doped substrate can also act as a collector, and the heavy doping makes forming ohmic contacts easier. Alternatively, as described below, a semi-insulating substrate of GaAs may be employed. The serai-insulating substrate permits closer spacing of devices and reduces parasitic capacitance.
An epitaxial layer 11, again of n-type GaAs, is formed on one surface of the substrate and is of the same conductivity type as the substrate, having a carrier concentration of about 5 x 1015 to 3 x 10l6 cm-3. The doping level in the collector region is kept in this range in order to achieve a reasonable doping level in the base region and also achieve a high emitter injection
efficiency. The epitaxial layer is formed by conventional liquid phase epitaxial growth processes and forms no part of this invention. The thickness of the epitaxial layer ranges from about 2 to 3 μm , s ince for a th ickness greater than about 4 μm, device isolation bombardment may pose problems.
In order to perform selective area ion implantation, a passivating layer 12 is formed on the semiconductor surface to a thickness of about 200 to 600 A. Less than 200 A results in patchy growth, while more than 600 A begins to impede the ion implantation processes used below. The passivating layer also protects the ion-implanted semiconductor surface, which is highly reactive. Dielectric materials such as silicon nitride (Si3N4), silicon oxynitride (SiχOyNz) and silicon dioxide (SiO2) may be employed. Preferably, Si3N4 is employed, since it forms a cap against gallium diffusion during a subsequent annealing step. Advantageously, the passivating layer is deposited by a low temperature plasma enhanced process.
A mask 13 of dense material is then formed over layer 12. Advantageously, the mask comprises any dense material, such as gold, silver, tungsten, tantalum, lead or palladium, which is capable of masking selected areas of the semiconductor from ion penetration during ion implantation, as described above. The thickness of the mask is dictated by the density of the mask material and the energy and mass of the implanted ion. For example, the thickness of the silver masks employed in the fabrication of the bipolar transistor described herein ranges from about 3,000 to 5,000 Å. Use of gold masks would require about the same thickness.
Prior to deposition of silver, a film of aluminum (not shown) is evaporated on the surface of layer 12. The use of gold as a metal mask employs a chromium film. The metal film enhances adherence cf the silver or gold to the dielectric layer 12.
An opening 14 is then formed completely through a portion of the mask 13, such as by chemical etching or a photolithographic lift-off process. In the lift-off process, a photoresist layer (not shown) is deposited on the passivating layer prior to depositing the mask and is exposed and processed to form openings therein. Following deposition of the mask, the remaining photoresis is dissolved, resulting in lift-off of metal at the appropriate areas to form, e.g., opening 14. Acceptor ions, such as beryllium or other Group II elements such as magnesium, cadmium and zinc are then implanted to form base region 15 in the epitaxial layer. The base region is formed, to a depth of about 0.6 to 0.9 μm to form p-n junction 16. The control of the base-width and consequently the location of the base/collector junction are extremely crucial to efficient operation of the device.
The mask 13 is then removed by chemical etching using appropriate etchants such as a mixture of phosphoric acid, ethylene glycol and water, and a new mask 17 of dense material is formed on layer 12 and is processed to form openings 18 and 19, using the process described above. As is typical, the collector forms a ring around the base and emitter regions. A donor implant employing silicon, or other donors such as sulfur, tellurium or selenium, forms emitter region 20. The emitter/base junction should lie between about 0.3 to 0.5 μm from the surface, thus resulting in a base-width between about 0.3 to 0.4 μm. A second p-n junction 21 is thereby formed. To facilitate making ohmic contacts, portions 22 of the surface of the collector are doped rather heavily through openings 19 by the donor implant used to form the emitter. For transistor action, it is necessary for the emitter to be shallower than the base, with base-widths ranging from about 0.3 to 0.4 μm, and for the emitter to be completely surrounded by the base.
To remove the radiation damage and to activate the implanted impurities by restoring minority carrier lifetime, the wafer is annealed. For GaAs, annealing is advantageously carried out at about 850°C for about 30 minutes. The temperature is dictated by a desire to anneal at as high a temperature as possible, without causing diffusion of implanted dopants. The time of annealing is not so critical. Following the anneal, a passivating dielectric layer 23 is deposited. It is convenient to strip off passivating layer 12 and to clean the semiconductor surface before depositing passivating layer 23. However, layer 12 could be left on and layer 23 deposited directly over it. Conveniently, passivating layer 23 is silicon oxynitride, Siχ0 N2. Other dielectrics such as Si3 4 or Si02 may also he employed. The passivating layer is conveniently deposited by a low temperature plasma enhanced process or by other low temperature process such as by photochemical deposition.
Appropriate openings are then made in the dielectric layer 23 by chemical etching through which ohmic contacts to the n and p regions of the device are. made, as shown in FIG. 3.
Ohmic contact 24 is made to the base region employing Ag:Mn or Au:Zn, while ohmic contacts 25 and 26 to the emitter and collector regions, respectively, are made using Au:Ce/Ni. Ohmic contacts employing other materials to the p and n regions, may, of course, be alternatively used. Formation of these ohmic contacts is conventional and does not form a part of this invention. Electrical connection to external circuitry is then done, employing well-known procedures.
At least one emitter region is formed in each base region. More than one emitter region may be formed in each base region, with the ohmic contacts to base and emitter regions forming an interdigitated structure, which is useful for fabricating high frequency devices.
The device technology is a planar one, and individual devices can be isolated from each other by proton or other appropriate ion bombardment, employing conventional process technology. The foregoing process has been directed to epitaxial layers formed on highly doped substrates. In such cases, ion bombardment is employed to isolate the devices. In an alternate embodiment, a semi-insulating substrate, such as Cr-doped GaAs, may be employed. A collector region is formed in the substrate by ion implantation of donor species through a mask of dense material, employing the same procedures described above in connection with ion implantation to form base and emitter - regions. These regions are then formed as above, and the device is processed as above, except that ion bombardment for device isolation is not necessary, due to use of the semi-insulating substrate. Such a device is depicted in FIG. 4, which shows a fabricated planar bipolar transistor 40 formed in a semi-insulating sub- strate 10'. An ion-implanted, region 11' serves as the collector; the remaining numbers reference like-numbered parts in FIG. 3. Again, two p-n junctions, 16 and 21, are formed by sequential ion implantation.
An n-type collector may alternatively also be implanted into a p-type substrate, if desired. Again, the ion implantation in conjunction with a patterned mask of dense material is employed.
The planar process described herein permits full integration of bipolar transistors with planar FET structures, such as MESFETs, MISFETs (metal-insulatorsemiconductor FETs) and JFETs (junction FETs). In particular, the planar process disclosed herein permits integration of bipolar devices with the MESFET technology presently available to fabricate high speed signal processing circuits such as microwave analog-to-digital
converters. The planar bipolar devices of the invention are particularly useful as elements used for bias current cancellation in differential amplifiers which, are vital to the operation of analog-to-digital converters. FIG. 4, in cross-section, depicts a combination of a bipolar transistor and a MESFET device on one substrate. For the bipolar device 40, the emitter, base and collector are as described above, while the MESFET device 41 comprises a conventional source contact 42 and drain contact 43 formed over ion-implanted regions 45 and 46 having n-type conductivity; gate contact 44 is formed over channel 47, which connects the two regions. To achieve the required device isolation and to reduce parasitic capacitance, a semi-insulating substrate 10' is employed. EXAMPLE
A device substantially as shown in FIG. 3 was constructed. An epitaxial layer of n-GaAs about 3 μm thick and doped to about 1016 cm-3 was formed on one surface of an n+-GaAs substrate, about 250 μm thick, doped to about i018 cm-3. A passivating film of Si3N4 o about 400 A thick was formed on the epitaxial layer by plasma enhanced deposition process, followed by deposition and patterning of photoresist. Next, an aluminum film about 400 A thick and a silver mask about 4,000 A. thick were deposited. Portions of the silver mask were removed by photolithographic lift-off techniques and Be ions were implanted at multiple energies into a portion of the epitaxial layer to form a p-type region uniformly doped to about 5 x 10l6 cm-3 to a depth of about 0.9 μm. The metal mask was removed and a second silver over aluminum mask, again about 4,000 Å thick, was formed on a new patterned photoresist formed on the Si3N4 film. Portions of the silver mask were removed by lift-off, and Si ions were implanted at multiple energies into a portion of the
previously Be-implanted region to form an n+-type region uniformly doped to about 2 x 1018 cm-3 to a depth of about 0.4 to 0.5 μm. Si ions were also implanted at the same time into portions of the epitaxial layer to form n+-type regions therein. The silver mask was removed by chemical etching. A layer of Si02 about 1,500 Á thick was deposited on the Si3N4 layer and on the backside of the substrate. To remove radiation damage and activate the implanted impurities, the wafer was annealed at 850°C for 30 min. The SiO2/Si3N4 layers were removed and a passivating layer of Siχ0yNz was formed on the epitaxial layer by plasma enhanced deposition process. Appropriate openings were made in the passivating layer by etching in buffered hydrofluoric acid, and alloyed ohmic contacts were made, employing Ag:Mn for the p-type regions and AutGe/Ni for the n-type regions.
The epitaxial layer served as the collector through the implanted n+-type regions; the p-type region was the base and the n+-type region therein was the emitter. Completed devices were tested using a curve tracer. FIG. 5 presents the current/voltage (I/V) characteristics of the bipolar transistor. The vertical (current) scale is in 0.5 mA/division; the horizontal (voltage) scale is. in 1 V/division; the base current is 0.1 mA/step. Typical devices exhibited breakdown voltages in excess of 40 V across the collector-base junction and common emitter current gains of about 6. Gains as high as 16 were observed.