WO1987000693A1

WO1987000693A1 - Detector and mixer diode operative at zero bias voltage and fabrication process therefor

Info

Publication number: WO1987000693A1
Application number: PCT/US1986/001445
Authority: WO
Inventors: Mark P. Zurakowski
Original assignee: Hewlett-Packard Company
Priority date: 1985-07-12
Filing date: 1986-07-10
Publication date: 1987-01-29
Also published as: DE3650250D1; EP0235248A1; EP0235248B1; JPS63500347A; JPH0773130B2; EP0235248A4

Abstract

An NPIN (or PNIP) diode structure (10) and epitaxial process for fabricating same wherein the thickness and doping levels of the intermediate layers (16, 18) of the structure are such that these layers are substantially depleted of majority carriers and therefore enable the structure to be operated at zero volts DC bias. This structure may be utilized either as an efficient detector diode or a mixer diode substantially free of odd order harmonic mixing products, and both devices may be fabricated in a single molecular beam epitaxial process with the advantage of high control over epitaxial layer thickness and impurity concentration.

Description

DETECTOR AND MIXER DIODE OPERATIVE AT ZERO BIAS VOLTAGE AND FABRICATION PROCESS THEREOF

Technical Field This invention relates generally to semiconductor diode detectors useful for microwave and millimeter wave applications, and more particularly to such detectors which are operative at high detection efficencies with zero bias voltage. Additionally, the multi-functional structure of the diode detector disclosed and claimed herein is uniquely adapted to operate in the anti-parallel mixing mode with an identical reversely poled structure to form a highly symmetrical mixer diode pair, also operable with zero volts bias and substantially free of odd order harmonic mixing products.

Background Art

In the field of microwave and millimeter detection, it is a common practice to provide a predetermined bias voltage on a semiconductor diode in order to set the DC operating point on a nonlinear region of the diode's current-voltage (I-V) characteristic in order to provide for maximum detection efficiency. Typically, this DC bias voltage is on the order of 0.7 volts where the diode is highly conductive in a range on one side of the DC bias voltage point and rather lighty conductive in a range on the other side of this DC bias voltage.

There are, however, several significant disadvantages in the requirement for a DC bias voltage on the diode detector. First, the requirement per se of DC bias circuitry between a power supply and the detector diode adds cost and complexity to the detector arrangement. Secondly, there is a certain amount of noise associated with the DC bias voltage, and this noise degrades the sensitivity and decreases the dynamic range of the detection. Thirdly, the bias voltage is thermally sensitive and will therefore frequently cause the DC bias operating point on the diode's I-V charac teristic to be shifted in response to changes in ambient temperature at the DC bias source.

Similarly, when employing mixer diodes, it has been a common practice to provide a DC bias voltage across the mixer diode or diodes in order to establish a desired operating point on the I-V characteristic of these diodes. One such mixer diode structure is disclosed for example by Malik in U.S. Patent 4,410,902. However, an additional problem in the Malik mixer structure arises from the fact that there will be some extraneous and undesirable doping in the body of the Malik structure from impurities in the substrate moving upwardly into the epitaxial layers thereon. This doping results in an unevenly distributed and extraneous impurity profile across these layers, and this profile in turn produces dissimilar and asymmetrical I-V curves in the first and third quadrants of the device's comosite I-V characteristic. Such asymmetrical I-V characteristics ultimately result in the generation of unwanted odd harmonic signals of the fundamental mixing frequency.

For a further discussion of this problem of extraneous doping in mixer structures of the type disclosed by Malik in U.S. Patent 4,410,902, reference may be made to an article by S. C. Palmateer, et al., entitled "A study of substrate effects on planar doped structures in gallium arsenide grown by molecular beam epitaxy", Institute Physics Conference. Serial Number 65: Chapter 3, presented at the International Symposium of Gallium Arsenide and Related Compounds, Albuquerque, New Mexico, 1982, at page 149 et seq.

In the field of diode detection there has been at least one attempt to provide a detector diode which operates with zero bias. Such an attempt is evidenced for example in U.S. Patent 3,968,272 issued to Anand. However, the Anand device relies upon the reaction of a semiconductor surface with certain metals in a controlled manner. It is well known that such semiconductor surface chemistry is difficult to control and this fact will in turn affect device yields and repeatability of results. Additionally, using the Anand process stable barrier heights have only been demonstrated in silicon which has a lower electron velocity than gallium arsenide, and thus operates at slower speeds than GaAs. Also, silicon has a higher parasitic capacitance than GaAs, a fact which further contributes to the slower speeds of silicon devices.

Disclosure of Invention

The general purpose of the present invention is to provide a new and improved detector diode which is operative at zero volts bias with a high detection efficiency and which therefore overcomes the aforedescribed disadvantages of the prior art. An additional purpose of this invention is to simultaneously provide a diode structure which is uniquely suited and adapted to operate with an identical, reversely poled diode structure in an anti-parallel mixing mode, having a highly symmetrical I-V characteristic and substantially free of odd order harmonic generation. Thus, this multi-functional diode structure may advantageously be constructed in a single wafer fabrication process which, in a preferred embodiment, employs molecular beam epitaxial growth to achieve high quality, high purity, and high thickness control in the multiple epitaxial layers. The resultant diode structure may be operated either as an efficient detector diode, or it may be connected in the anti-parallel mode with an identical mixer diode to exhibit a highly symmetrical composite I-V characterisitic necessary for the prevention of odd harmonic frequency generation.

To accomplish the above purposes, I have discovered and developed a new and improved diode structure which includes an intrinsic or substantially intrinsic (or lightly doped) layer of semiconductor material of predetermined thickness upon which a thin critical layer of one conductivity-type semiconductor material is epitaxially formed sufficiently thin so that it is fully depleted of majority carriers therein. The intrinsic (or lightly doped) layer must also be substantially depleted of majority carriers in order to provide optimum symmetry of the I-V char aσteristic of the structure. And additional layer of opposite conductivity-type semiconductor material is exitaxially formed adjacent the thin critical layer and should be of sufficient thickness and impurity concentration so that it is not fully depleted of carriers during operation. This layer also provides a good barrier between an outer ohmic metal contact and the thin critical epitaxial layer of the one conductivity-type. Preferably, all of the above layers are formed on a substrate layer in successive steps of controlled molecular beam epitaxy.

The above purposes, advantages, and other novel features of this invention will become better understood in the following description of the accompanying drawings wherein:

Brief Description of the Drawings

Figures 1A and IB illustrate the basic structural configuration of the novel detector/mixer diode according to the invention.

Figures 2A and 2B illustrate the majority carrier potential barrier extending across the various layers of the diode.

Figures 2C and 2D illustrate the increasing majority carrier potential across the diode structure Figure 2A for forward bias and reverse bias conditions, respectively.

Figures 3A through 3D illustrate respectively four (4) alternative structural embodiments of the invention wherein either the exact position or the conductivity type of the thin critical layer are different in each structure, but are operative to functionally accomplish the same detection or mixing function to be further described.

Figures 4A and 4B show a diode detection network (and associated I-V characteristic) utilizing the diode according to the invention.

Figures 5A and 5B show an anti-parallel diode mixer pair (and associated I-V characteristic) for the mixer diode according to the invention. Best Mode For Carrying Out The Invention Referring now to Figures 1A and 1B, the diode structure is indicated generally as 10 and includes a semi- insulating gallium arsenide (GaAs) substrate 12 which is typically 10⁷ to 10⁸ ohm-centimeters in resistivity and 8 to 10 mils in thickness. An N type epitaxial layer 14 is deposited on the substrate 12 and is typically 5000 Angstroms in thickness and has a doping concentrati .on of about 10¹⁸ silicon dopant atoms per cubic centimeter. The next layer 16 is an intrinsic or "I" layer of approximately 3000 Angstroms in thickness and typically has a doping concentration of about 10¹⁴ dopant atoms per cubic centimeter.

The intrinsic layer 16 has a thin P type layer 18 of epitaxial material deposited on its upper surface, and the P type layer 18 is also referred to herein as the critical layer. This layer 18 may range in thickness between 25 and 140 Angstroms, but will normally be about 40-50 Angstroms in thickness, and thus sufficiently thin so as to be fully depleted of minority .carriers (or electrons) in this structure. The dopant concentration of layer 18 will be about 8x10¹⁸ beryllium atoms per cubic centimeter. A top N type layer 20 is epitaxially deposited as shown on the top surface of P type layer 18, and this upper layer 20 will typically be about 3000 Angstroms in thickness and will have a carrier concentration of about 5X10¹⁸ silicon atoms per cubic centimeter.

The type of semiconductor material of the structures described herein is not critical, and such type only determines the potential barrier height range that can be achieved for these structures which are all formed preferable using molecular beam epitaxial (MBE) deposition processes. Such known and commercially available computer controlled MBE processes are preferred in that they are presently capable of providing the best repeatable layer thickness control and doping uniformity of any presently available epitaxial processes. Although the critical layer 18 may range in thickness from 25 to 140 Angstroms, it is preferred that it be 40 to 50 Angstroms in thickness, with a maximum acceptable tolerance of + 5 Angstroms.

In the MBE best mode process used in constructing the devices disclosed and claimed here, both silicon and beryllium were alternated in the MBE epi system for N and P type doping respectively.

The presently known best mode for carrying out the MBE epitaxial deposition according to the invention is the use of an epitaxial reactor known in the trade as the VARIAN GEN-2 available from VARIAN ASSOCIATES of Palo Alto, California, with MBE shutter control provided by a Hewlett Packard HP1000 computer. For convenience, a summary table of the various layer thickness or thickness range and doping levels is given below for the device structure of Figure 1A. It will be understood, however, that these doping levels and thickness apply equally to corresponding layers of the devices subsequently described with reference to Figures 3A3D.

For a further discussion of this MBE technology, reference may be made to Proceedings of the Fifth Molecular Beam Epitaxy Workshop edited by John R. Arthur, attended 6-7 October, 1983, Georgia Institute of Technology, Atlanta, Ga. and published for the American Vacuum Society by the American Institute of Physics, New York 1984, incorporated herein by reference.

The semiconductor device of Figure 1A has electrons as its majority carriers, but if N layers are changed to P layers and vice versa, the majority carriers become holes. Ohmic or non rectifying contacts 22 and 24 are made to the N type layers 20 and 14 respectively in Figure 1A, and the diode structure is etched in a mesa-like configura tion with the side wall contours shown and isolated physically on the N layer 14 from the ohmic contact 24. However, if a conducting substrate 12 is used instead of a semiinsulating substrate material, then the N layer 14 can be omitted and the intrinsic layer 16 positioned directly atop the proposed or alternative conductive substrate material (not shown). The exact mesa type configuration of the NPI structural portion of Figure IB may be determined by mesa etching, ion milling or other similar known processing methods.

The proper thickness and doping levels of the above described semiconductor layers of the diode structure are essential to optimum device operation. The top layer 20 needs to be thick enough so that it is not fully depleted of majority carriers during operation, and it also must be sufficiently thick to provide an adequate barrier between the ohmic contact 22 (which diffuses partially into the top N layer during alloying) and the thin critical P layer 18. If the N layer 20 doping level is sufficiently high, non alloyed contacts can be utilized and the diffusion of the metal will not be a problem. The thickness and impurity doping level of the P type layer 18 and the impurity doping level in the top N layer 20 are all determinative of the height of the potential barrier at the PN junction between layers 18 and 20. This condition is quite different from the planar doped barrier diode of the above identified Malik U.S. Patent 4,410,902 where the potential barrier height is determined by the P layer doping, the P layer thickness and the thicknesses of the two (2) intrinsic layers used therein. The P layer 18 in Figures 1A and IB needs to be sufficiently thin so that it is fully depleted of majority carriers, and the intrinsic or lightly doped layer 16 must be appreciably depleted of majority carriers. so as to give optimum asymmetry to the I-V characteristic of the diode.

The thickness and doping level of the bottom N layer 14 is not critical as long as a good ohmic contact can be made to this layer and as long as the layer 14 is thick enough and sufficiently doped to provide a low series resis tance on the order of 0.5 to 10 ohms. In fact, these devices can be made on a conductive substrate, in which case no bottom N layer 14 is needed.

Referring now to Figures 2A through 2D, the diagrams shown in these figures illustrate the operation of the diode structure according to the invention. The shaded portion of Figure 2A indicates the depleted regions of the various epitaxial layers at zero bias, and the diagram of Figure 2B illustrates the varying majority carrier potential gradient horizontally across the structure of the device. Since the intrinsic layer 16 is relatively large in thickness and high in resistance compared to any other region of the device structure, most of the potential drop occurs across this region 16. Under forward bias, the depleted region of the top N layer 20 grows at the expense of the depleted region of the bottom N layer 14 thereby causing the bottom N layer 14 to increase its potential relative to that of the top N layer. This characteristic allows electrons to flow from region 14 to region 20.

Figure 2C illustrates the increasing majority carrier potential across the diode structure under forward bias conditions, whereas Figure 2D illustrates the increasing majority carrier potential across the diode structure under reverse bias conditions. The difference between forward and reverse bias is that it takes more voltage to acheive the same current level in reverse bias because most of the voltage is dropped across the intrinsic layer 16. Only a small fraction of the applied voltage is available to change the potential of layer 20, whereas most of the voltage is available to change the potential of layer 14.

Referring now to Figures 3A through 3D, there are shown respectively four (4) structural modifications of the diode structure in Figure 1B, and all of these diode structural modifications in Figures 3A through 3D are alternative embodiments of the invention. All of these alternative embodiments serve to establish the potential barrier where q is the charge on an electron. This value of is somewhere between zero voltage and the semiconductor bandgap voltage.

In Figure 3A the critical P layer has been moved from the top side of the intrinsic layer (Figure 1B) to the bottom surface thereof where it separates the intrinsic layer and the bottom N type layer of the device.

In Figure 3B, the diode structure has been modified to eliminate the bottom N type layer and dispose the intrinsic layer directly on the N type conductive substrate as shown.

In Figure 3C, the conductivity types and vertical geometry of the various layers of Figure 3A have been reversed, so that the critical layer is now an N type layer positioned between a top P type layer and the intrinsic layer as shown.

Finally, in Figure 3D the conductivity types of the layers of Figure 3A have been reversed in polarity as indicated.

The quantity

is the zero bias barrier heightin volts to overcome the potential barrier at the PN junction of the above devices and thus essentially turn on these devices to a fully conductive state. In Figure 2A and 2B for example, this barrier height is a potential somewhere between zero and the semiconductor bandgap voltage and will typically be about 0.25 volts. For this structure, it can be shown that:

where e_s is the dielectric constant of the semiconductor material, N_A is the doping concentration of the P+ critical layer 18, N_D is the doping concentration of the N+ layer 20, and t is the thickness of the critical layer 18.

Diode Detection Referring now to Figures 4A and 4B, there is shown a conventional diode detector network including an input impedance matching resister R, typically of about 50 ohms, a diode D, and an output capacitor C for developing a detection voltage envelope thereacross in accordance with well known detection principles . As is also well known, the detection efficiency of the diode D is directly proportional to the degree of asymmetry of its I-V characteristic as indicated in Figure 4B. Thus typical detector diodes of the prior art will have an I-V characteristic as indicated by the curved dotted line in Figure 4B and will have a very large video resistance, Rv, at zero basis on the order of about 600,000 ohms. Since it is generally accepted that this video resistance, Rv, must be on the order of 1000 ohms for maxiumum detection sensitivity, then these detector diodes of the prior art are typically DC biased to about 0.7 volts in order to reduce the video resistance to 1000 ohms.

However, as indicated by the solid dl/dV curve which is the I-V characteristic of the present invention, the video resistance at zero volts DC bias is about 1000 ohms, and thus no separate applied DC bias is required. It is to be understood however. that these comparisons to the prior art diodes are based upon the same very small diode area of about 20 square microns of anode surface area necessary to keep junction capacitance at a minimum value on the order of 10^-14 farads.

Diode Mixing When the diode structure according to the invention is. to be used in a mixer circuit, two (2) of these identical devices are connected in parallel and in reverse polarity as shown in Figure 5A.in the well known "antiparallel" mixing mode of operation. The connection of reverse poled diodes in such an anti-parallel diode network is well known in the art, and the advantage presented by the present invention is that two substantially identical mixer diodes fabricated side by side on a common semiconductor wafer will have substantially identical I-V characteristics. Thus, the net overall I-V characteristics of these two identical diodes connected in parallel is a completely symmetrical curve as shown in Figure 5B whose first quadrant contour is an identical match to its third quadrant contour, resulting in the complete elimination of mixing with odd order harmonics of the fundamental mixing frequency. This feature has the effect of lowering device noise inasmuch as it enables one to detect very small imput signals on a spectrum analyzer. Additionally, since the number of mixing products is decreased, it now becomes easier to identify the desired mixing product. This complete symmetry of I-V characteristics of the anti-parallel diode mixer pair according to this invention is not acheivable by way of the above identified Malik structure by reason of the extraneous and unwanted impurity doping previously described.

Thus, there has been demonstrated a new and improved semiconductor device, when used as a broadband nonbiased detector, has a 5db greater dynamic range and improved flatness than any known similar device on the market. The range of operation of this device has been demonstrated from DC to 110 GHz; however higher operational frequencies are expected.

When this device is used in the anti-parallel mixer configuration at millimeter wave frequencies as a high harmonic (greater than the 10th harmonic) mixer, no DC bias is needed to obtain state of the art conversion efficiencies across an entire waveguide band. This fact has been demonstrated through W band (110 GHz). Other mixers on the market today require a variable DC bias to obtain efficient mixing across these millimeter wave bands.

The present invention is not limited to the use of GaAs, and alternatively may be carried out using gallium phosphide (GaP), indium gallium arsenide (InGaAs) or other equivalent semiconductor materials or even silicon epitaxy in cases where the slower speeds of silicon are acceptable for certain applications.

Finally, the present invention is not limited to the use of MBE epitaxial processes and alternatively may employ other epitaxial processes which are capable of repeatably controlled epi growth of plus or minus (+) 25 Angstroms with uniform doping concentration. Such control and uniformity may be achievable using certain state off the art organo-metallic vapor phase epitaxial (OMVPE) processing techniques known and available to those skilled in the art.

Industrial Applicability This invention finds application in the field of electronic instruments, particularly of the test and measurment and millimeter wave type, in a wide variety of diode mixing and detection operations.

Claims

1. A diode structure operative at zero volts DC bias to provide both good detection efficiency in diode detector applications and to provide good harmonic suppression in anti-parallel mixer diode applications, comprising: a. an intrinsic or substantially intrinsic semiconductor layer of a predetermined thickness and impurity concentration and substantially depleted of majority carriers, b. a critical thin layer on one conductivity type disposed on one surface of said intrinsic layer and being sufficiently thin and of an impurity concentration level sufficient so that it is also substantially depleted of majority carriers, c. a layer of opposite conductivity type semiconductor material disposed on the surface of said critical thin layer and of a thickness sufficient to shield metal impurities from said critical thin layer, d. a first ohmic contact on said layer of opposite conductivity type semiconductor material, and e. a second ohmic contact disposed on either a conductive substrate member or a semiconductive substrate member, either of which are in electrical contact with said intrinsic layer, whereby the potential barrier height,

of said structure is established at a minimum value somewhere between zero volts and the bandgap voltage of the semiconductive material of said layers.

2. The structure defined in claim 1 wherein said layers of semiconductor material are formed by highly controlled molecular beam epitaxial growth.

3. The structure defined in claim 1 wherein said critical thin layer may range between 25 and 140 Angstroms in thickness.

4. The structure defined in claim 3 wherein said intrinsic layer is between 500 and 15,000 Angstroms in thickness.

5. The structure defined in claim 4 wherein said opposite conductivity type layer is on the order of 3,000 Angstroms in thickness.

6. The structure defined in claim 5 wherein the doping level in said intrinsic layer is on the order of 10¹⁴ atoms/cc, the doping level in said critical thin layer is on the order of 8x10¹⁸ atoms/cc and the doping level in said opposite conductivity type layer is on the order of 5X10¹⁸ atoms/cc.

7. The structure defined in claim 6 wherein all of said layers are deposited using molecular beam epitaxy.

8. A method for detecting microwave or millimeter wave signals which includes connecting the detector structure according to Claim 1 in a detection circuit having an output network for developing a voltage envelope proportional to an input signal, and requiring no DC bias for said detector structure.

9. A method for mixing microwave or millimeter wave signals which includes connecting two (2) diodes of the type defined in claim 1 in a reverse poled anti-parallel circuit configuration, whereby said diodes provide a composite highly symmetrical I-V characteristic sufficient for suppressing or substantially eliminating odd harmonic frequencies of the fundamental frequency of the incoming signal applied to said anti-parallel pair.

10. A process for fabricating a diode structure which may be operated with good detection efficiency at zero volts DC bias or which may alternatively be operated as a mixer diode free of odd order harmonic mixing products, also with zero volts DC bias, including the steps of: a. providing an intrinsic or substantially intrinsic semiconductor layer of predetermined thickness and impurity concentration and substantially depleted of majority carriers; b. epitaxially depositing a critical thin layer of one conductivity type on one surface of said intrinsic layer and of a chosen thickness in the range in the range of 25 to 140 Angstroms and of an impurity concentration sufficient to insure that this layer is substantially depleted of majority carriers; c. depositing a layer of opposite conductivity semiconductor material on the surface of said critical thin layer and of a thickness sufficient to shield metal impurities from said thin layer; d. depositing a first ohmic contact on said layer of opposite conductivity type semiconductor material; and e. depositing a second ohmic contact on either a conductive substrate member or a semiconductive substrate member, either of which are in electrical contact with said intrinsic layer, whereby the potential barrier height, SB, of said structure is established at a minimum value somewhere between zero volts and the bandgap voltage of the semiconductive material of said layers.

11. The process defined in claim 10 wherein said semiconductor material layers are epitaxially deposited using highly controlled molecular beam epitaxial deposition.

12. A process for minimizing the potential barrier height in NPIN or PNIP diode structures which includes the steps of: a. depleting or substantially depleting an intrinsic or I layer of said structure of majority carriers, b. providing a critical thin P or N layer on said I layer and also depleted of majority carriers, and c. providing an N or P layer respectively on the outer surface of said critical thin P or N layer respectively and of a thickness sufficient to shield metal impurities from said critical thin layer.

13. The process defined in Claim 12 wherein said critical thin layer is deposited to a thickness in the range of between 25 and 140 Angstroms.

14. The process defined in Claim 13 wherein said intrinsic or I layer is deposited between 500 and 15,000 Angstroms in thickness.

15. The process defined in Claim 14 wherein the doping level in said intrinsic layer is established on the order of about 10¹⁴atoms/cc and the doping level of said critical thin layer is established on the order of about 10¹⁸ to 10¹⁹ atoms/cc.

16. The process defined in Claim 12 wherein said intrinsic layer, said critical thin layer and said N or P layer are all deposited using molecular beam epitaxial deposition.

17. The process defined in Claim 13 wherein said intrinsic layer, said critical thin layer and said N or P layer are all deposited using molecular beam epitaxial deposition.

18. The process defined in Claim 14 wherein said intrinsic layer, said critical thin layer and said N or P layer are all deposited using molecular beam epitaxial deposition.

19. The process defined in Claim 15 wherein said intrinsic layer, said critical thin layer and said N or P layer are all deposited using molecular beam epitaxial deposition.

20. The process defined in Claim 19 wherein the PNIP or NPIN layers are comprised of either gallium arsenide, gallium phosphide or indium gallium arsenide.

21. A diode structure of either NPIN or PNIP layered configuration and having a minimum potential barrier height between zero volts and the semiconductor bandgap voltage, characterized in that the I or intrinsic layer is depleted or substantially depleted of majority carriers, and the interior P or N layer is between 25 and 140 Angstroms and also depleted of majority carriers.

22. The structure defined in Claim 21 wherein said intrinsic layer is between 500 and 15,000 Angstroms in thickness.

23. The structure defined in Claim 22 wherein said intrinsic layer has an. impurity concentration on the order of about 10¹⁴ atoms/cc and said interior P or N layer has an impurity concentration on the order of 10¹⁸ - 10¹⁹ atoms/cc.

24. The process defined in Claim 23 wherein said PNIP or NPIN layers comprise either gallium arsenide, gallium phosphide or indium gallium arsenide.