WO1981001073A1

WO1981001073A1 - Semiconductor devices controlled by depletion regions

Info

Publication number: WO1981001073A1
Application number: PCT/US1980/001234
Authority: WO
Inventors: W Cardwell
Original assignee: W Cardwell
Priority date: 1979-10-09
Filing date: 1980-09-24
Publication date: 1981-04-16
Also published as: IT1128701B; BR8008860A; EP0036887A1; EP0036887B1; EP0036887A4; US4698653A; JPH0468774B2; DE3071863D1; CA1153479A; IT8049846A0; JPS56501306A

Abstract

A field effect semiconductor device and method of controlling the device by merged depletion regions are provided. The device includes, in combination, first (22) and second (24) spaced apart PN junctions. Depletion regions (36, 38 and 40, 42) associated with the junctions have boundaries displaced from their respective junctions as a function of the doping concentration on either side of the junctions. The junctions are spaced apart by a distance with allows overlap of the depletion regions (38, 40) positioned therebetween. By applying a reverse bias to one (22) of the PN junctions the conductivity on the side (34) of the second PN junction (24) remote from the first PN junction (22) can be varied through the effect of merged depletion regions.

Description

SEMICONDUCTOR DEVICES CONTROLLED BY DEPLETION REGIONS

BACKGROUND OF THE INVENTION

The field of the invention is semiconductor devices and particularly junction field-effect devices controlled by merged depletion regions.

Modification of basic junction field-effect devices to improve performance and increase their applicability to integrated circuits has been an area of active investigation. However, in all junction field-effect devices of the past, gate control of the transistor has been provided by using a semiconductor material having an impurity type which is opposite to that which has been utilized for the drain and source electrodes and conse¬ quently the channel . This has required that the gate region be isolated to be useful in integrated circuits. In addition, it has prevented the construction of enhancement mode junction field-effect transistors; that is, devices whose channel current is zero with the gate and source electrodes connected together.

Devices of the past have been controlled by changes in the applied potential across junctions. It is not believed that any prior art device has been disclosed or suggested in which merged depletion regions are used for isolation and which is operated and controlled by merging depletion regions or by otherwise affecting the charge balance across the channel to substrate junction. Nor is any prior art device known which has all connections formed of the same conductivity type material with no intervening isolating regi ons . SUMMARY OF THE INVENTION

The invention incorporates the application of two new methods for operating semiconductor devices, merged depletion regions and charge balance control. These two methods are used for many new semiconductor devices and are combined to yield a new type of junction field-effect transistor that can be manufactured in integrated form without the use of isolation regions. The invention is better suited to integrated circuits than any other devices because all connections can be formed of the same conduc¬ tivity type material . No metal contacts, electrodes or metal interconnections are required to create integrated circuitry, which permits the construction of true three- dimensional circuits. The lack of the need for metal interconnections makes the invention suitable to the creation of ^'denser forms of logic than are known today.

The term "substrate" as used hereinbelow refers to the semiconductor material surrounding or between regions of semiconductor material of different conductivity type. In other words, the substrate is the bulk semiconductor material through which external current does not normally flow. For example, a device can be fabricated on a wafer having an N-type surface layer. A P-type layer is formed on the N-type surface layer, and N-type regions are formed on or in the P-type layer. The N type layer and N type regions are used for external connections such as the gate, source, drain and channel of a JFET; the P type epitaxial layer is defined as and referred to as the "substrate". Using the invention, a semiconductor device can be constructed in a configuration which is similar to a conventional junction field-effect transistor in that it employs a drain electrode and source electrode connected by a channel of one conductivity type. However, the material surrounding the channel forms a substrate instead of the gate of the device. A gate region of the same conductivity type material as the channel is placed near enough to the channel so that when the gate to substrate junction is reverse biased, the gate depletion region merges with the channel depletion region in the substrate. Gate control of the channel junction depletion region in the channel 'is accomplished by altering the channel junction depletion region in the substrate. Using the method of charge balance depletion region width control, the gate control of the channel depletion region in the substrate is reflected in the depletion region in the channel . Consequently, the gate controls the channel current in a manner similar to a conventional junction field-effect transistor. It also permits the device to be operated in an enhancement mode where, since the gate and channel are isolated, the drain and source are biased with respect to the substrate to turn off the device while the gate to substrate bias is used to turn on the device. This is possible because the gate electrode is isolated from the channel by the merged depletion region. As a discrete device, the junction field-effect transistor controlled by merged depletion regions is characterized by lower noise and lower feedback capacitance than any other type of device and, with a submi crometer channel length obtained in a vertical structure, a cutoff frequency well above all other known transistor types is obtained. Because the gate is separated from the channel , high-voltage operation is possible while the self- ballasting nature of field-effect devices permits large-area, high-current devices to be constructed. As in the integrated form, the separation of the gate and channel permits operation of the device in an enhancement mode and, by changing the gate to channel spacing and/or the impurity concentrations, the threshold voltage can be changed.

Properly constructed, one embodiment of the device, referred to as a linear device and characterized by equal doping in the channel and gate regions, has a low or lower distortion than any other device. A second embodiment of the device, referred to as a quadratic device and charac¬ terized by unequal doping in the channel and gate regions, has lower threshold voltage variations with temperature than other devices while its distortion is comparable to other junction field-effect transistors. In addition, both versions are as radiation resistant and as rugged as other junction field-effect devices.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 represents a method of depletion region width control of the depletion region on one side of a junction; FIG. 2, comprising FIGS. 2A and 2B, di agrammatically illustrates a top view and a sectional side view, respec¬ tively, of a junction field-effect transistor which can be controlled by merged depletion regions in the substrate;

FIG. 3, comprising FIGS 3A-3C, di agrammatically illustrates a sectional view of a semiconductor device controlled by depletion regions showing the depletion regions and demonstrating the principles of operation particularly as it relates to the merging and interaction of the depletion regions. FIGS. 3B and 3C represent the electric field (E) and potential associated with the E field, respectively, for the operations described in connection with FIG. 3;

FIG. 4, comprising FIGS. 4A-4C, di agrammatical ly illustrates a sectional view of the device of FIG. 3 further demonstrating the principles of operation particularly as they relate to the merging and interaction of the depletion regions. FIGS. 4B and 4C represent the E field and potential associated with the E field, respec¬ tively, for the operation described in connection with FIG. 4; FIG. 5, comprising FIGS. 5A-5C, di agrammatically illustrates a sectional view of the device of FIG. 3

_/< RI

OMPI further demonstrating the principles of operation particularly as they relate to the merging and interaction of the depletion regions. FIGS. 5B and 5C represent the E field and potential associated with the E field, respec¬ tively, for the operation described in connection with FIG. 5; and

FIG. 6, comprising FIGS. 6A-6T, shows in FIGS. 6A-6D several configurations of a vertical junction field-effect transistor for use in integrated circuitry. FIGS. 6E-5T graphically show the changes in E field of the device due to changes in configuration, doping, and applied bias.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

For purposes of illustration and description the method of operating a semiconductor device by merging depletion regions in the substrate of the device is first discussed; next follows a description of the method of operating a semiconductor device by charge balance depletion region width control , that is, altering the depletion region on one side of a junction and using the charge balance across the junction to control the width of the depletion region on the other side of the junction. To aid in this illustration and description a discrete device is disclosed, a junction field-effect transistor having all connections -- gate, source and drain -- of the same conductivity type material with a substrate of opposite conductivity type material . New insights into the channel operation of such a device are presented. Depletion region width control in the channel is achieved by charge balance; the depletion region in the substrate associated with the channel junction is altered by merging in the substrate the gate junction depletion region with the channel junction depletion region. The following subscripts are used for purposes of illustrating the relationships between the various -δ-

sections: c (channel ), g (gate), and s (substrate). However, it is not intended that such subscripts should limit the invention as claimed. Similarly, semiconductor regions and depletion regions and the like which are functionally similar or which can be put to complementary uses are uniformly designated by numbers assigned to each to assist in the association and assimilation of the description of the preferred embodiments and the drawings. Again, this is not meant to limit the invention, but instead to enable others to apply and use the invention in configurations not specifically disclosed in this description but within the scope and intendment of the claims to the invention.

By semiconductor material of a different or opposite conductivity type is meant N type or P type for intrinsic material , P type or intrinsic for N type material and N type or intrinsic for P type material. Therefore, while the discussion is mainly directed to devices having N and P type material this is because of the extent of its present use and is not meant to exclude devices which may be fabricated using intrinsic materials.

Also, while the invention is described in conjunction with a junction field-effect device controlled by merged depletion regions, it should be readily apparent that there are many varied applications of merged depletion regions and charge balance depletion region width control alone or together. The invention is much broader than just a discrete field-effect transistor. Emphasis placed on a junction field-effect transistor is due to the widespread use of the junction field-effect transistor and the fact that it readily demonstrates the special characteristics of the invention and is therefore illustrates a preferred embodiment of the invention.

In conventional devices, particularly the junction field-effect transistor (JFET), depletion regions have been merged in the channel of a JFET to pinch off the channel . However, until now, the merging of depletion regions other than in the channel of a JFET has been something to avoid due to the high currents that result when the regions merge, known as "punch through". In most cases punch through occurred inadvertently in devices where one junction was heavily reverse biased while the other was either unbiased or forward biased. The only consideration given to merged depletion regions made use of the high currents obtained when one junction depletion region punched through to the other depletion region with a resulting high current. Punch through has been used for MOS input protection and as a non-critical voltage regulator. One embodiment of this invention, however, uses merged depletion regions as a unique method of controlling the operation of semiconductor devices which results in new and unique structures, devices, and applications.

For any junction formed from materials with uniform doping, the charge on one side of a junction must equal the charge on the other side of the junction. That this must be the case can be seen by understanding the origin of the bound charge in the depletion regions. Any semiconductor material doped to produce excess holes or electrons is still electrically neutral. The mobile charge carriers are- balanced by bound charges of opposite polarity. When materials of different conductivity types are joined to form a junction the excess carriers result in a concentra¬ tion gradient which results in a diffusion current toward the junction. When two opposite type carriers meet, they annihilate each other resulting in the depletion of mobile charge carriers in the depletion regions on each side of the junction. In these depletion regions there is left behind the bound charge of opposite polarity to that of the mobile carriers in the material. Since one hole combines with one electron, the fixed charges that remain on each side of the junction must be equal in magnitude but opposite in polarity. In all junction devices charge balance must be maintained across a junction in equilibrium no matter what the method of depletion control used. The potential across the junction will be determined by the charges in the depletion regions. The use of charge balance across the junction has not been used in the past as a controlling method.

A preferred method to achieve charge balance depletion width control in accordance with the invention is to use a merged depletion region 38, 40 on the gate side 40 of the junction 24 to control the depletion region 42 on the channel side as shown in FIG. 1. Channel region 34 is bounded on two sides by substrate 32. Reverse bias of junction 24 causes depletion regioos 40 and 42 to extend into the substrate and into one side of the channel, respectively. Similarly, reverse bias of junction 25 causes depletion regions 41 and 43 to extend into the substrate and into an opposite side of the channel , respectively. The combined depletion regions 42, 43 pinch off and prevent current flow along the channel . Reverse bias of junction 22 causes depletion regions 36 and 38 to extend into the gate and substrate, respectively. As illustrated, the reverse bias on junctions 22 and 24 is sufficient to cause depletion regions 38 and 40 to merge in the substrate. Further increases in the reverse bias on junction 22, in accordance with the invention, causes depletion regions 42, 43 to retreat, thus allowing current to flow through channel 34.

A junction field-effect transistor controlled by merged depletion regions, to be called a MFET (merged- depletion junction field-effect transistor), has the potential of being the most flexible and useful of all merged-depl eti on charge-balance devices. It should be readily apparent, however, that many other structures and combinations of structures that provide and utilize a

O method of controlling a depletion region through charge bal ance .

For the purposes of illustration and description the operation and construction of a discrete device is disclosed. There are, of course, many varied applications of the invention including the construction of vertical devices and, because the gate 30 and channel 34 may be constructed of the same type material, it is possible to extend the drain diffusion of one device to form the gate 30 of another. It is also possible for a channel 34 to act as the gate 30 for a second device. These two examples illustrate how metal interconnections can be minimized in accordance with the present invention. In fact, because of input and output compatibility many types of mul tie! ectrode compound structures are possible including for the first time three-dimensional structures. Because of these features it is anticipated that the invention will find its greatest applicability in very large scale integrated (VLSI) circuitry. FIG. 2, consisting of FIGS. 2A and 2B, shows a test device 90 constructed in accordance with the invention. Referring to FIG. 2B , the test device consists of an N-type region 30 utilized as the gate of the device. A P-type epi-layer substrate 32 contacts the gate 30 and contains an N region diffused into the epi-layer 32 which forms the source 92, drain 94, and channel 34. A P-type diffusion 96 is applied over the N diffusion to narrow the N diffusion to produce the channel 34. As shown in FIG. 2A the P diffusion 96 connects directly to the P-type epi-layer 32. P-N junctions 22, 24 are formed between gate and substrate and between substrate and channel, respectively.

FIGS. 3, 4 and 5 di agram ati cal ly show a cross section of the device 90 in operation with the associated depletion regions indicated by the dotted lines. Operation of the device is discussed in more detail below. Basically, FIG. 3 shows the device 90 with the gate voltage equal to the

OMPI /., WIPO . , source voltage and the source voltage high enough to pinch off the channel 34. The drain 94 is at a higher voltage so that the device 90 is operating like a conventional pinched-off junction field-effect transistor (JFET). In FIG. 4 the gate voltage has been increased to equal the drain voltage. At this higher potential the depletion region 38 of the gate 30 in the P epi-layer substrate 32 merges with the depletion region 40 of the channel 34. This shifts the point of zero E field 52 in the epi-layer 32 toward the channel 34. Because the overall charge balance must be maintained the depletion region 42 in the channel 34 decreases and opens up the channel 34. However, as shown in FIG. 4, the channel 34 is still pinched off at the drain end 74 and the device 90 operates in the saturation region of its transfer curve.

FIG. 5 shows what occurs when the drain voltage is lowered to the source voltage. At this point the entire channel 34 is open and the device operates in the linear region of its transfer curve. A detailed description of the principles of operation of the device is as follows:

The application of reverse biases to the source 92 and drain 94 of an N channel device 90 with respect to the P substrate 32 causes a depletion region to spread into the channel 34. When the depletion regions meet in the center of the channel 34, the potential at the center of the channel 34 is fixed and cannot increase. However, the depletion regions also spread into the P region 32. In conventional devices the P regions are more heavily doped than the channel so that most of the bias voltage appears across the depletion region of the channel.

In a field-effect device 90 controlled by merged depletion regions, the P region 32 is not heavily doped, which allows the channel junction depletion region 40 to spread sufficiently into the P region 32. The control of the channel depletion region 40 on the P side of the

O channel junction is used to control the depletion region 42 on the N side of the channel junction 24 by using charge balance.

FIGS. 3, 4, and 5 show a semiconductor device 90 which employs the methods set forth above to control the P-side channel depletion region 40 of the channel junction 24 in the P substrate 32. The embodiment shown in FIGS. 2-5 is one embodiment of the invention and is illustrated as shown for ease of discussion. However, in other applications, the device 90 can utilize a gate 30 or any combination of gates 30 separated from any side of the channel 34.

FIG. 3A shows the source 92, drain 94, channel 34, gate 30, and substrate 32, 96 of a field-effect device 90, with the associated depletion regions 36, 38, 40, 41, 42 and 43 indicated by the dotted lines. The drain 94 and source 92 are of N-type material , and are joined by a channel 34 of the same N-type material and impurity concentration as the source 92 and drain 94. The concen¬ tration of the N region 34 is equal to that of the P region 32; thus the depletion region 40, 42 will spread out an equal distance on each side of the channel junction 24. The gate 30 is a heavily doped N region, placed close enough for the depletion regions 38, 40 in the substrate 32 to merge. FIG. 3B shows the magnitude and direction of the E field 50, 52 at the line b-b drawn from the gate 30 to the opposite side of the channel 34 in FIG. 3A. FIG. 3C shows the potential corresponding to the E field of FIG. 3B.

In FIG. 3 the gate voltage Vg is equal to the source voltage Vs and the source voltage is high enough to pinch off the channel 34. The drain 94 is at a higher voltage so that, in this configuration, the device 90 is operating like a pinched-off JFET. With the low gate voltage, the gate depletion region 38 does not meet the channel depletion reg^'ion 40 and is shown in FIG. 3A by the dotted lines, in FIG. 3B by the zero E field between the two P depletion regions, and in FIG. 3C by the section of the P region 33 between the gate and channel at zero potential . Because the channel 34 is pinched off, the potential at the center of the channel 34 is the pinch-off voltage which is labeled Vs in FIG. 3C. For a device constructed in the manner shown in FIGS. 2-5, it is important for the source 92 to be at pinch-off because if it is increased much above pinch-off, the performance of the device 90 will suffer. For this reason the source is taken to be at the pinch-off potential .

FIG. 4 has the same configuration as FIG. 3 except that the gate voltage has been increased to equal the drain voltage At this higher potential, the depletion region 40 of the channel 34 in the P region 32 now merges with the depletion region 38 of the gate 30. The dotted line in FIG. 4A in the P region 32, at the reference line, is the point where the E field is zero and assigns a portion of the depletion region 38, 40 to the gate 30 and channel 34. The E field of FIG. 4B shows how the increase in potential of the gate 30, with respect to the P epi-layer 32, has shifted the point of zero E field in the P region 32 toward the channel 34. Due to overall charge balance, the depletion region 42 in the channel 34 must decrease and open up the channel 34. The gate 30 must be adjacent to the entire channel 34, otherwise only a localized pocket of the channel 34 will open up and no current will flow. When the gate 30 opens the entire channel 34, the device 90 turns on and conducts. FIG. 4A shows the channel 34 pinched off at the drain end 94; thus the device operates in the saturation mode in a manner similar to a conventional junction field-effect transistor.

As stated above, if the source 92 is raised above the pinch-off voltage of the channel 34 there will be a decrease in performance if a gate 30 is used on only one side of the channel 34 as is shown in FIG. 3. If the gate 30 is not opposite both sides of the channel 34, and the

O ^■ 13 -

source 92 is above pinch-off, the depletion region 43 on the opposite side of the channel 34 will spread into the channel 34 as the depletion region 42 from the gate side retreats. In effect, the device will operate as a limited source follower.

That this is the case can be seen by examining the channel 34 when it is conducting. The channel potential must be higher than the source if the E field is not to reverse along the channel 34. The effect of the gate depletion region 38 merging with the channel depletion region 40 is to increase the potential on the P side 32. As is shown in FIG. 4, the channel depletion region in the channel 42 retreats and the channel 34 opens. This occurs because the source is just at pinch-off. If it is higher, the channel voltage must be increased until it is above the source. This is done by increasing the potential on the P side. The increased channel potential causes the reverse bias of the junction 25 opposite the gate 30 to increase, since there is no merged region on that side. The channel voltage continues to increase, without the channel 34 opening up until the channel potential is equal to the source 92. As the gate voltage is increased, the channel depletion region 40 on the gate 30 side retracts until the point when it is equal to its unbiased value, at which time any further increase causes punch through from the gate 30 to the source 92 and a characteristically high punch through current results. When the source voltage is above the channel pinch-off voltage, the depletion region 43 from the opposite side 25 moves into the channel 34 and the on-resi stance of the device 90 is higher because of the partial closing of the channel 34. This is not a problem when the gate 30 is applied to both sides, since both depletion regions 42, 43 are forced to withdraw from the channel 34. FIG. 5 shows what occurs when the drain voltage is lowered to the source voltage. At this point the entire channel 34 opens, and the device 90 operates in the nonsaturation region.

This method of control can be utilized to implement certain logic functions. For example, if a vertical device has two gates 30 on opposite sides of the channel 34 and the source 92 is just at pinch-off, if either gate potential is increased, current will flow in the channel 34. This implements an OR function. If the channel 34 is a square with four equal channel sides, the channel 34 will pinch-off from all sides. If the gates 30 are on opposite sides of the channel 34 they will open the channel 34 from that side but the channel 34 will still pinch-off from the sides at right angles to the gate side of the channel 34. It is thus necessary for two gates 30 to be at right angles to each other and both gates 30 to have a high potential if the current is to flow in the channel 34. This produces an AND function.

FIG. 6 shows a vertical JFET formed to allow the construction of the device 90 in integrated form. The device 90 shown is only approximate and is only for use as an illustration of the design. Regions A, B, C, D, E, F, G, and H are all N type regions while I, J, K, and L are P type regions. Regions F and G are the gate 30 while regions A and D are interchangeable as the source 92 and drain 94.

The device can be constructed, for example, by first implanting impurities in the semiconductor material 103 to form regions A, E, and I and then growing an epi-layer 105 before any extensive drive-in diffusion of these lower regions occurs. The regions D, H, and L on the surface of the epi-layer 105 are then implanted followed by a drive-in diffusion step. The impurities on the surface diffuse into the epi-layer 105 while those impurities in the bottom semiconductor layer 103 out diffuse into the epi-layer 105. Regions B and F are the out diffusions from the N implanted regions A and E respectively. Regions C and G are the in

"STG diffusions of the implanted N surface 101 regions D and H respectively. For the P regions, J is the out diffusion of region I while K is the in diffusion of region L on the surface 101. Where regions B and C meet, the concentration is selected to give the required channel concentration to produce pinch-off at the required voltage and this concen¬ tration can be controlled by the implanted impurities and by the drive-in step.

Because of the diffusion profiles, both the N and P regions are lightly doped near the center of the epi-layer 105 with the doping becoming heavier toward the upper and lower, 101, 102 surfaces. These heavily doped regions provide a low resistance path from the lighly doped regions to the surfaces 101, 102. The heavily doped regions also prevent the gate and channel depletion regions 38, 40 from merging in areas where the gate 30 does not control the channel 34. This capacitive coupling is an unwanted parasitic capacitance.

In this structure the length of the channel 34 is only that portion of the channel 34 where the impurity concen¬ tration is low enough to allow the channel 34 to pinch-off at the required voltage.

For the purpose of analysis of operation it can be assumed that the P regions 32 are grounded. The plots of the E fields in FIG. 6, FIGS. 6A-6T, represent the E fields of the device 90 sectioned through the minimum doping level near the middle of the epi-layer 105 as shown by the dotted line. All the potentials are normalized to the voltage necessary to pinch-off the channel ; that is, all voltages are measured as multiples of the pinch-off voltage, Vp . All dimensions when given are normalized to the channel height, H.

In most applications the device will be operated in the enhancement mode; thus a potential must be applied to the source 92 equal to Vp to pinch-off the channel 34. Since this is a bias voltage across a reverse biased P-N ^■ In ¬

junction 24, the current that must be supplied will only be the leakage current of the junctions. This potential can be supplied by a bias supply off the chip or by a standard on-chip charge pump technique as is now commonly used for substrate bias in NMOS circuits.

To maximize performance, the channel 34 should have as low a resistance as possible and the influence of the gate 30 on the channel depletion region 42, should be maximized. To lower the resistance of the channel 34, it should be as heavily doped as possile. Since the doping is fixed by the pinch-off voltage requirements and the channel height, H, the channel resistance is reduced by shortening the length of the channel pinch-off region. The limits on the length of the channel 34 are set by the avalanche breakdown characteristics of the channel 34 at the applied drain to source voltage.

The second method of enhancing the performance of the device, that of maximizing the influence of gate 30 on the channel depletion region 42, is illustrated in the series of plots of the E fields in FIG. 6. The plots, FIGS. 6E-6T, are arranged horizontally according to the configuration of the device 90 and the threshold voltage and arranged vertically according to the normalized applied gate voltage. In these figures the degree of channel opening is a relative measure of the gain of the device. The gain of all field-effect devices is determined by the number of charges under the gate area and is thus a function of the gate capacitance and the impurity concen¬ tration under the gate. In FIG. 6 the gate capacitance increases from left to right as does the gain.

The first column in FIG. 6, FIGS. 6E-6T, consists of the E field plots of a linear device, the gate concentra¬ tion equal to the channel concentration, with a threshold voltage of .5Vp. The threshold voltage is that voltage applied to the gate 30 just necessary to start the channel 34 conducting when the source 92 is just biased to pinch-off. In all the figures, the area under the E field 52 will be equal to the pinch-off voltage while the area under the E field 50 will be equal to the gate voltage. The first row in FIG. 6, FIGS. 6E, I, M and Q, shows the threshold voltage and is equal to area 50 minus area 52. The second row, FIGS. 6F , J, N and R, indicates the E fields 50, 52 when the input voltage is .2Vp and for this case since it is below the threshold voltage, the gate and channel depletion regions 38, 40 have not merged. The third row, FIGS. 6G, K, 0 and S, shows the input voltage equal to Vp and for this configuration the channel opening is .2H/2. The last row, FIGS. 6H, L, P and T, shows the maximum voltage that can be applied to the gate 30 without the gate 30 punching through to the channel 34 and in this case it is equal to about 3.9Vp. For this linear device the channel doping and the substrate doping are equal .

The second column FIGS. 6I-L, shows a linear configu¬ ration with the threshold voltage equal to zero. As can be seen from the figures, the channel 34 opens further than for the configuration shown in the first column. To lower the threshold voltage the gate 30 to channel 34 spacing D has been decreased and is now equal to the channel height, H; thus the gate capacitance increases and the increase is reflected in the channel opening. It also results in a reduction in the maximum voltage that can be applied to the gate to about 3Vp.

The third column FIGS. 6M-P shows a quadratic configuration with the threshold voltage equal to zero and with the channel concentration equal to the substrate concentration. That this is a quadratic device, the gate concentration not equal to the channel concentration, can be seen by the vertical drop of the E field 50 in the gate 30 region. The gain also increases because there is greater charge adjacent to the gate depletion region which increases the gate capacitance. ^_{m a} is now reduced to 1.9Vp.

^■- iϊ^E

OMPI /,, WIFO .v The last column FIGS. 6G-T shows a quadratic configuration with the threshold voltage equal to zero but with the substrate doping twice that of the channel . The gate 30 to channel 34 separation D decreases; thus the capacitance increases. ^_{m a}x is now 1.5Vp.

The actual performance of these structures is deter¬ mined primarily by the size of the channel height, H, or the gate 30 to channel 34 spacing, D, whichever is less. The minimum dimension is determined by the limits of the photolithographic techniques and as this dimension gets smaller the performance increases due to the higher channel doping for a given pinch-off voltage and to a reduction in the parasitic capacitances of the gate and drain junctions. A comparison of the MFET to an MOS transistor gives some indication of the relative merits of the two devices. When comparing the gains of the two devices, the MOS device will usually have a higher gate capacitance while the MFET will have a lower channel resistance due to the high mobility and shorter channel lengths. Three different comparisons have been made based on the minimum photo¬ lithographic feature.

For devices where 5um is the minimum dimension which corresponds to conventional LSI MOS transistors, the MOS channel length is 5ym with the oxide thickness equal to .lμ . For the MFET the effective channel length is .2ym while the gate to channel spacing is 2.5μm for a device with the substrate doping twice that of the channel. The MOS gate capacitance is 8 times that of the linear MFET and 4 times greater than the quadratic. The MFET's channel resistance is 50 times less than that of the MOS device; thus the MFET has a gain advantage over an MOS device of 6 times for the linear MFET and 12 times for the quadratic device. With a minimum dimension of 2.5μm the MOS channel length is 2.5ym with an oxide thickness reduced to ,04ym which is close to the minimum due to the dielectric strength of the gate oxide. The MFET has an effective channel length of .2μm and a gate to channel spacing of 1.25μm. The MOS gate capacitance is 10 times that of the linear MFET and 5 times greater than the quadratic MFET. The MFET channel resistance is 25 times less than that of the MOS device; thus, the linear MFET's gain is 2.5 times that of the MOS device while the quadratic's gain is 5 times greater than that of the MOS transistor. At this level the MOS performance has improved due to the very thin oxide.

At lμm minimum dimension the MOS channel length is lμm and the gate oxide thickness is still .04μm. For the MFET the channel length is still .2μm while the gate to channel spacing is .5μm. . The MOS gate capacitance is 4 times that of the linear MFET and only twice the capacitance of the quadratic MFET. The MFET channel resistance is 10 times less than that of an MOS device. The relative gain of the linear device is 2.5 times greater than an MOS device while the quadratic device has a gain 5 times greater than the MOS device.

These figures show the superiority of the MFET over an MOS device when the gains of the device are compared. What is not shown is the fact that the MOS device has a higher propagation delay than the gain terms would indicate due to the higher gate capacitance of the MOS device which must be driven by the preceding stages. In addition, the MFET has much less drain to gate feedback capacitance which again further improves performance. As can be seen from the above considerations, the MFET will always have greater gain than an MOS device and at the same time a large reduction in operating device capacitance.

A test device 90 is shown in FIG. 2 and was constructed as follows. A P type epitaxial layer 32 was formed on an N type wafer 30. The N type wafer was doped

OMPI ;_h WIFO Λ with antimony to a resistivity of 0.02 ohm-centimeter. The P-type epi-layer 32 was boron doped with an impurity concentration of 5 x lO*-*^ _and was 20 micrometers thick. Construction of the device was completed by diffusing a light N region 34, 92, 94 into the P-type epi-layer 32. A P⁺ diffusion 96 was then applied over the N diffusion which narrowed the N diffusion and produced the channel 34. This P diffusion 96 connected directly to the epi-layer and thus became part of the substrate 32.

The N regions 34, 92, 94 were diffused to a depth of 5 micrometers and had a surface concentration of 2.5 x 10*6 A/cm^. P region 96 was diffused to a depth of 2.5 micrometers and a surface concentration of 1 x 10*⁷ A/cm3. For this device 90 the pinch-off voltage was 8.2 volts, and the threshold voltage Vt was calculated to be 15 vol ts .

In operation, the channel 34 of the device 90 is pinched off by applying a voltage to the source 92 that is greater than Vp, the voltage that causes the depletion region 42 in the N region 34 to close off the source end 92 of the channel 34. If the drain 94 is then made more positive than the source 92 and with the gate 30 unbiased, the entire channel 34 is pinched off and no current flows. If the drain voltage is increased the current still will not change. If a positive voltage is applied to the gate 30, the gate depletion region in the substrate 38 merges with the depletion region of the channel in the substrate 40. If the gate potential is further increased, the depletion region in the channel narrows and the channel 34 opens and current flows. When operated in this manner, the device operates like an enhancement- mode JFET, one that is normally off until a gate signal turns it on.

A junction field-effect transistor is inherently stable with temperature. The junction field-effect device controlled by merged depletion regions has the same channel

O resistivity characteristics as that of a conventional JFET with similar doping but the variation of threshold voltage with temperature is much less for the merged-depl eti on devices since the threshold voltage variations are due to the change in contact potential with temperature.

Thermal noise generated in the channel 34 is less than that of a conventional JFET since the shorter channel length increases the transconductance which lowers the channel noise resistance. Because the device can be constructed with a short channel , and since noise components are a linear function of the channel length, a low voltage device has almost an order of magnitude improvement over the current JFET devices. Since the JFET has lower noise than MOS or bipolar devices, the MFET has better performance than these other devices.

Those skilled in the art will immediately recognize that semiconductor devices of the present invention can assume many varied embodiments, other than those specifically disclosed herein, without departing from the spirit of the invention. The invention will have many uses both as a discrete device and particularly in integrated circuitry. It should be understood that any such modifications and variations that may be resorted to are considered to be within the purview and scope of the present invention, and it is intended by the appended claims to cover all such variations and modifications as fall within the true spirit and scope of the invention. I claim as my invention:

^■ _ΥEA

OMPI /,, WIPO

Claims

C l a i ms

1. A semiconductor device comprising: first and second regions of a first conductivity type; a third region of opposite conductivity type interposed between and forming junctions with said first and second regions, said junctions having depletion regions associated therewith and said third region being of a width and impurity doping concentration to allow merging of said depletion regions within said third region; and reverse bias means between said second and third regions controlling depletion width in said first region.

2. In a semiconductor device including a channel region of first conductivity type, a second region of second conductivity type forming a first junction with said channel region, and a third region of said first conductivity type forming a second junction with said second region, said second region being sufficiently thin that depletion regions associated with said first and second junction can merge in said second region, the method of operating said device which comprises: applying a reverse bias to said second junction, said reverse bias being sufficient to cause said depletion regions to merge in said second region; and varying said reverse bias to adjust the depletion spread into said channel region and thereby to adjust the conductivity of said channel region.

3. In a semiconductor device wherein current can flow through a channel region of first conductivity type between source and drain, the improvement which comprises: a gate region of said first conductivity type separated from said channel by an intermediate region of second conductivity type; first and second junctions being formed between said intermediate region and said channel region and between said intermediate region and said gate region, first depletion regions having a width extending into said channel and intermediate regions from said first junction, and second depletion regions having a width extending into said gate and intermediate regions from said second junction, said intermediate region being of width and doping to allow merging within said intermediate region of said first and second depletion regions; and bias means for reverse biasing said second junction and for controlling the width of said first depletion region in said channel regi on .

4. In a semiconductor device, the combination compri si ng: first and second PN junctions having approximately parallel portions and having depletion regions associated therewith ; said depletion regions having boundaries displaced from the associated junction as a function of the doping concentrations on either side of said junctions; said junctions being spaced such that the depletion regions therebetween merge; and reverse bias means connected to a first PN junction for varying the lateral conductivity on the side of the second PN junction remote from said first PN junction.

5. The semiconductor device as set forth in claim 4 wherein said side of the second PN junction remote from said first PN junction comprises a fi st channel .

6. The semiconductor device as set forth in claim 5 and further comprising: a third PN junction having depletion regions associated therewith; said third PN junction being spaced from said first PN junction such that the depletion regions therebetween merge; and wherein the side of the third PN junction remote from said first PN junction comprises a second channel.

7. The semiconductor device as set forth in claim 6 wherein said first channel is not parallel to the second channel .

8. The semiconductor device as set forth in claim 7 wherein said first channel is not coplanar with said second channel, thereby providing a three dimensional device.

9. The semiconductor device as set forth in claim 6 wherein the reverse bias means and the currents in said first and second channels provide a logic function.

10. The semiconductor device as set forth in claim 4 wherein the depletion regions between said junctions merge with little or no externally applied voltage across either junction.

11. The semiconductor device as set forth in claim 6 wherein the depletion regions between said first and third PN junctions merge with little or no externally applied voltage across either junction.

12. A semiconductor device comprising a region of semiconductor material of a first conductivity type; a region of semiconductor material of a second conductivity type opposite to that of the first forming a junction with the region of semiconductor material of the first conductivity type; and means for controlling the width of the depletion region on one side of the junction only so that the width of the depletion region on the other side of the junction is altered in proportion to the amount of control applied to the depletion region on the controlled side of the uncti

- O