Power FET with a resonant transistor gate
10651167 ยท 2020-05-12
Inventors
Cpc classification
H01L23/5228
ELECTRICITY
H01L27/0629
ELECTRICITY
H01L2924/0002
ELECTRICITY
G06F1/3203
PHYSICS
H01L23/5227
ELECTRICITY
H01L29/0634
ELECTRICITY
H01L29/43
ELECTRICITY
H01L23/5222
ELECTRICITY
H01L29/7803
ELECTRICITY
H01L2924/0002
ELECTRICITY
H01L2223/6627
ELECTRICITY
H01L29/435
ELECTRICITY
H01L2924/00
ELECTRICITY
H01L25/16
ELECTRICITY
H01L2924/00
ELECTRICITY
International classification
H01L21/00
ELECTRICITY
H01L25/16
ELECTRICITY
H01L29/43
ELECTRICITY
H01L29/423
ELECTRICITY
H01L29/739
ELECTRICITY
H01L29/06
ELECTRICITY
H01L27/06
ELECTRICITY
G06F1/3203
PHYSICS
Abstract
A semiconductor FET provides a resonant gate and source and drain electrodes, wherein the resonant gate is electromagnetically resonant at one or more predetermined frequencies.
Claims
1. A semiconductor FET, comprising a resonant gate electrode that modulates transconductance between source and drain electrodes, wherein: a plurality of passive circuit elements including one or more inductor elements are embedded within the structure of the gate electrode that causes the gate electrode to respond electrically as a resonant transmission line; and, said resonant transmission line further consists of one or more resonant transmission line segments that are in turn connected in series to operate as a distributed network filter, such that, the semiconductor FET modulates transconductance at a resonant frequency or resonant frequency band characteristic to each resonant transmission line segment forming the distributed network filter.
2. The semiconductor FET of claim 1, wherein the passive circuit elements embedded within the structure of the gate electrode to form the distributed network filter are electrically connected in series or in parallel.
3. The semiconductor FET of claim 1, wherein the resonant gate electrode comprises a serpentine path and the passive circuit elements are formed by localized reactive loading within regions along the serpentine path is generated by the electromagnetic coupling between parallel or anti-parallel alignment of instantaneous current vectors flowing between adjacent portions of serpentine path.
4. The passive circuit elements of claim 3, wherein electroceramic elements are incorporated within the regions of localized coupling.
5. The passive elements of claim 4, wherein the electroceramic elements comprise high dielectric density electroceramic with relative permeability .sub.r10 or relative permittivity .sub.r10.
6. The semiconductor FET of claim 2, wherein the embedded passive circuit elements are discrete dielectric passive circuits elements comprising an electroceramic element forming a resistive element, a capacitive element, or an inductive element.
7. The capacitive electroceramic elements of claim 6, wherein capacitive electroceramic elements comprise a high energy density electroceramic with relative permittivity .sub.r10.
8. The inductive electroceramic elements of claim 6, wherein the inductive electroceramic elements comprise a high energy density electroceramic with relative permeability .sub.r10.
9. The inductive electroceramic elements of claim 6, wherein the inductive element is a toroidal inductor, a spiral inductor, or a curved current path.
10. The semiconductor FET of claim 1, wherein a resonant transmission line segment forming the distributed network filter comprises a resistive element that tunes the resonant transmission line segment to be critically dampened at a characteristic frequency or frequency band.
11. The semiconductor FET of claim 1, wherein a resonant transmission line segment forming the distributed network filter comprises a terminal resistive element that broadens the frequency bandwidth of the resonant transmission line segment.
12. The semiconductor FET of claim 1 configured as a vertical FET or surface FET.
13. The semiconductor FET of claim 1 formed on silicon, silicon germanium, III-V compound or II-VI compound semiconductor.
14. The semiconductor FET of claim 1 embedded within a power management module.
15. The semiconductor FET of claim 14, wherein the power management module is a monolithic power management module.
16. The semiconductor FET of claim 1 that comprises an elongated gate width (W.sub.gate) wherein the ratio of gate width to gate length (L.sub.gate) is W.sub.gate/L.sub.gate10.sup.2, preferably W.sub.gate/L.sub.gate10.sup.6.
17. The power management module of claim 14, wherein the semiconductor FET comprises a resonant gate electrode that has a frequency bandwidth that has a resonant frequency or resonant frequency band that overlaps the frequencies of interest to the control topology.
18. The power management module of claim 14 incorporated within an electric motor, radio base station, power distribution network or handheld wireless appliance.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
(1) The present invention is illustratively shown and described in reference to the accompanying drawings, in which:
(2)
(3)
(4)
(5)
(6)
(7)
DESCRIPTION OF THE PREFERRED EMBODIMENT
(8) This application is copending with de Rochemont U.S. Ser. No. 13/168,922, entitled SEMICONDUCTOR CARRIER WITH VERTICAL FET POWER MODULE, filed Jun. 24, 2011 (de Rochemont '922), and de Rochemont U.S. Ser. No. 13/163,654, entitled FREQUENCY-SELECTIVE DIPOLE ANTENNA, filed Jun. 17, 2011 (de Rochemont '654), which are incorporated herein by reference. The current application instructs the incorporation of passive components and/or high dielectric density electroceramic elements within transistor gate structures to cause a power FET to become resonant at a desired frequency or range of desired frequencies. One counterpart application (de Rochemont '922) instructs means to fully integrate a high efficiency, power management system as a monolithic structure on a semiconductor carrier to modulate high current levels using a resonant three-dimensional gate structure enabled by serpentine windings. The other counterpart application, (de Rochemont '654), instructs methods to form a conducting element as a serpentine winding by folding the conducting element in ways that introduce localized regions of capacitive or inductive loading, such that the combination of localized reactive loads along the length of the folded conductor form a distributed network filter. It goes on to illustrate how two mirror image serpentine elements so formed function as a dipole antenna that is resonant over selective frequencies. The counterpart application de Rochemont '654 also instructs the insertion of tight-tolerance electroceramic material within the regions of localized reactive loading to increase or more precisely tune the coupling strength of localized reactive loads. The current application is also copending with de Rochemont U.S. provisional application No. 61/409,846, QUANTUM DOT FIELD EFFECT TRANSISTOR IN A FULLY INTEGRATED SILICON CARRIER AND METHOD OF MANUFACTURE THEREOF, filed Nov. 3, 2010 (de Rochemont '846), which is incorporated herein by reference.
(9) The current application incorporates by reference all matter contained in de Rochemont, U.S. patent application Ser. No. 11/479,159, filed Jun. 30, 2006, entitled ELECTRICAL COMPONENT AND METHOD OF MANUFACTURE (the '159 application), de Rochemont, U.S. patent application Ser. No. 11/620,042 filed Jan. 6, 2007 entitled POWER MANAGEMENT MODULES (the '042 application, de Rochemont and Kovacs, U.S. patent application Ser. No. 12/843,112 filed Jul. 26, 2010, LIQUID CHEMICAL DEPOSITION PROCESS APPARATUS AND EMBODIMENTS, (the '112 application), and de Rochemont U.S. patent application Ser. No. 13/152,222, entitled MONOLITHIC DC/DC POWER MANAGEMENT MODULE WITH SURFACE FET, filed Jun. 2, 2011 (the '222 application). The '159 application discloses how LCD methods fabricate a monolithic integrated circuit comprising tight tolerance passive networks. The '042 application discloses how liquid chemical deposition (LCD) methods fabricate a monolithic integrated power management module that includes a tunable inductor coil. The '112 application discloses preferred apparatus used in applying LCD methods. The '222 application instructs the monolithic integration of a low-loss power management circuit containing a surface FET.
(10) The present invention applies LCD manufacturing methods to integrate tight-tolerance passive circuit elements within the gate structure of a field effect transistor to produce a resonant gate signal response at a desired frequency or band of frequencies. The modulation of large currents typically requires large gate structures that increase gate capacitance, which, in turn, decreases switching speed. Large gate structures can be switched at higher speeds by applying more energy to the gate to displace the charge that collects underneath it. However, the displacement of large quantities of electrical charge under a large gate will require more energy to be applied to the gate electrode than is passed from the source to the drain to achieve the switching speeds desired in many modern microelectronic applications. This paradox results in negative efficiency devices. It is a specific objective of the present invention to modulate FET power output at arbitrarily high frequencies (switching speeds) in FETs having arbitrarily large gate structures. It is an additional objective to produce high-speed/high-current power management devices by embedding resistive, capacitive and inductive elements to form an RLC circuit within the gate enable the transistor gate to be operated under conditions of gate resonance. At gate resonance the switching response is effortless because the embedded inductance neutralizes the reactance of the gate and any embedded capacitors. As discussed further below, the passive circuit elements embedded within the gate structure are used to tune the gate's resonant frequency response to match any desired switching speed or band of switching frequencies regardless of the gate's surface area and internal capacitance. This resonant frequency response is achieved by embedding tight-tolerance, high-density dielectric materials that precisely tune the performance values of the embedded passive circuit elements located at select locations within the gate structure. The resonant transistor gate thereby allows power FET to manage arbitrarily large currents, limited only by the resistance and patterning of the conductors within in the device. The shape and elemental chemistry of the device conductor elements should selected to minimize resistive loss and electromigration to ensure reliable operation.
(11) LCD manufacturing methods allow compositionally complex and mismatched materials having atomic-scale chemical uniformity and stoichiometric precision to be integrated in selected areas on the surface of a semiconductor substrate. The process temperatures (400 C.) used by LCD do not alter dopant profiles of active components buried within the semiconductor substrate. These low deposition temperatures also allow the microstructure of LCD deposits to be restricted to nanoscale dimensions. Nanoscale microstructural controls are a necessary condition for producing electroceramic compositions that have functional properties that remain stable with varying temperature. These combined attributes allow complex electroceramics to be formed into higher performance passive components that hold the critical performance tolerances needed to make passive circuit integration economically viable. LCD manufacturing technology enables monolithic integration of passive components by ensuring their operational performance values are held to within 1% of their desired performance over standard operating temperatures without disturbing active circuitry embedded within a semiconductor substrate.
(12) The invention relates primarily to the formation of a resonant transistor gate structure fabricated by placing electroceramics as a passive component or a dielectric element on the surface of a semiconductor. The invention is not dependent on a particular doping structure within the semiconductor substrate and can be applied to any and all currently known FET junction doping patterns or any contemplated power FET doping structures without limitation to marginal gains that might be achieved by one doping profile relative to another when applied to a specific application.
(13) Reference is now made to
(14) The resonant gate surface FET 11 need not be formed using a series-connected resonant transmission line 40 as shown in
(15)
(16) In general, a vertical FET consists of a drain electrode 61, a drain layer 73 comprising a highly doped semiconductor material, (usually doped with charge carrier densities in the range of 10.sup.18 to >10.sup.19 cm.sup.3), and a lightly doped (usually n-type) intrinsic semiconductor layer 75 having charge carrier density in the range of 10.sup.10 to <10.sup.17 cm.sup.3 depending upon the semiconductor material and design considerations. The intrinsic semiconductor layer 75 will be selectively doped with opposite type (usually p-type) dopants in barrier diffusion regions 77 to produce active junctions. The diffusion regions 77 may be trenched or take the form of pillars used to form a superjunction (not shown) as design alternatives. Similarly, the drain layer 73 may be n-type to form a conventional double-diffused junction (DDMOSFET) or may be p-type to form an insulated gate bipolar transistor (IGBT). Source diffusion regions 79 (usually n.sup.+-type with charge carrier density >10.sup.17 cm.sup.3) are formed to establish ohmic contact with the source electrode 63. Voltage applied to the gate electrode 81 forms a conductive channel 83 within the barrier diffusion regions 77 immediately below the gate insulator 84 of the gate structure segment 72 to modulate current flow 85 between the source electrode 63 and drain layer 73.
(17) In a resonant gate vertical FET 59, LCD methods are used to insert passive components comprising one or more resistor elements 65, capacitor elements 67,69, and inductor elements 71 at select locations between the vertical gate structure segments 72. Performance values of the passive components and their electrical interconnection, whether in series or in parallel, are selected using methods of RLC lumped circuit or distributed network filter analysis to cause the gate to operate as a resonant transmission line having frequency response optimized (resonant) at a desired frequency or band of frequencies. Vertical gate structure segments 72 that modulate current between the source electrode 63 and drain layer 73, such as that shown in B-B cross-section in
(18) Reference is now made to
(19) Certain design criteria may require embedded parallel capacitors 120, (shown in
(20) Representative embedded inductor elements 31,71, (shown in
(21) Representative embedded resistive elements 150 (shown in
(22) For reasons relating to the critical dampening of the transistor's resonant response discussed below, it is a preferred embodiment of the invention to configure a resistive component as an embedded terminating resistor 160 as shown in
(23) It is further recommended that barrier layers (not shown) be inserted between the passive components' dielectric material and the semiconductor substrate to prevent migration of any element within the dielectric material that might readily diffuse into the semiconductor while operating the device.
(24) On resistance is a critical operational parameter of all power FETs, since higher resistivity generates more heat, which, if not properly managed, will produce higher temperatures in the channel region that will degrade transistor performance. The On resistance (R.sub.ON) of a standard vertical FET is the sum of the channel resistance (R.sub.Ch) and the drain resistance (R.sub.Drain) and is mathematically characterized using:
(25)
(26) This model representation of the present invention depicts a FET gate electrode that has a 30 m gate length (L.sub.gate) and a 1 meter wide gate width (W.sub.gate) to make it easier to visualize pictorially. It could just as easily comprise an FET gate electrode that a 1 m long (or smaller) gate length (L.sub.gate) and a 30 meter wide (or wider) gate width (W.sub.gate). A specific embodiment of the invention is to establish a gate electrode structure wherein the gate width (W.sub.gate) is at least two orders of magnitude, preferably more than 6 orders of magnitude, greater than the gate length (L.sub.gate). i.e., 10.sup.2W.sub.gate/L.sub.gate, preferably 10.sup.6W.sub.gate/L.sub.gate. Making reference to equations 1a, 1b, it is quite evident that the gate geometry described within the present invention enables a substantial reduction in On Resistance (R.sub.ON) by reducing its channel resistance (R.sub.Ch) component, which is inversely proportional to the ratio W.sub.gate/L.sub.gate. For example, a gate geometry wherein W.sub.gate/L.sub.gate is 10.sup.6 will have 1 1/millionth the channel resistance of a gate electrode where W.sub.gate/L.sub.gate=1. There are corresponding reductions in the drain resistance R.sub.Drain, since the geometrical factor, k, for the source electrode 63 is the ratio of the thickness of the intrinsic semiconductor layer 75 to the source electrode surface area. When there is a donut hole (not shown) at the center of the source electrode 63, its total surface area is determined as the area spanned by the outer radius minus the area spanned by the inner radius, or:
A=(R.sub.out.sup.2R.sub.in.sup.2)(2)
where, R.sub.in=0 when there is no hole at the center. Ordinarily, the thickness of the intrinsic semiconductor layer 75 is 5 m (510.sup.4 cm) or less. When the source electrode has an outer radius on the order of 1 cm with an inner radius of 0.4 cm, the total surface area of the donut is 2.51 cm.sup.2 making the geometrical factor k=210.sup.4, assuming an intrinsic semiconductor 75 layer thickness of 5 m. In designs where the layer thickness can be reduced to 2 m and the source electrode is expanded to 2 cm radius with no donut hole, the geometrical factor becomes k=1.610.sup.5. Consequently, the greatly expanded size of the source electrode 63 that is enabled adding greater economic value to the semiconductor real estate by fully integrating the entire system onto the semiconductor dramatically reduces the On Resistance (R.sub.ON). The expanded source electrode, and the elongated gate widths underneath it, reduces loss in the system, thereby mitigating the need for additional system costs, such as coolant loops, to be incorporated into the larger system. The greater overall efficiency of this one fully integrated circuit lowers the cost and improves the intrinsic value of diverse systems, such as an electric motor, a radio base station, or power distribution network, in which the power management system may ultimately be applied. These concepts may also be applied to lower power systems, such as handheld wireless appliances, by shrinking the scale of the device to match the lower power requirement of those systems.
(27) A drawback to elongated gate widths is that they will also have higher gate capacitance (C.sub.gate). While higher gate capacitance reduces channel resistance R.sub.Ch, the higher gate capacitance C.sub.gate also reduces the gate switching speed. Lower gate switching speeds are undesirable since they require larger inductance and capacitance values to be integrated into circuit, necessitating the integration of large scale components. Therefore, an additional aspect of the present invention is to integrate tight tolerance LCD passive components in series or in parallel into that gate electrode to minimize the gate input capacitance C.sub.INgate, while maintaining elongated gate widths, C.sub.gate. As explained in greater depth below, gate input capacitance can be minimized by adding additional capacitor elements in series with the gate electrode, which itself functions as a capacitor. The input capacitance of a transmission line is reduced when capacitors are added in series according to:
1/C.sub.gate=.sub.lj(1/C.sub.FET.+1/C.sub.SERIES CAPj)(3)
(28) However, limited instantaneous bandwidth is a drawback of series capacitors, so it is desirable to add additional series capacitance elements to the gate electrode to reduce the gate's input capacitance and improve gate switching speeds.
(29) A further aspect of the invention is to arrange the passive components, preferably tight tolerance passive components, in such a way as to cause the gate electrode to function as a critically dampened transmission line that is resonant at the desired switching speed(s) .sub.switch. The low loss, tight component tolerances enabled by LCD manufacturing methods allow high-Q transmission line structures to be constructed. When power management control topology uses time-based switching modes, it is preferable to tune the resonant gate electrode to have a narrow pass band. However, when control topologies utilize frequency-based switching modes, it is preferable to tune the resonant gate electrode to have a bandwidth that overlaps the frequencies of interest to the control topology.
(30) As mentioned above, transmission lines and circuits consisting of a plurality of series capacitors will have reduced capacitance at the expense of narrower instantaneous bandwidth. Therefore, it is desirable to introduce inductor elements 35,71 within the resonant gate transistors 11,59 that cause the gate signal to have broader instantaneous bandwidth and to be resonant at the desired switching frequencies .sub.switch pre-determined by the inductance of the inductor elements 35,71 and the total capacitance of the gate structure segment 19,72 and embedded capacitors 33,67,69 as defined by equation 3.
(31) Reference is now made to
(32) Resonant RLC circuits may also be constructed by embedding the passive circuit components in parallel as depicted in
(33) Transmission lines and circuits that contain inductors and capacitors, in series or in parallel, will display resonant characteristics. While it is advantageous for the resultant transmission line to be resonant at the switching frequency, .sub.switch1, resonance can cause deleterious effects when not properly damped. The characteristic solution of an RLC circuit is expressed in terms of its natural resonant frequency, .sub.o, and damping factor, , is:
x=Ae.sup.st(4a) where,
s=(.sup.2.sub.o.sup.2)(4b) and the circuit's natural frequency .sub.o is determined by
.sub.o=2.sub.o=1/(LC)(4c) x applies to the gate voltage when a series RLC circuit is considered and applies to the source current in the case of a parallel RLC circuit. Resistive elements, which include the conductive electrode element in the active and passive components, act as dampening force on the circuit. In parallel RLC circuits the damping coefficient is given by:
=1/(2R(C/L)),(4d) and in series RLC circuits the damping coefficient is given by:
=R/(2(L/C)).(4e)
(34) It is often desirable to tune the resonant gate transistors 11,59 to be critically damped at the switching frequency to minimize settling times. Over-dampening occurs in RLC circuits when >.sub.o and typically produces long settling times and large steady state errors. Under-dampening occurs when <.sub.o and are prone to signal ring, and have oscillations that decay over longer than desirable time periods. Critical dampening occurs when =.sub.o. Often the resistance of the conductive elements in the active and passive components forming the gate electrode is insufficient to achieve critical dampening. In these instances, it is desirable to introduce resistive elements 31,66 within the resonant gate transistors 11,59.
(35) It is advisable to broaden the frequency bandwidth of the gate electrode when frequency-based control topologies are applied to manage power in these circuits. Bandwidth is proportional to the damping coefficient in LCR circuits and is given, in units of Hertz, by:
=/2=/=R/2L.(5)
(36) In many instances it is desirable design objective to broaden the bandwidth by adding extra resistance to the resonant gate transistor 11,59 using a terminal resistor 160 as shown in
(37) The resonant gate transistors need not be constructed from simple series-only and parallel-only lumped circuits. These simple circuits allow for convenient expressions because they can be reduced to closed analytical formula. Numerical methods now provide means to determine resonant frequency response in more complex structures that contain a plurality of embedded passive components configured in series and parallel connections within a plurality of resonant transmission line segments as shown in
(38) Reference is now made to
(39) A close up top view of
(40)
(41) The inductance generated by inductive coupling between two parallel wire segments in the absence of a ground plane can be given by:
(42)
C=l.sub.o.sub.r ln(d/r)(7) where l is the coupling length, d is gap between the wires and r is the radius of the wire, all in meters, .sub.o is the permittivity of free-space, and .sub.r is the relative permittivity of the material separating the parallel wires.
(43) Equations 6&7 linearly correlate the strength of the reactive loading (inductive or capacitive) with the coupling length l. (the physical separation of electromagnetically coupled gate segments) and the dielectric density (.sub.r,.sub.r) of the material located between the co-linear portions of adjacent gate segments. Therefore, it is a distinct embodiment of the invention to use LCD methods to insert tight tolerance dielectric material with co-linear portions of adjacent gate segments to tune the resonant characteristics of the dielectrically loaded serpentine resonant gate transistor 212. LCD methods are used to tune the gate's resonant response by inserting high permeability dielectric loads 236, having .sub.r10, within the spacing between adjacent gate segments that exhibit parallel instantaneous current vector alignments 232A,232B,232C,232D,232D (where desired), and inserting high permittivity dielectric loads 238, having .sub.r10, with the spacing between adjacent gate segments that exhibit anti-parallel instantaneous current vector alignments 234A,234B (where desired).
(44)
(45) This technique allows a simple circuit layout to easily construct a fairly complex resonant RLC circuit. The complexity of the circuit shown in
(46) A final embodiment of the present invention integrates monolithic power management modules comprising a resonant gate transistor using any of the methods outlined above onto a semiconductor chip carrier.
(47) The present invention is illustratively described above in reference to the disclosed embodiments. Various modifications and changes may be made to the disclosed embodiments by persons skilled in the art without departing from the scope of the present invention as defined in the appended claims.