(1) Field of the Invention
This invention relates to field effect transistors, in particular to a multiple channel, ultra-linear field effect transistors.
(2) Description of the Related Art
A field effect transistor (FET) normally is a square-law device. In the current saturation of the drain current (ID) vs gate-to-source voltage (VGS) of the characteristic of a FET, the basic relationship is given as:
ID=K(VGS−Vt)2 (1)
where K is a transconductance parameter, and Vt is the threshold voltage. When an input voltage is applied as a change in VGS, the output current, which appears as the change in ID, does not vary linearly with VGS. This square-law relationship causes non-linearity (harmonic distortion) and limits the dynamic range of amplifiers.
In equation (1), the threshold voltage Vt is assumed to be constant, based on a uniform impurity concentration N(x) of the semiconductor substrate. This threshold voltage is the voltage required to creating a maximum depletion layer in the substrate after strong inversion.
When the substrate concentration is not uniform, equation (1) must be modified, it was revealed by R. A. Pucel in a paper “Profile design for distortion reduction in microwave field-effect transistors” Electronic Letters, vol. 14, P. 204, 1978, that the ID can be characterized in terms of the non-linear distortion components as:
ID=Ido+Gm(0VGS+½gm(1)VGS2+⅙gm(2)VGS3+ (2)
where gm(n)VGSn is the nth order distortion and gm(n) is the transconductance and its derivatives with respect to the gate voltage. Linear device operation occurs at signal levels (VGS˜0) or when the higher order terns have been eliminated Signal distortion in amplifiers is due to the higher order terms, which become important at high signal levels.
To improve the linearity of a FET, the transconductance must be constant with varying gate voltage VGS. The transconductance is strongly dependent on the electron distribution in the channel of the FET. Thus the design of a linear transistor demands paying attention to carrier distribution.
A measure of the linearity of a FET is the third order intermodulation ratio (IMR), which is the ratio of the power generated in a spurious third order intermodultion signal relative to the power in the dc desired signal. IMR has been shown to be proportional to the ratio of the third order component relative the first order (fundamental) component in Equation (2). Further more, Pucel has shown that this ratio can be expressed as:
IMR∝|gm(2)/gm(0)|=(κεo/q)2|d/dx(1/x3N(x)|/N(x) (3)
where N(x) is the charge distribution in the channel, κ is the relative dielectric constant, εo is the permittivity of free space, q is electronic charge and equation (3) is evaluated at the depletion edge boundary. The requirement for a highly linear device is for IMR to be small (close to 0). Therefore, from equation (3), either x3N(x) is constant or x3 tends to infinity. The former occurs when N(x) varies as 1/x3. In the U.S. patent application Ser. No. 09/618,884, we proposed to implement such a doping profile using multi-channel to achieve linearity in Hetrojunction Field Effect Transistors (HFETs). However, the invention was limited to specific doping profile in selected semiconductor materials such as InP and GaAs. Further, the lower mobilities of the HFET limit its applications to lower frequencies.
In a standard single channel High Electron Mobility Transistors (HEMTs), higher mobilities and sheet concentrations can be achieved as compared to HFETs. Therefore HEMT devices offer higher breakdown voltages and cut-off frequencies, low noise and higher power. There are numerous prior arts on single conducting channel HEMTs and its application to low noise and power applications. In U.S. Pat. No. 6,121,641 Ohno addressed the shielding of traps and suppression of short channel effect by insertion of p-type layer in a single conducting channel HEMT, and U.S. Pat. No. 5,767,539, Onda disclosed a field effect transistor having various designs for donor supply layer but with only one channel. Further, in the U.S. Patents to Mishima (U.S. Pat. No. 5,633,516), to Nakayama (U.S. Pat. No. 5,856,685), to Hida (U.S. Pat. No. 6,049,097), and to Matloubian (U.S. Pat. No. 5,663,583) variations to the single channel HEMT such as; two charge supply layer on either side of the channel layer, different material for spacer, various compositions of buffer and or channel have been proposed. However, there is no provision that the device can improve the linearity of the drain characteristic. U.S. Pat. No. 5,739,559 (the '559) to Isheda et al discloses a HEMT device having improved linearity. The patent '559 relates to a HEMT having asymmetrical carrier supply layers sandwiching a channel layer. More specifically, the heterojunction barrier height between the lower carrier supply layer and the channel layer is greater than the heterojunction barrier height between the upper carrier supply layer and the channel layer. Further, the patent '599 discusses only a single channel HEMT and silent regarding a HFETs in general, or having multiple channels in HFET and HEMT, or any specific carrier concentration and type of doping, or different materials for the multiple channels.
U.S. Pat. No. 6,015,981 to Gluck (“the patent '981”) relates to a HFET having high modulation efficiency.
Thallium (Tl) compounds of varying composition can be lattice matched to GaAs, InP and InAs. High electron mobilities have been predicted for TlP, TlAS, and TlSb (Schilfgaard et al, Applied Physics Letter, Vol. 65, pp 2714, 1994). Therefore FET with thallium compounds channels has the advantage of achieving highest gm, and cut-off frequencies exceeding the current state-of-the-art. Recently Tl compounds (U.S. Pat. No. 5,841,156 and references therein) have been proposed as detector, FET and HBT material. However, multi-channel HEMT or HFET for linearity has not been proposed.
An object of this invention is to obtain a linear dc ID variation as a function of VGS and a constant transconductance variation in the characteristics of a FET over a wide range of VGS. Another object of this invention is to obtain a sharp impurity gradient in the channel of a FET. Still another object of this invention is to apply the multi-channel concept to other semiconductor material systems such as TlInP, TlGaInP, InAs, InAsSb etc selected from the Table 1. A further object of this invention is to optimize the impurity gradient in the channels of a HFET to obtain constant cut-off frequency variation over a wide range of VGS.
These objects are achieved by using multiple channels for a FET. Alternate layers of doped and undoped different kinds of semiconductors form heterojunctions in the multiple channels. The heterojuncitons confine the electrons in separate thin spikes. A number of spikes of different electron concentrations can result in a sharp overall electron concentration gradient such as 1/x3 electron concentration profile. Such an electron concentration gradient can result in a linear variation of drain current with gate voltage to obtain a wide dynamic range.
In the second embodiment of the invention is to use multi-channel HEMT structure to increase the dynamic range of a FET. Specifically, one of the objects of this invention is to optimize the impurity gradient in the channels of a FET to obtain constant cut-off frequency variation over a wide range of VGS.
These objects are achieved by heterojunction FET having multiple two dimensional electron (or hole) gas channels formed by using alternate layers of narrow band gap and wide band gap different kinds of semiconductors (form heterojunctions) deposited on a substrate with a buffer layer. In the invention multiple two dimensional electron (or hole) channels are formed in narrow band gap semiconductor by inserting a thin planar or spikes doping in the multiple wide band gap semiconductor (charge supply) layers. The multiple number of spikes of different acceptor/donor concentrations in the wide band gap semiconductor can result in any arbitrary overall hole/electron concentration gradient or specifically 1/x3 electron concentration profile. Such an electron concentration gradient (1/x3) can result in a linear variation of drain current with gate voltage to obtain a wide dynamic range. In the third embodiment of the invention is to implement a multi-channel Insulated Gate HFET and HEMT to further increase the breakdown voltage and dynamic range of a FET.
Table 1 shows the material structures of the multi-channel FET (HFET/HEMT) of the present invention for the first and second embodiment.
An example of the multiple channels GaN/AlN HFET having n-type conducting is shown in
While
In the second embodiment of the invention is to use multi-channel HEMT structure to achieve ultra-linear device. A conventional and generic single channel HEMT structure uses a buffer layer, channel layer, spacer layer, charge supply layer, Schottky barrier, cap, source, drain, and gate formed on a substrate. In the single channel HEMTs higher mobilities and sheet concentrations can be achieved as compared to HFETs. HEMT devices offer higher breakdown voltages and cut-off frequencies, low noise and higher power. In the prior art variations to the single channel HEMT such as; two charge supply layer on either side of the channel layer, different material for spacer, various compositions of buffer and or channel have been proposed. Our device design consists of multiple channels (2-DEG or 2-DHG) and the location and number of the conducting channel distinguish from the prior art. In the multi-channel HEMTs higher mobilities and sheet concentrations can be achieved as compared to HFETs, the first embodiment of the invention. Therefore the proposed devices can be used where higher breakdown voltages and higher operating frequencies, low noise and higher power are required.
A generic structure of the said device is shown in the
The spike dopings 32, 35, 40, and 43 can be Si, Sn, Te and Be, C for n and p-type FETs. Spike provides the donor or acceptor supply layer for the FET. Using the spike doping or delta or planar doping reduces the distance between the gate and the channels that need to be optimize for sub-micron devices. Further, the breakdown voltage of the multi-channel HEMT will be higher due to quantum confinement In the
The multi-channel HEMT is capped with doped or undoped layer 46, or in combination of oxidation barrier. The cap can be N+GaAs, N+GaInAs and the oxidation barrier can be AlInAs for AlAsSb. For example the optional oxidation barrier can be used.to protect layers such as AlSb and AlAsSb. The doped GaInAs (or InAs) used as the cap requires recess etching before forming metal gate. A thin layer of undoped GaAs cap can be used to improve the uniformity of the threshold voltage of the HEMT. Over the layer 46, the source, gate and drain are formed laterally. Ohmic contacts are sintered to form the source and drain, which make contacts to the multiple channels. Specific layers that can be used to implement the linear device are shown in the Table 1. In the table multi-channel HEMTs can be implemented on InP, GaAs, and sapphire substrates using lattice matched structures or mismatched structures using metamorphic buffer to accommodate the lattice mismatch. The number of channels can be selected depending on the linearity requirement, or other applications. The conductivity of the channels can be either n or p-type. The HEMT structure can be designed to give optimum noise figure parameters (peak gm at 10% Idsat), and uniform gain, Ft, and Fmax with gate voltage. For example by proper selection of charge density, peak in gm close to pinch-off and or zero gate voltage can be realized in multi-channel HEMT (
An example of multiple channel AlInAsSb/GaInAs HEMT structure is shown in
Thallium (Tl) compounds of varying compositions can be lattice matched to InP, GaAs, and InAs substrates. For example TlGaInP, TlInP and TlGaInAs can be used lattice matched to InP and TlGaP, TlGaInP to GaAs and TlInAs to InAs substrate. In this invention we also propose Tl compounds for realizing multi-channel HEMT and HFET for linearity.
In the third embodiment of the invention is to use multi-channel HEMT and HFET structures to create an insulated gate FET (IGFET) (shown in
A variation of IGFET is to use all undoped wide band gap/narrow band gap layers and implants 49 (n or p) self aligned with gate to accumulate electron or holes in the channels. For example Si/SiGe, InP/GaInAs, GaAs/AlGaAs, GaN/AlN etc pairs can be used. Further, by selective implantation of n and p, a complimentary FET can be implemented. A further variation of the IGFET is to use the Type II band line up pair such a GaSb/InAs, GaAsSb/GaInAs to form complimentary FETs by accumulating holes in GaSb (GaAsSb) and electrons in InAs (GaInAs). In these material system, a single implant (Si) can be used for n and p doping (for Sb>0.6). The combination of wide band gap/narrow band gap, barrier, and of the cap can be extended to other semiconductor.
While
This application is a continuation of application Ser. No. 10/176,787, filed Jun. 24, 2002, now U.S. Patent No. 6,992,319, which is a continuation-in-part of application Ser. No. 09/618,884, filed Jul. 18, 2000, now abandoned.
Number | Name | Date | Kind |
---|---|---|---|
5227644 | Ueno | Jul 1993 | A |
5254863 | Battersby | Oct 1993 | A |
5266506 | Green, Jr. | Nov 1993 | A |
5767539 | Onda | Jun 1998 | A |
5770868 | Gill et al. | Jun 1998 | A |
6064082 | Kawai et al. | May 2000 | A |
6121641 | Ohno | Sep 2000 | A |
6232624 | Matloubian et al. | May 2001 | B1 |
6414340 | Brar | Jul 2002 | B1 |
Number | Date | Country | |
---|---|---|---|
20050285098 A1 | Dec 2005 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 10176787 | Jun 2002 | US |
Child | 11205484 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 09618884 | Jul 2000 | US |
Child | 10176787 | US |