Patent Application
20230207677
This disclosure relates generally to depletion mode High Electron Mobile Field Effect Transistor (HEMT) semiconductor devices and more particularly to depletion mode HEMT semiconductor devices having Schottky contact layers.
As is known in the art, active semiconductor devices such as for example, diodes and FETs (Field Effect Transistors) based on Group III-Nitride materials, such as GaN, AlGaN, InN, AlN, and ScAlN, have demonstrated excellent RF/Microwave performances, which are achieved with an optimization of material, process, device, and design technologies.
Transistor efficiency determines RF system range, sensitivity, power consumption, and reliability and operating at higher power densities comes with high power dissipation (Pdiss). It is desirable to improve transistor power and efficiency performance, otherwise applications are potentially limited due to thermal management.
In accordance with the present disclosure, a semiconductor device is provided comprising: an orthorhombic polar crystalline oxide κ-Al2O3 layer epitaxially and heterogeneously integrated above a wurtzite single-crystal Group III-Nitride comprising AlN; a source and drain electrodes disposed over a conductive region in the polar κ-Al2O3 layer and partway into the AlN; and a gate electrode disposed between the source electrode and the drain electrode on top of polar κ-Al2O3 layer.
In one embodiment, the upper layer comprises κ-B2O3 or an alloy of κ-Al2O3 and κ-B2O3 (e.g., κ-(AlxB1-x)2O3).
In another embodiment, the semiconductor structure is provided wherein a polarization discontinuity at the heterointerface between the κ-Al2O3 and AlN exists to induce a two-dimensional electron gas (2DEG) in the AlN side of the heterointerface.
The inventors have recognized that the benefits of using an AlN channel layer with a higher bandgap Schottky barrier layer in place of a GaN channel layer to take advantage of large breakdown and thermal conductivity properties of AlN to support ultra-high power density radio frequency (RF) transistors.
Still further, the inventors have recognized that the benefits of an AlN channel layer with a κ-Al2O3 barrier layer provides an overall larger polarization discontinuity and bandgap.
The details of one or more embodiments of the disclosure are set forth in the accompanying drawings and the description below. Other features, objects, and advantages of the disclosure will be apparent from the description and drawings, and from the claims.
Like reference symbols in the various drawings indicate like elements.
Before departing on a detailed description of the invention, it should be noted the inventor has recognized that the benefits of replacing a GaN channel with AlN and using a polar oxide (kappa-phase Al2O3 or κ-Al2O3) as a barrier layer to supply greater than ten times carrier density to AlN via polarization. Using AlN as a channel layer has greater than four times the electric field strength of GaN and therefore can scale voltage by four times and has a greater than two and one half times higher thermal conductivity than GaN for improved thermal management. Homoepitaxy is possible with AlN substrates that currently exist.
The disclosure involves a heterostructure of kappa-phase Al2O3 as the barrier layer to an AlN channel. The very strong polarization of Al2O3 (greater than 9 times higher that of GaN and greater than 3 times of AlN) with the AlN channel (greater than 2.5 times larger thermal conductivity and greater than 4 times electric field strength of GaN (P. Ranga et al., “Highly tunable, polarization-engineered two-dimensional electron gas in ε-AlGaO3/ε-Ga2O3 heterostructures”, Appl. Phys. Exp. 13 061009 (2020))) solves the problem of delivering up to 40 times power density of current state-of-the-art GaN structures with improved thermal conductivity to handle the higher heat dissipation that other technologies cannot achieve without resorting to integrating to diamond. Moreover, AlN is closer lattice matched to diamond so the process to integrate kappa-Al2O3/AlN onto diamond is simpler. AlN substrates exist commercially, so homoepitaxy is possible.
As to be described, the structure enables up to 40 times higher power density of today's GaN structures with improved power dissipation handling integral to the transistor. The latter addresses and alleviates the thermal heat dissipation problem with ultra high power transistors today being pushed for power performance exceeding 8 W/mm where there are currently limited thermal cooling solutions to address which limit application to short duty cycle pulse applications only.
Diamond despite its higher thermal conductivity and potential for ultra-high power density performance has for decades not lived to the promise as it suffers from low carrier density and mobility resulting in transistors that are far less performance than current state-of-the-art GaN structures. Other semiconductor technologies that promise ultra-high power density performance over GaN suffer from poor thermal conductivity relying on complex process to integrate these semiconductors to diamond. None of these technologies have the combined high expected carrier density and mobility combined with high thermal conductivity in one semiconductor platform.
As to be described, with the high polarization properties of orthorhombic κ-Al2O3 (˜3× of AlN) but also large bandgap of 7.7 eV (>1.5 eV higher than AlN); κ-B2O3 has even larger bandgap of 11.3 eV and can also be used as a barrier layer (See Appl. Phys. Lett. 117, 012104 (2020); https://dol.org/10.1063/5.0005808; Submitted: 25 Feb. 2020. Accepted 28 Jun. 2020. Published Online: 9 Jul. 2020).
Referring now to
More particularly, the semiconductor structure 10 includes a single crystal substrate 18, here for example silicon (Si), silicon carbide (SiC), diamond or aluminum nitride (AlN), and an epitaxially grown AlN layer 22 forming nucleation, buffer and channel regions of a HEMT structure, and barrier layer 24 being one or more epitaxial grown polar oxide barrier materials, here κ-Al2O3 or as to be described below, alternatively, κ-B2O3. The oxide-nitride heterolayers 22 and 24 form a heterojunction that results in a 2DEG channel (indicated by dotted line 23) in the AlN layer 22. It is noted that the polar oxide layer 24 extends from an ohmic contact region 27 of the source electrode 26 to an ohmic contact region 29 of the drain electrode 28. The gate electrode 25 is in Schottky contact with the polar oxide layer 24. It is also noted that the source electrode 26 and drain electrode 28 are in ohmic contact with the AlN channel layer 22 via ohmic contact regions 27 and 29, respectively. Traditional ohmic contact formation involving thermal anneal of metal contacts in direct contact with the Schottky barrier is not viable as the mechanism involving the formation of nitrogen vacancies and elemental intermixing of the electrode and barrier material during thermal anneal that renders barrier layer in Group III-Nitride HEMTs conductive forming ohmic contact regions between the electrode and barrier material is absent with κ-Al2O3 barrier. The formation of the ohmic contact regions 27 and 29 in this structure is achieved either by ohmic recess or regrowth process wherein the ohmic contact regions 27 and 29 is either same source and drain electrode material 26 and 28 deposited by e-beam evaporation for example in the case of ohmic recess or a conductive AlN layer deposited by epitaxy techniques (e.g. molecular beam, metal organic chemical vapor phase, physical vapor, etc.) in the case of ohmic regrowth for which either material contacts the 2DEG on the side as indicated in
Referring now to
It should be noted high density 2DEG is possible provided polarization of κ-Al2O3 barrier layer 24 is in same direction as AlN channel layer 22. In one embodiment, the κ-Al2O3 barrier layer is in a range of 3 to 15 nm and here approximately 4 nm thick.
Having described a κ-Al2O3/AlN HEMT structure above, it should be noted that the κ-Al2O3 barrier layer 24 can be replaced with a κ-B2O3, κ-(AlxB1-x)2O3 or alternatively κ-(AlxByGa1-x-y))2O3 layer, provided the bandgap of the latter is larger than AlN.
A number of embodiments of the disclosure have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the disclosure. The composition of the channel and barrier layers do not need to be homogeneous throughout and may consist of multiple layers, multiple combinations of Group III elements, or gradients of Group III element compositions. Further, it should be understood that other single crystal substrates 18 may be used that enables the deposition of AlN with a single well defined crystalline orientation with respect to the substrate 18 crystal structure. This includes heterojunction structures formed via the deposition of one or more crystalline materials on another crystalline material or formed by bonding one or more layers together to define a surface region that is crystalline and supports the crystalline growth of AlN. Accordingly, other embodiments are within the scope of the following claims.
