GaN VERTICAL-CHANNEL JUNCTION FIELD-EFFECT TRANSISTORS WITH REGROWN p-GaN BY METAL ORGANIC CHEMICAL VAPOR DEPOSITION (MOCVD)

Abstract

Fabricating a vertical-channel junction field-effect transistor includes forming an unintentionally doped GaN layer on a bulk GaN layer by metalorganic chemical vapor deposition, forming a Cr/SiO.sub.2 hard mask on the unintentionally doped GaN layer, patterning a fin by electron beam lithography, defining the Cr and SiO.sub.2 hard masks by reactive ion etching, improving a regrowth surface with inductively coupled plasma etching, removing hard mask residuals, regrowing a p-GaN layer, selectively etching the p-GaN layer, forming gate electrodes by electron beam evaporation, and forming source and drain electrodes by electron beam evaporation. The resulting vertical-channel junction field-effect transistor includes a doped GaN layer, an unintentionally doped GaN layer on the doped GaN layer, and a p-GaN regrowth layer on the unintentionally doped GaN layer. Portions of the p-GaN regrowth layer are separated by a vertical channel of the unintentionally doped GaN layer.

Claims

1.-10. (canceled)

11. A vertical-channel junction field-effect transistor comprising: a doped GaN layer; an unintentionally doped GaN layer on the doped GaN layer; and a p-GaN regrowth layer on the unintentionally doped GaN layer, wherein portions of the p-GaN regrowth layer are separated by a vertical channel of the unintentionally doped GaN layer.

12. The vertical-channel junction field-effect transistor of claim 11, further comprising gate electrodes on the p-GaN regrowth layer.

13. The vertical-channel junction field-effect transistor of claim 12, wherein the gate electrodes are separated by the vertical channel of the unintentionally doped GaN layer.

14. The vertical-channel junction field-effect transistor of claim 13, further comprising a source electrode on the vertical channel of the unintentionally doped GaN layer.

15. The vertical-channel junction field-effect transistor of claim 14, further comprising a drain electrode on the doped GaN layer.

16. The vertical-channel junction field-effect transistor of claim 15, wherein a regrowth interface is defined between the p-GaN regrowth layer and the unintentionally doped GaN layer.

17. The vertical-channel junction field-effect transistor of claim 16, wherein the vertical channel of the unintentionally doped GaN layer forms a vertical p-n junction at the regrowth interface between a gate electrode and a drain electrode.

18. The vertical-channel junction field-effect transistor of claim 17, further comprising a lateral p-n junction at the regrowth interface perpendicular to the vertical p-n junction, wherein the lateral p-n junction is between the gate electrode and a source electrode.

19. The vertical-channel junction field-effect transistor of claim 11, wherein the vertical channel is rectangular in shape.

20. The vertical-channel junction field-effect transistor of claim 11, wherein the vertical channel defines 90° angles with respect to the p-GaN regrowth layer.

Description

BRIEF DESCRIPTION OF DRAWINGS

[0017] FIGS. 1A-1L depict steps in an operation to form a vertical-channel junction field-effect transistor (VC-JFET).

[0018] FIGS. 2A and 2B show crystal planes in GaN wurtzite structure and schematics of the fin alignment direction, respectively. FIGS. 2C and 2D show SEM cross-sectional images of the VC-JFETs in m-plane and a-plane, respectively.

[0019] FIG. 3 is a schematic of a cross section of a GaN VC-JFET.

[0020] FIG. 4A shows I-V curves for vertical p-n junctions between the gate and the drain. FIG. 4B shows I-V curves for the lateral p-n junctions between the gate and the source. The insets show schematics of the p-n junctions in GaN VC-JFETs. FIG. 4C shows reverse leakage current of a planar diode and a p-n junction between the gate and the drain.

[0021] FIG. 5A shows representative ID, IG-VGS transfer characteristics in linear scale. FIG. 5B shows representative ID-VDs output characteristics.

[0022] FIG. 6 shows and OFF-state I-V curve of the GaN VC-JFETs.

[0023] FIGS. 7A and 7B show transfer characteristics of VC-JFETs aligned to m-plane and a-plane, respectively.

[0024] FIGS. 8A and 8B show cross-sectional SEM images of VC-JFETs devices aligned to a-plane and m-plane, respectively. FIGS. 8C and 8D show monochromatic CL images at 3.25 eV for devices aligned a-plane and m-plane, respectively. FIGS. 8E and 8F show monochromatic CL images at 2.9 eV for devices aligned to a-plane and m-plane, respectively.

DETAILED DESCRIPTION

[0025] FIGS. 1A-1L depict steps in an operation to form a vertical-channel junction field-effect transistor (VC-JFET) 100. In FIG. 1A, an unintentionally doped (UID) GaN layer 102 is homoepitaxially grown on a heavily doped bulk GaN substrate 104 by a metalorganic chemical vapor deposition (MOCVD) process. In FIG. 1B, a Cr/SiO.sub.2 hard mask is formed on the UID GaN layer 102. In one example, SiO.sub.2 layer 106 has a thickness of about 700 nm and Cr layer 108 has a thickness of about 50 nm. In FIG. 1C, fins 110 are patterned by electron beam lithography and aligned to either a-plane or m-plane. In FIG. 1D, reactive ion etching (ME) is used to define the Cr and SiO.sub.2 hard masks 106′, 108′. To improve the regrowth surface, the fins 110 are formed by inductively coupled plasma (ICP) etching as shown in FIG. 1E. In FIG. 1F, the hard mask residuals 106′, 108′ are removed by Cr etchant and hydrofluoric acid (HF). The sample is cleaned by a variety of processes (e.g., dipping in tetramethylammonium hydroxide (TMAH), immersing in piranha, treating with UV-ozone, performing a buffered oxide etch (BOE) and hydrochloric (HCl) acid for 5 minutes. The cleaned sample depicted in FIG. 1G is re-loaded into a MOCVD chamber for p-GaN regrowth to yield the structure depicted in FIG. 1H with regrowth interface 112′. A photoresist planarization process is conducted to selectively etch away the p-GaN on top of the fin as depicted in FIG. 1I. Top p-GaN etching is performed to expose the n-GaN 102′ for source electrodes, as depicted in FIG. 1J. In FIG. 1K, the gate electrodes 114 were formed by electron beam evaporation. FIG. 1L depicts the source and drain electrodes 116, 118 deposited by electron beam evaporation.

EXAMPLES

[0026] Device epilayers were grown by MOCVD where trimethylgallium (TMGa) served as the Ga precursor and ammonia (NH.sub.3) was the source for nitrogen. The carrier gas was hydrogen (H.sub.2). A 4-μm-thick unintentionally doped (UID) GaN was homoepitaxially grown on heavily doped bulk GaN substrates by metalorganic chemical vapor deposition (MOCVD). Device fabrication started with the deposition of Cr (50 nm)/SiO.sub.2 (700 nm) hard mask. The fins were patterned by electron beam lithography and aligned to either a plane or m plane. A chlorine (Cl.sub.2) and a fluorine (F.sub.2) based reactive ion etching (RIE) were used to define the Cr and SiO.sub.2 hard masks, respectively. To improve the regrowth surface, the fins were formed by a two-step inductively coupled plasma (ICP) etching: a six-minute fast etching (˜280 nm/min), and a three-minute slow etching (˜20 nm/min). The hard mask residuals (Cr and SiO.sub.2 ) were removed by Cr etchant and hydrofluoric acid (HF). The sample was then dipped in 75° C. 25% hot tetramethylammonium hydroxide (TMAH) for 5 minutes to further recover etching damages created by ion bombardments during ICP etching. The samples were then immersed in piranha for 15 minutes to further remove possible organic contaminants. The samples were then treated by UV-ozone for 1 hour, followed by buffered oxide etch (BOE) and 10% hydrochloric (HCl) acid for 5 minutes, respectively, to remove surface charges. The cleaned samples were re-loaded into MOCVD chamber for p-GaN regrowth. 1 μm p-GaN (10E17 cm.sup.−3) was successively grown on the sample with bis(cyclopentadienyl)magnesium (Cp.sub.2Mg) as the Mg precursor. After the regrowth, the activation of the regrown p-GaN was conducted at 700° C. for 20 minutes. A photoresist planarization process was used to selectively etch away the p-GaN on top of the fin and expose the n-GaN for source contacts. The gate electrodes were formed by Pd/Ni/Au (30/20/100 nm) by electron beam evaporation and annealed in 450° C. for 5 minutes. The source and drain electrode Ti/Al/Ni (30/100/30 nm) were deposited by the electron beam evaporation.

[0027] The VC-JFET devices with fins aligned to either a-plane or m-plane were fabricated. FIG. 2A shows the schematic of the GaN crystal planes. The alignment direction relative to orientation flat (OF) is shown in FIG. 2B. The wet etching of TMAH in GaN is strongly anisotropic, which results in different sidewall profiles for the two different crystal orientation. The device aligned to m-plane in FIG. 2D shows a narrower and more vertical fin 110 than that the device aligned to a-plane in FIG. 2C.

In the GaN VC-JFETs 100, two types of p-n junctions were formed by the p-GaN regrowth: the lateral p-n junction 300 and the vertical p-n junction 302 at regrowth interface 304, as shown in FIG. 3. The lateral p-n junction was between the gate electrode and the source electrode, and the vertical p-n junction was between the gate electrode and the drain electrode. The vertical channel 306 is rectangular in shape, and defines 90° angles with respect to the p-GaN regrowth layer 112.

[0028] The current-voltage (I-V) characteristics of the two junctions were measured by a Keithley 2410 source meter. FIGS. 4A and 4B show the representative I-V curves of the vertical and lateral p-n junctions. The currents were normalized by the effective gate region. Similar forward rectifying characteristics with an on/off ratio of —10.sup.5 were observed for both junctions. At forward bias, strong electroluminescence (EL) was observed, which showed the p-GaN regrowth was effective. The turn-on voltages, as extracted from linear extrapolation were 3.4 V and 3.1 V in the vertical and lateral p-n junctions, respectively. This larger turn-on voltage of the vertical p-n junction is likely due to thicker n-GaN layer in the vertical direction. The two junctions showed abnormal ideality factors of over 4. This is likely caused by the regrowth process where defects, etching damages and impurities can serve as non-radiative recombination centers at the regrowth interfaces. The lateral and vertical p-n junctions show similar leakage currents.

[0029] To understand leakage paths for the regrown p-n junctions, the reverse leakage characteristics of a planar diode (without fins) and a vertical p-n junction between the gate and the drain (with fins) on the same wafer are compared in FIG. 4C. Both p-n diodes show larger leakage currents compared with as-grown diodes. The discrepancy between the two diodes become larger with higher reverse bias. This suggests that both non-polar and polar regrowth interfaces contribute to the leakage current. The total leakage is the superposition of leakage currents from all etched surfaces. The c-plane interface may be dominate at low bias (<2 V), and the leakage current through non-polar planes increases with increasing reverse bias. The leakage currents continue to increase after 10 V with no signs of saturation.

[0030] The transistor characteristics were measured by a Keithley 4200 SCS parameter analyzer. The effective area for the source region is 960 μm.sup.2 consisting of six fins each with an area of 1 μm×160 μm. The total device area, including the source/gate electrode areas, is ˜2.5×10.sup.5 μm.sup.2. FIG. 5A shows the transfer characteristics of the VC-JFET with fins aligned to m-plane. The threshold voltage (Vth) determined by linear extrapolation of the drain current is ˜1.4 V. At a drain voltage of 10 V, the on/off ratio is ˜100. As the gate voltage increased to 2.2 V, the gate current began to increase due to the turn-on of the gate to source p-n junction, and as a result, the drain current slightly decreased. A hysteresis of ˜0.4V was observed in the I.sub.G-V.sub.GS transfer curve due to charges at the regrowth interface.

[0031] FIG. 5B shows the output I.sub.D-V.sub.DS family curves. Gate modulation was obtained, but no current saturation was observed in the VC-JFET mainly due to the gate leakage. The drain current density reached 740 A/cm.sup.2 with an R.sub.on,sp of 10.5 mΩ.Math.cm.sup.2 (normalized by the fin area) at a V.sub.G bias of 3 V and VDS bias of 12 V. The on-resistance normalized by the total device region is 2.8 Ω.Math.cm.sup.2. Since non-alloyed contacts for the source and the drain were used, thermal annealing may further decrease the on-resistance.

[0032] A typical OFF-state I-V curve of the GaN VC-JFETs is shown in FIG. 6. The drain current and gate current were normalized by effective fin region and gate region, respectively. The gate current was mainly the superposition of leakage current from both lateral and vertical p-n junctions. The drain current consists of leakage current from vertical p-n junctions and the leakage current induced by the weak pinch-off effects of lateral p-n junctions. The drain current is around ten times higher than the gate current, indicating the drain leakage contributed to the main leakage in drain current at off-state. The drain current broke down at 45 V, while no breakdown was observed at the gate. These results suggest that the off-state characteristics of this device are primarily limited by weak pinch-off effects of the lateral p-n junctions.

[0033] FIGS. 7A and 7B compare the transfer characteristics of the VC-JFETs aligned to m-plane and a-plane, respectively. The peak transconductances are 625 S/cm.sup.2 and 916 S/cm.sup.2 (normalized by fin areas) in m-plane and a-plane devices, respectively. The threshold voltage (V.sub.th) voltages are determined by the following method: first the point of maximum transconductance on the I.sub.D-V.sub.GS curve was found, and the maximum slope was extrapolated to I.sub.DS=0 to find the x-axis intercept. The m-plane and the a-plane devices showed turn-on voltages at ˜1.5 V and ˜0.5 V, respectively.

[0034] The a-plane device showed higher on-current compared to m-plane. This discrepancy is likely due to non-alloyed source contacts. The UID-GaN was treated with ICP etching to created donor-like surface defects to facilitate drain contacts. Different surface conditions before electrode deposition may lead to non-uniform contact resistance in different devices. The gate leakage in a-plane devices is around ten times higher than that of m-plane devices. This difference in gate leakage is likely related to different crystal orientations. It is possible that the defect and impurity level at m-plane could be lower than a-plane after TMAH treatment, resulting in lower gate leakage.

[0035] Interfacial impurities such as silicon (Si) and oxygen (O) have been shown to strongly effect GaN regrown p-n diodes. Using electron holography, the electrostatic potential profile at the regrowth interface was obtained. The energy band diagram showed large band bending at the regrowth interface. This indicates the formation of p.sup.+/n.sup.+ tunneling junction at the regrowth interface. The n.sup.+-GaN is due at least in part to high concentration of Si and O impurities acting as shallow donors, and the p.sup.+ doping at the regrowth interface is likely due to Si—Mg and/or O—Mg co-doping effects. The overlapping of Mg and Si/O at the interface region can enhance the hole concentration up to two orders of magnitude. The formation of tunneling junctions at the regrowth interface may be responsible for the large leakage, premature breakdown in regrown p-n junctions, and weak pinch-off effect of the lateral p-n junctions.

[0036] Another factor that may be limiting the device performance is the nonuniform distribution of acceptors in GaN epilayers grown on fin structures. Cathodoluminescence (CL) spectroscopy was used to study the optical properties of p-GaN in VC-JFET devices. The CL was carried out in a JEOL 6300 scanning electron microscope. The electron beam current was 100 pA and the acceleration voltage was 7 kV. CL mappings were obtained by recording the spatial variation of luminescence intensity over an area at a certain wavelength. FIGS. 8A-8F show the secondary electron (SE) images, and CL mappings at 3.25 eV and 2.9 eV of m-plane and a-plane devices. 2.9 eV emission intensity in CL can be used as an indicator for acceptor concentrations. The higher the CL intensity at 2.9 eV, the higher the acceptor concentration.

[0037] FIGS. 8A and 8B show cross-sectional SEM images of VC-JFETs devices aligned to a-plane and m-plane, respectively. FIGS. 8C and 8D show monochromatic CL images at 3.25 eV for devices aligned a-plane and m-plane, respectively. FIGS. 8E and 8F show monochromatic CL images at 2.9 eV for devices aligned to a-plane and m-plane, respectively. FIGS. 8E and 8F indicate that weaker intensity was observed on the sidewalls of the fins, indicating that the p-GaN grown on the sidewall of the fins had lower acceptor concentration. The interface between acceptor deficient p-GaN and normal p-GaN makes a 45° angle with the basal plane in both m-plane and a-plane devices. This indicates that this phenomenon is not related to crystal orientations, and possibly due to regrowth conditions and/or surface treatment.

[0038] All the aforementioned non-ideal factors indicate that devices regrowth with trenches is much more complicated and difficult than regrowth on planar surfaces. Methods should be explored to eliminate impurities on c-plane and non-polar surfaces simultaneously, as non-polar planes have highly different material properties from the polar c-planes such as dangling bond densities, surface states and impurities incorporation. In addition, the regrowth conditions (e.g., temperature, pressure, III/V ratio) also need to be optimized to improve acceptor concentrations at the sidewall.

[0039] GaN VC-JFETs were demonstrated through p-GaN regrowth on the patterned fin-like channel regions by MOCVD. A subsequent photoresist self-planarization process was applied to etch away the p-GaN on top of the n-GaN fins. The regrown lateral and vertical p-n junctions were characterized to verify the effectiveness of the regrowth process. The VC-JFETs show an on-off ratio ˜100 and excellent transconductances. Devices with vertical channels aligned to a-plane and m-plane were also compared. The nonideal factors such as interfacial impurities and non-uniform acceptor distribution were also discussed.

[0040] Although this disclosure contains many specific embodiment details, these should not be construed as limitations on the scope of the subject matter or on the scope of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in this disclosure in the context of separate embodiments can also be implemented, in combination, in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments, separately, or in any suitable sub-combination. Moreover, although previously described features may be described as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can, in some cases, be excised from the combination, and the claimed combination may be directed to a sub-combination or variation of a sub-combination.

[0041] Particular embodiments of the subject matter have been described. Other embodiments, alterations, and permutations of the described embodiments are within the scope of the following claims as will be apparent to those skilled in the art. While operations are depicted in the drawings or claims in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed (some operations may be considered optional), to achieve desirable results.

[0042] Accordingly, the previously described example embodiments do not define or constrain this disclosure. Other changes, substitutions, and alterations are also possible without departing from the spirit and scope of this disclosure.