60 GHz Double Deck T-Gate AlN/GaN/AlGaN HEMT for V-Band Satellites

The influence of double deck T-gate on LG = 0.2 μm AlN/GaN/AlGaN HEMT is analysed in this paper. The T-gate supported with Silicon Nitride provides a tremendous mechanical reliability. It drops off the crest electric-field at gate edges and postponing the breakdown voltage of a device. A 0.2-μm double deck T-gate HEMT on Silicon Carbide substrate offer fMAX of 107 Giga Hertz, fT of 60 Giga Hertz and the breakdown voltage of 136 Volts. Furthermore, it produces the maximum-transconductance and drain-current of 0.187 Siemens/mm and 0.41 Ampere/mm respectively. In addition, the lateral electric-field noticed at gate-edge shows 2.1 × 106 Volts/cm. Besides, the double deck T-gate AlN/GaN HEMT achieves a 45% increment in breakdown voltage compared to traditional GaN-HEMT device. Moreover, it reveals a remarkable Johnson figure-of-merit of 7.9 Tera Hertz Volt. Therefore, the double deck T-gate on AlN/GaN/AlGaN HEMT is the superlative device for 60 GHz V-band satellite application.


Introduction
The constraints of traditional semiconductor devices for Radio frequency and high power microwave application have fashioned the opportunity for wide-bandgap semiconductor materials such as Silicon Carbide, Gallium Nitride (GaN), etc. [1][2][3][4][5]. Gallium Nitride based HEMT devices are attractive for high-frequency and high-power application due to its ultimate material features such as excessive mobility of electron, high breakdown strength, high thermal-conductivity and high saturation velocity of electron [5][6][7][8][9][10]. Gate geometric engineering is a well known technique to enhance the performance of RF and high power devices. Assorted gating effects are achieved using i) T-gate [11] ii) Pi-gate [12] iii) gamma gate [13] iv) camel gate [14] v) gate field plate [15] vi) discrete field plate [16] vii) and multiple grating field plate [17]. The extension of gate head serves as a gate field plate. It suppresses the dispersion and electric field to improve the device reliability [18]. Pi-gate restricts the electron acceleration near the gate edges reducing the hot electron effects [19]. On the other hand, T gate diminishes the gate resistance and enhances the cut-off frequency (f T ). But it also increases the electric field near the gate edges leading to early breakdown of the device [20]. To solve this issue, a double deck T-gate is used in the HEMT device. The double deck T-gate is formed using T-gate tied up with Si 3 N 4 sheath as shown in Fig. 1. It provides an excellent mechanical reliability. In addition, it enhances the device breakdown voltage without sacrificing its maximum oscillation frequency (f MAX ) and cut off frequency (f T ). In this article, we demonstrate the impact of double deck T-gate on the DC and RF behaviour of L G = 0.2 μm AlN/ GaN HEMT using Silvaco Atlas TCAD tool. Furthermore, the merits and the drawbacks of added Silicon Nitride in the device is also summarized. This article is scheduled as follows. Section II provides the physics of a device structure and its dimension. Section III describes the physical model and calibration of a HEMT device. In section IV, the DC and RF performance of double deck T-gate HEMT is compared with the performance of conventional HEMT devices. Section V concludes this article.

SCHEMATICS OF L G = 0.μm AlN/GaN HEMT DEVICES
The device structure of assorted AlN/GaN/AlGaN HEMTs are shown in Fig. 1. All the devices are simulated on 5 μm Silicon Carbide Substrate. The epitaxy comprises of a 200 nm Aluminium Nitride (AlN) nucleation layer, 1.5 μm Aluminium Gallium Nitride (AlGaN) buffer, 150 nm Gallium Nitride channel, 6 nm Aluminium Nitride barrier and 3 nm Silicon Nitride (Si 3 N 4 ) passivation layer. The length of source and drain is defined as 0.5 um. The length (L G ) and width (W G ) of a gate is fixed to 200 nm and 200 μm respectively. The distance between the source and gate (L SG ) = 0.4 μm. The length between the gate and drain (L GD ) is fixed to 2 μm. The height of a stem (H S ) is defined as 200 nm. The height of first head (H H1 ) and the height of second head (H H2 ) are fixed as 50 nm. The length of first head (L H1 ) and the length of second head (L H2 ) are defined as 400 nm and 800 nm respectively. The Silicon Nitride (Si 3 N 4 ) sheath supporting the T-gate enhances the mechanical reliability of a device. It also reduces the maximum electric field (E) near the gate edges to increase the off state breakdown voltage of a device. The double deck T-gate provides better gate control without sacrificing the operating frequency of a device. Owing to spontaneous and piezo-electric polarization, the two dimensional electron gas occurs at the junction of AlN barrier and GaN channel.
Hence, doping is not needed in this device [21]. The AlGaN buffer smoothes the lattice mismatch between the SiC substrate and Gallium Nitride channel. The material parameter used for TCAD simulation is described in Table 1.

Physical Models Used and Calibration of Simulator
The investigations of double deck T-gate on AlN/GaN/AlGaN HEMT is carried out using Silvaco Atlas TCAD simulator. To generate the electrons and holes, continuity equations, Poisson equations and drift diffusion models are included in the device simulation. Without any potential bias, the polarization charges are present at the interface of AlN/GaN layers. These charges are opposite in sign and equal in magnitude to sustain the device charge neutrality. The abstraction method to estimate the polarization charges was reported in [22]. A positive sheet charge density (+σ pol ) of 2 × 10 13 cm −2 is fixed at the interface between the GaN and AlN region. The same concentration of equivalent negative sheet charge (−σ pol ) was defined at the interface of SiN/AlN layer. It is evident from [23] that the surface-donors are accountable for the creation of 2D-elctron gases in Gallium Nitride HEMT device. Therefore, donor-like surface traps with a density of 3.5 × 10 13 cm −2 is defined at the interface of Si 3 N 4 /AlN having the energy of 0.2 eV over mid-gap band. The capture cross sections of hole (σ p ) and electron (σ n ) were presumed to be σ p = σ n = 1 × 10 −15 cm −2 . To generate the charge carriers, constant-mobility and field-dependent mobility models are incorporated in the simulation. To include the electron generation and recombination, Shockley-Read-Hall model is used. In order to perform physical 2D simulation, it is necessary to carry out the  calibration of physical simulator. In addition, the process of calibration is complicated and iterative. The important parameters for the device calibration are i) Velocity saturation of Gallium Nitride ii) mobility iii) surface charge iv) spontaneous polarization v) piezo-electric polarization vi) contact resistance of drain and source terminals vii) work function of gate, etc. The initial calibration is completed by fixing the work function of gate to 4.25 eV [21]. By adjusting the saturation velocity and low field mobility, the saturation and linear regions of drain current (I D ) are then calibrated. To enable the impact ionization effect in breakdown simulation, Selberherr model is used. In addition, Block Newton method with lattice heating models are incorporated to solve drift-diffusion equations [24]. The breakdown simulation shown in Fig. 12 have been calibrated with the experimental results reported in [25].

Results and Discussion
In this paper, the influence of T-gate and double deck T-gate on drain current (I DS ), transconductance (g m ), maximum oscillation frequency (f MAX ), Cut-off frequency (f T ) and breakdown voltage (BV) were analysed using Silvaco Atlas TCAD tool. The device model is simulated and calibrated for the W G = 200 μm HEMT with gate length L G = 200 nm, gate-source distance L GS = 400 nm and the drain-gate distance L GD = 2 μm. The I D -V D curve of [21] is reproduced in this paper using Atlas TCAD simulation. It is obtained for the sweep of drain voltages (V DS ) between −1.4 V and + 1.4 V with a fixed gate bias of V GS = 1 V. It is evident from Fig. 2 that the result of simulated I D -V D characteristics are in good agreement with the result reported in [21]. The conventional AlN/GaN HEMT exhibits the I DS-max of 380 mA/mm, where as AlN/ GaN HEMT with T-gate and double deck T-gate exhibit a I DSmax of 440 mA/mm and 410 mA/mm respectively. T-gate based AlN/GaN/AlGaN HEMT shows 14% improvement in drain-source current (I DS ) compared to conventional HEMT device. It is because of small gate length and large gate area of T-shaped gate, reducing the overall gate access resistance [24]. In Gallium Nitride HEMT, gate to source/drain access region functions as curvilinear resistance which restricts the peak drain current.
The source/drain resistance comprises of bias free access region resistance (R s/d ) and contact resistance (R c ). The flow of current through the access region is expressed in eq. (1).
Where γ is a fitting parameter, V R is the potential drop across access region, V sat is the saturation velocity of an electron, V S is the velocity of electron and Qacc is the charge at access region. The source/drain resistance (R s/d ) can be given in terms of drain source current (I DS ) as expressed in eq. (2).
It is clear from eq.
Where, g m-max is the maximum transconductance in Siemens, ΔI D is the variation in drain-current and ΔV GS is the change in gate-source voltage. The extrinsic gate capacitance consists of two elements 1) fringing capacitances between the access region and gate stem and 2) parallel plate capacitances between the gate and neighbouring metals [26]. The variation of gate-source capacitance (C GS ) is obtained for the sweep of gate voltage (V GS ) between −7.5 V to 0 V at V DS = 1 V as shown in Fig. 5. It is clear from Fig. 5 that the AlN/ GaN/AlGaN HEMT with T-gate offers very low source to gate capacitance of 445 fF/mm. It is due to the high aspect ratios of T-gate stem to diminish the extrinsic capacitance [26]. If the height of T-gate stem is increased beyond 200 nm, there is no significant reduction in extrinsic capacitance is observed. It is because of the parasitic capacitance totally dominated by the capacitance owing to the stem alone. The traditional AlN/GaN HEMT reveals the source-gate capacitance of 5.9 × 10 −13 F/mm, where as the HEMT with double deck T-gate exhibit the C GS of 4.6 × 10 −13 F/mm.
The double deck T gate uses SiN sheath to improve the gating efficiency, but it includes additional parasitic capacitance to the device affecting the overall AC performance of AlN/GaN/AlGaN HEMT. The change in gate-drain capacitance (C GD ) for the gate voltage sweep between −7.5 V to 0 V at V DS = 1 V is illustrated in Fig. 6. It shows the maximum drain to gate-capacitance (C GD ) of 125 fF/mm for conventional HEMT device. Contrarily, the T-shaped gate HEMT and double deck T-gate HEMT offers the gate-drain capacitance (C GD ) of 108 fF/mm and 115 fF/mm respectively. In addition, the gate capacitance (C G ) = C GS + C GD exhibited for the conventional HEMT, T-gate HEMT and double deck T-gate HEMT are 692 fF/mm, 604 fF/mm and 668 fF/mm respectively as shown in Fig. 7.
The cut off frequency of Gallium Nitride HEMT is generally extracted as the sum of parasitic and intrinsic components given in Eq. (4) [20].
Where R d , R g , R s , g d , g m , C GD , C GS denotes drain resistance, gate resistance, source resistance, drain conductance, transconductance, drain to gate capacitance and source to gate    Fig. 9. It offers the maximum-frequency f MAX = 99 GH Z for the conventional HEMT device. On the other hand, it produces the f MAX of 118 GHz and 107 GHz for the AlN/ GaN/AlGaN HEMT with T shaped gate and double deck Tgate respectively. The simulation results reveal higher f MAX and f T for T-gate HEMT compared to the conventional HEMT and double deck T-gate HEMT. It is due to small gate resistance and capacitance offered by T-gate HEMT. It is clear from Fig. 9 that the f MAX of double deck T-gate HEMT is less compared to the f MAX of T-gate HEMT. It is mainly due to the inclusion of Si 3 N 4 sheath under tri-layer T-gate head, which additionally contributes the parasitic capacitance to the device.
The electric field profiles of assorted AlN/GaN/AlGaN HEMTs are shown in Fig. 10. It exhibits the peak electricfield at drain-edge of gate terminal. It offers the electric-field E = 2.9 MV/cm for T-gate based HEMT, whereas the electric field achieved for conventional HEMT and double deck Tgate HEMT are 2.65 MV/cm and 2.1 MV/cm respectively. The simulation results confirm that the double deck T-gate HEMT exhibit very low electric field compared to conventional HEMT and T-gate HEMT. It is due to the incorporation of Si 3 N 4 sheath and additional T-gate head in the device.
It reduces the crest electric-field at drain-edge of gate terminal that improves the device breakdown voltage and mechanical reliability. In Fig. 10, the electric field of the proposed device have additional peaks at about 1.2 um; it is due to the second head of T-gate on the top of the S i3 N 4 sheath acts like a field plate with a short distance to the semiconductor, effectively suppressing the peak field around the edges. The potential distribution of assorted AlN/GaN/AlGaN HEMTs are depicted in Fig. 11. As the drain to source voltage increases, the distribution of potential at drain edge is also increases.
It is because of the increased electric field between the drain and gate regions [27]. Figure 12 shows the off state breakdown behaviour of L G = 200 nm and W G = 200 μm AlN/GaN/AlGaN HEMT devices at V GS = −10 V. The two major components of breakdown mechanism in HEMT devices are 1) Gate associated breakdown and 2) Bulk associated breakdown. Gate associated breakdown is mainly due to the gate-drain leakage current. It is because of the contamination or defects creating a path way for surface conduction [28][29][30]. In order to reduce the gate-drain leakage, Si 3 N 4 sheath is used under double deck T-gate head. It suppresses the crest electric-field at drain-edge of gate terminal, which enhances the device breakdown voltage. On the other hand, the bulk associated breakdown is due to the vertical leakage current between the drain and bulk region [28]. It can be reduced by choosing thick AlGaN buffer layer. Figure 13 shows the drain leakage characteristics of L G = 200 nm AlN/GaN/AlGaN HEMT devices. It offers a very low drain-source leakage current of 9 × 10 −10 A/mm for the HEMT with double deck Tgate. It is due to the sandwich of strong Si 3 N 4 sheath and second T-gate head. At the same time, the conventional and T-gate based HEMT exhibits the drain-source leakage current of 8.5 × 10 −9 A/mm and 9 × 10 −8 A/mm respectively. Hence, a significant improvement in breakdown voltage is achieved for the proposed L G = 200 nm AlN/GaN high electron mobility transistor. It exhibits the off-state breakdown voltage (BV OFF ) of 136 V, whereas the conventional HEMT and Tgate HEMT offers the breakdown voltage (BV OFF ) of 75.4 V and 54 V respectively. HEMT with double deck T-gate exhibits the lowest peak electric fields at the edges of Si 3 N 4 sheath, gate foot, and gate head. This suppression of peak field helps to efficiently reduce the impact ionization rate which in turn needs to the reduction of leakage current and improvement of breakdown voltage.
In order to achieve, the optimal breakdown voltage and cutoff frequency in the proposed double deck T-gate HEMT, the stem height (H S ), second head height (H H2 ) and second head length (L H2 ) are varied. Varying the stem height from 0 to 200 nm causes the significant degradation in the device capacitance. Further any increase in stem height has no effect on device capacitance. Hence, the height of T-gate stem is fixed to 200 nm. On the other hand, to check the trade-off between  Fig. 15 respectively. When the height of second head is increased from 10 nm to 90 nm, it reduces the gate access resistance gradually thereby increasing the transconductance and cut-off frequency of the proposed device. Contrarily, it reduces the device breakdown voltage as shown in Fig. 14, owing to the increased electric field near gate edges. As the length of second head increases from 400 nm to 1200 nm, it improves the smooth distribution of electric field over the device surface.
Hence, the breakdown voltage is enhanced from 110 V to 162 V as shown in Fig. 15. At the same time, it degrades the device cut-off frequency owing to its increased terminal capacitances (C GS and C GD ). In both the plots (Fig. 14 and Fig. 15), the trade-off between the breakdown voltage and cut-off frequency is observed for different H H2 and L H2 . Therefore, Johnson figure of merit (breakdown voltage×cut-off frequency) is computed for the proposed device with various lengths and heights of the second head. It reveals the highest JFOM value of 7.9 Tera Hertz Volt at L H2 = 800 nm and H H2 = 50 nm.

Conclusion
The AlN/GaN/AlGaN HEMT combined with double deck Tgate and Si 3 N 4 sheath provides an excellent AC and DC performance over conventional HEMT and T-gate based HEMT. It reveals 61% improvement in breakdown voltage over Tgate HEMT and 18% improvement in cut-off frequency over Conventional AlN/GaN HEMT. In addition, it exhibits a remarkable Johnson figure-of-merit (f T × BV) of 7.9 × 10 12 V/ s for L G = 200 nm and W G = 200 μm AlN/GaN/AlGaN HEMT device. Furthermore, it has many advantages over Tgate HEMT and conventional HEMT. It exhibits high breakdown voltage, positive threshold voltage, enhanced gating efficiency and very good mechanical reliability. Hence, a double deck T-gate based AlN/GaN/AlGaN HEMT is an excellent device for 60 GHz V-band satellite application.