Design of a Schottky Metal-Brim Structure to Optimize 180–220 GHz Broadband Frequency Doubler MMIC

Abstract

The present work proposes a 180–225 GHz broadband frequency doubler monolithic microwave integrated circuit (MMIC) based on a novel Schottky barrier diode (SBD) terminal structure denoted as a Schottky metal-brim (SMB). Compared with an MMIC adopting the conventional SBD terminal structure, preliminary measurements show that the maximum output power of the MMIC adopting the SMB structure increases from 0.216 mW at 206 GHz to 0.914 mW at 208 GHz. Analysis of the nonlinear current–voltage and capacitance–voltage characteristics of the two terminal structures based on an extended one-dimensional drift-diffusion model, indicates that the SMB structure provides significantly better conversion efficiency than the conventional SBD structure by eliminating the accumulation of charge and additional current paths near the Schottky electrode edge. It provides a feasible scheme for the optimization of MMIC applications requiring high power and high efficiency.

1. Introduction

Terahertz science and technology have developed rapidly in recent decades, and this has promoted the use of THz radiation in many high-requirement applications, such as astrophysics and earth science, high-speed communication, biomedicine, and safety imaging [1,2,3]. A critical component in THz systems is electronic sources based on a local oscillator (LO) followed by a combination of amplifiers and frequency multipliers with sufficient output power [4,5]. Accordingly, the tremendous interest in THz technology has led to the development of a series of local oscillators operating in the THz band (i.e., 100 GHz to 30 THz) [6]. Among these, frequency multiplier chains based on Schottky diodes have been commonly employed due to their advantages of low cost, frequency agility, easy integration, and good functionality both at room temperature and at cryogenic temperatures [7,8,9]. Meanwhile, a number of advanced semiconductor technologies have been developed worldwide for frequency multiplier applications in the THz band. For example, Virginia Diodes Inc. (VDI), the Jet Propulsion Laboratory (JPL), and Advanced Compound Semiconductor Technologies (ACST) GmbH have already developed all-solid-state frequency multiplier monolithic microwave integrated circuits (MMICs) with high efficiency and an output power of a dozen milliwatts in the frequency range of 200–600 GHz [10,11]. In contrast, frequency multiplier research in China has mainly focused on hybrid integrated circuits with discrete Schottky diodes operating at a frequency of around 200 GHz. As such, the development of frequency multiplier MMICs based on Schottky diodes has been rarely reported in China. Among the very few examples, Ren et al. developed a 330–500 GHz broadband tripler MMIC [12]. However, the MMIC exhibited low efficiency, with an output power that was only on the order of a hundred microwatts.

Broadband frequency multipliers function at low forward and reverse biases by applying a small input signal, where the capacitance–voltage and current–voltage relationships are the dominant nonlinear terms. Conversion efficiency is an important criterion for evaluating the performance of frequency multipliers. This efficiency is affected primarily by circuit mismatching, parasitic parameters, increasing resistance, and the power saturation effect. The fact that Schottky diodes form the core device of frequency multipliers suggests that efforts to increase the power output and efficiency of broadband frequency multipliers should focus on increasing the efficiency of Schottky diodes.

The present work addresses the above-discussed issues by developing a 180–225 GHz broadband frequency doubler MMIC based on a novel Schottky barrier diode (SBD) terminal structure denoted as Schottky metal-brim (SMB). Preliminary measurement results demonstrate that the maximum efficiency and maximum output power of an MMIC adopting the proposed SMB structure are 8.39% and 0.92 mW, respectively, at 208 GHz, while those of an MMIC adopting the conventional SBD structure are 2.81% and 0.20 mW, respectively, at 212 GHz. The cause of the improved performance of the proposed SMB terminal structure is evaluated through a comparative analysis of the nonlinear current–voltage and capacitance–voltage characteristics of the two terminal structures based on an extended one-dimensional (1-D) drift-diffusion (DD) model [13]. The models were employed directly in design automation tools, such as Advanced Design System (ADS) simulation environment, to evaluate the frequency responses of the SBDs accurately.

2. Materials and Methods

The proposed SMB terminal structure is illustrated in Figure 1a, while the conventional SBD terminal structure is illustrated in Figure 1b. The proposed SBD structure is designed to eliminate the accumulation of charge and additional current paths near the Schottky electrode edge by better isolating the n-type GaAs epitaxial layer (epilayer), which is beneficial for MMIC applications requiring high power and high efficiency. We selected GaAs as a commonly used material for frequency multipliers based on Schottky diodes due to its high electron mobility and the small reverse saturation current provided by the resulting diodes. Based on our previously published studies [12,14], the diode structure was fabricated on a GaAs wafer with epitaxial lattice-matched GaAs layers, including a 1.5-μm-thick heavily doped n-type GaAs buffer epilayer with an Si doping concentration of ~10¹⁸ cm⁻³ and a 0.3-μm-thick lightly doped n-type GaAs epi-layer with an Si doping concentration of ~10¹⁷ cm⁻³.

A scanning electron microscopy (SEM) image of a fabricated frequency doubler MMIC with the proposed SMB terminal structure is shown in Figure 2a, and the detail of the diode cell in MMIC is shown in Figure 2b. The fabrication process began with mesa isolation using wet etching. Subsequently, Schottky metal and cathode ohmic contacts were deposited by electron-beam evaporation. The metal-brim structure was then formed by specified dry etching and a planarization process. Then, a 300-nm-thick SiO₂ layer was deposited by plasma-enhanced chemical vapor deposition (PECVD) for the passivation of the SMB structure, and the dielectric layer on the top of the metal contacts was removed by SF₆ plasma dry etching. Next, a 1.5-μm-thick gold layer was electroplated to form the air bridge fingers and passive circuitry, such as waveguide probes and filter structures. Finally, the GaAs substrate was thinned to the desired thickness (34–37 μm) by lapping and polishing.

3. Model Analysis

We first evaluated the reverse and forward bias DC I–V characteristics of the conventional SBD and the proposed SMB devices using an Agilent B1500 semiconductor device parameter analyzer. In addition, the extended 1-D DD model was applied for comparison with the experimental I–V results. Figure 3 presents the measured and simulated reverse and forward bias DC I–V characteristics of the two SBDs on a log scale, which has the best agreement with the measured data at larger forward bias.

However, the most striking features of the results in Figure 3 are that the measured currents of conventional SBD structure at a given bias voltage are much larger than those of SMB structure at reverse and low forward biases and that these observed deviations from ideal curve are greater for the conventional SBD structure than for the SMB terminal structure. This is not conducive to the power conversion of frequency multiplier. In addition, various parameters were extracted from the SBD devices using the Agilent B1500A system, as listed in

Table 1. Parameters of SBDs with different terminal structures.

Parameter	SMB-SBD Structure	Conventional SBD Structure
Ideal factor	1.14	1.17
Diameter (μm)	3	3
Junction potential (V)	0.75	0.74
Zero-biased capacitance (fF) by	11.9	12.1
I–V	-	-
Series resistance (Ω)	15.56	14.00
Reverse saturation current (A)	1.27 × 10−14	1.89 × 10−14

, and were used in the simulations and for analyzing the effects of the different terminal structures on the frequency conversion efficiency of the MMICs.

Generally, the current model was based on thermionic emission theory through the Schottky contact [15]. The common I–V SPICE model used in the harmonic balance simulation is given as follows:

$$\{\begin{array}{c}\mathrm{I}\left(\mathrm{V}\right)={\mathrm{I}}_{\mathrm{S}}\left({\mathrm{e}}^{-10}-1\right)+\mathrm{g}\left(\mathrm{V}+10\times \mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}\right) \mathrm{V}<-10\times \mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}\\ {\mathrm{I}}_{\mathrm{S}}\left({\mathrm{e}}^{\frac{\mathrm{V}}{\mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}}}-1\right) -10\times \mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}<\mathrm{V}<{\mathrm{V}}_{\mathrm{j}}\\ {\mathrm{I}}_{\max }+\left(\mathrm{V}-{\mathrm{V}}_{\mathrm{j}}\right)\times \mathrm{gmax} \mathrm{V}>{\mathrm{V}}_{\mathrm{j}}\end{array},$$

(1)

Here, ${\mathrm{I}}_{\mathrm{S}}$ is the reverse saturated current, $\mathrm{g}=\frac{{\mathrm{I}}_{\mathrm{S}}}{\mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}}\times {\mathrm{e}}^{-10}$, $\mathrm{V}$ is the applied bias, $\mathrm{N}$ is the ideal factor, ${\mathrm{V}}_{\mathrm{t}}$ is the thermal voltage, ${\mathrm{V}}_{\mathrm{j}}$ is the built-in voltage of the Schottky contact, where ${\mathrm{V}}_{\mathrm{j}}=\mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}\times \ln \left(\frac{{\mathrm{I}}_{\max }}{{\mathrm{I}}_{\mathrm{S}}}+1\right)$, and $\mathrm{gmax}=\frac{\left(\mathrm{Imax}+\mathrm{Is}\right)}{\mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}}$ [16,17].

To explain this, we first note that random variations in the epitaxial material and processing technology result in leakage currents, which are not accounted for in Equation (1). This affects the simulation results for both SBD structures equally. Secondly, unlike the proposed SMB terminal structure, the conventional SBD terminal structure introduces an additional direct current path between the anode/depletion layer and the cathode, as illustrated in Figure 4, which is also not accounted for in Equation (1), and directly impacts the accuracy of the simulation results for the conventional SBD structure.

The above analysis indicates that the SBD with the proposed metal-brim terminal structure provides measured I–V curves that conform better with the ideal model given by Equation (1) than the conventional SBD due to its elimination of Schottky fringing electric field effects that lead to an additional current path. These fringing field effects are one reason why the performances of frequency multiplier MMICs tend to deviate from their ideal design results [18]. Therefore, the I–V model of the conventional SBD must be modified to account for the additional direct current path between the anode/depletion layer and the cathode. The modified equation is given as follows:

$$\mathrm{I}\left(\mathrm{V}\right)=\{\begin{array}{c}{\mathrm{e}}^{\frac{\mathrm{N}\times {\mathrm{V}}_j}{{\mathrm{V}}_{\text{t}}}}\times {\mathrm{e}}^{\frac{\mathrm{V}}{N_{tunnel}}} \mathrm{V}<-10\times \mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}\\ {\mathrm{I}}_{\mathrm{S}}\left({\mathrm{e}}^{\frac{\mathrm{V}}{\mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}}}-1\right)+{\mathrm{I}}_{edge} -10\times \mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}<\mathrm{V}<{\mathrm{V}}_{\mathrm{j}}\\ {\mathrm{I}}_{\max }+\left(\mathrm{V}-{\mathrm{V}}_{\mathrm{j}}\right)\times \mathrm{gmax} \mathrm{V}>{\mathrm{V}}_{\mathrm{j}}\end{array},$$

(2)

Here, $N_{tunnel}$ is a tunneling factor that determines the inverse breakdown characteristic of actual diodes and ${\mathrm{I}}_{edge}$ is a nonlinear component indicative of Schottky fringing effects. The ranges given for the first two conditions are different from those given in Equation (1). This could be explained by the tunneling current deviation from thermionic emission theory when the diode works at negative bias.

Figure 5 presents the measured and simulated reverse and forward bias DC I–V characteristics of the conventional SBD on a log scale along with the simulated curves obtained from Equations (1) and (2). We note that the curve of Equation (2) is in good agreement with the measured data under low reverse and forward bias voltages, indicating that the revised model reliably represents the I–V characteristics of the conventional SBD. Note that the method of harmonic balance was used to extract the second component of the nonlinear curve to realize the analysis of the frequency multiplier MMIC.

We also plotted the simulated C–V characteristics of the conventional SBD and the proposed SMB devices based on the ideal 1-D drift-diffusion (DD) model, as shown in Figure 6. Because the broadband doubler operates in the small signal behavior regime of the SBDs, the C–V characteristics of an SBD below the built-in voltage ${\mathrm{V}}_j$ can be given according to the following conventional equation:

$$C=\frac{A{\epsilon }_s}{\sqrt{\frac{2{\epsilon }_s\left(V_j-V\right)}{q{N}_D}}} ,$$

(3)

where $A$ is the area of the anode, ${\epsilon }_S$ is the dielectric constant of the semiconductor, and $N_D$ is the doping concentration of the epilayer. Equation (3) represents ideal conditions and completely neglects any consideration of fringing effects and interface states. We note from Figure 6 that the results of Equation (3) are in very good agreement with the simulated results of the proposed SMB device. This indicates that the SMB structure completely eliminates the depletion layer capacitance fringing effect that is evident in conventional SBDs. The depletion layer capacitance fringing effect of the conventional SBD can be accounted for through a straightforward extension of Equation (3) as follows:

$$C=\frac{A{\epsilon }_s}{\sqrt{\frac{2{\epsilon }_s\left(V_j-V\right)}{q{N}_D}}}+2D_1{\epsilon }_{s}L\frac{\beta \left(V_S\right)}{{\beta }_0} ,$$

(4)

where the parameters $D_1$ (a correction term) and $L$ (the radius of Schottky junction) are determined by the report in [19]. $\frac{\beta \left(V_S\right)}{{\beta }_0}$ includes the fringing factors, and its value is 1 without considering the capacitance effect of the interface states.

Accordingly, the source of operating frequency current can be expressed as follows [20]:

$$\mathrm{I}\left(\mathrm{V}\right)={\mathrm{I}}_{\mathrm{S}}\left({\mathrm{e}}^{\frac{\mathrm{V}}{\mathrm{N}\times {\mathrm{V}}_{\mathrm{t}}}}-1\right)+{\mathrm{I}}_{edge}\left(t\right)+C\left(t\right)\times \frac{dV\left(t\right)}{dt} ,$$

(5)

Equation (5) indicates that the second harmonic was the result of the thermionic emission current and displacement current through the undepleted semiconductor. The symbolically defined device (SDD) component was employed to quickly and accurately analyze the effect of the SMB structure on the frequency multiplier efficiency based on the I–V and C–V characteristics of the SBDs.

4. Results

The cavity of the designed doubler is made of brass material, and the entire surface is gold-plated. The doubler MMIC is fixed and grounded by conductive adhesive, which is shown in Figure 7. The measurement setup employed for testing the frequency doubler MMICs is presented in Figure 8. Here, an Agilent E8257D analog signal generator (Santa Clara, CA, USA) was followed by a W-band, active ×8 frequency multiplier to generate an input signal in the 86–106 GHz band with a typical output power of +10 dBm. The output power of the frequency multipliers was measured using a VDI Erickson PM4 power meter (Charlottesville, VA, USA). In addition, a KFD55 Sub-Miniature version A (SMA) coaxial RF connector (Xi’an, China) was connected to the main transmission circuit using gold wire bonding, and a low forward bias (+0.8 V) was applied to the SMA port to bias the SBDs in the doubler MMICs. Two doublers based on different Schottky barrier diode (SBD) terminal structures were tested.

The measured output power of the frequency doubler MMICs constructed using the conventional SBD and the proposed SMB structures is shown in Figure 9a with respect to frequency along with the input power. The measured output power of the frequency doubler MMIC constructed using the proposed SMB structure is greater than 0.5 mW in the range from 188 to 208 GHz, and the maximum output power is about 0.914 mW at 208 GHz. In contrast, the maximum output power of the frequency doubler MMIC constructed using the conventional SBD structure was only 0.216 mW at 206 GHz.

Naturally, the output power of both frequency doubler MMICs would be greater with a larger input power. Figure 9b shows the measured efficiency of the frequency doubler MMIC based on the proposed SMB terminal structure versus input power operating at 99 GHz. We note that the efficiency monotonically increases with increasing input power. The efficiency would be higher with a sufficient power source.

5. Conclusions

In summary, 180–225 GHz broadband frequency doubler MMICs based on the conventional SBD and a novel SMB-structured SBD were fabricated and systemically evaluated. Compared with the MMIC adopting the conventional SBD terminal structure, the maximum output power of the MMIC adopting the SMB structure increased from 0.216 mW at 206 GHz to 0.914 mW at 208 GHz. Analysis of the nonlinear I–V and C–V characteristics of the two terminal structures based on an extended 1-D DD model demonstrated that the SMB structure provides better conversion efficiency than the conventional SBD structure by eliminating the accumulation of charge and additional current paths at the Schottky electrode edge, which is beneficial for MMIC applications requiring high power and high efficiency. The application of the terminal structure could provide a feasible scheme for the optimization of frequency multiplier performances. Future work will focus on analyzing the response mechanism of different terminal structures to high-frequency signals and increasing the efficiency further by reducing series resistance and the epilayer thickness. In addition, the fabrication process will be optimized and the relatively large parasitic capacitance originating from the electroplating technology employed will be addressed.