Capacitance Characteristics and Breakdown Mechanism of AlGaN / GaN Metal–Semiconductor–Metal Varactors and Their Anti-Surge Application

: The AlGaN / GaN materials with a wide band gap, high electron mobility, and high breakdown voltage are suitable for manufacturing high-power and high-frequency electronic devices. In this study, metal Schottky contact electrodes of di ﬀ erent dimensions are prepared on AlGaN / GaN wafers to fabricate metal–semiconductor–metal (MSM) varactors. Voltage-dependent capacitance and breakdown voltages of the varactors are measured and studied. The corresponding breakdown mechanisms of varactors with di ﬀ erent electrode gaps are proposed. Furthermore, an anti-surge application using GaN-based MSM varactors in a signal transmission module is demonstrated, and its surge suppression capability is shown. We believe that our study will be beneﬁcial in developing surge protection circuits for RF applications.


Introduction
The advent of the Internet of Things has placed difficult frequency and power requirements on electronic devices. GaN has been widely used in high-frequency and high-power applications. It has a wide band gap (3.4 eV), high electron mobility, and high breakdown voltage [1]. Because of these distinct advantages, it is more widely used in manufacturing devices for high-power and high-frequency applications compared to other materials [2] (Table 1). Moreover, advances in quantum confinement technology have led to the development of heterostructures of two-dimensional electron gas (2DEG), which can solve the serious problem of decreased channel electron mobility. Consequently, high-electron-mobility field-effect transistors (HEMT) made of AlGaN/GaN are extensively used in high-frequency electronic components [3][4][5]. Thermal conductivity (W cm −1 K −1 ) 1.5 4.9 0.56 1.5 Electron mobility (cm 2 V −1 s −1 ) 1350 700 8500 2000 Hole mobility (cm 2 V −1 s −1 ) 450 120 330 300 Breakdown voltage (MV cm −1 ) 0.25 3.5 0.4 4 Saturation electron speed (10 7 cm s −1 ) 1 2.1 1.3 2.5 Dielectric constant 11.9 10 12.5 9

Fabrication of the GaN-Based 2DEG MSM Varactor
First, the mask of the MSM varactors was designed. Six different component lengths (2000, 1500, 1000, 500, 250, and 150 μm) were used for the patterns in the mask. Next, for each specific electrode length, six different gap widths (30, 25, 20, 15, 10, and 5 μm) were also designed, as shown in Figure 2a. The wafer used to fabricate devices in this experiment was an AlGaN/GaN heterostructure epitaxial wafer, as shown in Figure 2b. The 2DEG layer carrier concentration obtained by Hall measurement at room temperature was 9.7 × 10 12 cm −2 , and the electron mobility was 1541 cm 2 V −1 s −1 .

Fabrication of the GaN-Based 2DEG MSM Varactor
First, the mask of the MSM varactors was designed. Six different component lengths (2000, 1500, 1000, 500, 250, and 150 µm) were used for the patterns in the mask. Next, for each specific electrode length, six different gap widths (30, 25, 20, 15, 10, and 5 µm) were also designed, as shown in Figure 2a.
The wafer used to fabricate devices in this experiment was an AlGaN/GaN heterostructure epitaxial wafer, as shown in Figure 2b. The 2DEG layer carrier concentration obtained by Hall measurement at room temperature was 9.7 × 10 12 cm −2 , and the electron mobility was 1541 cm 2 V −1 s −1 .

Fabrication of the GaN-Based 2DEG MSM Varactor
First, the mask of the MSM varactors was designed. Six different component lengths (2000, 1500, 1000, 500, 250, and 150 μm) were used for the patterns in the mask. Next, for each specific electrode length, six different gap widths (30, 25, 20, 15, 10, and 5 μm) were also designed, as shown in Figure 2a. The wafer used to fabricate devices in this experiment was an AlGaN/GaN heterostructure epitaxial wafer, as shown in Figure 2b. The 2DEG layer carrier concentration obtained by Hall measurement at room temperature was 9.7 × 10 12 cm −2 , and the electron mobility was 1541 cm 2 V −1 s −1 .  Initially, all of the wafers were cut into a 1 × 1 cm 2 area by laser cutting. They were then subjected to a standard RCA cleaning procedure in an ultrasonic bath with acetone, isopropanol, and deionized water for three minutes each, sequentially. This was followed by spin-coating the S1813 photoresist on top of the wafers. An exposure machine was then used to complete the lithography process. Subsequently, an electron-beam evaporation machine was used to deposit the Ni/Au (20 nm/70 nm) Schottky junction metal layer. Finally, all the wafers were subjected to a lift-off procedure to obtain MSM varactors of different dimensions (Figure 3). In addition to the normal-sized varactors mentioned above, we reduced the sizes of all varactors to one-fifth by the exposure-scaling method. Next, to investigate the best component structure characteristics of the varactors, the capacitance-voltage (C-V), current-voltage (I-V), and breakdown characteristics of all these varactors were investigated.
wafers. An exposure machine was then used to complete the lithography process. Subsequently, an electron-beam evaporation machine was used to deposit the Ni/Au (20 nm/70 nm) Schottky junction metal layer. Finally, all the wafers were subjected to a lift-off procedure to obtain MSM varactors of different dimensions (Figure 3). In addition to the normal-sized varactors mentioned above, we reduced the sizes of all varactors to one-fifth by the exposure-scaling method. Next, to investigate the best component structure characteristics of the varactors, the capacitance-voltage (C-V), current-voltage (I-V), and breakdown characteristics of all these varactors were investigated. The measurements were made using an Agilent E4980A LCR-meter (Agilent Technologies, Santa Clara, California, USA) and a Keithley 2410 SourceMeter (Keithley Instruments, Solon, Ohio, USA) to obtain the curves of electrical characteristics. The voltage measurement range of the C-V characteristic curve was from −15 to 15 V, measured once for every 0.05 V interval, and the frequency of the AC measurement signal was set to 2 MHz, 1 MHz, and 500, 200, 100, 50, and 10 kHz. The voltage measurement range of the I-V characteristic curve was done from −100 to 500 V, and the maximum current was limited to 21 mA because of equipment constraints.

Surge-Protection Circuit Design and Measurement
After completion of the experiment on the variabilities for the GaN-based MSM varactor and breakdown voltage, we used the design with an electrode length of 2000 μm and a gap width of 30 μm to complete the varactors. The wafer was then cut and bare-died by laser for use in the signal transmission and anti-surge protection circuit, as shown in Figure 4a. For the anti-surge module, we designed the 50 Ω microstrip line on the glass-reinforced epoxy laminate material (FR4) to be the signal transmission path. The GaN-based MSM varactor was connected in series between the microstrip line by flip-chip method with silver glue, and the low-parasitic-capacitance GDT was shunted in front of the varactor. The overall anti-surge module was completed as shown in Figure 4b. Surge current pulse injection measurements were performed, as shown in Figure 4c. Finally, the robustness of the anti-surge module was checked again using a network analyzer. The measurements were made using an Agilent E4980A LCR-meter (Agilent Technologies, Santa Clara, CA, USA) and a Keithley 2410 SourceMeter (Keithley Instruments, Solon, OH, USA) to obtain the curves of electrical characteristics. The voltage measurement range of the C-V characteristic curve was from −15 to 15 V, measured once for every 0.05 V interval, and the frequency of the AC measurement signal was set to 2 MHz, 1 MHz, and 500, 200, 100, 50, and 10 kHz. The voltage measurement range of the I-V characteristic curve was done from −100 to 500 V, and the maximum current was limited to 21 mA because of equipment constraints.

Surge-Protection Circuit Design and Measurement
After completion of the experiment on the variabilities for the GaN-based MSM varactor and breakdown voltage, we used the design with an electrode length of 2000 µm and a gap width of 30 µm to complete the varactors. The wafer was then cut and bare-died by laser for use in the signal transmission and anti-surge protection circuit, as shown in Figure 4a. For the anti-surge module, we designed the 50 Ω microstrip line on the glass-reinforced epoxy laminate material (FR4) to be the signal transmission path. The GaN-based MSM varactor was connected in series between the microstrip line by flip-chip method with silver glue, and the low-parasitic-capacitance GDT was shunted in front of the varactor. The overall anti-surge module was completed as shown in Figure 4b. Surge current pulse injection measurements were performed, as shown in Figure 4c. Finally, the robustness of the anti-surge module was checked again using a network analyzer.

Results and Discussion
The three-dimensional structure of the MSM varactor is shown in Figure 5. The MSM varactor can be regarded as a double-gate (G) structure with two back-to-back Schottky junction diodes. When the bias voltage is zero or less than the threshold value, we can obtain a large capacitance value when measuring the C-V characteristic curve. However, when the bias voltage exceeds the threshold voltage, the capacitance of the varactor suddenly drops to an extremely small value. This is in accordance with the defining characteristic of varactors: their capacitance changes with the applied voltage [19].

Results and Discussion
The three-dimensional structure of the MSM varactor is shown in Figure 5. The MSM varactor can be regarded as a double-gate (G) structure with two back-to-back Schottky junction diodes. When the bias voltage is zero or less than the threshold value, we can obtain a large capacitance value when measuring the C-V characteristic curve. However, when the bias voltage exceeds the threshold voltage, the capacitance of the varactor suddenly drops to an extremely small value. This is in accordance with the defining characteristic of varactors: their capacitance changes with the applied voltage [19].

Results and Discussion
The three-dimensional structure of the MSM varactor is shown in Figure 5. The MSM varactor can be regarded as a double-gate (G) structure with two back-to-back Schottky junction diodes. When the bias voltage is zero or less than the threshold value, we can obtain a large capacitance value when measuring the C-V characteristic curve. However, when the bias voltage exceeds the threshold voltage, the capacitance of the varactor suddenly drops to an extremely small value. This is in accordance with the defining characteristic of varactors: their capacitance changes with the applied voltage [19].

Capacitor Characteristics and Breakdown Voltages of the Normal-Sized MSM Varactors
In this study, six electrode sizes for the MSM varactors were considered. The calculated areas, from large to small ( Figure 3A-F), were 375,000, 325,000, 275,000, 225,000, 170,000, and 160,000 μm 2 . The electrode area, low-voltage capacitance, high-voltage capacitance, and the capacitance conversion ratio are summarized in Table 2.

Capacitor Characteristics and Breakdown Voltages of the Normal-Sized MSM Varactors
In this study, six electrode sizes for the MSM varactors were considered. The calculated areas, from large to small ( Figure 3A-F), were 375,000, 325,000, 275,000, 225,000, 170,000, and 160,000 µm 2 . The electrode area, low-voltage capacitance, high-voltage capacitance, and the capacitance conversion ratio are summarized in Table 2. From the experimental results, it can be seen that the larger the electrode area was, the higher the capacitance was. In addition, the component length of 2000 µm gave the highest capacitance conversion ratio. We further measured the C-V and I-V characteristics of this design for different frequencies and gap widths. All of the results are given in Figures 6 and 7. The obtained maximum capacitance value of the varactor was 283 pF, and the maximum breakdown voltage was 390 V. In contrast to conventional metal-oxide-semiconductor field-effect transistor (MOSFET) devices, the gate has a breakdown voltage of only approximately 30 V. As a component, the breakdown voltage of varactors in this study was much higher. capacitance was. In addition, the component length of 2000 μm gave the highest capacitance conversion ratio. We further measured the C-V and I-V characteristics of this design for different frequencies and gap widths. All of the results are given in Figures 6 and 7. The obtained maximum capacitance value of the varactor was 283 pF, and the maximum breakdown voltage was 390 V. In contrast to conventional metaloxide-semiconductor field-effect transistor (MOSFET) devices, the gate has a breakdown voltage of only approximately 30 V. As a component, the breakdown voltage of varactors in this study was much higher.   Figure 6).

Capacitor Characteristics and Breakdown Voltages of the Reduced-Size MSM Varactors
With the same manufacturing parameters, we further reduced the all dimensions of the normal-sized varactors by a factor of five. The dimensions of the reduced-size varactors are shown in Table 3. Since the varactor with a length of 2000 μm and a gap width of 30 μm can achieve the maximum capacitance and best capacitance conversion ratio, we chose the varactor that was shrunk from this design (reducing the length to 400 μm and the width to 6 μm) as a reduced-size sample for measuring the capacitance and breakdown voltage. The maximum capacitance of this varactor is 10.2 pF, the minimum is 0.589 pF, and the capacitance conversion ratio (C C ⁄ ) is 17.3. The measurement data, given in Figure 8, show that the maximum capacitance of the reduced varactor decreases by a factor of 30 as compared with the normal size. This also explains why, after reducing the electrode to one-fifth of its size, the overall electrode area decreases from 375,000 to 15,000 μm 2 . According to the capacitance formula, C = εA/d, the values of ε and d are constant, but as the electrode area A becomes smaller, the corresponding overall capacitance value C decreases. In addition, compared with the normal-sized varactors, the capacitance of the reduced-size varactor becomes more stable as the frequency of the measuring signal changes (as shown in Figure 8). This phenomenon is worth further investigation and discussion.   Figure 6).

Capacitor Characteristics and Breakdown Voltages of the Reduced-Size MSM Varactors
With the same manufacturing parameters, we further reduced the all dimensions of the normal-sized varactors by a factor of five. The dimensions of the reduced-size varactors are shown in Table 3. Since the varactor with a length of 2000 µm and a gap width of 30 µm can achieve the maximum capacitance and best capacitance conversion ratio, we chose the varactor that was shrunk from this design (reducing the length to 400 µm and the width to 6 µm) as a reduced-size sample for measuring the capacitance and breakdown voltage. The maximum capacitance of this varactor is 10.2 pF, the minimum is 0.589 pF, and the capacitance conversion ratio (C max /C min ) is 17.3. The measurement data, given in Figure 8, show that the maximum capacitance of the reduced varactor decreases by a factor of 30 as compared with the normal size. This also explains why, after reducing the electrode to one-fifth of its size, the overall electrode area decreases from 375,000 to 15,000 µm 2 . According to the capacitance formula, C = εA/d, the values of ε and d are constant, but as the electrode area A becomes smaller, the corresponding overall capacitance value C decreases. In addition, compared with the normal-sized varactors, the capacitance of the reduced-size varactor becomes more stable as the frequency of the measuring signal changes (as shown in Figure 8). This phenomenon is worth further investigation and discussion.  For the I-V characteristic curve measurement results, the breakdown voltages of the reduced-size varactors are shown in Figure 9. Among these, the varactor with a gap width of 6 μm achieves the highest breakdown voltage of 321 V.

Breakdown Mechanism for MSM Varactors with Different Gap Widths
The measurement results obtained from Figures 7 and 9 show that the width of the electrode gap of the varactor greatly affects the breakdown voltage. Overall, the larger the gap width was, the higher the breakdown voltage was.
In order to further analyze the relationship between the electrode gap width and the breakdown voltage, we measured the I-V characteristic curve of the normal-sized varactor (length of 500 μm and gap widths of 30, 25, 20, 15, 10, and 5 μm), as shown in Figure 10. Next, combining these measurement results with the reduced-size varactor (length of 400 μm and widths of 1, 2, 3, 4, 5, and 6 μm; shown in Figure 9), we can obtain the breakdown voltage relationship between the electrode gap width from 1 to 30 μm based on the similar length value, as shown in Figure 11. For the I-V characteristic curve measurement results, the breakdown voltages of the reduced-size varactors are shown in Figure 9. Among these, the varactor with a gap width of 6 µm achieves the highest breakdown voltage of 321 V. For the I-V characteristic curve measurement results, the breakdown voltages of the reduced-size varactors are shown in Figure 9. Among these, the varactor with a gap width of 6 μm achieves the highest breakdown voltage of 321 V.

Breakdown Mechanism for MSM Varactors with Different Gap Widths
The measurement results obtained from Figures 7 and 9 show that the width of the electrode gap of the varactor greatly affects the breakdown voltage. Overall, the larger the gap width was, the higher the breakdown voltage was.
In order to further analyze the relationship between the electrode gap width and the breakdown voltage, we measured the I-V characteristic curve of the normal-sized varactor (length of 500 μm and gap widths of 30, 25, 20, 15, 10, and 5 μm), as shown in Figure 10. Next, combining these measurement results with the reduced-size varactor (length of 400 μm and widths of 1, 2, 3, 4, 5, and 6 μm; shown in Figure 9), we can obtain the breakdown voltage relationship between the electrode gap width from 1 to 30 μm based on the similar length value, as shown in Figure 11.

Breakdown Mechanism for MSM Varactors with Different Gap Widths
The measurement results obtained from Figures 7 and 9 show that the width of the electrode gap of the varactor greatly affects the breakdown voltage. Overall, the larger the gap width was, the higher the breakdown voltage was.
In order to further analyze the relationship between the electrode gap width and the breakdown voltage, we measured the I-V characteristic curve of the normal-sized varactor (length of 500 µm and Crystals 2020, 10, 292 8 of 12 gap widths of 30, 25, 20, 15, 10, and 5 µm), as shown in Figure 10. Next, combining these measurement results with the reduced-size varactor (length of 400 µm and widths of 1, 2, 3, 4, 5, and 6 µm; shown in Figure 9), we can obtain the breakdown voltage relationship between the electrode gap width from 1 to 30 µm based on the similar length value, as shown in Figure 11.
We can see from this measurement result that, for an electrode gap width less than 10 µm, the slope of the breakdown voltage curve of the varactor is positive. This means that as the electrode gap width becomes larger, the breakdown voltage of the varactor increases. Under this small gap width condition, the bias voltage causes an extremely high electric field within the anode and cathode. This high electric field causes the surface material, the GaN cap layer, to break down and provide a path for the leakage current (as shown in Figure 11, path A). Conversely, because the electrode gap width is large (greater than 10 µm), the slope of the breakdown voltage curve of the varactor is nearly zero. This means that the electrode gap width will no longer be the main condition influencing the breakdown voltage. Although the electrode gap width becomes larger, the breakdown voltage remains at approximately 300 V. At this stage, the GaN cap layer on the surface can sustain a smaller electric field without breakdown, and so the breakdown path exists under the epitaxial wafer (as shown in Figure 11, path B). Consequently, the epitaxial thin-film quality of the GaN wafer becomes the most important condition for increasing the breakdown voltage. We can see from this measurement result that, for an electrode gap width less than 10 μm, the slope of the breakdown voltage curve of the varactor is positive. This means that as the electrode gap width becomes larger, the breakdown voltage of the varactor increases. Under this small gap width condition, the bias voltage causes an extremely high electric field within the anode and cathode. This high electric field causes the surface material, the GaN cap layer, to break down and provide a path for the leakage current (as shown in Figure 11, path A). Conversely, because the electrode gap width is large (greater than 10 μm), the slope of the breakdown voltage curve of the varactor is nearly zero. This means that the electrode gap width will no longer be the main condition influencing the breakdown voltage. Although the electrode gap width becomes larger, the breakdown voltage remains at approximately 300 V. At this stage, the GaN cap layer on the surface can sustain a smaller electric field without breakdown, and so the breakdown path exists under the epitaxial wafer (as shown in Figure 11, path B). Consequently, the epitaxial thin-film quality of the GaN wafer becomes the most important condition for increasing the breakdown voltage.   We can see from this measurement result that, for an electrode gap width less than 10 μm, the slope of the breakdown voltage curve of the varactor is positive. This means that as the electrode gap width becomes larger, the breakdown voltage of the varactor increases. Under this small gap width condition, the bias voltage causes an extremely high electric field within the anode and cathode. This high electric field causes the surface material, the GaN cap layer, to break down and provide a path for the leakage current (as shown in Figure 11, path A). Conversely, because the electrode gap width is large (greater than 10 μm), the slope of the breakdown voltage curve of the varactor is nearly zero. This means that the electrode gap width will no longer be the main condition influencing the breakdown voltage. Although the electrode gap width becomes larger, the breakdown voltage remains at approximately 300 V. At this stage, the GaN cap layer on the surface can sustain a smaller electric field without breakdown, and so the breakdown path exists under the epitaxial wafer (as shown in Figure 11, path B). Consequently, the epitaxial thin-film quality of the GaN wafer becomes the most important condition for increasing the breakdown voltage. Figure 10. I-V measurements of the MSM varactors with the same length (500 μm) and different widths (the same as those in Figure 6).

The Anti-Surge Module Application
Since the capacitance of MSM varactors changes with the applied bias voltage, when an MSM varactor is connected in series in the signal transmission path, the voltage of the signal transmission range is low under normal working conditions. Therefore, the high capacitance of the varactor offers low impedance to the signal. However, when facing strong high-voltage pulse injection, the capacitance of the MSM varactor rapidly decreases because of the high bias voltage. Therefore, the surge signal cannot couple through the transmission path, thereby protecting the back-end circuit module. This prevents ESD and MEMP surges from directly affecting the circuit and causing permanent damage. In this way, damage to the back-end components can be suppressed or blocked.
The C-V measurement results for a varactor with an electrode length of 2000 µm and a gap width of 30 µm are shown in the Figure 12a. The frequency of the measurement signal varies from 1 kHz to 2 MHz. The maximum capacitance (C max ) of the varactor is 455 pF, the minimum capacitance (C min ) is 5.08 pF, and the capacitance conversion ratio (C ratio ) can be up to 89.5. In addition, as described in the literature, it can be observed that the capacitance of the varactor is dependent on the measurement signal frequency. As the frequency decreases, the C-V curve assumes a "batman-like" curve. The results of network analyzer measurement of the overall anti-surge module are shown in Figure 12b. The insertion loss of the module is approximately −2 dB within the GPS devices' operating band (1.2 to 1.6 GHz). This proves that under normal operation, the capacitance is large enough, and therefore the signal transmission is still very efficient.

The Anti-Surge Module Application
Since the capacitance of MSM varactors changes with the applied bias voltage, when an MSM varactor is connected in series in the signal transmission path, the voltage of the signal transmission range is low under normal working conditions. Therefore, the high capacitance of the varactor offers low impedance to the signal. However, when facing strong high-voltage pulse injection, the capacitance of the MSM varactor rapidly decreases because of the high bias voltage. Therefore, the surge signal cannot couple through the transmission path, thereby protecting the back-end circuit module. This prevents ESD and MEMP surges from directly affecting the circuit and causing permanent damage. In this way, damage to the back-end components can be suppressed or blocked.
The C-V measurement results for a varactor with an electrode length of 2000 μm and a gap width of 30 μm are shown in the Figure 12a. The frequency of the measurement signal varies from 1 kHz to 2 MHz. The maximum capacitance (Cmax) of the varactor is 455 pF, the minimum capacitance (Cmin) is 5.08 pF, and the capacitance conversion ratio (Cratio) can be up to 89.5. In addition, as described in the literature, it can be observed that the capacitance of the varactor is dependent on the measurement signal frequency. As the frequency decreases, the C-V curve assumes a "batman-like" curve. The results of network analyzer measurement of the overall anti-surge module are shown in Figure 12b. The insertion loss of the module is approximately −2 dB within the GPS devices' operating band (1.2 to 1.6 GHz). This proves that under normal operation, the capacitance is large enough, and therefore the signal transmission is still very efficient. The anti-surge module application was used next. The standard injection current pulses are shown in Figure 13a. According to the testing requirements, the current specifications can be performed from 600 A to 2.5 kA. As per the MIL-STD-188-125-2, with a 50 Ω dummy load resistor and a 600 A injected current pulse, all of the residual current value can be suppressed to less than 5 A. The best result is 3.84 A, as shown in Figure 13b. Furthermore, the network analyzer measurement is performed again to check whether or not the anti-surge module has been damaged. All of the S-parameters show that after 600 A current pulse injection, the module functionality was still significantly normal, as shown in Figure 14. In addition, all of the test results show that this anti-surge module can effectively suppress the surge energy as the injection current value keeps increasing up to 2.54 kA. The residual current values are not only less than 10 A, as shown in Figure 15a, but the S-parameters show no significant variation, as shown in Figure  15b. Hence, the GaN-based MSM varactor can be used in signal transmission paths; when the surge pulse attack occurs, the MSM varactor can block the surge energy in time because of its excellent capability to The anti-surge module application was used next. The standard injection current pulses are shown in Figure 13a. According to the testing requirements, the current specifications can be performed from 600 A to 2.5 kA. As per the MIL-STD-188-125-2, with a 50 Ω dummy load resistor and a 600 A injected current pulse, all of the residual current value can be suppressed to less than 5 A. The best result is 3.84 A, as shown in Figure 13b. Furthermore, the network analyzer measurement is performed again to check whether or not the anti-surge module has been damaged. All of the S-parameters show that after 600 A current pulse injection, the module functionality was still significantly normal, as shown in Figure 14. In addition, all of the test results show that this anti-surge module can effectively suppress the surge energy as the injection current value keeps increasing up to 2.54 kA. The residual current values are not only less than 10 A, as shown in Figure 15a, but the S-parameters show no significant variation, as shown in Figure 15b. Hence, the GaN-based MSM varactor can be used in signal transmission paths; when the surge pulse attack occurs, the MSM varactor can block the surge energy in time because of its excellent capability to withstand voltage and because of its fast response abilities; hence, the GDT has sufficient time to start up and shunt, and the bulk of the surge energy can be discharged. withstand voltage and because of its fast response abilities; hence, the GDT has sufficient time to start up and shunt, and the bulk of the surge energy can be discharged.  withstand voltage and because of its fast response abilities; hence, the GDT has sufficient time to start up and shunt, and the bulk of the surge energy can be discharged.

Conclusions
We have fabricated GaN/AlGaN MSM varactors as surge protection components in this study for their suitable material properties. Accordingly, different electrode patterns of MSM varactors were designed, and the effects on their C-V and I-V characteristics were investigated.
Experimental results show that the maximum capacitance was 283 pF, the capacitance conversion ratio (Cratio) was 71, and the breakdown voltage was as high as 390 V for an electrode length of 2000 μm and an electrode gap width of 30 μm. We found that when the electrode gap width was less than 10 μm, surface breakdown phenomena occurred or dominated and the correspondent breakdown voltage was proportional to the electrode gap width. Conversely, under path breakdown occurred as the electrode gap width exceeded 10 μm. At this condition, the epitaxial layers' quality and thickness of the GaN wafer became important conditions for increasing the breakdown voltage of varactors.
The anti-surge module application with the GaN-based MSM varactor placed in series with the signal transmission path (the arrangement exhibited a low insertion loss of −2 dB) suppressed a 2.54 kA injection surge current to a value less than 10 A. This meets the requirements of MIL-STD-188-125-2 and can effectively protect the back-end RF circuit from damage.

Conclusions
We have fabricated GaN/AlGaN MSM varactors as surge protection components in this study for their suitable material properties. Accordingly, different electrode patterns of MSM varactors were designed, and the effects on their C-V and I-V characteristics were investigated.
Experimental results show that the maximum capacitance was 283 pF, the capacitance conversion ratio (C ratio ) was 71, and the breakdown voltage was as high as 390 V for an electrode length of 2000 µm and an electrode gap width of 30 µm. We found that when the electrode gap width was less than 10 µm, surface breakdown phenomena occurred or dominated and the correspondent breakdown voltage was proportional to the electrode gap width. Conversely, under path breakdown occurred as the electrode gap width exceeded 10 µm. At this condition, the epitaxial layers' quality and thickness of the GaN wafer became important conditions for increasing the breakdown voltage of varactors.
The anti-surge module application with the GaN-based MSM varactor placed in series with the signal transmission path (the arrangement exhibited a low insertion loss of −2 dB) suppressed a 2.54 kA injection surge current to a value less than 10 A. This meets the requirements of MIL-STD-188-125-2 and can effectively protect the back-end RF circuit from damage.