Evaluation of Pulsed Spark Discharge for Triggering GaAs Photoconductive Semiconductor Switches

Abstract

In this study, a 3 mm gap GaAs photoconductive semiconductor switch (GaAs PCSS) was triggered by pulsed spark discharge. The typical linear mode of GaAs PCSS was fulfilled at a low bias voltage. The on-state current waveform was similar to that of an optical pulse. The “nonlinear mode” was demonstrated at a bias voltage of 4 kV; concurrently, the peak current and the carriers’ multiplication rate were 33 A and 179, respectively. This study indicates that pulsed spark discharge is a promising candidate light source for the direct triggering of GaAs PCSSs.

1. Introduction

Gallium arsenide photoconductive semiconductor switches (GaAs PCSS) are widely utilized in a variety of high-power applications, including ultrafast photoelectric control, high-power technology, and the generation and detection of terahertz [1,2,3,4]. As an optically triggered semiconductor switch, GaAs PCSSs have better advantages over electrically activated switches in terms of electrical insulation and electromagnetic immunity. Customarily, pulsed desktop lasers are the primary optical source for triggering GaAs PCSS, which satisfies requirements for a coherent, homochromous, and ultra-high peak power density optical pulse. In contrast, the disadvantages of solid-state lasers, including their considerable size, immobility, and high price, severely restrict the application of GaAs PCSSs. For the past few years, pulsed semiconductor laser diodes (LDs) have been put to use as compact light sources instead of solid-state lasers for triggering GaAs PCSSs, which has led to the possibility of highly compact switching systems at a low cost (i.e., USD 500) [5,6,7]. However, LDs have a relatively lower energy in the μJ range in comparison with that of desktop lasers in the mJ range. LDs with a pulse energy as low as 24.3 nJ can successfully trigger GaAs PCSSs to operate in nonlinear mode, and the bias electric field can reach up to 78 kV/cm [8]. Using LDs with pulse energy in the nJ range, the bias electric field is significantly greater than that of desktop laser excitation. It is a challenge for the insulated fabrication and protection of GaAs PCSSs under the higher operating voltages in comparison with the excitation of a desktop laser system. Therefore, a compact, lightweight, and low-cost high-energy pulsed optical source is essential to promote the application of GaAs PCSSs. Recently, pulsed ultraviolet light-emitting diodes (LEDs) have been used to trigger wide-bandgap 4H-SiC PCSSs [9,10,11]. A nanosecond pulse can generated using a red LED bulb for triggering GaAs PCSSs [12]. However, due to the lower working power of commercially available LEDs, it is not successful in operating PCSSs in quintessential linear modes. Over the past couple of decades, there has been a rapid increase in the research of atmospheric pressure gas discharges. Compared with low-pressure or vacuum gas discharge systems, atmospheric-pressure gas discharge systems have a wide range of applications in scientific, industrial, and medical areas, considering their virtues such as lower costs, portability, and simplicity of operation [13,14,15].

In this paper, a pulsed spark discharge emitted light in the air is proposed to trigger a GaAs PCSS. The behavior of pulsed spark discharge radiation for triggering the GaAs PCSS is evaluated. Compared with LDs or LEDs, the energy of spark-discharge-emitted light is robust, in the mJ range. Meanwhile, it also satisfies the requirements of miniaturization and modularization.

2. Experiment Setup

The experimental facilities mainly consisted of a GaAs PCSS and a pulsed high-voltage generator module. The material of the test PCSS was semi-insulating (SI) GaAs with a dark resistivity of 5 × 10⁷ Ω·cm⁻³ and higher electron mobility of 8000 cm²/V·s, as shown in Figure 1a. Two 6 mm × 3 mm ohmic contact electrodes comprising Au/Ge/Ni alloy were deposited on the surface of the GaAs substrate, forming a 3 mm insulation gap. The GaAs PCSS was coated with a 900 nm Si₃N₄ insulation protection layer. The PCSS was placed on a copper board with planar transmission lines and was linked to an external circuit by coaxial transmission lines.

The pulsed high-voltage generator module (USD 2, 2022) is shown in Figure 1b. It was fabricated on the basis of a DC power boost module and operated under overvoltage self-breakdown circumstances. Analogous modules have been used for other gas discharge studies [16,17]. In this experiment, 3–6 V DC voltage was boosted to 80 kV by the high-voltage transformer. The needle-shaped stainless-steel electrodes connected to the high-voltage generator were placed 5 mm above the GaAs PCSS. The distance of gas separation between the needle tip could be adjusted in the range of 10 to 15 mm. The optical energy of spark discharge was measured by a pulsed light energy meter (KSDP2210-CAS-1). The spark discharge emission spectrum was recorded with a fiber optic spectrometer (Ocean Optics HR4000). The light emitted from the spark discharge was channeled to the fiber optic spectrometer. The scanning range of the fiber optic spectrometer was from 200 nm to 1100 nm. The time waveform of the spark-discharge-emitted light was recorded by a PIN.

In the test circuit, a GaAs PCSS with a 3 mm gap was charged by a 0.1 nF ceramic capacitor connected to a 15 ΜΩ resistor, as shown in Figure 2. It supplied sufficient power to the GaAs PCSS discharge circuit. The output electrical pulse of the GaAs PCSS was attenuated by a 20 dB attenuator and recorded with the high bandwidth digital storage oscilloscope. The whole device operated in the air at room temperature.

3. Results and Discussion

3.1. Spark Discharge Characterization and Evaluation

Figure 3 represents the total emitted light energy for various air gap distances. According to Figure 3, the general trend was that the emitted light energy increased for longer air gap lengths. The electrode distance was chosen to be 14 mm in a follow-up experiment. Optical emission spectroscopy is a valuable diagnostic tool for determining the collisional process in a plasma environment. The spark discharge emission spectrum was mainly located at 350–900 nm, with three prominent peaks (464 nm, 501 nm, and 568 nm), as shown in Figure 4. Furthermore, the peak at 501 nm was significantly higher than the other wavelengths. The waveform of the spark discharge optical pulse is shown in Figure 5. The pulse width of the spark discharge was approximately 130 ns, and the optical excitation energy was approximately 4 mJ. The rising time of the optical pulse was roughly equal to 47 ns.

When the air gap between the two electrodes was completely broken down, there was a linear spectrum of atomic emissions, a band spectrum of molecular emissions, and a continuous spectrum of hot electrode tip emissions. N₂ spectra are the predominant captured signals for the atmospheric air discharge. In this study, several lines from NⅡ (464.7 nm), NⅡ (501.64 nm), and NⅡ (568.62 nm) were identified from the air discharge spectrum. The spectra of oxygen atoms $O_2(b^1{\mathsf{\Sigma }}_g^{+}\to X^1{\mathsf{\Sigma }}_g^{-})$ with relatively low peaks were also observed at 777 nm. Meanwhile, the spectra of Fe atoms were also captured in the ultraviolet band. These results suggest that initial light emissions are dominated by primary nitrogen discharge; in contrast, the latter contributions may result from the excitation of iron vapor in the spark gap as a result of electrode erosion.

The Streamer theory is recommendable to interpret the gas breakdown processes in spark discharge. In any breakdown mechanism, an electron avalanche is the primary and inescapable element. The free electrons, accelerated by the electric field, obtain sufficient energy to produce more free electrons and positive ions by collision ionization with gas molecules. The electron number of avalanche increases exponentially, in accordance with the electron multiplication law,

$$n=n_0{e}^{{\displaystyle \underset{0}{\overset{x}{\int }}\alpha dx}}$$

(1)

where α is the first Townsend coefficient, n₀ is the initial number of electrons, and x is the gap distance. The Streamer theory is based on gas photoionization and the secondary electron avalanche. When the primary electron avalanche reaches a certain size, more free electrons are produced by gas photoionization around the head of the primary electron avalanche. The free electrons further create numerous secondary avalanches in a strong electric field. Electrons of the secondary avalanche are pulled into the positively charged trail left by the primary avalanche, and a weak, ionized channel is formed. The streamer, exhibiting a certain conducting, can ultimately lead to a spark discharge in the air gap. Spark discharge is a rapid transient process, not a steady one. The inter-electrode gap is pierced by a “fast as lightning” branching-out light channel, which immediately ceases.

3.2. Triggering the GaAs PCSS

The relationship between the output current of the GaAs PCSS and the bias voltage is shown in Figure 6; the storage capacitance was 0.1 nF. The waveforms of the output current were similar at low bias voltage (0.3 kV to 0.9 kV); moreover, the output current increased with the increase in the bias voltage. The rising time of the output current was approximately 43 ns, which was comparable to the rising time of the optical pulse: approximately 47 ns. The output current rose during the optical pulse and then dropped after the optical pulse disappeared. It can be estimated that the PCSS operated in linear mode at this moment. In linear mode, each incident photon generated one electron–hole pair, and the waveform of the on-state current was analogous to that of the optical pulse. The rising time of the output current was primarily determined by the rise time of the optical pulse. In the linear mode, the impact ionization of the carriers could be ignored; the energy of carriers increased with the increasing bias voltage; obviously, so did the on-state current. In the test circuit, the load resistance of the oscilloscope was 50 Ω, and the on-state current ranged between tens and hundreds of milliamps. The fall time of the current was approximately 1.5 µs, which is relatively a long time. The fall time is primarily determined by the lifetime of photogenerated carriers.

The energy of the incident photon was significantly greater than the bandgap of GaAs material over the broad band spectrum (hγ > Eg). The valance band electrons were excited by the conduction band to form electron–hole pairs by absorbing photon energy. The effective temperature of the carriers was higher than the lattice temperature. These photo-excited electrons, called “hot electrons”, were located in the higher energy levels of the conduction band. The excess energy was principally lost by electron–phonon scattering—ionized impurity scattering during the thermalization process in some cases—and finally, recombination of the carriers by direct radiation. The lattice scattering time via the electron–phonon interactions was very short (~10⁻¹² s) compared with the direct radiation recombination time. The lifetime of the carriers τ could be calculated using the radiative recombination time [18],

$$\tau =\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$rn$}\right.$$

(2)

where r represents the recombination coefficient of direct radiation and n is approximately equal to the donor impurity concentration. In general, the coefficient r is in the range of $1.2\sim 7.2\times {10}^{-10}\mathrm{c}{\mathrm{m}}^3{\mathrm{s}}^{-1}$, and the n is $\sim {10}^{15}\mathrm{c}{\mathrm{m}}^{-3}$. According to Equation (2), the carriers’ lifetime τ ranged from 1.4 µs to 8.3 µs; the larger range of carriers’ lifetime was because the recombination coefficient is not a constant, and the experimental results basically coincide with the theoretical calculation.

The energy storage capacitor was changed in order to make the GaAs PCSS operate smoothly in a nonlinear mode in the new test circuit. Figure 7 shows the GaAs PCSS being charged by a 4 nF ceramic storage capacitor. The on-state current was attenuated by a 60 dB attenuator and recorded with a digital storage oscilloscope.

The GaAs PCSS operated in typical linear mode at a low bias voltage of 200 V, as shown in Figure 8a. As stated previously, the PCSS was activated by absorbing a single photon in order to generate an electron–hole pair at a low bias voltage in linear mode. When the bias electric field and the energy of the optical pulse exceeded the nonlinear thresholds, the switch operated in nonlinear mode (also called the “high-gain mode”). Compared with the linear mode, the nonlinear mode could produce as many as 10³~10⁵ electron–hole pairs by absorbing a single photon because of the carrier avalanche multiplication in the bulk of the switch. When the bias voltage increased to 4 kV, the corresponding on-state current waveform appeared to be that of “nonlinear mode”, as shown in Figure 8b. The output peak current was 33 amperes with a rise time of 17 ns, which is much greater than the linear current at a bias voltage of 200 V. The linear and nonlinear mode could not be distinguished merely by the waveform in the strict sense. In order to estimate whether the PCSS operated in the nonlinear mode, the multiplication rate, M, for the nonlinear mode was defined to demonstrate the magnitude of avalanches and multiplication in the photoconduction process [8].

$$M=N_{non}/N_{lin}$$

(3)

where N_lin is the number of photons absorbed by the GaAs PCSS and N_non is the number of photogenerated carriers. In effect, the area under the on-state current waveform is equal to the total amount of charge flowing through the circuit, and it can be acquired using an oscilloscope. The numerical values of the carriers’ multiplication rates were determined to be 1.53 and 274 for bias voltages of 200 V and 4.0 kV, respectively. The value of M was 179. In other words, 179 electron–hole pairs were generated by absorbing a single incident photon. Although this value is smaller than the consequence of the nonlinear mode found in the literature, it still illustrates that the avalanche and multiplication effects occurred in the bulk of the GaAs material, which indicates that the GaAs PCSS operated in nonlinear mode. The lower energy density of the optical pulse may be the possible reason for this experiment result. Theoretically, the multiplication rate can be increased by further increasing the bias voltage; however, the instability of the PCSS and the system complicacy can be increased at a higher-bias electric field, which also results in a breakdown of the PCSS. One can see that the current dropped steeply when the PCSS operated at a bias voltage of 4 kV. The physical mechanism of this phenomenon is mainly relevant to the quenched mode of the photo-activated charge domain (PACD) [19]. If the energy in the energy storage capacitor is unable to maintain the electric field across the GaAs PCSS for the avalanche and multiplication processes of the photo-activated carriers, the GaAs PCSS will operate in “quenched nonlinear mode”, and quickly turn off.

Spark-discharge-emitted light is polychromatic and incoherent; nonetheless, the GaAs PCSS was sensitive to the optical pulse of the spark discharge. Despite a more inexpensive and straightforward system offered by the spark discharge, the maximum on-state current used was only 33 amperes in this experiment. The insulation protection of the switch and the enhancement of optical energy density are areas for further research and testing.

4. Conclusions

In this study, the behavior of the pulsed spark discharge for triggering a GaAs PCSS was investigated. The experimental results show that the spark-discharge-emitted light was polychromatic, with the most conspicuous peak emission occurring at 501 nm. The spectral characteristic and the breakdown process of the spark discharge were analyzed. The optical pulse exhibited a width of 130 ns and a rising time of 47 ns at the optical excitation energy of 4 mJ. Two different operating modes of GaAs PCSS were obtained. The GaAs PCSS was successfully triggered into a typical linear mode at low bias voltages (0.3 kV to 0.9 kV). When the bias voltage increased to 4 kV, the PCSS operated in a nonlinear mode, with a carriers’ multiplication rate of 179. Our research shows that spark-discharge-based optical pulses are an attractive, portable, and compact optical source for triggering GaAs PCSSs.