1. Introduction
Compact, affordable, and efficient sources of high-power laser pulses with sub-nanosecond pulse widths are currently of growing interest. Specifically, such sources can be used as seed lasers for fiber or solid-state amplifiers, for precision ToF (Time-of-Flight) rangefinders [
1], for unmanned Aerial Vehicle (UAV) sensing and mapping application [
2], mm precision laser radar [
3], laser ranging [
4] and in various fields of metrology [
5]. One technique for the generation of high-power sub-nanosecond laser pulses is the use of semiconductor lasers operating in gain-switching mode.
A number of theoretical, zero-dimensional challenge [
6], one-dimensional challenge [
7], and experimental studies [
8,
9] on the development of semiconductor lasers for the generation of high-power sub-nanosecond laser pulses have been published in recent years. In [
8], for a laser chip design with a saturable absorber and a specially optimized laser heterostructure design, the peak power reached 35 W with a pulse width of 80 ps. In [
9], the authors simplified the laser chip design, but also optimized the laser heterostructure and demonstrated laser pulses with a peak power of 20 W and a pulse width of 120 ps. It is clear that, in most cases, the focus of the research has been on the development and optimization of the laser heterostructure. However, optimizing the designs of both the semiconductor laser chips and the current pulse generator is also important.
In order to create compact, affordable, and efficient sources of high-power laser pulses, several important conditions that must be met can be highlighted: (1) a compact generator of short current pulses, which can be integrated with a semiconductor laser chip, and (2) a laser chip featuring a heterostructure design typical of most commercially available high-power semiconductor lasers. For example, external pulse generators were used to pump semiconductor lasers in the studies presented in [
10,
11,
12]. Work [
10] shows the use of the simplest avalanche drivers; work [
11] demonstrates drivers using GaN transistors that can receive nanosecond and picosecond pulses [
12]. However, these generators are relatively large compared to the semiconductor laser chips and require a 50 Ω matching load. This study presents a high-speed current switch based on a heterothyristor designed to pump semiconductor lasers. The advantage of this approach lies in its ability to support vertical mounting, allowing the creation of “current switch–semiconductor laser” integrated vertical stacks with minimal overall dimensions and parasitic inductances in the circuit. Additionally, the proposed design does not impose significant requirements on the control current pulse profile needed to turn on the thyristor switch. As will be shown below, the developed high-power laser pulse sources are also based on semiconductor heterostructures typical of commercially available high-power laser diodes, so that, in the future, no significant modifications of the laser heterostructure technology will be required for the production of pulsed sources.
Optimal pumping conditions for semiconductor lasers (in terms of current pulse width and amplitude) for the generation of short optical pulses in gain-switching mode should provide a regime in which the optical pulse contains only the first relaxation oscillation. A further increase in the pump current amplitude results in pulse broadening due to the onset of subsequent oscillations. In this case, for a given laser heterostructure design and a fixed cavity length, an effective way to enhance peak power is to widen the emitting aperture.
In our study presented in [
13], the dynamics of laser generation in the near field for high-power semiconductor lasers with an 800 μm aperture width was investigated for the first time. The studies showed a turn-on delay in the central region of the emitting aperture in the range of 100–300 ps, the magnitude of which depends on the cavity length and the pump current amplitude. One reason for the observed dynamics may be due to the complex onset dynamics of higher-order modes that fill the cavity volume differently (in terms of spatial distribution and onset time). This delay severely limits the feasibility of creating high-power sub-nanosecond laser pulse sources based on ultra-wide emitting aperture designs.
In [
14], an approach to addressing this challenge was studied using minibars of single-mode lasers with weak optical coupling. The proposed approach did reduce the turn-on spatial inhomogeneity, but it did not result in the desired increase in the peak power.
Previously, laser pulse sources based on “high-speed current switch–semiconductor laser” vertical stacks were developed [
15]. These sources demonstrated the ability to generate laser pulses with a duration of 3 ns and a peak power of 33 W. However, the emitting aperture of the semiconductor laser was formed by two optically uncoupled mesa 100 μm wide stripes, resulting in a turn-on delay of 160 ps between the mesa stripes. This means that the proposed laser part design cannot be used to create sub-nanosecond laser pulse sources with high peak power.
Thus, existing approaches use discrete circuitry, making it impossible to create compact integrated circuits, which increases device dimensions and limits the dynamic properties for generating short pump currents and sub-nanosecond laser pulses. Furthermore, the main results have been demonstrated for laser pulse sources with durations in the range of nanoseconds, without the optimization of the current switch parameters and the laser chip design for the generation of high-power sub-nanosecond laser pulses.
In this study, the feasibility of creating high-power sub-nanosecond laser pulse sources using vertical stacks of high-speed current switches and semiconductor lasers with a wide monolithic emitting aperture is investigated for the first time. The proposed approach uses a heterothyristor as a high-speed switch, enabling the development of compact vertical stacks. In addition, a laser chip emitting aperture optimal design is presented, which is fundamentally important for operation in the high-power sub-nanosecond laser pulse generation mode. This study demonstrates the effect of the emitting aperture width on the laser pulse width and shows that the peak power in a single sub-nanosecond pulse can be significantly increased by optimizing the aperture width.
2. Experimental Results
A scheme of the laser pulse source is shown in
Figure 1. The developed design includes a laser diode chip (LD) on which a heterothyristor-based high-speed current switch (HS) is mounted (
Figure 1a). The basic electrical diagram of the device is shown in
Figure 1b and includes a storage capacitor (C) (the capacity of which was in the range of 105 pF to 392 pF) and a power supply (PS) (with a maximum voltage of 60 V). The experimental study of dynamic and optical characteristics was carried out using the setup shown in
Figure 1c, including PS and a control pulse generator (CPG) with a synchronization line (SL) for a stroboscopic oscilloscope (OC), two aspherical lenses (AL), a fiber delay line (FDL), an optical fiber on a travel compact 3-axis manual stage (FMS), and a high-speed photodetector (PD). Typically, widely used designs rely on field-effect transistors, which do not allow for vertical mounting and require high-speed control drivers [
10,
11,
12], thereby reducing the compactness of the design and increasing manufacturing complexity. The proposed HS-based switch enables vertical integration. In addition, the HS is driven by low-amplitude current pulses that do not have strict profile requirements. This is because the HS switching speed is governed by the feedback characteristics of the structure, and the control signal is only necessary to initiate the transition to the on-state.
Thus, the operation cycle of the laser pulse source based on the “HS-LD” vertical stack includes the following main steps: (1) charging the C using the external PS while the HS remains in the high-impedance off-state, so that the C can be charged with minimal loss; (2) generating a control pulse by the CPG; the control current pulse initiates the HS switching-on process with a turn-on delay time, which depends on the control current amplitude; and (3) switching the HS on, characterized by a high-speed transition to a low impedance state, which also allows for the fast discharge of the C in the HS-LD circuit; during the C discharge, a current pulse is generated in the circuit, which pumps the semiconductor laser, so that the optical pulse is generated. When the C is discharged, the HS automatically returns to the off-state, allowing the PS to recharge the C and prepare the circuit to generate the next laser pulse. In the experimental setup developed, a CPG was used to pump with current pulses of amplitudes ranging from 10 to 400 mA, durations from 0.1 to 1000 µs, and repetition rates from 0.1 to 100 kHz. It is important to note that the control current pulse profile has no effect on the dynamic characteristics of the developed switch, since the heterothyristor turn-on speed is related to the excess electron and hole accumulation rate generated by a very fast impact ionization process, unlike field-effect transistors where the turn-on speed is directly related to the speed of the control pulse applied to the gate. In the case of HSs, the control current pulse amplitude has a primary effect on the turn-on delay [
16]. For the HSs developed, the turn-on delay, depending on the operating voltage and the amplitude of the control current pulse, ranged from 10 to 200 ns. Studies of the developed HSs showed that the turn-on dynamics remained unchanged for the used control current pulses. Therefore, for the experiments carried out, a control current pulse amplitude of 200 mA, a pulse width of 100 µs, and a repetition rate of 10 kHz were chosen, giving a turn-on delay of 10 ns at an operating voltage of 50 V.
The HS heterostructure design was developed to create a laser pulse source based on the “HS-LD” vertical stack. This design included a doped n-AlGaAs emitter, a base region consisting of a high-doped 100 nm thick p-GaAs layer, and a low-doped 4 μm thick p-GaAs layer, a 0.5 μm thick n-GaAs collector, and a 0.5 μm thick p-GaAs emitter. The proposed design had a composite base region, allowing for blocking voltages up to 60 V. Increasing the blocking voltage without sacrificing speed is an important requirement in the development of a thyristor switch. This is due to the fact that, despite the device compactness, there are still residual inductances and resistances remain in the circuit. These inductances and resistances limit the C discharge rate. To mitigate the impact of this issue, it is necessary to charge the capacitor to a higher voltage. In this case, the developed HS must be able to block this voltage to ensure efficient C charging. In addition, unlike previous studies, the thickness of the collector was reduced, which increased the feedback efficiency by reducing the recombination of holes injected from the p-emitter. This improvement increased the HS control efficiency by reducing the turn-on delay and the control current.
An asymmetric heterostructure design was used for the semiconductor laser, which included wide-bandgap n-Al0.25Ga0.75As and p-Al0.35Ga0.65As claddings, a 2 μm thick Al0.2Ga0.8As waveguide layer, and an InGaAs quantum well active region placed 0.8 µm from the p-cladding. The asymmetry helps suppress the lasing of higher-order modes in the vertical waveguide and ensures operation at the zero-order vertical mode. The HS and LD heterostructures were grown on n-GaAs substrates using the metal–organic chemical vapor deposition (MOCVD) technique. Subsequently, within the framework of the developed post-growth technique, the HS and LD chips were fabricated. A reliminary characterization of the LD heterostructure was carried out in continuous-wave mode (CW) using 100 µm wide aperture LD chips. Internal optical loss and internal quantum efficiency were demonstrated to be 0.2 cm⁻1 and 99%, respectively. The threshold current and external quantum efficiency of lasing for chips with a cavity length of 2000 µm were 500 mA and 1 W/A, respectively. The CW test was limited by a pump current of 5 A, which provided 5 W in CW in the linear range of the light–current curve.
The HS chips were tri-electrode: an anode contact with a mesa diameter of 150 μm formed on the p-GaAs emitter layer, a control electrode formed on the n-GaAs collector layer, and a cathode contact formed on the substrate side. The semiconductor lasers had mesa stripe designs with aperture widths of 200 μm and 400 μm. For the experiments, two-element HS chips with a linear size of 800 × 800 μm and LD chips with a cavity length of 2000 µm were fabricated. “HS-LD” vertical stacks were fabricated and mounted on a copper carrier with the C and contact pads to form the circuit shown in
Figure 1.
In general, the increase in peak optical power of semiconductor lasers is achieved by increasing the amplitude of the pump current. In gain-switching mode, this principle also applies. However, the main challenge in developing high-power sub-nanosecond laser pulse sources based on semiconductor lasers operating in gain-switching mode is the onset of a low-power “tail”, which degrades the laser pulse profile when the pump current amplitude and/or duration exceeds a certain threshold [
13,
17]. In practice, this low-power tail can introduce errors in signal processing and range measuring. Therefore, this study involved optimizing the designs of both the LD chips (emitting aperture widths) and the HS-LD-C circuit to find the optimal parameters that provide maximum peak power with minimal pulse profile distortion.
A specially designed arrangement to measure the integral profile of high-power sub-nanosecond laser pulses was used for the experimental studies. The fact that high-speed photodetectors have a light-sensitive area with a diameter of less than a few tens of microns is the main challenge in the characterization of such pulses. This small diameter makes it impossible to collect the entire beam from high-power multimode semiconductor lasers that have an emission aperture width of several hundred microns. Consequently, the acquisition of only a portion of the beam will not provide an accurate description of the pulse profile due to distortions associated with the multimode nature of the laser beam, as will be discussed later.
Accordingly, an acquisition scheme was used in the study wherein the integral profile of the laser pulse was obtained with a high-speed photodetector by sequential measurements and scanning along the near field. A magnified image of the LD near field was built using a lens system, and the image was scanned using a photodetector mounted on a micrometric translation stage. The laser pulse profile was obtained using a photodetector NewFocus 1444-50 (New Focus, SAN Jose, CA, USA) and a stroboscopic oscilloscope Agilent 86117A (Agilent Technologies, Santa Clara, CA, USA), while the average optical power was measured using a bolometer OPHIR 3A-P-FS-12 (Ophir Optronics, Jerusalem, Israel). The integral profile of the laser pulse was obtained by averaging optical pulse profiles measured in the near field. The peak power can be calculated from the resulting integral laser pulse profile and the average optical power. A detailed description of the measurement technique is presented in [
9,
15].
In
Figure 2a,b, the integral laser pulses of 200 μm and 400 μm wide aperture LDs are shown for a circuit with C = 105 pF. The maximum peak power achieved was 30.5 W and 27 W with pulse widths at half maximum of 90 ps and 100 ps, respectively, for 200 μm and 400 μm wide apertures.
It can be observed that the 200 μm wide aperture LD begins to form a low-power tail, which increases the pulse width. In contrast, the 400 μm wide aperture LD, despite its lower peak power, maintains an undistorted pulse profile. It can be concluded that this design is suitable for use in situations where higher amplitude and/or longer duration pump current pulses are required to further increase the optical peak power. The simplest approach is based on using an HS-LD-C circuit with a larger C rating.
Figure 2c shows the integral optical pulses of a 400 μm wide aperture LD with C = 165 pF, obtained at different operating voltages of the thyristor switch. As a result, the peak power increased to 39.5 W, and the pulse width was 120 ps. In this case, the contribution of the low-power tail to the laser pulse profile is negligible. Further increase in the C value to 390 pF led to a maximum peak power increase to 71 W (
Figure 2d). However, in this case, the contribution of the low-power tail becomes comparable to the energy of the first short pulse, which is unacceptable for generating sub-nanosecond laser pulses. Notably, a significant contribution begins to manifest at a peak power of 31 W, which is substantially lower than in the previous case with a C value of 165 pF. The reduction in efficiency and peak power when generating sub-nanosecond laser pulses in the circuit with a higher C is due to the fact that not only the amplitude of the pump current pulse increases, but also its duration.
A more detailed analysis of the dynamics of laser generation along the 400 µm wide aperture obtained in the optimal HS-LD-C circuit with a capacitance of 165 pF will now be presented.
Figure 3a shows near-field lasing along the aperture.
It is evident that irrespective of the operating voltage (current pulse amplitude), lasing starts at one aperture edge and then propagates to the opposite edge.
Figure 3b shows the dependence of the onset of lasing along the aperture. In this regard, the turn-on delay time along the aperture shows a nearly linear dependence. The maximum turn-on delay between the left and right aperture edges is only slightly dependent on the pump power/current and is approximately 90 ps, while the pulse width for a given near-field segment can be as low as 66 ps. This indicates that synchronous lasing turn-on along the aperture not only reduces the pulse width but also increases the peak power. This fact also emphasizes the importance of using the correct technique for measuring the integral laser pulse profile, whereas analyzing the lasing dynamics of high-power, wide-aperture LDs based solely on measurements in a single near-field segment can lead to significant overestimations of the peak power and underestimations of the pulse width.
Figure 4 presents typical lasing spectra and the far field for the optimal circuit with C = 165 pF. The wavelength of the laser spectrum peak was 940 nm with a full width at half maximum of 10 nm. The far field has a width at half maximum of 19 degrees and 4.5 degrees for the fast and slow axes, respectively. It is evident that with different pumping conditions, there is little change in the far fields, spectrum shape, and peak position.