Semiconductor lasers and devices based on them are an important part of modern optics and electronics. Research on semiconductor lasers aims to improve their characteristics, expanding their applications, and driving the development of new technologies^[1,2].

One topical area of research lies in semiconductor lasers operating at wavelengths near 1550 nm. The interest in such lasers is driven by their wide range of applications in telecommunications, medicine, scientific research, industrial sectors, and consumer devices. In our recent study^[3], we presented a series of semiconductor lasers with different emission apertures, fabricated using an AlGaInAs/InP heterostructure with an ultra-narrow waveguide. By varying the emission aperture width, it is possible to select a laser with the most suitable parameters and output characteristics for specific practical requirements. Among the lasers in the presented series, those with a stripe width of 20 µm demonstrated the highest lateral brightness of 1.09 W/(mm · mrad) at the operating current, which is an important parameter for many applications. Due to their dimensions, these lasers operate in a few-mode lasing regime, showcasing switching behavior between lateral modes that leads to abrupt changes in optical power. In applications involving beam coupling into various optical systems and signal transmission, such operational characteristics may be undesirable. Consequently, further research is crucial to better understand and control the few-mode operation of semiconductor lasers.

The goal of this study is to experimentally investigate the unique features exhibited by lasers with a 20-µm-wide emission aperture. Emphasis will be placed on examining their radiative characteristics directly related to the mode structures of the emitted light, such as the spatial distribution of radiation and lasing spectra.

The heterostructure design is described in Ref. [3]. Based on the provided heterostructure, semiconductor lasers with a 20-µm-wide stripe were fabricated using the shallow mesa approach^[3]. The choice of laser cavity parameters has significant importance, as they impact various internal processes such as the photon distribution along the cavity axis, threshold gain, and threshold density, ultimately affecting the laser’s output characteristics. From a scientific research perspective, the study of a set of lasers with varying cavity parameters enables us to identify features within experimental data and describe the specific characteristics of 20-µm-wide-stripe lasers. To address this objective, several series of semiconductor lasers with different cavity lengths had been fabricated: 1200, 2400, 3600, and 4800 µm. In the longitudinal direction of the laser cavity, photons are distributed non-uniformly. Specifically, for a cavity with symmetric mirror reflectivity coefficients, the distribution follows a curve with a bend in the middle of the cavity length. Conversely, when the cavity has different mirror reflectivity coefficients, the photon density near the output mirror reaches its maximum^[4]. In the case of lasers emitting through the front anti-reflection coated mirror, the non-uniform distribution of the photon density leads to the longitudinal spatial hole burning (LSHB) effect, which, albeit weak, does impact the laser’s output characteristics^[5]. To eliminate the influence of the LSHB effect and simplify post-growth operations, the facets of the laser cavity were uncoated. The laser mirrors were naturally formed (by cleaving) with power reflectivities of around 30%.

The study of electro-optical characteristics involved measurements of light-current and voltage-current curves (L-I and V-I curves) of the laser. The power of the laser emission was measured using a Thorlabs S405C thermal detector (5 W) and placed in close proximity to the laser mirror. Simultaneously, with the power measurements, V-I curves were measured using a voltmeter. The thermal detector and voltmeter were connected to a computer, which automatically controlled the pumping process and generated L-I and V-I graphs.

The study of light characteristics included measurements of the spatial intensity distribution profiles of the electromagnetic radiation in the near and far fields, as well as measurements of laser spectra.

The near-field profiles were obtained using a system of aspherical lenses, providing 100-fold magnification, and a high-resolution CCD camera with a phosphor coating (Ophir-Spiricon SP620U-1550). To calibrate the experimental setup, a 100-µm-wide-stripe semiconductor laser, fabricated using the deep mesas approach, was measured. The deep mesas approach completely eliminates lateral current spreading, resulting in the lateral near-field width being identical to the stripe width. Subsequently, the recorded spot sizes in pixels were converted to their corresponding real values in micrometers for the 20-µm-wide-stripe lasers, based on these calibration results.

The far-field measurements were conducted by scanning the radiation pattern. To achieve this, the laser was positioned on a rotating platform, and during the rotation, a photodetector continuously registered the laser emission signal.

For accurate laser spectra measurement, accounting for all radiation components, the laser beam was collected using an integrating sphere and coupled into an optical fiber with a core diameter of 100–200 µm. The signal was then recorded using the Advantest Q8384 spectrum analyzer, which operated within a spectral range of 600 to 1700 nm.

All measurements were performed in continuous operation mode of the laser at several heat sink temperatures (15°C, 25°C, and 35°C). To maintain a laser temperature with an accuracy of 0.1°C, a thermal stabilization system was employed, consisting of a Peltier element, a water-cooled radiator, and a temperature sensor.

In the upcoming analysis, we will examine the measurement results obtained at a heat sink temperature of 25°C for each series of lasers with different cavity lengths. Figure 1 illustrates representative V-I curves for each laser series, showcasing the overlapping and common profiles observed across all lasers within a series. At room temperature, for all lasers of any length, the cutoff voltage was approximately 0.9 V, and the series resistance decreased as the cavity length increased due to the enlarged contact area of the laser.

Table 1 presents the L-I curves of all the studied lasers, demonstrating that the L-I curves of all the lasers exhibit sharp bends and breaks. These bends, specific to laser diodes with a narrow aperture^[6,7], are called kinks and are caused by switching between lateral modes. It is important to note that upon repeated measurements under the same conditions, the profiles of the L-I curves and the positions of the kinks remain unchanged for each specific laser. It should be particularly emphasized that for all the measured samples, the kinks exhibit dynamically stable behavior.

For different laser samples of the same length, the position of the kinks in the L-I curves and, correspondingly, the appearance of the L-I curve differ slightly. This is likely due to the always-present minor differences between any two real crystals, which emphasizes the high sensitivity of mode competition to the laser cavity parameters. For each series of lasers, it is possible to distinguish a range of pump currents within which the L-I curves behave differently. We will refer to this range as the “L-I curve spread range”. For the series of 1200-µm-long cavity lasers, the L-I curve spread range lies within the pump current up to 1.8 A; for the 2400-µm-long cavity lasers, up to 2.8 A; for 3600-µm-long cavity lasers, up to 3.8 A; and for the 4800-µm-long cavity lasers, up to 5.2 A. It is evident that as the cavity length increases, the L-I curve spread range also increases.

Beyond the L-I curve spread range for lasers of the same length, the power kinks either become less pronounced or are not observed, and the L-I curves become more similar across the samples. We will refer to this range as the “L-I curve overlap range.” Accordingly, it can be stated that mode competition within the L-I curve overlap range leads to the same result, irrespective of the specific characteristics of the individual sample.

Table 1 also presents typical spectra for each laser series. Each transverse mode corresponds to a set of equidistantly spaced lines, representing longitudinal modes. When operating in a few-mode regime, these lines overlap, resulting in a broad emission spectrum with a sharp long-wavelength edge and a gradual short-wavelength edge.

The emitted wavelength, in relation to the pump current, was plotted alongside the L-I curves using the measured spectra. The wavelength was determined at the peak intensity of the spectra. Within the L-I curve spread range for lasers of the same series, some changes in the spectrum shape and wavelength associated with the switching of transverse modes were observed. In the L-I curve overlap range, the spectral shape and the wavelength do not differ for all lasers of the same series, and a linear wavelength dependence on the pump current is observed. The slope of the wavelength-current curve differs for lasers with varying cavity lengths, and as the length increases, the slope decreases due to a decrease in thermal resistance.

All measurements were also carried out with a change in the heat sink temperature by ±10°C. Figure 2 illustrates the L-I curves and spectra of a laser with a cavity length of 2400 µm at heat sink temperatures of 15°C, 25°C, and 35°C. As the temperature increases, the threshold current rises for all laser series samples, while the external differential efficiency and optical power decrease accordingly. With slight temperature variations, the general profile of the L-I curve (position of kinks) is preserved over the entire range of pump currents, and the spectral shape also remains unchanged, except for the shift in position (the wavelength shift coefficient amounted to 0.6 nm/K). This further confirms the stability of the mode structure of each individual laser under minor variations in external parameters.

The recorded intensity distribution profiles of electromagnetic radiation in the near and far fields along the lateral axis at a temperature of 25°C are presented in Table 2. Variations in intensity distribution profiles are observed for different samples of the same length, both in the near and far fields within the current range corresponding to the L-I curve spread range. As mentioned earlier, these differences are related to the sensitivity of the mode-switching process to the specific features of each laser. In the L-I curve overlap range, the intensity distribution profiles in the near and far fields also overlap for lasers of the same length.

The near-field patterns of all lasers exhibit complex intensity distributions, which is a result of the superposition of intensities from multiple competing lateral modes. Within the L-I curve spread range, distinct individual intensity maxima are observed in the near-field patterns, while the intensity distribution in the far field shows an asymmetric structure with multiple peaks, which is common to the few-mode operation regime.

Within the L-I curve overlap range, the near-field patterns become smoother for lasers with cavity lengths of 1200 and 2400 µm, while lasers with cavity lengths of 3600 and 4800 µm continue to show distinct individual maxima. In the far field, the radiation distribution for all lasers takes on a more or less regular dome-shaped form and broadens with a further pump current. This could indicate the involvement of a greater number of lateral modes and the laser operating in a multimode regime.

Vertical far fields are the same for all lasers and indicate the fundamental mode lasing. In the vertical direction, the far-field profile is governed by the thicknesses of the epitaxial layers and their refractive indices, ensuring similarity across different samples. As the pump current increases, the vertical divergence (along the fast axis) does not change. The beam divergence at full-width at half-maximum (FWHM) was approximately 28°.

To facilitate the comparison of the obtained results, it is convenient to examine the dependencies of the lateral beam width and angular beam divergence on the pump current (Table 3). The lateral beam width (W_lat) was measured from the near-field intensity profiles at the level of 1/e² (13.5%) of the maximum intensity. The laser beam divergence (θ) was determined as the width of the far-field pattern at the level of 1/e² of the maximum signal intensity.

For all lasers near the lasing threshold (up to a pump current of 1 A), the lateral near-field beam width is approximately 40 µm. Beyond that, within the L-I curve spread range, the lateral beam width diverges significantly among different laser samples of the same length. Within the L-I curve overlap range, the lateral beam width for lasers of the same length becomes indistinguishable between samples and narrows. With further pump current increase, the lateral beam width continues to decrease smoothly due to the increased temperature in the active region caused by self-heating and enhanced mode localization near the stripe contact area.

The lateral beam divergence (along the slow axis) as a function of the pump current for all laser series exhibits a curved profile with a minimum at the current point where the L-I curve overlap range begins. The lateral divergence continues to increase with the current, associated with a narrowing lateral beam width and the onset of an increasing number of lateral modes.

Based on the obtained data, the lateral brightness of the beam, Blat, can be determined. Blat is defined as Blat=Pout/BPPlat,(1)where Pout represents the optical output power, and BPPlat is the lateral beam propagation parameter, calculated as 0.25·wlat·θ^[8]. Figure 3 presents the results. As the pump current increases, the lateral brightness increases for all laser series. The slope of the dependence of lateral brightness on the pump current decreases with the cavity length increase.

Thus, for low pump currents, lasers with a cavity length of 1200 µm exhibit the highest lateral brightness. Such lasers demonstrated a lateral brightness of approximately 0.38 W/(mm · mrad) at a pump current of 2.5 A. However, at high pump currents, shorter lasers experience more pronounced heating, which results in thermal rollover and a subsequent reduction in lateral brightness^[9,10]. In the presented range of pump currents (up to 6 A), the series of lasers with a cavity length of 2400 µm achieved the highest lateral brightness from the front mirror, reaching a value of 0.57 W/(mm · mrad).

As a result of the study, the following characteristics and patterns of operation in continuous-wave 20-µm-wide aperture semiconductor lasers have been identified:

1. •There are two ranges of pump currents in which the laser characteristics behave differently. In the first range, corresponding to the operation in the few-mode regime and mode switching, the specific features of each individual sample strongly influence the mode competition, resulting in a spread in the characteristics (output power, spectra, near- and far-fields) among identical samples. In the second range, characterized by negligible spread in characteristics among samples, it can be concluded that the mode competition consistently produces identical results, regardless of the local features of the sample;
2. •As the cavity length increases (ranging from 1200 to 4800 µm), the first range of pump currents expands, meaning the current at which the transition to the second range occurs shifts towards higher pump currents;
3. •The most distinct separation between these two ranges is observed in the wavelength-current curves, where a sharp non-thermal shift of the spectra towards the longer wavelength region by 20–40 nm occurs, with the transition to the second range happening at a pump current of 1.8 A for the 1200-µm-long cavity lasers, 2.8 A for the 2400-µm-long cavity lasers, 3.8 A for the 3860-µm-long cavity lasers, and 5.2 A for the 4800-µm-long cavity lasers;
4. •In the second range of pump currents, the laser characteristics become more predictable, with the near-field width approaching the specified dimensions of the lateral waveguide (23 µm for the 1200-µm-long cavity lasers, 22 µm for the 2400-µm-long cavity lasers, and 32 µm for the 3600-µm-long cavity lasers), while also exhibiting an increased uniformity in the far field;
5. •Temperature variations within ±10°C do not significantly affect the mode competition in each specific sample, with the range of pump currents where spread in characteristics is observed remaining unchanged for each laser, indicating the stability of the mode structure to small changes in external parameters;
6. •All characteristics of each laser remain unchanged over time, indicating the dynamic stability of the mode structure; and
7. •Reproducibility of the characteristics is observed for each sample in repeated measurements.

The commercial application of these lasers in continuous mode is primarily viable within the second range of pump currents, which corresponds to the L-I curve overlap range. First, in this current range, the laser operates in a relatively predictable manner, and its characteristics are reproducible. Second, within this current range, it is possible to achieve an optimal balance between optical output power and efficiency. Third, this range allows for the highest lateral brightness of the emitted beam, which is crucial for many applications. The lasers with a cavity length of 3600 µm demonstrated the highest optical output power, reaching 0.84 W from a single mirror (1.68 W total) at a pump current of 7 A. The 1200- and 2400-µm-long cavity lasers exhibited the minimum near-field lateral width (25–22 µm), while 3600-µm-long cavity lasers showed the minimum far-field lateral divergence (8.8°–10°). The 2400-µm-long cavity lasers demonstrated the maximum lateral brightness within the range of 0.47–0.57 W/(mm · mrad) from a single mirror.