1. Introduction

Photonic-crystal surface-emitting lasers (PCSELs) are promising as next-generation semiconductor lasers for various applications, including laser processing, sensing, and optical communications. PCSELs have exhibited single-mode, high power, and high-beam-quality operation [1–9], which cannot be achieved with conventional semiconductor lasers. A two-dimensional stable broad-area cavity mode is formed at a singularity (Γ) point of the 2D photonic band structure, through direct (180°) and indirect (90°) optical couplings among four fundamental Bloch waves. In addition, light resonating in the cavity mode is coupled out in the vertical direction, forming a surface-emitted beam.

Single-mode operation with high output powers of 10-50 W under continuous-wave (CW) conditions has already been demonstrated at wavelengths of around 940 nm using GaAs-based PCSELs. These GaAs-based PCSELs are aimed at applications including light detection and ranging (LiDAR), laser-based material processing, and freespace optical communications [4,5,10]. InP-based PCSELs have also been developed for operation in the wavelength range from 1.3 µm to 1.55 µm, which is suitable for fiber-based telecommunications and eye-safer LiDAR. Recently, the fabrication of InP-based PCSELs using an epitaxial regrowth technique to form an air/semiconductor PC structure with a large refractive index contrast has been reported [11–14]. In our previous work, we formed deep air holes in a thick InGaAsP photonic crystal (PC) layer to achieve strong optical confinement within this layer, while also avoiding dry-etching damage to the adjacent active layer [13,14]. In addition, we have introduced double-lattice PC structures to suppress the oscillation of high-order modes [4,15], leading to single-mode CW operation at room-temperature of 1.3-µm-wavelength InP-based PCSELs with high output powers of over 200 mW [16,17].

In this work, we introduce a reflective mirror to improve the output power and emission efficiency of InP-based PCSELs. In PCSELs, light is emitted from both top and bottom surfaces of the PC layer. In order to efficiently extract light from a single side, a backside mirror that reflects the downward-emitted light is desirable. For GaAs-based PCSELs, semiconductor distributed Bragg reflectors (DBRs) have been effectively used for this purpose, owing to the ability of this material to form layer pairs with a high refractive index contrast [15,18]. Unfortunately, such high contrast cannot be achieved using InP-based materials, necessitating the use of many more layer pairs to achieve a sufficiently high reflectivity, which increases the differential resistance and lowers the wall-plug efficiency (WPE). Consequently, the application of semiconductor DBRs to InP-based PCSELs is challenging. On the other hand, the use of a highly reflective p-electrode acting as a metallic mirror has been demonstrated to effectively enhance the output power and WPE of GaN-based PCSELs at a wavelength of 430 nm [19]. The advantage of such a metallic mirror is the simplicity of its fabrication; unlike a DBR, the metallic mirror does not require additional, complicated fabrication steps, nor the growth of additional layers which increase the differential resistance. Leveraging these advantages, we introduce metallic reflectors into InP-based PCSELs. By adjusting the phase of the reflected light, we achieve 400-mW single-mode CW operation and 10-W-class pulsed operation of the PCSELs at room temperature. The WPE of the PCSELs under CW operation is 20%.

2. Effect of metal reflector on slope efficiency

First, we calculate the effect of introducing a reflector into PCSELs on the slope efficiency. Figure 1(a) shows a schematic of the InP-based PCSEL. The slope efficiency ${\mathrm{\eta }_{\textrm{SE}}}$ can be calculated as a function of the mirror reflectivity and the phase of the reflected light as follows [15].

Here, λ and η_i are the lasing wavelength and the internal quantum efficiency of the active layer, respectively; R is the reflectivity of the metal reflector; θ is the phase of the p-side-reflected light relative to that of the n-side-emitted light; and α₀, α_//, and α_V are the intrinsic loss, the in-plane cavity loss, and the vertical radiation constant, respectively. For the PCSEL considered in this work, we set λ = 1339 nm, η_i = 0.95, α₀ = 6.8 cm^-1, α_// = 6.2 cm⁻¹, and α_V= 12.5 cm⁻¹, where α_// and α_V were calculated by using three-dimensional coupled wave theory [20,21]. In Fig. 1(b), we show ${\mathrm{\eta }_{\textrm{SE}}}$ calculated as a function of θ for a family of R, indicating that ${\mathrm{\eta }_{\textrm{SE}}}$ is maximum at θ = 0°, corresponding to constructive interference of the p-side-reflected and n-side-emitted waves. In Fig. 1(c), we show ${\mathrm{\eta }_{\textrm{SE}}}$ calculated as a function of R when θ = 0°, indicating that ${\mathrm{\eta }_{\textrm{SE}}}$ increases monotonically with R. Based on these calculations, for a targeted ${\mathrm{\eta }_{\textrm{SE}}}$ of at least 0.4 W/A, which is twice as high as that of previously fabricated PCSELs without a metal reflector [16,17], we require R > 0.3. To attain this reflectivity, we fabricated the p-side electrode by using the commonly used Au-based metal for InP-based optical devices, employing a suitable metal structure and alloying temperature. Figure 2 shows the reflectivity of the fabricated metal reflector. A reflectivity of more than 30% was verified at around the targeted wavelength of approximately 1.3 µm. In addition, the reflector had a low contact resistivity of approximately 1 × 10⁻⁵ Ωcm², which was almost the same as that of the p-electrodes used in previous InP-based PCSELs. Hence, this reflector could effectively serve the dual purposes of a mirror and an electrode.

 

Fig. 1. (a) Schematic of a InP-based PCSEL with a metal reflector. (b) Calculated slope efficiency ${\mathrm{\eta }_{\textrm{SE}}}$ as a function of the phase of reflected light $\mathrm{\theta }$ for a family of mirror reflectivities R, and (c) calculated ${\mathrm{\eta }_{\textrm{SE}}}$ as a function of R when $\mathrm{\theta }$ = 0°.

Download Full Size | PDF

 

Fig. 2. Reflectivity R of the fabricated metal reflector.

Download Full Size | PDF

3. Device structure and fabrication

Based on our design, we introduce the metal reflector into the InP-based PCSEL as shown in Fig. 1(a). The fabrication process is as follows. First, an n-InGaAsP PC layer is grown atop an n-type InP substrate by metal-organic vapor phase epitaxy (MOVPE). Next, a two-dimensional double-lattice PC, consisting of pairs of elliptical and circular air holes (as shown in Fig. 1(a)), is formed by using electron-beam lithography and dry etching processes. This double-lattice structure suppresses the oscillation of high-order modes, which enables single-mode operation at a higher maximum output power [4,15]. The lattice constant (a) of the PC is set to approximately 400 nm in order to achieve surface emission at a wavelength of 1.3 µm. After the PC is formed, an InP overgrown (spacer) layer, an four-layer InGaAsP multiple-quantum-well (MQW) active layer, a p-InP cladding layer, and a p ⁺-GaInAs contact layer are grown atop the PC by MOVPE regrowth. The air-hole shapes after the regrowth process are shown in Refs. [14,22]. After regrowth, a circular isolation mesa with a diameter of 200 µm is formed to restrict the current injection area, then a p-side metal reflector with a reflectivity of approximately 40%, and which also serves as the p-electrode, is deposited on the p-InGaAs contact layer. Finally, a circular window n-electrode and an anti-reflection coating are deposited on the substrate using photolithography and evaporation processes. The advantage of this fabrication process is that, by growing the active layer atop the PC layer after the dry-etching step, the risk of damage to the active layer by dry etching is avoided. Furthermore, this process enables deep air holes with a high aspect ratio (of over 5) to be formed close to the active layer, with a separation of under 100 nm by the thin InP spacer layer, which strengthens optical coupling in the PCSEL. To investigate the dependence of the device characteristics on the phase of the reflected light, the above process is used to fabricate various PCSELs while varying the thickness of the p-InP cladding layer in increments of 50 nm, which corresponds to a phase change of π/2.

4. Device characteristics

First, we evaluated the lasing characteristics under pulsed conditions to avoid thermal effects. Figure 3(a) shows the light-output-current (L-I) characteristics under pulsed conditions with a pulse width of 10 µs and a duty cycle of 1% at 25°C as a function of the thickness changes for p-InP cladding layer. The dependences of slope efficiency (η_SE) and threshold current (I_th) on the changes in the p-InP cladding layer thickness are shown in Figs. 3(b) and (c), respectively, where the red circles indicate experimentally measured values. In Fig. 3(b), the black dashed line indicates η_SE calculated using Eq. (1) with the parameters described in Section 2 and the experimentally measured mirror reflectivity of R = 0.4. In Fig. 3(c), the black dashed line indicates I_th calculated using the equation I_th= qVCN_th³/η_i×exp(3g_th/Γg₀), where g_th = $\left( {{1 + }\sqrt {R} {\cos\theta }} \right){{\alpha }_{V}}{ + }{\mathrm{\alpha }_{{/{/}}}}{ + }{{\alpha }_{0}}$, q is the elementary charge, V is the volume of the active layer, C is the Auger coefficient, N_th is the threshold carrier density, Γ is the optical confinement factor, g₀ is the gain coefficient, and g_th is the threshold gain [23]. The calculations of I_th used C = 3 × 10⁻²⁹ cm⁶/s, N_th = 2.85 × 10¹⁸cm⁻³, and g₀ = 1500 cm⁻¹. The measured and calculated values of η_SE and I_th are seen to be in good agreement. These results show that, at a p-InP cladding thickness of around +100-150 nm, the targeted η_SE of over 0.4 W/A is achieved, indicating that the metal reflector works as designed.

 

Fig. 3. (a) L–I characteristics under pulsed conditions with a pulse width of 10 µs and a duty cycle of 1% at 25°C for changes of the p-InP cladding layer thickness ranging from +0 nm to +150 nm. (b) Slope efficiency and (c) threshold current as functions of the change of p-InP cladding layer thickness with theoretical values (black dotted line) and experimental results (red circles).

Download Full Size | PDF

We evaluate the fabricated devices with the thickness change of +150 nm in the p-InP cladding layer, which gives relatively higher value in the slope efficiency of Figs. 3(a) and (b), under CW conditions at temperatures between 15°C and 50°C. The L-I characteristics, current-voltage (I-V) curves, and WPE are shown in Figs. 4(a), (b), and (c), respectively. The maximum output powers at 25°C and 50°C are 397 mW and 188 mW, respectively. The differential resistance, calculated from the I-V curve, is 0.13 Ω. This value is almost equal to our previously fabricated PCSELs, indicating that a low contact resistance is maintained for the p-side electrode, resulting in a high WPE of 20.7% at 25°C. These output powers and WPE are nearly twice as high as those of previous works [16,17], demonstrating that the metal reflector is effective at serving its dual purpose as a p-electrode and a reflector. The decrease in WPE at high injection currents is due to the increase in energy consumption from Joule heat.

 

Fig. 4. (a) L-I characteristics, (b) I-V curves, and (c) WPE of a fabricated InP-based PCSEL with change of the p-InP cladding layer thickness of + 150 nm evaluated under CW conditions at temperatures between 15°C and 50°C.

Download Full Size | PDF

The lasing spectra are measured in temperature range from 15 to 50°C. Figure 5 shows typical lasing spectra under CW conditions at 15°C, 25°C, and 50°C (a) over a wide wavelength range and (b) close to the lasing wavelength. The injection current is set to 1000 mA, which is close to the peak WPE. In Fig. 5(a), emission peaks corresponding to each of the four band-edge modes (A, B, C, and D in order of decreasing wavelength) of the square-lattice PC are observed. Lasing operation occurs in the mode of band-edge B. Furthermore, as indicated in Fig. 5(b), high-order modes within band-edge B are suppressed owing to the double-lattice structure, resulting in single-mode lasing with a side-mode suppression ratio (SMSR) of over 43 dB at temperature of 15–50°C. The lasing wavelength is confirmed to lengthen due to temperature-borne change of the refractive index by 0.10 nm/K, which is comparable to that of InP-based distributed feedback (DFB) lasers.

 

Fig. 5. Lasing spectra of a fabricated InP-based PCSEL under CW conditions at 15°C, 25°C, and, 50°C (a) over wide wavelength range and (b) close to lasing wavelength. Each spectrum is evaluated at an injection current of 1000 mA, which is close to the peak WPE.

Download Full Size | PDF

Figure 6(a) and (b) show the far-field patterns and divergence angles, respectively, of the InP-based PCSEL under CW conditions. A circular beam is maintained and the beam divergence angle remains below 1.6° (evaluated using the 1/e² beam width) even as the injection current and temperature increase. These results demonstrate that the fabricated InP-based PCSELs enable single-mode, high-power lasing with narrow circular beams. Owing to such narrow circular beams, these PCSELs can be integrated into modules with simple, lens-free configurations for efficient coupling of light to optical fibers. Furthermore, these PCSELs are expected to improve the spatial resolution and ranging distance of light detection and ranging systems for sensing applications.

 

Fig. 6. (a) Far-field patterns and (b) divergence angles of a fabricated InP-based PCSEL under CW conditions.

Download Full Size | PDF

Finally, we evaluated the InP-based PCSEL with a p-InP cladding thickness of +150 nm under short-pulsed operation with a pulse width of 20 ns and a pulse repetition rate of 1 kHz over a wider range of injection currents. The device was operated at room temperature without the use of a thermoelectric controller. The measured optical power is shown as a function of injection current in Fig. 7. Remarkably, a high output power exceeding 12 W was achieved at a current injection of 35 A, resulting in a high slope efficiency of 0.43 W/A. To the best of our knowledge, this output power is the highest ever reported for pulsed InP-based PCSELs. At high injection currents, the slope efficiency began to slightly decrease, due to the slight accumulation of heat in the uncooled device. This decrease in the slope efficiency at high injection currents could be suppressed by using shorter pulse conditions and employing a thermoelectric cooler.

 

Fig. 7. Optical power versus injection current of the fabricated InP-based PCSEL with a p-InP cladding thickness change of +150 nm under short-pulsed conditions with a pulse width of 20 ns and a pulse repetition rate of 1 kHz at room temperature.

Download Full Size | PDF

5. Conclusion

We have demonstrated high-power and high-efficiency operation of InP-based PCSELs with a metal reflector. By optimizing the p-InP cladding layer thickness to achieve constructive interference of the p-side-reflected and n-side-emitted light, we have achieved a high output power of approximately 400 mW and a high wall-plug efficiency of 20% under CW conditions at room temperature. The PCSEL was confirmed to operate in a single mode and to emit a circular beam with a narrow divergence angle of 1.6° even at high CW output powers at temperatures from 15°C to 50°C. We have also achieved an output power of over 12 W under pulsed conditions. These results represent a significant advancement of InP-based PCSELs towards application to optical communications and sensing.