1. Introduction

Directly modulated lasers (DMLs) have attracted significant interest as potential light sources for short-reach data transmission due to their lower cost, more compact size, and lower power dissipation. The modulation bandwidth of semiconductor lasers is typically limited by the relaxation oscillation frequency, which is determined by the conversion rate of carriers and photons. To improve the modulation bandwidth, AlGaInAs multi-quantum wells (MQWs) active regions are widely adopted to increase differential gain [1–8]. Additionally, the active region volume is reduced using the buried heterostructure [3,4,9], and the shortened laser cavity length [5–8,10–13]. The modulation bandwidths up to 29 GHz and 30 GHz are reported for 1.3-µm InGaAlAs distributed feedback (DFB) lasers with a short cavity length of 100 µm [14] and 850-nm vertical-cavity surface-emitting laser (VCSEL)with a small oxide aperture [15], respectively. Furthermore, optical injection locking [16–21], push-pull modulation [22–24], photon-photon resonance (PPR) effects [25–36] and detuned loading effects [30,31,37–40] have been explored to further increase the modulation bandwidth.

In addition to the commonly known lasers, the coupled-cavity laser composed of a whisper-gallery-mode (WGM) microcavity and a Fabry-Pérot (FP) cavity has been investigated to achieve single-mode lasing, all-optical signal processing, and wavelength tunability [41–47]. Unlike traditional coupled-cavity lasers [48], the butt-coupled WGM-FP lasers have a self-consistent mode field pattern across the whole cavity, as they don't have a distinct boundary between the two cavities, making the laser more robust. The WGM-FP lasers can be fabricated with a standard planar fabrication process similar to the FP cavity, which doesn't require complicated multiple regrowth steps or high-precision sub-wavelength grating preparation, making it cost-effective and improving the manufacturing process tolerance in practical industrial applications.

In this paper, we present the latest advancements made in the development of the directly modulated deformed-square-FP coupled cavity (DFC) laser, which is a type of WGM-FP coupled-cavity laser. Using the two-dimensional finite element method (FEM), we simulate the equivalent reflection spectra of the deformed square microcavity (DSM) with different deformation parameters for the light incident from the FP cavity. We optimize the DSM's parameters to achieve a comb-like and narrow-peak reflection spectrum. The wavelength interval between two adjacent maximum reflectivities is equivalent to the free spectral range (FSR) of the DSM, which is much larger than that of the FP cavity suitable for single-mode operation. We fabricate the DFC laser on an AlGaInAs/InP compressively strained MQWs epitaxial wafer using standard planar fabrication processes. We achieve single-mode lasing and demonstrate high-speed direct modulation with the 3-dB modulation bandwidth of up to 30 GHz for the DFC laser, thanks to the intracavity-mode PPR effect. We also present clear eye diagrams at the modulation rates of 10 Gbps and 25 Gbps.

2. Device design and numerical simulation

Figure 1 shows the schematic diagram of the DFC laser. The resonator cavity is created by connecting the traditional FP cavity to one vertex of the DSM, which is a type of coupled cavity with a strong butt coupling. The DSM acts as an equivalent resonant-enhanced wavelength-sensitive reflector of the FP cavity and exhibits good mode selection properties, resulting in a self-consistent mode field pattern in the entire cavity. The reflection spectrum of an undeformed square microcavity is relatively complex, with many peaks related to the cavity's high-order transverse mode within one FSR. As the field distribution of higher-order modes is more concentrated near the corners, their influence can be suppressed by cutting the corners while having almost no effect on the reflection characteristics of the fundamental mode.

 

Fig. 1. Schematic diagram of the DFC laser and the simulation model used in FEM simulation.

Download Full Size | PDF

We define the geometric dimensions of the DFC using the following parameters: the square side length of a, the chamfer deformation parameter δ, the length L, and the width d of the FP cavity as shown in Fig. 1. Additionally, we introduce a rectangular hole with lengths l and widths w into the DSM for further control of the reflection spectrum. The FEM (commercial software: COMSOL Multiphysics) is used to simulate the transverse-electric (TE) modes in the DFC laser, as the TE modes dominate the modal gain in the experimentally used compressively strained MQW wafer. In the numerical simulation, the resonator cavity has an effective refractive index of 3.2, surrounded by the benzocyclobutene (BCB) with a refractive index of 1.54. The outermost boundary of the entire model is set to the perfectly matched layer (PML) condition to absorb scattered light and terminate the calculation.

The reflection spectrum and mode field pattern of DSM can be controlled by varying its deformation parameter to suppress the high-transverse-order WGMs and the interaction between different transverse modes. Figure 2(a) shows the reflection spectrum map near 1310 nm of the DSM with a = 13 µm and different δ and without a hole, for the light incident from the FP cavity with d = 1.5 µm. In the simulation, the right end of the FP cavity is set as a numerical port to provide a wave excitation source (first-order plane wave) and record the reflectivity at different wavelengths. The reflection peaks relative to the high-transverse-order WGMs are quickly suppressed with an increase in δ, and coupled modes with higher reflectivity appear within a specific range of deformation (δ = 3.6 ∼ 4.2 µm). However, as the deformation parameter continues to increase, it becomes difficult to form stable high-Q modes in the DSM, resulting in the reflection spectrum showing some broad envelope peaks.

 

Fig. 2. (a) The reflectivity map near 1310 nm of the DSM with a = 13 µm and different δ for the light injection from the FP cavity with d = 1.5 µm. (b) Simulated reflection spectra for the DSM with δ = 0 and 3.8 µm. (c) Mode intensity profiles of |H_z| corresponding to the modes s₁, s₀, s_0′, s₂, r₀ and r₁ in (b).

Download Full Size | PDF

Figure 2(b) shows the simulated reflection spectra for the DSM with δ = 0 and 3.8 µm as the blue dashed line and the red solid line, respectively. The spectra show maximum reflectivity around the DSM resonances due to the coupling between WGMs and the FP modes. The DSM is an undeformed square at δ = 0, and several typical modes in its reflection spectrum are denoted as s₁, s₀, s_0′, and s₂, respectively. Similarly, the main high-reflectivity peaks are denoted by r₀ and r₁ in the reflection spectrum of the DSM with δ = 3.8 µm. The corresponding mode intensity profiles of |H_z| are shown in Fig. 2(c). The modes in the undeformed square cavity can be described according to a quasi-analytical model. The adjacent longitudinal modes of WGMs have different symmetries and distinct field distributions in the corner regions [49]. When the waveguide is directly connected to the vertex, the adjacent high-Q WGMs have opposite symmetry relative to the waveguide. We use the first-order plane wave as the source to excite the symmetric modes in the waveguide, as the symmetric fundamental mode in the FP cavity has the lowest propagation loss. s₁ and s₀ represent the first-order and fundamental WGMs, which appear as a peak and a dip in the reflection spectrum, respectively. s_0′ represents the position of the anti-symmetric fundamental WGM, and a field distribution distinct to s₀ is obtained because of the symmetric excitation source. s₂ represents the high-transverse-order modes with the highest reflectivity but relatively low Q factor. Overall, the complex reflection spectrum structures and wide reflection peaks are not conducive to single-mode operation and of lasing mode manipulation for the square-FP coupled-cavity lasers. r₀ represents the high-Q WGM inside the DSM with the reflectivity of 0.75 at the wavelength of 1310 nm, and its mode field distribution is essentially the same as the modes at wavelengths of 1295 nm and 1325 nm with the wavelength interval of one FSR, which is beneficial for enhancing the continuous tunability for laser. There is remaining a wide reflection peak denoted as r₁.

To suppress mode r₁, a rectangular hole with l = 2 µm and w = 1 µm is introduced to the area where mode r₁ has strong field, but mode r₀ has weak field. The position of the hole is optimized in simulation, and the resulting reflection spectrum is shown in Fig. 3(a). The reflection spectrum of the DSM only has one set of narrow reflection peaks relative to mode r₀. This feature enhances the single-mode yield and Q factor of the hybrid lasing mode. Moreover, the reflection spectra of the DSM at different gain levels are compared. The relationship between the gain coefficient of g and the imaginary part of the refractive index of n can be expressed as

 

Fig. 3. (a) Simulated reflection spectra at different gain levels for the DSM after introducing inner hole with l = 2 µm and w = 1 µm. (b) Comparison of the simulated reflection spectra and the mode Q factor distribution. (c) Mode wavelengths and Q factors versus the variations of Δn_FP. (d) Normalized mode intensity profiles of |H_z|² and detailed view of local field in FP cavity of coupled hybrid mode h₀ and h_0′ at Δn_FP = 0.0003.

Download Full Size | PDF

In Fig. 3(b), the simulated reflection spectra of the DSM and the mode Q factor distribution in the DSM-FP coupled cavity are compared. The high reflectivity modes match very well with the high-Q modes. The mode Q factors are calculated using the model shown in Fig. 1, with the FP cavity length L = 300 µm, by setting the horizontal symmetry plane (the white dashed line in Fig. 1) as a perfect electrical conductor (PEC) boundary for the symmetric TE modes. The mode coupling between the WGMs and the FP modes and the modulation of Q factors for the coupled mode in a DFC are further studied numerically. The mode wavelengths and Q factors versus the variations of the refractive index of the FP cavity (Δn_FP) are plotted in Fig. 3(c). The red horizontal dotted line at 1309.42 nm corresponds to the mode wavelength of the fundamental WGM in the DSM without FP cavity. The blue dashed lines indicate the mode wavelengths of the fundamental transverse mode versus Δn_FP in the FP cavity. Wavelength crossing and mode Q factor anticrossing occur due to the mode coupling between the WGM and the FP modes, the resulting coupled hybrid modes are marked as h₀ and h_0′. Modes h₀ and h_0′ correspond to the WGM and the FP modes respectively when they are far away from the crossing point. In Fig. 3(c), by tuning Δn_FP from -1.5 × 10⁻³∼1 × 10⁻³, the wavelength of the high-Q hybrid mode h₀ fluctuates slightly around 1309.42 nm, and its Q factor is modulated due to the mode coupling effect. At the same time, the wavelength of the low-Q hybrid mode h_0′ increases linearly, and its Q factor also experiences a certain increase from 2.5 × 10³ to approximately 4 × 10³ at Δn_FP = 0.0004, which is slightly off from the wavelength crossing point at Δn_FP = 0.0001. It's important to note that in the DFC, the mode coupling coefficient is a complex number, which means that the extreme values of the mode Q factor can appear at a position deviation from the mode wavelength crossing point. The normalized mode intensity profiles of |H_z|² and the detailed view of the local field in the FP cavity for hybrid modes h₀ and h_0′ at Δn_FP = 0.0003 near the crossing point are shown in Fig. 3(d). The optical confinement factors in the DSM are 80% and 33% corresponding to a strong localization inside the microcavity and a distribution throughout the entire coupled cavity for modes h₀ and h_0′, respectively. The two hybrid modes with close wavelength can be used to obtain photon-photon resonance to enhance the bandwidth for the directly modulated laser.

3. Device fabrication and packaging

The DFC laser is fabricated on an AlGaInAs/InP epitaxial wafer which is grown using metal-organic chemical vapor deposition (MOCVD) on an N-doped InP substrate. The active layer consists of AlGaInAs MQWs that include eight compressively strained quantum wells and nine barriers situated between two AlGaInAs separate confinement layers. Compared to InGaAsP, AlGaInAs MQWs have higher differential gain and perform better at high temperatures. The i-line projection photolithography method is used to produce high-quality device patterns. During fabrication, a 600-nm SiO₂ layer is grown using plasma-enhanced chemical vapor deposition (PECVD) to serve as a mask. Inductively coupled plasma (ICP) is then used to etch the cavity to a depth of over 4.2 µm to ensure the formation of the WGM resonance. Scanning electron microscopic (SEM) images of the DFC after the ICP etching process are shown in Figs. 4(a) and 4(b). The isometric view indicates that the etched resonator cavity's pattern fidelity is relatively nice. At the same time, there are only some very shallow vertical stripes on the sidewalls, and the overall smoothness is also relatively good. From the perspective of steepness, there is a certain tailing at the part of the cavity near the base, which may lead to increased vertical loss in the laser. Next, a passivation layer consisting of a 200-nm silicon nitride layer (SiN_x) is deposited to protect the exposed active region interface from oxidization, and a BCB cladding layer is coated to form a flat surface. It is worth mentioning that the dielectric constant of BCB is very small, and the patterned P-electrode formed on the BCB layer can reduce the parasitic capacitance, thereby reducing the RC limitation in high-speed direct modulation. Then, an electrical isolation trench is made by ICP etching off the heavily doped InGaAs ohmic contact layer between the DSM and FP cavity. After that, the contact window is opened by ICP etching for current injection, and the Ti/Pt/Au patterned P-electrode is deposited using the e-beam evaporation and lift-off process. The wafer is thinned to about 120 µm using mechanical polishing. Finally, the Au/Ge/Ni metallization layer is magnetron sputtered and rapidly thermally annealed (RTA) on the N-side of the wafer to form the N-electrode.

 

Fig. 4. SEM image of the DFC after the ICP etching: (a) isometric view, (b) details of sidewalls. (c) Optical microscope image of the fabricated DFC laser. The (d) top view and (e) front view of the DFC laser with sintered on a double-layered AlN GCPW.

Download Full Size | PDF

The optical microscope image of the DFC laser fabricated under a microscope is displayed in Fig. 4(c). The DSM and FP cavities are injected with current through two patterned P-type electrodes separately. The devices are cleaved over the FP region, resulting in a length of about 300 µm, which allows for lasing resonances to occur as light bounces between the DSM and the cleavage facet. The laser is then sintered onto an aluminum nitride (AlN) ground coplanar waveguide (GCPW) to match with the RF probe in the test system, which also helps to improve the laser's heat dissipation capability. The GCPW has a bandwidth over 50 GHz and its surface signal trace is designed asymmetrically, corresponding to the center of the two separate patterned electrodes of the DFC laser. To minimize the influence of parasitic parameters, a double-layer GCPW is used to place the laser in a recess, with its top surface on the same plane as the GCPW. The top and front view images of the DFC laser sintered on the GCPW are shown in Figs. 4(d) and (e), respectively.

4. Experimental results and discussion

The packaged DFC laser is placed on a copper heat sink for testing and the temperature is maintained at 288 K controlled by a thermoelectric cooler (TEC). Continuous-wave (CW) injection currents are applied to the DSM (I_DSM) and the FP (I_FP) regions separately. For a DFC laser with a = 13 µm, δ = 3.8 µm, l = 2 µm, w = 1 µm, d = 1.5 µm, and L = 300 µm, the output powers coupled into a tapered single-mode fiber (SMF) and the applied voltage versus I_FP when I_DSM is fixed at 5, 10 and 20 mA are plotted in Fig. 5(a), respectively. A series resistance of 11 Ω is estimated for the FP cavity after deducting the matching resistance of 35 Ω on the GCPW. The threshold current decreases and the output power increases with the rise of I_DSM due to the increased gain of the DSM.

 

Fig. 5. (a) Single-mode fiber coupled powers and the applied voltage versus I_FP when I_DSM is fixed at 5, 10 and 20 mA, respectively. (b) The lasing spectra of the DFC laser at different values of I_FP (14, 40, 51, and 54 mA) and I_DSM = 20 mA. (c) Lasing spectra versus I_FP when I_DSM is fixed at 20 mA.

Download Full Size | PDF

Figure 5(b) shows the lasing spectra of the DFC laser at different values of I_FP (14, 40, 51, and 54 mA) and I_DSM = 20 mA, corresponding to several special states of the laser. Figure 5(c) shows the lasing spectra versus a wide range of I_FP at I_DSM = 20 mA. Due to the high gain of the DSM, the WGMs lase when the FP cavity is below the threshold. The wavelengths of the FP modes experience a blue shift because of the carrier dispersion effect as the current of the FP cavity increases when I_FP < 20 mA. When the WGM lasing peak approaches the FP modes, strong interactions occur such as the periodic oscillations in the spectrum at I_FP = 14 mA shown in Fig. 5(b), which is similar to the interaction between different order transverse modes within a WGM microcavity laser [50]. This process corresponds to a significant absorption loss of the FP cavity, which leads to fluctuations of the output power when I_FP is small in Fig. 5(a). As I_FP increases beyond 20 mA, the FP cavity gradually provides gain instead of absorption, and the FP modes appears as shown in the spectrum at I_FP = 40 mA of Fig. 5(b). The FP cavity, acting as a traveling-wave amplifier, contributes to a certain enhancement of the laser output power, and the lasing mode can also be regarded as the hybrid mode h₀ that is mainly localized in the WGM cavity. Due to the clamping of the carrier concentration, the FP mode wavelength exhibits a red shift with increasing current due to the thermal effect. When I_FP exceeds 47 mA, hybrid mode h_0′ reaches its threshold and begins to lase, leading to a rapid increase in the output power. The hybrid mode h_0′ has a significant competitive advantage due to the gain of the entire cavity, which consumes more carriers in the DSM cavity because of the higher photon density, which results in the low intensity of the WGM and being enveloped within the hybrid mode in the spectrum at I_FP = 51 mA shown in Fig. 5(b).

The reflection peak of the DSM is very narrow, and the red-shift rate of its central wavelength is significantly smaller than that of the FP modes as I_FP increases, then a deterioration in the phase matching between these two cavities appears. As a result, the output power of the laser exhibits some regions of saturation. Within the gain spectrum range, the reflection peaks of adjacent longitudinal modes of the DSM will separately produce dominant hybrid modes through the vernier effect with the FP modes as I_FP increases. Therefore, there is a noticeable mode-hopping phenomenon in the spectra in Fig. 5(c), corresponding to a clear change in the linearity of the output-power curve in Fig. 5(a). When the hybrid mode begins to form, there is a certain frequency difference between these two hybrid modes, as shown in the spectrum at I_FP = 54 mA in Fig. 5(b). As I_FP increases, this frequency difference gradually decreases until the optical spectra of the two modes overlap, because mode h_0′ at the short-wavelength side has larger redshift rate compared to mode h₀ considering their distinct field distributions as shown in Fig. 3(d).

The small signal electro-optic (EO) modulation responses of the DFC laser are demonstrated by injecting an RF signal with the power of -10 dBm together with the bias current into the DSM cavity through a 65-GHz bandwidth bias-tee and RF GSG probe. A 50-GHz bandwidth high-speed photodetector and 40-GHz bandwidth vector network analyzer (VNA) are used in the experiment. The EO modulation responses and corresponding lasing spectra of the DFC laser with various I_FP with I_DSM = 20 mA are plotted in Fig. 6(a) and (b), respectively. The output spectrum of the DFC laser at I_FP = 40 mA corresponds to the hybrid mode h₀, with a side mode suppression ratio (SMSR) of 42.2 dB, and the EO modulation response curve conforms to the conventional response function. The 3-dB bandwidth is 19.3 GHz, which represents a significant improvement compared to previous lasers of the same type. This enhancement is attributed to the optimized design of the laser resonator cavity and the very small parasitic parameters introduced by the packaging structure. The hybrid mode h₀ acts as a weak peak located at the red side of the main lasing hybrid mode h_0′ at I_FP = 54, 61, and 74 mA, with wavelength differences of 0.14, 0.148, and 0.16 nm and 28.8, 40.5, and 37.8 dB lower than the main lasing mode, respectively. Additional side mode appears in the blue side of the main lasing mode at I_FP = 74 mA, which may be the four-wave mixing peak results from the nonlinear interaction between modes h₀ and h_0’. The PPR peaks with the center frequencies of 24.5, 25.9, and 28 GHz at the EO modulation response curves are observed in Fig. 6(a), which results in the 3-dB bandwidth of the laser being increased to 29.6, 26.9, and 30 GHz, respectively.

 

Fig. 6. (a) The small signal electro-optic responses and (b) the corresponding lasing spectra of the DFC laser at I_FP = 40, 54, 61 and 74 mA with I_DSM = 20 mA. (c) The comparison of the small-signal modulation responses at I_DSM = 20 mA and I_FP = 75 mA corresponding to the RF signals applied to the DSM and FP cavities respectively (the main lasing mode is h_0′ at 1323.112 nm).

Download Full Size | PDF

Figure 6(c) shows the comparison of the small-signal EO modulation responses at I_DSM = 20 mA and I_FP = 75 mA corresponding to the RF signals injected into the DSM and FP cavities, respectively. The relaxation oscillation frequency of the device and the center frequency of the PPR peaks are consistent under the same bias currents. However, when the RF signal is injected into the FP cavity, the intensity of the EO modulation response is very weak, leading to the 3-dB bandwidth of only 10.7 GHz. In fact, from the simulation results shown in Fig. 3(d), we know that the photon number ratio of the high-Q hybrid mode within the DSM is significantly larger than that within the FP cavity. At the same time, the field of the low-Q hybrid mode is distributed in the entire coupled cavity. When the RF signal is injected into the DSM, both modes are modulated, which results in a strong EO modulation response of the laser, even affecting the flatness of the curve. Conversely, when the RF signal is injected into the FP cavity, only a smaller portion of the hybrid mode photons are modulated, and the lower modulation efficiency leads to a weak overall response of the device.

Utilizing a 64 Gb/s pulse-pattern generator (PPG) to supply the modulation current, we perform large-signal modulation measurement using a high-speed photodetector with 70-GHz bandwidth and a 90-GHz digital sampling oscilloscope. It should be noted that since the effect of large-signal modulation is highly related to the linearity of the output-power curve, we find that the quality of the eye diagram is better when the modulation signal is injected into the FP cavity, which limits the modulation rate of the DFC laser. In the subsequent work, it is necessary to improve the design of the laser to regulate the spatial distribution of the mode field to achieve a flat small-signal modulation response, and fully utilize the modulation bandwidth of the laser.

Figure 7 shows the eye diagrams for back-to-back (BTB) configuration with a nonreturn-to-zero (NRZ) pseudo-random binary sequence (PRBS) lengths of 2¹⁵-1 input into the FP cavity. The modulation voltage (Vpp) of the PPG is amplified from 580 mV to 1.78 V after passing through a microwave amplifier with a gain of 22 dB and a fixed attenuator of -12 dB. The optical eye diagrams at large-signal modulation rates of 10 Gbps and 25 Gbps at I_DSM = 20 mA and I_FP = 43 mA are shown in Fig. 7(a) and (b), and the extinction ratios (ERs) are 4.5 and 4.2 dB, respectively. The rising edge of the transmitted signal in the eye diagram has a noticeable overshoot, which may be related to the high relaxation-oscillation peak when the WGM is lasing. This limits the further improvement of the modulation rate. The optical eye diagrams at modulation rates of 10 Gbps and 25 Gbps at I_DSM = 20 mA and I_FP = 64 mA are shown in Fig. 7(c) and (d), and the ERs are 5.4 and 4 dB, respectively.

 

Fig. 7. The optical eye diagrams at large-signal modulation rates of (a) 10 Gbps and (b) 25 Gbps at I_DSM = 20 mA and I_FP = 43 mA (the main lasing mode is h₀ at 1309.084 nm). The optical eye diagrams at large-signal modulation rates of (c) 10 Gbps and (d) 25 Gbps at I_DSM = 20 mA and I_FP = 64 mA (the main lasing mode is h₀ at 1322.756 nm). The electrical eye diagram after digital signal processing at large-signal modulation rates of (e) 10 Gbps and (f) 25 Gbps at I_DSM = 20 mA and I_FP = 64 mA.

Download Full Size | PDF

As the hybrid mode is the main lasing mode at this time, the small-signal EO modulation response bandwidth is relatively large, which results in a significant reduction of the overshoot in the eye diagram. However, multipath effects appear on both the rising and falling edges in the eye diagram at the modulation rate of 25 Gbps, which may be related to the clock control of the test system. We have also investigated the use of an SFP + (small form-factor pluggable) transceiver combined with a matching testing host board to perform digital signal processing (DSP) on the optical signal to achieve eye diagram reshaping, as shown in Fig. 7(e) and (f), which are corresponding to the test results of the optical eye diagrams in Fig. 7(c) and (d), respectively. We can see that the noise spots are significantly reduced in the eye diagram, thereby enhancing the overall quality of the eye diagram, which is very meaningful for the practical application of the laser.

5. Conclusion

In conclusion, we have proposed and demonstrated a high-speed directly modulated DFC laser that can achieve mode Q factor control, stable single-mode operation, and a high direct modulation bandwidth. The hybrid mode with a strong mode field distribution inside the DSM enables mode Q factor enhancement, as observed in the simulated results. The reflection spectrum of DSM is clean, with excellent comb-like and narrow-peak characteristics, which significantly increases the lasing mode Q factor and the single-mode yield of the device. For the first time in a WGM-FP type coupled cavity semiconductor laser, a small-signal modulation bandwidth of up to 30 GHz based on the PPR effect has been achieved by controlling the WGM and the hybrid mode. The large signal modulation performance can be significantly enhanced by optimizing the design to regulate the mode coupling between the WGMs and FP cavity mode, which allows for better control over the spatial distribution of the mode field. When both hybrid modes are distributed throughout the coupled cavity, the modulation bandwidth can be improved through the PPR effect, even when applying the RF signal to the FP cavity. The laser structure does not involve precise subwavelength grating structures or epitaxial regrowth and has the advantage of a simple fabrication process and a compact structure. Unlike other serial coupled-cavity lasers, there is no distinct boundary between the two cavities in the WGM-FP lasers, which makes the laser more stable and robust. With its high direct modulation bandwidth, simple fabrication process, and compact structure, the laser has great potential for use in optical communication systems and photonic integrated circuits.