High-Speed Directly Modulated Laser Integrated with SOA

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Please see the attached.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

We sincerely thank the reviewer for the constructive comments and valuable suggestions, which have greatly helped us improve the clarity and quality of the manuscript. We have carefully addressed each point, and corresponding revisions have been made in the updated version. The detailed responses are provided below.

Point-by-point response to Comments and Suggestions for Authors:

This manuscript presents a directly modulated laser (DML) design integrated with a semiconductor optical amplifier (SOA) and a partially corrugated grating (PCG), achieving a bandwidth of 25.8 GHz, output power > 25 mW, and a side-mode suppression ratio (SMSR) > 44 dB. The work focuses critical challenges in high-speed optical communication systems, such as bandwidth limitations and output power constraints. Here, publication of this manuscript could be possible after addressing the following concerns.

Comments 1: Mechanistic Explanation of SOA-DML Interaction

While the quasi-high-pass filtering effect of the SOA is mentioned, the manuscript lacks a detailed theoretical analysis of carrier dynamics and gain saturation in the SOA-DML system. A more rigorous discussion of how the SOA’s carrier lifetime (τ0) and saturation power (Pinsat) influence bandwidth enhancement would strengthen the work. Additionally, the trade-off between SOA-induced chirp and SNR improvement warrants deeper exploration.

Response 1:  

We appreciate the reviewer’s insightful comments. In the revised manuscript, we have expanded the theoretical discussion on the SOA’s dynamic behavior. A frequency response model of the SOA operating in the gain saturation regime has been introduced (Eq. (2) and (3)), and the physical origin of the quasi-high-pass filtering effect has been clarified. Specifically, we explain how the carrier lifetime  and saturation power P_insat affect the differential gain response for different frequency components of the modulated input signal. These discussions are provided in Lines 54–80.

Experimentally, we show that our SOA, due to its short cavity length and strong optical confinement, enters saturation easily (Line 141, Fig. 2(b)). Static gain compression is evident even under moderate modulation, confirming a low P_insat. This behavior supports the conditions required for quasi-high-pass filtering, which contributes to bandwidth enhancement by compensating the high-frequency roll-off of the DFB section. This mechanism and its effect on the composite -3 dB bandwidth are further discussed in Lines 188–202.

Regarding chirp, we acknowledge the theoretical possibility of SOA-induced phase variation. However, as shown in Fig. 6, increasing the SOA bias leads to a consistent decrease in TDEC and clearer eye openings, indicating improved signal quality. These results suggest that, under our operating conditions, the positive effects of power and SNR enhancement outweigh any negative impact from chirp. The relevant discussion is added in Lines 240–248.

Comments 2: Comparative Analysis

The manuscript cites recent works (e.g., different bandwidth DMLs [Refs. 4,6,12]) but does not directly compare key metrics (e.g., bandwidth, output power, SMSR) with similar SOA-integrated DMLs. A table summarizing performance benchmarks would contextualize the advancements claimed here.

Response 2:

We appreciate the reviewer’s suggestion. In the revised manuscript, we have added a new performance comparison table (Table 1) to summarize the key metrics—modulation bandwidth, output power, and side-mode suppression ratio (SMSR)—of representative SOA-integrated DMLs, including our work and recent publications such as Refs. [12] and [19].

Our work focuses on exploiting the quasi-high-pass filtering effect of the SOA to enhance modulation bandwidth, while also leveraging its amplification capability to achieve high output power under relatively low bias currents. In contrast, many recent studies have primarily focused on bandwidth enhancement using Photon–Photon Resonance (PPR) and Detuned Loading (DL) effects, without integrating SOA in a dynamically active role. Therefore, comparable works featuring SOA-induced bandwidth shaping are relatively scarce.

For example, in Ref. [12], a two- DBR laser integrates an SOA that operates in the linear gain regime to boost optical power while avoiding nonlinear distortion. The SOA in that work does not exhibit frequency-selective gain response and is not employed for bandwidth shaping. The modulation bandwidth exceeds 50 GHz due to the cavity design and DL effect, but the maximum output power is only 14.5 mW.

In comparison, our design features a monolithically integrated DFB-SOA structure with independently biased regions. The SOA operates in the gain saturation regime, exhibiting a frequency-dependent dynamic gain response. This response acts as a quasi-high-pass filter, effectively compensating for the DFB’s strong low-frequency modulation and high-frequency roll-off, thereby increasing the overall -3 dB modulation bandwidth. Furthermore, SOA-induced power enhancement improves the signal-to-noise ratio (SNR) at the receiver, resulting in a lower TDEC and clearer eye diagrams, as shown in Fig. 6.

In Ref. [19], a mutually coupled dual-DFB structure is presented, with the SOA again operating in saturation and independently biased. The bandwidth enhancement is attributed to mutual optical injection between the two DFBs, triggering strong PPR effects and achieving a bandwidth of 38.7 GHz. The SOA in that design serves to strengthen the mutual coupling and introduces nonlinear spectral shaping. In contrast, our design achieves comparable performance improvements in a more compact and simplified DFB-SOA layout by relying on SOA’s quasi-high-pass filtering rather than mutual injection, thus improving both modulation bandwidth and output power without added structural complexity.

We believe that this updated comparison clarifies the uniqueness and advantages of our approach and contextualizes our results within the current state of the art.

Comments 3: Fabrication Details

(1) It seems that you have practically completed the device fabrication. If so, it is necessary to introduce more details about the fabrication process flow in the section “Device Structure and Fabrication”, and show the images (at least the top view image) of device structure under electron microscope.
(2) Critical fabrication parameters, such as the active layer material (e.g., InGaAsP/InP), grating coupling coefficient (κ = 7788 m-1) optimization, and coating techniques, are briefly mentioned but not elaborated. Including cross-sectional view of fabricated device would aid reproducibility.

Response 3:

Thank you for your suggestion. We confirm that the DFB-SOA integrated laser was fully fabricated and characterized. To address your comments, we have revised the section “Device Structure and Fabrication” (Lines 106–125) with a more comprehensive process description. This includes the epitaxial growth of an InGaAsP/InP MQW structure for 1310 nm emission, electron beam lithography and shallow etching for defining partial gratings, dry etching of 2.4 μm-wide ridge waveguides, wet etching of a 10 μm-wide and 200 nm-deep isolation trench, metal electrode patterning using the lift-off, and SiO₂ passivation with contact pad definition.

Due to the SiO₂ passivation on the device surface, high-resolution SEM imaging was not feasible. However, a top-view optical microscope image of the fabricated device has been added as Fig. 1(b) to aid visual understanding. Additionally, we elaborated on several key parameters: the grating coupling coefficient was designed as κ = 7788 m⁻¹, optimized through simulation to balance feedback and bandwidth. The device facets were coated with anti-reflection (AR) and high-reflection (HR) films using vacuum evaporation to suppress reflection-induced instability and enhance power efficiency. A schematic cross-sectional diagram is also provided in Fig. 1(a) for reproducibility.

 

Comments 4: Other Minor Issues

(1) Figure references in the text (e.g., Fig. 6(c), rather than Fig. 6) do not fully align with the described content.
(2) The title of the section should be "Device Structure and Fabrication".
(3) The significance of TDEC for system performance could be briefly explained for broader readability.

Response 4：

We appreciate the reviewer’s attention to these minor but important issues. We have carefully reviewed all figure references in the manuscript and revised them to ensure consistency and accuracy in formatting and content alignment (e.g., Fig. 6(c) now correctly corresponds to the eye diagram content). The title of Section 2 has also been corrected to “Device Structure and Fabrication” as suggested.

Additionally, in response to the reviewer’s suggestion, we have briefly introduced the significance of TDEC in the context of system performance(Lines 218–223). Specifically, TDEC is a standardized metric that reflects the signal quality of the transmitter after considering dispersion and filtering effects. A lower TDEC value indicates a more open eye diagram and stronger tolerance to transmission impairments, thus directly correlating with improved system performance. This explanation has been added in the revised manuscript to aid broader readability.

We are grateful for the reviewer’s thoughtful comments and suggestions, which helped us improve the clarity, rigor, and completeness of our manuscript. We hope the revised version adequately addresses all concerns.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The submitted manuscript is focused on study of ways to increase the speed of a semiconductor optical amplifier (SOA) during its direct modulation. I can provide the following considerations on this subject:

1. The developed SOA achieved a record bandwidth of 25.8 GHz. It is necessary to indicate how this achieved parameter compares to other SOAs.
2. Increasing the SOA speed may imply generation of shorter laser pulses in laser configurations where the SOA serves as the active medium and gain may be modulated. Some estimates regarding the pulse durations that may be generated in lasers where the new SOA could act as the active medium with direct gain modulation should be provided.
3. It is necessary to specify the power consumption of the SOA at the maximum modulation frequency and full modulation amplitude. Is cooling required for the SOA in this operating mode? What can be said about the lifetime of the SOA operating continuously in this mode?
4. The proposed directly modulated laser (DML) essentially consists of two independent parts: the SOA and the DFB. The DML without the SOA demonstrates decent characteristics: they are slightly worse than with the SOA, but may be considered satisfactory. In this regard, the question arises: can the high-speed module (DFB) be used separately in series with the SOA? That is, can it be used as a separate device with existing SOAs? This needs some discussion.

If the Authors address the above comments in a further revision of their manuscript, it may be published in Photonics.

Comments on the Quality of English Language

n/a

Author Response

We sincerely thank the reviewer for the constructive comments and valuable suggestions, which have greatly helped us improve the clarity and quality of the manuscript. We have carefully addressed each point, and corresponding revisions have been made in the updated version. The detailed responses are provided below.

Point-by-point response to Comments and Suggestions for Authors:

The submitted manuscript is focused on study of ways to increase the speed of a semiconductor optical amplifier (SOA) during its direct modulation. I can provide the following considerations on this subject:

Comments 1: The developed SOA achieved a record bandwidth of 25.8 GHz. It is necessary to indicate how this achieved parameter compares to other SOAs.

Response 1:  

We thank the reviewer for the detailed and valuable comments. Before addressing the specific points, we would like to clarify a potential misunderstanding. In our work, the SOA is not directly modulated, nor does it serve as the main gain medium of the laser. The electrical modulation is applied solely to the DFB section, while the SOA operates under constant bias as a post-amplifier. The core objective of our study is to investigate the frequency-selective gain response of the SOA under gain saturation conditions—referred to as a quasi-high-pass filtering effect—which reshapes the overall frequency response and enhances the modulation bandwidth of the integrated system.

The reported 25.8 GHz bandwidth refers to the modulation bandwidth of the monolithically integrated DFB-SOA laser, not the intrinsic modulation speed of the SOA itself. The frequency response discussed in our work pertains to the dynamic gain behavior of the SOA when it amplifies intensity-modulated optical signals from the DFB section, rather than its own direct modulation capability.

As detailed in Lines 54–59, we have added a theoretical expression describing the frequency-dependent gain of the SOA and explained how the interplay between carrier depletion and replenishment under saturation leads to stronger gain at high frequencies and compression at low frequencies. This produces a quasi-high-pass filtering behavior. In our device, the SOA has a short cavity and strong optical confinement, which leads to early saturation, as shown in Fig. 2(b). Under moderate DFB modulation, clear gain compression is observed, consistent with a low saturation power (P_insat), enabling the high-pass filtering effect to emerge under practical bias conditions. This behavior aligns with dynamic saturation models reported in Refs. [15–17].

Through this filtering effect, the SOA reshapes the DFB's frequency response, mitigating low-frequency dominance and suppressing high-frequency roll-off, thus extending the $-3$ dB bandwidth from 18.5 GHz (without SOA) to 25.8 GHz (with SOA). To our knowledge, this is the first demonstration of such SOA-enabled bandwidth shaping without relying on external feedback structures.

 

Comments 2: Increasing the SOA speed may imply generation of shorter laser pulses in laser configurations where the SOA serves as the active medium and gain may be modulated. Some estimates regarding the pulse durations that may be generated in lasers where the new SOA could act as the active medium with direct gain modulation should be provided.

Response 2:

We appreciate this thoughtful perspective. However, as clarified above, our work does not involve direct modulation of the SOA nor pulsed laser generation. The laser operates under continuous wave (CW) biasing, with modulation applied only to the DFB section. The SOA serves solely as a gain-shaping amplifier. Therefore, pulse generation and associated duration estimates fall outside the scope of this study.

 

Comments 3: It is necessary to specify the power consumption of the SOA at the maximum modulation frequency and full modulation amplitude. Is cooling required for the SOA in this operating mode? What can be said about the lifetime of the SOA operating continuously in this mode?

Response 3:

Thank you for raising this practical and important point. At the operating condition where the SOA bias current is 30 mA and the voltage is 1.228 V, the measured power consumption is approximately 36.84  mW (Line 140 of the revised manuscript). During all experiments, a thermoelectric cooler (TEC) was employed to maintain the chip temperature at 25 °C, ensuring thermal stability and measurement repeatability.

While long-term lifetime testing was beyond the scope of this work, we note that the SOA operates in a quasi-CW regime with modest electrical drive levels and without pulsed or high-amplitude modulation. Under such moderate conditions, the lifetime is expected to be consistent with that of conventional InP-based SOAs.

 

Comments 4: The proposed directly modulated laser (DML) essentially consists of two independent parts: the SOA and the DFB. The DML without the SOA demonstrates decent characteristics. Can the high-speed DFB module be used separately in series with existing SOAs? This needs some discussion.

Response 4：

We appreciate this important observation. Indeed, our device features a monolithically integrated DFB and SOA using the same epitaxial active layer, enabling structural continuity and eliminating regrowth. In principle, the DFB section could be operated independently and coupled to an external SOA module. However, this would require additional packaging efforts and mode-matching design to ensure efficient coupling.

It is also important to distinguish functional roles: commercial SOA modules typically operate in the linear gain regime for power amplification, while our integrated SOA is intentionally biased into the gain saturation regime to achieve frequency-selective shaping of the DFB output via quasi-high-pass filtering. This dynamic interaction enables both bandwidth enhancement and power amplification in a tightly integrated structure.

Thus, while external SOA coupling is feasible in theory, our monolithic integration strategy provides clear advantages in both performance synergy and packaging simplicity. It demonstrates that SOA-based bandwidth shaping can be achieved without introducing structural complexity or external optical feedback mechanisms.

We are grateful for the reviewer’s thoughtful comments and suggestions, which helped us improve the clarity, rigor, and completeness of our manuscript. We hope the revised version adequately addresses all concerns.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript presents research results on a directly modulated laser (DML) using a partially corrugated grating (PCG) and integrated with a semiconductor optical amplifier (SOA). The findings indicate a relatively high laser output power exceeding 25 mW, achieved with the DML laser at a modulated bandwidth of approximately 25 GHz. This report will be beneficial for readers in this technical field. After minor improvements, I recommend that this manuscript be accepted for publication in a technical journal. Detailed technical comments are provided below:

 

- Section 2 discusses the device structure without specifying the exact device dimensions and parameters. Additional comments are necessary if the device’s details adhere to the same conditions as reference 18. The reference 18 paper describes the numerically simulated device performance of a direct modulation laser with integrated active feedback through partially corrugated gratings. Yet, it does not specify the detailed composition of the active region. The title of Section 2 includes the word "fabrication" but lacks a description of how the device was fabricated. In Section 4 Conclusions, the second sentence mentions "fabricated" and "manufacturing”. The authors should provide a comprehensive description of the fabricated devices.

- In Fig. 2(b), the gain peaks appear at approximately 6 to 7 mA modulation current. I am curious about DML’s threshold current and whether the gain peaks occur in the lasing mode rather than in the incoherent emission mode (i.e., amplified spontaneous emission, ASE) below the threshold condition. The authors may need to examine the output spectrum of the DML+SOA emission to determine the threshold condition.

- Figures 3 and 4 illustrate the frequency response of the fabricated DML and SOA devices. No information about the laser wavelength corresponding to these frequency response characteristics is provided. The text following Figure 4 indicates that the DMLs without the SOA region were fabricated. Was the SOA device without the DML not fabricated to test its frequency response before assessing the combined frequency response of the DML and SOA?

End

Author Response

We sincerely thank the reviewer for the constructive comments and valuable suggestions, which have greatly helped us improve the clarity and quality of the manuscript. We have carefully addressed each point, and corresponding revisions have been made in the updated version. The detailed responses are provided below.

Point-by-point response to Comments and Suggestions for Authors:

This manuscript presents research results on a directly modulated laser (DML) using a partially corrugated grating (PCG) and integrated with a semiconductor optical amplifier (SOA). The findings indicate a relatively high laser output power exceeding 25 mW, achieved with the DML laser at a modulated bandwidth of approximately 25 GHz. This report will be beneficial for readers in this technical field. After minor improvements, I recommend that this manuscript be accepted for publication in a technical journal. Detailed technical comments are provided below:

Comments 1: Section 2 discusses the device structure without specifying the exact device dimensions and parameters. Additional comments are necessary if the device’s details adhere to the same conditions as reference 18. The reference 18 paper describes the numerically simulated device performance of a direct modulation laser with integrated active feedback through partially corrugated gratings. Yet, it does not specify the detailed composition of the active region. The title of Section 2 includes the word "fabrication" but lacks a description of how the device was fabricated. In Section 4 Conclusions, the second sentence mentions "fabricated" and "manufacturing”. The authors should provide a comprehensive description of the fabricated devices.

Response 1:  

We thank the reviewer for this valuable suggestion. In the revised manuscript (Lines 106–125), we have included detailed structural parameters of the device, such as the ridge waveguide width (2.4 μm), the isolation trench dimensions (10 μm wide and 200 nm deep), and the grating coupling coefficient (κ = 7788 m⁻¹). A full description of the fabrication process has also been added to Section 2, including epitaxial growth of the InGaAsP/InP MQW structure, grating definition via electron beam lithography and shallow etching, dry/wet etching for ridge and trench formation, metal electrode deposition, and front/rear facet coatings.

Regarding Reference [18], we adopted the partially corrugated grating concept as a structural basis. However, our device differs from Ref. [18] in several key aspects, including the bandwidth enhancement mechanism, grating length and coupling coefficient, electrical isolation structure, and the transmittance of the output facet coatings. These differences are now clarified in the revised text.

 

Comments 2: In Fig. 2(b), the gain peaks appear at approximately 6 to 7 mA modulation current. I am curious about DML’s threshold current and whether the gain peaks occur in the lasing mode rather than in the incoherent emission mode (i.e., amplified spontaneous emission, ASE) below the threshold condition. The authors may need to examine the output spectrum of the DML+SOA emission to determine the threshold condition.

Response 2:

We appreciate the reviewer’s observation. The threshold current of the DML is approximately 5 mA, as shown in the Fig. 2(a). The gain peaks shown in the Fig. 2(b) occur at modulation currents of 6–7 mA, which are above threshold. Therefore, the observed gain behavior corresponds to net gain under lasing conditions, rather than to amplified spontaneous emission (ASE) below threshold.

 

Comments 3: Figures 3 and 4 illustrate the frequency response of the fabricated DML and SOA devices. No information about the laser wavelength corresponding to these frequency response characteristics is provided. The text following Figure 4 indicates that the DMLs without the SOA region were fabricated. Was the SOA device without the DML not fabricated to test its frequency response before assessing the combined frequency response of the DML and SOA?

Response 3:

We thank the reviewer for this important clarification. The frequency response measurements shown in Figures 3 and 4 were conducted at the device’s operating wavelength of 1310 nm, determined by the InGaAsP/InP MQW epitaxial design. This information has been explicitly added in Line 109 of Section 2.

Regarding whether a standalone SOA was fabricated: in this work, we did not fabricate an SOA-only device. The term “SOA frequency response” in our manuscript does not refer to the SOA's direct modulation capability, but rather to its dynamic gain response under gain saturation, when amplifying intensity-modulated signals from the DFB section. We have clarified this definition in Lines 46–50 to avoid potential misunderstandings.

Our research focuses on a monolithically integrated DFB-SOA structure. To evaluate the SOA’s contribution to system response, we compare the bandwidth performance of the same device under two conditions: with SOA current off (I_S=0) and on (I_S>0), as shown in Figure 4. Additionally, a standalone DFB laser without SOA integration was fabricated as a control. The measured bandwidth of the DFB-only device is 18.5 GHz, while the DFB-SOA integrated device achieves 25.8 GHz, indicating an approximate 7 GHz improvement attributable to the SOA’s quasi-high-pass filtering effect.

We are grateful for the reviewer’s thoughtful comments and suggestions, which helped us improve the clarity, rigor, and completeness of our manuscript. We hope the revised version adequately addresses all concerns.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

In response to my observations, important information was added to the manuscript that made it

more interesting and comprehensible. My comments have been fully addressed by the Authors in

the revised manuscript, which may be now published.

Comments on the Quality of English Language

n/a