1. Introduction

The wavelength division multiplexed-passive optical network (WDM-PON) is considered as one of the solutions for future access networks because of its large transmission capacity, network security, and data transparency. The reflective semiconductor optical amplifier (RSOA) used in a WDM-PON has attracted much attention due to its simple structure, low-cost, and flexibility of use. The gain, saturation characteristics, and noise figure (NF) of an RSOA play important roles in an RSOA-based WDM-PON^[1-3]. When an RSOA is employed in the optical network unit of a WDM-PON, it is required to become saturated easily to erase the incident downstream data and have large gain to amplify the upstream data^[1]; and in order to suppress the extinction ratio of downstream optical signal, and decrease noise and nonlinear spectral distortion, the RSOA is required to work in a deep saturation regime^[2]. In an RSOA-based WDM/SCM (subcarrier multiplexing) PON, the gain-saturated RSOA used in the optical line terminal can reduce optical beat interference noise^[3]. Therefore, it is necessary to develop an RSOA with variable gain and saturation power, and low NF.

The multi-section method is used for quantum dot semiconductor super-luminescent diodes for broad bandwidth and in laser diodes for optical bistability. In addition, multi-section traveling-wave semiconductor optical amplifiers have also been reported in recent years^[4-7]. It can enhance gain recovery rate and cross-gain modulation bandwidth^[4], reduce pattern effects in wavelength conversion^[5], control gain and amplified spontaneous emission (ASE) spectrum^[6], and equalize optical spectrum^[7]. For a multi-section RSOA, Kim et al. proved it improved small-signal modulation bandwidth as a colorless light source in a WDM-PON^[8]. Valicourt et al. reported that a multi-section RSOA reduced chirp when it was used as remote modulator in a PON^[9] and enhanced reception responsivity and bandwidth when used as in-line photodetector^[10]. However, there is no report on manipulating the gain, saturation power and low NF of a two-section multi-quantum well reflective semiconductor optical amplifier (MQW-RSOA) by adjusting the injection current density of the two sections.

In this paper, we design and fabricate a 1.3 $\mu$m two-section MQW-RSOA. The gain and noise characteristics at different current density ratios of the two sections are investigated experimentally and theoretically; and the effect of reflectivity is studied theoretically. The rest of this paper is organized as follows. Section 2 presents the experimental setup and a brief introduction to the model of the two-section MQW-RSOA. The corresponding experimental and theoretical results are described in Section 3. Finally, a brief conclusion is given in Section 4.

2. Experimental setup and theoretical model

The fabricated two-section RSOA is a 1.3 $\mu$m MQW SOA, whose schematic diagram is shown in Fig. 1. The MQW active region consists of 3 tensile-strain InGaAsP wells and 4 compressively strained InGaAsP wells, which make the gain polarization insensitive. The RSOA has a ridge waveguide structure with stripe width $W$ of 2 $\mu$m. The material epitaxy of the RSOA is performed by metal-organic chemical vapor deposition on an n-InP substrate. Wet-etching and plasma-enhanced chemical vapour deposition is used to form a 20 $\mu$m SiO$_{2}$ insulated channel, which separates the RSOA into two sections. The two sections share the same MQW active region and other internal structures. The total length of the active region $L$ is 1200 $\mu$m and the length ratio of the two sections, $L_{1}$/$L_{2}$, is 8/2, in which $L_{1}$ is the length of the section near the input facet, and $L_{2}$ is that of the section near the reflective facet. Two separate current sources $I_{1}$ and $I_{2}$ are used to independently drive each section of the MQW-RSOA. The current density ratio is $J_{1}$/$J_{2}$, where $J_{1}$ $=$ $I_{1}$/$(WL_{1})$ and $J_{2}$ $=$ $I_{2}$/$(WL_{2})$. The reflectivity of the input facet $R_1$ is about 1 $\times$ 10$^{-5}$ and the reflectivity of the reflective facet $R_{2}$ is 4.8 $\times$ 10$^{-3}$ or so.

Figure 2 illustrates the schematic diagram of the experimental setup for gain and NF measurements. The wavelength of the continuous wave (CW) input signal is 1330 nm. The total injection current into the RSOA is 350 mA. The variable optical attenuator can continuously regulate the input power. The CW light is coupled into the MQW-RSOA via an optical circulator and the reflected optical signal from the reflective facet of the RSOA through the optical circulator is sent into a Yokogama AQ6370 optical spectrum analyzer (OSA). The input optical power is measured by an optical power meter and the output signal from the RSOA is measured by the OSA. The MQW-RSOA device is held at a constant temperature of 20 ℃ .

In the simulation, a segmentation model is used to account for the longitudinal distribution of the carrier density in the active region. The discretization method is employed to describe the wide spectrum characteristics of the ASE. For the material gain coefficient of the quantum-well structure, gain compression factors^[11] with contributions from spectral hole burning and carrier heating are considered; and a logarithmic dependence on carrier density is used^[12]. NF is calculated using the ASE power spectral density and the gain at the signal frequency^[13-15].

3. Results and discussion

In this section, the corresponding experimental and theoretical results are discussed. The influences of current density ratio $J_{1}/J_{2}$ and the reflectivity of the reflective facet $R_{2}$ on the gain, saturation power, and NF are studied. The current density ratio between the two sections, $J_{1}/J_{2}$, is chosen to scale the nonuniformity of injection current density distribution. $J_{1}/J_{2}$ < 1 means that the current density near the input facet is smaller than that near the reflective facet; while $J_{1}/J_{2}$ $>$ 1 indicates the current density near the input facet is larger; and $J_{1}/J_{2}$ $=$ 1 describes the uniform distribution of current density in the two sections.

3.1 Influence of current density ratio

3.1.1   Influence of current density ratio on gain and noise figure

Figures 3(a) and 3(b) give the measured and calculated results of gain and NF as a function of $J_{1}/J_{2}$ at input power of $-30$ dBm and $-6$ dBm. It can be found that with $J_{1}/J_{2}$ increasing, the gain increases firstly and then decreases, whereas the NF shows a contrary tendency with the gain. For smaller input power, the gain has the maximum value and the NF has the minimum value when $J_{1}/J_{2}$ $=$ 1 (current density uniform distribution in the two sections). However, for bigger input power, the maximum gain and the minimum NF will be achieved when $J_{1}/J_{2}$ < 1. With the input optical power increasing, it depletes more carriers, which speeds up the depleting rate of carriers along the propagation direction. In order to keep the amplification of photons, the carrier density near the reflective facet needs to be increased and the corresponding current density also needs to be enhanced. As a result, the maximum gain is obtained when $J_{2}$ is larger than $J_{1}$, i.e., $J_{1}/J_{2}$ < 1. According to Ref. [13], NF has an inverse relationship with gain, therefore the minimum NF occurs when gain is at its maximum. Figures 3(a) and 3(b) also show that gain decreases and the NF increases with input power increasing. This is because large input power makes RSOA saturate, thus causing the reduction of gain and the increase of NF. It can also be seen that the measurement results deviate from the simulation results to a larger extents when the input power and $J_{1}$/$J_{2}$ are larger, this is because the heating effect decreases gain and increases NF.

Figure 4 presents the results of input power versus output power at $J_{1}/J_{2}$ of, 0.6, 1 and 2.5. It clearly shows that when input power is smaller, the gain has the maximum value at $J_{1}/J_{2}$ of 1; and with the input power increasing, the maximum gain shifts to $J_{1}/J_{2}$ of 0.6.

3.1.2   Influence of current density ratio on saturation power

Figure 5 presents the saturation input power and saturation output power versus $J_{1}$/$J_{2}$. The saturation power is the optical power at which the gain drops by 3 dB with respect to its small-signal value. The measured and calculated results show that saturation input power has a minimum value and saturation output power has a maximum value when $J_{1}/J_{2}$ $=$ 1. This is because when current density is uniformly distributed, the RSOA gets saturated more easily. Then the saturation input power is at a minimum. And lightwave with smaller input power has larger gain, as a result, the RSOA gets maximum saturation output power when $J_{1}/J_{2}$ $=$ 1. Compared with common an MQW-RSOA ($J_{1}/J_{2}$ $=$ 1), the two-section MQW-RSOA displays variable saturation power when changing $J_{1}/J_{2}$; various saturation input power varying from –14.0 to –8.0 dBm and saturation output power varying from 7.5 to 12.0 dBm are measured.

3.2 Influence of reflectivity

In the simulations of Section 3.1, the reflectivity of the input facet and that of the reflective facet (i.e, $R_{1}$ and $R_{2})$ are 1 $\times$ 10$^{-5}$ and 4.8 $\times$ 10$^{-3}$, respectively, which are set to have the same value as the corresponding reflectivity of the fabricate RSOA. In this section, the effect $R_{2}$ on the gain, saturation power and NF are studied theoretically.

3.2.1   Influence of reflectivity on gain and noise figure

Figures 6(a) and 6(b) show the calculated gain and NF versus $R_{2}$ at different $J_{1}$/$J_{2}$. The input power of the continuous light is –30 dBm and the bias current is 350 mA. It can be seen from Fig. 6(a) that with the reflectivity $R_{2}$ increasing, the gain first increases quickly and then keeps nearly constant due to gain saturation. In Fig. 6(b), with $R_{2}$ increasing, the NF decreases rapidly due to the large reflection power suppressing ASE, and then it increases as a result of gain saturation. It can be seen that the gain is relatively larger and NF is smaller when $R_{2}$ is around 10%–40%. So the reflectivity of the two-section MQW-RSOA can be optimized in order to make the gain and NF simultaneously have better characteristics.

3.2.2   Influence of reflectivity on saturation power

Figures 7(a) and 7(b) give the calculated saturation power versus $R_{2}$ at different $J_{1}/J_{2}$. The bias current is 350 mA. The results in Fig. 7(a) indicate that the saturation input power first decreases quickly until $R_{2}$ is larger than 40%, and then the saturation input power decreases slowly. This is because a small increase of reflectivity results in large increase of reflective optical power, which makes the RSOA saturate more easily, so the saturation input power decreases quickly; and when $R_{2}$ is larger than 40%, the reflective optical power is large enough to make RSOA work in a deep saturation regime, therefore the saturation input power decreases slowly. And because small input power has large gain while large input power has small gain, the saturation output power in Fig. 7(b) shows a contrary tendency with the saturation input power. The results show that we can choose higher reflectivity to obtain larger saturation output power.

4. Conclusions

A 1.3 $\mu$m two-section MQW-RSOA with variable gain and saturation power is demonstrated. The measured input saturation power varying from –14.0 to –8.0 dBm and output saturation power varying from 7.5 to 12.0 dBm are obtained by changing $J_{1}/J_{2}$. Furthermore, it is found that gain has the maximum value and the NF has the minimum value at $J_{1}/J_{2}$ of 1 when input power is smaller. While when input power is bigger, the maximum gain and the minimum NF are obtained at $J_{1}/J_{2}$ lower than 1. Calculation results show that better gain and noise characteristics can be achieved when $R_{2}$ is around 10%–40%.