**4.1 Carrier lifetime analysis**

10 Selected Topics on Optical Amplifiers in Present Scenario

At first, the increase of the cavity length induces higher optical gain (from 300 µm to 700µm) however when it reaches 850 µm, the gain drops back. Therefore a maximum gain is obtained for 700 µm long devices. The optical gain versus the output power is presented in Fig. 6. (b) at the current density J = 10 kA/cm2. We can notice that increasing the gain leads to higher saturation power. It can be explained by the fact that we are at a constant current density therefore the electrical bias current increases with the length of the device leading to an improvement of the saturation power. For one specific optical confinement (Γ = 20%), an optimal length can be found in order to obtain the best static performances (high optical gain). At first, the optical gain increases linearly with the length. In fact, the forward and backward amplifications control the single pass gain. Figure 7. (a) represents the SE measurements where an optical fibre is placed along the active zone at the input/output, centre and mirror region. Then SE measurements as a function of the injected current are measured. SE measurements are performed in 700 µm long RSOA in order to confirm the

Fig. 7. (a) SE schematic and measurements; LSHB effect on (b) the optical gain in RSOA device At low input bias current, no difference is observed due to the flat carrier density. The saturation effect starts to appear above 50 mA when the carrier density spatial distribution becomes non-homogeneous. Low SE power is collected at the input region due to the saturation effect which means low carrier density in the region. However the mirror region emits more SE power due to the high carrier density value. This demonstrates the presence

Gain (dB)

5

10

15

20

25

 J = 10 KA/cm2 J = 20 KA/cm2

30

300 400 500 600 700 800 900

Length (µm)

**Current**

**Input/output center mirror**

In longer RSOAs, the depletion becomes stronger which induces a lower overall carrier density and a larger absolute difference in the carrier density between input and mirror facet. When varying the length of the RSOA, those several effects account for the existence of an optimum length where the optical gain is maximised. The optical gain versus the

RSOA devices have limited electro-optical (E/O) bandwidth between 1 to 2 GHz (Omella et al., 2008) compared to laser devices usually between 8 to 10 GHz. The difference can be

length of the device is plotted on Figure 7. (b) for two current densities.

**4. Modulation characteristics and performances** 

0 50 100 150 200

Current (mA)

*a) b)*

presence of the saturation effect.

 mirror center input

0,0 5,0x10-6 1,0x10-5 1,5x10-5 2,0x10-5 2,5x10-5 3,0x10-5 3,5x10-5 4,0x10-5 4,5x10-5 5,0x10-5

SE power (a.u.)

of a strong saturation effect in the device.

The objective is to obtain a first order approximation of the carrier lifetime for the steady state condition. We can demonstrate that the carrier lifetime can be approximated by:

$$\frac{1}{\tau\_{eff}} = A + B.n + \mathcal{C}.n^2 + \Gamma \times a \times \mathcal{S} \times v\_g \tag{13}$$

Where the differential gain is defined by ܽ ൌ డ డே, Γ is the optical confinement factor and S is the total photon density including the signal and the ASE.

The carrier lifetime is inversely proportional to the recombination rate. The recombination rate can be described using two different terms: one directly proportional to the spontaneous emission and non-radiative recombination (due to the defect or Auger process as described in section 2.2) and the second one depending on the stimulated recombination.

Fig. 8. Carrier lifetime simulation along 700 µm RSOA device at (a) low (Pin = -40 dBm) and (b) high (Pin = 0 dBm) optical injection

Simulations of the carrier lifetime have been carried out along the active region. Figure 8 represents the results with the bias current as parameter at Pin = -40 dBm (Figure 8 (a)) and Pin = 0 dBm (Figure 8 (b)). Obviously, in both cases, carrier lifetime decreases by increasing

Next Generation of Optical Access Network Based on Reflective-SOA 13

However at low bias current (c), both phenomena balance each other and both are responsible for the carrier lifetime. They are more or less equal and do not vary that much over z. This analysis is crucial for digital modulation as the input conditions change over time, therefore the dynamic of the device will depend on which recombination rate is

In order to validate our simulation, a comparison with experimental measurements should be done. High-frequency characterization is then needed. The experimental set-up and

We realize a RSOA-based microwave fibre-optic link as depicted in figure 10. All different devices of this experimental set up can be considered as two-port components and classified according to the type of signal present at the input and output ports. E/E, E/O, O/E or O/O are possible classifications where an electrical (E) signal or an optical (O) signal power are modulated at microwave frequencies (Iezekiel et al., 2000).The RSOA is considered as an E/O two-port device which is characterized by the electro-optic conversion process, i.e. the

A full two-port optical characterisation of the complete set up is important to quantify the system performances. Dynamic characterization allows the measurement of the electrical response of the two-port network. A high-frequency signal is sent to the RSOA and the optical modulation is detected by a photodiode. The |S21|2 parameter (link gain) is measured over a range of frequency from 0 to 10 GHz. Figure 11 shows the electrical

*E/O device*

*ECL*

*Receiver RSOA*

*Opticalfibre*

*E/O device*

*Microwave modulated electrical signal*

*RSOA drive circuit*

*E/E device*

dominant at a precise time.

results are described in the next section.

*Microwave output*

Fig. 10. High speed fibre-optic link

*E/E device*

*Transimpedance amplifier*

*E/O device*

response of a typical RSOA device.

**4.2 High-frequency experimental set-up and characterization** 

conversion of microwave current to modulated optical power.

the input electrical current. It is mainly due to the increase in all recombination terms. The second important observation is the non-uniformity of the carrier lifetime along the device. At large optical input power (Pin = 0 dBm), the saturation effect described in section 3.2 is much stronger than with low input injection at low bias current. The average carrier lifetime is also smaller in this condition, due to a larger photon density. In order to understand the influence of the different recombination mechanisms on the carrier lifetime, it is important to follow the evolution of the different recombination terms depending on the bias current and the input optical power.

Fig. 9. Spatial distribution of spontaneous and non-radiative recombination rate compared to stimulated recombination rate in 700 µm long RSOA at different input conditions. (a) Pin = -40 dBm and I = 40 mA, (b) Pin = -40 dBm and I = 40 mA, (c) Pin = 0 dBm and I = 40 mA and (d) Pin = 0 dBm and I = 80 mA

Figure 9 represents the spatial distribution of the two terms at various operation conditions. At low input optical power ((a) and (b)), the spontaneous and non-radiative recombination rates are dominant even at high bias current. Therefore the carrier lifetime depends on this recombination term. At high input optical power ((c) and (d)), the photon density is much higher than in the previous situation, thus the stimulated recombination rate tends to overcome the spontaneous and non-radiative recombination terms. This is also confirmed at high input bias current (d) when the signal and ASE are strongly amplified along the RSOA.

the input electrical current. It is mainly due to the increase in all recombination terms. The second important observation is the non-uniformity of the carrier lifetime along the device. At large optical input power (Pin = 0 dBm), the saturation effect described in section 3.2 is much stronger than with low input injection at low bias current. The average carrier lifetime is also smaller in this condition, due to a larger photon density. In order to understand the influence of the different recombination mechanisms on the carrier lifetime, it is important to follow the evolution of the different recombination terms depending on the bias current

Fig. 9. Spatial distribution of spontaneous and non-radiative recombination rate compared to stimulated recombination rate in 700 µm long RSOA at different input conditions. (a) Pin = -40 dBm and I = 40 mA, (b) Pin = -40 dBm and I = 40 mA, (c) Pin = 0 dBm and

Figure 9 represents the spatial distribution of the two terms at various operation conditions. At low input optical power ((a) and (b)), the spontaneous and non-radiative recombination rates are dominant even at high bias current. Therefore the carrier lifetime depends on this recombination term. At high input optical power ((c) and (d)), the photon density is much higher than in the previous situation, thus the stimulated recombination rate tends to overcome the spontaneous and non-radiative recombination terms. This is also confirmed at high input bias current (d) when the signal and ASE are strongly amplified along the RSOA.

I = 40 mA and (d) Pin = 0 dBm and I = 80 mA

and the input optical power.

However at low bias current (c), both phenomena balance each other and both are responsible for the carrier lifetime. They are more or less equal and do not vary that much over z. This analysis is crucial for digital modulation as the input conditions change over time, therefore the dynamic of the device will depend on which recombination rate is dominant at a precise time.

In order to validate our simulation, a comparison with experimental measurements should be done. High-frequency characterization is then needed. The experimental set-up and results are described in the next section.
