WO2022174927A1

WO2022174927A1 - Optimized semiconductor optical amplifier

Info

Publication number: WO2022174927A1
Application number: PCT/EP2021/054270
Authority: WO
Inventors: Romain Brenot; Antonin GALLET
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2021-02-22
Filing date: 2021-02-22
Publication date: 2022-08-25
Also published as: EP4233138A1

Abstract

Described is a semiconductor optical amplifier (600, 900) comprising an active layer extending between an input (601, 901) and an output (602, 902) of the optical amplifier, the active layer having a noise suppression part (603, 903) operative to reduce noise of the optical amplifier and a second part (604, 904), the noise suppression part (603, 903) being adjacent to the input (601, 901) of the optical amplifier and having an epitaxial structure different to the second part (604, 904), the noise suppression part (603, 903) being configured to, during operation of the optical amplifier, present a gain that is blue-shifted by at least 20 nm compared to the second part (604, 904). This may allow the noise of the optical amplifier to be reduced, which may improve the performance of the device.

Description

OPTIMIZED SEMICONDUCTOR OPTICAL AMPLIFIER

FIELD OF THE INVENTION

This invention relates to optical amplifiers, in particular to reducing the noise and increasing the saturation power in such amplifiers.

BACKGROUND

Semiconductor Optical Amplifiers (SOAs) have been studied for many years, but have generally not been considered as a viable alternative to Erbium Doped Fibre Amplifiers (EDFAs). In particular, their Noise Factor (NF) and saturation power (P_sat) have generally not been good enough for many practical applications.

In an SOA, the 3 dB saturation output power is determined by the following relationship:

Psat “ (1)

These parameters are schematically illustrated in Figure 1 , showing a cross-sectional view along the ridge waveguide 101 of an SOA. w is the SOA ridge width of the waveguide 101 , d is the active region thickness and r the quantum well confinement, i.e. the overlap between the optical mode 102 and the active structure 103 of the SOA.

Several research groups have reduced the optical confinement in SOAs with the aim of improving P_sat. Figure 2 shows a summary of some previously achieved results. Each point corresponds to a research result or a product. The target P_sat value is 25 dBm, which corresponds to a current of 2 A.

However, there are several drawbacks to this option. Firstly, these large P_sat SOAs only amplify one polarization state of the light. To achieve a polarization independent SOA, this requires two SOAs in a polarization diversity scheme, which implies a current of 4 A. Reducing the power consumption of each SOA is therefore crucial to achieve reasonable power consumption of the SOA module in such a scheme.

In such cases, the chip length has to be very large, for example at least 5 mm. Otherwise, the gain can drop below 20 dB, which is the minimum gain value required for transmission. Having a long SOA length also increases the cost of the SOA module significantly. This is expected from theory and can be experimentally confirmed by the plots of device length and gain as a function of P_sat shown in Figures 3(a) and 3(b) respectively.

Thus, there is a trade-off between P_sat on one side, and power consumption, gain, footprint and cost on the other side.

The trade-off between P_sat and gain is described in P. Juodawlkis and J. Plant, ‘Gain-Power T rade-Off in Low-Confinement Semiconductor Optical Amplifiers’, in NUSOD 2007. The same team (J. Klamkin et al, ‘High-Output Saturation Power Variable Confinement Slab-Coupled Optical Waveguide Amplifier’ in OFC’11 , paper JThA25) has proposed a scheme to increase P_sat by narrowing the active stripe at the output of the SOA, as illustrated in Figure 4. In this case, confinement decreases when the ridge width decreases. Therefore, the ratio between volume and confinement does not vary significantly.

Thus, the main drawbacks of standard SOAs are their low P_sat and large NF compared to EDFAs. Previous research has demonstrated that a NF value of <6dB and a P_sat value of >20dBm are achievable in an SOA, but at the expense of a very large power consumption. Since SOAs generally also need to be cooled down, this implies an even greater power consumption increase, especially in high temperature cases.

Therefore, a large P_sat implies long SOA devices and a large current, leading to large power consumption, and still the gain decreases for large P_sat.

It is desirable to develop an optical amplifier which overcomes such problems.

SUMMARY OF THE INVENTION

According to one aspect there is provided a semiconductor optical amplifier comprising an active layer extending between an input and an output of the optical amplifier, the active layer having a noise suppression part operative to reduce noise of the optical amplifier and a second part, the noise suppression part being adjacent to the input of the optical amplifier and having an epitaxial structure different to the second part, the noise suppression part being configured to, during operation of the optical amplifier, present a gain that is blue-shifted by at least 20 nm compared to the second part. This may allow the noise of the optical amplifier to be reduced, which may improve the performance of the device.

Preferably, the noise suppression part is operative to reduce the noise factor and/or noise figure of the optical amplifier, as described herein.

In some embodiments, the noise suppression part may have a material composition different to the second part. This may allow the noise of the optical amplifier to be reduced.

The noise suppression part may be composed of a material with larger band-gap energy than the second part. Optionally, the noise suppression part may be composed of a material with larger separation energy between quasi-Fermi levels than the second part.

The second part may be adjacent to the noise suppression part. The second part may be spaced from the input of the optical amplifier. This may allow the region at the input of the optical amplifier to be conveniently modified.

The noise suppression part may be configured to present a gain that is blue-shifted by at least 35 nm compared to the second part. The noise suppression part may be configured to present a gain that is blue-shifted by at least 50 nm compared to the second part. Achieving a gain that is blue-shifted by a larger amount may result in improved performance.

The noise suppression part may be configured so as to, during operation of the optical amplifier, reduce the inversion factor at the input of the optical amplifier. The noise figure of an optical amplifier is proportional to the invention factor, n_sp. Improved inversion corresponds to a smaller n_sp (1 being the smallest possible value). Thus, reducing the inversion factor at the input can reduce the noise in the optical amplifier.

During operation of the optical amplifier, the noise figure of the optical amplifier may be less than 6 dB in the first 20 nm of the operating band. This may allow the performance of such a device to be improved in the low-wavelength region of the optical amplifier.

The optical amplifier may have a central region remote from the output of the optical amplifier comprising a cladding layer configured to increase the optical confinement in the central region and promote concentration of the optical mode in the central region so as to increase the gain per unit length of the optical amplifier. This may additionally allow for the saturation power of the optical amplifier to be increased without requiring a long chip length or very large current.

According to another aspect there is provided a semiconductor optical amplifier comprising an active layer extending between an input and an output of the optical amplifier, the optical amplifier having a central region remote from the output of the optical amplifier comprising a cladding layer configured to increase the optical confinement in the central region and promote concentration of the optical mode in the central region so as to increase the gain per unit length of the optical amplifier. This may allow for the saturation power of the optical amplifier to also be increased without requiring a long chip length or high current.

The active layer may comprise a multiple quantum well layer, wherein the cladding layer is located above the multiple quantum well layer. This may be convenient for constructing the device during manufacturing.

The central region may have a smaller volume per unit length than regions of the optical amplifier remote from the central region. This may result in increased gain per unit current of the central region.

The central region may be configured so as to adapt the gain of the optical amplifier independently of the saturation power of the optical amplifier.

The optical confinement at the output may be lower than the optical confinement in the central region. The volume per unit length at the output may be larger than the volume per unit length in the central region. This may result in an increased saturation power.

The optical amplifier may comprise a waveguide. The waveguide may be, for example, a ridge waveguide or a buried heterostructure waveguide. The optical amplifier may comprise an exit region adjacent to the output of the optical amplifier, the exit region having a greater waveguide width than the waveguide width of the central region. Thus, the optical amplifier may have a wider active stripe at the exit region of the optical amplifier than at the central region. This may result in an increased saturation power.

The exit region may be optically coupled to the central region by a tapered section configured to adiabatically transfer the optical mode between the central region and the exit region. This may allow the optical mode to be efficiently transferred between the different regions of the device.

BRIEF DESCRIPTION OF THE FIGURES

The present invention will now be described by way of example with reference to the accompanying drawings.

In the drawings:

Figure 1 shows a schematic illustration of the cross-section of a ridge waveguide, illustrating the ridge width w, the active region thickness d and the quantum well confinement.

Figure 2 shows a plot of current versus saturation power (P_sat) showing examples of previously achieved results.

Figure 3(a) shows a plot of device length versus saturation power (P_sat) for previous devices.

Figure 3(b) shows a plot of gain versus saturation power (P_sat) for previous devices.

Figure 4 illustrates a previous approach for an SOA (MIT, J. Klamkin et al, ‘High-Output Saturation Power Variable Confinement Slab-Coupled Optical Waveguide Amplifier’ in OFC’11 , paper JThA25).

Figure 5 shows a plot of noise figure versus wavelength for a polarization independent, wideband and high power SOA exhibiting a relatively low noise figure (from A. Verdier et al, “Wideband material for Low linewidth Widely Tunable Laser and Reach Extender for Optical Access Networks ”, ECOC17).

Figure 6 schematically illustrates an optical amplifier according to an embodiment of the invention.

Figure 7 schematically illustrates the basic concept of increasing saturation power in an optical amplifier. Figures 8(a) to 8(d) schematically illustrate an optical amplifier according to another embodiment of the invention. Figure 8(a) shows a top view, Figure 8(b) shows a side view, Figure 8(c) shows cross-sectional views and Figure 8(d) shows optical mode simulations.

Figures 9(a) to 9(d) schematically illustrate an optical amplifier according to a further embodiment of the invention. Figure 9(a) shows a top view, Figure 9(b) shows a side view, Figure 9(c) shows cross-sectional views and Figure 8(d) shows a summary of the properties in the different regions of the optical amplifier.

Figure 10 shows the theoretical results of calculations of the noise figure in dependence on the SOA wavelength range for a standard SOA structure compared to the input section of the SOA being configured to, during operation, present a gain that is blue-shifted by 50 nm compared to the remaining part of the active region.

DETAILED DESCRIPTION

The optical amplifier described herein targets an increase in saturation power and/or a decrease in noise in an optical amplifier. Advantageously, in some embodiments, both of these effects may be achieved simultaneously.

The noise in an optical amplifier, such as a SOA, may be characterised by parameters including the noise factor or noise figure. These parameters are a measure of the degradation of the signal-to-noise ratio (SNR), caused by components in a signal chain, with lower values indicating better performance.

As will be known to those skilled in the art, the noise factor of an optical amplifier is defined as the ratio of the input SNR to the output SNR.

In an optical amplifier device, a large saturation power (P_sat) requires a low optical confinement. However, traditionally, this reduces the gain in the device. To achieve sufficient gain for a given P_sat, the device length generally has to be increased, which increases the injected current. Therefore, a large P_sat traditionally requires a large power consumption to achieve sufficient gain.

Gain and P_sat may be expressed as below:

Gain = exp{(r. g — a)L) (2) Psat ^K^/p (3) where G = optical confinement, L = device length, V = active volume, a = propagation losses and g = material gain.

The noise factor (NF) of the optical amplifier can be expressed as:

where:

where C = coupling efficiency, a = propagation losses, n_sp= inversion factor, hv = photon energy, k = Boltzmann constant, T = temperature and AE_F = separation between Fermi levels.

Concerning the NF of such devices, low confinement and low propagation losses can allow the NF of the optical amplifier to be reduced below 5.5 dB. This value is generally regarded as the upper limit for acceptable performance. However, this approach may only reduce the NF below this value on part of the optical band. Shown in Figure 5 is a plot of NF versus wavelength for a polarization independent, wideband and high power SOA exhibiting a relatively low NF (from A. Verdier et al, “Wideband material for Low linewidth Widely Tunable Laser and Reach Extender for Optical Access Networks ”, ECOC17). This SOA has a large Psat (>20 dBm) and large gain (>18 dB). However, the NF increases towards the S-band, which impacts transmissions and can degrade performance.

Examples of an approach for reducing the noise in an optical amplifier will now be described.

As schematically illustrated in Figure 6, in one embodiment of the present invention, the semiconductor optical amplifier 600 comprises an active layer, or active stripe, extending between an input 601 and an output 602 of the optical amplifier. The active layer may comprise multiple quantum well (MQW) semiconductor material. To reduce the noise in the SOA (specifically the noise factor and/or noise figure), the active layer comprises a noise suppression part 603 operative to reduce noise of the optical amplifier and a second part 604. The noise suppression part is located adjacent to the input 601 of the optical amplifier 600. The noise suppression part 603 has an active layer with an epitaxial structure different to the epitaxial structure of the active layer of the second part 604. Preferably, the noise suppression part 603 has a material composition different to that of the second part 604. The noise suppression part is configured to, during operation of the optical amplifier, present a gain that is blue-shifted by at least 20 nm compared to the second part.

The noise suppression part of the active region may have a larger AE_F for a given current density (which may be thanks to smaller photoluminescence wavelength).

When cascading two SOAs, the total NF mainly depends on the NF of the first SOA, provided the gain of the first SOA, Gi, is sufficient:

Therefore, at the input of the device, the growth of a different active layer having a different material composition to that of the remainder of the device (i.e. a part of the device adjacent to the input region but spaced from the input of the optical amplifier) is used to reduce the NF.

The noise suppression part may be composed of a material with a larger band-gap energy than that of the second part. For example, the band-gap energy of the material comprising the noise suppression part may typically be 25 meV larger than the band-gap energy of the material comprising the second part (a wavelength difference of 50 nm corresponds to an approximate difference in band-gap energy of 25-28 meV in the band of interest). Optionally, the noise suppression part may be composed of a material with larger energy separation between Fermi levels, or larger AE_F, than the second part.

The noise suppression part may be configured so as to, during operation of the optical amplifier, decrease the inversion factor at the input of the optical amplifier. The NF of an optical amplifier is proportional to the invention factor, n_sp. Improved inversion corresponds to a smaller n_sp (1 being the smallest possible value). Thus, reducing the inversion factor at the input can reduce the noise in the optical amplifier.

During operation of the optical amplifier, the noise figure of the optical amplifier may be less than 6 dB in the first 20 nm of the operating band. This may allow performance to be improved in the low-wavelength region of the optical amplifier. To be most effective, the layer at the input of the optical amplifier preferably presents a gain blue-shifted by at least 20 nm, preferably by 35 nm, more preferably by 50 nm, compared to the main active layer of the remaining part of the SOA. In some implementations, this layer can reduce the NF by improving inversion factor at the input of the SOA, particularly at the low wavelength range of the SOA. For example, in the first 20 nm of the operating band of the SOA.

To achieve this shift, in some implementations, both the noise suppression part and the second part of the active region may be composed of Quantum Wells (QW), or Multiple Quantum Wells (MOW), made with InGaAsP quaternary alloys, which may be separated by barriers made with another InGaAsP quaternary alloy having a larger band-gap energy. The InGaAsP may have inclusions of N or Al. The PL (photoluminescence) of these materials can be varied by modifying the composition of these alloys, or the thickness of the QW. Thus the composition and/or QW thickness of the epitaxial structures of each part can be controlled to tune the gain of the different parts.

In order to increase the gain of the optical amplifier, it is also desirable to increase the optical confinement in the central region of the device.

In previous implementations of SOAs where the confinement has been varied along the SOA, confinement increases when the ridge width increases, so the ratio between volume and confinement does not vary significantly. In contrast, the approach described herein leads to the opposite effect, i.e., confinement decreases when the ridge width increases, so the ratio between volume and confinement will vary a lot.

The basic concept is schematically illustrated in Figure 7. In this top view, the input and output of the optical amplifier 700 are shown at 701 and 702 respectively.

The optical confinement is adapted along the active stripe region of the optical amplifier to maximize the conversion efficiency of the optical amplifier. Adjacent to the output 702, the optical confinement G is low to give a high P_sat and remote from the output, in at least the central region of the device, the optical confinement G is larger than at the output to give a larger gain per unit length than at the output.

Examples of ways in which this may be done will now be described. As schematically illustrated in Figures 8(a) to 8(c), a semiconductor optical amplifier 800 comprises an active layer 806 extending between an input 801 and an output 802 of the optical amplifier. The optical amplifier has a central region 803 remote from the output 802 of the optical amplifier. The central region 803 comprises a cladding layer 804. The cladding layer is configured to increase the optical confinement in the central region and promote concentration of the optical mode in the central region so as to increase the gain per unit length of the optical amplifier. The optical confinement at the output is lower than the optical confinement in the central region. This may allow for the saturation power of the optical amplifier to also be increased without requiring a long chip length or high current. The central region may also at least partially overlap with the input region 805 of the optical amplifier. Therefore, the central region is remote from the output 802 of the optical amplifier, but may in some implementations, extend to the input 801 of the optical amplifier. Alternatively, the central region comprising the cladding layer, configured as described above, may extend across a region that is remote from both the input and output of the optical amplifier.

A high gain region is therefore introduced in the center of the device as a result of the upper cladding layer 804. The upper cladding layer attracts the optical mode on the active layer and thus may increase the gain per unit length. By adjusting the design of this upper cladding, the gain of the optical amplifier can be adapted without having to reduce P_sat. The central region 803 may therefore be configured so as to adapt the gain of the optical amplifier independently of the saturation power of the optical amplifier.

To increase the gain in the central region 803, the central region has a smaller volume per unit length than regions of the optical amplifier remote from the central region. The volume per unit length at the output 802 may be larger than the volume per unit length in the central region 803. This can result in an increase in P_sat.

Thus there is a larger optical confinement in the center of the device than at the output, with a smaller volume (for example, cross-sectional dimensions of 3 pm * 0.1 pm and/or a beam waist (1/2e diameter of the optical mode) of 1.9 pm * 1.3 pm), so the conversion efficiency will be large. There is a smaller confinement at the output than in the center of the device with a larger volume (for example, cross-sectional dimensions of 5 pm * 0.1 pm and/or a beam waist of 4.4 pm * 2.6 pm), in order to achieve a large saturation power.

The active layer of the optical amplifier is shown at 806 in Figure 8(b). The active layer may comprise a multiple quantum well layer. The cladding layer 804 is preferably located above the multiple quantum well layer. The active layer and cladding layer are grown on a lower cladding layer 809. The upper cladding layer, active layer and lower cladding layer may be positioned on a substrate 810, as shown in Figure 8(c). The substrate is preferably a semiconductor substrate, such as a Si or InP wafer.

Both the lower and upper cladding layers may be made from InGaAsP quaternary materials. The lower and upper layers may each be made from a different material. The upper cladding layer is preferably made from a material having a larger PL wavelength (i.e. a larger refractive index) than the lower one.

It has been demonstrated experimentally that only the end of the device influences the P_sat. As can be seen in Figure 8a, in the preferred implementation, the exit region 807 of the device is optically coupled to the central region 803 by a tapered section 808. The tapered section 808 allows for adiabatic transfer of the mode between the exit region 807 and the central region 803 of the SOA. The length of the tapered section (in the direction perpendicular to the input and output faces of the optical amplifier) is preferably at least 200 pm. For example the length of the tapered section may be in the range of 200-600 pm. In one example, the width (in a direction perpendicular to the length of the optical amplifier) of the upper cladding layer may decrease from 3 to 0.5 pm along the tapered section (towards the output region), while the active waveguide width may increase from 3 to 5 pm (towards the output region).

Therefore, in this implementation, the exit region 807 adjacent to the output 802 of the optical amplifier that has a greater waveguide width than the waveguide width of the central region. The exit region 807 is optically coupled to the central region 803 by the tapered section 808 configured to adiabatically transfer the optical mode between the central region and the exit region.

In some implementations, the exit region may have a lower optical confinement than the central region and other regions of the device remote from the exit region.

Figure 8(d) shows simulations of the optical mode confinement in each region of the device (center, mode transfer region (i.e. tapered section), and output from left to right respectively). The optical confinement can be seen to be greatest in the central region of the device.

In some implementations, the input section may have the same tapered design as the output section. In this case there would be two tapered sections: one as shown at 808 in Figure 8(a) and another tapered section at the input of the optical amplifier. This may reduce the input coupling losses, which may consequently reduce the NF of the optical amplifier further.

In a preferred embodiment of the present invention, the above-described features may be applied simultaneously to the SOA designed for high P_sat. There will be a trade-off between gain and P_sat for a given power consumption. The combination of these elements in the SOA can lead to an optimized SOA that exhibits low noise and increased saturation power, particularly in the low-wavelength region.

An exemplary implementation is shown in the top and side views of Figures 9(a) and 9(b) respectively, and also in the cross-sectional views of Figure 9(c) at four different areas along the device length.

As schematically illustrated in Figures 9(a) to 9(c), in this embodiment of the invention, the semiconductor optical amplifier 900 comprises an active layer extending between the input 901 and an output 902 of the optical amplifier. To reduce the noise in the SOA (specifically the noise factor (NF) or noise figure), the active layer comprises a noise suppression part 903 operative to reduce noise of the optical amplifier and a second part 904. The noise suppression part is located adjacent to the input 901 of the optical amplifier 900. The noise suppression part 903 has an epitaxial structure different to that of the second part 904. The noise suppression part 903 preferably has a material composition different to that of the second part 904. The noise suppression part is configured to, during operation of the optical amplifier, present a gain that is blue-shifted by at least 20 nm compared to the second part. This may be achieved by controlling the epitaxial structure of the active layer, as described previously.

As schematically illustrated in Figure 9, the semiconductor optical amplifier 900 also comprises a central region 905 remote from the output 902 of the optical amplifier. The central region 905 comprises a cladding layer 906 configured to increase the optical confinement in the central region and promote concentration of the optical mode in the central region so as to increase the gain per unit length of the optical amplifier. This may allow for the saturation power of the optical amplifier to also be increased without requiring a long chip length or high current. The central region may also at least partially overlap with the input region of the optical amplifier (i.e. the region of the device adjacent to the input 901). Therefore, the central region is remote from the output 902 of the optical amplifier, but may extend to the input 901 of the optical amplifier. Alternatively, the central region 905 may extend across a region that is remote from both the input 901 and output 902 of the optical amplifier. A high gain region is therefore introduced in the center of the device as a result of the upper cladding layer as well as the noise suppression region at the input of the device. The upper cladding layer attracts the optical mode on the active layer and thus may increase the gain per unit length. By adjusting the design of this upper cladding, the gain can be adapted without having to reduce P_sat. The central region may therefore be configured so as to adapt the gain of the optical amplifier independently of the saturation power of the optical amplifier.

To increase the gain in the central region 905, the central region has a smaller volume per unit length than regions of the optical amplifier remote from the central region. The volume per unit length at the output 902 may be larger than the volume per unit length in the central region 905. This can result in an increase in P_sat.

The optical amplifier may comprise a waveguide. The active region of the optical amplifier is preferably part of the waveguide. For example, the waveguide may be a ridge waveguide or a buried hetero-structure waveguide. The exit region 907 adjacent to the output of the optical amplifier has a greater waveguide width than the waveguide width of the central region.

As can be seen in Figures 9(a) to 9(d), in the preferred implementation, the exit region 907 is optically coupled to the central region 905 by a tapered section 908 (mode transfer region). This tapered section may therefore be present with (as in Figure 9) or without (as in Figure 8) the noise suppression region at the input. The tapered section allows for adiabatic transfer of the mode between the exit region and the central region of the SOA having the upper cladding layer. In the tapered region, the optical mode is transferred smoothly from the center to the output of the SOA.

The active layer of the optical amplifier (including parts 903 and 904 in this example) comprises a multiple quantum well layer. The cladding layer 906 is preferably located above the multiple quantum well layer. The active layer and cladding layer are grown on a lower cladding layer 909. The upper cladding layer, active layer and lower cladding layer may be positioned on a substrate 910, as shown in Figure 9(c). The substrate is preferably a semiconductor substrate, such as a InP wafer.

Figure 9(d) summarizes the different regions of the device in this preferred implementation. At the input, a material having a different composition is used to reduce the noise in the device during operation. At the center, the upper cladding layer is inserted for large gain. The output has a high P_sat design, with a larger waveguide width than the central region. These features combined may result in an optical amplifier with particularly advantageous properties in terms of low noise and high saturation power.

Figure 10 shows the theoretical results of calculations of the NF in dependence on the SOA wavelength range for a standard SOA structure compared to the input section of the SOA being configured to, during operation, present a gain that is blue-shifted by 50 nm compared to the remaining part of the active region. This demonstrates that in the central region the gain per unit length may in some implementations be increased by a factor of two, while the injected current may in some implementations be reduced by more than a factor of two, thus theoretically reducing the power consumption in the central region by a factor of four for a given gain target.

Therefore, embodiments of the present invention may allow for increasing the SOA gain for a given power consumption, and a reduction in its NF. This may be achieved by simultaneously inserting a gain detuned active section at the input of the SOA and inserting an upper cladding layer in the center of the SOA to improve the conversion efficiency.

Embodiments of the present invention may therefore have one or more of the following features, which can result in improved performance of the SOA relative to traditional SOAs. At the input, a different epitaxial structure and/or material can be used in the active region to improve the noise (for example, the NF) of the device. In the central region of the device (remote from the output), a large confinement and a small volume can allow a large gain for a limited current. At the output, the standard configuration (low confinement, large volume) can be kept to achieve a large P_sat. In addition to this, a large optical confinement at the input can be implemented to reduce the length and power consumption of this input section. Alternatively, the input might be identical to the output for a reduced length, followed by a taper to transfer the mode to the central region. This may reduce the input coupling losses, which may consequently reduce the NF of the optical amplifier further.

By varying the confinement and stripe width along the active stripe, the power consumption may advantageously be reduced by at least a factor of two.

The NF in the lower part of the spectrum may in some implementations be reduced by 1 dB by using a different material at the SOA input region. The combination of these elements in the SOA can lead to an optimized SOA that exhibits low noise and increased saturation power, particularly in the low-wavelength region.

The applicant hereby discloses in isolation each individual feature described herein and any combination of two or more such features, to the extent that such features or combinations are capable of being carried out based on the present specification as a whole in the light of the common general knowledge of a person skilled in the art, irrespective of whether such features or combinations of features solve any problems disclosed herein, and without limitation to the scope of the claims. The applicant indicates that aspects of the present invention may consist of any such individual feature or combination of features. In view of the foregoing description it will be evident to a person skilled in the art that various modifications may be made within the scope of the invention.

Claims

1. A semiconductor optical amplifier (600, 900) comprising an active layer extending between an input (601 , 901) and an output (602, 902) of the optical amplifier, the active layer having a noise suppression part (603, 903) operative to reduce noise of the optical amplifier and a second part (604, 904), the noise suppression part (603, 903) being adjacent to the input (601 , 901) of the optical amplifier and having an epitaxial structure different to the second part (604, 904), the noise suppression part (603, 903) being configured to, during operation of the optical amplifier, present a gain that is blue-shifted by at least 20 nm compared to the second part (604, 904).

2. The semiconductor optical amplifier (600, 900) of claim 1 , wherein the noise suppression part (603, 903) is composed of a material with larger band-gap energy than the second part (604, 904).

3. The semiconductor optical amplifier (600, 900) of any preceding claim, wherein the second part (604, 904) is adjacent to the noise suppression part (603, 903), the second part being spaced from the input (601 , 901) of the optical amplifier.

4. The semiconductor optical amplifier (600, 900) of any preceding claim, wherein the noise suppression part (603, 903) is configured to present a gain that is blue-shifted by at least 35 nm compared to the second part (604, 904).

5. The semiconductor optical amplifier (600, 900) of any preceding claim, wherein the noise suppression part (603, 903) is configured so as to, during operation of the optical amplifier, reduce the inversion factor at the input (601 , 901) of the optical amplifier.

6. The semiconductor optical amplifier (600, 900) of any preceding claim, wherein, during operation of the optical amplifier, the noise figure of the optical amplifier is less than 6 dB in the first 20 nm of the operating band.

7. The semiconductor optical amplifier (900) of any preceding claim, the optical amplifier having a central region (905) remote from the output (902) of the optical amplifier comprising a cladding layer (906) configured to increase the optical confinement in the central region (905) and promote concentration of the optical mode in the central region (905) so as to increase the gain per unit length of the optical amplifier (900).

8. A semiconductor optical amplifier (800, 900) comprising an active layer (806, 903, 904) extending between an input (801 , 901) and an output (802, 902) of the optical amplifier, the optical amplifier having a central region (803, 905) remote from the output of the optical amplifier comprising a cladding layer (804, 906) configured to increase the optical confinement in the central region (803, 905) and promote concentration of the optical mode in the central region (803, 905) so as to increase the gain per unit length of the optical amplifier.

9. The semiconductor optical amplifier (800, 900) of claim 7 or claim 8, wherein the active layer (806, 903, 904) comprises a multiple quantum well layer, wherein the cladding layer (804, 906) is located above the multiple quantum well layer.

10. The semiconductor optical amplifier (800, 900) of any of claims 7 to 9, wherein the central region (803, 905) has a smaller volume per unit length than regions of the optical amplifier remote from the central region.

11. The semiconductor optical amplifier (800, 900) of any of claims 7 to 10, wherein the central region (803, 905) is configured so as to adapt the gain of the optical amplifier independently of the saturation power of the optical amplifier.

12. The semiconductor optical amplifier (800, 900) of any of claims 7 to 11 , wherein the optical confinement at the output (802, 902) is lower than the optical confinement in the central region (803, 905).

13. The semiconductor optical amplifier (800, 900) of any of claims 7 to 12, wherein the volume per unit length at the output (802, 902) is larger than the volume per unit length in the central region (803, 905).

14. The semiconductor optical amplifier (800, 900) of any of claims 7 to 13, wherein the optical amplifier comprises a waveguide and an exit region (807, 907) adjacent to the output (802, 902) of the optical amplifier, the exit region (807, 907) having a greater waveguide width than the waveguide width of the central region (803, 905).

15. The semiconductor optical amplifier (800, 900) of any of claims 7 to 14, wherein the exit region (807, 907) is optically coupled to the central region (803, 905) by a tapered section (808, 906) configured to adiabatically transfer the optical mode between the central region (803, 905) and the exit region (807, 907).