Device for modulating the amplitude of laser radiation and method for manufacturing same

Abstract

wavelength λ i A device (10, 100) for modulating the amplitude of incident laser radiation (1), a semiconductor layer (4) comprising a stack of multiple quantum wells, a metal underlayer (3) on which a structured metal upper layer (2) is placed, the two metal layers (2, 3) reflect incident laser radiation (1), the structuring of the upper layer and the distance L between these two metal layers being small enough that the device forms an optical microcavity with at least one resonant mode, At least a portion of the quantum well, called the active well, has a central wavelength λ ISB =hc / E ISB has intersubband absorption at the center wavelength λ ISB The coupling between this intersubband transition at and one of the modes of the microcavity leads to the excitation of a cavity polariton and the energy E ISB ±hΩ Rabi (However, Ω Rabi drives the Rabi splitting at The device (10, 100) comprises an electrical circuit (5) configured to apply two different voltage differences V0 and V1 between two metal layers, the device (4) being characterized in that it absorbs incident radiation (1) when the voltage difference is V0, and the device reflects or transmits the incident radiation when the voltage difference is V1.

Description

[Technical Field]

[0001] The present invention relates to the field of electro-optic modulation, and more particularly to electro-optic modulators, especially in the infrared. [Background technology]

[0002] Ultrafast modulation of the amplitude of laser radiation in the mid-infrared is an area of ​​research that has generated intense interest, particularly in the field of telecommunications for the transmission of information using optical carriers. Indeed, using optical carriers with wavelengths within one of two atmospheric windows (3-5 μm and 8-12 μm), Mie theory provides a high degree of robustness for airborne information transmission against atmospheric disturbances (water vapor, atmospheric particles, etc.). This robustness is due to the fact that the average size of atmospheric particles is generally much smaller than the wavelength of the carrier.

[0003] A second area of ​​interest is metrology and high-resolution spectroscopy, since ultrafast modulation can generate sidebands that are precisely positioned in frequency relative to the carrier frequency.

[0004] Currently, most lasers in the mid-infrared are quantum cascade lasers. However, there are no commercially available sources within this wavelength range that can be modulated at high frequencies (sub-GHz and above). The few solutions available and proposed in the scientific literature are still in the complex and expensive demonstration stage, which severely limits their use and applications.

[0005] From the prior art, it is known to fabricate ultrafast electro-optic modulators in the mid-infrared that rely on charge transfer between two coupled quantum wells (Nevou, L., et al. (2007). Short-wavelength intersubband electroabsorption modulation based on electron tunneling between GaN / AlN coupled quantum wells. Applied Physics Letters, 90(22), 223511). In this device, GaN quantum wells are coupled by an ultrathin potential barrier formed of AlN. One well acts as a reservoir, and the other active well is designed to exhibit intersubband (ISB) transitions at the illumination wavelength. Under application of a positive voltage, electrons are transferred from the reservoir well to the active well by tunneling, and the modulator then enters an absorbing state. The modulator is then re-entered into a transmissive state by applying a negative voltage, which returns the electrons to the reservoir well. Because the charge transfer mechanism is fast (a few picoseconds), the device can achieve excellent response times (several GHz). However, the device only allows a low modulation depth (about 30% at λ=2.1 μm).

[0006] Furthermore, it is known to fabricate ultrafast modulators in the mid-infrared that rely on the increase / decrease of quantum wells under the action of an electric field (Machhadani, H., et al. (2009). GaN / AlGaN intersubband optoelectronic devices. New Journal of Physics, 11(12), 125023). By inserting an active region consisting of three GaN / AlN quantum wells into a waveguide made of AlGaN, a modulation depth of 13.5 dB at 1.55 μm using a voltage of ±7 V has been demonstrated. In this type of device, the response frequency is limited only by the RC constant of the electrical circuit applying the gate voltage capable of emptying the quantum wells. In fact, another phenomenon that limits the response time is the charge transfer between coupled quantum wells, which is very fast because it is generated by tunneling (on the order of a few picoseconds). However, this device introduces significant propagation losses into the waveguide.

[0007] Finally, it is known to those skilled in the art to fabricate electro-optic modulators based on a stack of quantum wells arranged in a microcavity (the quantum wells are configured to exhibit a significant quantum Stark effect) (Lee, J., et al. (2014). Ultrafast Electrically Tunable Polaritonic Metasurfaces. Advanced Optical Materials, 2(11), 1057-1063). In this device, modulation is achieved by applying an electric field between the two elements of the microcavity and via the stack of quantum wells, which can shift the ISB absorption peak via the quantum Stark effect. Furthermore, the response time of such a device is limited only by the RC constant of the electrical circuit. However, the modulation depth remains low (30% for λ = 7.12 μm).

[0008] The object of the present invention is to alleviate certain problems from the prior art. To this end, one subject of the present invention is an electro-optic modulator based on a stack of quantum wells arranged in a microcavity using a strong coupling topology in order to obtain a good response time of the device and a good modulation depth. [Prior art documents] [Non-patent literature]

[0009] [Non-Patent Document 1] Nevou, L., et al. (2007).Short-wavelength intersubband electroabsorption modulation based on electron tunneling between GaN / AlN coupled quantum wells.Applied Physics Letters,90(22),223511 [Non-patent document 2] Machhadani,H.,et al.(2009).GaN / AlGaN intersubband optoelectronic devices.New Journal of Physics,11(12),125023 [Non-patent document 3] Lee, J., et al. (2014). Ultrafast Electrically Tunable Polaritonic Metasurfaces. Advanced Optical Materials, 2(11), 1057-1063 Summary of the Invention [Means for solving the problem]

[0010] For this reason, one subject of the present invention is to i 1. A device for modulating the amplitude of incident laser radiation (1), comprising a semiconductor layer (4) containing a stack of multiple quantum wells, a metal underlayer (3) on which there is a structured metal upper layer (2), the two metal layers (2, 3) reflect the incident laser radiation (1), the structuring of the upper layer and the distance L between these two metal layers are small enough for the device to form an optical microcavity with at least one resonant mode, at least some of the quantum wells, called active wells, have a central wavelength λ ISB =hc / E ISB has intersubband absorption at the center wavelength λ ISB The coupling between this intersubband transition at and one of the modes of the microcavity leads to the excitation of a cavity polariton, and the energy

number

[0011] According to a particular embodiment of the invention, the following is true: The metal underlayer is not discontinuous and the device reflects incident radiation for a voltage difference V1. - The upper metal layer is separated by a distance a and has a width s (where s + a < λ i / 2) of multiple metal strips. - The width s of the metal strip is s=λ i / 2.n ±30% (where n is the refractive index of this semiconductor layer). - width of metal strip s=λ i / 2.ns is s=3.λ i / 2.n ±30% (where n is the refractive index of this semiconductor layer). The contact between the semiconductor layer and the metal layer is created by a Schottky contact or by the introduction of an insulating layer. - The distance L between these two metal layers is λ i / 30 ±30% is equal to. The distance L between the two metal layers is chosen by digital simulation so that the device operates in a non-dispersive regime. - the quantum wells, or the barriers separating them, are electronically doped so that when a first voltage difference V0 is applied between these two metallic layers by this electrical circuit (5), the device generates a

number

number

number

number

[0012] Other characteristics, details and advantages of the invention will become apparent on reading the description given with reference to the accompanying drawings, given by way of example and respectively illustrating: [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a schematic illustration of an ultrafast modulation device for modulating the amplitude of incident laser radiation according to a first embodiment of the present invention; [Figure 2] FIG. 1 is a band diagram of the device in the non-dispersive regime. [Figure 3] FIG. 1 is a band diagram of the device in a distributed configuration. [Figure 4] 4 is a schematic diagram of an ultrafast modulation device for modulating the amplitude of incident laser radiation according to a second embodiment of the present invention. [Figure 5] 1 is the reflectivity of a device of the present invention with and without an applied voltage. [Figure 6] 10 is a schematic diagram of an ultrafast modulation device for modulating the amplitude of incident laser radiation during transmission according to a third embodiment of the present invention. [Figure 7] 1 is a metal overlayer of a modulating device according to an embodiment of the present invention. [Figure 8] 1 is an illustration of a modulation device according to an embodiment of the present invention. [Figure 9] 1 is an illustration of a modulation device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0014] In particular, the subject of the present invention is a device consisting of a semiconductor layer inserted into a microcavity that allows the amplitude of incident laser radiation to be modulated by applying an external electric field to the microcavity, creating a strong optical material coupling between the microcavity and the semiconductor layer.

[0015] Figure 1 is a diagram of an ultrafast modulation device 10 for modulating the amplitude of incident laser radiation 1 according to a first embodiment of the invention. This device 10 operates in reflection and is capable of obtaining high speed modulation (greater than 10 MHz) with a modulation depth of more than 90%. In the embodiment of Figure 1, the incident laser radiation 1 whose amplitude it is desired to modulate has a wavelength λ i (However, λ i ∈[3-20 μm]), thus providing modulated laser radiation 6.

[0016] In the embodiment of Figure 1, the laser radiation 1 is continuous wave radiation. Alternatively, the laser radiation 1 can be pulsed radiation.

[0017] The device 10 of Figure 1 is composed of a first metal lower layer 3, an intermediate semiconductor layer 4, and an upper metal layer 2. The thickness of the two metal layers is formed between 80 nm and 300 nm. The intermediate semiconductor layer is also called the active region.

[0018] The lower metal layer is uniform and has no discontinuities. The upper metal layer is spaced a distance a apart and has a width s (where s + a < λ i / 2). This criterion is important because it ensures that the device does not act as a diffraction grating and therefore does not exhibit the appearance of additional diffraction orders in the modulation of the incident laser radiation.

[0019] 1, the distance between the two metal layers, or the maximum thickness L of the active region, is much smaller than the wavelength of the incident laser radiation (L≦λ / 10). The cavity formed by the two metal layers therefore functions as a microcavity or optical microresonator with a single resonant mode at wavelength λ. Hereinafter, the thickness of the active region and the distance between the two metal layers are considered to be equal and have the value L.

[0020] The device of FIG. 1 includes an electrical circuit 5 configured to apply a voltage difference between two metal layers from an incident RF signal 7. Contact between the intermediate semiconductor layer and the metal layer is created either by a Schottky contact, by directly depositing a metal contact (e.g., titanium / gold) on the semiconductor layer, or by introducing an insulating layer (e.g., SiN or SiO2 or any other insulator) compatible with the fabrication of optoelectronic devices to avoid too large a current flowing through the semiconductor layer. Below, we will discuss in detail the effects of applying or not applying a voltage difference between the two metal layers (hence, generating an electric field). We will now briefly describe how a change in the absorption of a semiconductor layer at a given wavelength can be induced.

[0021] Depending on the thickness L of the semiconductor layer, the microcavity can operate in a non-dispersive or dispersive mode. In the dispersive mode, the resonant frequency or frequencies of the microcavity depend on the angle of the incident radiation, while in the non-dispersive mode, the resonant frequency or frequencies do not depend on the angle of incidence of the incident radiation. In this non-dispersive mode, the resonant frequency is set solely by the width s of the metal strips. It should be noted that L does not affect the operating wavelength of the device, only the dispersive or non-dispersive operating regime.

[0022] In a first embodiment, the thickness L of the semiconductor layer is set to a value which is proportional to the wavelength λ of the incident laser radiation whose amplitude it is desired to modulate. i , and to operate in a non-dispersive mode, λ iλ is approximately equal to λ / 30. Therefore, each metal strip belonging to the upper metal layer forms a "single" cavity together with the lower metal layer, because the electromagnetic field is located in each single cavity only below each metal strip. Therefore, there is no interaction between the single cavities due to the strong lateral confinement of the resonance. In this case, the reflectivity of the device is very high. Generally, when the thickness of the active region is λ i / 30 ±30% The microresonator is considered to operate in the non-dispersive regime when ρ is equal to ρ. Preferably, the active region thickness L is selected by digital simulation of the resonator strip structure to operate in the non-dispersive regime.

[0023] In a non-dispersive embodiment, the width s of the strips can set the wavelength of use of the amplitude modulation device 10. If the semiconductor layers do not absorb the incident laser radiation, the device 10 will exhibit two resonant modes TM corresponding to two different wavelengths λ=2.0 ns and λ=2.0 ns / 3, respectively, where n is the refractive index of the semiconductor layers. 00 and TM 02 Alternatively, according to another embodiment, the device 10 may operate at a wavelength λ i = 2.ns / (i+1) corresponds to each mode of resonance mode TM 0i (where i∈N)

[0024] Figure 2 illustrates schematically the resonant modes in the case of non-dispersive operation. Figure 2 shows the reflectivity of the device of Figure 1 as a function of frequency and angle of incident laser radiation (when the intermediate semiconductor layer does not absorb the incident laser radiation). Two horizontal strips 20 and 21 define two resonant modes TM 00 and TM 02 and . Each resonant mode has a specific linewidth δE. As mentioned above, over the short distance between the two metal layers, the frequency of the cavity resonant mode is independent of the angle of incidence of the laser radiation. For example, in the embodiment of FIG. 1, the wavelength λ cav =hc / E cav Resonant mode TM at =9.5 μm 00The semiconductor layer consists of a stack of quantum wells made of GaAs with a refractive index n≈3.3. The width of the metal strip is then s=1.44 μm. The linewidth of the resonant mode is given by the value δE cav It has.

[0025] After setting the wavelength of use by setting the width s of the metal strips, the spacing a between the metal strips is selected by simulation to be located as close as possible to the critical optical coupling. In this state, the loss of the resonator (which gives the linewidth of the resonant mode) is equal to the coupling ratio of photons generated from the incident radiation, which means that the total energy of the incident radiation is absorbed. In fact, a sets the metal strip density. Specifically, a is optimized with s set to obtain the best contrast in reflectivity depending on the wavelength of the incident radiation. The spacing a is optimized by simulation, and the optimal value can vary depending on the semiconductor active region. In the embodiment of Figure 1, a = 1.1 μm.

[0026] Alternatively, device 10 can be operated in a distributed mode. In this mode of operation, the semiconductor layer thickness L is set to λ i λ / 10. Indeed, for such a thickness, the electromagnetic field is no longer located only under the upper metal strip in a single cavity that is coupled, which causes a dispersive behavior. i If the angle of incidence is greater than or equal to 10, the microresonator is said to operate in a dispersive regime, in which the frequency of the optical mode depends on the angle of incidence.

[0027] FIG. 3 shows the device 10 when operated in a distributed mode, i.e., when the active region thickness L is λ i3 illustrates two optical resonant modes of device 10 when the angle s / a exceeds s / 10. Similar to FIG. 2, FIG. 3 shows the reflectivity of device 10 as a function of the frequency of the incident radiation when the intermediate semiconductor layer does not absorb the incident laser radiation. As mentioned above, the dependence of the frequencies of the two resonant modes 40 and 41 as a function of wavelength is observed. The distance between the two branches is set by the ratio s / a. Specifically, device 10 in the dispersive regime of the embodiment of FIG. 3 has an optimal operating frequency that is tunable as a function of the angle of incidence. However, in the following, it is assumed that the thickness of the active region is such that device 10 of FIG. 1 operates in the non-dispersive regime.

[0028] In another embodiment illustrated in FIG. 4, the device 100 has a semiconductor layer 4 contained only under the metal strips (in the single cavities). There is no semiconductor layer between the single cavities (between the metal strips on the metal top layer). In other words, the active region is etched in the parts of the active region that are not covered by metal. This allows to create an air / semiconductor interface between the single cavities, thus obtaining better confinement of the electromagnetic modes in the single cavities. Therefore, better coupling of the incident laser radiation in the cavity is obtained. In this embodiment, the thickness L of the active region is selected to operate in a non-dispersive regime. The thickness of the semiconductor layer is therefore λ i It should be noted that in the embodiment of Figure 4, it is difficult in practice to achieve a distributed operating configuration.

[0029] In the embodiment of FIG. 1, the semiconductor layer has a width L QW Seven identical quantum wells formed of GaAs with a thickness of 9.5 nm and Al with a width of 20 nm. 0.3 Ga 0.7 The quantum well is composed of barriers formed of As. The quantum well is formed of Al in the form of delta doping. 0.3 Ga 0.7 6.10 introduced into the As barrier 11 cm 2 (n-type) electron doping. Alternatively, the doping can be introduced into the well. The quantum well then has a linewidth

number

number

[0030] In the second exemplary embodiment of Figure 4, device 100 includes a semiconductor layer configured with a stack of multiple pairs of coupled quantum wells. In this embodiment, the coupled quantum wells are formed by a narrow, undoped well, called the active well, and a wide, electronically doped well, called the reservoir, separated by a first thin barrier. Each pair of coupled quantum wells is separated by a second barrier that is wider than the first barrier. Alternatively, electronic doping can be introduced into the first thin barrier that separates the wide well from the active well.

[0031] For example, the following structure (6.10 11 cm -2 or 1.2.10 12 cm -2 Al from a wide 14.5 nm well in GaAs with doping 0.3 Ga 0.7 Width L separated by a thin 4 nm barrier formed of As QWThe semiconductor layer can be constructed by stacking five layers of Al. 0.3 Ga 0.7 They are separated from each other by an 18 nm barrier of As.

[0032] In this embodiment, the active well has a linewidth δE ISB λ ISB = 9.5 μm. However, the semiconductor layers of device 100 do not absorb at 9.5 μm because the charge resulting from the doping is in the reservoir well and not directly in the active well.

[0033] Alternatively, the structure of the active region of the embodiment of FIG. 4 may be identical to the structure used in device 10 of FIG. 1, and the structure of the active region of the embodiment of FIG. 1 may be identical to the structure used in device 100 of FIG. 4.

[0034] As mentioned above, the formation of quantum wells in a resonant cavity modifies the interaction between photons and intersubband excitations, hereafter referred to as ISB plasmons. Indeed, a semiconductor microcavity can selectively couple a single mode of the cavity's electromagnetic field to a single ISB plasmon mode. In the embodiment of FIG. 1, the single mode is mode TM coupled to the ISB transition at 9.5 μm. 00 The elements of this optical material coupling matrix are the quantities Ω, which are called Rabi frequencies. rabi The quantity Ω is expressed by rabi The higher the value, the stronger the bond.

[0035] It is defined as follows:

number

[0036] However, L QW,tot is the total length of the active quantum wells in the structure, L is the total length of the semiconductor layers, and L QW is the length of the active quantum well, and f 12is the oscillator force of the ISB transition of the active quantum well (determined by digital calculation), ε0 is the dielectric constant of vacuum, and ε slab is the dielectric constant of the active region, and m * is the effective mass of the electron in the active quantum well. Finally, n d is the surface electron doping introduced into each quantum well or barrier in the case of delta doping.

[0037] In the embodiment of FIG. d Al 0.3 Ga 0.7 In the second embodiment (FIG. 4), n d corresponds to the doping introduced into the reservoir well or barrier. In this embodiment, L QW,tot =L.

[0038] To obtain the highest modulation contrast between the absorbing and non-absorbing states of the device, the non-absorbing state is operated in the strong photon (ISB plasmon) coupling regime. Indeed, under certain conditions, this coupling can become so-called strong coupling, and then new eigenmodes of the system, called cavity polaritons, appear. The linewidth of the coupled cavity resonant modes and the ISB transitions is defined as δE cav and δE ISB It is defined as:

[0039] E rabi <<δE cav , δE ISB , spontaneous emission is irreversible and photons emitted in radiative recombination are "lost" in the cavity, resulting in Ω rabi The oscillator force coupled to E is so weak that it can reabsorb this photon. This is a weak coupling. Conversely, E rabi >>δE cav , δE ISBIf , the emitted photon stays in the cavity long enough to be reabsorbed. This is a strong coupling regime. Therefore, there is a coherent energy exchange between the quantum well and the cavity photon modes. This phenomenon is reversible. These are Rabi oscillations. In this regime, the eigenstates of the system are mixed photon ISB plasmon states called polaritons (upper and lower polaritons). In the spectral range, there is an energy separation of 2.Ω, called the Rabi splitting. rabi and by the anticrossing observed between the bound states. ISB =2πc / λ ISB , the frequencies of simple intersubband transitions ±Ω, which are isolated from any coupling. rabi This strong coupling and this Rabi splitting are reflected by the splitting at . Conversely, the weak coupling between the ISB transition and the cavity resonance mode is reflected only by the broadening of the cavity resonance.

[0040] In all embodiments of the present invention, δE cav , δE ISB ≒E ISB The quantum wells and microcavities are parameterized to have a .DELTA..times ...

[0041] Therefore, different parameters of the semiconductor layer structure affect the Rabi splitting (L QW,tot , L, L QW , f 12 , ε slab , n d ) to E Rabi >δE ISB , δE cav Therefore, δE ISB , δE cav <E ISB For / 10, E rabi >>E ISB / 10 is desirable (see [Equation 1]).

[0042] Next, without the electric field generated between the metal layers by the electric circuit 5, the device 10 of FIG. ISB ±±Ω rabiabsorbs the incident laser radiation at frequency ω cav The frequency ω corresponding to the ISB transition of the quantum well close to the resonant mode of the cavity ISB It should be understood that the electrons do not absorb at 1000 .mu.m. In fact, the electrons reside in quantum wells within the cavity, and the strong coupling leads to the formation of cavity polaritons and Rabi splitting.

[0043] Similarly, without an electric field generated between the metal layers, the device 100 of FIG. 4 would have a cavity resonant mode ω cav In fact, the electrons emerging from the doping reside in the reservoir well and are absorbed by the cavity's resonant mode ω cav A frequency close to ω ISB There is no active quantum well that exhibits an ISB transition at the frequency ω cav The cavity is transparent to incident laser radiation at frequencies other than the resonant modes of the microcavity. However, the cavity allows coupling (and therefore absorption) of incident laser radiation at such frequencies. Conversely, incident radiation having a frequency different from the resonant modes of the microcavity is reflected by device 100.

[0044] The electric circuit 5 is adapted to apply a voltage difference between the two metal layers.

[0045] 1 and 4, the electrical circuit is configured to apply a voltage difference V1=F*L between the two metal layers, where F is the electric field required to "empty" or deplete the quantum well of charge arising from doping. In effect, the application of an electrical bias allows for the rapid transfer of charge from the quantum well to one of the metal layers.

[0046] In the embodiment of FIG. 1 having the exemplary GaAs structure described above, V1=6 V. Thus, application of such a voltage difference causes the semiconductor layer to

number

[0047] Thus, when a voltage difference V1 = F*L is applied between the two metal layers, the device functions as a "conventional" microcavity, and the device of Figure 1

number

[0048] The wavelength λ is equal to the resonant mode of the cavity i Taking the incident laser radiation at as a reference, the "absorption" state is used to indicate the state of device 10 corresponding to a voltage difference V=F*L applied between the two metal layers, and the "reflection" state is the state corresponding to zero voltage difference (V=0V) between the two metal layers.

[0049] Thus, using strong coupling regimes and Rabi splitting, the device of FIG. 1 can achieve very good modulation depth (>90%) between absorbing and reflecting states.

[0050] 5 illustrates the reflectivity of device 10 as a function of the frequency of the incident laser radiation and as a function of the voltage difference applied between the metal layers. Curve 50 (thin line) represents the reflectivity of the device for the absorbing state (voltage difference V1 = F * L between the two metal layers). As mentioned above, in this particular case, there is no longer any charge in the quantum wells, and therefore there is no Rabi splitting, and the device functions as a "conventional" microcavity. That is, when the wavelength of the incident laser radiation is equal to the resonant mode of the cavity, the device absorbs the incident laser radiation.

[0051] Curve 51 (bold line) represents the reflectivity of the device for the reflective state (zero voltage difference between the two metal layers). In this particular case, doping is present in the quantum wells within the cavity, leading to the formation of cavity polaritons and ω cav Near the resonant mode of the cavity at ω ISB Frequency ω from a simple ISB transition to ISB ±Ω R This splitting causes a Rabi splitting in the energy

number

[0052] It is important to note that in the embodiment of Figure 1, the distance L between the two metal layers (and therefore the maximum thickness of the semiconductor layer here) is not simply limited by the desire to operate in the cavity-non-dispersive regime. Furthermore, the thickness of the semiconductor layer is also limited by the maximum gate voltage V that the Schottky contact between the semiconductor layer and the metal layer can withstand. max This means that V1=F*L <V max =F*L max means that L <L max So that the upper limit L max By using Ti / Au for the undoped GaAs contact, a V on the order of ±5 V can be achieved. maxThis value can be exceeded by using insulating layers (SiN, SiO2, Al2O3, ...).

[0053] Although the charge transfer induced by the application of a voltage difference is a fast phenomenon, it is limited by the time it takes for the electrons to travel from the quantum well to one of the metal layers, which can be several tens or even hundreds of nanometers. Furthermore, in the embodiment of Figure 1, the response frequency of device 10 remains below 1 GHz. To improve the response time of this embodiment, the thickness of the semiconductor layers can be reduced.

[0054] The inherent response time limitations of the device 10 of the embodiment of FIG. 1 do not apply to the embodiment of FIG. 4. Indeed, in device 100, the electrical circuit 5 is configured to apply a voltage difference selected to generate a bias or electric field that can move electrons introduced into the reservoir well during doping into the active well by tunneling. This charge transfer means that electron doping is present in the active quantum well within the cavity, resulting in the formation of cavity polaritons and Rabi splitting, and the structure then operates in a strong coupling mode. Thus, for a suitably chosen voltage difference V≠0, the device 100 of FIG. 4 will respond at a frequency ω ISB ±±Ω R absorbs the incident laser radiation at frequency ω cav The frequency ω corresponding to the ISB transition of the active quantum well close to the resonant mode of the cavity ISB Using the same criteria as for device 10, it is proposed that the device is in a reflective state when the voltage difference between the two metal layers, V≠0, and in an absorbing state when V=0V.

[0055] 5 also schematically illustrates the reflectivity of device 100 as a function of the frequency of the incident laser radiation and as a function of the voltage difference applied between the metal layers. Curve 50 represents the reflectivity of the device for the absorbing state (zero voltage difference between the two metal layers). As mentioned above, in this particular case, electrons are in the reservoir, therefore there is no Rabi splitting, and the device functions as a "conventional" microcavity. That is, when the wavelength of the incident laser radiation is equal to the resonant mode of the cavity, the device absorbs the incident laser radiation.

[0056] Curve 51 represents the reflectivity of the device for the reflective state (voltage difference V≠0 between the two metal layers). In this particular case, doping is transferred from the reservoir to the active well in the cavity, leading to the formation of cavity polaritons and ω cav Near the resonant mode of the cavity at ω ISB Frequency ω from a simple ISB transition of the active well in ISB ±±Ω R This splitting causes a Rabi splitting in the energy

number

[0057] Similar to the active region of FIG. 1, the use of strong coupling configurations and Rabi splitting in the device of FIG. 3 allows for very good modulation depth (>90%) between absorbing and reflecting states.

[0058] The barrier between the active well and the reservoir is very thin (a few nanometers), so tunneling is very fast (on the order of picoseconds). Furthermore, the response frequency of the device in Figure 3 is very high (over 1 GHz), limited only by the RC constants of the electrical circuit.

[0059] Another advantage of the embodiment of Figure 3 is that the voltage difference required to induce charge transfer from the reservoir to the active well, and thus the transition from the absorbing state to the reflecting state, is lower than in the embodiment of Figure 1. Thus, in device 100, a voltage V1 = 2 V is sufficient at λ = 9 μm to induce charge transfer into the active well. This value is much lower than the value required to empty the quantum well in the embodiment of Figure 1 (where V1 = 6 V), and V1 <V max can be assured that this is indeed the case.

[0060] Furthermore, in the embodiment of FIG. 1 or FIG. 4, the overall width d x-tot and total length d y-tot is preferably less than 200 μm, which reduces the RC constant of the device. x-tot and total length d y-tot indicates the maximum distance between

[0061] Similarly, in all embodiments, the contacts and microjunctions are adapted so as not to limit the response time of the device.

[0062] In another embodiment, the device can be cooled to 78 K, for example using a cryostat or Peltier element. This cooling is especially necessary if it is desired to modulate THz waves. THz is defined as λ>30 μm.

[0063] In another embodiment, for operation in a non-dispersive regime, s=3.λ i / 2.n (where λ i is the wavelength of the incident laser radiation whose amplitude it is desired to modulate, thus increasing the overlay factor between the electromagnetic mode and the semiconductor active region, and thus the TM 02 To increase the Rabi splitting of the ISB transitions of the active quantum wells at frequencies close to the frequency of the mode, the resonant mode TM of the cavity is 02 This allows the cavity mode TM in the absorbing state to be 03This allows for a better frequency separation between the absorption line corresponding to the ISB transition and the two lines corresponding to the Rabi splitting of this ISB transition in the active quantum well. This results in a better modulation depth. Furthermore, the larger dimensions of the metal strip also simplify the fabrication of the resonator.

[0064] In another embodiment, the quantum wells are produced in GaAs / AlGaAs, InGaAs / AlInAs, InAs / AlSb, Si / SiGe, GaN / AlGaN, or any other material capable of producing quantum wells with ISB transitions comprised between 3 μm and 200 μm.

[0065] 6 shows another embodiment of a device 1000 for modulating the amplitude of incident laser radiation. Unlike the other embodiments of the invention, this device operates in transmission mode. For this, the metal underlayer 3 is discontinuous and is structured in the same way as the metal upper layer 2, so that the metal strips of the metal upper layer and the metal underlayer face each other. The assembly formed by the two metal layers and the intermediate semiconductor layer therefore modulates the amplitude of the semiconductor layer 3 at wavelength λ i If there is no absorption for λ i does not correspond to a resonant mode of the cavity, the wavelength λ i Conversely, if the incident radiation has a wavelength λ corresponding to a resonant mode of the cavity, iIf the active region does not absorb at this wavelength, the incident radiation will be coupled into the cavity (and therefore absorbed). The thickness L of the active region 3 can be selected to operate in either the dispersive or non-dispersive regime. The device includes a substrate 8, a component below and in contact with the metal underlayer that is transparent to the incident laser radiation. This substrate can be made of CaF2, ZnSe, BaF2, or any other material that is transparent to the mid-infrared. Fluorine-based materials (CaF2, BaF2) are particularly suitable because they have a low refractive index (n<1.5) in the mid-infrared spectral range. In one embodiment, the substrate 8 is bonded to this metal underlayer 3. The operating principle remains the same as in other embodiments. That is, the "absorption state" can be changed to the "transmission state" (or vice versa, depending on the structure of the active region) by applying a voltage difference between the two metal layers using an electrical circuit 5, causing the transfer of charges introduced into the active region by doping.

[0066] Alternatively, device 1000 may simply include semiconductor layer 4 below the metal strips. As a result, there is no semiconductor layer in the portions of the active region not covered by metal. For example, an etching process may be used to remove the desired portions of the semiconductor layer. It should be noted that in this embodiment, it may be difficult to operate in a distributed mode.

[0067] In another embodiment illustrated in Figure 7, the metallic upper layer 2 is structured in two directions x and y to form an electrically connectable two-dimensional resonator (or cavity) R. In this embodiment, the device operates in a non-dispersive regime (L<λ / 30). The structuring of the layers is adapted to form square, rectangular, triangular or other shaped resonators together with the metallic layer 3 and the semiconductor layer 4, all of which are identical and have a wavelength λ i, which supports resonant modes. As is known, the resonant wavelengths of these resonators depend on the structuring of layer 2, more particularly on the shape and dimensions of the cavities. These resonant modes can be determined digitally or analytically for simple geometries. In order to maximize the contrast of the modulator, the spacings px and py between the resonators in the directions x and y are preferentially chosen so that they are located as close as possible to the critical optical coupling.

[0068] Preferentially, as illustrated in Fig. 7, each resonator is connected to its neighbor in the direction x by a metal connection C whose width is much smaller than the dimension of the resonator in order to reduce disturbances to the electromagnetic modes. More precisely, the width of the connection is between 1 / 5 and 1 / 20 of the characteristic dimension of the cavity (diagonal in the case of a rectangular cavity, side in the case of a square cavity, and diameter in the case of a circular cavity). Alternatively, according to another embodiment, each resonator is connected to its neighbor by a connection in the direction x and / or in the direction y.

[0069] The active area 4 may or may not be etched in the areas not covered by layer 2 .

[0070] This embodiment allows a reduction in the RC constant of the device due to the reduction in the surface of the metal layer 2, and therefore an increase in the modulation speed.

[0071] In another embodiment illustrated in Figure 8, the metallic top layer 2 is structured to allow operation of the modulation over a wide spectral band. In this embodiment, the device operates in the non-dispersive regime (L<λ / 30). The top layer 2 has a width s different from the other metallic strips of the subset. i The structure is such that it is composed of multiple identical subsets SE each containing multiple metal strips of a distance a in the direction x. iThe subsets are separated from adjacent strips by a periodicity Px in the direction x. In the example of FIG. 8, by way of non-limiting example, each subset includes three strips. Alternatively, according to another embodiment, each subset includes at least two strips. The width s of the subsets i Each metal strip emits a frequency that is different from the frequencies associated with the other strips.

number

[0072] The upper part of Figure 8 shows the reflectivity of the device as a function of the frequency of the incident radiation with and without an applied voltage difference. For zero bias (voltage difference V0 = 0 V), the absorption of the device, corresponding to the sum of the curves C1, each associated with a strip of subset SE, is maximum over a broad spectral band δω associated with the absorption of the semiconductor cavity alone. The parameter a i and s i By varying the thickness L of the semiconductor layer 4, the absorption can be adjusted / maximized by digital simulation.

[0073] The width of the broad spectral band δω is approximately |ω s-max -ω s-min |(However, ω s-min and ω s-max is the maximum width (s max ) and minimum width (s min ) are the resonant frequencies associated with the metal strips of

[0074] Similarly, for a positive bias (voltage difference V1 ≠ 0 V), the absorption of the device corresponding to the sum of the curves C2, each associated with a strip of subset SE, as described above, is maximum over the broad spectral bands δω' and δω'' associated with the absorption of the quantum well and the excitation of polaritons.

[0075] (strip width s iDue to the dependence of the polariton excitation frequency on the resonant frequency of the cavity (coupled to the

[0076] This embodiment allows for broadband operation of the modulation device because the modulation device is absorbent over a wide frequency range δω for zero voltage difference, and over wide frequency ranges δω″ and δω′ for positive voltage difference.

[0077] Alternatively, in another embodiment, the metallic top layer 2 is structured in both directions x and y in a manner similar to the embodiment illustrated in Figure 8, thus forming two-dimensional resonators (patch cavities). In this embodiment, each patch cavity has a different frequency ω that depends on the shape and dimensions of the patch cavity. patch_i 8, which produces structuring only in the y direction, the modulator operates over a wide frequency range.

[0078] In another embodiment, the semiconductor layer 4 has a different quantum well thickness L for each period i. QW,i and thus, N tot As a non-limiting example, illustrated at the top of FIG. 9, three different thicknesses L QW,1 , L QW,2 and L QW,3 Therefore, N1+N2+N3=N tot There are N periods i is a different frequency ω for each period i. ISB,i is the number of periods of each quantum well stack, i, that cause intersubband absorption at . In this case, the Rabi frequency Ω is equal to the frequency obtained with a single thickness quantum well stack. RabiTo obtain an absorption at 1000 s, the doping in the quantum wells must be multiplied by three. Alternatively, in another embodiment, numbers other than three different thicknesses can be used. In the upper embodiment of Figure 9, a single quantum well is used for each period. Alternatively, in another embodiment, the quantum wells in each period are coupled wells, as in the embodiment of Figure 4.

[0079] The bottom part of Figure 9 shows the reflectivity of the device as a function of the frequency of the incident radiation, with and without an applied voltage difference.

[0080] At zero bias (voltage difference V0 = 0 V), the reflectivity of the device (curve C3) is the cavity frequency ω associated with the absorption of the semiconductor cavity alone. cav where the absorption spectrum width δω ISB is the linewidth of the resonant mode δE cav Simply concatenate to

[0081] For a positive bias (voltage difference V≠0 V), the reflectivity of the device (curve C4) varies with the number of intersubband absorption frequencies ω associated with each period i. ISB,i For this reason, abbreviated ω cav ±Ω Rabi is the maximum over two wide ranges of width δω'.

Claims

1. wavelength λ ｉ A device for modulating the amplitude of laser radiation (1) incident on the device, comprising: a metal underlayer (3) on which a semiconductor layer (4) comprising a stack of multiple quantum wells is disposed, and a structured metal upper layer (2) on which the semiconductor layer (4) is disposed, the structured metal upper layer (2) and the metal underlayer (3) reflect the incident laser radiation (1), and the structuring of the structured metal upper layer (2) and the thickness L of the semiconductor layer (4) are sufficiently small so that the device forms an optical microcavity having at least one resonant mode; At least a portion of the quantum well, called the active well, has a central wavelength λ ＩＳＢ = hc / E ＩＳＢ and has intersubband absorption at the center wavelength λ ＩＳＢ Coupling between the intersubband absorption at [Equation 1] (However, Ω Ｒａｂｉ drives the Rabi splitting at The device is configured to apply two different first voltage differences V between the structured metal upper layer (2) and the metal lower layer (3). ０ and the second voltage difference V １ and the device comprises an electric circuit (5) configured to apply the first voltage difference V ０ and the device absorbs the incident laser radiation (1) when the second voltage difference V １ reflecting or transmitting the incident laser radiation (1) when A device characterized in that

2. The metal underlayer (3) is not discontinuous and the device is １ 2. The device according to claim 1, which reflects the incident laser radiation (1) when

3. The structured metal overlayer (2) is spaced a distance a apart and has a width s (where s+a<λ ｉ 3. The device according to claim 1, wherein the device is formed by a plurality of metal strips.

4. The width s of the metal strip is s=λ ｉ / 2. The device according to claim 3, wherein n±30% is the refractive index of the semiconductor layer (4).

5. The width s of the metal strip is s=3.λ ｉ / 2. The device according to claim 3, wherein n±30% is the refractive index of the semiconductor layer (4).

6. 6. The device according to any one of claims 1 to 5, wherein the contact between the semiconductor layer (4) and the structured metal upper layer (2) and the metal lower layer (3) is made by a Schottky contact or by the introduction of an insulating layer.

7. The thickness L of the semiconductor layer (4) is λ ｉ 7. The device of claim 6, wherein the .lambda. / 30±30%.

8. 8. A method for manufacturing a device according to claim 7, wherein the thickness L of the semiconductor layer (4) is selected by digital simulation so that the device operates in a non-dispersive regime.

9. The quantum wells or the barriers separating the quantum wells are electronically doped, and the first voltage difference V applied by the electric circuit (5) between the structured metal upper layer (2) and the metal lower layer (3) ０ If the device [Equation 2] and the central wavelength λ is equal to the resonant mode of the optical microcavity. ＩＳＢ and the quantum well is selected so as to exhibit no absorption with respect to the second voltage difference V １ When the charge of all the quantum wells of the semiconductor layer (4) is reduced, the device exhibits absorption for the resonant modes and the energy [Equation 3] 9. The method of claim 8, wherein the quantum wells are selected to exhibit no absorption with respect to .

10. The thickness L of the semiconductor layer (4) is １ is less than the maximum voltage supported by the contact between the semiconductor layer (4) and the structured metal upper layer (2) and the metal lower layer (3).

11. The thickness L of the semiconductor layer (4) is １ 11. The method of claim 10, wherein is selected to be less than 10V.

12. The thickness L of the semiconductor layer (4) is λ ｉ 7. The device according to claim 1, wherein the axial length is equal to 10±30%.

13. The semiconductor layer (4) comprises a stack of coupled quantum wells consisting of a narrow undoped well called an active well separated by a first barrier from a wider well called a reservoir, the coupled quantum wells being separated on each side from a second barrier wider than the first barrier, the first barrier or the reservoir of each coupled quantum well being electronically doped and having a central wavelength λ close to a resonant mode of the optical microcavity. ＩＳＢ the active well is selected so that it exhibits intersubband absorption at The first voltage difference V applied by the electric circuit (5) between the structured metal upper layer (2) and the metal lower layer (3) ０ In this case, the charges arising from the doping are transferred by tunneling into the narrow undoped well, and the device is [Equation 4] and the central wavelength λ is equal to the resonant mode of the optical microcavity. ＩＳＢ and the coupled quantum well is adapted to exhibit no absorption with respect to the second voltage difference V １ when the device is at the center wavelength λ equal to the resonant mode of the optical microcavity. ＩＳＢ shows absorption at the energy [Equation 5] 7. A device according to any one of claims 1 to 6, wherein the coupled quantum wells are adapted to exhibit no absorption for .gamma.

14. The thickness L of the semiconductor layer (4) is ０ 14. The device of claim 13, wherein is selected to be 2 V or less.

15. The device according to any one of claims 3 to 5, wherein the semiconductor layer (4) is structured in such a way that the stack of quantum wells is located only directly beneath the metal strips of the structured metal overlayer (2).

16. The device according to any one of claims 1 to 7, 12 to 15, wherein the quantum wells of the semiconductor layer (4) are made of GaAs / AlGaAs, InGaAs / AlInAs, InAs / AlSb, Si / SiGe, GaN / AlGaN.

17. The device according to any one of claims 1 to 7, 12 to 16, wherein a characteristic dimension of the device is less than 200 μm.

18. The metal underlayer (3) is discontinuous and the device is １ 18. The device according to any one of claims 3 to 7, 12 to 17, wherein the device is structured to be transparent to the incident laser radiation (1) when

Images