2. Materials and Methods
Figure 1a–c sequentially illustrates the front view with an energy band diagram, a three-dimensional view, and a top view of the long-wavelength GaAs/AlGaAs QWIPs. The quantum well region consists of 10 periods of Al
0.27Ga
0.73As/GaAs single quantum wells. This material was employed in the research conducted by Jeong in 2005 [
14], with the experimental devices operating in the wavelength region around 8 μm. The author has verified that reliable etching depth can be achieved for this material using wet etching methods in experiment. In our QWIPs, the quantum well has a width of 4.8 nm and a barrier width of 50 nm. The energy difference between the ground state energy level
E1 and the first excited state energy level
E2 for the inter-subband transition is 153 meV. The active region thickness is
h3 = 0.6 μm, and the thickness of the bottom and top contact layers is
h1 =
h4 = 1.1 μm. The contact layer is composed of GaAs doped with Si at a concentration of 1 × 10
18 cm
−3. The gold stripe grating has a width of
s = 0.5 μm and a thickness of
h2 = 0.2 μm. The unit cell of the chiral metasurface array is composed of two structurally identical L-shaped gold structures with central symmetry. The unit cell period is
Px =
Py = 2.5 μm. The double L-shaped structure has a thickness of
h5 = 0.2 μm, a width of
a = 0.4 μm, a short side length of
b = 1 μm, a long side length of
c = 1.8 μm, and a separation distance of
d = 0.8 μm.
The fabrication of the proposed chiral metasurface-enhanced QWIPs has been demonstrated to be feasible in previous works, where they are hybridized with a fanout circuit through standard flip-chip bonding processes [
15]. Initially, dielectric layers are grown on GaAs wafers by molecular beam epitaxy (MBE). Subsequently, wet etching is employed to obtain photosensitive mesas, controlling the mesa depth to reach the top contact layer. Inductive coupled plasma etching is used to etch the bottom contact layer. Then, Au is deposited via electron-beam evaporation to form gold stripe grating and a metal reflection layer. Finally, the double L-shaped metal array can be formed through electron-beam evaporation and lift-off processes.
In this study, the Finite-Difference Time-Domain (FDTD) method was used for calculations. The relative permittivity of gold was obtained by the Drude model [
16]:
In the equation, ω represents the frequency of the incident wave, ωp = 1.37 × 1016 rad/s is the plasma frequency, and ωτ = 4.65 × 1013 rad/s is the damping constant of gold.
The quantum well active region is an anisotropic material, and its dielectric constant can be represented as diag(
εx,
εy,
εz), where
εx and
εy follow the material model of GaAs, as shown in the following equation [
17]:
In this equation, ε∞ = 11 represents the high-frequency dielectric constant, fL = 8.757 THz represents the longitudinal optical phonon frequency, and fT = 8.056 THz represents the transverse optical phonon frequency. γT = 72 GHz is the phonon damping constant.
In the z-direction, the equation is given as follows:
where
fw represents the peak frequency of the response, while
γw = 0.05
fw corresponds to the damping factor [
18].
ε′ denotes the real part of the complex permittivity at the peak wavelength, which is typically assigned a value of 0.5 [
19].
The coupling efficiency (CE) of the active region in the quantum wells structure can be calculated using the following equation [
13]:
The equation includes E0 as the initial electric field and Ez as the electric field component in the z-direction. In this study, CPER represents the ratio of coupling efficiency between LCP light and RCP light. A higher value of CPER indicates better circular polarization detection performance of the detector.
3. Results and Discussion
Figure 2 shows the real and imaginary parts of the relative permittivity in the
z-direction for the active region of the designed quantum well structure.
Figure 3a depicts the absorption spectrum of the long-wavelength GaAs/AlGaAs QWIPs, where the black line represents the LCP light source and the red line represents the RCP light source. Three peaks are observed at wavelengths of 7.5 μm, 7.9 μm, and 8.0 μm, corresponding to the microcavity mode [
20,
21], the SPP mode [
22,
23], and a hybrid mode [
24] of the two. In the SPP mode at 7.9 μm, the maximum absorption for LCP light source is 0.8, while the absorption for RCP light source is only 0.25. Across all three operating modes, the device exhibits significantly higher absorption for LCP light sources compared to RCP light sources.
Figure 3b displays the coupling efficiency of the active region of the quantum well as a function of wavelength. When the LCP light source is incident, the coupling efficiency reaches 2700% at 7.9 μm and 870% at 8.0 μm. On the other hand, when the RCP light source is incident, the coupling efficiency at 7.9 μm and 8.0 μm is 60% and 540%, respectively.
Figure 3c represents the CPER of this QWIP device with a maximum value of 45 at 7.9 μm.
Figure 3d–f show the |
Ez|/|
E0| (where
Ez represents the electric field component in the
z-direction and
E0 represents the initial electric field) under LCP light illumination for the three operating modes. This is the
xoz cross-section at
y = 0, and the
z-axis coordinates are zero on the upper surface of the bottom gold base. When the incident light passes through the top chiral metasurface layer, the microcavity mode is excited. Selected by the chiral structure, most of the RCP light is reflected, while a significant amount of LCP light transmits into the QWIPs and excites the SPP mode in the bottom gold stripe grating layer, where it is subsequently absorbed. The electric field enhancement of the microcavity mode is mainly concentrated at the top of the device, while the SPP mode exhibits enhanced electric fields over the entire semiconductor layer. Although the hybrid mode is a combination of the two, the excited electric field is weaker and primarily concentrated on the metal surface. Among the three operating modes, the dominant mode is the SPP mode. Under the SPP mode, the maximum enhancement of |
Ez|/|
E0| is 8.7 times with a higher amplitude and a broader range compared to the other modes, which is consistent with the higher absorption rate observed in
Figure 3a for the SPP mode.
Figure 3g–i show the |
EZ/
E0| in
xoz cross-section (
y = 0) under three operating modes when RCP light is incident, setting the upper surface of the bottom gold base as the coordinate zero plane. In the microcavity mode at 7.5 μm, although the electric field enhancement amplitude of RCP light is smaller than that of LCP light, its radiation range is wider, resulting in higher overall absorption of the QWIPs, consistent with the results in
Figure 3a. After passing through the top double L-shaped chiral structure, circularly polarized light of different handedness experiences different phase shifts. LCP light effectively excites SPP waves, enhancing the localized electric field, while SPP waves excited by RCP light are suppressed. At 7.9 μm and 8.1 μm, the electric field enhancement of RCP light is smaller than that of LCP light. With the combined effects of the top chiral metasurface and the bottom gold grating, the QWIPs can absorb most of the LCP light and a small amount of RCP light. The significant difference between the two results in a remarkable CPER of up to 45, demonstrating the device’s capability to effectively discriminate the handedness of circularly polarized light.
Due to the dominant role of SPP waves excited by the bottom gold stripe grating in the three operating modes, this study investigates the influence of the grating width (
s) and thickness (
h2) on the device performance. As shown in
Figure 4a,b, we examine the relationship between the coupling efficiency (CE) and the response wavelength for different grating widths (
s). When RCP light is incident, the CE initially decreases and then increases at the SPP wave response wavelength of 7.9 μm as
s varies from 0.3 μm to 0.5 μm, reaching a local minimum at
s = 0.5 μm. When LCP light is incident, the CE reaches a local maximum at
s = 0.5 μm. It can be observed that at
s = 0.5 μm, the SPP effect is suppressed under RCP light and enhanced under LCP light, resulting in a significant difference in CE between the two polarizations. Therefore, the QWIPs exhibits the ability to distinguish the rotational direction of circularly polarized light. Due to the unchanged parameters of the top double L-shaped chiral structures, the response wavelength and intensity of the microcavity effect excited at 8.1 μm tend to stabilize, regardless of whether it is RCP light or LCP light. From
Figure 4c, it can be visually observed that the CPER reaches its maximum value at
s = 0.5 μm. It should be noted that when
s is between 0.3 μm and 0.4 μm, the CPER value between 7.6 μm and 8 μm is less than 1, indicating that under this condition, the CE for RCP light is stronger than that for LCP light.
Figure 5 illustrates the impact of different grating thicknesses (
h2) on the QWIPs. As shown in
Figure 5a,b, under RCP light incidence, the CE initially decreases and then increases at the SPP wave response wavelength of 7.9 μm as
h2 varies from 0.1 μm to 0.3 μm, reaching a local minimum at
h2 = 0.2 μm. Under LCP light incidence, CE maintains a relatively high value in the range of
h2 = 0.2–0.3 μm. When
h2 is less than 0.2 μm, CE rapidly decays, resulting in weaker CE for LCP light compared to RCP light, which is consistent with the results in
Figure 5c. The device achieves the maximum CPER when
h2 = 0.2 μm. In addition to exciting SPP waves on the surface of the gold grating, the incident light also generates transverse diffracted light and reflected light, leading to interference phenomena [
25]. When
h2 and
s are small, the destructive interference caused by LCP light weakens the SPP effect, while the constructive interference caused by RCP light enhances the SPP effect, resulting in a CPER value less than 1. Changes in the grating structure parameters can affect the excitation of the SPP effect, making the selection of appropriate values for
s and
h2 crucial for device performance.
In this study, the impact of L-shaped structures on CPER was investigated through a comparison of individual L-shaped structures,
L1 and
L2, as well as the double L-shaped structure. The coupling efficiency curves for
L1 and
L2 structures are depicted in
Figure 6a,b. Due to their central symmetry, the coupling efficiency curves of
L1 and
L2 structures exhibit similarities. At a wavelength of 7.9 μm, the coupling efficiency (CE) for LCP light is 2400%, while the CE for RCP light is 1060%.
Figure 6c presents a comparison of the CPER between the double L-shaped structure and the single L-shaped structures. The illustration shows the coupling efficiency of the double L-shaped structure. In comparison to the single L-shaped structures, the double L-shaped structure does not exhibit a significant enhancement in the CE for LCP light. However, it significantly suppresses the CE for RCP light. The maximum CPER for the single L-shaped structure is only 3.2, whereas the CPER for the double L-shaped structure is 14 times higher than that of the single L-shaped structures. These findings indicate that double L-shaped long-wavelength GaAs/AlGaAs QWIPs can effectively improve CPER and enhance the circular polarization detection capabilities of the device and be applied in medical in spectral imaging [
26].
Subsequently, we analyzed the electric field distribution of the single L-shaped structure.
Figure 7a and
Figure 7b, respectively, present the |
Ez|/|
E0| of the
L1 structure under LCP and RCP light at the peak wavelength of 7.9 μm in
xoz plane. The maximum enhancement of |
Ez|/|
E0| for the LCP incident reaches 9.6 times, surpassing the 8.7 times observed for the double L-shaped structure in
Figure 3c. This is attributed to a considerable portion of incident light directly transmitting into the cavity through the uncovered top contact layer by the L-shaped structure, thereby enhancing the local field. However, this portion of incident light is not converted by the top chiral metasurface and the excited SPP effect is limited, resulting in an overall coupling efficiency lower than that of the double L-shaped structure. This also explains why the peak coupling efficiency in
Figure 6a,b is lower than 2500%. In
Figure 7b, the maximum enhancement of |
Ez|/|
E0| caused by the SPP waves reaches 7.4 times, which is weaker than that of LCP light. This confirms that the single L-shaped structure also possesses a certain ability to discriminate the rotation direction of circularly polarized light, although with a smaller difference between the LCP and RCP light compared to the double L-shaped structure. This explains why the CPER of the single L-shaped structure is lower than that of the double L-shaped structure in
Figure 6c.
The |
Ez|/|
E0| values of the
L2 structure under LCP and RCP light at the peak wavelength of 7.9 μm, as shown in
Figure 7c,d, are consistent with the enhancement amplitudes in
Figure 7a,b due to the central symmetry between
L1 and
L2 structure. However, the excitation position of the SPP mode is different: for
L1, it is on the right (left) side of the unit, while for
L2, it is on the left (right) side of the unit under LCP (RCP) light. This is because, as mentioned earlier, the incident light is transmitted through the top contact layer, not covered by the metal stripes. The mirrored symmetric electric field distribution also explains why the coupling efficiency curves of
L1 and
L2 in
Figure 6a,b are consistent. Therefore, through a comparative study of single L-shaped and double L-shaped structures, we can observe that the performance of the double L-shaped structure is superior than that of the single L-shaped structure in terms of the coupling efficiency and circular polarization discrimination capability.
In order to investigate the circular polarization selectivity of double L-shaped chiral structures, we manipulated the position of the top double L-shaped structure by symmetrically flipping it about the
x-axis, thereby modifying the device’s target polarization to RCP light.
Figure 8a illustrates the absorption spectrum of the flipped device, with an inset schematic of the right-handed chiral double L-shaped structure targeted towards RCP light. The numerical values of absorption peaks, peak wavelengths, and peak absorption ratios remain consistent with the results in
Figure 3a. However, RCP light plays a dominant role rather than LCP light.
Figure 8b and
Figure 8c, respectively, demonstrate the coupling efficiency curve and the CPER of the right-handed chiral double L-shaped structure. At the peak wavelength of 7.9 μm, the coupling efficiency under RCP light hits 2700%, which is significantly higher than the 60% under LCP light. Similarly, at the wavelength of 8.1 μm, the coupling efficiency of RCP light surpasses that of LCP light. This stands in stark contrast to the results in
Figure 3b. The peak CPER in
Figure 8c is consistent with that in
Figure 3c; however, the CPER here is obtained by calculating the ratio of coupling efficiencies between RCP and LCP light. These results confirm that this design can achieve free selection of detecting the target rotation of circularly polarized light by altering the layout of the double L structure, thereby exhibiting wide applicability.
In the application of QWIPs, it is essential to evaluate the performance of chiral metasurfaces with various incident angles, including coupling efficiency and CPER. Investigating the device performance under different incident angles allows for the assessment of angle independence, thereby providing guidance for the design and optimization in practical applications.
Figure 9a depicts the diagram of infrared light under an incident angle, and φ represents the angle between the incident light and the
z-axis.
Figure 9b demonstrates the CPER of the device under different incident angles ranging from 0° to 75°. It is observed that when the incident angle increases, the CPER slightly decreases. This is because an inclined incidence weakens the excitation of the SPP effect in the double L structure. Even under the incident angle of 75°, the CPER hits above 36, indicating that the device maintains high circular polarization selectivity under large incident angles. The results demonstrate that the device’s angular stability is minimally affected by different incident angles.
In addition, we calculated the coupling efficiency of the device under different incident angles ranging from 0° to 75°, as illustrated in
Figure 10a,b. For LCP light, the coupling efficiency significantly decreases with an increasing incident angle. When the incident angle φ = 75°, the coupling efficiency decreases to only 780%. However, when incident angles φ < 60°, the coupling efficiency remains at a relatively high level, exceeding 1000%. This indicates that the device maintains good coupling efficiency over a wide range of operating angles, demonstrating excellent angle independence. In contrast, under an RCP infrared incident, the polarization selectivity of the bottom gold stripe grating suppresses the excitation of SPP modes, resulting in generally lower coupling efficiency. Under normal incidence, the peak coupling efficiency under RCP light incidence is much lower than that under LCP light incidence. As the incident angle gradually increases, the coupling efficiency of the device continuously decreases. When the incident angle φ = 75°, the coupling efficiency drops less than 200%. The big difference in the coupling efficiency between LCP and RCP light confirms that the device has high circular polarization discrimination capability, which is consistent with the CPER results in
Figure 9b.