1. Introduction

Semiconductor array of sub-wavelength size can act as not only the optical grating and dielectric cavity medium, but also the transition medium to convert light into electricity, or vice versa. The multi-role advantages have been demonstrated in the visible light range by nanometer sized semiconductors particularly the nanowire arrays [1–3]. Firstly, with the radial size smaller than the light wavelength, the surface reflection of NW array is drastically eliminated, which is commonly lower than that of traditional anti-reflecting surface [4]. Secondly, nearly 100% absorption can be reached in the nanowire forest due to the optical mode resonance and the near-field coupling among periodic NWs with high refractive index [5,6]. Thirdly, the length direction of NWs serves as one-dimensional channel in the length direction for efficient collection of photo-generated carriers [3]. While flourishing progresses have been made on solar cells, lasers and photodetectors with function region all composed of sub-wavelength semiconductors, relevant studies are rarely reported in the infrared range.

Due to its mature growth and processing technologies, AlGaAs/GaAs quantum well infrared detector is a viable candidate for sensitive imaging in the long and very long wavelength range, but it is restricted by the ISBT selection rule and low quantum efficiency of n-doped QWs [7–10]. Grating structures are generally constructed which typically enables absorbing around 20% of the incident light [11]. The quantum efficiency could be doubled when heavily doped QWs are coupled by the crossed gratings with waveguide layers [12]. In recent years, photonic crystal structures as well as the plasmon microcavities have been adopted to improve the performance of QW detectors [13,14]. With the capability of confining the light in sub-wavelength scale, absorption in a very few QWs can be enhanced by more than ten times at certain resonant wavelengths [15,16]. Critical coupling status was then obtained for perfect light trapping in the active region when the ohmic loss is unneglectable [17,18]. On the other way, the metallic cavities were designed as sub-wavelength antenna patches to reduce the physical area of the device with much larger photon collection area [19]. The wired patches sandwiched with QWs show not only increased responsivity, but also the ability to work at room temperature for detecting long wavelength light [20]. Although these photonic structures exhibit superiority in controlling the light propagation in QW devices, efficient infrared detection still relies on several unfavorable factors. For instance, the active region should be very thin for the plasmon cavities, this limits the periods of QWs (1-5 periods) and results in low resistance of the detector in dark condition. Also the QWs are usually doped with high concentration (≥1 × 10¹⁸cm^-3) to reach desirable absorptance, or relatively high electrical field (2-5 V/µm) is applied to raise the conductive gain. All-dielectric subwavelength array offers a solution to avoid those deficiencies.

In this paper, we design and demonstrate a low filling factor III–V multiple-quantum-wells absorber with absorption-enhanced working at 8.7 µm. The micropillar array with the optimal geometry has a minimum reflectivity of 1.9% and a high absorption quantum efficiency of 49.5%. In addition, we discuss the influence of structural parameters on the main resonance peak position, which is expected to design infrared absorbers of other wavelengths with multi-period QWs.

2. Theoretical models and methods

As sketched in Fig. 1(a), two dimensional array of cylindrical pillars is designed to enhance the absorption in quantum wells with reduced volume ratio for infrared light centered at 8.7 µm. This is achieved by diffracting and condensing the incident light in the sub-wavelength AlGaAs/GaAs columns with high refractive index of 3.3. The inset of Fig. 1(a) shows the cross-sectional view of the periodic columns. The height, diameter of micro-pillars and period of the array are defined as L, D, and P, respectively. The diameter and period will be optimized by numerical simulation. The pillar height was set to 4.85 µm which matches the thickness of active region consisting of (50 QW stacks) and the contact layers. Compared with the plasmonic cavity that contains typically no more than 5 stacks of QWs [15,16,21], here, the thick active region in micro-pillars holds advantages both in enhanced light absorption and reduced dark current.

 

Fig. 1. (a) 3D schematic diagram of micropillar array and the inset show that schematic diagram of the cross section of a periodic unit. (b) Relative permittivity of the anisotropic and isotropic media of GaAs quantum well materials. (c) The energy band diagram and the confined levels of AlGaAs/GaAs single quantum well.

Download Full Size | PDF

According to the selection rule of intraband transition in n-type quantum wells, only the photons with electric field component perpendicular to the x-y plane can be effectively absorbed, so anisotropic dielectric parameters should be adopted in simulating the optical properties of the micro-pillar array. The relative permittivity of quantum wells can be expressed as: ε_QW= diag(ε_x,ε_y,ε_z). GaAs is a transparent medium in long-wave infrared, and the dielectric constant of GaAs changes slowly with wavelength [22]. For the polarization component of the electric field in x-y plane, the light response of the quantum well is consistent with that of GaAs (ε_x=ε_y= ε_GaAs) [23]. For the polarization component of the electric field in z direction, the quantum well subband transition is excited, and ε_z have Lorentz linear [23–25].

Figure 1(c) presents the band diagram and energy levels of single AlGaAs/GaAs quantum well by solving the Schrödinger-Poisson equations with Nextnano++ software package. When the conduction band offset is set as 0.22 eV, quantized energy levels are formed in the 5.8 nm wide QW with the inter-level difference of 0.143 eV corresponding to the photon wavelength of 8.7 µm.

Optical simulation is done by solving the Helmholtz equation in the Lumerical FDTD Solutions 2020 with the perfect matched layer (PML)conditions along the z-axis and the periodic boundary conditions adopted in x and y directions. The total absorption of the device can be calculated by A = 1-R-T, where R, T is the reflectance and transmission of the array.

3. Results and analysis

It’s known that the peak wavelength of cavity mode is mainly determined by the diameter of the pillars [26,27]. As plotted in Fig. 2(a), the HE₁₁ resonant wavelength varies in a wide range from 3 µm to 12 µm when the diameter increases from 1 µm to 3.5 µm. Also, period affects the resonance peak, a global scan was performed to find out the optimal D and P for the maximum absorption at 8.7 µm. As shown in Fig. 2(b), the absorptance of QWs is simulated to change for several times when the diameter and period varies between 1.8-2.8 µm and 3-6 µm respectively. Certainly, the alignment between the peak wavelength of optical mode and that of the electronic transition should also be responsible for this numerical finding.

 

Fig. 2. (a) Dependence of HE₁₁ mode peak wavelength on the diameter of micro columns. The period of array is optimized for each diameter. (b) The absorption rate of QW column arrays at 8.7 µm with the diameter range from 1.8 to 2.8 µm and the period from 3 to 6 µm. (c) The red dashed line and red solid line represent the absorption rate of the 45° edge facet coupled QWs and the QWs in micro-pillar array with optimal sizes respectively; the black solid line represents the relative enhancement of absorption in the micro-pillar array compared to that of the 45° standard geometry; the insets show the optical modes and their electric field intensity profiles at the two main absorption peaks displayed above correspond to the mode excitation of HE₁₁ and HE₁₂ with the azimuth mode excitation, m = 1 and the radial mode excitation, n = 1,2. (d)The black solid line represents the integral of |Ez| at different periods with a fixed diameter of 2.2 µm under the same photon flux; the solid blue line represents the ratio of the |Ez| integrals of the active area in a single pillar. (e) and (f) are the cross-sectional diagrams of the electric field vector and field strength under the structures of D = 2.2 µm, P = 4 µm and D = 2.2 µm, P = 28 µm, respectively. The solid black box outlines the micron pillar and the QW active area is squared by the red dashed box.

Download Full Size | PDF

The highest absorption of micro-pillar array can reach 58.5% for incident light of 8.7 µm with D = 2.2 µm and P = 4 µm. For comparison, Fig. 2(c) presents the absorption spectra of 50 period QWs in column array and in 45° edge facet geometry. the latter can be calculated by the following formula [28].

To further discuss the inter-pillar coupling effect, we investigate the variation of the Ez integral in the active region for different periods of the array at a fixed diameter of 2.2 µm under the same photon flux. The black data points in Fig. 2(d) show that The ∫|Ez| of active area reaches its maximum at P = 4 µm and then decreases rapidly. In contrast, when the |Ez| integral of the active region accounts for the largest proportion in a single microcolumn, the corresponding period is only 3.5 µm. The difference just explains that the resonance of the photonic crystal mode enables the array to achieve the maximum value of ∫|Ez| in the active region at a relatively larger period. Of course, when the period is very large, the obvious antenna effect of the single micro-column can be seen through the electric field x-z cross-sectional view in Fig. 2(f).

Figure 2(e) shows the electric field vector and |Ez|’s intensity distribution diagram of five periods x-z section with the first-rank structure. We can see that the direction of plane wave incident light is effectively changed by sub-wavelength columns, furthermore, the photonic crystal mode strongly enhances the Ez vector perpendicular to the quantum well plane, which is crucial for coupling with the eletronic transition in n-type quantum wells. In addition, unlike the plasmonic cavity mode where the field strength is decaying rapidly above the metal surface [30,31], the electric field of the light is mainly concentrated at height of 1.2 µm ∼ 3.5 µm overlapping with the active area of the quantum wells. As a result, the intensity of the horizontally propagating light namely absorptance has been increased by 4.84 times by the all-dielectric self grating and the photonic crystal mode.

The QWIP layer sequence was grown on the GaAs substrate by molecular beam epitaxy(MBE), followed by a 200-nm-thick GaAs buffer layer, a 300-nm-thick Al_0.5Ga_0.5As sacrificial layer, a 1.2 µm thick GaAs bottom contact layer, the 3.35 µm thick active region, and a 0.5 µm GaAs top contact layer. The active region was composed of fifty 5.8nm-thick GaAs wells doped with 3 × 10¹⁷cm^-3 Si donors in the center 2-nm-thick region alternate with fifty-one 60nm-thick Al_0.24Ga_0.76As barrier layers. The bottom and top contact layer were Si-doped with concentration of 1 × 10¹⁸cm^-3 and 1.5 × 10¹⁸cm^-3 respectively.

To prepare all-dielectric micropillar arrays, we use dense SiN_x instead of metal masks [32]. The etching exchange ratio between SiN_x and GaAs is 1:6. The minimum thickness of SiN_x we need is 850 nm for etching 4.85 µm thick GaAs. To obtain the patterned SiN_x, the thickness of the photoresist (BCI3511) has to be more than 1.1 µm, since the etching exchange ratio between photoresist (BCI3511) and SiN_x is 1.3 : 1. Therefore the fabrication process is vital to achieve the goal of reaching the minimum linewidth of the pattern with micron-level thick glue under the lower limit of the accuracy of the ultraviolet lithography technology.

After washed with acetone, isopropanol, and DI water in sequence, high dense SiN_x of 900 nm in thickness was deposited on the quantum well wafer by ICPCVD at low temperature of 75°C. Next, BCI35111 was spin-coated on the SiN_x to the thickness of ∼ 1.1 µm and baked at 135°C for 3.5 minutes in order to increase the hardness of the photoresist. The wafer was exposed to UV light for 3.4s in hard contact mode and soaked in AZ400k developer for 11 s, then immediately rinsed with DI water for 60s. The development time was found critical in determining the shape and size of the photoresist columns. As shown in Fig. 3(a), perfect resin array with clear space between columns can be obtained, whereas the over-developing condition results in the incomplete pattern, and the columns remain interconnected under insufficient development. Reactive ion etching (RIE) was applied to etch the SiN_x film for 12 min under CHF₃ and O₂ flow, finally the micropattern was transferred to the semiconductor in Oxford PlasmaTherm ICP-RIE system using BCl₃/Cl₂/Ar/N₂ chemical gas. The optimized plasma conditions was set as 5 sccm of BCl₃, 5 sccm of Cl₂, 20 sccm of Ar and 3 sccm of N₂, where the introduction of N₂ allows the surface passivation of the micropillars to avoid the burring on the side wall [33]. The HF power and the ICP power was selected as 35 W and 350 W correspondingly, which can ensure the anisotropic etching and prevent the undercutting [34].

 

Fig. 3. (a) Photographs of the patterned AlGaAs/GaAs surface. (b)-(k) SEM images of the QW pillar arrays in different sizes. The periods of the arrays are 4.0 µm (d), 4.5 µm (e), and 5.0 µm (f) respectively with D = 2.2 µm. The diameters of the pillars are designed as 2.5 µm (g), 3.0 µm (h), and 3.5 µm (k) respectively with P = 5.5 µm.

Download Full Size | PDF

The SEM images of micropillar arrays with different diameters and periods are presented in Figs. 3(b)–3(f). The geometric features of the large area photonic structures will heighten the intraband absorption in quantum wells in several aspects. i) The diameter is nearly same at the top and bottom ends of the columns, which ensures the invariability of HE₁₁ peak wavelength. Cone type pillars with wide angles are often prepared in the fields of solar energy harvesting as well as the perfect absorber in which broadband light trapping is required [35–37]. Here identical size of the subwavelength structures is prefered to maximize the resonance effect since the narrowband transition of quantum well could be entirely covered by the low-order cavity mode. Also ii) the diameter fluctuation among different columns is less than 5% which guarantees the resonant coupling. iii) The roughness of the bottom surface is only about 2.0 nm measured by AFM. This, together with the smooth sidewalls certify the ideal effects of dielectric cavity mode.

One can note that hollow internal structure is formed in the micro-pillars as shown in the inset of Fig. 4(a). Due to the diffraction effect, the center of the photoresist column top is exposed and developed to form a small hole (Fig. 3(a)), which is transferred to the semiconductor column after being etched twice during the etching process. Firstly, as the diameter decreases close to the ultraviolet wavelength, the strengthening of diffraction causes the hole diameter to increase. As shown by the red line in Fig. 4(a), the hole diameter rapidly increases from 0.40 µm to 1.128 µm when the diameter of the pillar decreases from 3.5 µm to 2.1 µm. Besides, as a periodic array of photonic crystal structure, the distance between neighboring columns is close enough, so that the interference effect further enhances the diffraction. The blue line in Fig. 4(a) represents the diameter of the pore varies in a range from 0.69 µm to 1.25 µm while the period decreases from 5.5 µm to 3.6 µm. As a result, the electrical volume of the array is further reduced via the holes.

 

Fig. 4. (a) The diameter of the hole in micro-pillars along with the array period and the pillar size, the inset shows x-z sectional view of the hollow structure in a pillar. (b) The solid lines are reflection spectra measured on micro-pillar array (in black) and planar surface (in blue) of QW samples respectively, while the dashed lines represent the simulated counterparts. (c) The measured reflection spectra of micro-pillar arrays with different diameters and periods. (d) The relationship between the minimal-reflectance wavelength and the diameter of the micro-pillars.

Download Full Size | PDF

Micro spectroscopy is generally applied to characterize the optical property of the sub-wavelength structures [35,38]. Here, a Fourier transform infrared spectrometer (Thermo-Nicolet 6700 FTIR) was used to study the local reflection and transmission behaviors of samples within wavelength band from 6 µm to 15 µm. The aperture size was set as 100 µm × 100 µm. The area of each micropillar array is 2mm × 2 mm. All the experimental spectra were normalized with a gold mirror and the air for eliminating the influence of the environment.

As shown in the Fig. 4(b), the average reflectivity measured on the quantum well film is 28.5% as expected, where the oscillation of the curve with wavelength results from the interference between the upper and lower surfaces of epitaxial layer. Compared with the thin film structure, the array has an obvious anti-reflection effect without additional AR coating [39]. On the one hand, the consistency of simulation and experimental results further demonstrates that the modulation effect of the array with the optimal size on specific wavelengths, which is manifested in the following two aspects: 1) three troughsexist in the bands of 6.6∼7 µm, 8.5∼9 µm and 10∼11 µm; 2) the measured minimum reflectance is 1.9% at the wavelength of 8.7 µm. In the range of long-wave infrared detection, the thick film potentially has flaking and delaminating problems during the complex growth process due to the internal stress of materials. It may lead to the low yield of FPA even though the traditional multi-layer coating structure can obtain a lower AR effect [39–41]. For comparison, the array structure formed by etching does not encounter puzzles above while possessing low reflectivity. On the other hand, as shown in Fig. 4(c), the minimum reflection value can be obtained at different wavelength by adjusting the diameter of the microcolumn. Valley 2 varies in a wide range from 8.7 µm to 10.5 µm as the diameter increases from 2.2 µm to 3 µm. These three valleys varied with the diameter are more intuitively reflected in Fig. 4(d). The 6∼14um band is almost covered. It is expected to achieve antireflection at any wavelength over the long wavelength band by adjusting the array size.

To evaluate the absorbance of the QWs in micro-pillar arrays, it’s necessary to assess the light loss caused by various experimental factors during the spectrometric measurement. First, the collection angle of the micro-FTIR is 71° in our setup both for reflection and transmission study. As illustrated in the inset of Fig. 5(a), numerical simulation has been performed to calculate the light leakage beyond the cone angles, in which the planar light with periodic boundary conditions was replaced by the Gaussian light source with the perfectly matched layers (PML). Second, hollow configuration was adopted in the simulation to reproduce the actual structure of micro-pillars shown in the inset of Fig. 4(a); Third, the SiN_x layer on top of each GaAs columns may cause absorption in the long wavelength range, which can be calculated with the parameters from J. Kischkat’s work [42].

 

Fig. 5. (a) The simulated light loss during the micro-FTIR measurement on the QW pillar array. The dashed lines in green, and purple represent the reflected rate(L_Relf) and the transmitted rate(L_Tran) in 180°≥θ≥71° respectively; The solid blue line, L_R + T= L_Relf+ L_Tran, includes the sum of the leaked light that goes beyond the collection angle of the monitors of 71°. The absorption of SiN_x is plotted in red line. The total light loss is shown by the black line. (b) The absorption by QWs in the pillar array is shown in black solid line that is estimated by subtracting the simulated light loss from the experimental data 1-R_71°-T_71°.) The black dashed line represent the absorption rate of the QWs in micro-pillar array The sizes of micropillar array are D = 2.2 µm and P = 4.0 µm. (c) The dependence of the peak absorption wavelength on the diameter and period of the micro-pillar arrays, where the solid line and the dashed line represent the experimental and theoretical results respectively. (d) The normalized absorption spectra of the QW pillar arrays with different periods. The insets show the electric field distributions in the x-y cross sections of the pillars at the peak wavelengths.

Download Full Size | PDF

Figure 5(a) shows the calculated light loss and its spectral properties caused by various experimental factors: 1) The reflection collected beyond 71 degrees is estimated to be less than 3% over 6-12 µm wavelength range. This firstly benefits from the superior AR performance of the sub-wavelength array. Meanwhile, the Scattered light under the large angle can also be suppressed via the strong lateral confinement formed by the photonic mode within the micropillars. In particularly, the light leakage of reflectance is as low as 1.3% at the resonant wavelength of 8.7 µm. 2) In contrast, the high angle transmission will lead to considerable light loss during the micro-spectroscopic investigation. The leakage rate could be more than 30% in the short wavelength side according to the simulation. This is due to the combined effect of the array on the incident light, namely to guide the entrance in the vertical direction and to cause divergent diffraction in the columns. In the meantime, the impact of resonant modes can also be clearly seen in the spectrum that is the peak transmission at 6.5 µm and the minimum transmission rate of 6% at 8.7 µm for intraband absorption in quantum wells. 3) SiN_x has noticeable absorption when the wavelength is longer than 9 µm, but it is no more than 1.2% in the range of 8∼9 µm. And this remnant SiN_x will be removed before preparing the top electrode contact.

The absorption spectrum of the array sample with optimal sizes is obtained by 1-R_71°(λ)-T_71°(λ)-L_total as shown by the black solid line in Fig. 5(b), where R_71°(λ) and T_71°(λ) are the reflectance and transmittance measured via Thermo-Nicolet 6700 respectively. The absorption spectrum of the QWs has a significant enhancement at 8.7 µm, which is quite consistent with the simulated spectrum based on the real structure. Among them, the actual absorption of the QWs’ array at the 8.7 µm is 49.5%, twice that of the traditional grating structure [9], slightly lower than the simulated value of 58.5%. The difference may come two factors, may be that the Gaussian light source cannot fully simulate the convergence characteristics of the actual light source or the fitting of the quantum well material parameters is a bit smaller than the experimental value.

Figure 5(c) confirms that the diameter of the microcolumns plays a major role in the absorption peak position within the long wave band. The resonance wavelength varies in a wide range from 8.5 µm to 12 µm under the control of guided mode when the experimentally prepared diameter increases from 2.1 µm to 3.5 µm, which is almost the same as the simulated result. In addition, the results of theory and experiments have proved that the period of array also has a weak effect on the red shift of the formant. The peak wavelength varies in a wide range from 8.45 µm to 9.47 µm while the array period increases from 3.5 µm to 5.5 µm. The intuitive explanation of period’s influence on the absorption spectrum is shown in Fig. 5(d), which shows the normalized absorption curve for different periods at fixed diameter. In the inset, the position of the strongest electric field gradually shifts outwards as the period increases, resulting in the weakening of the resonance coupling of adjacent micropillars accompanied by spectral shifting red shifting. Meanwhile, as the period decreases, the radiative loss of the mode decreases, which proves the adjustment of the period further enhances the absorption effect of the array.

Besides compressing the optical field to enhance the effective absorption of the quantum well, the impact of the subwavelength structure on the electrical property of the detection device is also worthy of attention. Table 1 lists the representative progress in boosting the performance of quantum well infrared detection devices with different optical coupling schemes. Selective etching of the photosensitive material results in considerable reduction of the electrical area of the device which is described by the filling factor(F). Among them, the antenna metal structures proposed by D. Palaferri and Y. Todorov brings the filling factor down to 0.13, so that the dark currents of the devices are expected to decrease by several times [43]. Further, the combined effects of the subwavelength couplers on the signal to noise ratio of QW detectors can be evaluated via M=η*(1/F), where the external quantum efficiency (η) represents the optical and photoelectric merit that also includes the influence of the doping level, the period numbers, as well as that of the working conditions of the device.

Table 1. The principles, structures, parameters and the effects of different approaches applied on semiconductor QWs for long wavelength detection

View Table

In the early days, the quantum efficiency of the 45° edge-coupled QW detector reached 16% with very-high doping concentration in 50-period of quantum wells [10]. Subsequently, the efficiency of 29% was achieved by introducing the waveguide and grating coupler structure. Recently, various metal cavities were designed to enhance the light absorption of QWs, the external quantum efficiency can be raised up to 61% in single QW device [21]; moreover, the antenna structure helps to lower down the filling factor, then the M value was optimized as high as with 99%. In this paper, the M is estimated as 83.3% by using the micro-pillar array integrating the absorber and the coupler. The photoconductive gain is 0.4 from A. Rogalskis’ work [11]. It’s worth noting that the near-field feature of the plasmonic modes in metallic cavity generally requires a thin active region (less than 5 periods), also high-dose doping in QWs or high voltage was often adopted to increase the quantum efficiency. In comparison, the dielectric cavity mode and its field distribution allow for much thicker active layer that contains 50 periods of QWs and relatively lower doping concentrations, both of which will bring additional advantages to improve the electrical merit of the device that is not involved in the M factor.

With the strong absorbency of columnar array been verified, the top electrode contact will be a major issue to be addressed to realize it’s photoelectric merit. One possible scheme is to deposit oxide and disperse polymer around the pillars if the transparency of polymer and stable interface at the sidewall of pillars could be obtained [45]. Another choice is to connect the pillars with thin walls as reported in literatures [20,21], in which the top contact can be made in traditional way. The dark current is another key point that may neutralize the optical benefit of the pillar array since it has large area of sidewall surface. Here for QWIP, as an intra-band device, its photoelectric function relies on the majority electrons which can hardly be affected by the surface defects. Moreover, the charged surface states at sidewall tends to form radial potential distribution that facilitates the electrons to transport in the core of the pillars [46]. This prevents the large surface area from deteriorating the electrical property of micropillars, so that high performance of detectivity could be expected for the columnar array of QWIPs.

4. Conclusions

We proposed a micro-array structure of semiconductor pillars that integrates the capacities of grating, compressing and absorption for incident light, which will elevate the photoelectric efficiency of quantum well infrared detectors while diminishing the electrical area of the devices. FDTD calculation has been performed to simulate the optimal sizes of micro-pillar arrays for light wavelength centered at 8.7 µm. With the diameter and period of the pillars set as 2.2 µm and 4 µm respectively, the peak absorption of QWs in arrayed pillars is predicted to be enhanced up to 5.1 times compared with that of the 45° facet edge coupled structure. The simulation discloses that the normal incident light will be guided in the pillars by HE₁₁ resonant cavity mode to form strengthened E_z electrical field perpendicular to the QW plane, which is essential for exciting the inter-level transition in conduction band. High uniform cylindrical arrays with 50 periods of AlGaAs/GaAs QWs in each column are fabricated in 2 mm 2 mm area by UV exposure and ICP etching. Optical merits of the geometric features are revealed by spectroscopic investigation: i) The peak wavelength of resonant cavity mode can be adjusted in a wide range mainly by the diameter of the pillars. Meanwhile the optical coupling between neighboring pillars is modulated by the spacing, namely the period of arrays. ii) Both effects contribute to the ultralow reflectivity as 1.9% and the high absorption rate of 49.5% to the light at 8.7 µm measured on the arrays with optimal sizes. In addition, several advantages are expectable for the electrical merit of the dielectric structures: 1) The electrical area is reduced by more than 4 times relative to its planar counterpart; 2) The thickness of active region is 3.3 µm which is preferable to contain more narrow QWs and suppress the dark current; 3) The doping concentration in the QWs is fairly low as 3 × 10¹⁷cm^-3. All of these are mighty prefered for semiconductor QWs toward high sensitive infrared detection. Last but not least, our scheme avoids the metal planes that are needed in the widely studied plasmonic cavities, this allows it to be compatible with the focal plane array (FPA) configurations of infrared detectors, and the line-widths appeared in this study are feasible for the traditional lithography.