Metamaterial Engineering for Superior HgTe cQD Photodetector Performance

Highly responsive, low noise, and inexpensive photodetectors that operate in the mid‐infrared (MIR) wavelength regime are in high demand for applications ranging from fundamental science to large scale industries. However, simultaneously achieving all this in one device architecture is very challenging. In this work, mercury telluride (HgTe) colloidal quantum dot (cQD) based photodetectors are systematically improved by the introduction of new metamaterial designs. The new designs are found by utilizing simulations. Thereby the structures are optimized to increase the responsivity and simultaneously decrease the noise spectral current density. This is achieved by focusing on improving the photogenerated charge carrier collection efficiency while reducing the active material volume without altering the near unity absorption. A standard metamaterial perfect absorber architecture based on disc resonators is used as a starting point for the optimization process. By optimizing the carrier extraction through contact engineering, resulting in a narrow slot metamaterial, an overall ≈13‐fold responsivity and ≈345‐fold detectivity increase is achieved. The final metamaterial design reaches a responsivity of 16.2 A W−1 and detectivity of 6×108 Jones at a wavelength of 2710 nm. The analysis therefore provides a route to improve the responsivity and noise characteristics of mid‐infrared photodetectors based on cost‐efficient colloidal quantum dots.


Introduction
Photodetectors are essential components for various applications such as thermal imaging, [1] biomedical sensing, [2] DOI: 10.1002/adom.202303223spectroscopy, [3] environmental [4,5] and gas monitoring. [6,7]Given their great importance, ongoing efforts focus on enhancing their performance metrics, including responsivity and detectivity, while simultaneous reducing their cost.Achieving improvements across all these metrics is particular challenging for detectors operating in the low energy photon MIR wavelength regime. [8]ommercially available highperformance MIR-photodetectors are based on low band gap materials such as indium antimonide (InSb), [9] indium gallium arsenide (In 1-x Ga x As), [10] lead selenide (PbSe) [11,12] and mercury cadmium telluride (Hg 1-x Cd x Te) [13] or multi quantum well structures (MQW). [10,14]These material systems are well established and can even reach performances close to the physical limitations. [15]However, these materials and their fabrication are often expensive and rely on sophisticated equipment.Additionally, such detectors often require operation at cryogenic conditions that typically entails further costs. [16,17]romising alternatives to these materials and MQW structures are cQDs.They have the advantage that they can be fabricated and deposited using simple methods. [18,19]HgTe cQDs are among the most promising materials for detectors operating the infrared regime. [20]This can be attributed to their absorption tuneability which ranges from the shortwave infrared (SWIR 1.5-2.5 μm) [21,22] to the MIR (3-5 μm) and even the THz [23,24] spectral wavelength range. [25,26]Although cQD materials hold great promise, they have unwanted limiting properties that have to be considered when designing a photodetector.Such limitations include low charge carrier mobility, [27] charge carrier traps and large 1/f noise. [28]The low mobility in combination with traps, which can act as recombination centers, can result in photogenerated electron-hole pairs recombining before reaching the contacts where they can be collected, thus decreasing the photoresponse. [27,29]ecently, what was referred to as band-like transport has been shown to occur in HgTe cQD clusters.It was observed that in small clusters of strongly coupled CQDs the mobility is increased compared to long range transport. [30]Audrey Chu et al. [22] utilized the observed effect to drastically increase the photoresponse while simultaneously reducing the noise in nanogap photodetectors.[33] Further, if the distances are extremely small, the mentioned mobility increase can be observed, leading to an additional increase of the responsivity.It has been also shown that the nanogap photodetectors exhibit a decrease of the noise spectral current density due to smaller 1/f noise since the active material volume is reduced. [22,34]While such nanogap photodetectors can drastically increase the internal quantum efficiency and device performance, efficient optical coupling to such nanogaps and achieving high absorption in the cQD layer is difficult and thus limits the responsivity and practicability.
A method to overcome these limitations is by combining thin cQD layers with absorption enhancement schemes such as metamaterials.37][38] While absorption enhancement strategies such as metamaterials can improve the absorption, the photodetector can still be ineffective if the photogenerated charge carriers must travel large distances and thus recombine before reaching the contacts.Demonstrations of combining optical metamaterial resonators with electrical contact lines have shown improved performance for devices based on novel materials such as 2D materials and cQD. [12,39,40]owever, metamaterial designs combining short carrier traveling lengths and high absorption are still scarce for lateral detector configurations.
In this work, we systematically improve the performance of metallic metamaterial enhanced HgTe cQDs photodetectors.We show that by optimizing the extraction of photogenerated charge carriers we can improve the responsivity and decrease the noise spectral current density while simultaneously maximizing the absorption.As an initial device a simple disc resonator-based metamaterial perfect absorber structure was combined with a thin HgTe cQD layer to increase the absorption.We then introduce electrical charge extraction in the form of straight interdigitated contact lines in between individual disc shaped resonators.These contacts were bent around the disc resonators in the following design to extract as many photogenerated carriers as possible.The contact lines were placed as close as possible to the areas with the strongest light absorption without distorting the metamaterial resonance, enabling more efficient charge carrier extraction and a volume reduction of active material, thus increasing the responsivity, and decreasing noise.Ultimately, to achieve a narrow slot metamaterial, the contact lines were merged with the resonators forming a meander-shaped slot metamaterial.This simulation supported systematic engineering of the carrier extraction of the metamaterial enhanced photodetector led to an ≈ 13-fold increase in responsivity, reaching 16 A W −1 at a wavelength of 2710 nm, and an average ≈ 50fold reduction of the noise current spectral density when comparing simple disc resonators with the slot metamaterial.These improvements resulted in a ≈ 345-fold increase of the specific detectivity reaching ≈ 6×10 8 Jones at room temperature and a wavelength of 2710 nm.Furthermore, simulations showed for all metamaterials good angular stability and little polarization dependent absorption.All presented design improvements are sup-ported by optical and electrical simulations giving a qualitative and intuitive understanding of concepts used to improve the photodetectors.This demonstration therefore enables the efficient design of future metamaterial-enhanced photodetectors operating at various wavelengths.

Device Architecture
The metamaterial cQD photodetector optimization process was carried out with a metamaterial perfect absorber architecture as visualized in Figure 1a.The devices were fabricated on a silicon (Si) substrate covered with a 250 nm thick thermally grown silicon dioxide (SiO 2 ) layer.On to this oxidized substrate the metamaterial stack was deposited consisting of a 100 nm thick gold (Au) backplane a 17.5 nm thick alumina (Al 2 O 3 ) spacer layer and a 60 nm thick gold (Au) metamaterial top layer.The resonating top layer was covered by a 5 nm thin alumina layer, see inset in Figure 1a.This thin insulating alumina layer that is only located on top of the contacts and resonating elements restricts the current flows through the absorbing QD layer to a lateral direction.Furthermore, this layer may reduce exciton quenching that has been observed in other works. [41]Lastly, the metamaterials are covered by a 60 nm HgTe cQD absorber layer that is passivated by a 120 nm thick PMMA layer.The photodetectors were fabricated in two sizes with an active area of 15×15 and 30×30 μm 2 .In the visualizations shown in Figure 1.In the visualizations shown in Figure 1 the cQD layer is only placed in between the contacts and resonators and the PMMA layer has been omitted to increase comprehensibility.The HgTe cQDs used in this work had an average diameter of ≈ 9 nm and an absorption onset at ≈ 4650 nm, see Figure S1c (Supporting Information).A detailed description of their synthesis, ligand exchange and the used deposition method can be found in the Section S1 (Supporting Information).
The metamaterial resonator layer was systematically improved to achieve near unity optical absorption and high carrier extraction efficiency at the same time.Figure 1b illustrates the initial metamaterial structure that consists solely out of gold disc resonators.The metamaterial design in Figure 1b can only enhance the absorption and does not contribute to improving the carrier extraction.In order to improve the carrier extraction electrical contacts were added to the metamaterial.Figure 1c shows the disc resonators combined with electricals contact lines in between the resonators.Next, we bend the contact lines around the disc resonators as illustrated in Figure 1d.The final design merges the contact lines and resonators, thereby consisting only out of contact lines that form a narrow-slot metamaterial themselves, Figure 1e.Based on these four designs, we show the systematic enhancement of the photodetector performance.

Device Simulation
To optimize the metamaterial photodetectors optical and electrical simulations were performed using CST Suite 2022.A schematic of a unit cell implemented in CST can be seen Figure 2a.The metamaterial was first simulated with respect to  its optical characteristics and designed in such a way that near unity total absorption could be achieved.More precisely, 40% of the absorption is lost in the metal discs, whereas 60% of the absorption occurs in the HgTe cQD layer at a wavelength of 2710 nm.Identical absorption values in the HgTe layer on the order of 60% could also be achieved for all designs further below.It was observed that the absorption in the cQD layer decreased for larger wavelengths although still reaching overall near unity absorption.This decrease of absorption in the active layer coincides with the decreasing absorptivity of the cQDs, see Figure S1c (Supporting Information) for the absorption spectra of the HgTe cQDs.The losses in the metal layer were mainly limited to the disc resonators to which the incoming light resonantly coupled, no grating effects or coupling of the incoming light to the contact lines was observed.Furthermore, all designs were simulated in such a way that the peak position can be tuned depending on the geometrical design parameters.The dependency plots of the absorption spectra on these parameters and the dependence of the absorption spectra on the polarization and illumination angle can be found in detail in the Section S2.2 (Supporting Information) for all designs.All designs were observed to demonstrate absorption that is independent of the polarization angle, coupled with excellent stability for varying illumination angles.
The simulated absorption spectra of the metamaterial for different disc diameters d can be seen in Figure 2b, additionally in Figure 2c the spatial optical absorption distribution for unpolarized light in the cQD layer is shown.
Electrical simulations were used to investigate the electric field (E DC -field) distribution resulting from an externally applied static DC bias.The normalized static E DC -field of the disc metamaterial without contact lines can be seen in Figure 2d.From the electrical stimulation we extract the distribution of the normalized transition time  tr of photogenerated carriers in the QD layer.This distribution can be seen in Figure 2e.Details on how the transition time distribution was calculated can be found in the Section S2.1 (Supporting Information).
Ideally the areas of greatest absorption, surrounding the disc resonator, and the areas with the shortest transition time overlap.This corresponds to the most efficient way to transport the photogenerated carriers to the contacts.Clearly, when analyzing Figure 2c-e, the absorption distribution (Figure 2c) does not show a strong overlap with the regions corresponding to the fastest carrier extraction pathways (Figure 2e).Further, in the case of simple disc resonator the carriers must travel across several unit cells until they reach the contacts.This long travel paths can have the consequence that a large number of photogenerated carriers recombine before they can be collected.
A simple method to increase the carrier extraction in a disc metamaterial is to place straight interdigitated contacts in between the individual disc resonators.A unit cell of such a metamaterial with interdigitated contacts can be seen in Figure 3a, with a signal contact on the left and a ground contact on the right.Optical simulations showed that the interdigitated contact lines hardly distort the metamaterial disc resonance, as shown in Figure 3b,c.Only when the contact lines are closer than 100 nm a broadening and shift of the resonance can be observed, see Section S2.2.2 (Supporting Information).In Figure 3c-e, the fields resulting from the optical and electrical simulations are shown.It can be seen that the normalized E DC -field, absorption and the transition time distribution within a cell do not significantly change compared to the disc metamaterial without contact lines presented in Figure 2. The benefit of this design is that due to the metal contact lines, the carriers get extracted within each cell.Thus, the extraction pathways are cut down that should improve the responsivity.
To further improve the carrier extraction, we propose a new design consisting of interdigitated contact lines wrapped around the disc resonators.This design and the resulting field distribution can be seen in Figure 4.It was discovered that the meandering contact lines hardly influence the disc resonator resonance, as shown in the spectral absorption plot Figure 4b,c.Similar to the metamaterials with straight contact lines, only when the contact lines are very close, i.e., below 80 nm, to the resonators a shift and broadening of the resonances can be observed, see Section S2.2.3 (Supporting Information).
The design with the bent disc lines exhibits an improved overlap of the area of absorption (Figure 4c) and short carrier extraction (Figure 4e), which is expected to increase the device performance.The largest overlap is achieved in the area where the contact lines are bent around the discs forming a narrow slot.However, this area only covers half of the resonator; therefore, the design can be further improved.
To achieve the highest overlap of short carrier transit times and absorption, we propose the final design presented in this work, depicted in Figure 5.This metamaterial only consists of deformed interdigitated electrodes forming a narrow-slot metamaterial.With this design near unity absorption with ≈ 60% absorption in the HgTe cQD layer could be achieved at a wavelength of 2710 nm-same as for the all disc metamaterials mentioned above.The absorption, static E DC -field and resulting transition time distribution of this metamaterial can be seen in Figure 5b-d.This metamaterial exhibits an ideal overlap of the static E DC -field resulting from the external applied bias and area of absorption.Furthermore, the slots are very narrow therefore the transition times are short.A slight asymmetry in the plotted field can be observed, this is the result of an asymmetry of the design for narrow slots.The metamaterial design was scaled by a factor of 1.1 in the x-direction for gaps smaller than 80 nm achieving a minor increase of absorption and Q-factor.

Passive Characterization
Following the metamaterial design and simulation process the envisioned metamaterials were fabricated, see scanning electron micrographs (SEM) in Figure 6.For detailed information on the fabrication process, we refer to the Experimental Section.The HgTe cQD metamaterial detectors were characterized with a Fourier-transform infrared (FTIR) spectrometer to gain the spectral absorption after fabrication.The FTIR measurements were done in a reflection setup configuration.A schematic of the setup can be found in the Section S3 (Supporting Information), Figure 6.
In Figure 6 the measured and simulated spectra of the different metamaterials are plotted.The radii of the disc metamaterials and the radius of the bend of the slot metamaterials were varied, whereas the gap between discs, discs and contact lines or the slot width were kept the same at 80 nm.It can be seen that the simulated and measured absorption spectra generally match well.A small shift of the peak resonance position for the measured against the simulated slot metamaterials as well as lower absorption maxima is observed.We attribute this to fabrication deviations such as angled metal sidewalls that result in broadening of the resonance as well as deviating materials parameters of the cQD.
Another difference between simulated and measured absorption spectra can be seen for all designs at smaller wavelengths.Here the measured absorption is larger than simulations predicted.This is attributed to material property data used for the simulations, which was for cQDs with a smaller diameter and therefore a different absorption onset.
In summary, all absorption spectra exhibit a narrow peak with a maximum overall absorption between 70 -95%.Further, from these measurements it is clear that the absorption peaks can be tuned over a wide range.

Metamaterial cQD Photodetector Performance Characterization
The optical characterizations revealed that for all four metamaterial variations a similar optical absorption enhancement can be achieved.To verify the expected beneficial contributions of the contact lines, we subsequently characterize the photodetectors in terms of their active opto-electric (O-E) performance.This was done with a similar setup configuration as used for the passive characterization where the sample was illuminated from the top, same as for the O-E characterization.A schematic of the setup configuration can be found in the Section S3 (Supporting Information), Figure 7.
We illuminate the different photodetectors under identical conditions with a laser source operating at a wavelength of 2710 nm.Electrically, the detectors were biased at their ideal operation voltage, i.e., in case of the disc metamaterials with contact lines and the slot metamaterial this meant that they were biased at a voltage where they exhibited the greatest photoresponse.For the disc metamaterials without contact lines a voltage was selected ensuring reliable operation, close but below the breakdown voltage.Further details on the voltage dependent photoresponse and current of different metamaterial photodetectors is shown in the Section S4 (Supporting Information).
We first compare the metamaterial cQD photodetector performance in terms of achievable responsivity.In Figure 7a the maximum responsivity of each of the metamaterials is plotted.The plot compares the best performing devices of each metamaterial design operating at its ideal conditions.In the plot it can be seen that there is a continuous increase of the responsivity for each of the improvement steps.The lowest responsivity is measured for the disc metamaterial with no contact lines with 1.25 A W −1 .By introducing the straight contact lines an increase of the responsivity by a factor of 2.4x could be achieved, resulting in a responsivity of 3 A W −1 .This increase is attributed to the increased carrier extraction, since the carriers are collected in between each disc.An additional increase of the photoresponse by a factor of 2x could be achieved by bending the contact lines around the disc resonators, highlighting the importance of the overlap of the external applied static E DC -field and the area of absorption.This design change step resulted in a responsivity of 6 A W −1 .The largest responsivity was recorded for the slot metamaterial, which was yet another factor 2.65x greater than for the disc metamaterial with bent meander shaped contact lines, reaching a responsiv-ity of 16 A W −1 .We therefore achieved an overall factor 13x improvement over the disc resonator metamaterial, even though the overall absorption of the narrow slot metamaterial only reached ≈76% at a wavelength of 2710 nm.These results are in line with our theoretical analysis on the extraction of the photo generated carriers.
To verify the here reported responsivity enhancement, we have additionally performed measurements on several devices to gain a statistical analysis.The resulting distributions and a discussion can be found in the Section S4.3 (Supporting Information).
The photodetectors were also characterized with respect to their frequency response as shown in Figure 7b.For these measurements different detectors were characterized compared to the plot in Figure 7a, but with the same design parameters.All photodetectors exhibited a constant photoresponse between a frequency range of 4-10 000 Hz.The measurements were limited to this frequency range due to the limitations of the mechanical chopper used to modulate the incoming light.Time dependent photoresponse measurements at the maximum frequency of the mechanical chopper were additionally performed, see Section S6 (Supporting Information).From these measurements we were able to extract a rise time of  rise = 14 μs and a fall time of  fall = 17 μs for all devices.No significant difference of the rise and fall time for the different photodetectors could be observed.
The engineering of the metamaterials was not only focused on increasing the responsivity but also on reducing the noise current spectral density by decreasing the active material volume.In the low frequency regime, where 1/f noise is dominant, it is possible to reduce the noise by reducing the volume of the cQD material.This is the case since 1/f noise can be seen as a superposition of Lorentzian shaped generation and recombination noise with different time constants and characteristics. [42]The generation and recombination noise is often dependent on trapping and detrapping processes that are related to defect sights in cQDs.By decreasing the volume of the cQD material the number of defect sights can be reduced resulting in an decrease of noise. [33,43,44]he noise reduction with the decrease of the active material volume between the contacts could also be observed throughout our measurements, as depicted in Figure 7c.It can be seen that the disc metamaterial without any contact lines exhibits the largest noise.The introduction of straight contact lines in the disc metamaterial results in an average current noise reduction by a factor of 46, while bent contact lines yield a reduction by a factor of 30.Similarly, slot metamaterials exhibit a noise reduction by a factor of 27 compared to disc metamaterials without contact lines.The difference of the noise current of the photodetectors with contact lines and the slot metamaterial is attributed to the lower resistance.The different current noise of the different photodetectors therefore stems from the correlation between noise current and dark current magnitude. [45]he reduced noise and increased responsivity result ultimately in an increase of the specific detectivity as shown in Figure 7d.The specific detectivity D* was calculated using the equation , with A being the active area and R the responsivity.The detectivity was increased by a factor of ≈110x, ≈145x, and ≈345x when comparing the metamaterial with straight contacts, bent contacts and the slot metamaterial with the disc metamaterial, respectively.
This clearly highlights the importance of charge extraction engineering in cQD photodetectors.Furthermore, these findings are not only limited to design of photodetectors operating in the MIR spectral range.A short discussion on the applicability of these findings can be found in the Section S7 (Supporting Information).

Conclusion
In this work we designed, simulated, fabricated, and characterized metamaterial enhanced HgTe cQD photodetectors operating in the MIR spectral range.The focus of this work was set on enhancing the photodetector performance by increasing the photogenerated collection efficiency and therefore the responsivity while simultaneously decreasing the noise and using costefficient cQDs.To this end a metallic disc metamaterial enhanced photodetector was stepwise improved using simulations and then verified experimentally.The carrier collection efficiency was improved by reducing the carrier extraction time.This was partially achieved by introducing contacts in between the disc resonators, thus drastically reducing the carrier extraction path lengths.To further improve the devices the contacts were deformed and wrapped around the resonator to gain a greater overlap between the externally applied E DC -field, responsible for carrier extraction, and the area of greatest absorption, concentrated around the disc resonators.Finally, a metamaterial was designed consisting only out of slots with a uniform absorption overlapping with the extraction E DC -field and short carrier extraction times.This systematic by simulations supported improvements resulted in a ≈ 13-fold increase of the measured responsivity comparing the disc metamaterial and the slot metamaterial.Further, the introduction of the contact lines leading to final metamaterial design resulted also to a decrease of the noise by an average of a factor of ≈ 50x, which is attributed to a reduced volume of the active material between the contacts.Overall, the design improvements lead to a 345-fold increase of the specific detectivity highlighting the importance of the contact engineering of metamaterial enhanced photodetectors.
Although a drastic improvement of the device performance could be shown by optimizing the metamaterials, there is still potential for further improvements especially concerning the absorber material.The observed responsivity increase is a result of the decreased carrier transition time due to shorter transition path lengths.An additional decrease of the transition time could be achieved by optimizing the cQD synthesis and ligand exchange, which could lead to increased mobilities. [46,30]Further, it has been observed that the dark current can be decreased with a narrow size distribution of the cQD, leading to a reduced noise spectral density. [47]We expect that these improvements, coupled with more efficient infilling of the metamaterial slots, have the potential to enhance the performance of the presented narrow slot metamaterial photodetector beyond current state-of-the-art detectors in terms of detectivity and cost-effectiveness.
The demonstrated design improvements highlighting the importance of the carrier extraction efficiencies and the reduction of material volume are not limited to photodetectors operating in the MIR spectral range but are of interest for photodetectors operating in various wavelength regimes.

Experimental Section
Simulations: The electrical and optical device simulations were carried out using CST Studio Suite 2022.The material data was either used from the CST library or in case of HgTe extracted from the work of Prachi Rastogi et al. [48] The distributions of the transition time were gained by post processing of the electrical simulation results and is discussed in more detail in the Section S2 (Supporting Information).
CQD Synthesis: A detailed description of the synthesis of HgTe CQDs used in this work, transmission electron microscopy images, XRD characterization and FTIR absorption spectra can be found in the Section S1 (Supporting Information).
Device Fabrication: For the metamaterial enhanced photodetector fabrication, a Si substrate with 200 nm thermally grown SiO 2 layer on top was used.A 100 nm thick Au backplane for the metamaterial was deposited on to the substrate followed by atomic layer deposition of a 20 nm alumina spacer layer.Subsequently, standard e-beam lithography and lift-off process was used to define the resonant metasurface.The 55 nm thick Au layer and a 5 nm alumina layer on top were deposited using e-beam evaporation.On to the metamaterial structures a ≈ 60 nm thick ligand exchanged HgTe cQDs layer was deposited using spin coating.The ligand exchange was done in optimized fashion to previous reports. [22]A discussion on the choice of the absorber layer thickness can be found in the Section S8 (Supporting Information).The fabricated devices were 15×15 and 30×30 μm 2 in size.For the passive characterization large metamaterial areas with 175×175 μm 2 were fabricated.
Metamaterial and HgTe cQDs Absorption Measurements: The ligand exchanged cQDs were purified and drop cast onto an Au covered Si substrate and the characterized using a custom-built setup (Section S3, Supporting Information) with an free-space coupled Arc Optics FTIR-Rocket spectrometer in a reflection measurement configuration.The same setup was used to perform the passive characterization of the different metamaterials covered with 60 nm cQDs and a 120 nm PMMA layer.
O-E Characterization: A schematic of the used setup for the optical excitation of the photodetectors can be found in the Section S3 (Supporting Information), Figure 7.A Keysight B2902A SMU was used for biasing the photodetector, which was then connected in series to the transimpedance amplifier current input of a MFLI lock-in amplifier from Zurich Instruments.With the help of the lock-in amplifier it was possible to acquire the noise current spectral density and the photoresponse of the device under test.
A reference photodetector with an area of 15×15 μm 2 was used to characterize the power distribution of the incoming light on the sample stage.The distribution of the power can be found in the Section S6 (Supporting Information).It was found that the spot size was larger than 15×15 μm 2 .Further, a calibrated powermeter EO TH5B-BL-DZ-D0 from Gentec was used to measure the overall incoming power.These two measurements were used to calculate the power illuminating the DUT.
XRD Characterization: The ligand exchanged HgTe CQDs were deposited on to a Si substrate using drop casting and then characterized using a Rigaku SmartLab 9KW XRD-diffractometer.

Figure 1 .
Figure 1.a) Illustration of a metamaterial enhanced HgTe photodetector with source and drain contacts.The detectors consist of a 100 nm Au back reflector (Mirror), a 17.5 nm alumina spacer layer (Spacer), followed by 60 nm thick Au resonator with 5 nm alumina on top.The resonator layer is imbedded in a ≈ 60 nm HgTe layer that is covered by 120 nm PMMA nm layer for passivation.For comprehensibility the cQD layer was only infilled between the resonators and the PMMA was omitted in the main image.The inset shows the layer stack in more detail.Illustration of b) disc resonator top layer, c) disc resonator top layer with straight interdigitated finger contacts, d) indicates the area disc resonator top layer with bent interdigitated contacts and e) of narrow slot metamaterial.

Figure 2 .
Figure 2. a) Illustration of a disc metamaterial unit cell.b) Simulated absorption spectra with varying diameter d ranging from 380 to 590 nm.The disc diameter d is increased in increments of 30 nm.The distance g between the disc resonators was kept constant at 80 nm.The spectra shift to larger wavelengths with increasing disc diameter c) Normalized optical losses (absorption) of a disc resonator surrounded by HgTe cQDs.d) Normalized static electric field resulting from external DC biasing including streamlines.e) Normalized transition time for single type of charge carrier.

Figure 3 .
Figure 3. a) Illustration of disc metamaterial unit cell with straight contact lines.b) Simulated absorption spectra with varying diameter ranging from 380 to 590 nm.The disc diameter d s is increased in increments of 30 nm.The distance g s between the disc resonators is kept constant at 80 nm.The spectra shift to larger wavelengths with increasing disc diameter.c) Normalized optical losses (absorption) of a disc resonator with straight contact lines surrounded by HgTe cQDs.d) Normalized static electric field resulting from external DC biasing including streamlines.e) Normalized transition time for single type of charge carrier.

Figure 4 .
Figure 4. a) Illustration of a disc metamaterial unit cell with bent contact lines.b) Simulated absorption spectra with varying diameter ranging from 380 to 590 nm.The disc diameter d b is increased in increments of 30 nm.The distance g b between the disc resonators was kept constant at 80 nm.The spectra shift to larger wavelengths with increasing disc diameter.c) Normalized optical losses (absorption) of a disc resonator with bent contact lines surrounded by HgTe cQDs.d) Normalized static electric field resulting from external DC biasing including streamlines.e) Normalized transition time for single type of charge carrier.

Figure 5 .
Figure 5. a) Illustration of a narrow slot metamaterial unit cell.b) Simulated absorption spectra with varying bend radius r s ranging from 240 to 330 nm.The radius r b is increased in increments of 15 nm.The slot width s w of the metamaterial is kept constant at 80 nm.The spectra shift to larger wavelengths with increased radii.c) Normalized optical losses (absorption) of a narrow slot metamaterial filled with HgTe cQDs.d) Normalized static electric field resulting from external DC biasing including streamlines.e) Normalized transition time for single type of charge carrier.

Figure 6 .
Figure 6.SEM image of the metamaterials before cQD deposition, measured (top) and simulated absorption spectra (bottom) for different radii of a,b) disc metamaterial, c,d) disc metamaterial with straight contact lines, e,f) disc metamaterial with bend contact lines and g,h) narrow slot metamaterial.All spectral measurements were done with metamaterials covered with HgTe cQDs and PMMA.The scalebar in the SEM images indicates a length of 500 nm.

Figure 7 .
Figure 7. O-E characterization of the four different metamaterials."Disc" refers to the disc metamaterial without contact lines, "Disc lines" refers to disc metamaterials with straight contact lines, "Disc bent" refers to disc metamaterials with bent meander shaped contact lines and "Slots" refers to narrow slot metamaterials.a) Maximum responsivity of the four different metamaterials.b) Frequency dependent responsivity of typical metamaterial devices.c) Current noise spectral density i n versus frequency.d) Detectivity versus frequency.