Opto-electrical decoupled phototransistor for starlight detection

doi:10.21203/rs.3.rs-4705743/v1

Download PDF

Article

Opto-electrical decoupled phototransistor for starlight detection

https://doi.org/10.21203/rs.3.rs-4705743/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Highly sensitive shortwave infrared (SWIR) detectors are essential for detecting weak radiation (typically below 10^− 8 W·Sr^− 1·cm^− 2·µm^− 1) with high-end passive image sensors. However, mainstream SWIR detection technology is based on epitaxial photodiodes, which cannot effectively detect ultraweak infrared radiation due to the lack of inherent gain. Here, we developed a heterojunction-gated field-effect transistor (HGFET) consisting of a colloidal quantum dot (CQD)-based p-i-n heterojunction and a carbon nanotube (CNT) field-effect transistor, which achieves a high inherent gain based on an opto-electric decoupling mechanism for suppressing noise. The stacked heterojunction absorbs infrared radiation and separates electron-hole pairs. Then, the generated photovoltage tunes the drain current of the CNT FET through an Y₂O₃ gate insulator. As a result, the HGFET significantly detects and amplifies SWIR signals with a high inherent gain while minimally amplifying noise, leading to a recorded specific detectivity above 10¹⁴ Jones at 1300 nm and a recorded maximum gain-bandwidth product of 69.2 THz. Direct comparative testing indicated that the HGFET can detect weak infrared radiation at 0.46 nW/cm² levels; thus, compared to commercial and reported SWIR detectors, this detector is much more sensitive and enables starlight detection or vision. As the fabrication process is very compatible with CMOS readout integrated circuits, the HGFET is a promising SWIR detector for realizing passive night vision imaging sensors with high resolutions that are high-end, highly sensitive, and inexpensive.

Physical sciences/Engineering/Electrical and electronic engineering

Physical sciences/Nanoscience and technology/Nanoscale devices/Electronic devices

infrared photodetector

carbon nanotubes

weak infrared radiation detection

opto-electric decoupling

sensitivity

Extensive efforts have been dedicated to improving the detectivity and resolution of shortwave infrared (SWIR, within the band of 0.9–1.7 µm) photodetectors, which enable remote imaging, night vision, spectroscopy and object tracing^1–4. Generally, the detection of weak infrared radiation (typically below 10^− 8 W·Sr^− 1·cm^− 2·µm^− 1) places extremely high demands on the signal-to-noise ratio (SNR), also known as the detectivity of sensors; due to these demands, it is necessary to achieve high optical responses and low electrical noise⁵. Mainstream SWIR detectors rely on the diode configuration and are currently dominated by epitaxial semiconductors, such as indium gallium arsenide (InGaAs); through continuously improving material and interface quality, researchers have developed detectors with high specific detectivity (~ 10¹³ Jones) based on an ultralow dark current below 1 nA/cm² and electrical noise approaching the technological limit^5–8. The optical response of a diode detector is determined by the external quantum efficiency (EQE), which can reach as high as ~ 90%⁵; thus, there is little room for improvement. Further improving the detectivity beyond 10¹³ Jones, which enables the detection of weak infrared radiation at sub-1 nW/cm² levels (starlight)⁹, is challenging in a photodiode structure that lacks inherent gain and provides only one pair of carriers per single incident photon^{5, 10}.

Implementing an inherent gain mechanism is essential for enhancing the optical response, as exemplified in device architectures, such as avalanche photodiodes (APDs) and nanostructured field-effect transistors (FETs)^11–13. Under a high electric field, photocarriers in APDs are accelerated and multiplied through impact ionization, achieving a high gain (M) that exceeds 10⁵ when operating in Geiger mode for single-photon detection^{14, 15}. However, due to the random ionization of charge carriers, APDs tend to exhibit significantly amplified excess noise ^{11, 16}. Emerging nanomaterials with facile solution processability and integration capability offer greater flexibility in infrared detector design, holding significant potential for constructing novel detectors with high gain and low costs^{1, 2}. Recently, extensive infrared radiation detectors with nanostructured FET architectures have been reported to have inherent gains^17–21. Quantum dot-sensitized graphene FET, a typical type of nanostructured phototransistor, achieved high photogain (up to 10⁸) based on a trap-dominated photogating effect, but usually suffers from high dark current, noise levels, and limited response speed^{9, 12}. More recently, researchers have also reported sensitized FETs based on large bandgap semiconductor channel materials with reduced dark current (0.1–100 nA) and high photogain^{18, 22–24}. Almost all the reported FET detectors with high inherent gain have also been no substantial breakthrough in detectivity, since the noise and signal are simultaneously amplified in the same channel in charge of optical sensing and electrical amplification. Therefore, a SWIR detector with a well-designed device structure and operating mechanism is highly desired to achieve opto-electric decoupling, facilitating ultrahigh detectivity beyond 10¹³ Jones and enabling passive starlight detection or vision.

In this work, we present a heterojunction-gated field-effect transistor (HGFET) that achieves ultrahigh photogain and exceptionally low noise in the SWIR region, benefiting from a design that incorporates a comprehensive opto-electric decoupling mechanism. The stacked heterojunction absorbs infrared radiation and separates electron-hole pairs. Then, the generated photovoltage tunes the drain current of the CNT FET through an electrostatic gate coupling effect. As a result, optical sensing process and electrical amplification process are completely isolated in space. The opto-electric decoupling HGFET significantly detected and amplified SWIR signals with a high inherent gain of 1.5×10⁶ while barely amplifying noise; in addition, the HGFET exhibited a recorded specific detectivity above 10¹⁴ Jones at 1300 nm and a maximum gain-bandwidth product (GBW) of 15.5 GHz, enabling the detection of weak infrared radiation at 0.46 nW/cm². The use of aligned CNTs in HGFETs further promoted the GBW to 69.2 THz, surpassing the theoretical limit of sensitized Si FETs and state-of-the-art InGaAs APDs^{18, 25–27}. The results demonstrated that the HGFET is a promising SWIR detector for realizing high-end passive night vision image sensors with high resolution, high sensitivity, and low cost.

Figure 1a shows a schematic of the HGFET infrared detector, which consists of a CNT thin-film transistor gated by a PbS CQD/ZnO heterojunction (the detailed fabrication processes are provided in Methods and Supplementary Section 1). Local bottom-gated metal-oxide- semiconductor (MOS) FETs are fabricated on a randomly oriented semiconducting CNT film covered by an yttrium oxide (Y₂O₃) film which acts as gate insulator to separate the photoelectric conversion component above and the electronic component below. A sputtered 40-nm layer of n-doped ZnO (with a dopant concentration of ~ 10¹⁷ cm^− 3)^{28, 29} is deposited and patterned on a Y₂O₃/CNT film channel through photolithography and a lift-off process. A 150 nm-thick intrinsic PbS-halide CQD layer and a 30 nm-thick ethanedithiol (EDT)-treated p-doped PbS CQD layer (with a dopant concentration of 10¹⁷ cm^− 3) are spin-cast to form a p-i-n heterojunction, which generates a negative photovoltage through the absorption of infrared photons, the separation of photogenerated electron-hole pairs through the built-in field, and the accumulation of electrons at the ZnO/gate dielectric interface. The generated photovoltage tunes the Fermi level of the semiconducting CNT film in the FET through the electrostatic gate-coupling effect. Then, the optical response in the heterojunction can be efficiently coupled to the CNT channel and further amplified by the drain current.

Spatially resolved photocurrent response measurements (Fig. 1b) were obtained on an as-fabricated device to explore the working mechanism of the HGFETs. Obvious photocurrent generation only occurs when the laser beam irradiates the region covered by the CQD/ZnO heterojunction gate (HG); this result indicates that the HG plays a crucial role in establishing the built-in potential necessary to separate the photocarriers and generate a photovoltage for tuning the covered semiconducting CNT channel through the intermediate Y₂O₃ gate insulator. As a highly selective optical response was observed only in the predefined photosensitive zone, the HGFET configuration provides significant advantages for minimizing electrical crosstalk between pixels in high-end applications, such as image sensors. Spectral responsivity measurements (Fig. 1c) illustrate that the HGFET responds to light with wavelengths ranging from 600 nm to 1440 nm with a prominent peak at approximately 1300 nm, which is consistent with the absorption peak at 1286 nm in solution-processed PbS-halide CQDs (inset of Fig. 1c).

Opto-electric decoupling, i.e., realizing optical sensing and electrical amplification at different regions, is the core goal of designing HGFETs with ultrahigh detectivity. To further elucidate the fundamental mechanism, we have presented a comparative analysis of the photoelectric conversion and primary noise components between the APD and HGFET structures (Fig. 1d). In a typical APD with a p-doped absorption region and a lightly doped n-type multiplication region, the depletion region is established within the n-region and is intensified through reverse bias (V_bias) to separate photoelectrons and induce avalanche multiplication for carrier multiplication¹⁵. The light absorption, charge separation, multiplication, and collection processes occur within the same semiconducting channel, which was named the opto-electric mixing structure; in this structure, noticeable excess noise, or amplified shot noise, arises from random ionization of carriers. The shot noise is generally expressed as I_s=[2qIM²F(M)]^1/2, where M is the gain, I is the current for M = 1, q is the elementary charge, and F(M) is the excess noise factor typically exceeding unity¹⁶. Compared to a typical p-n junction photodiode, the shot noise in an APD is amplified by \(\:M\sqrt{F\left(M\right)}\), which results in a proportionally lower signal-to-noise ratio. Additionally, fluctuations in thermally generated carriers (corresponding to thermal noise, I_nt) and fluctuations in carrier concentration related to traps or interface states (flicker noise, I_nf) can be amplified, although they may be negligible compared to the amplified shot noise under high electric field conditions. Therefore, compared to normal photodiodes, APDs exhibit no obvious advantage in terms of the signal-to-noise ratio for low-light-level imaging.

To improve the signal-to-noise ratio of the detector, the HGFET was specially designed to operate with an opto-electric decoupling mechanism, i.e., light absorption and photovoltage generation within the HG-based optical module, and photovoltage amplification and charge transport occur in the CNTs FET; the two components are separated by the Y₂O₃ gate insulator. The optical response in the heterojunction can be efficiently coupled to the CNT channel and further amplified in the drain current through the electrostatic gate-coupling effect, and the photogain is primarily determined by the transconductance (g_m) of the photovoltage-gated CNT FETs (as discussed in Supplementary Section 2). Additionally, since the p-i-n heterojunction operates in an open-circuit configuration, current-related noise, including shot noise or flicker noise, cannot be generated. As a result, only thermal noise is coupled into the device and amplified by the CNT FET. Due to the opto-electric decoupling design, the HGFET can amplify the photovoltage while barely amplifying the electrical noise, which significantly improved the signal-to-noise ratio and detectivity. Recently, various phototransistor-based infrared detectors have been reported based on trap-dominated photoconductive gain mechanisms (as depicted in Fig. S3) since photosensitive nanomaterials are directly grown on the semiconducting channel of FETs without an insulator layer^{12, 13, 17, 22, 30}. Current-related noise, such as shot noise and flicker noise in these phototransistors, must be amplified, and carrier traps can be introduced to increase the flicker noise of the semiconducting channel. Thus, compared to InGaAs photodiodes, the detectivity of these phototransistors with the trap-dominated gain mechanisms was not greatly improved.

A typical fabricated HGFET infrared detector (Fig. 2a) was characterized under preoptimized bias conditions (V_bg= -0.6 V, V_ds= -0.1 V) to achieve high gains and low dark currents (Fig. S4). As shown in Fig. 2b, the dynamic photocurrent responses were determined by increasing the irradiation power density (P_light) from 0 to 15.4 W/cm² at 1300 nm. An obvious increase of 75.9 pA at background current (2.8 nA) indicates that the HGFET detector can respond to weak infrared radiation with a P_light as low as 0.46 nW/cm², corresponding to an irradiance of approximately 3007 photons/second. With increasing irradiation power density P_light, the photocurrent I_ph monotonically increases but follows a positive power function in the low P_light region and a logarithmic function in the high P_light region (Fig. 2c and Supplementary Section 4). Notably, I_ph does not show obvious saturation even at P_light values as high as 15 W/cm², which leads to a wide dynamic range (210 dB) in our HGFETs.

For a photovoltage-gated HGFET (Fig. S5), the relationship between the open-circuit voltage (\(\:{V}_{oc}\)) of the p-i-n heterojunction and P_light is expressed as follows:

\(\:{V}_{oc}=\frac{{k}_{B}T}{q}ln\left(\frac{\eta\:qA{P}_{light}}{hv{I}_{0}}+1\right)\) (1),

where \(\:{k}_{B}\) is the Boltzmann constant, \(\:T\) is the temperature, \(\:A\) is the optical area, \(\:{I}_{0}\) is the reverse saturation current, \(\:\eta\:\:\)is the external quantum efficiency, \(\:q\) is the elementary charge, and hν is the energy of one photon. Then, the photocurrent \(\:{I}_{ph}\) of the HGFET can be determined as follows:

\(\:\:{I}_{ph}=\frac{{g}_{m}\alpha\:{k}_{B}T}{q}ln\left(\frac{\eta\:qA{P}_{light}}{hv{I}_{0}}+1\right)\) (2),

where α is the gate coupling coefficient between the HG and FET, and g_m is the transconductance of the FET. Based on equations (1) and (2), a photoelectric response model of the HGFET was built and verified (see details in Section 4 of the supplementary information), which was also used to fit the measured photoresponse data for retrieving key parameters. Typically, we extracted HG parameters, such as η of 10%, from an ohmic contacted ZnO/CQD p-i-n photodiode with identical geometrical dimensions and structures (Fig. S7). The inherent gain (\(\:G\)) of the HGFET can be expressed as follows:

\(\:G=Rhv/q\eta\:\) (3),

where \(\:R=\frac{{I}_{ph}}{A{P}_{light}}\) is the photoresponsivity. Considering an R of up to 1.6×10⁵ A/W with a V_ds bias of -0.1 V under 1300 nm irradiation at P_light of 0.46 nW/cm² (Fig. 2c), the corresponding \(\:G\) is calculated to be ~ 1.5×10⁶.

The response time of the HGFET detector is strongly dependent on the V_ds applied to the CNT FET, as shown in Fig. 2d, which demonstrates that increasing V_ds from − 0.1 to -5 V results in a continuous decrease in the rise and decay times. The lowest rise time t_Rise (95 µs) and fall time t_Fall (415 µs) are achieved with a V_ds bias of -5 V. According to the full photoelectric process of the HGFET shown in Fig. 1d, the response time is dominated by two main processes, namely, the drift of photocarriers in the p-i-n heterojunction and the propagation delay in the CNT channel. The drift of photocarriers in the p-i-n heterojunction will occur at sub-microseconds, comparable to that of reported photodiodes with similar dimensions²⁸, and CNT network FETs with a channel length of 10 µm typically exhibit a propagation delay of ~ 100 µs³¹. Therefore, the speed of the detector is mainly constrained by carrier transport in long-channel CNT FETs. The photocurrent decay is much slower than the photocurrent rise because the detrapping process of charged trap states is slower than the trapping process^{32, 33}. To enable high-speed detection, the photocurrent rise and fall time should be further reduced through other strategies, such as adopting aligned CNTs, reducing the channel length, and applying an electric pulse at the gate to facilitate the escape of trapped carriers¹².

The current noise spectrum of the HGFET detector (see detailed measurements in the Supplementary Information and Fig. S8a) shows a typical 1/f frequency-dependent character, indicating that the total current noise is mainly contributed by the flicker noise. As evidence, the calculated lines of shot noise and thermal noise (Supplementary Sections 5 and 8) are far below the measured curve of the noise spectrum, while the simulated flicker noise is almost consistent with the measured noise curve. Considering the opto-electronic decoupling mechanism of HGFETs, electric noise can be categorized into two sources: noise originating from the heterojunction and noise stemming from the underlying CNT FET (the developed noise model is detailed in Supplementary Section 8). In the open-circuit state, only the preexisting thermal noise (V_nT−HG) of the heterojunction can be amplified by g_m and then coupled to the detector. Compared to the flicker noise of the CNT FET, this noise is negligible (Fig. 2e), which was determined by comparing the current noise of the device before and after depositing the p-i-n heterojunction on the high-κ layer of the CNT channel (Fig. S11). Based on the measured photoresponsivity R and electrical noise spectrum, the room-temperature specific detectivity (D*) of the HGFET detector is plotted as a function of P_light at 1300 nm (Fig. 2e). D* increases with decreasing P_light, and the peak value exceeds 10¹⁴ Jones under weak light, which is recorded in all reported SWIR detections with measured noise data^{3, 12, 13, 28}.

Gate-tuneable operating mode of the HGFET

The gain of the FET relies on the gate voltage, which will provide a factor to tune the detectivity of the HGFET detector. Figure 3a displays the transfer curves of an HGFET under an infrared radiation power density ranging from 0 to 1.08 µW/cm² at 1300 nm. The threshold voltage (V_th) shifts by approximately 0.16 V due to the photovoltage-gated response. The photocurrent for the HGFETs has a positive power function dependent on the light intensity (Fig. 3b) when gated in the subthreshold region (V_bg =-0.05 to -0.15 V) but exhibits a logarithmic function (Fig. 3c) when gated in the linear region (V_bg =-1.2 to -1.3 V). Further analysis (Figs. S9 and S10) was conducted to determine the V_bg-dependent responsivity and gain (Fig. 3d). Both the responsivity and gain increase with a power-law dependence on V_bg in the subthreshold region and reach saturation at a V_bg of approximately 0.5 V (see the fitted gain detailed in Supplementary Section 7); this is because the values are directly proportional to transconductance g_m, which increases with V_bg until the device enters the linear region. When operating in the linear region (V_bg ranging from − 1 to -2 V), a high inherent gain (~ 10⁷) is achieved to amplify the photovoltage at the heterojunction of the HGFET in situ and provide a higher output current.

The measured current noise of the HGFET is also dependent on V_bg, which exhibits characteristics similar like the transfer characteristics of a CNT FET (Fig. 3e). The relation between D* and V_bg was extracted by combining the R-V_bg curve in Fig. 3d and the noise-V_bg curve in Fig. 3e, as shown in Fig. 3f; these results demonstrate that the D* of the HGFET strongly depends on the gate voltage with a similar behaviour to that of the simulation results (see details in Supplementary Section 9 and Fig. S12). Corresponding to the operating region of the FET tuned by V_bg, the HGFET can operate in two typical modes, i.e., high-gain mode (in the linear region) and high-detectivity mode (in the subthreshold region). In high-gain mode, the HGFET detector can provide a high output current, which is beneficial for simplifying the readout circuit design. In high detectivity mode, the HGFET detector can provide high sensitivity for weak light detection with low power dissipation.

Performance comparison and benchmarking

We compared the performance of the HGFET detector and a commercial InGaAs photodiode (FGA015)³⁴ by directly testing the SWIR response characteristics under the same testing conditions (see details in Supplementary Section 10). Figure 4a illustrates the detailed measurement system, enabling precise photoelectric testing and laser beam analysis. The output beam spot can be adjusted using two different objective lenses with amplification factors of 10× and 50× to match the optical areas of the two devices. The output power was modulated using adjustable slots and neutral density filters, and it was meticulously calibrated using a germanium power metre and a standard InGaAs photodiode (Table S1). The time-resolved photoresponses under weak infrared irradiation at 1300 nm (Fig. 4b) show that the HGFET demonstrates a high SNR (I_ph/I_n) (approximately 70) at a light intensity as low as 0.46 nW/cm², at which the InGaAs photodiode exhibits no response. As the InGaAs photodiode begins to show a weak current response (with an SNR of approximately 7), the light intensity increases to 3.78 nW/cm². The photocurrent in the HGFET device (in the subthreshold region) exhibited a clear power-law relationship with the incident light intensity (Fig. 4c), even when the power intensity was below 1 nW/cm². However, the photocurrent in the InGaAs photodiode exhibits a linear relation with the light intensity and begins to deviate from the linear relation as the light intensity decreases below 100 nW/cm²; these results indicate that the responsivity is uncertain under weak light, mainly owing to the absence of inherent gain-induced low R (Fig. 4c and 4d).

To preliminarily demonstrate the array ability of the HGFET detectors for passive night vision imaging, we fabricated an image sensor featuring 64×64 pixels of HGFETs, and the pixel array was connected to the readout circuits on the printed circuit board (PCB, see details in Supplementary Section 11). Without an image lens, we displayed the “PKU” images acquired at different irradiation power densities using an assembled test system (Fig. 4e). Despite far from optimizing device structure (lack of a gate switch transistor for each pixel and non-ideal electrical cross-talk within each row of the 64 pixels), the HGFET imager successfully captured the “PKU” pattern (Fig. 4f) at a low power density (100 nW/cm²), demonstrating the potential of HGFET for high sensitivity.

We performed benchmarking tests for our HGFET detector using commercial photodiodes and other thin-film-based infrared photodetectors, focusing on important metrics such as specific detectivity and speed. In Fig. 5a, the wavelength-dependent D* characteristics of our HGFET device are compared with those of InGaAs photodiodes (from Thorlabs and Hamamatsu) and Ge photodiodes (from Thorlabs) measured under identical setup and conditions. The HGFET achieved a peak D* that exceeded 10¹⁴ Jones, which is nearly two orders of magnitude greater than that of InGaAs photodiodes. This is the highest reported peak D* for thin-film-based photodetectors in the SWIR range, facilitating detection under moonlight and even starlight conditions, as depicted in Fig. 5b and detailed in Table S2^{34, 35}. Moreover, the sensitivity of HGFETs can be substantially improved by optimizing the EQE of the CQD heterojunction and enhancing the gate efficiency of the FET; thus, it is possible to approach the fundamental limits defined by the signal fluctuation limit (SFL) and background limited infrared photodetection (BLIP) of approximately 10¹⁸ Jones at 1300 nm⁵.

Similar to all reported electrical devices with internal gain¹⁷, a trade-off occurs between the gain (G) and bandwidth in photodetectors, including the HGFET, and the gain-bandwidth (GBW) product is the key metric for characterizing the comprehensive performance in terms of the sensitivity and speed of a photodetector with gain. Figure 5c presents the GBWs of five typical detectors with different architectures, including photoconductors³⁶, photodiodes^{18, 34}, APDs^{26, 27}, photogating devices¹², and HGFETs. A maximum GBW of 15.5 GHz was achieved in the HGFET with randomly oriented CNT films, which outperforms all reported thin-film-based infrared detectors and InGaAs PIN diodes^{13, 18, 25, 28} (Fig. S20). With the much higher carrier mobility than randomly oriented CNT film, aligned semiconducting CNT films have been considered as a superior semiconducting channel in constructing FETs with high g_m and speed³⁸. Thus, the HGFET based on aligned CNTs exhibits a GBW of up to 69.2 THz alongside a D* of 6.7×10¹³ Jones (Figs. S20 and S21) and far exceeds that of all existing infrared detectors, including the latest advanced APD devices^{13, 26, 27}. This outstanding GBW performance of the HGFET primarily results from the physical mechanism of opto-electric decoupling, specifically rapid photocarrier separation facilitated by the built-in potential, ultrafast electrostatic coupling, swift charge transport within the CNT channel, and excellent signal amplification capability of the CNT FET.

Opto-electrical decoupling enables HGFETs provide greater controllability and can help alleviate trade-offs between photon absorption and electrical noise, as well as between gain and response speed. Excellent optical and electrical properties can be achieved simultaneously in one device by employing two solution-processed semiconductors rather than relying on single-crystalline materials that necessitate precise lattice matching; this process may promote the monolithic integration of HGFET detector arrays with silicon ROICs through a back end of line (BEOL)-compatible process. Furthermore, HGFETs can replace the light absorber and serve as a versatile platform for photodetection across various wavelength bands; thus, the detector can be adapted to different optical applications.

We have reported an opto-electric decoupling HGFET SWIR detector that presents a high inherent gain for amplifying signals while barely amplifying noise for detecting weak infrared radiation. The HGFET devices exhibit a recorded detectivity of 1.27×10¹⁴ Jones and a maximum gain-bandwidth product of 69.2 THz to 1300 nm irradiation at room temperature. The opto-electric decoupling device can detect weak infrared radiation of 0.46 nW/cm², which is much more sensitive than the commercial and all reported SWIR detectors and especially enables a starlight detection or vision. As the fabrication process of HGFET is highly compatible with CMOS readout integrated circuits, the HGFET offers a universal platform for achieving high-end passive night vision image sensors in thin-film semiconductors; this technology paves the way for innovative optoelectronic circuits and future monolithically integrated systems with high resolution, high sensitivity and low cost.

Materials preparation

Polymer-sorted, solution-derived CNTs with a semiconductor purity exceeding 99.999% were prepared using dispersion and centrifugation, as detailed previously in the literature³⁷. The polymer-wrapped CNT solution was then diluted with toluene. Randomly oriented CNT films with high uniformity were formed by immersing the cleaned substrate in a diluted CNT solution for 48 hours. A dimension-limited self-alignment (DLSA) procedure was employed to deposit aligned CNT films, as previously described in the literature³⁸. Next, a cleaning process involving coating with yttrium oxide (Y₂O₃) and then decoating to remove the polymer was carried out. Finally, thermal annealing was conducted at 600°C in an Ar/H₂ atmosphere.

The synthesis of PbS CQDs with oleylamine ligands was executed following a solution-processed procedure detailed in prior studies^{28, 29}. In the ambient atmosphere, a solution-phase ligand exchange process was conducted to synthesize PbS-halide CQDs. These CQDs were then dispersed in a mixture of amylamine, hexylamine, and butylamine (in a volume ratio of 3:2:10) at a concentration of 150 mg/mL to prepare the PbS-halide CQD ink for film fabrication. The construction of the ZnO/PbS CQD heterojunction began with the sputtering of a 40 nm-thick ZnO thin film utilizing a previously described method³⁹. Next, a 150 nm thick absorbing layer was deposited onto the ZnO thin film via spin coating the PbS-halide CQD ink at 2500 rpm for 20 seconds. The final step involved depositing two layers of PbS CQDs treated with 1,2-ethanedithiol (EDT, 0.02 vol% in an acetonitrile solution) to serve as the p-doped layer. The doping concentrations of the prepared ZnO film and PbS-EDT CQDs were previously measured to be ~ 10¹⁷ cm^− 3.

HGFET fabrication

The HGFET detectors were constructed on Si/SiO₂ substrates. As illustrated in Fig. S2, the local bottom-gate metal was patterned using photolithography and electron beam evaporation (EBE) of a Ti/Pd film (1/20 nm) via a standard lift-off process. A 10 nm thick (100 cycles) layer of hafnium oxide (HfO₂) dielectric was deposited onto the gate metal using atomic layer deposition. Subsequently, a CNT film was deposited onto the HfO₂ dielectric and patterned through photolithography and reactive ion etching (RIE) to create the channel region. Source (S) and drain (D) pads were then created through photolithography, EBE of a Ti/Pd/Au film (0.3/40/30 nm), followed by a standard lift-off process. An ultrathin Y₂O₃ layer (~ 6 nm) was deposited on the CNT channel through thermal oxidation at 270 ℃ in air, serving as the dielectric for the floating gate. Finally, the p-i-n heterojunction was constructed by sequentially depositing the ZnO film, PbS-halide CQD film, and PbS-EDT CQD film.

HGFET imager fabrication

The fabrication process, as illustrated in Fig. S15, began with the deposition of a randomly oriented CNT film on a 4-inch Si/SiO₂ substrate, followed by patterning using photolithography and RIE. Next, the source/drain electrodes and the lower interconnect line were patterned by photolithography, followed by EBE of a 30 nm Pd film and subsequent lift-off processing. An isolation layer of 60 nm SiO₂ was deposited using photolithography, EBE, and a subsequent lift-off process. The upper interconnect line was then formed by photolithography and deposition of a 20/600 nm Ti/Au film via EBE. Following this procedure, similar steps were performed to deposit the ultrathin Y₂O₃ layer and p-i-n heterojunction. Finally, an HGFET array with 64×64 pixels was constructed and wire bonded to the readout circuit board.

Characterization of the materials and device

SEM images were obtained using a FEI ZEISS Sigma 300 scanning electron microscope. The optical absorption spectra of the PbS CQDs were measured with a UV-3600 plus spectrophotometer. The Raman spectrum of the CNT film was collected by a LabRAM HR800 instrument (Jobin-Yvonr). Optoelectronic measurements of the HGFET devices and the InGaAs diode were conducted using the measurement system depicted in Fig. 4a. A supercontinuous spectrum laser from the NKT Company was utilized to characterize the device responses, with the laser wavelength adjustable from 640 to 2100 nm. The spectral responsivity of the devices was assessed using a monochromator and calibrated against a standard InGaAs diode (FGA21, Thorlabs). Photocurrent spectroscopy was performed using an SR830 lock-in amplifier in conjunction with an SR540 chopper and a tungsten lamp equipped with monochromator, all of which were controlled by a LabVIEW program. The transient response and − 3 dB bandwidth were measured using a high-speed oscilloscope (Agilent DSO7054A). Current noise spectra were obtained using the PDA FS380 (Bodawei Electronic Technology Co., Ltd.) and E4727B (Keysight Technologies Co., Ltd.) systems. All electronic transport measurements were carried out using a Keithley 4200 semiconductor analyser at room temperature under ambient conditions.

Image capture

A schematic diagram of the assembled test system is shown in Fig. 4e. The setup involves collimated infrared light emitted from a 940 nm LED source (M940L3-C4 from Thorlabs) with a beam diameter of 44 mm. The power density of the light was modulated using neutral density filters. A photomask containing the designed patterns was used to provide a spatial light distribution, which was subsequently captured by the 64×64 pixel HGFET imager.

Detector characteristics

The calculated gain of the HGFET, as outlined in Section 2 (Supporting Information), is directly linked to the transconductance (g_m). The calculated photocurrent, which was described in Section 4, exhibits a positive power function in the subthreshold region of the CNT FET and a logarithmic function in the linear region. These distinctive response characteristics were further validated by comparing the extracted g_m and subthreshold swing (SS) of HGFETs with those of top metal-gated CNT FETs. TCAD simulations were constructed and illustrated a continuously fitted photocurrent under various irradiation conditions (refer to Fig. S7). The specific parameters used for the TCAD simulation are detailed in Section 4. The noise characteristics are described in Section 8, and the calculated expressions for the HGFET noise are provided. The detectivity characteristics of HGFETs are presented in Section 9, illustrating how the signal-to-noise ratio (SNR) is proportional to g_m/I_ds in the subthreshold region and g_m²/I_ds in the linear region.

Data availability. The data that support the plots within this paper and other findings of this study are available from the corresponding authors upon reasonable request.

Acknowledgements

This work is supported by the Natural Science Foundation of China (62225101 and U21A6004) and the Peking Nanofab Laboratory.

Author contributions

Z.Z. and Y.W. advised on and led this project. S. Z. carried out the device fabrication and experimental measurements. X. Z. carried out the aligned CNT-based HGFET fabrication and measurements. S.Z. and P.Z. provided the TCAD simulation. D.L. and J.J. designed and provided the readout circuit boards. L. X. and D. Z. synthesized the PbS CQDs and fabricated the CQDs photodiodes. S.Z., Y.W., and Z.Z. wrote the paper. All authors discussed the results and prepared and commented on the manuscript.

Competing interests

The authors declare no competing interests.

Additional information

The supplementary material is contained.

1. Xia F., Wang H., Di Xiao, Dubey M. & Ramasubramaniam A. Two-dimensional material nanophotonics. Nat. Photonics. 12, 899–907 (2014).
2. Saran R. & Curry R. J. Lead sulphide nanocrystal photodetector technologies. Nat. Photonics. 10, 81–92 (2016).
3. Liu C., Guo J., Yu L., Li J., Zhang M. & Li H. et al. Silicon/2D-material photodetectors: from near-infrared to mid-infrared. Light-Sci. Appl. 10, 123 (2021).
4. Rogalski A., Antoszewski J. & Faraone L. Third-generation infrared photodetector arrays. J. Appl. Phys. 105, 091101 (2009).
5. Rogalski A. Infrared Detectors. CRC press: (2000).
6. Onat B. M., Huang W., Masaun N., Lange M., Ettenberg M. H. & Dries C. Ultra-low dark current InGaAs technology for focal plane arrays for low-light level visible-shortwave infrared imaging. 2007 2007-01-01: SPIE; 2007. p. 65420L-65420L-9.
7. InGaAs photodiode 800–1700 nm. FGA21; https://www.thorlabs.com
8. InGaAs PIN photodiode. G6854-01; https://www.hamamatsu.com.
9. Goossens S., Navickaite G., Monasterio C., Gupta S., Piqueras J. J. & Pérez R. et al. Broadband image sensor array based on graphene–CMOS integration. Nat. Photonics. 11, 366–371 (2017).
10. Rudan M. Physics of Semiconductor Devices. Springer Science & Business Media, 2005
11. Kang Y., Liu H., Morse M., Paniccia M. J., Zadka M. & Litski S. et al. Monolithic germanium/silicon avalanche photodiodes with 340 GHz gain–bandwidth product. Nat. Photonics. 3, 59–63 (2009).
12. Konstantatos G., Badioli M., Gaudreau L., Osmond J., Bernechea M. & Garcia D. A. F. et al. Hybrid graphene-quantum dot phototransistors with ultrahigh gain. Nat. Nanotechnol. 7, 363-8 (2012).
13. Bielecki Z., Achtenberg K., Kopytko M., Mikołajczyk J., Wojtas J. & Rogalski A. Review of photodetectors characterization methods. Bulletin of the Polish Academy of Sciences Technical Sciences, 70, 140534–140534 (2022).
14. Jiang X., Itzler M., ODonnell K., Entwistle M., Owens M. & Slomkowski K. et al. InP-Based Single-Photon Detectors and Geiger-Mode APD Arrays for Quantum Communications Applications. IEEE J. Sel. Top. Quantum Electron. 21, 5–16 (2015).
15. Wang B. & Mu J. High-speed Si-Ge avalanche photodiodes. PhotoniX. 3, (2022).
16. McIntyre R. J. Multiplication noise in uniform avalanche diodes. IEEE Trans. Electron Devices. ED-13, 164–168 (1966).
17. Fang H. & Hu W. Photogating in Low Dimensional Photodetectors. Adv. Sci. 4, 1700323 (2017).
18. Adinolfi V. & Sargent E. H. Photovoltage field-effect transistors. Nature. 542, 324–327 (2017).
19. Zhang M., Chi Z., Wang G., Fan Z., Wu H. & Yang P. et al. An Irradiance-Adaptable Near‐Infrared Vertical Heterojunction Phototransistor. Adv. Mater. 34, 2205679 (2022).
20. Yang B., Wang Y., Li L., Zhang J., Wang J. & Jiao H. et al. High Performance Ternary Organic Phototransistors with Photoresponse up to 2600 nm at Room Temperature. Adv. Funct. Mater. 31, 2103787 (2021).
21. Akinwande D., Huyghebaert C., Wang C., Serna M. I., Goossens S. & Li L. et al. Graphene and two-dimensional materials for silicon technology. Nature. 573, 507–518 (2019).
22. Zhang C., Yin X., Qian G., Sang Z., Yang Y. & Que W. Gate Voltage Adjusting PbS-I Quantum‐Dot‐Sensitized InGaZnO Hybrid Phototransistor with High‐Sensitivity. Adv. Funct. Mater. 34, 2308897 (2024).
23. Kim S. H., Lee D., Moon S., Choi J. H., Kim D. & Kim J. et al. Sulfurized Colloidal Quantum Dot/Tungsten Disulfide Multi-Dimensional Heterojunction for an Efficient Self‐Powered Visible‐to‐SWIR Photodetector. Adv. Funct. Mater. 33, 2303778 (2023).
24. Luo P., Zhuge F., Wang F., Lian L., Liu K. & Zhang J. et al. PbSe Quantum Dots Sensitized High-Mobility Bi2O2Se Nanosheets for High-Performance and Broadband Photodetection Beyond 2 ∝m. ACS Nano. 13, 9028–9037 (2019).
25. He J., Liu H., Huang C., Jia Y., Li K. & Mesli A. et al. Analytical Transient Responses and Gain–Bandwidth Products of Low-Dimensional High-Gain Photodetectors. ACS Nano. 15, 20242–20252 (2021).
26. Chen D., March S. D., Jones A. H., Shen Y., Dadey A. A. & Sun K. et al. Photon-trapping-enhanced avalanche photodiodes for mid-infrared applications. Nat. Photonics. 17, 594–600 (2023).
27. Shi Y., Li X., Chen G., Zou M., Cai H. & Yu Y. et al. Avalanche photodiode with ultrahigh gain–bandwidth product of 1,033 GHz. Nat. Photonics, 1–7 (2024).
28. Liu J., Liu P., Chen D., Shi T., Qu X. & Chen L. et al. A near-infrared colloidal quantum dot imager with monolithically integrated readout circuitry. Nat. Electron. 5, 443–451 (2022).
29. Xia Y., Chen W., Zhang P., Liu S., Wang K. & Yang X. et al. Facet Control for Trap-State Suppression in Colloidal Quantum Dot Solids. Adv. Funct. Mater. 30, 2000594 (2020).
30. Zhang X., Huang H., Yao X., Li Z., Zhou C. & Zhang X. et al. Ultrasensitive Mid-wavelength Infrared Photodetection Based on a Single InAs Nanowire. ACS Nano. 13, 3492–3499 (2019).
31. Yang Y., Ding L., Chen H., Han J., Zhang Z. & Peng L. Carbon nanotube network film-based ring oscillators with sub 10-ns propagation time and their applications in radio-frequency signal transmission. Nano Res. 11, 300–310 (2018).
32. McNeill C. R., Hwang I. & Greenham N. C. Photocurrent transients in all-polymer solar cells: Trapping and detrapping effects. J. Appl. Phys. 106, 024507 (2009).
33. Nissan-Cohen Y., Shappir J. & Frohman-Bentchkowsky D. Dynamic model of trapping-detrapping in SiO2. J. Appl. Phys. 58, 2252–2261 (1985).
34. InGaAs photodiode 800–1700 nm. FGA015; https://www.thorlabs.com
35. Clifford J. P., Konstantatos G., Johnston K. W., Hoogland S., Levina L. & Sargent E. H. Fast, sensitive and spectrally tuneable colloidal-quantum-dot photodetectors. Nat. Nanotechnol. 4, 40–44 (2009).
36. Konstantatos G., Howard I., Fischer A., Hoogland S., Clifford J. & Klem E. et al. Ultrasensitive solution-cast quantum dot photodetectors. Nature. 442, 180–183 (2006).
37. Bishop M. D., Hills G., Srimani T., Lau C., Murphy D. & Fuller S. et al. Fabrication of carbon nanotube field-effect transistors in commercial silicon manufacturing facilities. Nat. Electron. 3, 492–501 (2020).
38. Liu L., Han J., Xu L., Zhou J., Zhao C. & Ding S. et al. Aligned, high-density semiconducting carbon nanotube arrays for high-performance electronics. Science. 368, 850–856 (2020).
39. Zhou S., Wang Y., Deng C., Liu P., Zhang J. & Wei N. et al. Highly sensitive SWIR photodetector using carbon nanotube thin film transistor gated by quantum dots heterojunction. Appl. Phys. Lett. 120, 193103 (2022).

There is NO Competing Interest.

SISYzhouHGFETforweakinfraredradiationdetection20240708.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Opto-electrical decoupled phototransistor for starlight detection

Status:

Version 1

Abstract

Figures

Introduction

HGFET structure

Opto-electric decoupling mechanism

Merits of the HGFET

Gate-tuneable operating mode of the HGFET

Performance comparison and benchmarking

Conclusion

Methods

Materials preparation

HGFET fabrication

HGFET imager fabrication

Characterization of the materials and device

Image capture

Detector characteristics

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1