We present an unsupervised representation learning approach using videos without semantic labels. We leverage the temporal coherence as a supervisory signal by formulating representation learning as ...a sequence sorting task. We take temporally shuffled frames (i.e., in non-chronological order) as inputs and train a convolutional neural network to sort the shuffled sequences. Similar to comparison-based sorting algorithms, we propose to extract features from all frame pairs and aggregate them to predict the correct order. As sorting shuffled image sequence requires an understanding of the statistical temporal structure of images, training with such a proxy task allows us to learn rich and generalizable visual representation. We validate the effectiveness of the learned representation using our method as pre-training on high-level recognition problems. The experimental results show that our method compares favorably against state-of-the-art methods on action recognition, image classification, and object detection tasks.
Using thermal annealing process, graphene oxide (GO) films were synthesized into reduced graphene oxide (rGO) composite films. Compared with the NO2 gas sensors using ZnO and rGO sensing membranes, ...the performances of the NO2 gas sensors using ZnO-rGO sensing membrane were improved. The improvement mechanisms were attributed to the removal of oxygen-containing functional groups, the supply of electrons from the oxygen vacancies of ZnO material, and the formation of COZn bonds. To study the dependence of sensing performances on the ZnO/GO ratio, various ZnO/GO ratios in the ZnO-rGO sensing membrane were used in the NO2 gas sensors. For the best ratio of 0.08, the sensing responsivity, response time, and recovery time of the ZnO-rGO NO2 gas sensors were respectively improved to 47.4%, 6.2 min, and 15.5 min compared with 19.0%, 10.3 min, and 75.9 min of the rGO NO2 gas sensors operated under 100 ppm NO2 environment at room temperature. Furthermore, the minimum detective NO2 concentration of 5 ppm and linear sensing responsivity from 10 ppm to 100 ppm were achieved using the ZnO-rGO gas sensors operated at room temperature.
•Sensing performances improved with an increase of ZnO/GO ratio until 0.08.•High responsivity of 47.4% for ZnO-rGO NO2 gas sensors at room temperature.•Reduction of oxygen-containing functional group and formation of COZn bonds.•Reaction enhancement by suppling electron from oxygen vacancy of ZnO material.•Linear sensing responsivity from 10 ppm to 100 ppm NO2 gas environment.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
To modulate the cutoff wavelength of the metal-semiconductor-metal deep ultraviolet photodetectors (MSM DUV-PDs), the aluminum content of the aluminium oxide-alloyed gallium oxide (Ga 2 O 3 :Al 2 O 3 ...) absorption layers was deposited using radio frequency (RF) magnetron cosputter system with dual targets, in which the Ga 2 O 3 target was sputtered with an RF power of 100 W and the Al 2 O 3 target sputtered with various RF powers. The optical bandgap energy of the Ga 2 O 3 :Al 2 O 3 films increased with an increase of the Al content deposited with a larger sputtered RF power of the Al 2 O 3 target. In view of the suppression effect of the oxygen vacancy and the defect by the more Al-O bonds in the Ga 2 O 3 :Al 2 O 3 absorption layer, the noise performances and the detectivity of the MSM DUV-PDs were improved. For depositing the Al atomic percentage of 5.9% in the Ga 2 O 3 :Al 2 O 3 absorption layer, the resulting MSM DUV-PDs revealed the cutoff wavelength of 230 nm, the noise equivalent power of 2.80 × 10 -12 W, and the detectivity of 1.37 × 10 11 cmHz0.5W -1 . Furthermore, the dominant low-frequency noise source was the flicker noise.
In this letter, monolithic red, green, and blue (RGB) micro light-emitting diodes (LEDs) were fabricated using gallium nitride based blue micro LEDs and quantum dots (QDs). Red and green QDs were ...sprayed onto individual region surrounded by patterned black matrix photoresist on the blue micro LEDs to form color conversion layers. Owing to its light-blocking capability, the patterned black matrix photoresist improved the contrast ratio of the micro LEDs from 11 to 22. To enhance the color conversion efficiency and the light output intensity, a hybrid Bragg reflector (HBR) was deposited on the bottom side of the monolithic RGB micro LEDs, thus reflecting the RGB light emitted to the substrate. To further improve the color purity of the red and green light, a distributed Bragg reflector (DBR) with high reflection for the blue light was deposited on the top side of the QDs/micro LEDs. The red and green light output intensities of the micro LEDs with HBR and DBR were enhanced by about 27%.
Recent image-to-image (I2I) translation algorithms focus on learning the mapping from a source to a target domain. However, the continuous translation problem that synthesizes intermediate results ...between two domains has not been well-studied in the literature. Generating a smooth sequence of intermediate results bridges the gap of two different domains, facilitating the morphing effect across domains. Existing I2I approaches are limited to either intra-domain or deterministic inter-domain continuous translation. In this work, we present an effectively signed attribute vector, which enables
continuous
translation on
diverse
mapping paths
across
various domains. In particular, we introduce a unified attribute space shared by all domains that utilize the sign operation to encode the domain information, thereby allowing the interpolation on attribute vectors of different domains. To enhance the visual quality of continuous translation results, we generate a trajectory between two sign-symmetrical attribute vectors and leverage the domain information of the interpolated results along the trajectory for adversarial training. We evaluate the proposed method on a wide range of I2I translation tasks. Both qualitative and quantitative results demonstrate that the proposed framework generates more high-quality continuous translation results against the state-of-the-art methods.
Full text
Available for:
CEKLJ, DOBA, EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, IZUM, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, SIK, UILJ, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
Due to its high carrier mobility and electron transmission, the phenyl-C61-butyric acid methyl ester (PC61BM) is usually used as an electron transport layer (ETL) in perovskite solar cell (PSC) ...configurations. However, PC61BM films suffer from poor coverage on perovskite active layers because of their low solubility and weak adhesive ability. In this work, to overcome the above-mentioned shortcomings, 30 nm thick PC61BM ETLs with different concentrations were modeled. Using a 30 nm thick PC61BM ETL with a concentration of 50 mg/mL, the obtained performance values of the PSCs were as follows: an open-circuit voltage (Voc) of 0.87 V, a short-circuit current density (Jsc) of 20.44 mA/cm2, a fill factor (FF) of 70.52%, and a power conversion efficiency (PCE) of 12.54%. However, undesired fine cracks present on the PC61BM surface degraded the performance of the resulting PSCs. To further improve performance, multiple different thicknesses of ZnO interface layers were deposited on the PC61BM ETLs to release the fine cracks using a thermal evaporator. In addition to the pavement of fine cracks, the ZnO interface layer could also function as a hole-blocking layer due to its larger highest occupied molecular orbital (HOMO) energy level. Consequently, the PCE was improved to 14.62% by inserting a 20 nm thick ZnO interface layer in the PSCs.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
In this study, intrinsic Ga
2
O
3
(i
-
Ga
2
O
3
) film was deposited at about 80 K using a vapor cooling condensation system. Its bandgap energy was 5.0 eV. Low oxygen vacancy and defects were ...verified by using photoluminescence and Hall measurements. When a 40-nm-thick i
-
Ga
2
O
3
film was used as the gate dielectric layer of AlGaN/GaN metal-oxide-semiconductor high-electron mobility transistors (MOS-HEMTs), threshold voltage, and gate breakdown voltage were − 3.5 V and − 538.0 V, respectively. The associated gate leakage current of the devices operating at a gate-source voltage of − 100 V was 0.57 μA. Furthermore, a saturation drain-source current of 186.2 mA/mm and a maximum extrinsic transconductance of 85.8 mS/mm were obtained for the devices operating at a gate-source voltage of 0 V and a drain-source voltage of 10 V. The unit gain cutoff frequency and the maximum oscillation frequency were 5.7 GHz and 11.0 GHz, respectively. The normalized noise and Hooge’s coefficient were 3.79 × 10
−14
Hz
−1
and 5.06 × 10
−5
, respectively, when the devices operated at a frequency of 100 Hz, with a drain-source voltage of 1 V and a gate-source voltage of 5 V.
Full text
Available for:
EMUNI, FIS, FZAB, GEOZS, GIS, IJS, IMTLJ, KILJ, KISLJ, MFDPS, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, SBMB, SBNM, UKNU, UL, UM, UPUK, VKSCE, ZAGLJ
To extend the sensing wavelength band, triple ultraviolet-band metal-semiconductor-metal ultraviolet photodetectors (MSM-UVPDs) were fabricated and studied in this study. Using a radio frequency ...magnetron sputtering system, the ZnO films, TiO 2 films, and AlN films were sequentially stacked on quartz substrates. The cutoff wavelengths of the ZnO-based, TiO 2 -based, and AlN-based MSM-UVPDs were 370, 310, and 220 nm, respectively. The associated photoresponsivity of 47, 142, and 83 mA/W and the associated specific detectivity of 2.02×10 10 , 9.15×10 10 , and 2.20×10 11 cm·Hz 1/2 · W -1 were obtained, respectively. By probing the common pad and each UV pad, the performance of the UVA, UVB, and UVC wavelength band could be measured, respectively. The performance of dualband could be obtained by probing the common pad and the two connected UV pads. Moreover, by probing the common mode and the three connected UV pads in the stacked MSM-UVPDs, the performance of triple ultraviolet bands could be obtained. Because the photoresponsivity in the visible wavelength band was very low, the triple-band MSM-UVPDs can work as solar blind UVPDs. The dominant noise was the flicker noise.
Photodetectors based on reduced graphene oxide (rGO) have attracted much attention owing to their simple and low‐cost fabrication process. However, the aggregation and defects of rGO flakes still ...limit the performance of rGO photodetectors. Controlling the composition of rGO has become a vital factor for its prospective applications. For example, the interconnection between rGO and polymers for modified morphologies of rGO films leads to an enhanced performance of devices. In this work, a practical approach to engineer surface uniformity and enhance the performance of a photodetector by modifying the rGO film with hydrophilic polymers poly(vinyl alcohol) (PVA) is reported. Compared with the rGO photodetector, the on/off ratio for the PVA/rGO photodetector shows 3.5 times improvement, and the detectivity shows 53% enhancement even when the photodetector is operated at a low bias of 0.3 V. This study provides an effective route to realize PVA/rGO photodetectors with a low‐power operation which shows promising opportunities for the future development of green systems.
A practical approach to enhance the performance of a reduced graphene oxide (rGO) photodetector by modifying the rGO film with poly(vinyl alcohol) (PVA) is reported. Compared with the rGO photodetector, the on/off ratio for the PVA/rGO photodetector shows 3.5 times improvement, and the detectivity shows 53% enhancement at a low operating bias of 0.3 V.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SBCE, SBMB, UL, UM, UPUK
In this article, to study the features of the quadruple gate and quadruple T-gate structures in AlGaN/ GaN fin-nanochannel array metal-oxide-semiconductor high-electron-mobility transistors ...(MOSHEMTs), an ELS-7500 electron-beam lithography system was used to pattern the 80-nm-wide fin-nanochannel array, and the quadruple gate and quadruple T-gate structures. Compared with the conventional single-gate structure, the saturation drain-source current was improved from 1153 to 1958 mA/mm and 2009 mA/mm by using the quadruple gate structure and quadruple T-gate structure, respectively. Furthermore, the maximum extrinsic transconductance was also improved from 265 to 323 mS/mm and 340 mS/mm, respectively. Compared with the unit gain cutoff frequency (<inline-formula> <tex-math notation="LaTeX">{f}_{T}{)} </tex-math></inline-formula> of 7.7 GHz and the maximum oscillation frequency (<inline-formula> <tex-math notation="LaTeX">{f}_{{\text {max}}}{)} </tex-math></inline-formula> of 15.1 GHz of the conventional single gate, the <inline-formula> <tex-math notation="LaTeX">{f}_{T} = {12.0} </tex-math></inline-formula> GHz and <inline-formula> <tex-math notation="LaTeX">{f}_{{\text {max}}} = {18.6} </tex-math></inline-formula> GHz were obtained by using the quadruple gate structure. If the quadruple T-gate structure was utilized, the <inline-formula> <tex-math notation="LaTeX">{f}_{T} </tex-math></inline-formula> and <inline-formula> <tex-math notation="LaTeX">{f}_{{\text {max}}} </tex-math></inline-formula> were further improved to 14.8 and 29.5 GHz. By simulating the equivalent circuits, the improved mechanisms were attributed to the enhancement of transconductance and the reduction of parasitic gate resistance and parasitic gate capacitance.