Defect inspection is critical in semiconductor manufacturing for product quality improvement at reduced production costs. A whole new manufacturing process is often associated with a new set of defects that can cause serious damage to the manufacturing system. Therefore, classifying existing defects and new defects provides crucial clues to fix the issue in the newly introduced manufacturing process. We present a multi-task hybrid transformer (MT-former) that distinguishes novel defects from the known defects in electron microscope images of semiconductors. MT-former consists of upstream and downstream training stages. In the upstream stage, an encoder of a hybrid transformer is trained by solving both classification and reconstruction tasks for the existing defects. In the downstream stage, the shared encoder is fine-tuned by simultaneously learning the classification as well as a deep support vector domain description (Deep-SVDD) to detect the new defects among the existing ones. With focal loss, we also design a hybrid-transformer using convolutional and an efficient self-attention module. Our model is evaluated on real-world data from SK Hynix and on publicly available data from magnetic tile defects and HAM10000. For SK Hynix data, MT-former achieved higher AUC as compared with a Deep-SVDD model, by 8.19% for anomaly detection and by 9.59% for classifying the existing classes. Furthermore, the best AUC (magnetic tile defect 67.9%, HAM10000 70.73%) on the public dataset achieved with the proposed model implies that MT-former would be a useful model for classifying the new types of defects from the existing ones.

Monolithic multi-freeform optical structures play significant roles in advanced optical systems by simplifying system structures and enhancing optoelectronic performance. However, manufacturing and measurement present significant challenges, which require the simultaneous assurance of form quality and relative positioning of multiple functional surfaces. Consequently, a deterministic form-position deflectometric measuring method is proposed based on Bayesian multisensor fusion, which effectively overcomes the inherent limitation of deflectometry in absolute positioning. Calibration priors were marginalised in the measurement model to improve fidelity, and a fully probabilistic measurement framework was proposed to eliminate numerical bias in conventional sequential optimisation approaches. Finally, a geometric-constraint-based registration method was developed to evaluate the form-position quality of freeform surfaces. The experimental results demonstrated the measurement accuracy could achieve a level of one hundred nanometres for surface forms and a few microns for surface positions.

Halide perovskite-based photodiodes are promising for efficient detection across a broad spectral range. Perovskite absorber thin-films have a microcrystalline morphology, characterized by a high density of surface states and defects at inter-grain interfaces. In this work, we used dielectric/ferroelectric poly(vinylidene-fluoride-trifluoroethylene) (P(VDF-TrFE)) to modify the bulk interfaces and electron transport junction in p-i-n perovskite photodiodes. Our complex work demonstrates that interface engineering with P(VDF-TrFE) induces significant Fermi level pinning, reducing from 4.85 eV for intrinsic perovskite to 4.28 eV for the configuration with dielectric interlayers. Modifying the interfaces in the devices resulted in an increase in the key characteristics of photodiodes compared to pristine devices. The integration of P(VDF-TrFE) into the perovskite film didn’t affect the morphology and crystal structure, but significantly changed the charge transport and device performance. IV curve analysis and 2-diode model calculations showed enhanced shunt properties, a decreased non-ideality factor, and reduced saturation dark current. We have shown that the complex introduction of P(VDF-TrFE) into the absorber’s bulk and on its surface is essential to reduce the impact of the trapping processes. For P(VDF-TrFE) containing devices, we increased the specific detectivity from 10¹¹ to ~10¹² Jones, expanded the linear dynamic range up to 100 dB, and reduced the equivalent noise power to 10⁻¹³ W·Hz^−1/2. Reducing non-radiative recombination contributions significantly enhanced device performance, improving rise/fall times from 6.3/10.9 µs to 4.6/6.5 µs, and achieved photo-response dynamics competitive with state-of-the-art analogs. The cut-off frequency (3dB) increased from 64.8 kHz to 74.8 kHz following the introduction of the dielectric. We also demonstrated long-term stabilization of PPD performance under heat-stress. These results provide new insights into the use of organic dielectrics and an improved understanding of trap-states/ion defect compensation for detectors based on perovskite heterostructures.

Detecting electrolyte leakage is an effective early warning approach for abnormal faults in lithium-ion batteries (LIBs) and can help mitigate safety risks such as fires and explosions. However, detecting electrolyte leakage in the early stages of LIB faults presents a significant challenge, as leaks in LIBs produce volatile organic compounds (VOCs) at parts per million levels that are difficult to detect using conventional VOC sensors. Here, an effective LIB VOC sensor using micro-nano optical fibres (MNFs) has been developed for the first time, coated with an in situ self-assembled zeolitic imidazolate framework-8 (ZIF-8) membrane as an electrolyte-sensitive layer. The abundance of pores in ZIF-8 is excellent for adsorbing a variety of VOCs, including diethyl carbonate, ethyl methyl carbonate, dimethyl carbonate, and propylene carbonate. The MNFs possess high refractive index sensitivity, enhancing the online monitoring of electrolytes. MNFs with a diameter of approximately 7 μm were assembled with four-cycle ZIF-8 of approximately 500 nm thickness, as the fabricated sensor. Through wavelength demodulation, the LIB sensor demonstrated high sensitivity, detecting 43.6 pm/ppm of VOCs and exhibiting rapid response and recovery times of typically within 10 min and 23 s, respectively, as well as a low theoretical detection limit of 2.65 ppm for dimethyl carbonate vapor with excellent reversibility. The first on-site verification of online LIB leakage monitoring demonstrated that the sensor achieved a 35 h early warning prior to full-load leakage, thus exhibiting promising prospects for applications in scenarios such as car batteries.

Deformable mirrors are essential dynamic wavefront compensation. Among the various types of deformable mirrors with different actuation mechanisms, the unimorph piezoelectric deformable mirror (UPDM) offers distinct advantages owing to its compact size and low cost. The two most critical challenges in UPDM are electrode fabrication and deformation control. This study proposes an integrated electrode fabrication and sensorless feedback control scheme for UPDM, which simplifies the manufacturing process and enhances its performance. The electrode array is created using silkscreen printing combined with flexible printed circuit board technology, integrating electrode fabrication and electrical connection into a single step. The desired mirror deformation is achieved by introducing a closed-loop direct deformation control method based on piezoelectric self-sensing. The feedback mechanism utilizes the local strain-induced charge generated by the piezoelectric plate, effectively addressing the nonlinear behavior of the piezoelectric material. Experimental results confirm the feasibility and effectiveness of the proposed method, with the maximum relative error in the steady state phase remaining below 2%.

Melanoma, a highly malignant and complex form of cancer, has increased in global incidence, with a growing number of new cases annually. Active targeting strategies, such as leveraging the α-melanocyte-stimulating hormone (αMSH) and its interaction with the melanocortin 1 receptor (MC1R) overexpressed in melanoma cells, enhance the concentration of therapeutic agents at tumor sites. For instance, targeted delivery of plasmonic light-sensitive agents and precise hyperthermia management provide an effective, minimally invasive treatment for tumors. In this work, we present a comparative study on targeted photothermal therapy (PTT) using plasmonic gold nanorods (Au NRs) as a robust and safe nanotool to reveal how key treatment parameters affect therapy outcomes. Using an animal model (B16-F10) of melanoma tumors, we compare the targeting abilities of Au NRs modified with two different MC1R agonists, either closely mimicking the αMSH sequence or providing a superior functionalization extent of Au NRs (4.5% (w/w) versus 1.8% (w/w)), revealing 1.6 times better intratumoral localization. Following theoretical and experimental assessments of the heating capabilities of the developed Au NRs under laser irradiation in either the femtosecond (FS)- or nanosecond (NS)- pulsed regime, we perform targeted PTT employing two types of peptide-modified Au NRs and compare therapeutic outcomes revealing the most appropriate PTT conditions. Our investigation reveals greater heat release from Au NRs under irradiation with FS laser, due to the relaxation rates of the electron and phonon temperatures dissipating in the surrounding, which correlates with a more pronounced 17.6 times inhibition of tumor growth when using FS-pulsed regime.

In this paper, we experimentally demonstrate a non-volatile switchable infrared stealth metafilm based on high temperature resistant metal Molybdenum (Mo) and phase change material Ge₂Sb₂Te₅(GST). By controlling the phase state of GST, the switch between the infrared stealth and the non-stealth states can be realized. Specifically, when the GST is in the amorphous state, the emissivity of the film in the 3−5 μm and 8−14 μm atmospheric window band is suppressed and can realize infrared stealth, together with a high absorption peak of 94% at 6.08 μm, which enables radiative heat dissipation; While for the crystalline state of the GST, the average emissivity is more than 0.7 in the band of 8−14 μm, and the infrared stealth function cannot be realized. When the background temperature is 100°C, the temperature difference between the two samples reaches as high as 28°C under an infrared thermal imager. Therefore, our proposed metafilm can flexibly regulate the infrared thermal radiation of the target so as to realize the switch between the infrared stealth and non-stealth state. We have fabricated the metafilm on both hard and flexible substrates. Our work holds profound significance for the study of dynamic thermal radiation control and it is set to pave the way for the practical implementation of intelligent infrared stealth technology.