Precision, high‐speed measurement techniques based on optical frequency combs have rapidly advanced, demonstrating broad utility in distance metrology, spectroscopy, and interferometry. In light of these developments, we developed a microscopic interferometry using a 10‐GHz electro‐optic comb broadened by nonlinear effects, enabling high‐accuracy 3D profiling over several hundred micrometers to several millimeters without mechanical scanning and with electronically controlled comb‐spacing sweeps. However, symmetric comb‐spacing sweeps fix the center wavelength, causing phase localization and fringe‐free interference peaks whose amplitudes vary with phase, producing an unmeasurable region at π/2 as reported previously. To address this limitation, we incorporated sinusoidal phase modulation (SPM) technique, thereby eliminating the unmeasurable region by rendering the π/2‐phase null interference signal detectable. Furthermore, aligning amplitude‐peak and stationary‐phase distributions obtained from comb‐spacing sweeps and SPM enabled submicron‐accurate 3D full‐field topography. This phase‐stabilized tomography approach achieved displacement accuracy beyond the limit set by the interference‐peak width. The technique’s effectiveness is demonstrated through measurements of a step‐height standard, a 10‐yen coin surface, and a glass plate, confirming its capability for high‐speed, high‐precision full‐field profiling and tomography.

Abstract A new method for thin-film thickness measurement based on the spectral phase of a white-light interference signal is proposed. In this method, the refractive-index distribution of the substrate is not required. The method is particularly useful when the refractive-index distribution of the substrate cannot be determined accurately because of its multilayer structure. Simulations indicate that the proposed method has a systematic bias of approximately 1.0 nm for a 294-nm-thick indium tin oxide (ITO) film. In experiments, the average measured thickness was 293.5 nm after correcting the bias, which agrees well with the value of 293.8 nm measured using ellipsometry. Furthermore, excellent repeatability of 0.1 nm was achieved experimentally.

Optical‑frequency‑comb‑based metrology has rapidly advanced, enabling high‑precision and high‑speed measurements in ranging, spectroscopy, and interferometry. Building on these developments, we previously developed a microscopic interferometry using a 10‑GHz electro‑optic comb broadened by nonlinear effects, enabling high‑accuracy 3D profiling over several hundred micrometers to several millimeters without mechanical scanning and with electronically controlled comb‑spacing sweeps. However, symmetric comb‑spacing sweeps fix the center wavelength, causing phase localization and fringe‑free interference peaks whose amplitudes vary with phase and vanish at π/2, result in unmeasurable blind regions as reported previously. To address this limitation, we incorporated sinusoidal phase modulation (SPM) technique, thereby eliminating the blind regions by rendering the π/2‑phase null interference signal detectable. Furthermore, aligning amplitude‑peak and stationary‑phase distributions obtained from comb‑spacing sweeps and SPM enabled submicron‑accurate 3D full‑field topography. This phase‑stabilized tomography approach achieved displacement accuracy beyond the limit imposed by the interference‑peak width. The system acquires a 640 × 518 × 865‑voxel volume within one second, enabling high‑speed full‑field tomography. The technique’s effectiveness is demonstrated through measurements of a step‑height standard, a non‑specular coin surface, and a glass plate, confirming its capability for high‑speed, high‑precision full‑field profiling and tomography.

With the aim of applying complex master/slave interferometry (CMSI) to polarization-sensitive optical coherence tomography (PS-OCT) for birefringent tomographic measurements of biological tissue, we present the impact of temperature instability on an all-fiber-based depth-encoded PS-OCT system and the practicality of temperature control for birefringence measurements. In our PS-OCT system, two orthogonally polarized interrogating beams were separated to a depth of approximately 1 mm using a 5-meter long polarization-maintaining fiber (PMF) as a passive delay unit, which is susceptible to temperature instability. The variation in resolution and delay due to temperature change of PMF were investigated. Furthermore, it is shown that tomographic birefringence can be measured under temperature-controlled operation utilizing the advantages of CMSI. We found that changes to the location of generated masks caused by an emerging temperature drift between the channels can be corrected with our presented characterization.

Ultrasound, or sound at frequencies exceeding the conventional range of human hearing, is not only audible to mice, microbats, and dolphins, but also creates an auditory sensation when delivered through bone conduction in humans. Although ultrasound is utilized for brain activation and in hearing aids, the physiological mechanism of ultrasonic hearing remains unknown. In guinea pigs, we found that ultrasound above the hearing range delivered through ossicles of the middle ear evokes an auditory brainstem response and a mechano-electrical transduction current through hair cells, as shown by the local field potential called the cochlear microphonic potential (CM). The CM synchronizes with ultrasound, and like the response to audible sounds is actively and nonlinearly amplified. In vivo optical nano-vibration analysis revealed that the sensory epithelium in the hook region, the basal extreme of the cochlear turns, resonates in response both to ultrasound within the hearing range and to harmonics beyond the hearing range. The results indicate that hair cells can respond to stimulation at the optimal frequency and its harmonics, and the hook region detects ultrasound stimuli with frequencies more than two octaves higher than the upper limit of the ordinary hearing range.

This study introduces a technique for generating stochastic electromagnetic (SEM) beams using a modified degenerate cavity laser in which one mirror is substituted with a spatial light modulator (SLM). We propose two methods to manipulate the spatial coherence of SEM beams: the first involves adjusting the size of a spatial filter within the laser cavity, which alters the number of oscillating transverse modes and thus varies the spatial coherence. The second method employs phase modulation by applying a dynamic random phase to the SLM. This dual approach allows for precise control over the spatial coherence properties of SEM beams. Experimental results demonstrate the generation of SEM beams using an SLM within a modified degenerate cavity laser and reveal a correlation between two orthogonal polarization components of the beams.

Measurements of phase refractive indexes of thick objects by using fitting method have been reported in many papers, but the fitting method produces errors in the fitted coefficients of a fitting function. In this paper it is made clear that the thickness of object, for which the phase refractive index can be measured exactly, is limited by the errors. Phase refractive indexes of three kinds of glass plates of 1 mm thickness are measured directly from spectral phases detected with a spectrally resolved interferometer. Instead of using the fitting method, the 2π phase ambiguity contained in a detected spectral phase is determined by using an assumption that an actual refractive index measured in experiment is almost the same as well-known data of the refractive index. The actual refractive indexes measured with an error less than 8 × 10 −5 are slightly different from the well-known data about the slope and the constant value in the distributions.

The need to repeatedly shuttle around synaptic weight values from memory to processing units has been a key source of energy inefficiency associated with hardware implementation of artificial neural networks. Analog in-memory computing (AIMC) with spatially instantiated synaptic weights holds high promise to overcome this challenge, by performing matrix-vector multiplications (MVMs) directly within the network weights stored on a chip to execute an inference workload. However, to achieve end-to-end improvements in latency and energy consumption, AIMC must be combined with on-chip digital operations and communication to move towards configurations in which a full inference workload is realized entirely on-chip. Moreover, it is highly desirable to achieve high MVM and inference accuracy without application-wise re-tuning of the chip. Here, we present a multi-core AIMC chip designed and fabricated in 14-nm complementary metal-oxide-semiconductor (CMOS) technology with backend-integrated phase-change memory (PCM). The fully-integrated chip features 64 256x256 AIMC cores interconnected via an on-chip communication network. It also implements the digital activation functions and processing involved in ResNet convolutional neural networks and long short-term memory (LSTM) networks. We demonstrate near software-equivalent inference accuracy with ResNet and LSTM networks while implementing all the computations associated with the weight layers and the activation functions on-chip. The chip can achieve a maximal throughput of 63.1 TOPS at an energy efficiency of 9.76 TOPS/W for 8-bit input/output matrix-vector multiplications.

Linear-wavenumber swept-source optical coherence tomography (SS-OCT) enables real-time, high-quality OCT imaging by eliminating the need for data resampling, as required in conventional SS-OCT. In this study, we introduced a high-performance linear-wavenumber swept source ( k -SS) with a broad scanning range and high output power. The linear k -SS is an acousto-optic-modulator-based external-cavity laser diode analogous to the Littrow configuration. The k -SS exhibits strong linearity in the 1.3 µm region, justified by a high goodness of fit R 2 value of 0.9998. Additionally, its scanning range, output power, and linewidth are 120 nm, more than 43 mW, and approximately 1.6 nm, respectively. The sweep rate is 280 Hz after the linear k compensation of the experimental equipment. We demonstrated the effectiveness of the linear k -SS by applying it to measure a sample distribution without k -domain resampling before the Fourier transform. This successful implementation indicates that the linear k -SS has practical potential for application in SS-OCT systems.

We discuss the process challenges such as the heater geometry effect, heater patterning processes and deep via formation. Based on the electrical data, we improve the processes and Phase Change memory (PCM) resistance distribution to meet Analog Computing requirement.

Analog non-volatile memory (NVM)-based accelerators for deep neural networks implement multiply-accumulate (MAC) operations – in parallel, on large arrays of resistive devices – by using Ohm’s law and Kirchhoff’s current law. By completely avoiding weight motion, such fully weight-stationary systems can offer a unique combination of low latency, high throughput, and high energy-efficiency (e.g., high TeraOPS/W). Yet since most Deep Neural Networks (DNNs) require only modest (e.g., 4-bit) precision in synaptic operations, such systems can still deliver “software-equivalent” accuracies on a wide range of models. We describe a 14-nm inference chip, comprising multiple 512×512 arrays of Phase Change Memory (PCM) devices, which can deliver software-equivalent inference accuracy for MNIST handwritten-digit recognition and recurrent LSTM benchmarks, and discuss various PCM challenges such as conductance drift and noise.

We discuss inline electrical testing to monitor the baseline of Analog Computing hardware using Phase Change Memory (PCM) technology. Tightening the PCM resistance distribution is necessary to meet analog computation requirement. A new yield methodology is introduced. A study of heater process variation, which will affect the heater height and the PCM resistance, will be discussed.

A polarization-sensitive optical coherence tomography (PS-OCT) system is able to not only show the structure of samples through the analysis of backscattered light, but is also capable of determining their polarimetric properties. This is an extra functionality to OCT which allows the retardance and axis orientation of a bulk sample to be determined. Here, we describe the temperature instabilities of a depth-encoded, multiple input state PS-OCT system, where two waves corresponding to two orthogonal states in the interrogating beam are delayed using a 5-meter long polarization-maintaning (PM) fiber. It is shown that the temperature not only affects the delay between the two relatively delayed waves, but also the amount of mismatched dispersion in the interferometer, which ultimately affects the achievable axial resolution in the system. To this end, the technique of complex master/slave interferometry (CMSI) can be used as an option to mitigate this effect.

Analog in- memory computing (AIMC) using memristive devices is considered a promising Non-von Neumann approach for deep learning (DL) inference tasks. However, inaccuracies in the programming of devices, that are attributed to conductance variations, pose a key challenge toward achieving sufficient compute precision for DL inference. Fortunately, conduction variations in memristive devices, such as phase-change memory (PCM) devices, exhibit a strong state dependence. This state dependence can be exploited in synaptic unit cells that comprise more than one memristive device, to encode positive or negative weights. In such multi-memristive unit cells, we propose a method that optimally maps the weights to the device conductance values, by maximizing the number of devices at the stable SET and RESET states. We demonstrate that this method reduces the matrix-vector multiplication (MVM) error and is more resilient to non-ideal device retention characteristics. With this approach, we increase the mean experimental inference accuracy of a network trained for MNIST classification by 0.71% on two PCM-based AIMC cores, and the hardware-realistic simulated top-1 accuracy of a network trained for ImageNet classification by 0.28%, while significantly reducing variability across multiple experiment instances.

We propose a method for simultaneous tomographic vibration visualization of an entire volume of a target object using a scanning low-coherence interferometric microscope. By combining the difference frequency component produced by the phase modulated reference beam and the object vibration into the time-domain OCT signal, we could capture the internal vibration distribution at a frequency of 43 kHz, which far exceeds the frame rate of the CMOS image sensor as a detector, with 25 nm sensitivity.

The precise programming of crossbar arrays of unit-cells is crucial for obtaining high matrix-vector-multiplication (MVM) accuracy in analog in-memory computing (AIMC) cores. We propose a radically different approach based on directly minimizing the MVM error using gradient descent with synthetic random input data. Our method significantly reduces the MVM error compared with conventional unit-cell by unit-cell iterative programming. It also eliminates the need for high-resolution analog-to-digital converters (ADCs) to read the small unit-cell conductance during programming. Our method improves the experimental inference accuracy of ResNet-9 implemented on two phase-change memory (PCM)-based AIMC cores by 1.26%.

Vortex beams (VBs), a type of light beam with a spiral wavefront, have unique properties, such as the orbit angular momentum (OAM), and diverse applications in optical communications and optical trapping and tweezers. Therefore, accurate measurements and estimations of the phase distribution and topological charge are essential for their applications to ensure VB quality. In this paper, we employed a sinusoidal phase modulation (SPM) interferometry to measure the phase distributions of VBs and the topological charge of VBs were estimated by mean of a method of the process of unwrapped phase. The phase measurement of optical vortices generated by a spatial light modulator (SLM) demonstrated that the SPM interferometry-based technique had a high measurement accuracy with a simplified configuration. The estimation errors of the topological charges for various orders of VBs were within approximately 4%. The fluctuation in the surface of the SLM leading to the flatness of the wavefront was estimated to be 0.06 rad by 10 consecutive measurements

In this study, a novel restoration model for the data of optical coherence tomography (OCT) is proposed. An OCT device acquires a tomographic image of a specimen at the scale of a few micrometers using a near-infrared laser and has been frequently adopted to measure the structures of bio-tissues. In certain applications, OCT devices face the problem of extremely weak reflected light and require the help of image processing to estimate the distribution of reflected light hidden in various noises. OCT identifies tomographic structures by searching for peak interference locations and their intensities. Therefore, the challenge of OCT data restoration involves the problem of identifying the interference function and its deconvolution. In this study, a restoration method is given by reducing the problem to a regularized least-squares problem with a hard constraint for the latent refractive index distributions, and an algorithm is derived using a primal-dual splitting (PDS) framework. The PDS has the advantage of requiring no inverse matrix operation and is able to handle high-dimensional data. The significance of the proposed method is verified through simulations using artificial data, followed by an experiment conducted using actual observation of $64 \times 64 \times 5000$ sized voxels.

A high-speed full-field surface profile measurement technique using frequency tunable supercontinuum multigigaherz comb was proposed. The phase modulation to the reference path of the microscopic interferometer resolved the previous non-measurable problem, and simultaneous detection of two modalities with amplitude and phase was realized. 3D measurements with a scan speed of 865 (depth) × 512 (length) × 640 (width) voxels/s was achieved. The depth resolution was improved to be approximately 30 μm.