1-Octanol-assisted ultra-small volume droplet microfluidics with nanoelectrospray ionization mass spectrometry.

BACKGROUND: Droplet microfluidics with push-pull and microdialysis sampling from brain slices, cultured cells and engineered tissues produce low volume mass limited samples containing analytes sampled from the extracellular space. This sampling approach coupled to mass spectrometry (MS) detection allows evaluation of time-dependent chemical changes. Our goal is an approach for continuous sampling and segregation of extracellular samples into picoliter droplets followed by the characterization of the droplets using nanoelectrospray ionization (nESI) MS. The main focus here is the optimization of the carrier oil for the microfluidic device that neither affects the stability of picoliter droplets nor compatibility with MS detection of a range of analytes. RESULTS: We developed and characterized a 1-octanol-assisted ultra-small volume droplet microfluidic nESI MS system for the analysis of neurotransmitters in distinct samples including cerebrospinal fluid (CSF). The use of a 1-octanol oil phase was effective for generation of aqueous droplets as small as 65 pL and enabled detection of acetylcholine (ACh) and gamma-aminobutyric acid (GABA) in water and artificial CSF. Continuous MS analysis of droplets for extended periods up to 220 min validated the long-term stability of droplet generation and analyte detection by nESI-MS. As an example, ACh response demonstrated a linear working range (R2 = 0.99) between 0.4 µM and 25 µM with a limit of detection of 370 nM (24 amol), enabling its quantitation in rodent CSF. SIGNIFICANCE: The established droplet microfluidics - nESI MS approach allows the analysis of microenvironments at high spatiotemporal resolution. The approach may allow microsampling and monitoring of spatiotemporal dynamics of neurochemicals and drugs in the brain and spinal cord of live animals.

Highly Localized Chemical Sampling at Subsecond Temporal Resolution Enabled with a Silicon Nanodialysis Platform at Nanoliter per Minute Flows.

Microdialysis (MD) is a versatile and powerful technique for chemical profiling of biological tissues and is widely used for quantification of neurotransmitters, neuropeptides, metabolites, biomarkers, and drugs in the central nervous system as well as in dermatology, ophthalmology, and pain research. However, MD performance is severely limited by fundamental tradeoffs between chemical sensitivity, spatial resolution, and temporal response. Here, by using wafer-scale silicon microfabrication, we develop and demonstrate a nanodialysis (ND) sampling probe that enables highly localized chemical sampling with 100 µm spatial resolution and subsecond temporal resolution at high recovery rates. These performance metrics, which are 100-1000× superior to existing MD approaches, are enabled by a 100× reduction of the microfluidic channel cross-section, a corresponding drastic 100× reduction of flow rates to exceedingly slow few nL/min flows, and integration of a nanometer-thin nanoporous membrane with high transport flux into the probe sampling area. Miniaturized ND probes may allow for the minimally invasive and highly localized sampling and chemical profiling in live biological tissues with high spatiotemporal resolution for clinical, biomedical, and pharmaceutical applications.

Highly localized chemical sampling at sub-second temporal resolution enabled with a silicon nanodialysis platform at exceedingly slow flows.

Microdialysis (MD) is a versatile and powerful technique for chemical profiling of biological tissues and is widely used for quantification of neurotransmitters, neuropeptides, metabolites, biomarkers, and drugs in the central nervous system as well as in dermatology, ophthalmology, and in pain research. However, MD performance is severely limited by fundamental tradeoffs between chemical sensitivity, spatial resolution, and temporal response. Here, by using wafer-scale silicon microfabrication, we develop and demonstrate a nanodialysis (ND) sampling probe that enables highly localized chemical sampling with 100µm spatial resolution and sub-second temporal resolution at high recovery rates. These performance metrics, which are 100X-1000X superior to existing MD approaches, are enabled by a 100X reduction of the microfluidic channel cross-section, a corresponding drastic 100X reduction of flow rates to exceedingly slow few nL/min flows, and integration of a nanometer-thin nanoporous membrane with high transport flux into the probe sampling area. Miniaturized ND probes may allow for the minimally invasive and highly localized sampling and chemical profiling in live biological tissues with unprecedented spatio-temporal resolution for clinical, biomedical, and pharmaceutical applications.

Pre-neuronal processing of haptic sensory cues via dispersive high-frequency vibrational modes.

Sense of touch is one of the major perception channels. Neural coding of object textures conveyed by rodents' whiskers has been a model to study early stages of haptic information uptake. While high-precision spike timing has been observed during whisker sweeping across textured surfaces, the exact nature of whisker micromotions that spikes encode remains elusive. Here, we discovered that a single micro-collision of a whisker with surface features generates vibrational eigenmodes spanning frequencies up to 10 kHz. While propagating along the whisker, these high-frequency modes can carry up to 80% of shockwave energy, exhibit 100× smaller damping ratio, and arrive at the follicle 10× faster than low frequency components. The mechano-transduction of these energy bursts into time-sequenced population spike trains may generate temporally unique "bar code" with ultra-high information capacity. This hypothesis of pre-neuronal processing of haptic signals based on dispersive temporal separation of the vibrational modal frequencies can shed light on neural coding of haptic signals in many whisker-like sensory organs across the animal world as well as in texture perception in primate's glabrous skin.

Enhancement of faradaic current in an electrochemical cell integrated into silicon microfluidic channels.

Implantable electrochemical sensors enable fast and sensitive detection of analytes in biological tissue, but are hampered by bio-foulant attack and are unable to be recalibrated in-situ. Herein, an electrochemical sensor integrated into ultra-low flow (nL/min) silicon microfluidic channels for protection from foulants and in-situ calibration is demonstrated. The small footprint (5 µm radius channel cross-section) of the device allows its integration into implantable sampling probes for monitoring chemical concentrations in biological tissues. The device is designed for fast scan cyclic voltammetry (FSCV) in the thin-layer regime when analyte depletion at the electrode is efficiently compensated by microfluidic flow. A 3X enhancement of faradaic peak currents is observed due to the increased flux of analytes towards the electrodes. Numerical analysis of in-channel analyte concentration confirmed near complete electrolysis in the thin-layer regime below 10 nL/min. The manufacturing approach is highly scalable and reproducible as it utilizes standard silicon microfabrication technologies.

Integrated silicon microfluidic chip for picoliter-scale analyte segmentation and microscale printing for mass spectrometry imaging.

A silicon single-chip microfluidics system that integrates microscale fluidic channels, an analyte segmentation device, and a nozzle for electrohydrodynamic-assisted printing is designed for hyphenation with MALDI mass spectrometry (MS) imaging. A miniaturized T-junction segments analytes into monodisperse picoliter oil-isolated compartments. The printing nozzle deposits generated droplets one-by-one into an array on a conductive substrate without splitting or coalescing. Virtually single-shot MS analysis is enabled due to the ultrasmall droplet volumes and highly localized printing. The signal-to-noise ratio indicates that detection limits at the attomole level are achieved for γ-aminobutyric acid.

Attomole-Level Multiplexed Detection of Neurochemicals in Picoliter Droplets by On-Chip Nanoelectrospray Ionization Coupled to Mass Spectrometry.

While droplet microfluidics is becoming an effective tool for biomedical research, sensitive detection of droplet content is still challenging, especially for multiplexed analytes compartmentalized within ultrasmall droplets down to picoliter volumes. To enable such measurements, we demonstrate a silicon-based integrated microfluidic platform for multiplexed analysis of neurochemicals in picoliter droplets via nanoelectrospray ionization (nESI)-mass spectrometry (MS). An integrated silicon microfluidic chip comprising downscaled 7 µm-radius channels, a compact T-junction for droplet generation, and an integrated nESI emitter tip is used for segmentation of analytes into picoliter compartments and their efficient delivery for subsequent MS detection. The developed system demonstrates effective detection of multiple neurochemicals encapsulated within oil-isolated plugs down to low picoliter volumes. Quantitative measurements for each neurochemical demonstrate limits of detection at the attomole level. Such results are promising for applications involving label-free and small-volume detection for monitoring a range of brain chemicals.

Droplet Microfluidics with MALDI-MS Detection: The Effects of Oil Phases in GABA Analysis.

Microfluidic and mass spectrometry (MS) methods are widely used to sample and probe the chemical composition of biological systems to elucidate chemical correlates of their healthy and disease states. Though matrix-assisted laser desorption/ionization-mass spectrometry (MALDI)-MS has been hyphenated to droplet microfluidics for offline analyses, the effects of parameters related to droplet generation, such as the type of oil phase used, have been understudied. To characterize these effects, five different oil phases were tested in droplet microfluidics for producing samples for MALDI-MS analysis. Picoliter to nanoliter aqueous droplets containing 0.1 to 100 mM γ-aminobutyric acid (GABA) and inorganic salts were generated inside a polydimethylsiloxane microfluidic chip and deposited onto a conductive glass slide. Optical microscopy, Raman spectroscopy, and MALDI-mass spectrometry imaging (MSI) of the droplet samples and surrounding areas revealed patterns of solvent and oil evaporation and analyte deposition. Optical microscopy detected the presence of salt crystals in 50-100 µm diameter dried droplets, and Raman and MSI were used to correlate GABA signals to the visible droplet footprints. MALDI-MS analyses revealed that droplets prepared in the presence of octanol oil led to the poorest detectability of GABA, whereas the oil phases containing FC-40 provided the best detectability; GABA signal was localized to the footprint of 65 pL droplets with a limit of detection of 23 amol. The effect of the surfactant perfluorooctanol on analyte detection was also investigated.

Droplet-assisted electrospray phase separation using an integrated silicon microfluidic platform.

We report a silicon microfluidic platform that enables monolithic integration of transparent micron-scale microfluidic channels, an on-chip segmentation of analyte flows into picoliter-volume droplets, and a nano-electrospray ionization emitter that enables spatial and temporal separation of oil and aqueous phases during electro-spray for subsequent mass spectrometry analysis.

Acceleration of Deep Neural Network Training with Resistive Cross-Point Devices: Design Considerations.

In recent years, deep neural networks (DNN) have demonstrated significant business impact in large scale analysis and classification tasks such as speech recognition, visual object detection, pattern extraction, etc. Training of large DNNs, however, is universally considered as time consuming and computationally intensive task that demands datacenter-scale computational resources recruited for many days. Here we propose a concept of resistive processing unit (RPU) devices that can potentially accelerate DNN training by orders of magnitude while using much less power. The proposed RPU device can store and update the weight values locally thus minimizing data movement during training and allowing to fully exploit the locality and the parallelism of the training algorithm. We evaluate the effect of various RPU device features/non-idealities and system parameters on performance in order to derive the device and system level specifications for implementation of an accelerator chip for DNN training in a realistic CMOS-compatible technology. For large DNNs with about 1 billion weights this massively parallel RPU architecture can achieve acceleration factors of 30, 000 × compared to state-of-the-art microprocessors while providing power efficiency of 84, 000 GigaOps∕s∕W. Problems that currently require days of training on a datacenter-size cluster with thousands of machines can be addressed within hours on a single RPU accelerator. A system consisting of a cluster of RPU accelerators will be able to tackle Big Data problems with trillions of parameters that is impossible to address today like, for example, natural speech recognition and translation between all world languages, real-time analytics on large streams of business and scientific data, integration, and analysis of multimodal sensory data flows from a massive number of IoT (Internet of Things) sensors.

Neural coding in barrel cortex during whisker-guided locomotion.

Animals seek out relevant information by moving through a dynamic world, but sensory systems are usually studied under highly constrained and passive conditions that may not probe important dimensions of the neural code. Here, we explored neural coding in the barrel cortex of head-fixed mice that tracked walls with their whiskers in tactile virtual reality. Optogenetic manipulations revealed that barrel cortex plays a role in wall-tracking. Closed-loop optogenetic control of layer 4 neurons can substitute for whisker-object contact to guide behavior resembling wall tracking. We measured neural activity using two-photon calcium imaging and extracellular recordings. Neurons were tuned to the distance between the animal snout and the contralateral wall, with monotonic, unimodal, and multimodal tuning curves. This rich representation of object location in the barrel cortex could not be predicted based on simple stimulus-response relationships involving individual whiskers and likely emerges within cortical circuits.

Binary phase-shift keying by coupling modulation of microrings.

We propose a coupling-modulated microring in an add-drop configuration for binary phase-shift keying (BPSK), where data is encoded as 0 and π radian phase-shifts on the optical carrier. The device uses the π radian phase-flip across the zero coupling point in a 2 × 2 Mach-Zehnder interferometer coupler to produce the modulation. The coupling-modulated microring combines the drive power reduction of resonant modulators with the digital phase response of Mach-Zehnder BPSK modulators. A proof-of-concept device was demonstrated in silicon-on-insulator, showing differential binary phase-shift keying operation at 5 and 10 Gb/s.

Cascaded Mach-Zehnder wavelength filters in silicon photonics for low loss and flat pass-band WDM (de-)multiplexing.

We present 1-to-8 wavelength (de-)multiplexer devices based on a binary tree of cascaded Mach-Zehnder-like lattice filters, and manufactured using a 90 nm CMOS-integrated silicon photonics technology. We demonstrate that these devices combine a flat pass-band over more than 50% of the channel spacing with low insertion loss of less than 1.6 dB, and have a small device size of approximately 500 × 400 µm. This makes this type of filters well suited for application as WDM (de-)multiplexer in silicon photonics transceivers for optical data communication in large scale computer systems.

High-speed receiver based on waveguide germanium photodetector wire-bonded to 90nm SOI CMOS amplifier.

The performance of a receiver based on a CMOS amplifier circuit designed with 90nm ground rules wire-bonded to a waveguide germanium photodetector is characterized at data rates up to 40Gbps. Both chips were fabricated through the IBM Silicon CMOS Integrated Nanophotonics process on specialty photonics-enabled SOI wafers. At the data rate of 28Gbps which is relevant to the new generation of optical interconnects, a sensitivity of -7.3dBm average optical power is demonstrated with 3.4pJ/bit power-efficiency and 0.6UI horizontal eye opening at a bit-error-rate of 10(-12). The receiver operates error-free (bit-error-rate < 10(-12)) up to 40Gbps with optimized power supply settings demonstrating an energy efficiency of 1.4pJ/bit and 4pJ/bit at data rates of 32Gbps and 40Gbps, respectively, with an average optical power of -0.8dBm.

Optimized light-matter interaction and defect hole placement in photonic crystal cavity sensors.

Photonic crystal slab cavities were investigated for increased light-matter interaction based on selective placement of sublattice hole sized defect holes inside L3 cavities. A multiple-hole defect (MHD) consisting of three defect holes placed in the regions of highest cavity mode field intensity were demonstrated through finite-difference time-domain simulations and experiments to exhibit the strongest light-matter interaction without introducing significant scattering losses. Compared to an L3 cavity without defect holes, these strategically designed three-hole MHD cavities presented higher quality factor and more than double the resonance wavelength shift upon exposure to a thin oxide and two small chemical molecules.

Drive-noise-tolerant broadband silicon electro-optic switch.

We report a broadband digital electro-optical switch, based upon a multi-stage Mach-Zehnder lattice design in silicon-on-insulator. A digital switching response is demonstrated, engineered through apodization of the coupling coefficients between stages. The digital switching behavior results in crosstalk lower than -15 dB for drive-voltage noise levels in excess of 300 mV(pp), which exceeds the noise tolerance of a conventional single-stage Mach-Zehnder switch by more than six-fold. In addition, the digital design enables a larger maximum 'on'-state extinction (below -26 dB) and lower 'on'-state free-carrier-induced insertion loss (less than 0.45 dB) than that of the single-stage switch. The noise-tolerant, low-crosstalk switch can thus play a key role within CMOS-integrated reconfigurable optical networks operating under noisy on-chip conditions.

Self-phase modulation and nonlinear loss in silicon nanophotonic wires near the mid-infrared two-photon absorption edge.

We report an experimental study of picosecond pulse propagation through a 4-mm-long Si nanophotonic wire with normal dispersion, at excitation wavelengths from 1775 to 2250 nm. This wavelength range crosses the mid-infrared two-photon absorption edge of Si at ~2200 nm. Significant reduction in nonlinear loss due to two-photon absorption is measured as excitation wavelengths approach 2200 nm. At high input power, self-phase modulation is clearly demonstrated by the development of power-dependant spectral fringes. Asymmetry and blue-shift in the appearance of the spectral fringes at 1775 nm versus 2200 nm is further shown to originate from a strong reduction in the intra-pulse density of two-photon absorption-generated free carriers and the associated free-carrier dispersion. Analysis of experimental data and comparison with numerical simulations illustrates that the two-photon absorption coefficient ß(TPA) obtained here from nanophotonic wire measurements is in reasonable agreement with prior measurements of bulk silicon crystals, and that bulk Si values of the nonlinear refractive index n(2) can be confidently incorporated in the modeling of pulse propagation in deeply-scaled waveguide structures.

Non-blocking 4x4 electro-optic silicon switch for on-chip photonic networks.

We present a 4x4 spatially non-blocking Mach-Zehnder based silicon optical switch fabricated using processes fully compatible with standard CMOS. We successfully demonstrate operation in all 9 unique switch states and 12 possible I/O routing configurations, with worst-case cross-talk levels lower than -9 dB, and common spectral bandwidth of 7 nm. High-speed 40 Gbps data transmission experiments verify optical data integrity for all input-output channels.

Statistics of light transport in 235-ring silicon coupled-resonator optical waveguides.

In contrast to recent reports of localization-impaired transport in long slow-light waveguides, we demonstrate light transport in silicon coupled-resonator optical waveguides (CROWs) consisting of up to 235 coupled microrings without localization over frequency bands that are several hundred gigahertz wide. Furthermore, from the unique statistical signatures provided by time-domain propagation delay measurements, we demonstrate the spectrally correlated nature of light propagation in CROWs.

CMOS-integrated high-speed MSM germanium waveguide photodetector.

A compact waveguide-integrated Germanium-on-insulator (GOI) photodetector with 10 +/- 2fF capacitance and operating at 40Gbps is demonstrated. Monolithic integration of thin single-crystalline Ge into front-end CMOS stack was achieved by rapid melt growth during source-drain implant activation anneal.