Research analysis

August 15, 2026 The analog bursting neuron as an array phantom

A new analog circuit realizes a phase-locked-loop neuron model and reproduces spiking and bursting in hardware. For array builders its value is as a physical stimulus generator, with clear limits.

August 15, 2026 The spin-oscillator neuron and array EMI

A spintronic array stores CNN weights in non-volatile MTJs and uses a 199 MHz vortex oscillator as its neuron. The recording-relevant story is radio-frequency self-emission and magnetic write crosstalk.

August 14, 2026 Electrolyte bonding and the analog trim store

Bang and colleagues suppress a compounded CBRAM variability figure by roughly three orders of magnitude through electrolyte composition alone, traced to the bonding network by Raman. The result buys verify-free programming, not one-shot per-channel trims, and it says little about the variance a hold-once analog element actually cares about.

August 14, 2026 Nanoparticle reservoirs meet the tissue temperature wall

Mensing and colleagues show in simulation that static control voltages retune a nanoparticle single-electron reservoir. Two coupled facts, a cryogenic operating point and a nanosecond memory, not the elegant design rules, decide whether such analog preprocessing could ever sit beside living tissue.

August 13, 2026 Crossbar selectors and array addressing limits

A materials study scales a nonlinear crossbar selector to a five-nanometre film with lower leakage and a sharper transition. Read for microelectrode arrays, it explains why dense recording grids keep paying for a transistor at every site instead.

August 13, 2026 Stochastic in-memory MAC and the array data path

A spintronic computational RAM runs neural-network matrix math inside memory, but probabilistic device switching collapses its accuracy until an error-aware co-design rescues it. Read as instrumentation, it is a warning about what happens when you move compute onto the electrode array.

August 12, 2026 CMOS-native PCM and on-array storage

A device paper demonstrates elemental germanium as a CMOS-compatible phase-change memory with fast crystallization, projected high retention, and lower resistance drift than the standard alloy, but modest measured endurance. For array hardware the defensible use is a write-cold on-die calibration store, not the always-programmed compute fabric the memory community imagines.

August 12, 2026 Top-K winners and the acquisition record

A measured 65 nm compute-in-memory macro reaches sub-picojoule efficiency by keeping only the top-K strongest outputs and stopping its converter early. That is fine for a classifier or a closed-loop controller, but as an on-array first pass for data you intend to keep it is a signal-dependent, non-invertible filter, and the authors inject noise to hide what it drops.

August 9, 2026 Coincidence gating and the egress wall

A collider photodetector team gates 320,000 single-photon channels with a one-neuron-per-tile coincidence stage before any learned classifier. The mechanism, and its measured shortfall, both map directly onto high-density MEA readout.

August 9, 2026 Nanopore memristance and the interface state

A wafer-scale silicon nanopore study decomposes hysteretic ionic current into resistive, capacitive, and genuinely memristive parts, and pins the memory on ion adsorption kinetics. The same physics lives at every MEA electrode and passivation surface.

August 8, 2026 Clock-driven spiking networks and the window an array never gets

A single-author tool paper extends hls4ml to compile PyTorch spiking networks into synchronous FPGA firmware, quoting a 33.6 microsecond full-window compute latency on a spiking audio benchmark. Read closely, that latency needs careful accounting, the sparsity dividend is left unclaimed, and the fixed-window state semantics are the real story for anyone wiring inference to a living culture.

August 8, 2026 The slow approach: VO2 oscillator jitter and the level-crossing front end

An eight-author experimental study pins VO2 oscillator linewidth on the stochastic incubation before the insulator-metal transition, and shows square-wave injection beats sinusoidal locking. The transferable result is a ruler for timing jitter in any threshold-crossing detector, including the ones between tissue and silicon.

August 7, 2026 Steered device noise and the at-array decoder

A single-author, pre-registered study shows that the intrinsic noise of analog neuromorphic silicon can be conditioned into a restoring force that protects previously learned weights, with a proof of concept on real BrainScaleS-2 hardware. For array systems it forces a partition of the noise budget: measurement noise remains a tax, while compute-substrate noise becomes a potential resource.

August 7, 2026 Hybrid neuron models can launder the recording chain

A Tubingen group embeds neural ODEs inside conductance-based neuron models to recover unknown ion channel gating from a single voltage trace. The same flexibility that compensates for model error will just as readily absorb instrumentation artifacts, unless the acquisition chain is modeled alongside the biology.

August 6, 2026 The gut sensor the recording chain was not built for

Neuropod cells are electrogenic gut sensory cells that form fast synapses onto the vagus, and a new grant proposes to record their activity and wire it to brainstem control. Read as instrumentation, it asks whether an acquisition chain tuned for cortical spikes can hear a sparse, secretory, barrier-embedded cell at all.

August 6, 2026 The Rett network deficit and the detection floor

This grant proposes to read Rett syndrome as a network-topology deficit in patient-derived cortical organoids grown at an air-liquid interface. Because those graph metrics are computed downstream of spike detection across a maturation window of weeks to months, the acquisition chain and its detection floor sit inside the phenotype it is trying to measure.

August 5, 2026 Closed-loop control of the electrode interface

A CAREER proposal treats the skin-electrode interface as something to be actively regulated rather than tolerated. The transferable idea for arrays is the control loop, not the number.

August 5, 2026 Repeat-expansion organoids and the longitudinal array

A funded plan to phenotype myotonic dystrophy in iPSC cortical organoids with high-density arrays. The interesting problem it hands the hardware is metrological, not electrophysiological.

August 4, 2026 Optical depth and the interior access wall

A two-photon pipeline pushes quantitative whole-mount imaging of packed organoids to roughly 200 microns of usable depth, but only in fixed, cleared, immunostained tissue. The measured depth wall and its fixation cost reframe the 3D interior-access problem that surface-biased microelectrode arrays also cannot solve.

August 3, 2026 Temporal interference and the recording front end

Temporal interference promises focal deep stimulation from two crossing kilohertz currents, and this grant proposes to test the mechanism on brain slices sitting on a microelectrode array. Whether the response at the beat frequency is read electrically or optically, any nonlinearity or aliasing in the instrument manufactures a signal at exactly that frequency, so the gating question is a control, not a sensitivity spec.

August 3, 2026 Two-photon printing and the 3D electrode gap

A shared-facility award adds a maskless two-photon nanoprinter that writes true 3D microstructures with rated sub-100 nanometre resolution. The freedom to print out-of-plane geometry addresses a real limit of planar arrays, but turning a printed scaffold into a stable, low-noise electrode in soft tissue is a chain of unsolved materials steps the tool does not perform.

August 2, 2026 On-chip inference and the transduction gap

GelNeuro deletes the host computer from a sense-to-classify loop at 19.6 mW. The result is a useful negative control for organoid hardware, because its clean event stream is the exact thing an electrode array cannot buy for free.

August 2, 2026 Learned event detection and the averaging budget

A compact detector pulls cut-locked potentials out of continuous EEG by stacking an average across subjects on top of the average across cuts. That borrowed subsidy is exactly what a single organoid preparation cannot reproduce.

August 1, 2026 Discriminant-guided targets and the stimulation array

A per-subject generative model of resting-state fMRI finds the most therapeutically responsive site dissociates from the most abnormal one. The transferable idea for arrays is that the electrode to drive is defined by effect, not by deviation.