Ultra-Short Electrical Pulses in TEM

Many of the most interesting processes in modern materials do not happen gradually. They begin only when a critical voltage, electric field or current density is reached.

A memristor switches. A phase-change material transforms. A magnetic domain wall starts to move. A dielectric breaks down. A new electronic phase emerges.

The challenge is that the thin, electron-transparent samples required for transmission electron microscopy are often particularly vulnerable to exactly the electrical conditions needed to trigger these effects. Apply the required current continuously and Joule heating can dominate the experiment. Maintain a high electric field for too long and the device may drift, restructure or fail completely.

Sometimes, the solution is not to reduce the excitation — but to make it shorter.

Electrical biasing inside a TEM is already well established. Modern commercial systems can apply voltage and current while imaging structural transformations, device operation and failure processes in real time. Operando TEM has consequently become an important tool for studying electronic devices, memristors, batteries and semiconductor structures.

But applying a voltage is not the same as delivering a fast electrical pulse.
Once pulse durations approach the nanosecond and picosecond regimes, the wiring between the pulse generator and the specimen becomes part of the experiment. Cables, connectors, contacts and the sample chip no longer behave simply as electrical wires. Impedance mismatches produce reflections. Parasitic capacitance and inductance distort pulse shapes. A pulse that looks sharp at the generator can arrive at the specimen broadened, ringing or with a substantially different amplitude.
This is where ISOTEM’s RF architecture becomes important.

Rather than treating electrical connections as conventional biasing wires, our sample-holder technology is designed around high-bandwidth coaxial signal transmission directly towards the TEM specimen. The same technological foundation has already been used for microwave excitation and GHz dynamics in the TEM.
Our current architecture extends this principle to a signal path designed for frequencies from DC to 67 GHz, with future concepts targeting approximately 150 GHz.

The result is not simply faster biasing. It is a route towards picosecond-scale electrical stimuli inside the TEM.

Short electrical pulses provide a particularly interesting possibility for electron-transparent devices.

Many electrically driven phenomena depend on crossing a threshold. Below it, little happens. Above it, the desired process can occur rapidly — while continued excitation mainly adds heat or ultimately destroys the specimen.

A sufficiently short pulse can therefore deliver a large instantaneous electric field or current density while keeping the total deposited energy comparatively small.
This distinction can be decisive.

Instead of asking whether a TEM lamella can withstand a particular current continuously, the more useful question becomes:

Can it withstand that current for 100 ns? 1 ns? 10 ps?

Our answer opens an experimental window that does not exist under steady-state conditions.

Recent experiments already demonstrate the principle. Electrically driven ultrafast TEM has followed structural changes in 1T-TaS₂ during voltage pulses as short as 20 ns, correlating the applied excitation with the evolution of a charge-density-wave state [1]. In magnetic materials, nanosecond current pulses have been used during in-situ Lorentz TEM to create and manipulate nanoscale skyrmionic structures [2,3], while also revealing where excessive current begins to introduce unwanted Joule heating.

Moving further into the high-bandwidth regime creates the opportunity to study increasingly rapid, strongly non-equilibrium processes.

The range of possible experiments is broad.

Memristors and neuromorphic devices can be investigated while resistance states are written electrically, potentially revealing how conductive filaments form, migrate and disappear. Operando TEM is already exposing these mechanisms in emerging memristive systems. Faster excitation could bring the experimental conditions closer to the short switching pulses used in practical devices.

Phase-change materials can be driven across structural or electronic transitions and examined during relaxation, rather than only before and after switching.

Semiconductor devices and interconnects can be exposed to controlled electrical stress to investigate dielectric breakdown, electromigration and other reliability-limiting processes at the scale where they originate. In-situ TEM is already being applied to nanoscale electrical degradation and device failure; precise pulsed excitation adds control over both the magnitude and duration of that stress.

Spintronic structures can be driven by short current pulses to investigate spin-transfer torque, current-induced domain-wall motion, skyrmion dynamics and other magnetic switching processes directly in Lorentz TEM or electron holography.

And in ultrafast TEM, electrical pulses can act as the pump in pump–probe experiments — complementing optical excitation with a stimulus that more closely reproduces the operating conditions of an electronic device.

Bandwidth provides an indication of how rapidly an electrical signal can change, but it does not mean that a holder with a given bandwidth automatically produces perfectly rectangular pulses on the corresponding timescale.

For example, for 67 GHz, rise time can be roughly 5 ps. At 150 GHz, the idealized value approaches 2–3 ps.

At these timescales, the TEM holder, sample carrier and specimen form one electrical system.

That is why ISOTEM’s approach is not limited to providing an RF connector on a sample holder. Contact geometry, impedance, chip design, thermal behaviour and the required excitation must be considered together.

A high-current experiment on a magnetic nanostructure needs a different electrode layout from a high-field experiment on a dielectric. A memristor requires different contacts from a phase-change cell. A pump–probe experiment may place particularly strict requirements on timing, reflections and synchronization.

Our modular architecture is designed so that the electrical pathway can be adapted to the experiment, including custom sample carriers and electrode geometries. This philosophy follows the broader move towards flexible, user-adaptable in-situ and operando sample environments that is already shaping the next generation of TEM instrumentation.

TEM has traditionally been exceptionally good at telling us what a material looks like before and after something happens. Ultra-fast electrical excitation offers something different: the possibility of accessing the brief states while it is happening.

For threshold-driven materials, that’s exactly where the interesting physics lies.

ISO-TEM provides the high-bandwidth electrical pathway needed to bring these experiments into the TEM — combining ultra-fast electrical excitation with imaging, diffraction and spectroscopy at the nanoscale.

Have a material that switches, moves, transforms or fails when you apply a pulse?

Let’s build the experiment around it.