Is Pulsed Field Ablation Really Non-Thermal? How Much Does Tissue Heat Up?
Written by Avi Berger and the Berger Thermal Research team
The short answer
Pulsed field ablation (PFA) is non-thermal in its mechanism of cell death, not in its physics. Cells die because high electric fields permanently disrupt their membranes (irreversible electroporation), not because they are cooked or frozen. But every pulse also deposits Joule heat in blood and tissue, and that heat accumulates from pulse to pulse.
How much the tissue heats up depends on the device and the waveform. Published measurements range from a few degrees at depth to tissue-surface temperatures in the mid-50s °C, and models show the oesophagus exceeding 60 °C at high voltage. Near electrode edges, local heating can reach thermal-damage levels. With some needle and catheter designs, we have observed a thermal lesion depth of up to around 3 mm. A well-designed PFA system keeps the thermal component small; it does not make it zero.
What "non-thermal" actually means
"Non-thermal" describes how the target cells die, not whether temperature rises. In thermal ablation (RF, microwave, cryo), the temperature itself kills. Proteins denature or ice crystals form, and everything in the zone dies by coagulative necrosis, including the extracellular matrix (ECM), vessels and nerves. In irreversible electroporation (IRE), the electric field drives the transmembrane voltage past a critical level. The pores fail to reseal, and the cell loses homeostasis.
The term entered tissue ablation in 2005, when Davalos, Mir and Rubinsky proposed IRE as a stand-alone, non-thermal ablation method. AngioDynamics commercialised it for soft-tissue tumours as NanoKnife, typically using monophasic pulses of about 70–100 µs delivered between needle electrodes. Cardiac PFA relies on the same biophysical phenomenon, but most current systems deliver short biphasic pulse trains from multi-electrode catheters sitting in blood.
Where the heat comes from
The heat is ordinary Joule heating, and it is local. Total delivered energy (V × I × total pulse on-time) tells you how much heat enters the system, but injury depends on where that heat is deposited. The local volumetric heating rate is set by conductivity and the local field:
Ignoring conduction and blood flow during a single short pulse, the adiabatic temperature rise is:
As a worked example, take blood (σ ≈ 0.7 S/m, ρc ≈ 3.8 MJ/m³·K) at 1,500 V/cm for a 100 µs pulse. It heats by about 0.4 °C. That looks harmless, but three things change the picture:
- Field concentration. At electrode edges, the local field can be several times the nominal value. Heating scales with E², so a 3× field means about 9× the heating: several degrees per pulse.
- Accumulation. An application contains tens to hundreds of pulses, and sites are often treated more than once. The heat stacks up if inter-pulse and inter-train delays are shorter than the local thermal relaxation time. For a pulse train with a high duty cycle and tens of pulses, the peak local temperature near the electrode edge can reach hundreds of °C for brief periods.
- Positive feedback. Tissue conductivity rises roughly 2% per °C, and electroporated tissue becomes more conductive again, often 2–3×. Higher σ draws more current, which produces more heat. Constant-voltage generators are especially exposed to this, and a rising current during a train is a useful early warning.
Where it gets hot
The hottest point is almost always the electrode edge, and from there the risks escalate in a predictable chain:
- Electrode edges. Edges and tips concentrate the field and the current density. This is the primary hot spot in every geometry we have modelled.
- Blood next to the electrode. Blood is a better electrical conductor than tissue, so it carries much of the current and heats first. If temperatures stay in the coagulation range during an application, coagulum can form on the electrode or thrombus in the blood pool.
- Gas formation, by two routes. Bubbles form either by electrolysis at the metal–blood interface, which needs no high temperature, or by local boiling once blood approaches roughly 100–105 °C. Mahnič-Kalamiza et al. (2024) found that monophasic protocols produce mainly electrochemical bubbles, while high-frequency biphasic protocols typical for PFA produce mainly boiling bubbles. Boiling bubbles collapse quickly; electrochemical ones persist longer.
- Arcing. A gas layer on the electrode can break down under the high electrical field and ionise into a plasma. Arcing damages the electrode and the tissue, creates a pressure wave, and disrupts current delivery from the generator.
- Tissue at the contact site. Tissue under the electrode heats too, especially in unipolar (monopolar) configurations, where all current passes through tissue to a return patch and generates Joule heat in the tissue (an effect similar to RF ablation). With some needles and catheters in use today, temperatures can reach true thermal ablation. In our own work, we have seen thermal lesion depths of about 3 mm with IRE ablation.
Contact force, electrode angle and irrigation all shift this chain. That is one reason bench data from one device rarely transfer to another.
What the literature has measured
Measured heating ranges from a few degrees to clearly thermal levels, depending on device, waveform, energy and cooling.
| Study | Setting | Key thermal finding |
|---|---|---|
| Mercado Montoya et al., Europace 2025 | 3D left-atrium model; bipolar, 50 × 100 µs pulses | Oesophagus barely heated at 1 kV; reached 46.3 °C at 1.5 kV and above 62 °C at 2 kV after one train |
| Variable-loop circular catheter study, 2025 | Bovine ventricle in vitro; two waveforms, two irrigation rates | Surface up to about 56 °C (ΔT about 19 °C); waveform choice and irrigation reduced it markedly |
| Verma et al., JACC Clin EP 2023 | Perfused swine muscle; focal monopolar biphasic PEF vs RF | PEF maximum average rise of 7.6 / 2.8 / 0.9 °C at surface / 3 mm / 7 mm; RF rose 16.6 / 39.8 / 9.5 °C |
| Faroja et al., Radiology 2013 | Porcine liver, high-dose IRE | Temperature rose in all experiments; thermal damage possible near critical structures |
| Davalos et al., Prostate 2015 | Commentary on oncology IRE | Thermal effects are real but controllable with protocol and device design |
The pattern is consistent: heating is dose-dependent, peaks at the electrode–tissue interface, and falls steeply with depth. The favourable numbers in the 2023 Verma study apply to one specific focal system and should not be generalised to all PFA.
What the clinic is telling us
Large registries confirm that PFA avoids the classic thermal complications, while revealing energy-related side effects that bench work had underweighted. MANIFEST-17K followed 17,642 patients treated with a pentaspline catheter. It reported no oesophageal complications, no pulmonary vein stenosis and no persistent phrenic palsy. Two unexpected complications appeared: coronary spasm in 0.14% of patients, and haemolysis-related renal failure requiring dialysis in 0.03%.
- Haemolysis scales with the number of applications and the energy delivered into the blood pool. It is the clearest reminder that blood, not only myocardium, absorbs PFA energy.
- Coronary spasm is mostly proximity-related, and pulse width matters: shorter pulses reduce smooth-muscle stimulation.
- Microbubbles are monitored because gas released into the circulation could embolise. Biphasic short pulses with a low duty cycle minimise both electrolytic and boiling gas.
How to design PFA to stay non-thermal
Thermal safety has to be designed in, by co-optimising the device and the pulse sequence. These are the main levers:
| Lever | Thermal effect | Trade-off |
|---|---|---|
| Peak voltage / field | Heating scales with E² | Lesion depth falls with lower field |
| Pulse width | Shorter pulses deposit less heat per pulse; biphasic pulses limit electrolysis | Short pulses need a higher field for the same kill |
| Duty cycle and inter-train delay | Longer pauses let heat dissipate between pulses | Longer application time |
| Electrode geometry and edges | Rounded edges and larger surfaces lower peak current density | Mechanical and size constraints |
| Irrigation / blood flow | Convective cooling of the interface | Saline load for the patient |
| Bipolar vs unipolar | Bipolar confines the current; unipolar drives it through tissue | Lesion shape and depth; less muscle contraction with bipolar mode |
| Feedback control | Monitoring current rise or impedance drop can stop a runaway train | Generator complexity |
The electric field, conductivity changes, Joule heating, blood flow and thermal damage are all coupled and time-dependent, so they cannot be optimised one at a time. Our approach is a time-domain 3D multiphysics model of the device and the ablated site in COMSOL Multiphysics that couples:
- The electric field, with conductivity that depends on temperature and electroporation
- The temperatures, with bioheat transfer in tissue and convective cooling by blood flow
- Sensitivity analysis on waveform, electrode geometry, contact and flow
- Optional analysis of boiling or clotting, using phase-change calculations that depend on local temperatures and energy
- Optimisation of device parameters with the COMSOL Optimization Module
For each candidate design, the model output is iso-field and isotherm maps. Benchtop validation is then used to calibrate model parameters and establish the model's credibility, which is required by the FDA for computational models included in regulatory submissions.
Key takeaways
- PFA kills cells non-thermally, but delivers real Joule heat to blood and tissue.
- Heating is local: it peaks at electrode edges and at the electrode–blood interface, and falls steeply with depth.
- Measured temperature rises range from a few degrees to tissue-surface temperatures above 55 °C. Poorly matched device–waveform combinations can produce true thermal lesions and clots.
- Gas forms by electrolysis and by boiling, and gas enables arcing. Biphasic short pulses with a low duty cycle minimise both.
- Haemolysis and coronary spasm show that the energy budget is a safety parameter.
- Coupled, time-dependent multiphysics modelling is the practical way to design devices and waveforms that keep PFA genuinely non-thermal.
Sources
Thermal side effects during pulsed field ablation: an analysis using computer modelling
Mercado Montoya M. et al. · EP Europace 2025
Assessment of temperature dynamics in PFA with a variable-loop circular catheter
2025
Thermal profiles for focal pulsed electric field ablation
Verma A. et al. · JACC Clin Electrophysiol 2023;9(9):1854–63
Irreversible electroporation ablation: is all the damage nonthermal?
Faroja M. et al. · Radiology 2013;266(2):462–70
Implications and considerations of thermal effects when applying IRE
Davalos R.V., Bhonsle S., Neal R.E. · Prostate 2015
Elucidating the mechanisms of microbubble formation in intracardiac pulsed field ablation
Mahnič-Kalamiza S. et al. · Electrochimica Acta 2024
Safety of pulsed field ablation in more than 17,000 patients (MANIFEST-17K)
Ekanem E. et al. · Nature Medicine 2024;30:2020–29
Microbubble formation and hemolysis in PFA: risks, mechanisms, and prevention
PMC
Complexity and perplexity of pulsed field ablation: an engineering perspective
Arrhythmia & Electrophysiology Review
