This page presents selected research results and public outputs of NEURALSENS, coordinated by the Institute of Electrical Engineering, Slovak Academy of Sciences (IEE-SAS), with Comenius University in Bratislava, Faculty of Mathematics, Physics and Informatics (CUB), and the Institute of Informatics, Slovak Academy of Sciences (II-SAS).
Selected research results
Hydrogen sensing elements and electrode design
KPB1, led by CUB, with IEE-SAS and II-SAS. The partners investigated Pt/TiO₂/Pt and TiN/TiO₂/Pt sensing structures, examining how electrode geometry, porosity and oxide preparation influence electrical behaviour and hydrogen response.
Work included diode-like sensing elements, nanoporous Pt electrodes and comparative measurements under controlled gas conditions. II-SAS lithographic patterning and complementary measurements supported the shared fabrication and characterisation programme.
Temperature sensing and sensory neural-network modelling
KPB2, led by II-SAS, with IEE-SAS and CUB. The project developed design rules and software models for Sensory Neural Networks (SeNNs), using temperature sensing to investigate sensor-dependent network behaviour.
Experimental work at IEE-SAS included 16 × 16 memristor test arrays, electrical mapping and device programming. Temperature measurements examined how device geometry, programmed states and read voltage affect thermal response. These activities connect experimental device characteristics with network design and simulation. Public Python scripts provide temperature-based modelling examples.
Sensor-array fabrication and integration
KPB3, led by IEE-SAS, with CUB and II-SAS. Work towards the laboratory SeNN prototype included sensor-array mask designs combining meander structures with patterned top and bottom electrodes, lithographic fabrication, array test structures and readout development.
Electrical and gas-response measurements informed the selection of sensing elements. Complementary studies explored programmable devices and the combination of sensing and memory functions as building blocks for in-sensor processing.
Published research
M. Patrnčiak et al.
Highly sensitive hydrogen gas sensor based on a capacitor-like Pt/TiO₂/Pt structure with large-scale nanoporous top electrode.
International Journal of Hydrogen Energy 99 (2025), 137–145.
This CUB-led study reports nanoporous Pt electrode preparation and hydrogen-sensing behaviour under different temperature and humidity conditions. It is included in the project deliverable register as D1.2.
Download journal article (PDF). Read the journal article — DOI: 10.1016/j.ijhydene.2024.12.142
Preprints
D2.2 — Paper on smart temperature sensor
Investigation of Temperature Sensitivity in Analog Memristors for Programmable Temperature Sensoric Neural Network
This preprint investigates the temperature sensitivity of analog memristors for use in a programmable temperature-sensing neural network. It constitutes Deliverable D2.2 of the NEURALSENS project.
View preprint (PDF) ; Publication record and DOI
T. Izsák, S. Krylov, T. Ščepka, O. Romanyuk, I. Píš, J. Fedor and B. Hudec.
Etch profile engineering of TiN thin films by dry reactive ion etching in fluorine- and chlorine-based gas chemistries.
SSRN, posted 20 April 2026.
The study compares etching processes for controlling TiN profiles, supporting device fabrication. The linked version is a preprint and has not been peer reviewed.
Download preprint (PDF) · Publication record — DOI: 10.2139/ssrn.6614723
Manuscript under review
D3.2 — Hydrogen-sensing SeNN laboratory prototype
Optimization of Pt electrode porosity for improved hydrogen gas sensing performance in vertical Pt/TiO₂/Pt chemiresistive gas sensors
Submitted to the International Journal of Hydrogen Energy — under review.
This work investigates the effect of top-electrode porosity on hydrogen sensing in vertical Pt/TiO₂/Pt chemiresistive sensors. It constitutes Deliverable D3.2 of the NEURALSENS project.
Open-source software
D2.3 — NEURALSENS SeNN simulation software
Developed by Jaromír Klarák at II-SAS, the Python scripts investigate sensor-dependent network models and compare them with conventional neural-network processing.
The public examples use modelled temperature signals and polynomial sensor characteristics. The code is available under the MIT licence.
View source code on GitHub · Download source code (ZIP)
Selected conference contributions
ICNCE 2024 · Aachen
M. Dehghan et al. — IEE-SAS and CUB
ALD H₂-sensor for in-sensor computing crossbar array.
Conference contribution associated with D1.1.
Download conference poster (PDF)
IEEE NAP 2025 · Bratislava
The following topic summaries link to the full titles, author lists and abstracts in the official conference book:
- M. Dehghan et al. — IEE-SAS and CUB: thin-film hydrogen-sensor development for crossbar arrays. Abstract 1648 — PDF
- M. Horský et al. — IEE-SAS: nonlinear thermal sensing elements for sensor-based networks. Abstract 1653 — PDF
- B. Hudec et al. — IEE-SAS and CUB: building in-sensor computing systems from device-level characteristics. Abstract 1793 — PDF
- P. Nemec et al. — II-SAS and IEE-SAS: Application of a TiO₂ Gas Sensor as a Component in Neural Networks. Abstract 1821 — PDF
MEMRISYS 2025 · Edinburgh
M. Bačo et al. — IEE-SAS
In-sensor computing thermometer using analog memristor crossbar array.
Conference poster MEM25-132 on programmed device states and temperature-dependent response. This conference contribution fulfils Deliverable D2.1 of the NEURALSENS project.
Download conference poster (PDF).
Related MEMRISYS 2025 posters
- N. K. Calvo et al. — Filamentary TaOₓ digital memristors made by Ta oxidation for fuzzy-logic applications. Poster MEM25-129. Download poster (PDF)
- M. Horský et al. — Fabrication and characterization of 16×16 analog memristor 1R crossbar arrays. Poster MEM25-131. Download poster (PDF)
- M. Dehghan et al. — Development of Electronic System Interface for Neuromorphic Arrays in In-Sensor Computing. Poster MEM25-109. Download poster (PDF)
SURFINT–SREN IX 2025 · Smolenice
M. Patrnčiak et al. — CUB
Concepts and prospects of hybrid memristive-chemiresistive gas sensors based on Pt/TiO₂/Pt sandwich structures.
Proceedings contribution, pp. 84–87, with an invited presentation by Tomáš Plecenik. The study explores combining hydrogen sensing and memory functions.
IEEE NAP 2026 · Budva
P. Nemec et al. — II-SAS and IEE-SAS
Geometry-Driven Temporal Signal Encoding in TiO₂ Gas Sensors for Neural Network Integration.
Abstract accepted in June 2026; poster presented in September 2026. The work examines how electrode geometry shapes temporal gas-response features for subsequent neural-network processing. The final project report identifies this contribution as supporting Milestone MS2.2 — design of a hydrogen-sensing SeNN using data from KPB1.
View the conference poster — PDF
IEEE IEDM 2025 · San Francisco
Direct Atomic Layer Processing TiO₂ RRAM Gas Sensor for Real-time All-RRAM Intelligent Gas Perception
Presented at the IEEE International Electron Devices Meeting (IEDM) 2025 in San Francisco, USA. This contribution constitutes Deliverable D3.1 of the NEURALSENS project.
View the paper in the Slovak Academy of Sciences repository ; Download IEDM paper (PDF).
Technical milestones
The final monitoring report records the following technical milestones as completed.
| ID | Milestone | Reported outcome |
|---|---|---|
| MS1.2 | Achievement of objective O1, providing input for KPB3 | Hydrogen sensors with different electrode sizes and shapes exhibited responses to 1000 ppm H₂ spanning more than one order of magnitude. The supporting measurements were documented in Appendix 2 of the first monitoring report. |
| MS2.1 | Validation of the SeNN design rule using temperature sensing | SeNN design rules were established and validated using temperature-dependent sensor characteristics. The supporting analysis was documented in Appendix 4 of the first monitoring report. |
| MS2.2 | Design of a hydrogen-sensing SeNN using data from KPB1 | The final report identifies the electrode-geometry-based contribution presented at IEEE NAP 2026 as supporting this milestone. View conference poster (PDF). |
| MS3.1 | Fabrication designs for a smart hydrogen sensor | Responses of sensors with different top-electrode shapes and porosities were used to simulate SeNN weights and prepare a CAD design for maskless lithography, documented in Appendix MS3-1 of the final report. |
Review milestones MS0.1, MS0.2 and MS0.3 are documented on the Project Review Meetings page.
Project deliverable register
The register follows the project’s deliverable plan. Responsible institutions and dissemination designations are shown below. Available public outputs are linked in the preceding sections; inclusion in the register does not indicate that every listed output is already published.
| ID | Deliverable | Responsible institution · Access |
|---|---|---|
| D0.1 | Project website | II-SAS · Public |
| D1.1 | International conference presentation — ICNCE 2024 contribution | CUB · Public |
| D1.2 | Paper on H₂-sensing element (PDF) | CUB · Public |
| D0.2 | Project interim report | CUB · Non-public project reporting |
| D2.1 | International conference presentation — MEMRISYS 2025, Edinburgh (poster PDF) | II-SAS · Public |
| D2.2 | Paper on smart temperature sensor — preprint (PDF) | IEE-SAS · Public dissemination |
| D3.1 | International conference presentation — IEEE IEDM 2025, San Francisco; Download IEDM paper (PDF). | IEE-SAS · Public dissemination |
| D2.3 | View source code on GitHub · Download source code (ZIP) | II-SAS · Public |
| D0.3 | Project final report | IEE-SAS · Non-public project reporting |
| D3.2 | Paper on H₂-sensing crossbar-array SeNN laboratory prototype — manuscript under review | IEE-SAS · Public dissemination |