Abstract
The transition toward a hydrogen-based economy demands reliable, low-cost sensors capable of selective detection in complex gas mixtures. Semiconducting metal oxide (SMOX) based chemiresistive gas sensors offer excellent sensitivity and scalability, yet their inherent cross-sensitivity to multiple gas species limits practical deployment in safety-critical applications such as hydrogen leak detection. This thesis investigates the use of cellulose nanocrystal (CNC) membranes as passive selective filters to enhance SMOX-based sensor discrimination toward hydrogen. Two central challenges defined the scope of this work. The first was reconciling the high operating temperatures of SMOX-based sensors (300–450 °C) with the limited thermal stability of CNC films, which degrade above approximately 200 °C. A membrane-on package strategy was developed to address this, spatially decoupling the membrane from the heated sensing element within a protective cap. Computational fluid dynamic simulations confirmed adequate thermal isolation across the operating power range, establishing feasibility before hardware fabrication. Six successive cap designs were developed, each iteration addressing limitations identified in the previous version. The second challenge was the absence of infrastructure for quantitative membrane permeability analysis at the host institutions. Three successive generations of custom-made permeation setups were designed and validated, progressing from initial pressure driven configurations to an automated, species-resolved system enabling simultaneous measurement of H2, O2, and CO2 under controlled humidity (0–60% RH). This methodological development enabled characterization of membrane transport under conditions directly relevant to sensor operation. CNC membranes were synthesized and fully characterized in terms of morphological, chemical, mechanical, and thermal properties. Integration into the sensor package yielded substantial selectivity enhancement: suppression factors of approximately 1700 fold for acetone and 190-fold for ethanol at 50 ppm, while hydrogen response remained largely unaffected. Correlation analysis revealed that this discrimination arises from the interplay of molecular size and polarity, extending beyond conventional size-based sieving. Agreement across three independent measurement approaches validated both the membrane transport properties and the analytical framework. Humidity-dependent permeation studies revealed an unexpected trend: gas permeability systematically decreased with increasing relative humidity for all tested species, contrary to the plasticization behavior typical of dense polymer membranes. Both permeability and diffusivity followed Langmuir-type relationships with humidity, supporting a competitive adsorption model in which water molecules preferentially occupy interstitial transport pathways. Water permeability was approximately two orders of magnitude higher than that of other measured species, further corroborating this mechanism. In conclusion, this work demonstrates a practical route to improving SMOX selectivity using low-cost, bio-derived materials. The correlation between barrier effect and molecular properties provides a predictive framework for other analytes, while the validated cross-platform methodology suggests that membrane-integrated sensors could serve as accessible proxies for permeability screening, bridging gas sensing and membrane science in a single experimental platform. The findings establish both the practical viability of membrane-integrated sensors and a validated methodology for characterizing gas trans port in colloidal nanocrystalline films.