A Highly Sensitive Hydrogen Sensor with Gas Selectivity Using a PMMA Membrane-Coated Pd Nanoparticle/Single-Layer Graphene Hybrid

Background

Hydrogen leaks need fast and selective detection. Commercial metal oxide H2 sensors usually operate above 400 °C, and decorating them with noble metal catalysts lowers the operating temperature but introduces cross-sensitivity to CO and hydrocarbons such as CH4. Graphene responds to polar gases but is insensitive to nonpolar H2, so graphene H2 sensors also rely on noble metal catalysts, and cross-sensitivity to CO, CH4 and NO2 remained a problem for these sensors.

Design and approach

SLG was grown on Cu foil by chemical vapor deposition, the graphene on the back of the foil was removed by O2 plasma, and the SLG/Cu sample was immersed in PdCl2 solution. Because Cu has a lower standard electrochemical potential than Pd, a galvanic displacement reaction occurs, and the graphene layer transports electrons from the underlying Cu so that Pd nucleates uniformly on the graphene surface rather than in irregular shapes on exposed Cu. The Pd NP/SLG hybrid was transferred onto Au/Ti electrodes on glass with a PMMA supporting layer and coated again with PMMA to form a sealing membrane. The membrane is intended to pass H2, whose kinetic diameter (0.289 nm) is smaller than those of NO2 (0.4 nm), CO (0.33 nm) and CH4 (0.38 nm), through the free volume of the PMMA matrix.

How it was evaluated

Pd coverage on graphene (cpd) was controlled by immersion time between 1 and 120 min and examined by field emission scanning electron microscopy and micro-Raman spectroscopy. Resistance changes were measured at a bias of 10 mV in a tube furnace with H2 or other gases diluted in N2 and with air for recovery. The authors compared sensors with different cpd under 2% H2, averaged ten sensors made at the optimized coverage, tested repeated cycles and H2 concentrations from 0.025 to 2%, compared PMMA thicknesses of 200, 400, 600 and 800 nm, and exposed sensors with and without the 400 nm PMMA membrane to 10% CH4, 0.5% CO, 0.05% NO2 and 0.025% H2, and the PMMA-coated sensor also to a mixture of 0.025% H2 and 0.05% NO2.

Key results

Below 15 min of immersion, Pd NPs about 20 nm in average diameter were separated on the graphene; longer immersion formed islands, and cpd increased from 0.28 to 0.97 between 1 and 120 min. The Raman intensity ratio IG/I2D of the graphene was 0.49 and remained nearly unchanged after a 15 min reaction, which the authors took to show that the graphene was single-layer and was not damaged by the deposition.

The response to 2% H2 rose with coverage to 65% at cpd = 0.66 and fell to 58% at cpd = 0.72, and at cpd = 0.97 the resistance decreased instead, giving -82%. The authors attribute the positive response at low coverage to electron transfer from Pd hydride into graphene, which carries the current, and the negative response at high coverage to hydrogen-induced Pd expansion that closes gaps in a film-like Pd path. The cpd = 0.97 sensor showed the largest absolute response but was unstable over repeated cycles because of expansion, contraction and crack propagation, whereas sensors below cpd 0.72 remained stable.

Ten sensors made at cpd = 0.66 gave average values of 66.37% response, 1.81 min response time and 5.52 min recovery time under 2% H2. A representative sensor gave 66.38% in each cycle of about 5 min on and 20 min off, with a 90% rise time of about 1.07 min and recovery within 11.28 min without significant drift. The response was roughly linear with H2 concentration on a logarithmic scale and was about 4.014% at 250 ppm.

A 200 nm PMMA layer did not seal the device, which then responded to 2% CO (about 10%) as well as 2% H2 (about 100%); membranes of 400 to 800 nm showed no response to 2% CO, while the H2 response decreased from about 65% to 26% with thickness, so 400 nm was chosen. Without PMMA, the Pd NP/SLG sensor responded to all gases tested (1.4% for 10% CH4, -2.08% for 0.5% CO, -5.17% for 0.05% NO2 and 4.11% for 0.025% H2). With the PMMA membrane, it showed no response to 10% CH4, 0.5% CO or 0.05% NO2 and gave about 4% both for 0.025% H2 alone and for the mixture of 0.025% H2 and 0.05% NO2.

Open questions

The authors reported sample-to-sample variation among the ten sensors, which they attribute to the solution-based galvanic reaction, polymer coating and manual transfer. The highest-coverage sensor showed unstable responses during repeated sensing, which the authors attribute to hydrogen-induced expansion and contraction and crack propagation, and a thicker PMMA membrane lowered the H2 response.

How it connects to other research

The PVDF–SnO2–rGO composite sensor listed below tunes the composition of the active sensing layer, whereas this sensor keeps a Pd NP/graphene sensing layer and obtains selectivity from a separate PMMA membrane that filters gases by molecular size.