In-operando FTIR study of ligand-linked Pt nanoparticle networks employed as catalysts in hydrogen gas micro sensors

Microporous networks of Pt nanoparticles (NP) interlinked by aromatic diamines have recently shown prospects of application as hydrogen combustion catalysts in H2 gas microsensors. In particular with respect to long-term sensor performance, they outperformed plain Pt NP as catalysts. In this paper, electron microscopy and Fourier transform infrared (FTIR) spectroscopy data on the stability of p-phenylene diamine (PDA) and of the PDA-linked Pt NP network structure during catalyst activation and long-term sensor operation at elevated temperature (up to 120–180 °C) will be presented. For the first time, all data were collected directly from microsensor catalysts, and FTIR was performed in operando, i.e., during activation and sensor operation. While the data confirm high long-term catalyst activity far superior to that of plain Pt NP over 5 days of testing, they reveal that PDA fully decomposed during long-term sensor operation and that the network of discrete Pt nanoparticles changed to a sponge-like Pt nanostructure already during catalyst activation. These findings are at variance with previous work which assumed that stability of the PDA-linked Pt NP network is prerequisite for catalyst stability and performance.

one nozzle was used to apply the catalytic layer on sensor chips, which required time-consuming cleanup of the nozzle between the application of the ligand solution and the Pt NP colloid.In the present work, one nozzle was used exclusively for the solution of bifunctional ligands, while the other one was only loaded with Pt NP colloid.
To achieve the required accuracy of positioning both depositions, an extended dispensing stage displayed in Fig. SI 1 was used.The two dispenser nozzles were attached to a moveable holder that enabled movements in one horizontal (x) and the vertical (z) direction.The sensor chip was fixed onto a 3D-movable microscope stage with micrometer screws, which was, in particular, used for the movement in the remaining horizontal (y) direction.A macro camera (CCD camera, by Teli) having a static focal length of about 5 cm was mount close to the stage and helped to arrange the dispenser nozzle tips accurately close to the sensor membrane and to ensure precise droplet transfer and a proper dispensing volume.
To load the dispensers, the filtered ligand solution and NP colloid were each given in a clean reservoir glass bottle and tightly mounted on the nozzle holder and connected to the dispenser nozzles via flexible tubing.By applying moderate air pressure to the reservoirs via a syringe, the nozzles could be loaded with the desired dispensing liquid until the liquid spilled out of the nozzle tip.Each nozzle was Electronic Supplementary Material (ESI) for Nanoscale Advances.This journal is © The Royal Society of Chemistry 2024 operated by one control unit (MD-E-3010) which allowed to adjust parameters for the droplet generation.For all ligand solutions as well as for Pt NP colloids in cyclohexanone, the best results for uniform droplet formation and droplet transfer to the membrane were obtained with the following parameters: Voltage = 100 V, pulse width = 200 μs and frequency = 50 Hz.Higher or lower voltages and pulse widths lead to uncontrolled scattering of the droplets or resulted in no droplet formation, respectively.The volume of each dispensed microdroplet can be estimated as 1.8*10 -10 L, assuming spherical droplets with diameters of 70 μm (cf.inner diameter of glass capillary).
During sensor fabrication, a polymer cylinder (40 μm height x 660 μm diameter) was fabricated via microtechnology on each sensor membrane as container for the dispensed liquids.Right: steps of a dispensing process, in this case with pure cyclohexanone (a-e).
For deposition of catalytic layers, the two loaded dispenser nozzles were moved one after the other above the sensor membrane so that both reactants were applied consecutively.More precisely, 50 microdroplets of a freshly prepared 123 mM ligand solution in cyclohexanone were dispensed in the polymer cylinder first.Directly afterwards, 50 microdroplets of Pt NP colloid (24 g/L, 123 mM) redispersed in cyclohexanone were added which resulted in a total liquid volume of around 1.8*10 -8 L with an equimolar ratio of PDA to Pt atoms.Afterwards, time was allowed for the reaction of ligand and Pt NP and evaporation of the solvent cyclohexanone.This procedure was carried out in 4 repetitive cycles to deposit a sufficient amount of catalytic material which was detectable by the FTIRspectrometer.The top section of Fig. SI 2 depicts three stages of one of these repetitive cycles.Due to the stability of the Pt NP redispersed in cyclohexanone, this colloid could be used for a maximum of 2 hours.After finishing 4 cycles of the dispensing process, the nozzles were thoroughly cleaned by purging several milliliters of pure acetone to completely remove the dispensing liquids.Subsequently, air was flushed through the nozzle in order to dry the capillary and avoid falsification of the concentration for any following dispensing series.
The total deposit after 4 cycles of deposition corresponded to approx.0.9 μg of Pt and 0.5 µg of PDA which were finally crosslinked within the polymer cylinder after drying.As fig.SI 2 shows, the PDAlinked Pt NP layer was fairly homogenously distributed within the cylinder.Note that more than 4 dispensing cycles resulted in deteriorated FTIR spectra due reduced IR transparency of thicker catalytic layers.Also note that properties of the catalytic layer could be varied systematically by varying the standard dispensing procedure described above in several aspects such as type of ligand solution, type of NP colloid, total count of microdroplets, count of droplets per dispensing cycle, concentration, and ratio of Pt to ligand.

II) Estimate of the thickness of a catalytic layer
The thickness of catalytic layers deposited according to the procedure described above (see section I) was estimated as follows: a total volume of 3.6*10 -8 L (200 droplets) of Pt NP colloid with a Pt atom concentration of 0.123 mol/L corresponds to 2.667*10 15 Pt atoms on a sensor membrane.The Pt NP had diameters of ~1.5 nm which indicates that the NP are composed of 2 atomic shells [4] with a total of about 55 Pt atoms per NP [31].Consequently, the number of Pt NP deposited on a sensor membrane can be estimated as 4.849*10 13 .
As determined by TEM in previous work, [4] the Pt NP within a PDA-linked network have a nearestneighbour distance of about 1 nm and are coordinated by -on average -6 nearest neighbours.Since this coordination corresponds to that expected for a simple cubic lattice of Pt NP, the average volume per Pt NP within the catalytic layer is estimated in analogy as (2.5 nm)³=15.625*10 -2 m³ (with 2.5 nm corresponding to NP diameter plus average nearest neighbour distance).Given the number of Pt NP deposited on a sensor membrane determined above, the total volume of the ligand-linked network on a sensor membrane can therefore be estimated as 7.577*10 -13 m³.Assuming the formation of a homogenously thick catalytic layer within the polymer cylinder (660 µm diameter, i.e., 3,421*10 -7 m² base area) on the sensor membrane, its thickness should correspond to 2.2*10 -6 m, i.e., to about 2 µm.preparation with an internal standard would have been necessary for optimal quantification by EDX.

III) In-operando FTIR data of a sensor with plain Pt NP catalyst
[32], [33] In addition to the EDX data presented in the main publication, the STEM/EDX data presented in the following figs.SI 5-15 were used for the elemental quantification summarized in table S1.Note that the carbon film of the TEM grid underneath 500 nm thick SiN membrane samples did not contribute significantly to the C(K) EDX signal.The EDX spectra were essentially the same, whether the carbon film was under the membrane or not (see fig.SI 5 -2 and 6 -1).
Exemplarily, 5 stages (a-e) of a dispensing process of 100 microdroplets of cyclohexanone into the polymer cylinder on a sensor membrane are shown in the right section of fig.SI 1. Picture a shows a clean sensor chip placed on the stage.In picture b, the nozzle tip was moved accurately over the sensor membrane.Picture c is taken after deposition of 100 microdroplets of pure cyclohexanone inside of the polymer cylinder.Due to surface tension, a stable hemisphere of cyclohexanone was formed on the polymer cylinder which allowed to hold 100 droplets in a reproducible manner.More applied volume eventually caused spill-over of the liquid.After the liquid was deposited, the nozzle was moved away (picture d), whereupon the cyclohexanone evaporated within 3 minutes (Figure e).

Fig. SI 1
Fig. SI 1 Left: Micro dispenser stage employed for deposition of catalytic layers on sensor membranes.

Fig. SI 2 :
Fig. SI 2: Thermoelectric sensor before and after deposition of Pt-PDA prepared by four repetitive cycles of a standard dispensing procedure.

Fig. SI 3 :
Fig. SI 3: In-operando FTIR spectra of pure Pt NP on sensor membrane after deposition, after activation and during catalytic H 2 gas sensing.

Fig. SI 5 :
Fig. SI 5: STEM EDX data of 1) carbon film of the TEM grid and 2) clean SiN sensor membrane with underlying carbon film of the TEM grid.

Fig. SI 6 :
Fig. SI 6: STEM EDX data of 1) clean SiN sensor membrane without any underlying carbon film and 2) empty space.

Fig. SI 8 :
Fig. SI 8: STEM EDX data of pure Pt NP after deposition on a sensor membrane.

Fig. SI 9 :
Fig. SI 9: STEM EDX data of Pt-PDA on sensor membrane after five days of sensor operation.EDX spectra acquired 1) directly on the catalyst and 2) on blank SiN membrane.

Fig. SI 10 :
Fig. SI 10: STEM EDX data of pure Pt NP on sensor membrane after five days of sensor operation.EDX spectra acquired 1) directly on the catalyst and 2) on blank SiN membrane.

Fig. SI 12 :
Fig. SI 12: STEM EDX data of Pt-DACH on sensor membrane after 5 days of sensor operation.EDX spectra acquired 1) directly on the catalyst and 2) on SiN membrane.

Fig. SI 13 :
Fig. SI 13: STEM EDX data of Pt-BEN on sensor membrane after 5 days of sensor operation.EDX spectra acquired 1) directly on the catalyst and 2) on SiN membrane.

Fig. SI 14 :
Fig. SI 14: STEM EDX data of Pt-DAN on sensor membrane after 5 days of sensor operation.EDX spectra acquired 1) directly on the catalyst and 2) on SiN membrane.

Fig. SI 15 :
Fig. SI 15: STEM EDX data of Pt-DATER on sensor membrane after 5 days of sensor operation.

note blank SiN sensor membrane with
S1 summarizes the results of EDX quantification (based on K-or L-lines) with respect to all relevant elements (C, N, O, Si, Cl, and Pt).All EDX quantifications, in particular the numbers for light elements such as C, N, and O should be considered as estimates because samples were analyzed without further treatment and without taking into account variations of sample texture.An extensive sample