Bulk-Processed Pd Nanocube-Poly(methyl methacrylate) Nanocomposites as Plasmonic Plastics for Hydrogen Sensing

ı Supporting Information ABSTRACT: Nanoplasmonic hydrogen sensors are predicted to play a key role in safety systems of the emerging hydrogen economy. Pd nanoparticles are the active material of choice for sensor prototype development due to their ability to form a hydride at ambient conditions, which creates the optical contrast. Here, we introduce plasmonic hydrogen sensors made from a thermoplastic nanocomposite material, that is, a bulk material that can be molded with standard plastic processing techniques, such as extrusion and three-dimensional (3D) printing, while at the same time being functionalized at the nanoscale. Speci ﬁ cally, our plasmonic plastic is composed of hydrogen-sensitive and plasmonically active Pd nanocubes mixed with a poly(methyl methacrylate) matrix, and we optimize it by characterization from the atomic to the macroscopic level. We demonstrate melt-processed deactivation-resistant plasmonic hydrogen sensors, which retain full functionality even after 50 weeks. From a wider perspective, we advertise plasmonic plastic nanocomposite materials for application in a multitude of active plasmonic technologies since they provide e ﬃ cient scalable processing and almost endless functional material design opportunities via tailored polymer − Transmission electron microscopy of Pd nanocubes after synthesis and compounding; small-angle X-ray scattering for Pd nanocube − PMMA; Pd nanocube isotherm and response/recovery to/from a 750 mbar hydrogen pressure; raw data for the Arrhenius analysis of hydrogen sorption kinetics; calculation of the apparent activation energies of hydrogen sorption; raw data for the derivation of sorption-order parameter; modeling of hydrogen sorption kinetics of Pd nanocube − PMMA; hydrogen di ﬀ usion and permeability measurements on neat PMMA and Pd nanocube − PMMA plates; and hydrogen gas sensing test setups (PDF)


■ INTRODUCTION
Nanoparticles supporting localized surface plasmon resonance (LSPR) enable a wide range of fascinating technologies, where optical bio-and chemosensors 1−6 are of particular interest, and where hydrogen sensors show great promise as a key technology in the emerging hydrogen economy, 6−8 since safety sensors are critical due to H 2 's wide flammability range in air. In this context, optical plasmonic detection is attractive since it generates no sparks due to the passive nature of the transducer and the remote readout capability of light. In this technology, the optical contrast that signals the presence of H 2 in the environment stems from the absorption of hydrogen species into interstitial sites of metal nanoparticles. Therefore, plasmonic H 2 sensors are intrinsically highly H 2 -selective. 6 Pd is the prototype-active H 2 sensing material due to its ability to dissociate H 2 gas without a sizable energy barrier at ambient conditions and its reversible phase transformation from metal to metal hydride at room temperature 9 that gives rise to a large optical contrast. 10 However, despite these advantages, the large-scale practical implementation of plasmonic sensors in commercial devices, in general, and of hydrogen sensors, in particular, is still lacking. A key reason for this situation is that nanofabrication approaches based on nanolithography and vacuum-based thin-film material deposition methods are mostly used for the generation of nanoplasmonic sensing functions aimed at device integration. This is both costly and limits the technology to two-dimensional (2D) arrays on flat surfaces. Thus, to enable the widespread utilization of plasmonic sensors, in general, and of hydrogen sensors, in particular, new ways to make the active plasmonic materials, as well as flexible and scalable material processing routes for three-dimensional (3D) device integration, are of critical importance. For this purpose, plasmonic nanoparticles made by colloidal synthesis 11 are very attractive since they offer unrivaled possibilities for size, composition, and structural engineering to, in turn, both maximize sensitivity and optimize optical performance in plasmonic hydrogen sensors. 12,13 However, even though the self-assembly of colloidal nano-crystals on surfaces 14−16 represents a reasonably mature approach in nanoscale processing, inherent deficiencies like nonscalability and limited adhesion call for alternative solutions for integration in sensor devices. Furthermore, hydrogen sensor deactivation due to "poisoning" by molecular species, 17 such as trace amounts of CO present in, for example, urban air, 18,19 or issues with cross-sensitivity 20 require urgent attention.
To address these challenges, we explore here the application of an inherently scalable paradigm of melt extrusion and 3D printing of plasmonic plastic nanocomposite materials, which are composed of a polymer matrix mixed with plasmonic metal nanoparticles that provide the sensing function. This approach only requires well-established and cost-effective bulk processing techniques to shape plasmonically active objects for device integration. However, neither plasmonic plastic nanocomposite materials as such nor their processing by, e.g., melt processing, such as 3D printing, has so to date been considered for application in nanoplasmonic sensors in general and hydrogen sensors in particular. Therefore, as the key step beyond the state of the art, we explore here a plasmonic plastic nanocomposite material for melt-processed hydrogen sensors. As building blocks for the nanocomposite, we select colloidal Pd nanocubes as the active hydrogen sensing material due to their well-documented selective interaction with hydrogen via hydride formation and thus the ideal suitability as a model system to demonstrate our new concept, 3,21,22 and poly(methyl methacrylate) (PMMA) as a matrix material due to its demonstrated molecular sieving function. 7,20,23 After detailed structural characterization of the material to optimize its composition in terms of sensor performance, we demonstrate fully functional melt-processed plasmonic H 2 gas sensors that we validate in the form of a 3D-printed sensor cap compatible with fiber optics and a melt-pressed plate integrated into a  ■ RESULTS AND DISCUSSION Nanoparticle Synthesis, Composite Compounding, and 3D Printing. The first step in the process toward a functional plasmonic nanocomposite material for hydrogen sensing is the colloidal synthesis of the desired plasmonic nanocrystals. For this purpose, here, we relied on the aqueoussolution-based Pd nanocube synthesis, using hexadecyltrimethylammonium bromide (CTAB) as the stabilizing surfactant, yielding single-crystalline Pd nanocubes with a 23.2 ± 3.3 nm average side length. In the next step, the nanoparticle dispersion was mixed with the PMMA powder and then dried for 16 h at elevated temperature to evaporate the aqueous solvent ( Figure 1). Hence, the mixing ratio of Pd colloidal nanoparticles and polymer powder determines the colloid concentration in the final composite. Subsequently, the nanoparticle−polymer mixture was compounded in a twinscrew microcompounder and extruded into a Pd nanoparticle− PMMA matrix composite filament of the desired dimension. This filament then served as the raw material for further processing of the composite into macroscopic plasmonic objects by melt pressing or 3D printing.
Pd Nanocube−PMMA Plasmonic Plastic Nanocomposite Characterization. As the active material for hydrogen detection, we have compounded Pd nanocube−PMMA composites with five different Pd nanocube loadings (0, 0.02, 0.03, 0.06, and 0.16 wt %) and melt-pressed the extruded filaments into 500 μm thick plates that exhibit increasingly darker coloring for increasing nanoparticle loading (Figure 2a). The grayish appearance is in good agreement with an LSPR resonance peak in the UV spectral range and the broad-band spectral absorbance of Pd nanoparticles due to interband damping of the LSPR 24 Table SI 1). Finally, elemental analysis reveals that the weight ratio of the Pd nanocubes in the synthesis suspension to the PMMA powder scales linearly with the weight ratio of the Pd to the PMMA in the Pd nanocube−PMMA nanocomposite ( Figure 2f). Moreover, the elemental analysis also implies that no Pd is lost during the processing.
Pd Nanocube−PMMA Nanocomposites for Plasmonic Hydrogen Detection. For the first global characterization of the Pd nanocube−PMMA nanocomposite interaction with H 2 and its corresponding optical response, we melt-pressed a 50 μm thick plate with a 0.03 wt % Pd nanocube loading. Using a vacuum chamber setup with optical windows, 26 we then measured the wavelength-resolved selfreferenced (with respect to the first measured spectrum of the experiment) optical extinction of the composite for increasing H 2 gas pressure ( Figure 3a). We focus on the wavelength range corresponding to the flank of the plasmonic "peak", which has been shown to yield the highest sensitivity. 26,27 The selected spectra reveal the corresponding distinct optical contrast of the composite. Continuous monitoring of the extinction difference at two selected wavelengths, λ min and λ max , thus enables the optical mapping of a pressure−composition isotherm of the   (Figure 3b). The obtained isotherm is in excellent agreement with other studies of Pd nanocube−H 2 interactions and exhibits the characteristic low-pressure αphase, the α + β-phase coexistence plateau, and the highpressure β-phase regions, as well as distinct hysteresis between absorption and desorption in the two-phase region. 3,21,22 This corroborates that the intrinsic material properties of the Pd nanocrystals are preserved in the nanocomposite, guaranteeing the anticipated functionality (for the corresponding analysis of neat Pd nanocubes, see Figure SI 3a,b). At the same time, this Pd-specific characteristic response to H 2 exposure has implications for sensor performance. As apparent from the measured isotherm, the most significant optical contrast change occurs at the α + β-phase coexistence plateau, while it is very small in the α-phase region. This means that the material in its current form, and as characteristic for most H 2 sensors based on pure Pd, is most useful when applied for the detection of a critical H 2 threshold pressure corresponding to the two-phase coexistence plateau. To resolve this limitation and thereby significantly improve sensor detection limit, as well as to eliminate hysteresis, alloying of Pd with, e.g., Au 26 or Cu 17 has proven very effective, and we thus propose this approach also for next-generation plasmonic plastics-based hydrogen sensors.
To quantify the kinetics of hydrogen sorption in the Pd nanocube−PMMA composite, we measured the temporal response to stepwise H 2 exposure and used t 50 , that is, the response time to 50% of the maximal steady-state signal at the new pressure (Figure 3c), as the descriptor. As the corresponding Arrhenius analysis reveals, we find apparent activation energies of 20 and 51 kJ/mol H 2 for hydrogen absorption and desorption in the composite, respectively (Figures 3d and SI 4, and Section SI 5), which is within the range of reported literature values for the Pd−H system. 7,30 Furthermore, analysis of the pressure dependence yields an order parameter, m = 1, for absorption and m = 0.1 for desorption (Figures 3e and SI 5), indicating pressure independence in the latter case.
Having established a fundamental mechanistic understanding of the Pd nanocube−PMMA nanocomposite−H 2 interaction, it is now interesting to investigate the impact of nanocomposite design parameters on H 2 sensing performance. We prepared a series of melt-pressed plates with (i) different Pd nanocube loadings (0.02−0.16 wt %) at constant 250 μm thickness and (ii) different thicknesses (100−800 μm) at a constant 0.02 wt % Pd nanocube concentration. Subsequently, we measured their response to a 750 mbar H 2 pressure step (see Figures 4a,c and SI 6 and 7 for recovery). Plotting t 50 vs nanocube concentration (Figure 4b) and vs plate thickness (Figure 4d) reveals distinct linear and quadratic dependencies, respectively. To understand these, we developed an analytical model based on propagation of a H 2 diffusion front through the composite (Figure 4e and Subsection SI 7.1). Our model suggests that the transient kinetics of hydrogen sorption in the composite depends on the balance between H 2 diffusion in the PMMA matrix and H sorption by the Pd nanocubes. Hence, we predict that the absorption time scales linearly with nanoparticle loading and shows a quadratic dependency on the plate thickness, and fit t 50 = k a conc C NP (eq S16) and t 50 = k a thick L 2 (eq S18) to the experimental data in Figure 4b for varying plate thickness. Here, c 0 is the H concentration in the hydride, V NP is the nanocube volume, c f is the H 2 ACS Applied Nano Materials www.acsanm.org Article concentration in the PMMA matrix, and cf * is the H 2 concentration in the PMMA matrix when the nanocubes are fully hydrated. cf * is thus proportional to the absorption plateau pressure of Pd. The obtained diffusion constants are D conc. = 4.4 × 10 −7 cm 2 /s and D thick = 1.6 × 10 −7 cm 2 /s. These values are in good agreement with the corresponding value D diff = 6.3 × 10 −7 cm 2 /s obtained from direct diffusivity measurements (Figure SI 8 and Table SI 3). This means that the predominant rate-limiting factor for H 2 absorption in the nanocomposite is H 2 diffusion through the PMMA matrix and not H 2 sorption by the Pd nanocubes, in agreement with the faster response of neat Pd nanocubes ( Figure SI 3c). Finally, we also quantified the optical contrast, defined as [extinction@λ max − extinc-tion@λ min ], between the composite in the unhydrogenated and hydrogenated states at a 750 mbar H 2 pressure (Figure 4f). Evidently, the measured optical contrast scales linearly both with increasing plate thickness at a constant Pd nanocube loading and with increasing loading at a constant thickness, in good agreement with the Beer−Lambert law. In summary, these observations imply a trade-off between the optical contrast and temporal response for sensing. Hence, there is an optimal Pd loading and plate thickness that provides the best compromise between those two figures of merit, and we argue that it is preferable to enhance the sensing contrast by increasing the Pd loading rather than the plate thickness because the former increases the temporal response linearly, while the latter shows a quadratic dependency. Also, we predict that further tailoring of the polymer matrix in terms of H 2 permeability is a viable strategy to significantly improve the temporal response of such systems in the future.
Melt-Pressed and 3D-Printed Deactivation-Resistant Plasmonic Plastic Hydrogen Sensors. To demonstrate the ease of integration of our plasmonic plastic into real devices, we tested two different systems as deactivation-resistant H 2 sensors in the CO-rich environment: 31 (i) a melt-pressed Pd nanocube−PMMA sensor plate of 100 μm thickness with a Pd nanocube loading of 0.03 wt % and (ii) a 3D-printed Pd nanocube−PMMA sensor cap of a 200 μm effective thickness and with a 0.02 wt % Pd nanocube loading. The 200 μm thickness of the 3D-printed cap is dictated by the nozzle of the used 3D printer. To compensate for the increased thickness and its detrimental impact on response time, we have therefore reduced the Pd loading slightly, according to the design rule introduced above.
For the first test, we inserted the melt-pressed Pd nanocube−PMMA plate into a gas sensor device (Insplorion AB, Goẗeborg, Sweden; Figure 5a). We then exposed the sensor to subsequent cycles of 10% H 2 , of 0.5% CO, and of 10% H 2 + 0.5% CO, all in the synthetic air carrier gas. Remarkably, thanks to the low diffusivity of CO in PMMA, 31 the system exhibits consistent and reliable response to H 2 even at severe CO poisoning conditions, which would readily deactivate the unprotected Pd control via strong CO chemisorption and the concurrent blocking of H 2 dissociation sites (Figure 5b). 18 For the second test, we opted for a fiber-optic sensor solution onto which we fitted the tailor-made 3D-printed sensor cap (Figure 5c). For sensing performance testing, the cap was mounted on the tip of an SMA-terminated optical fiber mounted inside a homebuilt flow reactor using a fiberoptic feed-through ( Figure SI 10c) and then exposed to subsequent cycles of 10% H 2 , of 0.5% CO, and of 10% H 2 + 0.5% CO, all in the synthetic air carrier gas. Similar to the melt-pressed sensor plate, the 3D-printed fiber cap sensor remained fully responsive even at severe CO poisoning conditions, thereby confirming the anticipated selective molecular sieving function of the PMMA matrix and the selective response of the Pd nanocubes to H 2 (Figure 5d, blue curve). 31 The 3D-printed neat PMMA control also in this case exhibited no response to any of the tested gases (Figure 5d, gray curve).
Finally, to test the long-term stability of our plasmonic plastic nanocomposite, we repeated the same experiment after aging of the 3D-printed sensor cap for 50 weeks at ambient conditions in air. Remarkably, the response to hydrogen, as well as the protection against deactivation by CO, is virtually unchanged (Figure 5d, red curve). This result thus advertises plasmonic plastic materials as an attractive solution for hydrogen sensors where frequent sensor recalibration due to aging and deactivation is one of the main limitations of existing commercial sensor solutions.

■ CONCLUSIONS
Plasmonic plastic nanocomposites provide a flexible material platform for melt-processed active plasmonic sensor devices, where functions like molecular sieving and ease of economic and scalable processing are intrinsically included. Furthermore, ACS Applied Nano Materials www.acsanm.org Article the melt-processed plasmonic plastic nanocomposite-H 2 gas sensors we have presented challenge existing plasmonic H 2 sensor platforms, utilizing 2D arrays of nanofabricated or selfassembled nanoparticles on flat surfaces, since they are a conceptually fundamentally different paradigm for the scalable production of 3D devices. In our plasmonic plastic nanocomposite material, the plasmonically active and hydrideforming Pd nanocubes respond selectively to H 2 through a distinct change in their optical properties. At the same time, the PMMA matrix: (i) acts as a molecular sieve that prevents sensor deactivation by other molecular species, such as CO; (ii) is a means to stabilize the shape of the embedded nanocrystals; (iii) prevents their aggregation; and (iv) ensures sensor long-term stability, thereby drastically reducing the need for recalibration. This breakthrough was corroborated by the successful testing of melt-pressed and 3D-printed sensor plates and caps, respectively, that were directly integrated with a miniaturized optical sensor device prototype and with an offthe-shelf fiber-optic solution. These implementations were then successfully used to demonstrate H 2 detection in simulated application conditions in synthetic air with deactivating CO contamination, and revealed virtually unchanged sensor response and deactivation resistance even after 50 weeks of plasmonic plastic sensor aging at ambient conditions in air.
In a wider perspective, our work enables next-generation optical gas sensors centered on the idea of melt-processed plasmonic plastic nanocomposite materials that combine tailored plasmonic nanocrystal signal transducers with polymer matrix materials, selected and optimized based on the specific analyte to be detected. Furthermore, we envision the application of these materials in other branches of plasmonics, including magnetoplasmonics, plasmonic metamaterials, and plasmon-enhanced photovoltaics, where cost-effective scalable fabrication is critical for large-scale commercial integration in real devices and where the almost endless opportunities provided by combining polymer and colloidal chemistry will enable compelling opportunities in terms of rational material design and function.
Pd nanocubes stabilized with CTAB were synthesized using a procedure reported elsewhere 32 with slight modifications where appropriate. A 10 mM H 2 PdCl 4 solution was prepared by dissolving 0.1773 g PdCl 2 in 10 mL of 0.2 M HCl; thereafter, the solution was diluted up to 100 mL in Milli-Q H 2 O (18.2 MΩ). H 2 PdCl 4 (0.5 mL) was added to a 12.5 mM CTAB solution. The mixture was heated up to 95°C, and 80 μL of 100 mM ascorbic acid was added under rapid stirring. Subsequently, the mixture was kept in the water bath for 30 min. Next, the reaction mixture was centrifuged at 8000 rpm, and the obtained pellet was dispersed in Milli-Q H 2 O to obtain 4 × 10 11 [NPs/mL]. To test the role of nanoparticle concentration in the nanocomposite in H 2 sensing, Pd nanocubes were also synthesized with the following concentrations: 1 × 10 11 , 5 × 10 11 , 1 × 10 12 , 2 × 10 12 , and 5 × 10 12 [NPs/mL] by changing the volumes while maintaining the ratio and concentrations of the reactants.
Composite Compounding. Poly(methyl methacrylate), PMMA, powder with a weight-average molecular weight, M w , of 75 000 g/mol, a polydispersity index, PDI, of 2.8 and a density of 1.2 g/cm 3 was obtained from Polyscience Inc. The Pd−PMMA composite was prepared by pouring the aqueous Pd nanoparticle dispersions onto the PMMA powder, followed by drying at 80°C overnight. The dry mixtures were compounded for 5 min in an Xplore Microcompounder MC5 at 200°C, extruded through a 1.0 mm die, and collected onto a conveyor belt at slow speed, resulting in filaments with a diameter of 1.75 ± 0.1 mm.
Melt Pressing of Plates. Melt-pressed plates were produced with a hot press from AB Nike Hydralik, using extruded filaments as raw material. The Pd nanocube−PMMA samples were pressed at 200°C and 5 tons for 3 min and then under 10 tons for 2 min, with metal spacers to control their final thickness ranging from 50 to 800 μm. For H 2 sensing with the prototype sensor device, 100 μm thick plates were melted onto a fused silica substrate for structural support.
3D Printing via Fused Filament Fabrication. For the fused filament fabrication of the Pd−PMMA composite, we used a Massportal Pharaoh XD printer. The printing was carried out at a nozzle temperature of 210°C, with a build-plate temperature of 60°C and a printing speed of 1000 mm/min.
Elemental Analysis. Elemental analysis for the Pd content in the composite was performed on samples dried at 60°C for 4 h at MIKROLAB, Kolbe Germany.
Transmission Electron Microscopy. All nanoparticle and nanocomposite samples were imaged using a Tecnai T20 TEM microscope with a LaB6 gun, operating at 200 kV or a Titan 80−300 TEM microscope with a field emission gun, operating at 300 kV. The nanoparticle solution sample was prepared on a pure carbon 200 mesh copper grid, and Pd−PMMA nanocomposites were prepared on a Lacey 400 mesh copper grid.
UV−Vis Spectroscopy. The UV−visible absorbance spectra were obtained with a Cary 5000 spectrophotometer within the 350−800 nm spectral range.
Scanning Small-Angle X-ray Scattering (Scanning-SAXS). Scanning-SAXS measurements were carried out at the cSAXS beamline of the Swiss Light Source (SLS) at the Paul Scherrer Institute (PSI) in Villigen, Switzerland. The sample was scanned at an energy of 11.2 keV, and the beam size was 40 μm × 40 μm. The sample-to-detector distance was 2.16 m. A Pilatus 2 M detector was used to acquire small-angle scattering patterns. The step sizes used were 40 μm × 40 μm, while the applied exposure time was 0.2 s. Data processing was carried out using the "cSAXS scanning SAXS package" developed by the CXS group, Paul Scherrer Institute, Switzerland. 33 The size determination was performed using the SASView 4.2.0 software.
Hydrogen Sensing Experiments. The H 2 sensing experiments were carried out in (i) a vacuum chamber described in detail elsewhere 7,26 ( Figure SI 9), (ii) a homebuilt flow reactor (Figure SI 10), and (iii) the hydrogen sensor prototype device. The measurements in the vacuum chamber and in the flow reactor were executed at 30°C (the temperature was maintained by external heating), while those in the homebuilt flow reactor and the prototype sensor device were done at 21°C (ambient temperature of the testing lab). We used multimode optical fibers and a spectrophotometer (Avantes SensLine AvaSpec-2048XL). In the flow reactor, the gas flow was regulated by mass flow controllers (Bronkhorst LowΔP). The total gas flow was 200 mL/min using synthetic air as the carrier gas. The 3D-printed fiber caps were mounted in a flow reactor equipped with a CF flange fiber-optic feed-through and a silica vacuum viewport connected to a light source (Avantes AvaLight-Hal Mini) and the spectrophotometer, respectively. The sensing was performed in transmission mode. The prototype sensor device is equipped with both an internal light source and a detector and operates in transmission mode. It was connected to the same set of mass flow controllers for the gas flow and composition regulation as the flow reactor. Transmission electron microscopy of Pd nanocubes after synthesis and compounding; small-angle X-ray scattering for Pd nanocube−PMMA; Pd nanocube isotherm and response/recovery to/from a 750 mbar hydrogen pressure; raw data for the Arrhenius analysis of hydrogen sorption kinetics; calculation of the apparent activation energies of hydrogen sorption; raw data for the derivation of sorption-order parameter; modeling of hydrogen sorption kinetics of Pd nanocube−PMMA; hydrogen diffusion and permeability measurements on neat PMMA and Pd nanocube−PMMA plates; and hydrogen gas sensing test setups (PDF)