You are on page 1of 53

Articles

SUPPLEMENTARY INFORMATION
https://doi.org/10.1038/s41563-019-0325-4

In the format provided by the authors and unedited.

Metal–polymer hybrid nanomaterials for


plasmonic ultrafast hydrogen detection
Ferry A. A. Nugroho   1*, Iwan Darmadi   1, Lucy Cusinato1, Arturo Susarrey-Arce1, Herman Schreuders2,
Lars J. Bannenberg2, Alice Bastos da Silva Fanta3, Shima Kadkhodazadeh   3, Jakob B. Wagner3,
Tomasz J. Antosiewicz   1,4, Anders Hellman1, Vladimir P. Zhdanov1,5, Bernard Dam   2 and
Christoph Langhammer   1*

1
Department of Physics, Chalmers University of Technology, Göteborg, Sweden. 2Department of Chemical Engineering, Delft University of Technology,
Delft, the Netherlands. 3Center for Electron Nanoscopy, Technical University of Denmark, Kongens Lyngby, Denmark. 4Faculty of Physics,
University of Warsaw, Warsaw, Poland. 5Boreskov Institute of Catalysis, Russian Academy of Sciences, Novosibirsk, Russia. *e-mail: ferryn@chalmers.se;
clangham@chalmers.se

Nature Materials | www.nature.com/naturematerials


Supplementary Information for
Metal - Polymer Hybrid Nanomaterials for Plasmonic Ultrafast Hydrogen
Detection

Ferry A. A. Nugroho*,†, Iwan Darmadi†, Lucy Cusinato†, Arturo Susarrey-Arce†, Herman


Schreuders‡, Lars J. Bannenberg‡, Alice Bastos da Silva Fanta§, Shima Kadkhodazadeh§,
Jakob B. Wagner§, Tomasz J. Antosiewicz†,⊥, Anders Hellman†, Vladimir P. Zhdanov†,║,
Bernard Dam‡ and Christoph Langhammer*,†


Department of Physics, Chalmers University of Technology, 412 96 Göteborg, Sweden.

Department of Chemical Engineering, Delft University of Technology, 2629 HZ Delft,
The Netherlands.
§
Center for Electron Nanoscopy, Technical University of Denmark, 2800 Kongens
Lyngby, Denmark.
⊥Faculty of Physics, University of Warsaw, Pasteura 5, 02-093, Warsaw, Poland.

Boreskov Institute of Catalysis, Russian Academy of Sciences, Novosibirsk 630090,
Russia.

*Correspondence to: ferryn@chalmers.se, clangham@chalmers.se

1
Table of Contents
1. Hydrogen Sensor Stakeholders, Performance Targets and State-of-the-Art ............. 3
2. Additional Structural and Material Characterization ................................................. 8
3. DFT Calculations ..................................................................................................... 14
4. FDTD Simulation of the Enhanced Sensitivity ....................................................... 18
5. Determination of Experimental λpeak ........................................................................ 20
6. Experimental Noise Evaluation ............................................................................... 21
7. Strain Induced by PTFE Coating ............................................................................. 24
8. Sensor@PTFE Sensitivity ....................................................................................... 26
9. Sensor@PTFE Kinetics ........................................................................................... 27
10. Theoretical Analysis of the Hydrogenation Kinetics of PdAu Nanoparticles ......... 30
11. Sensors@PMMA ..................................................................................................... 37
12. Variation in Sensor Sensitivity due to Difference in the Optical Properties ........... 41
13. Tandem Sensors ....................................................................................................... 43
14. Deactivation Tests .................................................................................................... 45
References ......................................................................................................................... 47

2
1. Hydrogen Sensor Stakeholders, Performance Targets and State-of-the-Art
Hydrogen sensors find application in various areas, including safety, industrial process
monitoring (e.g. the production of ammonia or methanol), as well as healthcare.1–3 The
drive towards a hydrogen economy, however, further accelerates both the need and the
development of hydrogen detection technologies, which are to find application wherever
hydrogen is produced, stored, transported or used. Already today, the direct contact of the
public with hydrogen is imminent or reality, e.g. in commercial hydrogen powered
vehicles.4 Hence numerous institutions, governmental agencies, and end users such as car
manufacturers and other industries have gathered and formulated guidelines listing the
desired performance of hydrogen sensors.
Table S1 provides an overview of the hydrogen sensor performance targets defined for
different applications by different stakeholders. It is clear that the required performance
varies depending on the level of associated risk. The toughest requirements are
formulated for hydrogen sensors for safety systems and were defined by U.S. Department
of Energy (DoE) in 2015.5 According to the DoE targets, a hydrogen sensor must respond
in less than 1 s to a hydrogen concentration in the range of 0.1 to 10%. A sensor accuracy
better than 5% is also demanded.

Table S1. Technical performance requirements for hydrogen sensors in different areas of
application.
Parameter Performance Requirement
Stationary Systems6 Automotive Safety Systems5
Systems7
Measurement Range up to 4% up to 4% 0.1% - 10%
Detection Limit < 100 ppm - -
Operating -20 oC – 50 oC -40 oC – 125 oC -30 oC – 80 oC
Temperature
Response Time < 30 s <3s <1s
Recovery Time < 60 s <3s -
Accuracy 25% - 5%

In attempts to meet these demands, a large number of hydrogen sensor architectures have
been developed in the past two decades. Table S2 presents a summary of the
performance metrics of the most relevant and competitive hydrogen sensors reported in
the literature to the best of our knowledge. To this end we note that in some cases it is
difficult to extract reliable sensor metrics since they are either not explicitly given and
rigorously derived, or because of generally poor data quality. Nevertheless, from our
survey it becomes clear that the biggest challenge lies in meeting the 1 s response time
target at 0.1% H2 at ambient temperature and that two other important traits of a
hydrogen sensor, i.e. hysteresis-free readout (related to accuracy in dynamic hydrogen
concentration) and resistance to poisoning gases (e.g. CH4, CO, NO2), are only very
rarely addressed and thus a widely unsolved problem (Table S3).

3
Table S2. Sensing metrics of state-of-the-art hydrogen sensors from the literature.
Sensor Platforma Transducing t90 t10 Pressureb Temperatureb LoD Hysteretic Resistance to Ref
Method (s) (s) (mbar) (oC) (ppm) Behaviourc Poisoning Gasesc
PdAu NP optical 1 5 1 RT 3 hysteresis- CO2, CH4, CO, this
@PTFE@PMMA free NO2 work
8
Pd ultrasmall grain electrical 12 19 30 RT 2.5 n.a. n.a.
9
Pd strip on 3D structures optical 20 - 1 RT 10 n.a. n.a.
10
CeO2-loaded In2O3 electrical 24.5 21 0.05 160 0.01 n.a. n.a.
hollow spheres
11
Pd NW/ZnO electrical 50 - 0.1 100 100 n.a. n.a.
12
Pd NP/C NW electrical 45 5 1 RT 10 n.a. n.a.
13
Pd-SnO2/MoS2 electrical 26 15 1 RT 30 n.a. n.a.
α-MoO3 NW electrical 3 2.7 15 260 100 n.a. n.a. 14

15
Pd/SiC NCf electrical 4 48 0.5 300 2 n.a. n.a.
16
SiO2 NR@Pd electrical 17 - 10 RT 10 n.a. n.a.
17
Pd NP/MoS2 electrical 780 900 10 RT 50 n.a. n.a.
18
MoS2 NP/rGO electrical 251 260 0.5 60 200 n.a. n.a.
19
PdAg NW electrical 120 102 1 RT 100 n.a. n.a.
20
PdCuSi mechanical 5 - 30 - - hysteresis- n.a.
free
21
Pt /YSZ/LaSrCrFeO3-δ electrical 4 24 1 450 20 n.a. n.a.
22
Pd NC/TiO2 NF electrical 25 1 6 150 6000 n.a. n.a.
23
MoS2 electrical 14 140 10 30 - n.a. n.a.
24
Pd-capped Mg film electrical 6 33 10 RT 1 n.a. n.a.

4
Active Elementa Transducing t90 t10 Pressure Temperature LoD Hysteretic Resistance to Ref
Method (s) (s) (mbar) (oC) (ppm) Behaviourb Poisoning Gasesb
25
UV-activated ZnO electrical 4 24 10 RT 5 n.a. n.a.
26
Pd NW@ZIF-8 electrical 13 6 1 RT 1000 n.a. n.a.
27
Pd-Pt/ZnO NR electrical 5 76 10 100 0.2 n.a. n.a.
28
Pd NP/TiO2 electrical 12 5 1 180 1 n.a. n.a.
29
Pd NP/MnO2 electrical 30 40 1 100 10 n.a. n.a.
30
Pt-TiO2 electrical 10 20 1 RT 30 n.a. n.a.
31
Pd NP/ZnO NR electrical 100 - 5 RT 10 n.a. n.a.
32
TiO2 nanostructures electrical 50 40 1 RT 1 n.a. n.a.
33
PdPt film electrical 4 5 10 150 10 n.a. n.a.
34
Pt NW electrical 100 - 1 25 1 n.a. n.a.
35
Pd NRb electrical 4 9 100 RT 20000 n.a. n.a.
36
PdY film optical 6 8 40 RT 1000 n.a. n.a.
37
PdMg film electrical 1 60 4000 100 - n.a. n.a.
38
Au@Pd NP electrical 15 18 200 RT 1000 n.a. n.a.
39
Pt@Pd NW electrical 2 2.5 40 103 4000 n.a. n.a.
40
polyurethane@Pd electrical 24 - 1 RT 20 n.a. n.a.
41
Pt@Pd/graphene electrical 180 72 10 RT 1 n.a. n.a.
42
Pd NP/graphene electrical 108 330 20 RT 250 n.a. CH4, CO, NO2
@PMMA
43
Pd NP/TiO2 NT electrical 120 90 10 RT - n.a. n.a.
44
Pd NW electrical 25 - 1 RT 50 n.a. n.a.
45
Pd/SiO2/Au optical 3 10 40 RT 5000 n.a. n.a.

5
Active Elementa Transducing t90 t10 Pressure Temperature LoD Hysteretic Resistance to Ref
Method (s) (s) (mbar) (oC) (ppm) Behaviourb Poisoning Gasesb
46
PdAu film optical 15 15 20 RT 5000 hysteresis- CH4
free
47
Pd NP/graphene electrical 300 - 1 RT 20 n.a. n.a.
48
PdNi NG electrical 0.5 0.5 20 RT 500 n.a. n.a.
49
Pd NG electrical 5 200 40 RT 5000 n.a. n.a.
50
Pd NW electrical 4 - 24 RT 1000 n.a. n.a.
51
Pd/Au film optical 4.5 13 40 RT - n.a. n.a.
52
Pd NG electrical 498 3000 30 RT 20000 n.a. n.a.
53
Pd NG electrical 0.07 - 20 RT 25 n.a. n.a.
54
Pd film optical 10 - 40 RT - n.a. n.a.
55
Pd microcantilever mechanical 90 300 10 RT - n.a. n.a.
56,57
fractured Pd NW electrical 0.07 - 40 RT 10000 n.a. CH4
a
NC = nanocube, NCf = nanocauliflower, NF = nanofiber, NG = nanogap, NP = nanoparticle, NR = nanorod, NRb = nanoribbon, NT
= nanotube, NW = nanowire. bPressure/temperature for response and recovery times measurements. cn.a. = not addressed.

6
Table S3. Technical performance requirements for hydrogen sensors for safety systems
and sensing metrics of poisoning gases-resistant hydrogen sensors from the literature.
Sensor Platform

Parameter Safety Pd Pd NPa @ Pd film @ PdAu NPa @


Systems5 a
NP /graphene PMMA58 HKUST-1 PTFE @
@ PMMA42 59
PMMA (this
work)
Measurement 0.1–10% 0.025–2% 10% Up to 10 ppm –
Range 100% 100%
Detection 1000 ppm 250 ppm 50 ppm n.a. 10 ppm
Limitb
Operating -30–80 oC RTc RTc 100 oCc 30 oCc
Temperature
Response <1 s 108 sd 11 se 170 sd <1 sf
Time
Uncertaintyb <5% n.a. n.a. Hysteresis Hysteresis-
free
Resistance to - CH4, CO,NO2 CH4, CO CO2, CH4 CO2, CH4,
Poisining CO, NO2
Gases
a
NP = nanoparticles. bn.a. = not addressed. cTemperature at response time
measurements. RT = room temperature. dResponse to 20 mbar H2. eResponse to 40
mbar H2. fResponse to 1–100 mbar H2.

7
2. Additional Structural and Material Characterization

Figure S1. EDX elemental linescan across a Pd70Au30 alloy nanoparticle. The counts are
normalized to the average elemental counts at the middle of the nanoparticle, i.e. distance
between -50–50 nm.

Figure S2. SEM image and particle diameter histogram of a Pd70Au30 alloy nanoparticle
array. The observed polydispersity is “inherited” from the polystyrene particles used
during fabrication. The scale bar is 1 µm.

8
Figure S3. (a) TKD grain orientation map of a Pd70Au30 alloy nanoparticle array. The
color code depicts the grain crystallographic orientation with respect to the out-of-plane
axis. (b-f) Zoomed-in TKD maps of individual Pd70Au30 nanoparticles. Panel (e) and (f)
also show the corresponding grain boundaries which are categorized as high-angle grain
boundaries (HAGB) and twin boundaries. Statistics of the nanoparticles microstructure is
presented in Table S4. Scale bars are 100 nm.

9
Table S4. Pd70Au30 nanoparticles microstructure obtained by TKD analysis.
Parameters Value
Number of particles included in the analysis 179
Number of single crystal particles 8
Average particle diameter (nm) 143.4
Average grain size diameter [NFa] (nm) 58.3
b
Average grain size diameter [AF ] (nm) 99.7
Length of twin boundaries (µm) 39.1
c
Total length of HAGB boundaries (µm) 57.9
Length of HAGB excluding twins (µm) 18.8
Length of LAGBd (µm) 3.1
2
Total area of particles (µm ) 3.1
Fraction of twin boundaries (µm-1) 12.8
Fraction of HAGB (µm-1) 18.9
Fraction of HAGB excluding twins (µm-1) 6.1
-1
Fraction of LAGB (µm ) 1.0
Length of all LAGB - misorientation > 5o (µm) 14.0
Fraction of LAGB - all boundaries (µm-1) 4.6
a
Number fraction. bArea fraction. cHigh-angle grain boundary. dLow-angle grain
boundary.

Figure S4. (a) In situ X-ray diffractogram of a Pd70Au30 alloy nanoparticle array
measured with Co-Kα, λ = 0.1789 nm. The obtained single peak lies between the
expected diffraction angles for pure Au and Pd,60 marked by the dashed lines,
corroborating the fully homogeneously alloyed nature of the nanoparticles. (b) d-spacing
of the Pd70Au30 alloy nanoparticles measured with in situ XRD as a function of hydrogen
pressure. A hysteresis-free isotherm is found, in good agreement with the optical and
QCM results shown in the main text. The d-spacing is increased by 1.4% with respect to
the as-prepared state at the maximum hydrogen pressure of 400 mbar.

10
Figure S5. Ex situ XRD diffractograms spectra of (a) pure Pd and (b) Pd70Au30 alloy
nanoparticle array measured with Cu-Kα, λ = 0.1541 nm. The diffraction peaks indicated
in the panels for reference are obtained from powder Pd and PdAu reference.60

Figure S6. Rocking curves around the <111> diffraction peak of pure Pd and Pd70Au30
alloy nanoparticle arrays measured at 28 oC. The curves denote the orientation of the
crystallites with respect to <111> direction.

11
Figure S7. AFM topographical image of a Pd70Au30@PTFE nanoparticle array.
A smooth and conformal PTFE layer with no sign of cracks is observed.

Figure S8. XPS spectra of Pd 3d, C and Au 4f peaks of Pd and Pd70Au30 sensors before
and after PTFE coating. Upon addition of PTFE, the Pd and Au peaks from both sensors
are suppressed, indicating homogeneous and complete coverage of the PTFE film.

12
Figure S9. High-resolution XPS spectra of Pd 3d peaks of Pd with and without PTFE
coating with different thicknesses. The spectrum shows a doublet peak with binding
energies (BE) around 335 eV and 340.25 eV, consistent with the Pd 3d5/2 and Pd 3d3/2
peaks, respectively.61 The Pd 3d BE shifts slightly to higher energies by 0.55 eV as the
PTFE thickness increases. The results are in good agreement with the expected values for
a Pd-Cx phase formation,62–64 evidencing the formation of Pd-C bonds upon PTFE
deposition.

Figure S10. As-deposited (solid line) and 1 bar H2-loaded (dashed line) FTIR spectra of
200 nm sputtered PTFE on a 10 nm Pd film on a KBr crystal. The spectra show the
conformational configuration of –CF2 with its various vibrational modes (i.e. stretching,
rocking, bending and wagging), as well as a C-C inflection point, in good agreement with
the bulk system reported in the literature.65 Additionally, vibrational signatures for –CF3
and unsaturated fluorinated carbons CF=CF (marked in red) also exist, in good agreement
with FTIR data obtained from sputtered thin PTFE films reported in the literature.66

13
3. DFT Calculations
1. Hydrogen-polymer interaction
To evaluate the interaction between H2 molecules and the polymer bulk in the coating,
chains with between 12 and 15 carbon atoms have been modeled for several polymers,
that is, PTFE 136, PTFE 157, PTFE 21, and atactic PMMA. The indices for the different
PTFE systems denote the symmetry of the helical carbon backbone. An “mn” helix will
consist of m building blocks, i.e. CF2 monomers for PTFE, evenly spaced along a helix
with n turns to form the periodically repeating pattern. This can be seen in Fig. S11 for
PTFE 136 and PTFE 157. PTFE 21 is a special kind of helix, with only one turn per
pattern. This leads to a planar carbon backbone, corresponding to the all-trans
conformation of PTFE.

Figure S11. Optimized geometries and adsorption energies for H2 interacting with PTFE
136, PTFE 157, PFTE 21, and atactic PMMA calculated by DFT.

The optimized geometry of the polymers interacting with H2 can be seen in Fig. S11,
together with the calculated interaction energy. With values between -0.07 eV and -0.04
eV, it appears that the interaction between H2 and the bulk polymer is very weak and
essentially independent of the kind of polymer considered. This goes hand-in-hand with
the conclusion that the relevant contribution to the faster response of a polymer coated Pd
and PdAu alloy surface in hydrogen sensing stems from the polymer-metal interface and
not the (bulk) polymer itself.

2. Pd-polymer interaction
In theoretical studies on Pd/Pd hydride systems slab models are often used. 67,68 However,
in recent studies cluster models have also been used.69,70 Here, the Pd nanoparticle was
modeled using a cluster model instead of a slab model for two main reasons: (i) to avoid
the periodicity introduced by slab models, which would lead to large supercells when
trying to match the metal and polymer lattices, thus making them computationally
expensive, and (ii) to avoid bias caused by lattice strain when one or several hydrogen

14
atoms are absorbed (or adsorbed) in (or on) palladium. It has previously been shown that
lattice expansion during hydrogen absorption strongly impacts ad- and absorption
energies and diffusion barriers.71 Therefore, this issue has to be taken into account when
using a slab model. In contrast, in a cluster model, the metal lattice is free to respond to
hydrogen and is not blocked by periodic conditions imposed by a slab model.
Furthermore, edges and kinks of a cluster are good models for special or defective sites
existing on real surfaces, and they have been shown to be of importance in catalytic
reactions.72,73 With this in mind, a cuboctahedral cluster of 55 palladium atoms was
chosen as the nanoparticle model (see Fig. S12). This cluster, of fcc stacking, is large
enough to exhibit (111) and (100) facets while being small enough to be computationally
affordable. Hydrogen was adsorbed on the surface, as well as absorbed in the cluster, and
optimized. For the latter, i.e. for one single H atom in a subsurface octahedral site, we
have not been able to locate a local minimum. Therefore the energy of the H-Oh structure
and of the transition state leading to it from H* (cf. Fig. 3e) were estimated by computing
the single point energy of a hydrogen atom progressively absorbing into a frozen Pd55
cluster. However, this minimum could be located for a fully hydrogen-covered cluster.

Figure S12. Cuboctahedral Pd55 cluster used in the DFT calculations.

To mimic the polymer-metal interface, a short-polymer model was used. For PTFE, it
consists of a 6-carbon chain of the PTFE fragment, from which two fluorine atoms have
been removed (C6F12) to chemically bind the fragment to the metal surface through Pd-C
bonds, in agreement with the corresponding experimental XPS data (Fig. S12). PTFE is
adsorbed on one edge of Pd55 on the boundary between (111) and (100) facets (Fig. S13).

15
Figure S13. PTFE fragment (C6F12) adsorbed on Pd55 cluster used in the DFT
calculations.

In the case of PMMA, we chose a monomer unit of PMMA (C5O2H10), similar in size to
the previously described PTFE fragment. It allows the interaction with the metal surface
through the C=O of the ester group. This leads to an adsorption energy of Eads=-0.28 eV
for the PMMA fragment. Similar to PTFE, the PMMA is adsorbed at the edge between
(111) and (100) facets of our model Pd55 nanoparticle (Fig. S14).

Figure S14. PMMA fragment (C5H2O10) adsorbed on Pd55 cluster used in the DFT
calculations.

DFT calculations for PMMA coating show similarly decrease apparent activation energy,
Ea, for both absorption and desorption processes (Fig. S15). However, compared to PTFE
coating, the reduction of the absorption activation barrier and the surface-adsorbed
hydrogen destabilization are lower, i.e. 10 kJ/mol and 5 kJ/mol, respectively. These
trends are in good agreement with the experimentally derived Ea for the hydrogen
sorption processes in PMMA-coated sensors (Figs. S41-42) and therefore explain their
lower kinetics-accelerating effect (Figs. 4d-e, S39-40).

16
Figure S15. Comparison between DFT-computed energy landscapes for hydrogen
absorption in Pd55 for the bare nanoparticle (black), with PTFE on the surface (green) and
with PMMA on the surface (blue). The reference is taken as Pd55+1/2 H2 (g),
Pd55@PTFE+1/2H2 (g) and Pd55@PMMA+1/2H2 (g), respectively.

17
4. FDTD Simulation of the Enhanced Sensitivity
The results for the single disk coated by a 30 nm layer (Fig. S16) indicate that as the
polymer coating layer RI increases for a single disk, the sensitivity in terms of peak
induced by the formation of PdH0.67 in the nanoparticles, increases by up to 20 nm for the
largest considered polymer layer RI. Hence, in the case of a PTFE coating with a RI of
ca. 1.38 we obtain a sensitivity increase on the order of 15 nm. This peak increase is
comparable to, but smaller than observed experimentally for a reason we will discuss
below. Mechanistically, the peak enhancement induced by the polymer layer can be
understood by the following analogy. In general, the larger a plasmonic nanoantenna is,
and, consequently, the more to the red it resonates, the larger is its peak response to bulk
RI changes in the environment of the nanoantenna, expressed in nm per RI unit.74 The
same process occurs here, except that what causes the resonance shift is a material
change which, when going from Pd to PdH (see e.g. ref. 75), causes effectively a redshift
of the permittivity function. The more material there is outside the antenna (in terms of
optical density), the larger is the response of the resonator.

Figure S16. (a) FDTD-simulated optical extinction spectra of a Pd nanodisk covered


with a 30 nm thick conformal coating layer with refractive index of n = 1-1.9 in the
metallic (solid lines) and hydrogenated state (PdH0.67 – dashed lines). (b) Enhancement
factor for the hydrogen sorption induced peak shift of the system as a function of the
coating film refractive index. The qualitative trend corroborates the experimental finding
shown in Fig. 2a-b where a polymer coating enhances the absolute sensor response.

Since in the simulation so far only a single disk has been considered, we seek the reason
for the difference in magnitude of the obtained signal enhancement compared to the
experiment in the fact that in the array of nanoparticles used, particle-particle near and
far-field coupling76 is enabled. Accordingly, approximating the amorphous-array inter-
particle coupling by a hexagonal array of disks, we observe an almost doubling of the
sensitivity enhancement, corroborating the reason for the quantitative mismatch in
sensitivity enhancement between theory and experiment (see Fig. S17a). In the
corresponding calculations, the lattice constant is chosen such that the resonance matches
that of the amorphous array used in the experiment.
The magnitude of the sensitivity enhancement is also affected by the thickness of the
coating. As shown in Fig. S17b, the data clearly support the conclusions drawn above.
The resonance position of the antenna is red shifted by both a larger refractive index of

18
the coating, as well as by a larger thickness. Hence, the larger both of them are, the larger
is the redshift, and the larger is the sensitivity enhancement obtained.

Figure S17. (a) FDTD-simulated sensitivity change dependence on the refractive index
of a 30 nm dielectric layer covering the Pd nanodisks when considering only a single disk
and an ensemble of disks arranged in a hexagonal lattice to mimic long range interaction
present in the amorphous array used in the experiments. The symbols are simulated data
and lines indicate best linear fits. (b) Sensitivity enhancement for various coating
thicknesses (5 to 75 nm, in 5 nm steps) as a function of coating layer refractive index for
a single Pd nanodisk. The general trend indicates that the enhancement increases in
magnitude with an increase of the coating layer thickness, due to the induced and
increasing spectral red-shift of the LSPR peak.

The magnitude of the sensitivity enhancement is also affected by the thickness of the
coating. As shown in Fig. S17b, the data clearly support the conclusions drawn above.
The resonance position of the antenna is red shifted by both a larger refractive index of
the coating, as well as by a larger thickness. Hence, the larger both of them are, the larger
is the redshift, and the larger is the sensitivity enhancement obtained.
We also note that enhancing the sensitivity via shifting the LSPR peak to longer
wavelength causes the figure-of-merit (FOM) to be decreased due to the subsequent peak
broadening. By deriving FOM the simulated data in Fig. S16, however, we found the
decrease to be small (Fig. S18).

Figure S18. Figure-of-merit of a Pd nanoparticle when hydrogenated to PdH0.67 as


function of the peak at the metallic state. The data is derived from the simulated results
plotted in Fig. S16.

19
5. Determination of Experimental λpeak
To obtain λpeak we used a Lorentzian fit to the optical spectra. In general, a Lorentzian
function does not provide a good fit to a rather broad and asymmetric spectrum as
characteristic of Pd and PdAu plasmon resonances that we have at hand here. However,
since we in our analysis only apply the fit to a quite narrow part of the spectrum centered
around the peak maximum where it is symmetric, the fit is good. In particular, we fit the
spectra only at ±60 nm from λpeak (Fig. S19) and thus obtain a very good fit with R2 >
0.95 that we use to derive λpeak.

Figure S19. Lorentzian function fitting to the optical spectra to extract λpeak. For our
analysis the fit is only applied within ±60 nm from the peak maximum (yellow shaded
areas) where it is symmetric and thus enables a good fit with R2 > 0.95.

20
6. Experimental Noise Evaluation
The evaluation of the experimental noise is critical especially in the context of limit of
detection (LoD), as its theoretical definition is directly related to the noise. As we
describe in the main text, we use 1 Hz sampling frequency for measurements related to
LoD, to comply with the targeted response time of 1 s. To provide extended experimental
determination of the noise, we carry out a prolonged measurement of the
Pd70Au30@PTFE sensor in the vacuum chamber. Fig. S20 shows the response of the
sensor under 30 min of vacuum followed by 30 min under 40 mbar H2. We first note that
the sensor is extremely stable throughout the measurement. Evaluating the noise
throughout the overall 30 min exposure reveal a noise in the metallic phase, σm, and
hydrogenated state, σh, of 0.007 nm and 0.009 nm, respectively, consistent with the one
we put forward in the main text (cf. Fig. 2e). From here we also note that the noise in the
hydrogenated state is slightly higher due to the LSPR peak of the sensor being shifted to
longer wavelength and broaden.

Figure S20. Lorentzian-fitted Δλpeak response of the 190×25 nm Pd70Au30@PTFE sensor


in vacuum and under exposure to 40 mbar H2 (red-shaded area). A stable signal is
observed in both environments over the course of 30 min. The insets show zoomed-in
regions of Δλpeak spanning 10 min with the derived peak-to-peak noise in the metallic
state, σm, and in the hydrogenated state, σh, identified as 0.007 nm and 0.009 nm,
respectively. The dashed lines and gray-shaded areas denote the mean of the signal and
±σ from the mean, respectively.

To discuss alternative way in determining the λpeak using commonly used method of
centroid,77 we use similar data as above and fit it using 20th degree polynomial which
results in excellent fit (R2 > 0.999). As shown in Fig. S21, the resulting noise is roughly
twice as higher as the one fitted with Lorentzian (Fig. S20). At this point it becomes clear
that having excellent fit (i.e. high R2) does not necessarily implies a high accuracy in
determining the LSPR peak shift. In particular for polynomial fit with high degree, it will
try to follow every detail of the data fitted (and hence gives a very high R 2 value)
including, for example, fluctuations induced by the CCD/CMOS detector readout at the
pixel level. This results in higher level of noise, as we show here.

21
Figure S21. ΔCentroid response of the 190×25 nm Pd70Au30@PTFE sensor in vacuum
and under exposure to 40 mbar H2 (red-shaded area). A stable signal is observed in both
environments over the course of 30 minutes. The insets show zoomed-in regions of
ΔCentroid spanning 10 min with the derived peak-to-peak noise in the metallic state, σm,
and in the hydrogenated state, σh, identified as 0.014 nm and 0.017 nm, respectively. The
dashed lines and gray-shaded areas denote the mean of the signal and ±σ from the mean,
respectively.

We also investigate the evolution of the noise as a function of sampling frequency. To


this end we systematically measure the response of the Pd70Au30@PTFE sensor in
vacuum with subsequently varied sampling frequency from 1 Hz to 6 Hz (the frequency
used for the kinetics measurements) corresponding to changing the number average from
1000 to 10 (see also Methods). As shown in Fig. S22, there is a positive correlation
between sampling frequency and noise. In the case of 6 Hz data, noise level up to 0.062
nm is observed (Fig. S23). We also analyze the distribution of the noise and find them to
be of normal-type distribution (Fig. S22b).

Figure S22. (a) The evolution of experimental signal noise, σ, of a Pd70Au30@PTFE


sensor measured at 0 mbar H2 upon increasing sampling frequency. It is apparent that
higher sampling frequency results in a signal with higher noise. The dashed line and
gray-shaded areas denote the mean of the signal and ±σ from the mean, respectively. (b)
The noise distribution for each sampling frequency, which are found to exhibit a normal
distribution.

22
Figure S23. The experimental signal noise, σ, of Pd70Au30@PTFE sensor measured at 0
mbar H2 as a function of sampling frequency.

For the LoD experiments in Ar and air, we carried out our measurements using flow
reactor setup. To this end, we employed similar light source and spectrometer used in the
vacuum experiments (see Methods) to preserve the optical properties of the samples, and
thus their noise. Parallel to the above, we evaluate the noise of the sensor in the flow
reactor under constant flow of pure Ar followed by 4% H2 in Ar. As shown in Fig. S24,
we observe quantitatively similar noise level to the one measured in vacuum chamber.

Figure S24. Lorentzian-fitted Δλpeak response of the 190×25 nm Pd70Au30@PTFE sensor


in 100% Ar and under exposure to 4% H2 in Ar (red-shaded area) at atmospheric pressure
measured in the flow reactor. A stable signal is observed in both environments over the
course of 30 min. The insets show zoomed-in regions of Δλpeak spanning 10 min with the
derived peak-to-peak noise in the metallic state, σm, and in the hydrogenated state, σh,
identified as 0.009 nm and 0.0011 nm, respectively. The dashed lines and gray-shaded
areas denote the mean of the signal and ±σ from the mean, respectively.

23
7. Strain Induced by PTFE Coating
Deposition of PTFE onto Pd nanodisks results in a slight downward shift of the isotherms
of H2 sorption in Pd (Fig. 2a and c). This effect is expected to be related to the PTFE-
induced strain in Pd. At equilibrium between dissolved hydrogen atoms and hydrogen
molecules in the gas phase, the chemical potentials of these species are related as
2µH = µH2 . (S1)
For the gas phase, we have
µH2 = 𝑘𝐵 𝑇ln(𝑃) + 𝑐𝑜𝑛𝑠𝑡, (S2)
where P is the H2 pressure. For absorbed hydrogen, the contribution of the Pd lattice
strain to the chemical potential is phenomenologically described as78
ΔµH = −𝑣𝜎𝑖𝑖 /3, (S3)
where v is the increment of the lattice volume per interstitial atom, and σii is the trace of
the stress tensor. Combining Eqs. (S1)-(S3), we obtain that the relative strain-induced
shift of the H2 pressure is given by
2𝑣𝜎𝑖𝑖
ln⁡(𝑃∗ /𝑃𝑜 ) = − 3𝑘 , (S4)
𝐵𝑇

where 𝑃∗ and 𝑃𝑜 are the plateau pressures in the PTFE-coated and uncoated Pd,
respectively. The downward shift of the isotherms of H2 sorption is indicative of tensile
PTFE-induced strain in Pd (σii > 0). Such strain can be induced provided the PTFE film is
in the compressed state (compared to the strain-free case).
The analysis of the strain in the PTFE film is complicated by the fact that the formation
of this film takes place under conditions far from equilibrium, and accordingly the film is
expected to be heterogeneous. The corresponding details are not known. In our rough
estimates, we will thus operate with average values. In particular, we take into account
that the scales of the elastic energies calculated per unit area of the Pd disk and the PTFE
film are, respectively, given by79
εPd ≅ 𝐸Pd 𝑢𝑖𝑖2 𝑑Pd and εf ≅ 𝐸f 𝑤𝑖𝑖2 𝑑f , (S5)
where EPd, Ef , uii, and wii are Young’s moduli and the traces of the strain tensor for Pd
and PTFE, respectively, and dPd and df are the corresponding disk and film thicknesses.
The derivative of these energies with respect to the trace of the strain tensor are ≅
2𝐸Pd 𝑢𝑖𝑖 𝑑Pd and ≅ 2𝐸f 𝑤𝑖𝑖 𝑑f (here and below we use the absolute value of wii). At elastic
equilibrium, these derivatives should be comparable, i.e.,
𝐸 𝑤𝑖𝑖 𝑑f
𝑢𝑖𝑖 ≅ 𝐸 f . (S6)
Pd 𝑑Pd

Using this relation and taking into account that σii ≅ 3𝐸Pd 𝑢𝑖𝑖 , we obtain 𝜎𝑖𝑖 ≅
3𝐸f 𝑤𝑖𝑖 𝑑f /𝑑Pd . Substituting the latter relation into Eq. (S4) yields
2𝑣𝐸f 𝑤𝑖𝑖 𝑑f
ln⁡(𝑃∗ /𝑃𝑜 ) ≅ − . (S7)
𝑘𝐵 𝑇𝑑Pd

The range of the reported values of Ef is wide, from 0.3 to 1.3 GPa.80,81 Using the highest
value in combination with vo = 2.6 Å3 and T = 300 K, and assuming that the PTFE

24
compression is significant, wii = 0.3, and the disk and film thicknesses to be comparable,
df ≅ dPd, we obtain ln⁡(𝑃∗ /𝑃𝑜 ) ≅ −0.5. This value is close to what we observe (Fig. 2a and
c). Thus, the PTFE-induced strain in Pd can indeed be the reason behind the observed
shift of the sorption isotherms.

25
8. Sensor@PTFE Sensitivity

Figure S25. Time-resolved Δλpeak response of a 190×25 nm Pd70Au30@PTFE sensor to


different hydrogen concentrations in Ar measured at 30 oC in flow mode (500 mL min-1).
Shaded areas denote the periods where the sensor is exposed to hydrogen. Note that there
are three cycles for each hydrogen concentration.

Figure S26. Time-resolved Δλpeak response of a 190×25 nm Pd70Au30@PTFE sensor to


different hydrogen concentrations in synthetic air measured at 30 oC in flow mode (325
mL min-1). Shaded areas denote the periods where the sensor is exposed to hydrogen.
Note that there are three cycles for each hydrogen concentration.

26
9. Sensor@PTFE Kinetics

Figure S27. Enhanced desorption kinetics are observed for PTFE-coated 190×25 (a) Pd
and (b) Pd70Au30 sensors. t10 is the recovery time defined as the time to reach 10% of the
signal and is marked by dashed lines. (c) Engineering the size of the nanoparticles to
smaller dimensions and more favourable volume/surface ratios enables even faster
desorption kinetics since the rate limiting step occurs at the surface, as shown for 140×25
and 100×25 Pd70Au30 sensors. The shaded areas mark the period when the sensors are
exposed to 40 mbar H2.

Figure S28. (a-d) Absorption and desorption kinetics of Pd sensors before and after
PTFE coating measured at 4 different temperatures (30–60 oC, in 10 oC steps). (e)
Derived absorption and desorption apparent activation energies, Ea, through an Arrhenius
plot. It is clear that a reduction in Ea for both absorption and desorption occurs upon
PTFE coating. For neat Pd, the found values are in good agreement with reports for the
bulk (see e.g. ref. 71 and references therein), validating our measurement approach.

27
Figure S29. (a-d) Absorption and desorption kinetics of Pd70Au30 sensors before and
after PTFE coating, measured at 4 different temperatures (30–60 oC, in 10 oC steps). (e)
Derived absorption and desorption apparent activation energies, Ea. Similar to the Pd data
above, a reduction in Ea for both absorption and desorption is induced by the PTFE
coating.

Figure S30. Raw absorption kinetics response of (a) 140×25 nm and (b) 190×25 nm
Pd70Au30@PTFE sensors to varying H2 pressure from 1000 to 1 mbar at 30 oC. Dashed
lines mark the t90 and shaded areas denote the period when the sensor is exposed to H2.

28
Figure S31. Raw desorption kinetics response of (a) 100×25 nm, (b) 140×25 nm and (c)
190×25 nm Pd70Au30@PTFE sensors to varying H2 pressure from 1000 to 1 mbar at 30
o
C. Dashed lines mark the t10 (recovery time), i.e. the time to reach 10% of the total
response, and shaded areas denote the period when the sensor is exposed to H2. (d)
Recovery time of Pd70Au30@PTFE sensors with different sizes to varying H2 pressure
pulses. Similar to absorption, engineering the size of the nanoparticles to smaller
dimensions enables faster kinetics through optimized volume/surface ratio. For the
100×25 sensor, a recovery time of less than 5 s throughout the investigated H2 pressure
range is achieved. The lines denote power-law fit based on Eq. (S24).

29
10. Theoretical Analysis of the Hydrogenation Kinetics of PdAu Nanoparticles
1. General remarks
Hydrogen absorption by and desorption from PdAu alloy nanoparticles with or without
polymer coating depend on a multitude of factors and may occur via various scenarios
depending on the fraction of Au in the alloy. For the Pd70Au30 nanoparticles used in our
experiments, there is no hysteresis (cf. Fig. 2b), and the time scales of the hydrogen
adsorption kinetics are weakly dependent on pressure (see e.g. Figs. 3g, 4e, and 5c) and
can be represented by power-law expressions with relatively low exponents. Both these
features can be explained by the fact that at this and higher fractions of Au, the PdAu
nanoparticles are energetically highly heterogeneous because the Au atoms, distributed
nearly at random (cf. Fig. 1b), decrease the H binding energy. In this case, the H
absorption is energetically most favorable at the sites fully surrounded by Pd. If the
fraction of Au is smaller than or equal to 0.25, the concentration of such sites is not too
low, and the PdAu nanoparticles exhibit hysteresis. If the fraction of Au is above 0.25,
the concentration of such sites is nearly negligible, and the heterogeneity dominates. The
role of this factor has been widely discussed in heterogeneous catalysis where the
reaction time scales are often weakly dependent on pressure, and the reaction kinetics are
fitted by power-law expressions. Phenomenologically, the effect of the energetic
heterogeneity of the substrate on the kinetics of various process can be described by
introducing a linear dependence of the binding and activation energies on adsorbate
and/or absorbate coverage.82 In this case, the dependence of the rates on coverage is
related primarily to that of the activation energy. Under this condition, the coverages in
the pre-exponential terms can be included into the pre-exponential factors of the
corresponding rate constants. Following this line, we interpret here the observed kinetics.
Our theoretical analysis is based on a model that includes the reversible hydrogen
adsorption on surface or subsurface sites, according to
H2 + 2𝑆 ↔ 2𝐻𝑆 , (S8)
and the reversible diffusion of hydrogen from the surface into the bulk of a PdAu alloy
nanoparticle,
H𝑆 ↔ 𝐻𝑎 . (S9)
For our treatment we assume that the fraction of Au in the alloy is large (i.e. 30 at.%) so
that the absorption-desorption hysteresis is negligible like in the experiments considered
here. As already noticed, we also assume that the Au atoms are distributed randomly in
the alloy, and hence, due to the presence of the Au atoms in the Pd matrix, the bulk and
the surface of a nanoparticle are energetically heterogeneous in terms of the energies of H
binding on the absorption and adsorption sites. In this case, the hydrogen uptake or, more
specifically, the average occupation of bulk sites by H, , can be phenomenologically
described by using the Temkin absorption isotherm,82
𝜃 = 𝑚⁡ln(𝐾𝑃), (S10)
where P is the H2 pressure, and K and m are the kinetic parameters (Fig. 2b shows that
one needs to change the pressure by about four orders of magnitude to change  by about
0.5, meaning that m is about 0.1). This isotherm implies that the energy distribution of the

30
bulk absorption sites for hydrogen inside the alloy is broad and linear. In this case, the
effect of energetic heterogeneity on the kinetics of rate processes can be described by (i)
assuming the distribution of H atoms inside a nanoparticle to be close to equilibrium (this
is reasonable because the H diffusion is rapid), by (ii) introducing a linear dependence of
the activation energies on the average occupation of sites and by (iii) ignoring the
dependence of the pre-exponential factors on site occupation. Below, we systematically
use this approximation for describing the hydrogen absorption and desorption kinetics in
the PdAu alloy nanoparticles at hand.

2. Absorption
Since hydrogen dissociation on Pd and PdAu surfaces essentially is non-activated, the
rate-limiting step for H absorption is related to the activated “jumps” of H atoms from the
surface layer to the NP bulk. The number of absorption and adsorption sites is given by
V/a3 and S/a2, where V and S are the nanoparticle volume and surface area, and a is the
size characterizing the absorption and adsorption sites (in reality, the corresponding sizes
are slightly different, but this fact is negligible in our context). With this specification, the
rate of this process is represented as
𝑉𝑑𝜃/𝑑𝑡 = 𝑆𝑎𝜅+ exp(𝜂𝜗 − 𝛾𝜃), (S11)
where 𝜗⁡is the average occupation of the surface sites, 𝜅+ is the occupation-independent
absorption rate constant, and 𝛾 and 𝜂 are positive dimensionless parameters related to the
energetic heterogeneity of the alloy system. Physically, the presence of H atoms in the
surface layer changes the activation energies of the transition and ground states along the
profile of the potential energy for a jump. The changes of both these energies are positive
for increasing surface site occupation. Accordingly, the change of the former energy
increases the activation energy, while the change of the latter energy decreases it and is
generally considered to be stronger. The first term in the argument of the exponent, 𝜂𝜗,
takes both these effects into account. Specifically, it describes the resulting decrease of
the activation energy of jumps with increasing 𝜗. The second term, 𝛾𝜃, corresponds to
the increase of the activation energy of jumps due to the presence of H atoms in the
nanoparticle bulk.
During absorption, the surface layer can be considered to be at equilibrium with the gas
phase, meaning that 𝜗 is a parameter determined by the applied H2 pressure, P. Under
such conditions, Eq. (S11) is integrated as
𝛾𝜅+ 𝑆𝑎
exp(𝛾𝜃) − 1 = exp(𝜂𝜗). (S12)
𝑉
This then yields
𝛾𝜅+ 𝑆𝑎
exp(𝛾𝜃0.5 ) − 1 = exp(𝜂𝜗)𝑡0.5, (S13)
𝑉

where 𝜃0.5 is 50% of the equilibrium bulk site occupation at given P, and 𝑡0.5 is the time
needed to reach 𝜃0.5. If the energy distribution of the absorption sites is broad as in the
present case for the PdAu alloy, 𝛾 is sizable, and consequently also 𝛾𝜃0.5 . Thus we keep
only the first term on the left-hand side of Eq. (S13) and get

31
𝑉
𝑡0.5 ≅ 𝛾𝜅 exp(𝛾𝜃0.5 − 𝜂𝜗). (S14)
+ 𝑆𝑎

According to Eq. (S10), we have


𝑚
𝜃0.5 = ln⁡(𝐾𝑃), (S15)
2
Furthermore, the adsorption-desorption equilibrium in the surface layer can be described
as
𝑑𝜗/𝑑𝑡 = 𝑘𝑎 𝑃exp(−𝛼𝜗) − 𝑘𝑑 exp(𝛽𝜗) = 0, or (S16)
ln⁡(𝑘𝑎 𝑃/𝑘𝑑 )
𝜗= , (S17)
𝛼+𝛽

where ka and kd are the occupation-independent rate constants, and 𝛼 and 𝛽 are positive
dimensionless parameters.
Substituting Eqs. (S15) and (S17) into Eq. (S13) results in
𝛾𝑚 𝜂
𝑡0.5 ∝ (𝑉/𝑆)𝑃 𝑥 where 𝑥 = − 𝛼+𝛽. (S18)
2

3. Desorption
When it comes to desorption, the hydrogen atom concentration in the nanoparticle bulk is
considered to be at equilibrium with the hydrogen at the surface, and the equation
describing the occupation of the bulk sites corresponds to
𝑉𝑑𝜃/𝑑𝑡 = −𝑆𝑎𝑘𝑑 exp(𝛽𝜗), (S19)
where 𝜗 is the surface-site occupation corresponding to a given bulk site occupation 𝜃.
The equilibrium between the bulk and surface is described as
𝑑𝜗/𝑑𝑡 = 𝜅+ exp(𝜒𝜃 − 𝜁𝜗) − ⁡ 𝜅− exp(𝜂𝜗 − 𝛾𝜃) = 0, or (S20)
ln⁡(𝜅+ /𝜅− ) 𝜒+𝛾
𝜗= + 𝜂+𝜁 𝜃, (S21)
𝜂+𝜁

where 𝜅+ and 𝜅− are the occupation-independent forward and backward rate constant,
and 𝜒, 𝜁, 𝜂, and 𝛾 are the positive dimensionless parameters. Substituting Eq. (S21) into
Eq. (S19) yields
𝜒+𝛾
𝑑𝜃/𝑑𝑡 ∝ −exp⁡(𝛽 𝜂+𝜁 𝜃). (S22)

Integrating the latter equation in analogy with Eqs. (S11)-(S14) above, we obtain
𝜒+𝛾
𝑡0.5 ∝ (𝑉/𝑆)exp⁡(−𝛽 𝜂+𝜁 𝜃0.5 ). (S23)

Taking Eq. (S15) into account, this expression can be rewritten as


𝛽𝑚(𝜒+𝛾)
𝑡0.5 ∝ (𝑉/𝑆)𝑃𝑦 where 𝑦 = − . (S24)
2(𝜂+𝜁)

The derived Eqs. (S18) and (S24) now allow us to discuss the experimentally obtained
kinetics by plotting log10(1/t0.5) versus log10(P) for the PdAu sensor in the presence and
absence of the PTFE and PMMA coatings (Figs. S32-33, respectively). As the key

32
observations we report (i) good agreement with the theoretically predicted power-law
dependence, (ii) essentially no effect of the polymer coating on the slope and (iii)
fractional ab/desorption orders. The latter is a consequence of the aforementioned wide
energy distribution of the bulk hydrogen absorption sites in the alloy. Specifically in our
model, such a wide site energy distribution during hydrogen absorption (Eq. S18) is
reflected in that the parameters 𝛼, 𝛽, 𝜂, and 𝛾 are expected to be comparable (e.g., ≅ 5, as
one often has in the case of elementary rate processes on solid surfaces). Furthermore,
since m ≅ 0.1, 𝜂/(𝛼 + 𝛽) and 𝛾𝑚/2 are on the order of 0.4–0.6 and 0.1–0.2, respectively.
Accordingly, in agreement with our experiments (Figs. S32-33), x is on the order of -
(0.4–0.5). Similarly for desorption (Eq. S24) with the scale of the parameters indicated
above, the scale of 𝛽/2 and (𝜒 + 𝛾)/(⁡𝜂 + 𝜁) are on the order of 2–3 and 1, respectively,
and we find that y is on the order of -(0.2 to 0.3), also in agreement with our data.

Figure S32. Absorption kinetics of Pd70Au30 sensor (a) before and (c) after PTFE coating
and also the corresponding desorption kinetics for the sensor (b) before and (d) after
coating measured at different H2 pressures at 30 oC. (e) The derived orders of the
hydrogen absorption and desorption obtained for the sensor by plotting the corresponding
decimal logarithmic time scales (here defined at t0.5) vs. decimal logarithmic pressure.
The application of the PTFE coating does not change the sorption orders. The lines
denote power-law fit based on Eqs. (S18) and (S24) for absorption and desorption,
respectively.

33
Figure S33. Absorption kinetics of Pd70Au30 sensor (a) before and (c) after PMMA
coating and also the corresponding desorption kinetics for the sensor (b) before and (d)
after coating measured at different H2 pressures at 30 oC. (e) The derived orders of the
hydrogen absorption and desorption obtained for the sensor by plotting the corresponding
decimal logarithmic time scales (here defined at t0.5) vs. decimal logarithmic pressure.
Similar to the PTFE case above, the application of a PMMA coating does not change the
sorption orders. The lines denote power-law fit based on Eqs. (S18) and (S24) for
absorption and desorption, respectively.

4. Particle Size Dependence


Based on the above treatment, we note that the time scales of the hydrogen absorption by
the nanoparticles under consideration are proportional to the volume/surface ratio (V/S).
Thus, plotting the experimentally obtained response times for different hydrogen partial
pressures as a function of the V/S for the three considered particle sizes reveals that this
indeed is the case for both polymer coatings (Fig. S34). Furthermore, we also find that
the absorption order is very similar for the different particle sizes, confirming that there is
no change in rate limiting elementary step as the particle size is decreased (Fig. S35).

34
Figure S34. Response time as a function of volume-to-surface ratio of (a)
Pd70Au30@PTFE and (b) Pd70Au30@PMMA sensors at different applied H2 pressures.
Linear correlation between the two parameters is established where the slope decreases as
the H2 pressure increases.

From these plots we also observe a pressure dependence of the amplitude of the impact of
V/S on response time, in other words, at higher pressures the effect is less pronounced.
This can be rationalized by the fact that the proportionality constant depends on the
specifics of the kinetics, which, as we discuss above, are dependent on hydrogen
pressure. Specifically, the parameter 𝜂 in Eq. (S11) describes a decrease in the activation
energy for a jump of H from the surface to the subsurface with increasing the surface-site
coverage, 𝜗. Since 𝜗 is H2 pressure dependent (Eq. S17) and increases with increasing
pressure, this explains the observed trend (Fig. S36).
These findings have an important implication for Pd-based hydrogen sensors, namely that
it is impossible to reach the DoE response time target at ambient temperature at 1 mbar
hydrogen pressure with macroscopic (bulk) or, to some extent, thin-film systems, since in
this low pressure and temperature regime the V/S is the critical factor determining the
response time. In particular for a thin-film system, a typical film thickness of 40 nm is
employed to ensure adequate optical signal contrast during hydrogenation; see e.g. refs.
62, 83–85. Comparing to nanoparticles of 100 nm diameter with similar 40 nm thickness,
the resulting V/S of the thin film is 2.6 times higher than the one of a nanoparticle;
translated into response time of ~9 s in the case of PdAu@PTFE system at 1 mbar (cf.
Fig. S34). In other words, only nanostructuring and optimization of the V/S of the sensing
material can lead to faster response time, together with interface engineering to reduce
the activation barrier, as the case when adding a polymer coating.

35
Figure S35. (a) Absorption and (b) desorption orders of 100×25 nm 140×25 nm and
190×25 nm Pd70Au30@PTFE sensors. Note that the sorption orders are very similar
despite the different particle dimensions. The lines in panel (a) and (b) denote power-law
fit based on Eqs. (S18) and (S24), respectively.

Figure S36. Slope of volume-to-surface-ratio – response time correlations (cf. Fig. S34)
as a function of hydrogen pressure.

36
11. Sensors@PMMA

Figure S37. Optical absorption and desorption isotherms of (a) Pd and (b) Pd70Au30
sensors before and after PMMA coating. The arrows denote the sorption direction. The
panels to the right depict the Δλpeak ratio of the coated and uncoated sensors, and reveal a
factor of two enhanced signal amplitude induced by the PMMA, similar to corresponding
data of PTFE (cf. Fig. 2a-b). Also, symmetric lowering of plateau pressures is observed,
hinting at a generic effect exerted by a polymer coating.

Figure S38. Gravimetric QCM isotherms of (a) a Pd and (b) a Pd70Au30 sensor before
and after PMMA coating. The obtained hydrogen concentration in the metal is consistent
with the corresponding data of the PTFE-coated sensors (cf. Fig. 2c-d). The identical
response of the coated and uncoated systems furthermore corroborates that no H2 is
absorbed in the PMMA itself.

37
Figure S39. Absorption kinetics of (a) Pd and (b) Pd70Au30 and desorption kinetics of (c)
Pd and (d) Pd70Au30 before and after PMMA coating. For all cases, an enhancement in
the kinetics is observed, however, to a lesser extent compared to the PTFE coating. The
shaded areas mark the period when the sensors are exposed to 40 mbar H2.

38
Figure S40. Raw absorption kinetics response of (a) 100×25 nm (b) 140×25 nm and (c)
190×25 nm Pd70Au30@PMMA sensors to varying H2 pressure from 1000 to 1 mbar at 30
o
C. Dashed lines mark t90 and shaded areas denote the period when the sensor is exposed
to H2.

39
Figure S41. (a-d) Absorption and desorption kinetics of Pd sensors before and after
PMMA coating measured at 4 different temperatures (30–60 oC, in 10 oC steps). (e)
Arrhenius plot and corresponding absorption and desorption apparent activation energies,
Ea, for neat and PMMA coated Pd. A reduction in Ea is also observed upon PMMA
coating, however, smaller in magnitude compared to PTFE.

Figure S42. (a-d) Absorption and desorption kinetics of Pd70Au30 sensors before and
after PMMA coating measured at 4 different temperatures (30–60 oC, in 10 oC steps). (e)
Arrhenius plot and corresponding absorption and desorption apparent activation energies,
Ea, for neat and PMMA coated Pd. A reduction in Ea is also observed upon PMMA
coating, however, smaller in magnitude compared to PTFE.

40
12. Variation in Sensor Sensitivity due to Difference in the Optical Properties
We note a difference in the absolute Δλpeak of the uncoated Pd70Au30 sensors upon
hydrogenation used in two different experiments shown in Figs. 2b and 4c. To explain
this observation, we recall that these sensors were fabricated at different points in time in
two separate batches. To this end we highlight that it is in the nature of the used
nanofabrication method86 that the particle surface density in the array, and thus the
average center-to-center distance, varies from batch to batch, as a consequence of the
method being self-assembly based. As we have shown, even minute variations in surface
density in the amorphous array can have a quite significant impact on the LSPR
wavelength, including oscillations in the spectral peak position, λpeak.76 Consequently, as
we also have reported recently, this array-to-array variation in λpeak, is expected to have a
sizable impact on the λpeak induced by hydrogen sorption since it is the spectral position
in the non-hydrogenated state that dictates the sensitivity in this respect.87 This is also
supported by our FDTD calculation above (cf. Fig. S17a). Hence, we argue that the
difference in the absolute Δλpeak of the uncoated Pd70Au30 sensors shown in Figs. 2b and
4c is a consequence of this effect. To corroborate this argument, we have fabricated two
additional Pd70Au30 sensors and investigated their response to hydrogen and their LSPR
properties as such. Indeed their LSPR peak maximum varies (Fig. S43a) and we observe
corresponding variation in the Δλpeak upon hydrogenation (Fig. S43b). Plotting the Δλpeak
upon hydrogenation vs λpeak in the non-hydrogenated state (Fig. S43c) then indeed reveals
the anticipated linear correlation in agreement with our recently discovered design rule,87
and confirms the origin of the different sensitivities. In a wider perspective, this aspect
also opens up further possibilities of engineering the limit of detection of our sensor by
optimizing the surface density of nanoparticles in the amorphous array.

41
Figure S43. (a) Normalized optical extinction spectra of uncoated Pd70Au30 sensors
fabricated at different batches. The data include those for the sensors shown in Figs. 2b
and 4c. In addition to different fabrication batches, sensor shown in Fig. 2b was
fabricated using a different batch of polystyrene beads (see Methods). (b) The
corresponding absorption and desorption isotherms of the sensors. (c) Δλpeak of the
sensors upon exposure to 1000 mbar of hydrogen as a function of their λpeak in the
metallic state. A linear correlation between the two parameters is observed which
indicates higher sensitivity for sensors with Δλpeak at longer wavelength.

42
13. Tandem Sensors

Figure S44. Raw absorption kinetics response of 100×25 nm Pd70Au30@PTFE@PMMA


tandem sensor to varying H2 pressure from 1000 to 1 mbar at 30 oC. The dashed line
marks the t90 and shaded area denotes the period when the sensor is exposed to H2.

Figure S45. (a-d) Absorption and desorption kinetics of Pd@PTFE sensors before and
after PMMA coating, measured at 4 different temperatures (30–60 oC, in 10 oC steps). (e)
Arrhenius plot and corresponding absorption and desorption apparent activation energies,
Ea. Clearly, the addition of the PMMA layer on top of the PTFE does not change the
obtained Ea, in excellent agreement with the observed identical response times.

43
Figure S46. Optical absorption and desorption isotherms of Pd70Au30@PTFE sensor
before and after PMMA coating. The arrows denote the sorption direction. The panels to
the right depict the Δλpeak ratio of the coated and uncoated sensors, which reveals around
30% enhanced signal amplitude, due to the increasing total coating thickness by adding
the second polymer layer, in agreement with the FDTD simulations (cf. Fig. S17b).

Figure S47 (a) Δλpeak response of the Pd70Au30@PTFE@PMMA tandem sensor upon
exposure to (b) a step-wise decreasing H2 pressure in the 7–1000 µbar range, measured at
1 Hz sampling frequency in a vacuum chamber at 30 oC. (c) Derived Δλpeak response as a
function of the H2 pressure. The black dashed line marks the limit of detection (LoD) at
3σ = 0.03 nm (cf. Fig. 2e). The orange dashed lines depict an extrapolation from the
lowest reliably attainable data point in our system (7 µbar) to the 3σ point, indicating a
LoD < 3 µbar. The obtained sensitivity is higher than the one exhibited by the
Pd70Au30@PTFE sensor (green, cf. Fig. 2f).

44
14. Deactivation Tests

Figure S48. Top: Time-resolved Δλpeak response of Pd70Au30, Pd70Au30@PTFE,


Pd70Au30@PMMA and Pd70Au30@PTFE@PMMA to 2 pulses of 4% H2 followed by 10
pulses of 4% H2 + 3% CO2, 4% H2 + 0.5% CH4, 4% H2 + 0.1% CO, and 4% H2 + 0.05%
NO2. All were measured using synthetic air as carrier gas. The uncoated Pd70Au30 sensor
is readily deactivated by CO and NO2. When coated with PTFE, there is a slight
resistance towards these gases. In contrast, both Pd70Au30@PMMA and
Pd70Au30@PTFE@PMMA sensors are completely resistant to all of the tested gases.
Bottom: Normalized sensor signal to the one obtained in pure 4% H2. The error bars
denote the standard deviation from 10 cycles. The shaded area indicates the ±20%
deviation limit from the normalized Δλpeak in pure 4% H2.

45
Figure S49. (a) Time-resolved Δλpeak response of Pd to 2 pulses of 4% H2 followed by 10
pulses of 4% H2 + 0.1% CO. Note that with increasing time the sensor is completely
deactivated by CO. (b) Normalized sensor signal to the one obtained in pure 4% H2. The
error bars denote the standard deviation from 10 cycles. The shaded area indicates the
±20% deviation limit from the normalized Δλpeak in pure 4% H2.

46
References
1. Hübert, T., Boon-Brett, L., Black, G. & Banach, U. Hydrogen sensors - A review.
Sensors Actuators, B Chem. 157, 329–352 (2011).
2. Boon-Brett, L. et al. Identifying performance gaps in hydrogen safety sensor
technology for automotive and stationary applications. Int. J. Hydrogen Energy
35, 373–384 (2010).
3. Manjavacas, G. & Nieto, B. Hydrogen sensors and detectors. Compend. Hydrog.
Energy 215–234 (2016).
4. Hydrogen Detectors Adopted in Toyota Fuel Cell Vehicle Mirai. (2014). Available
at: http://www.nissha.com/english/news/2014/12/12th_1.html. (Accessed: 9th
March 2018)
5. U.S. Department of Energy, Energy Efficiency and Renewable Energy (EERE), Fuel
Cell Technologies Office. Multi-Year Research, Development, and Demonstration
Plan, 2011-2020. Section 3.7 Hydrogen Safety, Codes and Standards. (2015).
6. ISO 26142:2010 Hydrogen detection apparatus - Stationary applications. (2010).
7. U.S. Department of Energy, Hydrogen, Fuel Cells & Infrastructure Technologies
Program. Multi-Year Research, Development, and Demonstration Plan, 2003-
2010. Section 3.4 Fuel Cells. (2005).
8. Cho, S. Y. et al. Ultrasmall Grained Pd Nanopattern H2Sensor. ACS Sensors 3,
1876–1883 (2018).
9. He, J. et al. Integrating plasmonic nanostructures with natural photonic
architectures in Pd-modified Morpho butterfly wings for sensitive hydrogen gas
sensing. RSC Adv. 8, 32395–32400 (2018).
10. Hu, J. et al. Highly sensitive and ultra-fast gas sensor based on CeO2-loaded In2O3
hollow spheres for ppb-level hydrogen detection. Sensors Actuators B Chem. 257,
124–135 (2018).
11. Lupan, O. et al. Ultra-sensitive and selective hydrogen nanosensor with fast
response at room temperature based on a single Pd/ZnO nanowire. Sensors
Actuators B Chem. 254, 1259–1270 (2018).
12. Seo, J., Lim, Y. & Shin, H. Self-heating hydrogen gas sensor based on an array of
single suspended carbon nanowires functionalized with palladium nanoparticles.
Sensors Actuators B Chem. 247, 564–572 (2017).
13. Zhang, D., Sun, Y., Jiang, C. & Zhang, Y. Room temperature hydrogen gas sensor
based on palladium decorated tin oxide/molybdenum disulfide ternary hybrid via
hydrothermal route. Sensors Actuators B Chem. 242, 15–24 (2017).
14. Luo, X. et al. Rapid hydrogen sensing response and aging of α-MoO3 nanowires
paper sensor. Int. J. Hydrogen Energy 42, 8399–8405 (2017).
15. Sanger, A., Jain, P. K., Mishra, Y. K. & Chandra, R. Palladium decorated silicon
carbide nanocauliflowers for hydrogen gas sensing application. Sensors Actuators
B Chem. 242, 694–699 (2017).
16. Shim, Y.-S. et al. Nanogap-controlled Pd coating for hydrogen sensitive switches
and hydrogen sensors. Sensors Actuators B Chem. 255, 1841–1848 (2018).
17. Baek, D.-H. & Kim, J. MoS2 gas sensor functionalized by Pd for the detection of
hydrogen. Sensors Actuators B Chem. 250, 686–691 (2017).
47
18. Venkatesan, A. et al. Molybdenum disulfide nanoparticles decorated reduced
graphene oxide: highly sensitive and selective hydrogen sensor. Nanotechnology
28, 365501 (2017).
19. Jang, J.-S. et al. Hollow Pd–Ag Composite Nanowires for Fast Responding and
Transparent Hydrogen Sensors. ACS Appl. Mater. Interfaces 9, 39464–39474
(2017).
20. Yamazaki, H., Hayashi, Y., Masunishi, K., Ono, D. & Ikehashi, T. High sensitivity
MEMS capacitive hydrogen sensor with inverted T-shaped electrode and ring-
shaped palladium alloy for fast response and low power consumption. J.
Micromechanics Microengineering 28, 094001 (2018).
21. Zhang, H., Yi, J. & Jiang, X. Fast Response, Highly Sensitive and Selective Mixed-
Potential H 2 Sensor Based on (La, Sr)(Cr, Fe)O 3-δ Perovskite Sensing Electrode.
ACS Appl. Mater. Interfaces 9, 17218–17225 (2017).
22. Woo, J.-A., Phan, D.-T., Jung, Y. W. & Jeon, K.-J. Fast response of hydrogen sensor
using palladium nanocube-TiO2 nanofiber composites. Int. J. Hydrogen Energy 42,
18754–18761 (2017).
23. Agrawal, A. V. et al. Fast detection and low power hydrogen sensor using edge-
oriented vertically aligned 3-D network of MoS 2 flakes at room temperature.
Appl. Phys. Lett. 111, 093102 (2017).
24. Hassan, K. & Chung, G.-S. Fast and reversible hydrogen sensing properties of Pd-
capped Mg ultra-thin films modified by hydrophobic alumina substrates. Sensors
Actuators B Chem. 242, 450–460 (2017).
25. Kumar, M. et al. Efficient room temperature hydrogen sensor based on UV-
activated ZnO nano-network. Nanotechnology 28, 365502 (2017).
26. Koo, W.-T. et al. Accelerating Palladium Nanowire H2 Sensors Using Engineered
Nanofiltration. ACS Nano 11, 9276–9285 (2017).
27. Hassan, K., Uddin, A. S. M. I., Ullah, F., Kim, Y. S. & Chung, G.-S.
Platinum/palladium bimetallic ultra-thin film decorated on a one-dimensional
ZnO nanorods array for use as fast response flexible hydrogen sensor. Mater.
Lett. 176, 232–236 (2016).
28. Moon, J., Hedman, H.-P., Kemell, M., Tuominen, A. & Punkkinen, R. Hydrogen
sensor of Pd-decorated tubular TiO2 layer prepared by anodization with
patterned electrodes on SiO2/Si substrate. Sensors Actuators B Chem. 222, 190–
197 (2016).
29. Sanger, A., Kumar, A., Kumar, A. & Chandra, R. Highly sensitive and selective
hydrogen gas sensor using sputtered grown Pd decorated MnO2 nanowalls.
Sensors Actuators B Chem. 234, 8–14 (2016).
30. Chen, W. P. et al. Extraordinary room-temperature hydrogen sensing capabilities
of porous bulk Pt–TiO2 nanocomposite ceramics. Int. J. Hydrogen Energy 41,
3307–3312 (2016).
31. Uddin, A. S. M. I. & Chung, G.-S. A self-powered active hydrogen gas sensor with
fast response at room temperature based on triboelectric effect. Sensors
Actuators B Chem. 231, 601–608 (2016).
32. Xia, X. et al. A hydrogen sensor based on orientation aligned TiO2 thin films with

48
low concentration detecting limit and short response time. Sensors Actuators B
Chem. 234, 192–200 (2016).
33. Hassan, K., Iftekhar Uddin, A. S. . & Chung, G.-S. Fast-response hydrogen sensors
based on discrete Pt/Pd bimetallic ultra-thin films. Sensors Actuators B Chem.
234, 435–445 (2016).
34. Yoo, H.-W., Cho, S.-Y., Jeon, H.-J. & Jung, H.-T. Well-Defined and High Resolution
Pt Nanowire Arrays for a High Performance Hydrogen Sensor by a Surface
Scattering Phenomenon. Anal. Chem. 87, 1480–1484 (2015).
35. Pak, Y. et al. Palladium Nanoribbon Array for Fast Hydrogen Gas Sensing with
Ultrahigh Sensitivity. Adv. Mater. 27, 6945–6952 (2015).
36. Song, H. et al. Optical fiber hydrogen sensor based on an annealing-stimulated
Pd–Y thin film. Sensors Actuators B Chem. 216, 11–16 (2015).
37. Sanger, A., Kumar, A., Chauhan, S., Gautam, Y. K. & Chandra, R. Fast and
reversible hydrogen sensing properties of Pd/Mg thin film modified by
hydrophobic porous silicon substrate. Sensors Actuators B Chem. 213, 252–260
(2015).
38. Rajoua, K., Baklouti, L. & Favier, F. Electronic and Mechanical Antagonist Effects in
Resistive Hydrogen Sensors Based on Pd@Au Core–Shell Nanoparticle Assemblies
Prepared by Langmuir–Blodgett. J. Phys. Chem. C 119, 10130–10139 (2015).
39. Li, X., Liu, Y., Hemminger, J. C. & Penner, R. M. Catalytically Activated
Palladium@Platinum Nanowires for Accelerated Hydrogen Gas Detection. ACS
Nano 9, 3215–3225 (2015).
40. Chen, R., Ruan, X., Liu, W. & Stefanini, C. A reliable and fast hydrogen gas leakage
detector based on irreversible cracking of decorated palladium nanolayer upon
aligned polymer fibers. Int. J. Hydrogen Energy 40, 746–751 (2015).
41. Phan, D.-T., Uddin, A. S. M. I. & Chung, G.-S. A large detectable-range, high-
response and fast-response resistivity hydrogen sensor based on Pt/Pd core–shell
hybrid with graphene. Sensors Actuators B Chem. 220, 962–967 (2015).
42. Hong, J. et al. A highly sensitive hydrogen sensor with gas selectivity using a
PMMA membrane-coated Pd nanoparticle/single-layer graphene hybrid. ACS
Appl. Mater. Interfaces 7, 3554–3561 (2015).
43. Xiang, C. et al. A room-temperature hydrogen sensor based on Pd nanoparticles
doped TiO2 nanotubes. Ceram. Int. 40, 16343–16348 (2014).
44. Lim, S. H. et al. Flexible Palladium-Based H 2 Sensor with Fast Response and Low
Leakage Detection by Nanoimprint Lithography. ACS Appl. Mater. Interfaces 5,
7274–7281 (2013).
45. Perrotton, C. et al. A reliable, sensitive and fast optical fiber hydrogen sensor
based on surface plasmon resonance. Opt. Express 21, 382 (2013).
46. Westerwaal, R. J. et al. Nanostructured Pd–Au based fiber optic sensors for
probing hydrogen concentrations in gas mixtures. Int. J. Hydrogen Energy 38,
4201–4212 (2013).
47. Chung, M. G. et al. Flexible hydrogen sensors using graphene with palladium
nanoparticle decoration. Sensors Actuators B Chem. 169, 387–392 (2012).
48. Lee, J., Shim, W., Lee, E., Noh, J.-S. & Lee, W. Highly Mobile Palladium Thin Films

49
on an Elastomeric Substrate: Nanogap-Based Hydrogen Gas Sensors. Angew.
Chemie Int. Ed. 50, 5301–5305 (2011).
49. Kiefer, T., Villanueva, L. G., Fargier, F., Favier, F. & Brugger, J. Fast and robust
hydrogen sensors based on discontinuous palladium films on polyimide,
fabricated on a wafer scale. Nanotechnology 21, 505501 (2010).
50. Offermans, P. et al. Ultralow-power hydrogen sensing with single palladium
nanowires. Appl. Phys. Lett. 94, 223110 (2009).
51. Monzón-Hernández, D., Luna-Moreno, D. & Martínez-Escobar, D. Fast response
fiber optic hydrogen sensor based on palladium and gold nano-layers. Sensors
Actuators B Chem. 136, 562–566 (2009).
52. Kiefer, T., Favier, F., Vazquez-Mena, O., Villanueva, G. & Brugger, J. A single
nanotrench in a palladium microwire for hydrogen detection. Nanotechnology 19,
125502 (2008).
53. Xu, T. et al. Self-assembled monolayer-enhanced hydrogen sensing with ultrathin
palladium films. Appl. Phys. Lett. 86, 1–3 (2005).
54. Villatoro, J. & Monzón-Hernández, D. Fast detection of hydrogen with nano fiber
tapers coated with ultra thin palladium layers. Opt. Express 13, 5087 (2005).
55. Baselt, D. R. et al. Design and performance of a microcantilever-based hydrogen
sensor. Sensors Actuators B Chem. 88, 120–131 (2003).
56. Favier, F., Walter, E. C., Zach, M. P., Benter, T. & Penner, R. M. Hydrogen sensors
and switches from electrodeposited palladium mesowire arrays. Science 293,
2227–2231 (2001).
57. Walter, E. C., Favier, F. & Penner, R. M. Palladium mesowire arrays for fast
hydrogen sensors and hydrogen-actuated switches. Anal. Chem. 74, 1546–1553
(2002).
58. Chen, M. et al. Response Characteristics of Hydrogen Sensors Based on PMMA-
Membrane-Coated Palladium Nanoparticle Films. ACS Appl. Mater. Interfaces 9,
27193–27201 (2017).
59. Szilágyi, P. Á., Westerwaal, R. J., Van De Krol, R., Geerlings, H. & Dam, B. Metal-
organic framework thin films for protective coating of Pd-based optical hydrogen
sensors. J. Mater. Chem. C 1, 8146–8155 (2013).
60. Maeland, A. & Flanagan, T. B. Lattice Spacings of Gold–Palladium Alloys. Can. J.
Phys. 42, 2364–2366 (1964).
61. Yang, S. et al. One-Pot Synthesis of Graphene-Supported Monodisperse Pd
Nanoparticles as Catalyst for Formic Acid Electro-oxidation. Sci. Rep. 4, 4501
(2015).
62. Ngene, P. et al. Polymer-Induced Surface Modifications of Pd-based Thin Films
Leading to Improved Kinetics in Hydrogen Sensing and Energy Storage
Applications. Angew. Chemie Int. Ed. 53, 12081–12085 (2014).
63. Balmes, O. et al. in Physical Chemistry Chemical Physics 14, 4796 (The Royal
Society of Chemistry, 2012).
64. Seriani, N., Mittendorfer, F. & Kresse, G. Carbon in palladium catalysts: A
metastable carbide. J. Chem. Phys. 132, 024711 (2010).
65. Ignatieva, L. N., Tsvetnikov, A. K., Livshits, A. N., Saldin, V. I. & Buznik, V. M.

50
Spectroscopic Study of Modified Polytetrafluoroethylene. J. Struct. Chem. 43, 64–
68 (2002).
66. Cioffi, N. et al. Deposition and analytical characterization of fluoropolymer thin
films modified by palladium nanoparticles. Thin Solid Films 449, 25–33 (2004).
67. Dong, W. & Hafner, J. H 2 dissociative adsorption on Pd(111). Phys. Rev. B 56,
15396–15403 (1997).
68. Herron, J. A., Tonelli, S. & Mavrikakis, M. Atomic and molecular adsorption on
Pd(111). Surf. Sci. 606, 1670–1679 (2012).
69. Ishimoto, T. & Koyama, M. Theoretical study of tetrahedral site occupation by
hydrogen in Pd nanoparticles. J. Chem. Phys. 148, 034705 (2018).
70. Liu, X., Tian, D. & Meng, C. DFT study on the adsorption and dissociation of H2 on
Pdn (n = 4, 6, 13, 19, 55) clusters. J. Mol. Struct. 1080, 105–110 (2015).
71. Grönbeck, H. & Zhdanov, V. P. Effect of lattice strain on hydrogen diffusion in Pd:
A density functional theory study. Phys. Rev. B 84, 052301 (2011).
72. Dahl, S. et al. Role of Steps in N2 Activation on Ru(0001). Phys. Rev. Lett. 83,
1814–1817 (1999).
73. Hendriksen, B. L. M. et al. The role of steps in surface catalysis and reaction
oscillations. Nat. Chem. 2, 730–734 (2010).
74. Miller, M. M. & Lazarides, A. A. Sensitivity of Metal Nanoparticle Surface Plasmon
Resonance to the Dielectric Environment. J. Phys. Chem. B 109, 21556–21565
(2005).
75. Silkin, V. M., Díez Muĩo, R., Chernov, I. P., Chulkov, E. V & Echenique, P. M. Tuning
the plasmon energy of palladium-hydrogen systems by varying the hydrogen
concentration. J. Phys. Condens. Matter 24, 104021 (2012).
76. Antosiewicz, T. J., Apell, S. P., Zäch, M., Zorić, I. & Langhammer, C. Oscillatory
optical response of an amorphous two-dimensional array of gold nanoparticles.
Phys. Rev. Lett. 109, 247401 (2012).
77. Dahlin, A. B., Tegenfeldt, J. O. & Höök, F. Improving the instrumental resolution of
sensors based on localized surface plasmon resonance. Anal. Chem. 78, 4416–23
(2006).
78. Larche, F. C. in Advances in Phase Transitions 193–203 (Pergamon, 1988).
79. Landau, L. D., Lifshit︠︡s, E. M., Kosevich, A. M. & Pitaevskiĭ, L. P. Theory of elasticity.
(Butterworth-Heinemann, 1986).
80. Bergström, J. S. & Hilbert, L. B. A constitutive model for predicting the large
deformation thermomechanical behavior of fluoropolymers. Mech. Mater. 37,
899–913 (2005).
81. McCook, N. L. et al. Wear resistant solid lubricant coating made from PTFE and
epoxy. Tribol. Lett. 18, 119–124 (2005).
82. Vannice, M. A. Kinetics of Catalytic Reactions. (Springer US, 2005).
83. Gremaud, R. et al. Hydrogenography: An Optical Combinatorial Method To Find
New Light-Weight Hydrogen-Storage Materials. Adv. Mater. 19, 2813–2817
(2007).
84. Boelsma, C. et al. Hafnium—an optical hydrogen sensor spanning six orders in
pressure. Nat. Commun. 8, 15718 (2017).

51
85. Ngene, P., Longo, A., Mooij, L., Bras, W. & Dam, B. Metal-hydrogen systems with
an exceptionally large and tunable thermodynamic destabilization. Nat. Commun.
8, 1846 (2017).
86. Fredriksson, H. et al. Hole–Mask Colloidal Lithography. Adv. Mater. 19, 4297–
4302 (2007).
87. Nugroho, F. A. A., Darmadi, I., Zhdanov, V. P. & Langhammer, C. Universal Scaling
and Design Rules of Hydrogen-Induced Optical Properties in Pd and Pd-Alloy
Nanoparticles. ACS Nano 12, 9903–9912 (2018).

52

You might also like