harnessing plasma environments for ammonia catalysis
TRANSCRIPT
doi.org/10.26434/chemrxiv.13296956.v1
Harnessing Plasma Environments for Ammonia Catalysis: MechanisticInsights from Experiments and Large-Scale Ab-initio Molecular DynamicsSharma Yamijala, Giorgio Nava, Zulfikhar A. Ali, Davide Beretta, Bryan Wong, Lorenzo Mangolini
Submitted date: 28/11/2020 • Posted date: 01/12/2020Licence: CC BY-NC-ND 4.0Citation information: Yamijala, Sharma; Nava, Giorgio; Ali, Zulfikhar A.; Beretta, Davide; Wong, Bryan;Mangolini, Lorenzo (2020): Harnessing Plasma Environments for Ammonia Catalysis: Mechanistic Insightsfrom Experiments and Large-Scale Ab-initio Molecular Dynamics. ChemRxiv. Preprint.https://doi.org/10.26434/chemrxiv.13296956.v1
By combining experimental measurements with ab initio molecular dynamics simulations, we provide the firstmicroscopic description of the interaction between metal surfaces and a low-temperature nitrogen-hydrogenplasma. Our study focuses on the dissociation of hydrogen and nitrogen as the main activation route. We findthat ammonia forms via an Eley-Rideal mechanism where atomic nitrogen abstracts hydrogen from thecatalyst surface to form ammonia on an extremely short timescale (a few picoseconds). On copper, ammoniaformation occurs via the interaction between plasma-produced atomic nitrogen and the H-terminated surface.On platinum, however, we find that surface saturation with NH groups is necessary for ammonia production tooccur. Regardless of the metal surface, the reaction is limited by the mass transport of atomic nitrogen,consistent with the weak dependence on catalyst material that we observe and has been reported by severalother groups. This study represents a significant step towards achieving a mechanistic, microscopic-scaleunderstanding of catalytic processes activated in low-temperature plasma environments.
File list (2)
download fileview on ChemRxivSharma-Mangolini_revised_bmwong.pdf (1.03 MiB)
download fileview on ChemRxivSharma-Mangolini-SI.pdf (2.17 MiB)
1
Harnessing Plasma Environments for Ammonia
Catalysis: Mechanistic Insights from Experiments
and Large-Scale Ab-initio Molecular Dynamics
Sharma S. R. K. C. Yamijala1†, Giorgio Nava2†, Zulfikhar A. Ali1, Davide Beretta3, Bryan M. Wong1,4*,
Lorenzo Mangolini2,4*
1Department of Chemical & Environmental Engineering, University of California-Riverside, Riverside,
CA USA
2Department of Mechanical Engineering, University of California-Riverside, Riverside, CA USA
3Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland
4Materials Science and Engineering Program, University of California-Riverside, Riverside, CA USA
†The first two authors equally contributed to this work
Corresponding authors: Bryan M. Wong ([email protected]) and Lorenzo Mangolini,
By combining experimental measurements with ab initio molecular dynamics simulations, we
provide the first microscopic description of the interaction between metal surfaces and a low-temperature
nitrogen-hydrogen plasma. Our study focuses on the dissociation of hydrogen and nitrogen as the main
activation route. We find that ammonia forms via an Eley-Rideal mechanism where atomic nitrogen
abstracts hydrogen from the catalyst surface to form ammonia on an extremely short timescale (a few
picoseconds). On copper, ammonia formation occurs via the interaction between plasma-produced atomic
nitrogen and the H-terminated surface. On platinum, however, we find that surface saturation with NH
groups is necessary for ammonia production to occur. Regardless of the metal surface, the reaction is limited
by the mass transport of atomic nitrogen, consistent with the weak dependence on catalyst material that we
observe and has been reported by several other groups. This study represents a significant step towards
achieving a mechanistic, microscopic-scale understanding of catalytic processes activated in low-
temperature plasma environments.
2
3
The use of low-temperature plasmas to drive heterogeneous chemical reactions continues to attract
significant interest by both the plasma and catalysis communities.1-14 Since these processes are electrically
driven, they are compatible with a decentralized, renewable energy-driven chemical manufacturing
infrastructure. To this end, many chemical reactions are currently being investigated by the plasma
community, such as methane reforming,1-3 CO2 hydrogenation,10-11 and others.7, 15 Among these reactions,
the case of plasma-driven nitrogen fixation is particularly interesting because its thermal counterpart – the
Haber-Bosch process,16-19 which is the standard for the industrial production of ammonia – is particularly
energy-intensive. Reasonably high degrees of nitrogen conversion to ammonia have been reported for a
broad range of reactor designs, varying from atmospheric pressure dielectric barrier discharges (DBDs) in
contact with a catalyst bed6, 20 to low-pressure plasmas driven by radio-frequency excitation.21-22 No external
heating is required in any of these cases, and good degrees of nitrogen fixation, even with the catalyst at
nominally room temperature, are observed. The exact mechanism leading to the reasonably efficient
fixation of nitrogen in these systems is still under intense debate. Mehta et al.6 attributed the reasonably
high yield of ammonia in a dielectric barrier discharge to the plasma-induced vibrational excitation of
nitrogen, effectively lowering the barrier for dissociation upon adsorption onto the catalyst surface. This
results in an increased activity for metals such as Co, i.e., catalysts that interact more weakly with nitrogen
(with a more positive nitrogen adsorption energy) compared to those optimized for the Haber-Bosch process
such as Ru. In addition, Mehta et al.6 predicted a strong dependence of reaction turnover frequency over
the catalyst material, varying by orders of magnitude. Iwamoto et al.23 compared the activity of different
materials by placing metallic wool within a tubular plasma reactor operated at atmospheric pressure, making
sure that the area exposed to the plasma is constant between the different materials. They found a very
similar ammonia yield (well within a factor of two) for metals such as Pt, Au, Ag, Cu, Pd, and Fe. In the
thermally activated case, the reactivity of these metals would differ by several orders of magnitude.
Iwamoto et al.23 attempted to gain a microscopic understanding of this behavior by performing atomistic
DFT calculations. They found a reasonable correlation between ammonia yield and the stability of surface
M3N groups, where M is the metal. Shah et al.21 impinged a low pressure (0.26 Torr) radio-frequency N2-
H2 plasma onto metallic meshes of Cu, Pd, Ag, and Au, and obtained effectively the same ammonia yield,
suggesting a weak dependence on catalyst material. The authors performed a detailed microkinetic analysis
for the case of Fe, supporting the hypothesis that nitrogen dissociation is the main activation mechanism in
the low pressure regime. Van Helden et al.24 generated a N2-H2 plasma via a high-temperature plasma torch,
followed by expansion of the ionized gas into a low-pressure chamber. The authors used emission
spectroscopy to determine the concentration of ammonia in the gas-phase and found the same value after
the stainless steel reactor walls were coated with a silicon nitride layer. The authors concluded that ammonia
was produced at the reactor walls via recombination of plasma-produced radicals, consistent with a lack of
4
material dependence. Finally, Hong et al.8 demonstrated a reasonably high yield of ammonia even when an
atmospheric pressure low-temperature plasma was in contact with a diamond-like carbon surface,
suggesting a reaction pathway that is drastically different from that of a thermally-activated process.
Overall, the scientific community agrees that excitation of nitrogen in the gas-phase is critical, with the
accepted dominant mechanism being vibrational excitation for atmospheric pressure processes because of
the reduced electric field strength and lower electron temperature.25 Nitrogen dissociation is generally
considered to be prevalent at lower pressure. Both Eley-Rideal and Langmuir-Hinshelwood mechanisms
are generally invoked to explain the formation of ammonia in plasma-assisted processes.21, 26
Still, significant gaps in knowledge exist with respect to the interaction between plasma-produced
species and catalyst surfaces. While reactor-level characterization of ammonia yield, in combination with
micro-kinetic models, can be useful from the point of view of process design and optimization, this strategy
is limited as it cannot provide a mechanistic, microscopic understanding of surface reaction pathways.
Common atomistic models such as DFT are limited to static, zero kelvin configurations, and therefore
cannot predict chemical configurations in a dynamic environment. In this manuscript, we address this
knowledge gap by combining a reactor-level characterization of a low-pressure RF plasma operating in a
hydrogen-nitrogen mixture complemented with large-scale Born-Oppenheimer Molecular Dynamics
(BOMD) simulations. We focus on a low-pressure, RF-sustained plasma since these conditions are
conducive to a spatially and temporally uniform plasma, as opposed to dielectric barrier discharges, which
are composed of transient and highly localized plasma filaments, complicating the plasma-surface
interaction and the interpretation of the measurements. To complement our experimental measurements,
we utilize BOMD calculations since this technique provides a direct, dynamical, and unbiased approach for
predicting reaction mechanisms on surfaces.27-29 Since BOMD includes both dynamic and steric effects, it
naturally gives unbiased results compared to static electronic structure methods such as conventional
density functional theory (DFT). Also, since the potential energy surfaces (which dictate the forces on the
nuclei) are computed in an ab initio manner during a BOMD simulation, it is superior to classical molecular
dynamics and empirical force-field techniques. Its capability of handling radicals such as atomic nitrogen
and hydrogen (via spin-polarization quantum mechanical effects) makes it essential for investigating
plasma-produced ammonia processes in low-pressure reactors, where the dissociation of N2 and H2 is the
dominant excitation mechanism.
The measurements are performed on a RF-driven plasma operating in a 1:3 N2:H2 mixture at a
pressure of 1 Torr. The system is comprised of a quartz cylindrical plasma reactor with three ring electrodes,
two of which are grounded while the center electrode is connected to a radiofrequency (RF) power source.
The tube diameter is 1”, with each electrode being 1” wide. The plasma is ignited in a mixture of nitrogen
5
flowing at 20 standard cubic centimeters per minute (sccm) and hydrogen flowing at 60 sccm. A schematic
and a photograph of the reactor are shown in Figure 1a. A Faraday cage is used to prevent interference with
the surrounding electronics. Part of the exhaust stream is sampled through a 50 m orifice expanded into a
low-pressure chamber onto which a residual gas analyzer (RGA) is mounted. The transmission coefficient
of the orifice is dependent on the atomic mass unit (amu) of the sampled species. We have carefully
calibrated the response by flowing known amounts of N2, H2, and NH3 though the reactor. Details of the
calibration procedure are given in the Supplementary Information document (see Figure S1).
The catalyst consists of an aluminum foil substrate (1”6”) onto which 100 nm metallic films are
applied via DC magnetron sputtering. The foil is then rolled into a cylinder to line a portion of the inside
wall of the reactor. This approach ensures that the catalyst-active area is identical when different materials
are compared. It is important to note that the position of the catalyst in the tube reactor is critical. We
observed a negligible ammonia yield when the catalyst is placed in the initial section of the reactor, i.e.,
under the grounded electrode placed immediately upstream of the RF biased electrode. This low yield is an
expected result since the ammonia produced at the entrance of the reactor rapidly dissociates as it travels
through the plasma. We have further verified this hypothesis using a control experiment, where we added
a known amount of ammonia to the gas stream without placing any catalyst (metal foil) into the reactor.
For this experimental setup, plasma ignition results in the decomposition of >90% of the ammonia supplied
at the inlet, strongly supporting our hypothesis. To ensure higher ammonia yields, in all our experiments,
we placed the catalyst under the downstream grounded electrode, i.e., closer to the exit of the plasma
volume. The residence time through the plasma volume at 1 Torr, based on the flow velocity, is 30 msec.
The yield of ammonia is calculated by monitoring the partial pressures of nitrogen at amu 28 and
ammonia at amu 17. The degree of molecular fragmentation induced by the electron gun of the RGA is
accounted for in the calculations. In this manuscript, we report the nitrogen fixation efficiency , i.e., the
atomic fraction of nitrogen carried by ammonia at the reactor exhaust, defined as:
𝜂 =𝑃𝑁𝐻3
2∙𝑃𝑁2+ 𝑃𝑁𝐻3
, (1)
where PNH3 and PN2 are the partial pressures of ammonia and nitrogen downstream of the plasma volume,
respectively. In Figure 1b, we show the measured fixation efficiency at a power of 150W for different
metals. In the absence of any catalyst (labeled as “No Catalyst” in Figure 1b) we reproducibly observed a
yield of roughly 0.5%, which can be attributed to ammonia production at the quartz surface. The Al and Ti
surfaces did not show any increase in the yield compared to the “no catalyst” case. We strongly suspect that
this inactivity is due to their native oxide layer, which makes their catalytic activity comparable to that of
quartz. Among the other metals, Mo yielded roughly 0.75% of ammonia, while Ag, Cu, Pd, Co, and Ni
6
behaved very similarly, yielding ~1.25% of ammonia. Pt and Fe produced around 1.6% of ammonia. As a
whole we observed a weak dependence on catalyst material, consistent with other reports in the literature.
Also, we found a roughly linear dependence of the ammonia yield on the RF input power (as shown in
Figure S2). It is important to note that using a too high input power (above 150W) led to sputtering of the
metal and contamination of the reactor tube. We, therefore, do not report those results. The error bars were
determined by repeating the measurement with the Al foil five times and picking the maximum spread
among those as the measured error.
We performed optical emission spectroscopy (OES) to estimate the density of atomic species in the
plasma since atomic nitrogen is the main activated species in the low-pressure plasma regime. A small flow
Figure 1. (a) Schematic and photograph of the flow-through RF-driven
plasma reactor for the fixation of N2 and production of ammonia. (b)
Histogram comparing the nitrogen fixation efficiency, as defined in
equation (1), for a power input of 150W and a pressure of 1 Torr, for
different catalytic surfaces. (c) Example of an optical emission spectrum
observed for a nitrogen-hydrogen mixture, with the small addition (5%)
of argon as an actinomer. For clarity, the amplitude of the H peak has
been reduced by a factor of 5. The right panel also shows the ratio between
atomic hydrogen and nitrogen densities, as obtained by the OES data,
following the procedure outline in the text.
7
of argon has been added to the nitrogen-hydrogen mixture as an actinomer,30 and the density of atomic
nitrogen and atomic hydrogen have been obtained by the emission line intensity ratios using the procedure
outlined by Tatarova et al.31 An example of the spectrum used for the calculation is shown in Figure 1c. We
have used the following emission lines for our calculation: the H hydrogen line at 656.2 nm (transition
3d2Dj to 2p2P0), the nitrogen line at 744.2 nm (transition 3p4S0 to 3s4P), and the 750.4 nm line for Ar (2p1
to 1s2 transition). The following relation was used to calculate the densities of atomic nitrogen and
hydrogen, respectively:
𝑛𝑁
𝑛𝐴𝑟=
𝐼(744)
𝐼(750)∙
744
750∙ (
𝑋𝐴𝑟
𝑋𝑁) ∙
𝐴750/(∑ 𝐴𝑗,𝐴𝑟+𝐾𝑀𝐴𝑟∙𝑛𝑀)
𝐴744/(∑ 𝐴𝑗,𝑁+𝐾𝑀𝑁∙𝑛𝑀)
, (2)
𝑛𝐻
𝑛𝐴𝑟=
𝐼(656)
𝐼(750)∙
656
750∙ (
𝑋𝐴𝑟
𝑋𝐻) ∙
𝐴750/(∑ 𝐴𝑗,𝐴𝑟+𝐾𝑀𝐴𝑟∙𝑛𝑀)
𝐴656/ ∑ 𝐴𝑗,𝐻, (3)
where nN and nH are the densities of atomic nitrogen and hydrogen, and nAr is the density of argon as
determined by ideal gas law. I(744), I(750), and I(656) are the emission intensities from the atomic species
at those wavelengths corresponding to nitrogen, argon, and hydrogen, respectively. A744, A750, and A656 are
the transition probabilities for the corresponding lines (numerical values are provided in the Supplementary
Information). Aj,Ar, Aj,N, and Aj,H are the transition probabilities for all other optically allowed transitions
from the same excited states for the corresponding atomic species, which need to be accounted for to obtain
the correct branching ratio. The transition probabilities are obtained from the NIST atomic spectra
database.32 KArM and KN
M are quenching rates that account for the non-radiative relaxation of the excited
state of Ar and N respectively due to collision with partner M, with M = Ar, N2, H2. We use the same values
utilized by Tatarova et al.31, with the numerical values provided in the Supplementary Information. nM is
the density of the collision partner, as determined by ideal gas law. XAr, XN, and XH are the electron impact-
induced excitation rates to the selected excited states of argon, nitrogen, and hydrogen. The excitation rates
are obtained by averaging the electron-induced excitation cross sections33 over the electron energy
distribution function (EEDF). While Tatarova et al.31 assumed a Maxwell-Boltzmann EEDF, here we use
Bolsig+,34,35 a commonly utilized freeware software for the solution of the Boltzmann transport equation,
to obtain the EEDF. It is well-known that the EEDF strongly depends on the electron temperature Te. Since
Te is an unknown variable, we calculated the EEDF over a wide range of Te (between 2 and 6 eV), and the
corresponding EEDFs are shown in Figure S3. The atomic nitrogen and hydrogen densities obtained
following this procedure are shown in Figure S4 as a function of input RF power and electron temperature.
For a power of 150W and Te = 6 eV, we find that the atomic hydrogen density nH is 41014 cm-3, and the
atomic nitrogen density nN is roughly one order of magnitude smaller, 41013 cm-3. In Figure 1c, we also
show the ratio between atomic hydrogen and nitrogen densities. For all conditions assumed here, the atomic
8
hydrogen density exceeds that of atomic nitrogen by at least a factor of 5, with a weak dependence on the
electron temperature. These density values are consistent with those from other groups.36-39 Tatarova et al.31
report a nitrogen degree of dissociation as high as 4%, although they investigate a more energy-dense,
microwave sustained plasma.
Overall, our measurements confirm the weak dependence of ammonia yield on the catalyst
material. Surprisingly, we observed effectively the same yield on Pt and Fe, despite Pt being an inactive
catalyst under thermally-activated conditions because of its more positive nitrogen adsorption energy of 0.6
eV6 (i.e., nitrogen does not absorb on Pt as strongly as on Fe). Similarly, we observed a good yield even for
Cu (1.1% compared to 1.6% for Pt at 150 W RF power), despite DFT calculations predicting a very high
nitrogen adsorption energy of 3.4 eV for this metal40 (i.e., nitrogen does not absorb on Cu). In addition, the
OES characterization suggests that for all the conditions under consideration here, the atomic hydrogen
density exceeds that of atomic nitrogen (see Figure 1c). Therefore the flux of atomic hydrogen to the catalyst
surface greatly exceeds that of atomic nitrogen, which implies a higher probability of formation of a
hydrogen-terminated surface under plasma exposure. As such, it is necessary to investigate the interaction
between the plasma-produced species and the hydrogen-terminated surfaces to advance our understanding
of plasma-driven heterogeneous catalysis. To this end, we performed large-scale BOMD simulations of
atomic nitrogen impinging onto hydrogen-terminated catalyst surfaces. Given the fact that these simulations
are computationally demanding, we focused on two metals, namely, Cu and Pt. Both of these metals adsorb
nitrogen more weakly compared to the optimal choice for a thermally driven process (they are on the right
branch of the well-known volcano plot for ammonia synthesis) and show good ammonia yield for a plasma-
driven reaction, consistent with other reports in the literature. They also have greatly different nitrogen
adsorption energies, with the case of Cu (+3.4 eV)40 being significantly more positive than that of Pt (+0.6
eV)6. Our own calculations of the nitrogen binding energies for Cu(111) and Pt(111) surfaces find values
of -3.681 eV and -5.115 eV, respectively (see table S1), consistent with the more positive adsorption energy
for copper. It is, therefore, interesting to investigate the microscopic behavior of these metals, given their
similar yields of ammonia under plasma exposure.
We performed spin-polarized, ab initio Born-Oppenheimer molecular dynamics (BOMD)
simulations with three-layer 6×6-Cu(111) and Pt(111) surfaces, where the atomic positions of the bottom
layer (36 metal atoms) are fixed. We passivated all the metal atoms in the top-layer with either hydrogen
or nitrogen. First, we optimized these passivated structures using DFT and conducted NVT simulations to
check their stability at room temperature. We used the Nosé-Hoover thermostat of the chain length three to
maintain the temperature at 300 K and subsequently simulated the impingement of nitrogen atoms onto H-
terminated surfaces. The initial positions and velocities for all the NVE simulations were obtained from the
9
NVT simulations, and a 0.5 fs time step was used to integrate the equations of motion. For the impinging
nitrogen atom, the initial velocity was set such that it has 0.1 eV kinetic energy towards the surface (we
tried with other incidence energies in the range of 0.03-0.3 eV, and all of them gave similar results (see
Figure S10)), and its initial position was kept 5 Å above the passivated surface. To avoid any spurious
interactions between the periodic images, 15 Å of vacuum space was used. On the hydrogen-terminated
surfaces, we impinged an N atom at 9 different angles, from 0 to 80 degrees at 10-degree intervals. The
results for the case of Cu are presented first in Figure 2. Figure 2a shows the hydrogen-terminated Cu
surface at 0 K and after equilibration at 300 K, confirming the stability of the hydrogen termination at room
temperature. A similar simulation with a nitrogen-terminated Cu surface at 300 K (Figure 2b) resulted in
the facile desorption of the nitrogen molecules. This observation is consistent with the high adsorption
energy of nitrogen on copper. In Figure 2c, we show the NVE simulation results of the atomic nitrogen
10
impinging onto an H-terminated Cu surface with six consecutive snapshots spanning an interval of 1 ps.
The corresponding energy profiles (see Figure S7) and movies are available in the Supplementary
Information. These NVE results clearly suggest that the incoming nitrogen atom readily abstracts hydrogen
atoms from the surface and rapidly desorbs as ammonia (within 1 ps). The time-dependent bond distance
and angles of the formed NH3 molecules are shown in Figure S11. The outcome of NVE simulations
performed at different angles of incidence is shown in Figure S5. For all the incident angles, we consistently
Figure 2. Optimized geometry (left) and the geometry after a 1 ps long NVT
simulation (right) of an (a) H-terminated and (b) N-terminated Cu(111) surface.
At room temperature, only hydrogen termination was found to be stable
whereas nitrogen termination led to the desorption of N2 molecules from the
Cu surface. (c) Atomic geometries at various time steps during an NVE
simulation of an atomic nitrogen impinging onto a H-terminated Cu(111)
surface at normal incidence, with a kinetic energy of 0.1 eV. In the initial frame
of this NVE simulation, the N atom was placed at 5 Å above the surface. The
outcomes for eight other angles of incidence are given in Figure S5. For all
conditions, either NH3 or NH2 is the reaction product. The three layers of the
copper surface are shown in different shades of orange/yellow color.
11
observed the formation and desorption of NH3 or NH2 (accompanied by the desorption of H2 in a few cases)
from the Cu surface.
In stark contrast, the BOMD simulation results on the Pt surface are drastically different from the
Cu surface, as summarized in Figure 3. Figure 3a and 3b show the results of NVT simulations on Pt for
determining the stability of H-termination and N-termination, respectively. As expected, hydrogen is stable
at room temperature on Pt. Nitrogen is also stable on Pt at room temperature, showing a very different
behavior compared to Cu (Figure 2b), which instead showed rapid desorption of nitrogen molecules from
the surface. This is consistent with the less positive adsorption energy of nitrogen on Pt compared to Cu.
Moreover, the NVE simulations for Pt (Figure 3c) have a very different outcome compared to the case for
Cu (Figure 2c). Impinging atomic nitrogen does not lead to the formation of NH2 or NH3 molecules; rather,
it sticks onto the Pt surface without formation of an N-H bond, at least within the time window considered
Figure 3. Optimized geometry (left) and the geometry after a 2 ps long NVT
simulation (right) of an (a) H-terminated and (b) N-terminated Pt(111) surface.
At room temperature, both hydrogen and nitrogen terminations were found to
be stable for a Pt surface. (c) Atomic geometries at various time steps during
an NVE simulation of an atomic nitrogen impinging onto a H-terminated
Pt(111) surface at a 20° angle to the surface normal, with a kinetic energy of
0.1 eV. In the initial frame of this NVE simulation, the N atom was placed at 5
Å above the surface. The outcomes for eight other angles of incidence are given
in Figure S6. For most of the cases, we observe a strong adsorption of the
nitrogen atom onto the Pt surface, without any immediate N-H bond formation.
The three layers of the platinum surface are shown in different shades of
cyan/blue color.
12
here (<2 ps). The corresponding energy profiles (Figure S8) and movies are available in the Supplementary
Information. The outcomes of the same NVE simulation performed for different nitrogen impinging angles
are shown in Figure S6. For the majority of cases, we observe strong adsorption of nitrogen onto the Pt
surface without immediate formation of an N-H bond. However, in one case, we observed the formation of
an N-H surface group, and in another case, the formation of NH3. So while these calculations do not rule
out the formation of ammonia as a result of N-impingement onto H-terminated Pt, they do suggest that the
likelihood of such reaction pathway is considerably smaller for Pt compared to Cu. This lower probability
is a consequence of the stronger interaction of atomic nitrogen with Pt compared to Cu (see Figures 2b and
3b, and table S1).
The above results suggest that the same microscopic mechanism, i.e., a direct abstraction of
hydrogen atoms (from an H-terminated catalyst surface) by the impinging nitrogen atom, cannot explain
the similar yields of ammonia for hydrogen-terminated Cu and Pt surfaces. To solve this discrepancy, we
first assumed that the continuous impingement of atomic hydrogen and nitrogen onto the Pt surface
ultimately leads to the formation of an NH-termination. This assumption is justified by the significantly
larger density of atomic hydrogen compared to atomic nitrogen (Figure 1c), with the corresponding thermal
flux of atomic hydrogen exceeding that of atomic nitrogen by an even larger factor. Next, we performed
BOMD calculations on this NH-terminated Pt, as shown in Figure 4. The NVT simulations in Figure 4a
suggest that the NH termination is not entirely stable at room temperature, with some NH and NH2
fragments desorbing from the surface. Most importantly, we observe that the direct formation of NH3 from
the NH-terminated Pt is possible under exposure to either atomic hydrogen or nitrogen within the
computational time scales accessible to the BOMD technique. In Figure 4b, we show results from the case
of atomic hydrogen impinging onto the NH-terminated surface. The formation and desorption of NH3 are
clearly observed (see Figures S12 and S13 for the time-dependent bond-distance and angles of the formed
NH3 molecules). In Figure 4c, we show the case of atomic nitrogen impinging onto the NH-terminated Pt
surface. Within 1 ps, we observe the formation and desorption of ammonia, similar to what is observed for
the case of H-terminated Cu. The corresponding energy profiles (Figure S9) and movies are available in
the Supplementary Information.
Overall, these results suggest that there are important differences in the mechanism of ammonia
formation for the case of Pt and Cu. For Cu, nitrogen does not adsorb onto the surface, and ammonia is
formed via an Eley-Rideal mechanism leading to the direct abstraction of surface-bound hydrogen by the
13
impinging atomic nitrogen. For Pt, impinging atomic nitrogen adsorbs onto the surface without the direct
abstraction of hydrogen. The saturation of the surface with NH groups is necessary for activating the direct
abstraction mechanism. Interestingly, NH3 formation has been observed for both cases of atomic hydrogen
and atomic nitrogen impinging onto the NH-terminated Pt surface. The computational results also explain
the similar yield of ammonia that we measure for Cu and Pt. Our OES measurements strongly suggest that
Figure 4. (a) Optimized geometry (left) and configurations after an 8-ps long
NVT simulation (right) of an NH-terminated Pt(111) surface. At room
temperature, the NH termination was found to be reasonably stable for a Pt
surface with the desorption of a few NH and NH2 groups during the simulation.
(b) The outcomes of NVE simulations for an atomic nitrogen impinging onto a
NH-terminated Pt(111) surface at 0° and 30° angles to the surface normal, with
a kinetic energy of 0.1 eV. (c) Atomic geometries at various time steps during
an NVE simulation of an atomic nitrogen impinging onto a NH-terminated
Pt(111) surface at normal incidence, with a kinetic energy of 0.1 eV. For both
H and N impingement, we observed the formation and desorption of ammonia
on the NH-terminated Pt surface.
14
the flux of atomic hydrogen to the surface largely exceeds that of atomic nitrogen. Assuming a temperature
of 300 K and using the densities reported earlier (nH~41014 cm-3 and nN~41013 cm-3), we find that the
atomic hydrogen flux to the surface is ~1020 cm-2s-1 while the atomic nitrogen flux is ~2.71018 cm-2s-1, i.e.,
the atomic hydrogen flux is ~40 times larger than that of atomic nitrogen. From the 66 cross-sectional area
of the cells used in our BOMD calculations, these fluxes imply that, on average, one hydrogen radical
impinges onto that surface every ~350 ns, and one nitrogen radical impinges every ~13 s. These timescales
are considerably longer than the ones over which we have observed relevant reactions in the BOMD
calculations (ps). This result is consistent with a mass-transport limited mechanism in which the formation
of ammonia is not limited by surface reaction kinetics (which occur on a ps time scale) but rather by the
flux of ammonia precursor to the surface (which occurs on a hundred-ns time scale).
While the difference in ammonia yields between Cu and Pt is relatively small, it is reproducible
with Pt producing more ammonia than Cu. While our work focuses on the gas-phase dissociation of nitrogen
as the dominant activation pathway in low-pressure plasma reactors, this does not rule out that vibrational
excitation of molecular nitrogen is also present and contributing to the total ammonia yield. Some
dissociative adsorption of nitrogen may occur on metal surfaces even at room temperature, consistent with
the proposal from Mehta et al.6 This process is likely to be more relevant to Pt compared to Cu because of
the negligible interaction between Cu and nitrogen (Cu has a much more positive nitrogen adsorption
energy of +3.4 eV40 compared to that of Pt, which is equal to +0.6 eV6). This explanation is consistent with
the larger ammonia yield that we reproducibly measure for the case of Pt.
To summarize, we have characterized a low-pressure RF-driven plasma reactor for the production
of ammonia from a nitrogen-hydrogen mixture. Consistent with previous reports, we observe a weak
dependence on catalyst material. Careful characterization suggests that under these conditions, the flux of
atomic hydrogen to the catalyst surface largely exceeds that of atomic nitrogen. We have performed large-
scale BOMD simulations focusing on two metals, Cu and Pt, which exhibit a significantly different
interaction with atomic nitrogen. For the case of Cu, ammonia is formed via direct abstraction of hydrogen
from the surface when it is exposed to plasma-produced atomic nitrogen. The case of Pt is more complex.
Direct abstraction of ammonia under either atomic nitrogen or hydrogen fluxes can occur rapidly for Pt, but
only after the surface is saturated with NH groups. In both scenarios, plasma-produced atomic nitrogen is
converted into ammonia on a ps time scale. While our observations are limited to short reaction time scales,
which do not capture slower pathways such as Langmuir-Hinshelwood processes, they are consistent with
a mass-transport-limited regime and the weak dependence on catalyst material, which we observe and has
been reported by several other groups. This work represents the first application of BOMD to the case of
plasma catalysis and suggests that this technique is valuable for providing microscopic and mechanistic
15
insight into the plasma-surface interaction, with potential future application to other relevant plasma-
activated chemical reactions.
ACKNOWLEDGMENTS
The authors would like to acknowledge the support of the U.S. Army Research Office under Grant No.
W911NF-17-1-0340.
REFERENCES
(1) Nozaki, T.; Muto, N.; Kado, S.; Okazaki, K., Dissociation of vibrationally excited methane on Ni
catalyst: Part 1. Application to methane steam reforming, Catal. Today 2004, 89 (1–2), 57-65.
(2) Nozaki, T.; Muto, N.; Kadio, S.; Okazaki, K., Dissociation of vibrationally excited methane on Ni
catalyst: Part 2. Process diagnostics by emission spectroscopy, Catal. Today 2004, 89 (1–2), 67-74.
(3) Kim, J.; Abbott, M. S.; Go, D. B.; Hicks, J. C., Enhancing C–H Bond Activation of Methane via
Temperature-Controlled, Catalyst–Plasma Interactions, ACS Energy Letters 2016, 1 (1), 94-99.
(4) Stere, C. E.; Anderson, J. A.; Chansai, S.; Delgado, J. J.; Goguet, A.; Graham, W. G.; Hardacre,
C.; Taylor, S. F. R.; Tu, X.; Wang, Z.; Yang, H., Non-Thermal Plasma Activation of Gold-Based Catalysts
for Low-Temperature Water–Gas Shift Catalysis, Angewandte Chemie International Edition 2017, 56 (20),
5579-5583.
(5) Xu, S.; Chansai, S.; Stere, C.; Inceesungvorn, B.; Goguet, A.; Wangkawong, K.; Taylor, S. F. R.;
Al-Janabi, N.; Hardacre, C.; Martin, P. A.; Fan, X., Sustaining metal–organic frameworks for water–gas
shift catalysis by non-thermal plasma, Nature Catalysis 2019, 2 (2), 142-148.
(6) Mehta, P.; Barboun, P.; Herrera, F. A.; Kim, J.; Rumbach, P.; Go, D. B.; Hicks, J. C.; Schneider,
W. F., Overcoming ammonia synthesis scaling relations with plasma-enabled catalysis, Nature Catalysis
2018, 1 (4), 269-275.
(7) Mehta, P.; Barboun, P.; Go, D. B.; Hicks, J. C.; Schneider, W. F., Catalysis Enabled by Plasma
Activation of Strong Chemical Bonds: A Review, ACS Energy Letters 2019, 4 (5), 1115-1133.
(8) Hong, J.; Aramesh, M.; Shimoni, O.; Seo, D. H.; Yick, S.; Greig, A.; Charles, C.; Prawer, S.;
Murphy, A. B., Plasma Catalytic Synthesis of Ammonia Using Functionalized-Carbon Coatings in an
Atmospheric-Pressure Non-equilibrium Discharge, Plasma Chem. Plasma Process. 2016, 36 (4), 917-940.
(9) Hong, J.; Prawer, S.; Murphy, A. B., Plasma Catalysis as an Alternative Route for Ammonia
Production: Status, Mechanisms, and Prospects for Progress, ACS Sustainable Chemistry & Engineering
2018, 6 (1), 15-31.
(10) Mu, Y.; Xu, S.; Shao, Y.; Chen, H.; Hardacre, C.; Fan, X., Kinetic Study of Nonthermal Plasma
Activated Catalytic CO2 Hydrogenation over Ni Supported on Silica Catalyst, Ind. Eng. Chem. Res. 2020.
(11) Xu, S.; Chansai, S.; Shao, Y.; Xu, S.; Wang, Y.-c.; Haigh, S.; Mu, Y.; Jiao, Y.; Stere, C. E.; Chen,
H.; Fan, X.; Hardacre, C., Mechanistic study of non-thermal plasma assisted CO2 hydrogenation over Ru
supported on MgAl layered double hydroxide, Applied Catalysis B: Environmental 2020, 268, 118752.
(12) Neyts, E. C.; Bogaerts, A., Understanding plasma catalysis through modelling and simulation-a
review, J. Phys. D-Appl. Phys. 2014, 47 (22).
(13) Somers, W.; Bogaerts, A.; van Duin, A. C. T.; Neyts, E. C., Interactions of plasma species on nickel
catalysts: A reactive molecular dynamics study on the influence of temperature and surface structure,
Applied Catalysis B: Environmental 2014, 154–155, 1-8.
(14) Neyts, E. C.; Ostrikov, K.; Sunkara, M. K.; Bogaerts, A., Plasma Catalysis: Synergistic Effects at
the Nanoscale, Chemical Reviews 2015, 115 (24), 13408-13446.
(15) Whitehead, J. C., Plasma–catalysis: the known knowns, the known unknowns and the unknown
unknowns, Journal of Physics D: Applied Physics 2016, 49 (24), 243001.
16
(16) Galloway, J. N., The global nitrogen cycle: changes and consequences, Environmental Pollution
1998, 102 (1, Supplement 1), 15-24.
(17) Baltrusaitis, J. Sustainable Ammonia Production; American Chemical Society: 2017/11/06, 2017;
pp 9527-9527.
(18) Cherkasov, N.; Ibhadon, A. O.; Fitzpatrick, P., A review of the existing and alternative methods
for greener nitrogen fixation, Chemical Engineering and Processing: Process Intensification 2015, 90, 24-
33.
(19) Schrock, R. R., Reduction of dinitrogen, Proceedings of the National Academy of Sciences 2006,
103 (46), 17087-17087.
(20) Barboun, P.; Mehta, P.; Herrera, F. A.; Go, D. B.; Schneider, W. F.; Hicks, J. C., Distinguishing
Plasma Contributions to Catalyst Performance in Plasma-Assisted Ammonia Synthesis, ACS Sustainable
Chemistry & Engineering 2019, 7 (9), 8621-8630.
(21) Shah, J.; Wang, W.; Bogaerts, A.; Carreon, M. L., Ammonia Synthesis by Radio Frequency Plasma
Catalysis: Revealing the Underlying Mechanisms, ACS Applied Energy Materials 2018, 1 (9), 4824-4839.
(22) Ben Yaala, M.; Saeedi, A.; Scherrer, D.-F.; Moser, L.; Steiner, R.; Zutter, M.; Oberkofler, M.; De
Temmerman, G.; Marot, L.; Meyer, E., Plasma-assisted catalytic formation of ammonia in N2–H2 plasma
on a tungsten surface, Phys. Chem. Chem. Phys. 2019, 21 (30), 16623-16633.
(23) Iwamoto, M.; Akiyama, M.; Aihara, K.; Deguchi, T., Ammonia Synthesis on Wool-Like Au, Pt,
Pd, Ag, or Cu Electrode Catalysts in Nonthermal Atmospheric-Pressure Plasma of N2 and H2, ACS
Catalysis 2017, 7 (10), 6924-6929.
(24) Helden, J. H. v.; Wagemans, W.; Yagci, G.; Zijlmans, R. A. B.; Schram, D. C.; Engeln, R.;
Lombardi, G.; Stancu, G. D.; Röpcke, J., Detailed study of the plasma-activated catalytic generation of
ammonia in N2-H2 plasmas, J. Appl. Phys. 2007, 101 (4), 043305.
(25) van ‘t Veer, K.; Reniers, F.; Bogaerts, A., Zero-dimensional modeling of unpacked and packed bed
dielectric barrier discharges: the role of vibrational kinetics in ammonia synthesis, Plasma Sources Science
and Technology 2020, 29 (4), 045020.
(26) Carrasco, E.; Jiménez-Redondo, M.; Tanarro, I.; Herrero, V. J., Neutral and ion chemistry in low
pressure dc plasmas of H2/N2 mixtures: routes for the efficient production of NH3 and NH4+, Phys. Chem.
Chem. Phys. 2011, 13 (43), 19561-19572.
(27) Yamijala, S. S. R. K. C.; Ali, Z. A.; Wong, B. M., Acceleration vs Accuracy: Influence of Basis
Set Quality on the Mechanism and Dynamics Predicted by Ab Initio Molecular Dynamics, The Journal of
Physical Chemistry C 2019, 123 (41), 25113-25120.
(28) Tsona, N. T.; Bork, N.; Loukonen, V.; Vehkamäki, H., A Closure Study of the Reaction between
Sulfur Dioxide and the Sulfate Radical Ion from First-Principles Molecular Dynamics Simulations, The
Journal of Physical Chemistry A 2016, 120 (7), 1046-1050.
(29) Lin, W.; Stocker, K. M.; Schatz, G. C., Mechanisms of Hydrogen-Assisted CO2 Reduction on
Nickel, J. Am. Chem. Soc. 2017, 139 (13), 4663-4666.
(30) Coburn, J. W.; Chen, M., Optical emission spectroscopy of reactive plasmas: a method for
correlating emission intensities to reactive particle density, J. Appl. Phys. 1980, 51 (6), 3134-3136.
(31) Tatarova, E.; Dias, F. M.; Gordiets, B.; Ferreira, C. M., Molecular dissociation in N2–H2
microwave discharges, Plasma Sources Science and Technology 2004, 14 (1), 19-31.
(32) database, https://www.nist.gov/pml/atomic-spectra-database
(33) database, www.lxcat.net
(34) Hagelaar, G. J. M.; Pitchford, L. C., Solving the Boltzmann equation to obtain electron transport
coefficients and rate coefficients for fluid models, Plasma Sources Science and Technology 2005, 14 (4),
722.
(35) Version 12/2017 available at http://www.bolsig.laplace.univ-tlse.fr/index.html
(36) Amorim, J.; Baravia, G.; Ricard, A., Production of N, H, and NH active species in N2-H2 dc
flowing discharges, Plasma Chem. Plasma Process. 1995, 15 (4), 721-731.
17
(37) Gordiets, B.; Ferreira, C. M.; Pinheiro, M. J.; Ricard, A., Self-consistent kinetic model of low-
pressure - flowing discharges: I. Volume processes, Plasma Sources Science and Technology 1998, 7 (3),
363-378.
(38) Gordiets, B.; Ferreira, C. M.; Pinheiro, M. J.; Ricard, A., Self-consistent kinetic model of low-
pressure - flowing discharges: II. Surface processes and densities of N, H, species, Plasma Sources Science
and Technology 1998, 7 (3), 379-388.
(39) Ricard, A.; Henriques, J.; Cousty, S.; Villeger, S.; Amorim, J., Determination of N-, H- and O-
Atom Densities in N2–H2 and in N2–O2 Gas Mixtures by Optical Actinometry in Flowing Microwave
Discharges and by NO Titration in Post-Discharges, Plasma Processes and Polymers 2007, 4 (S1), S965-
S968.
(40) Pang, X.-Y.; Xue, L.-Q.; Wang, G.-C., Adsorption of Atoms on Cu Surfaces: A Density Functional
Theory Study, Langmuir 2007, 23 (9), 4910-4917.
download fileview on ChemRxivSharma-Mangolini_revised_bmwong.pdf (1.03 MiB)
1
Harnessing Plasma Environments for Ammonia
Catalysis: Mechanistic Insights from Experiments
and Large-Scale Ab-initio Molecular Dynamics
Supplementary Information
Sharma S. R. K. C. Yamijala1†, Giorgio Nava2†, Bryan M. Wong1,3*, Lorenzo Mangolini2,3*
Department of Chemical and Environmental Engineering, University of California-Riverside, Riverside,
CA USA
Sharma S. R. K. C. Yamijala1†, Giorgio Nava2†, Zulfikhar A. Ali1, Davide Beretta3, Bryan M. Wong1,4*,
Lorenzo Mangolini2,4*
1Department of Chemical & Environmental Engineering, University of California-Riverside, Riverside,
CA USA
2Department of Mechanical Engineering, University of California-Riverside, Riverside, CA USA
3Swiss Federal Laboratories for Materials Science and Technology, Dübendorf, Switzerland
4Materials Science and Engineering Program, University of California-Riverside, Riverside, CA USA
2
Mass spectrometer calibration procedure
The gas exiting the reaction volume is sampled through a 50 μm orifice and into a turbo-pumped chamber
onto which a residual gas analyzer (RGA) is mounted. The amu-dependent transmission coefficient of the
orifice and response of the (RGA) are calibrated by flowing known amounts of nitrogen and ammonia
through the system over a range of flows and pressures (in the 1-3 Torr range, i.e., under the same conditions
in which the plasma is run). This procedure allows building a correction factor that is then applied to the
data measured during the plasma catalysis experiments. Figure S1 below shows the uncorrected ratio in the
partial pressures of NH3 and N2 (y-axis) as a function of the ratio between the flow rates (x-axis). The
accuracy of the mass flow controllers has been verified by carefully reporting the time necessary for the
system to go from base pressure to a target pressure. For a given constant volume, this is inversely
proportional to the volumetric flow rates supplied by different gas lines (N2, NH3, or H2).
The contribution to the ammonia signal at amu = 17 from the fragmentation of water is accounted for by
first measuring the water signal without plasma and by recording the ratio between the signal at amu 18 =
18 and amu = 17. This allows the removal of any contribution from water to the amu = 17 signal when the
plasma is ignited and the reactor is producing ammonia. This is summarized in the equation below.
𝑃𝑁𝐻3= 𝑎𝑚𝑢17,𝑝𝑙𝑎𝑠𝑚𝑎−𝑜𝑛 − 𝑎𝑚𝑢18,𝑝𝑙𝑎𝑠𝑚𝑎−𝑜𝑛 ∗ (
𝑎𝑚𝑢17,𝑝𝑙𝑎𝑠𝑚𝑎−𝑜𝑓𝑓
𝑎𝑚𝑢18,𝑝𝑙𝑎𝑠𝑚𝑎−𝑜𝑓𝑓)
Figure S1. Calibration factor for the response of the sampling
system (RGA + extraction orifice).
3
Ammonia yield at varying process conditions
Figure S2 shows the ammonia yield for different catalyst surfaces at varying input powers and pressures.
In general, we observe a lower yield as pressure increases. Higher input power increases yield because of
the higher density of plasma-produced radicals.
Values for the calculation of the atomic hydrogen and nitrogen densities
The following was used for calculating nH and nN (see equations 2 and 3 in the manuscript):
Constant Value Source
A744 1.06⋅107 s-1 [1][2]
A750 4.72⋅107 s-1 [1][2]
A656 4.41⋅107 s-1 [1]
KArN2 3.2⋅10-11 cm3s-1 [2]
KArH2 2.7⋅10-11 cm3s-1 [2]
KArAr 1.6⋅10-11 cm3s-1 [2]
KNN2 3.8⋅10-10 cm3s-1 [2]
KNH2 5⋅10-10 cm3s-1 [2]
KNAr 3.8⋅10-10 cm3s-1 [2]
Figure S2. Ammonia yield at varying power and reactor pressures.
4
5
Figure S3 below shows the electron energy distribution functions (EEDFs) calculated using Bolsig+. These
are needed to calculate the excitation rates for H, N, and Ar from the ground state to the specified excited
state, under the assumption that the dominant excitation mechanism is via electron collision.
Figure S4 shows the calculated densities of atomic hydrogen and nitrogen as a function of RF power and
assumed electron temperatures. Atomic hydrogen densities exceed atomic nitrogen density by roughly one
order of magnitude.
Figure S3. Variation in the EEDF with reduced electric field (E/N) and
corresponding electron temperature Te.
Figure S4. Variation of nH (atomic hydrogen density) and nN (atomic nitrogen density) as a function
of RF input power, assuming different values of electron temperature Te for the EEDF calculation.
6
Computational details
In both the BOMD and electronic structure calculations, we used the PBE exchange-correlation
functional [3] and molecularly optimized double-zeta quality (DZVP) basis-sets [4] as implemented in the
CP2K software package. [5] For the auxiliary plane-wave (PW) basis, we used 600 Ry for the PW energy
cutoff and 60 Ry for the reference grid cutoff. Goedecker-Teter-Hutter (GTH) pseudopotentials, [6, 7]
which are compatible with the employed basis-sets, were used for all the elements. Dispersion interactions
were included using Grimme's D3-dispersion correction. [8] Due to the large cell sizes, and longer
simulation duration (two weeks to run ~ 500 fs on 32 Haswell processors), all the BOMD calculations were
performed at the Γ-point of the Brillouin zone.
Figure S5. Trajectories of NVE simulations for an atomic nitrogen impinging onto a H-terminated
Cu(111) surface at varying incidence angles, with a kinetic energy of 0.1 eV. In the initial frame of
these NVE simulations, the N atom was first kept at 5 Å above the surface. Here, the angles from 0° to
80° are given with respect to the surface normal. Thus, the 0° simulation corresponds to a perpendicular
incidence, and in the 80° simulation, the nitrogen atom moves almost parallel to the surface. The three
layers of the copper surface are shown in different shades of orange/yellow color. For all conditions,
either NH3 or NH2 is the reaction product.
7
Figure S6. Trajectories of NVE simulations for an atomic nitrogen impinging onto a H-terminated
Pt(111) surface at varying incidence angles, with a kinetic energy of 0.1 eV. The three layers of the
platinum surface are shown in different shades of cyan/blue color. The hydrogen and nitrogen atoms
(topmost layer) are displayed in white and dark-blue colors, respectively. For only the 40° incidence
angle, we observed the formation of ammonia.
8
Figure S7: Energy profiles of the hydrogen-terminated Cu(111) surface when impinged with an N atom
at 0, 10, and 80° angles. For each incident angle, the entire system’s kinetic energy (KE), potential energy
(PE), and the total energy (TE) are shown as a function of time in the left panel. Since the absolute values
of both the PE and TE are roughly five orders of magnitude greater than the KE, we scaled them with
respect to the PE at the zeroth step, PE(0), to make the fluctuations in KE apparent. Due to the nature of
the NVE simulation, the TE of the system is roughly constant throughout the simulation duration. The
dip in the PE after the first few hundred femtoseconds of the simulation is due to the interaction between
the incoming nitrogen atom and the Cu-surface (and the H-atoms attached to it). The energy released due
to the formation of bonds (among the N, Cu, and H atoms) increases the surface temperature, which is
reflected as a rise in the KE of the system, which further helps the desorption of the newly formed
fragments (such as H2, NH, NH2, and NH3). In the middle (right) panel, the variation in the kinetic
energies of individual fragments are shown for the entire simulation time (the first 100 fs). As shown in
these panels, the KE of the impinging nitrogen atom increases as it approaches the surface due to the
strong attraction between the N and Cu/H atoms. After forming a bond with the hydrogen atoms, the KE
9
of the formed fragment fluctuates (due to the competition between the Cu and N atoms to form bonds
with the H atoms). Furthermore, as expected, these plots also show that the time taken for the N atom to
be in direct contact with H atoms (which roughly equals the first peak in the KE profile of the impinging
N atom) on the Cu surface increases with the incident angle (~ 200 fs for 0 and 10° incidences and ~ 400
fs for 80 degree incidence). Here, we re-emphasize that 0° corresponds to a normal incidence. In certain
cases, we had issues with the SCF convergence during the first few steps (< 20) of the NVE simulation,
and we omitted those steps for clarity.
Figure S8: Energy profiles of the hydrogen-terminated Pt(111) surface when impinged with an N atom
at 0 and 20° angles. See Figure S7 for the description of left, middle, and right panels. Due to the strong
adsorption of the N atom onto the Pt(111) surface, the fluctuations in the KE of the formed fragments
(i.e., NH) are lower compared to those exhibited by the same fragments on the Cu(111) surface.
10
Figure S9: Energy profiles of the NH-terminated Pt(111) surface when impinged with an H-atom at 0
and 30° angles. See Figure S7 for the description of left, middle, and right panels. For the 30° (0°)
incidence angle, the impinging H atom was part (not part) of the NH3 molecule that was formed during
the NVE simulation.
Table S1: Binding energies (in eV) of various atoms/fragments/molecules to the Cu(111) and Pt(111)
surfaces as obtained from our DFT simulations. The energies of the stable spin-configuration were always
used for the binding energy calculations. The binding energy of a fragment (BEFrag) towards a surface was
calculated using the formula, BEFrag = E(Surface+Frag) - E(Surface) - E(Frag), where E(Surface+Frag), E(Surface), and E(Frag)
are the DFT-energies of the entire system (when the fragment is adsorbed on the surface), surface, and
fragment, respectively, calculated using the same cell parameters. Except for the molecular nitrogen and
hydrogen, all the atoms/molecules/fragments bind strongly to the Pt surface compared to the Cu surface.
Due to its very strong binding energy (~1.6 eV) to the Pt(111) surface, an NH3 molecule would need a large
amount of kinetic energy to desorb from the Pt surface even after its formation (unlike the Cu(111) surface).
Note that the situation can be different when the Pt surface is terminated with other atoms/groups.
11
Atom/Molecule/Fragment Binding Energy to Cu(111)
(eV)
Binding Energy to Pt(111)
(eV)
N -3.681 -5.114
NH -3.513 -4.638
NH2 -2.310 -3.093
NH3 -0.698 -1.660
H -3.432 -3.891
N2 -0.136 0.082
H2 -0.133 -0.079
12
Figure S10: Trajectories of NVE simulations for atomic nitrogen impinging onto a H-terminated Cu(111)
surface at a 30° incidence angle with a kinetic energy of (a) 0.03 eV, (b) 0.1 eV, (c) 0.3 eV, and (d) 1 eV.
In the initial frame of these NVE simulations, the N atom was first kept at 5 Å above the surface. The
three layers of the copper surface are shown in different shades of orange/yellow color. For all cases, we
observed the formation of the NH bond. Except for the 1eV incidence, the desorption of an H2 molecule
was also observed in all cases.
13
Table S2: N-H bond distances and H-N-H bond angles of the ammonia molecule formed on the Cu and Pt
surfaces in the last frames of the NVE simulation. For comparison, the gas-phase bond distances and angles
of an optimized ammonia molecule (computed at the same level of theory) are also given. The time-
dependent plots of the bond distances and angles were shown in figures S11-S13.
System Bond distances (Å) Angles (degrees)
NH3 molecule (gas phase) 1.02, 1.02, 1.02 106.27, 106.27, 106.27
Cu(111)+36H+N (0° incidence) 1.08, 1.11, 1.25 86.39, 90.20, 120.73
Cu(111)+36H+N (20° incidence) 0.98, 1.15, 1.49 98.10, 99.0, 88.77
Pt(111)+36NH+H (0° incidence) 0.99, 1.08, 1.13 95.82, 93.04, 100.32
Pt(111)+36NH+H (30° incidence) 1.09, 0.96, 0.97 98.45, 104.56, 120.09
Pt(111)+36NH+N (0° incidence) 1.12, 1.04, 0.98 120.79, 86.27, 115.86
14
Figure S11: Time-dependent N-H bond distances and H-N-H bond angles of the ammonia molecule that
was formed during an NVE simulation of atomic nitrogen colliding with the H-terminated Cu(111)
surface at different angles. For comparison, the computed N-H bond distances and H-N-H angles of an
optimized ammonia molecule (optimized at the same level of theory) in the gas phase are 1.02 Å and
106.27°, respectively. As shown in the above plots, the time-dependent values of the bond lengths and
bond angles oscillate near the optimized values (1.02 Å and 106.27°), suggesting the complete formation
of the ammonia molecule.
15
Figure S12: Time-dependent N-H bond distances and H-N-H bond angles of the ammonia molecule that
was formed during an NVE simulation of atomic hydrogen colliding with the NH-terminated Pt(111)
surface at different angles. As shown in the above plots, the time-dependent values of the bond lengths
and bond angles oscillate near the optimized values (1.02 Å and 106.27°), suggesting the complete
formation of the ammonia molecule.
16
Figure S13: Time-dependent N-H bond distances and H-N-H bond angles of the ammonia molecule that
was formed during an NVE simulation of atomic nitrogen colliding with the NH-terminated Pt(111)
surface at 0° incidence. As shown in the above plots, the time-dependent values of the bond lengths and
bond angles oscillate near the optimized values (1.02 Å and 106.27°), suggesting the complete formation
of the ammonia molecule.
[1] database, https://www.nist.gov/pml/atomic-spectra-database
[2] Tatarova, E.; Dias, F. M.; Gordiets, B.; Ferreira, C. M., Molecular dissociation in N2–H2 microwave
discharges. Plasma Sources Science and Technology 2004, 14 (1), 19-31.
[3] Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized Gradient Approximation Made Simple. Physical
Review Letters vol. 77 3865–3868 (1996).
[4] VandeVondele, J.; Hutter, J. Gaussian Basis Sets for Accurate Calculations on Molecular Systems in
Gas and Condensed Phases. J. Chem. Phys. 2007, 127, No. 114105.
[5] VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Quickstep: Fast
and Accurate Density Functional Calculations Using a Mixed Gaussian and Plane Waves Approach.
Comput. Phys. Commun. 2005, 167, 103−128.
17
[6] Goedecker, S.; Teter, M.; Hutter, J. Separable Dual-Space Gaussian Pseudopotentials. Phys. Rev. B:
Condens. Matter Mater. Phys. 1996, 54, 1703−1710.
[7] Krack, M. Pseudopotentials for H to Kr Optimized for Gradient-Corrected Exchange-Correlation
Functionals. Theor. Chem. Acc. 2005, 114, 145−152.
[8] Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping Function in Dispersion Corrected Density
Functional Theory. J. Comput. Chem. 2011, 32, 1456−1465.
download fileview on ChemRxivSharma-Mangolini-SI.pdf (2.17 MiB)