doi.org/10.26434/chemrxiv.11768718.v1

Plasma-Catalytic Ammonia Synthesis Beyond the Equilibrium LimitPrateek Mehta, Patrick M. Barboun, Yannick Engelmann, David B. Go, Annemie Bogaerts, William F.Schneider, Jason C. Hicks

Submitted date: 30/01/2020 • Posted date: 31/01/2020Licence: CC BY-NC-ND 4.0Citation information: Mehta, Prateek; Barboun, Patrick M.; Engelmann, Yannick; Go, David B.; Bogaerts,Annemie; Schneider, William F.; et al. (2020): Plasma-Catalytic Ammonia Synthesis Beyond the EquilibriumLimit. ChemRxiv. Preprint. https://doi.org/10.26434/chemrxiv.11768718.v1

We explore the consequences of non-thermal plasma activation on product yields in catalytic ammoniasynthesis, a reaction that is equilibrium-limited at elevated temperatures. We employ a minimal microkineticmodel that incorporates the influence of plasma activation on N2 dissociation rates to predict NH3 yields intoand across the equilibrium-limited regime. NH3 yields are predicted to exceed bulk thermodynamic equilibriumlimits on materials that are thermal-rate-limited by N2 dissociation. In all cases, yields revert to bulk equilibriumat temperatures at which thermal reaction rates exceed plasma-activated ones. Beyond-equilibrium NH3yields are observed in a packed bed dielectric-barrier-discharge reactor and exhibit sensitivity to catalyticmaterial choice in a way consistent with model predictions. The approach and results highlight the opportunityto exploit synergies between non-thermal plasmas and catalysts to affect transformations at conditionsinaccessible through thermal routes.

File list (2)

download fileview on ChemRxivChemrxiv-manuscript.pdf (504.40 KiB)

download fileview on ChemRxivsupporting-info.pdf (396.20 KiB)

Plasma-Catalytic Ammonia Synthesis Beyond the Equilibrium

Limit

Prateek Mehta,1 Patrick M. Barboun,1 Yannick Engelmann,2 David B.

Go,1, 3 Annemie Bogaerts,2, ∗ William F. Schneider,1, † and Jason C. Hicks1, ‡

1Department of Chemical and Biomolecular Engineering,

University of Notre Dame, Notre Dame, Indiana 46556, United States

2Department of Chemistry, Antwerp University,

Campus Drie Eiken, Universiteitsplein 1, 2610 Wilrijk

3Department of Aerospace and Mechanical Engineering,

University of Notre Dame, Notre Dame, Indiana 46556, United States

(Dated: January 29, 2020)

Abstract

We explore the consequences of non-thermal plasma activation on product yields in catalytic

ammonia synthesis, a reaction that is equilibrium-limited at elevated temperatures. We employ a

minimal microkinetic model that incorporates the influence of plasma activation on N2 dissociation

rates to predict NH3 yields into and across the equilibrium-limited regime. NH3 yields are predicted

to exceed bulk thermodynamic equilibrium limits on materials that are thermal-rate-limited by N2

dissociation. In all cases, yields revert to bulk equilibrium at temperatures at which thermal

reaction rates exceed plasma-activated ones. Beyond-equilibrium NH3 yields are observed in a

packed bed dielectric-barrier-discharge reactor and exhibit sensitivity to catalytic material choice

in a way consistent with model predictions. The approach and results highlight the opportunity to

exploit synergies between non-thermal plasmas and catalysts to affect transformations at conditions

inaccessible through thermal routes.

∗ annemie.bogaerts@uantwerpen.be† wschneider@nd.edu‡ jhicks3@nd.edu

1

INTRODUCTION

A fundamental goal in catalysis is to obtain high yields of desired products at high pro-

duction rates. In thermally driven catalytic systems, the reaction conditions under which

this goal can be accomplished are constrained by the kinetic activity of the catalyst and the

thermodynamics of the reaction. For many important reactions, these constraints can only

be satisfied at severe conditions, particularly if they involve the transformation of unreac-

tive molecules. As an example, consider the ammonia synthesis reaction from dinitrogen

and hydrogen, N2 + 3 H2 −−⇀↽−− 2 NH3. This reaction is exergonic at near-ambient conditions

(∆G◦(298 K) = −32.8 kJ mol−1), so high yields are thermodynamically possible, but typical

heterogeneous catalysts are inactive near ambient temperature [1, 2]. Increasing the tem-

perature into the regime in which catalytic turnover rates become appreciable also pushes

the equilibrium of the exothermic reaction towards reactants, so that elevated pressures

and recycle are necessary to achieve practically useful yields. Typical industrial-scale NH3

synthesis processes thus operate at temperatures in the range of 700–800 K and pressures of

100–200 atm [1, 3, 4].

Recent reports have shown that ammonia synthesis can be performed at milder tempera-

tures and pressures (∼473 K, 1 atm) when catalysts are operated alongside non-thermal plas-

mas [5–19]. Such plasmas are characterized by a non-thermal distribution of energy between

different degrees of freedom, such that high populations of reactive species (e.g. vibrationally

or electronically excited molecules, as well as radicals and ions) are accessible without signif-

icantly increasing the bulk gas temperature [20–23]. In particular, non-thermally activated

N2 species can experience lower dissociation barriers on a catalyst surface [24–29].

In previous work, we considered the consequences of non-thermal N2 excitation on ammo-

nia synthesis rates using a microkinetic model [16]. This model augmented well-established

thermal catalytic ammonia synthesis steps with the dissociation of plasma-activated N2 at

the catalyst surface to predict volcano-curves for NH3 synthesis rates under plasma stim-

ulation. The model assumed the presence of a steady-state, non-thermal distribution of

vibrationally excited molecules, parametrized on the N2 vibrational temperature extracted

from optical emission spectroscopic measurements of a dielectric barrier discharge (DBD)

plasma [16, 17]. Model predictions indicated that (1) non-thermal excitation of N2 lifts the

volcano, such that NH3 synthesis rates per reaction site at mild conditions may approach

2

those at thermal Haber-Bosch conditions and (2) the optimal catalytic material shifts from

Fe and Ru towards intrinsically less reactive materials, like Co or Ni. Additionally, the

model suggested that terrace sites on metal nanoparticles, which are inactive in thermal

catalysis, may also become active in a plasma environment. These model predictions were

qualitatively consistent with plasma catalytic kinetic experiments performed in a differential

flow DBD reactor, which showed elevated site-time yields of NH3 at atmospheric pressure

and 438 K, as well as a shift in optimal catalyst material [16, 18].

Site-normalized rates are intrinsically difficult to compare across catalysts because of their

sensitivity to the identification and enumeration of reaction sites. Yields relative to bulk

equilibrium, in contrast, are readily and unambiguously observable. A potential implication

of non-thermal plasma-assisted reaction acceleration is access to product yields that exceed

limits based on apparent bulk thermal equilibrium [23, 30, 31]. This “beyond equilibrium”

phenomenon has been observed in plasma-enhanced catalysis of the endothermic dry re-

forming of CH4 with CO2. In this case, product yields are observed to exceed equilibrium

yields at low bulk gas temperatures in the presence of plasma with or without catalysts

[23, 32–45]. The contributions of the catalyst and plasma to observed product yields in

the apparent equilibrium-limited regime are unclear. Product yields are typically observed

to increase with increasing plasma power [33, 36, 38–43, 46], but are insensitive to [32, 36]

or even decrease with increasing bulk gas temperatures [37]. Similarly, the introduction

of catalyst packing into the reactor is reported sometimes to enhance [38, 40, 43, 46–48]

and sometimes to diminish overall yields [33, 36, 38, 39, 44, 45, 47, 49] relative to plasma

alone. Additionally, apparent beyond-equilibrium effects on product yields for endothermic

reactions like dry reforming are difficult to disentangle from local heating (hot spots) of the

catalyst surface or reactor walls by the plasma [21, 50].

To address this gap in understanding of the coupling of catalyst and plasma on observed

yields in the apparent equilibrium-limited regime, here we propose an extension to the pre-

viously reported plasma catalysis microkinetic model [16] to predict trends in product yields

and compare them to laboratory non-thermal plasma catalytic experiments. In order to

make the model tangible and to avoid the complications of interpreting experimental results

for an endothermic reaction, we develop the model and perform experiments to test predic-

tions in the context of exothermic NH3 synthesis. We replace the steady-state approximation

on plasma-generated non-thermal N2 vibrational distributions with an explicit kinetic treat-

3

ment of the rate of plasma-stimulated generation of excited species. We exercise the model

across a range of conditions and assumed catalyst types to highlight conditions under which

conversions can be expected to exceed apparent equilibrium and to explain the material

choice on these excursions. We show that beyond equilibrium behavior is expected to be

most pronounced on materials for which N2 dissociation is strongly rate-limiting (i.e. mate-

rials that bind N weakly) and at temperatures at which thermal rates do not out-compete

the rates of the non-thermally excited processes. Plasma-catalytic experiments conducted in

a DBD reactor display trends consistent with model predictions. NH3 yields exceeding the

bulk equilibrium limit are observed and are greater on materials with weaker N binding en-

ergies (e.g. Pt) that are inactive for thermal ammonia synthesis. Moreover, the experiments

highlight the potential to tune performance by independent manipulation of the non-thermal

and thermal rates—yields can be increased further past equilibrium by increasing the plasma

power, but re-equilibrate at high bulk temperatures where the thermal reaction pathways

become dominant. The work highlights how non-thermal plasma catalysis offers different

performance tradeoffs compared to conventional thermal catalysis, and offers further sup-

port for the contribution of plasma-induced promotion of N2 dissociation to the observed

plasma influence on catalytic NH3 synthesis rates.

RESULTS

Kinetic model

Figure 1 shows a condensed scheme for the plasma-catalytic ammonia synthesis reaction.

Blue arrows indicate the conventional thermal pathway for ammonia synthesis, which in-

cludes the dissociative chemisorption of N2 onto the catalyst surface followed by its sequential

reaction with adsorbed hydrogens to form NH3.

N2 + 2 ∗ −−⇀↽−− 2 N∗ (1)

N∗ +3

2H2 −−⇀↽−− NH3 + ∗ (2)

The hydrogenation of N* typically occurs through a series of elementary steps involving

the chemisorption of H2 followed by the reaction of the adsorbed H∗ with NHx∗ species

4

FIG. 1. Reaction scheme for plasma-catalytic ammonia synthesis. The thermal reaction

steps are shown with blue arrows. Homogeneous plasma phase reactions are shown in orange. The

plasma-catalytic adsorption of activated nitrogen is shown in green.

(where x = 0, 1, 2). For convenience and with no loss of generality, these steps are lumped

together in reaction (2).

During plasma-driven operation, N2 can be promoted by electron impact to a more reac-

tive form, which we denote as N2(act).

N2 + e −−→ N2(act) + e (3)

Reaction (3) is agnostic to the type of excitation—N2(act) may be vibrationally or electron-

ically excited, ionized, or even dissociated into N radicals depending on the energy of the

colliding electron. In any of these forms, we consider activated N2(act) to chemisorb onto the

catalytic surface (shown by the green arrow in Figure 1) with a smaller activation energy

than that for reaction (1),

N2(act) + 2 ∗ −−→ 2 N∗ (4)

Assuming that N∗ primarily desorbs as ground state N2, reaction (4) becomes an irreversible

step. In addition to the heterogeneous reaction pathway through reaction (4) and reac-

tion (2), the reactive N2(act) species may participate in homogeneous plasma phase reactions

to form NH3, as observed in plasma-only experiments (see Refs. [16, 18, 23] and experiments

discussed below). We write this reaction in a lumped fashion as,

N2(act) +

3

2H2 −−→ NH3 (5)

5

The simple scheme depicted in Figure 1 allows us to develop a conceptual picture of

plasma-catalytic NH3 synthesis. Whether NH3 is formed via thermal or non-thermal path-

ways depends on the relative rates along these pathways. When the rate of reaction (1),

r1, exceeds that of reaction (3), r3, no benefits of plasma-stimulation of N2 are expected;

the reaction proceeds along the thermal pathway. Such a situation may arise at low specific

electrical energy input, where the rate of plasma-driven activation of N2 is low, or at high

bulk temperatures, where thermal N2 dissociation on the catalyst is rapid. Figure 1 neglects

the quenching of plasma-activated species, which becomes increasingly important with in-

creasing pressure and temperature. Similarly, non-thermal activation of N2 does not impact

catalytic reaction rates when the hydrogenation of surface adsorbed N* to NH3 (reaction (2),

r2) is slow, as discussed in previous work [16].

Plasma-driven reaction pathways become relevant when r1 << r3 and r2 is not rate-

limiting. If the generation of N2(act) is slow relative to subsequent heterogeneous or homo-

geneous pathways, i.e., reaction (3) is rate-limiting, the reaction kinetics of plasma-only and

plasma-catalytic operation become indistinguishable. If reaction (3) is not rate-limiting, the

reaction is kinetically controlled by either reaction (4) or reaction (5). If r5 is fast relative to

r4,2, NH3 is formed predominantly via homogeneous plasma-phase mechanisms, and benefits

of plasma-catalytic operation are not expected. This scenario may arise at low temperatures

where surfaces are saturated by adsorbates, and higher gas pressures and/or low catalyst

loadings, which favor reactive or dissipative homogeneous collisions of the excited states

before they arrive at a catalytic site. Based on the above reasoning, the catalyst plays a

kinetically significant role when the dissociation of N2(act) on the catalytic surface is the

rate-limiting step.

The electron impact activation of N2 (reaction (3)) is the primary driver of the non-

thermal chemistry. We can write the net rate of N2(act) generation as

r3 = kenepN2 (6)

Here, ne is the electron density, ke is the electron impact rate coefficient, and pN2 is the

pressure of N2. Both ne and ke depend on the characteristics of the non-thermal plasma

[20]. ke is primarily a function of the electron energy distribution function, or the effective

electron temperature of the plasma, rather than the bulk temperature.

The rates of the forward and reverse processes in Figure 1 are thus controlled by different

6

physical characteristics of the reacting system. The forward rate can be increased beyond

what could be possible thermally by manipulation of the characteristics of the non-thermal

stimulus, e.g. increasing the plasma power is expected to increase ke, ne, and influence

both the concentration of N2(act) and its effective activation energy for further reaction.

Additionally, because the forward and reverse processes are not connected by the principle

of detailed balance, product yields will not be constrained by the equilibrium limitations

characteristic of thermally driven reactions [51].

To understand the implications of the decoupling of the forward and reverse rates on

NH3 yields, we implement the simple reaction scheme given by reactions (1) to (4) in a

reactor model to trace reaction progress from the kinetic regime to the equilibrium-limited

regime. Reaction (5) is not included in the model for simplicity and to isolate the catalytic

consequences of non-thermal stimulation of N2 on conversions. We model the system as a

continuously stirred tank reactor (CSTR) operating at atmospheric pressure and constant

volume (1 mL). The reactants are introduced into the reactor in their stoichiometric ratio,

at a flow rate of 10 mL min−1. The overall volumetric flow rate decreases with reaction pro-

gression, because the reaction proceeds with a decrease in the number of moles. We consider

that the reactor is packed with catalyst having only one type of active site (metal steps)

with a total site density of 10−3 mol mL−1. Any potential transport effects are neglected.

For the thermal reaction steps (reactions (1) and (2)), we use Brønsted-Evans-Polyani

(BEP) relationships to calculate activation energies as a function of the atomic nitrogen

binding energy relative to N2 [16, 52, 53]. Because the hydrogenation steps are assumed

to be unaffected by the plasma, they are treated in a lumped fashion [54] (as written in

reaction (2)). In this formulation, volcano curves for the reaction rates (see inset, Figure 2)

still show the same general behavior as previous work [2, 16, 53, 55], although modeled rates

are overestimated because N* is the only species included in the surface site balance. Futher

model details are available in the Methods and Supplementary Methods sections.

Figure 2 depicts the steady state NH3 pressure in the reactor as a function of the bulk

temperature in the thermal catalysis model (i.e. excluding plasma reactions). Product yields

are shown for three catalysts selected to span different characteristic regions (i.e. strong, in-

termediate, and weak N binding energies) of typical volcano curves. The catalyst that binds

N with intermediate strength (EN = −0.6 eV), sits at the top of the volcano curve (see inset)

and provides the optimal balance between N2 activation and its subsequent hydrogenation.

7

FIG. 2. Modeled steady-state NH3 partial pressures in thermal catalysis as function

of bulk temperature. Three catalysts are shown, characterized by strong (EN = −1.2 eV),

intermediate (EN = −0.6 eV), and weak (EN = 0.0 eV) nitrogen binding energies. The equilibrium

pressure of NH3 is shown by dashed line. The inset shows the volcano curve for NH3 synthesis

calculated at 473 K, 1 atm, and 5% conversion of N2.

This material thus displays the highest NH3 yields of the three catalysts in Figure 2 at the

specified reactor conditions. Ammonia yields increase with temperature until approaching

the equilibrium limit, beyond which they follow the thermal equilibrium curve of diminishing

NH3 productivity with increasing temperature.

We next elaborate the CSTR model to incorporate activation of N2 by a non-thermal

plasma (i.e. reactions (3) and (4)). To simplify the analysis, we include only one activated

state in the model. Further, we assume that upon excitation to this state, the N2 dissociation

barrier is reduced by 1 eV, corresponding to roughly three N2 vibrational quanta. The extent

to which the barrier is reduced is arbitrary and does not alter the qualitative conclusions of

the work, as discussed in the Supplementary Information (Supplementary Figures 1 and 2).

We take the electron density (ne) and the electron impact excitation rate constant (ke,

reaction (3)) to be insensitive to the bulk temperature and gas composition. We scan

over three representative values of the product kene, corresponding to a low, intermediate,

and high rate of excitation. We neglect spatio-temporal variations in kene. However, we

note that in dielectric barrier discharge plasmas most commonly used for plasma catalysis,

8

electron impact reactions are only expected to occur during short-lived (∼ 10 ns) filamentary

microdischarges, followed by an afterglow period on the order of 1µs [56]. To roughly account

for the short-lived nature of the microdischarges and the much longer afterglow period, the

kene values used here are chosen to be smaller than or at the lower limit of the product of

the ke and ne typical of DBD microdischarges (see Computational Details).

Figure 3 reports the steady-state NH3 partial pressures as a function of the bulk tem-

perature as predicted by the plasma-on CSTR model for the same three catalysts as above.

The representative strong N-binding catalyst along the left leg of the thermal volcano curve

(EN = −1.2 eV) is presented in Figure 3a. Because this catalyst readily activates N2, the

reaction occurs via the thermal pathway, and no enhancements in NH3 production during

plasma-driven operation are observed. That is, the plasma-on product pressures correspond

with the thermal-only results. During plasma-catalytic operation, the surface remains cov-

ered with N up to temperatures as high as 600 K and substantial amounts of unreacted

N2(act) remain in the gas phase. These excited molecules could react along homogeneous

pathways or relax and equilibrate with the bulk gas. Coupling of plasma and catalyst has

no direct benefit in this case.

Figure 3b and Figure 3c plot NH3 partial pressures for a material with near-optimal N

binding energy for the thermal reaction (EN = −0.6 eV) and for a material selected from the

right leg of the thermal volcano curve (EA = 0.0 eV), respectively. In the kinetic regime, the

plasma-on models predict enhancements in NH3 produced over thermal catalysis for both

materials if the rate of plasma-activation of N2 is sufficiently high (kene ≥ 10−2 s−1). Re-

sults also indicate that plasma-on product pressures can exceed the bulk thermal equilibrium

limit on both of these materials. Further, the models indicate that this apparent beyond

equilibrium behavior is catalyst-dependent. For the optimal thermal catalyst, NH3 yields

exceeding equilibrium for the optimal thermal catalyst are observed at kene = 10−1 s−1, in

the temperature window of 475–600 K. In this temperature window, the amount of NH3

produced first increases, reaches a maximum, and then collapses back to the equilibrium

line. While the NH3 yields show similar features as a function of the bulk temperature

for the material that binds N more weakly than the optimal thermal catalyst, the degree

of excitation required to observe beyond equilibrium behavior is lower—NH3 pressures are

predicted to go beyond the equilibrium limit for kene ≥ 10−2 s−1. Furthermore, the tem-

perature range over which NH3 pressures exceed the bulk equilibrium limit is significantly

9

FIG. 3. Modeled steady-state NH3 partial pressures in plasma-enhanced catalytic

operation as a function of bulk temperature and rate of plasma-excitation. (a-c) The

subpanels show catalysts characterized by (a) strong (EN = −1.2 eV), (b) intermediate (EN =

−0.6 eV) and (c) weak (EN = 0.0 eV) nitrogen binding energies. For each catalyst steady state

thermal (plasma-off), and plasma-on (for three values of kene) partial pressures of NH3 are plotted.

Plasma-excitation is assumed to reduce the barrier for N2 dissociation by 1 eV.

wider for this material. In effect, plasma-activation of N2 shifts the effective NH3 synthesis

equilibrium [51], and the ability to sustain that shift is a function of the kinetic limits to

the reverse NH3 decomposition reaction.

We interpret the temperature-dependence of the NH3 partial pressures for the two mate-

rials above as follows. Plasma-activation of N2 facilitates its dissociative chemisorption on

both materials at low temperatures, but slow hydrogenation at these temperatures limits

its conversion to NH3. Surface sites become available at temperatures ∼400 K, resulting in

light-off in the NH3 pressures. Because the non-thermal pathway for N2 dissociation (re-

actions (3) and (4)) is decoupled from the thermal desorption of N∗ as N2 (see Figure 1),

product pressures can extend beyond the equilibrium limit. As the bulk temperature in-

10

creases further, the steady-state partial pressures of NH3 reach maxima and then decrease

as the thermal desorption of N∗ as N2 becomes significant. The temperature at which NH3

pressures converge on the bulk thermal equilibrium curve, the reaction occurs only through

the thermal route, i.e. the rate of thermal N2 dissociation exceeds the one along the plasma-

induced pathway, and the net NH3 yield is limited by the forward and backward rates of

NH3 synthesis and decomposition. The reversion to the equilibrium curve occurs at lower

temperature on materials closer to the thermal-only volcano maximum—the thermal rates

in both the forward and backward directions are highest for this material [57].

Experimental

We performed temperature-sweep experiments in a DBD reactor to investigate ammonia

synthesis behavior near the equilibrium regime. The reactor was configured in a fashion

similar to previous work [16–18, 23] (further discussed in Experimental Methods). We

introduced N2 and H2 into the reactor in a 2:1 ratio at a total flow rate of 40 SCCM. Ini-

tial experiments were performed to evaluate background contributions, plotted in Figure 4.

These background measurements were made in a reactor packed with bare support material

(100 mg γ-Al2O3), as opposed to an empty reactor, due to the potential of the dielectric pack-

ing to influence the discharge characteristics (plasma volume, discharge gap, etc.) [17, 23].

We note, however, that the empty reactor behaves similarly to the γ-Al2O3-packed reactor,

as shown in Supplementary Figure 3. The bulk reactor temperature was varied between

423–1123 K. Reaction performance was evaluated at plasma powers of 5, 10, and 15 W,

corresponding to specific energy inputs of 7.5, 15, and 22.5 kJ L−1. In all experiments, we

observed significant background NH3 production, consistent with previous literature reports

[23]. Furthermore, NH3 production increased with increasing reactor temperature, and yields

well beyond the equilibrium limit were observed at high bulk temperatures. In both the low-

and high-temperature regimes, NH3 yields increased with increasing plasma power. These

experiments demonstrate that the plasma power can be used to manipulate the rates of

the non-thermal reactions and drive the reaction further beyond the bulk thermal equilib-

rium limit. At all plasma powers, NH3 yields maximized at about 920 K, after which they

decreased.

Next, we performed similar temperature-sweep experiments with the reactor loaded with

11

FIG. 4. Plasma background ammonia yields in a DBD reactor packed with γ-Al2O3 as

a function of bulk reactor temperature. Three powers, including 5, 10, and 15 W, are shown.

Reaction conditions: total flow rate = 40 SSCM, inlet N2:H2 = 2:1. The dashed line shows the

bulk thermodynamic equilibrium conversion at the specified reaction conditions.

Ni and Pt catalysts supported on γ-Al2O3 (100 mg material, 5 wt% metal). We selected

these materials because both bind N more weakly than the optimal thermal catalyst, Ru

[2, 16, 53, 55, 58] (the N-metal bond strength follows the order Ru > Ni > Pt). Based on the

kinetic analysis above, beyond equilibrium behavior may be observable on these materials.

In Figure 5a, we plot plasma-catalytic NH3 yields on these two catalysts as a function of

bulk gas temperature at 10 W plasma power. The corresponding NH3 yields observed in the

plasma reactor packed with only Al2O3 are included for comparison. In the low temperature

kinetic regime, the metal-based catalysts behave in a similar fashion—NH3 yields increased

with temperature on both materials, and both materials exhibited enhanced NH3 production

relative to the background. However, in the equilibrium-limited regime, the behavior of the

two catalysts diverged significantly. On the Ni catalyst, NH3 yields fell below those of the

background experiment and collapsed onto the equilibrium curve. In contrast, temperature-

sweep profiles on the Pt catalyst resembled those of the background experiment—NH3 yields

continued to increase with temperature, significantly exceeding the bulk equilibrium limit,

before ultimately decreasing. Pt continued to enhance NH3 production over the background

at higher temperatures until about 1000 K, after which the yields converged.

12

FIG. 5. Ammonia synthesis and decomposition temperature-sweep experiments.

(a) Plasma-catalytic ammonia yields in a DBD reactor packed with Al2O3, Ni/γ-Al2O3, and

Pt/γ-Al2O3. Reaction conditions: total inlet flow rate = 40 mL min−1, N2:H2 = 2:1. (b)

NH3 conversion in thermal ammonia decomposition experiments over γ-Al2O3, Ni/γ-Al2O3, and

Pt/γ-Al2O3. Reaction conditions: Inlet flow rate = 40 mL min−1. Inlet gas composition: 0.5%

NH3, 9.5% He, and balance of 2:1 N2 and H2.

The hypothesized reaction scheme in Figure 1 and the subsequent kinetic analysis sug-

gests that the collapse of the NH3 yields in Figure 5a is linked to the reaction proceeding

in the reverse direction. To test the hypothesis, we performed thermal NH3 decomposition

experiments at feed compositions comparable to the outlet stream of the background ex-

periment at 900 K (the temperature at which NH3 yields were the highest). A gas mixture

of 0.5% NH3 and 9.5% He with a balance of N2 and H2 in a 2:1 ratio was fed into the

reactor. Figure 5b plots the amount of NH3 converted over the same three materials as

a function of temperature. Of the tested materials, Ni/Al2O3 is the most active for NH3

decomposition, lighting off at around 750 K. The decomposition curves for Pt/γ-Al2O3 and

metal-free γ-Al2O3 are nearly identical, lighting off at approximately 900 K. NH3 is not

expected to be stable at these high temperatures [59], and we attribute the similarity of the

13

light-off points for these two materials to gas-phase NH3 decomposition. The correspondence

between the temperatures that NH3 decomposition light off in Figure 5b and the collapse

in the NH3 synthesis yields in Figure 5a provides strong evidence of the decoupling of the

thermally and non-thermally driven processes. Plasma-stimulated reactions allow access to

yields of NH3 in excess of bulk thermal equilibrium limits. These excess ammonia yields

can remain kinetically trapped until the temperature at which thermal NH3 decomposition

becomes kinetically relevant. The experiments thus support the model inference that the

benefits of plasma-assisted NH3 synthesis accrue most greatly for materials on the right

(weaker binding) side of the thermal volcano plot, that they promote N∗ hydrogenation

steps, and that performance with increasing temperature is limited by the reverse thermal

NH3 decomposition reaction.

CONCLUSIONS

The promise of plasma catalysis lies in the potential to carry out chemical conver-

sions at conditions and with efficiencies that are inaccessible by thermal-only catalytic

routes. Plasma stimulation can in principle bypass kinetic bottlenecks in thermal-only cat-

alytic transformations, enhancing net reaction rates and conversions. Here we explore the

consequences of plasma-stimulation across conditions at which thermal reactions become

equilibrium-limited, in the context of ammonia synthesis, which under typical conditions is

kinetically limited by the activation of N2. We show that the presence of a non-thermally

activated reaction channel can disturb the coupling of the forward and reverse rates, i.e. the

forward and reverse rate processes balance each other at conditions different from those

specified by the bulk thermal equilibrium. As a consequence, reaction products can be

kinetically trapped in concentrations exceeding the equilibrium limit, as observed here for

plasma-stimulated NH3 synthesis. The contributions of the thermal and non-thermal chan-

nels may be independently manipulated by separate control of the bulk temperature and

the plasma power (or the specific energy input) respectively. We find that, with increasing

plasma power, ammonia yields extend further beyond the equilibrium limit, while yields

re-equilibrate at temperatures at which thermal chemistry becomes dominant.

The model and experimental results further indicate that the choice of catalytic material

has a controlling influence on the extent of excursion from thermal equilibrium. The great-

14

est departures are expected and observed on catalytic materials for which N2 activation is

strongly rate-limiting and thus are not optimal under equivalent thermal conditions. The

above findings align with and support prior results that show that the greatest potential

for kinetic enhancements are on catalysts on the far-right of the thermal ammonia synthesis

volcano curve. Because NH3 yields are measured with less ambiguity than turnover fre-

quencies, the directional agreement between the kinetic model and the experiments provide

strong evidence of a reaction mechanism that includes contributions from non-thermally

activated N2 dissociation. The overall energy efficiency of this process is thus a function of

the efficiency at which plasma power is directed into reactant activation.

We chose here NH3 synthesis as a model reaction because it is exothermic, any adventi-

tious heating provided by the plasma will diminish rather than enhance NH3 conversions,

and conclusions regarding the influence of plasma stimulation are thus not confounded by

questions of temperature control. Similar concepts are expected to apply to endothermic

reactions that are equilbrium-limited at low temperatures, such as methane dry reforming.

Here, parallel plasma-promoted reaction channels will cause conversions to exceed bulk equi-

librium at low temperatures and approach the thermal-equilbrium limit behavior at higher

temperatures, as in fact is observed in experiments [37].

METHODS

Computational details

The ammonia synthesis reaction was modeled with ∆H (298 K) = −91.8 kJ/mol. Stan-

dard entropies of the gas phase N2, H2, and NH3 were obtained from the National Institute

of Standards and Technology (NIST) Chemistry WebBook (http://webbook.nist.gov/

chemistry). The microkinetic model for the thermal reactions was developed using the

approach of Grabow [54]. The rate expressions used for the reaction steps and Brønsted-

Evans-Polyani (BEP) relationships used to calculate activation energies are provided in

the Supplementary Methods. To obtain an estimate of typical values of the electron den-

sity and the electron impact excitation rate coefficients, we solved the Boltzmann equa-

tion for a N2 plasma at a reduced electric field of 100 Td [60] using the online solver at

https://us.lxcat.net/. ke typically ranges between 10−14–10−9 cm3 s−1 depending on

15

the type of excitation (e.g. vibrational or electronic excitation, ionization, etc), while ne is

typically in the range of 1014–1015 cm−3. As noted in the main text, the electron impact

reactions are only expected to occur during the short-lived DBD microdischarges, which

exhibit spatial and temporal variations across the reactor. We ignore the temporal and

spatial variations of the DBD plasma and assume that the product kene can be treated as a

constant independent of the bulk gas temperature. We scan the CSTR model over a number

of representative values of kene, which are taken to be smaller than or at the lower limit of

the product of the ke and ne values noted above, such that the model can crudely account

for the long afterglow periods where the electron density is zero.

Experimental details

Pt and Ni supported on γ-Al2O3 (5 wt%) were obtained from Riogen Inc. The structural

properties of the catalysts were characterized prior to experiments. Surface areas were

evaluated with N2 physisorption using a Quantachrome 2200e at 77 K. The crystallinity

of the materials was evaluated using X-ray diffraction (XRD) over 30–70° 2θ using a D8

Advance diffractometer and referencing to a standard silicon crystal peak. Metal particle

sizes were determined by transmission electron microscopy imaging using a Jeol 2011 electron

microscope. All particle sizes were based on at least 300 particles. In addition to physical

properties, the number of accessible surface metal sites were evaluated both before and after

the plasma reaction via CO pulse chemisorption using a Micromeritics Chemisorb 2750.

Catalysts were pretreated in 20 SCCM (standard cubic centimeter per minute) H2 at 773 K

for 30 min followed by 30 min in 20 SCCM He at 773 K to remove chemisorbed H2 and

prevent potential polycarbonyl formation on the during CO adsorption. Characterization

information can be found in Supplementary Table 1.

The plasma reactor is the same as reported previously [18, 37, 61]. The reactor consists of

a quartz tube (5 mm internal diameter and 1 mm wall thickness) wrapped in a stainless steel

mesh (McMaster-Carr 200-1400) that covers 6 cm of the length of the tube and a 1.5 mm

tungsten rod inserted through the center of the tube. Within this setup, the discharge gap

is 1.75 mm, and the discharge volume is 1.07 cm3. The plasma was generated by applying

a high voltage from an AC power source (PVM500). Plasma power was monitored with an

oscilloscope (Tetronix TDS3012B) using the Lissajous curve method described previously

16

[18, 39]. Gas flow rates into the reactor were controlled using Aalborg GFC 17 mass flow

controllers such that the flow rate of each gas could be controlled independently. For re-

actants, we used 99.999% N2 (Airgas NI UHP300), 99.999% H2 (Airgas HY UHP300), and

10% NH3 in He (Airgas) with 99.997% He (Airgas HE HP300) occasionally used as a carrier

gas or to purge the system. Gas chromatography (GC) measurements were calibrated using

the 10% NH3 in He gas mixture. The reaction temperature was externally controlled by

a furnace over the temperature range of 423–1123 K. Temperature was measured using a

thermocouple downstream from the discharge zone that was calibrated ex situ. For each

experiment 100 mg of material was packed into the reactor. Each catalyst was pre-treated

at 773 K for 30 min in 20 SCCM of H2 before data collection. We waited 15 minutes upon

reaching each temperature point measured before doing GC injections in order to ensure

that the reactor was at steady state prior to measurement. Each data point is based off

of an average of three injections by the GC and the error bars shown correspond to the

standard deviation of these three measurements.

Data availability

An external Zenodo repository hosts code and raw data [62]. The kinetic model and

associated utility functions are included as Python files. Model input and calculated results

are included in json format. Raw data for the plasma catalysis experiments is included in

an Excel spreadsheet. Additional details and python scripts showing how these data files

were used to create all the figures in the paper are provided in the Supporting Data pdf file.

REFERENCES

[1] Schlogl, R. Ammonia synthesis. In Handbook of Heterogeneous Catalysis. 2nd Edition, 2501–

2575 (Wiley-VCH Verlag GmbH & Co. KGaA, 2008).

[2] Vojvodic, A. et al. Exploring the limits: A low-pressure, low-temperature haber-bosch process.

Chemical Physics Letters 598, 108–112 (2014). URL https://doi.org/10.1016/j.cplett.

2014.03.003.

[3] Jennings, J. Catalytic Ammonia Synthesis: Fundamentals and Practice (Springer Science &

Business Media, 1991).

17

[4] Nørskov, J., Chen, J., Miranda, R., Fitzsimmons, T. & Stack, R. Sustainable ammonia

synthesis–exploring the scientific challenges associated with discovering alternative, sustain-

able processes for ammonia production. Tech. Rep., US DOE Office of Science (2016).

[5] Mizushima, T., Matsumoto, K., ichi Sugoh, J., Ohkita, H. & Kakuta, N. Tubular membrane-

like catalyst for reactor with dielectric-barrier-discharge plasma and its performance in am-

monia synthesis. Applied Catalysis A: General 265, 53–59 (2004). URL http://dx.doi.org/

10.1016/j.apcata.2004.01.002.

[6] Mizushima, T., Matsumoto, K., Ohkita, H. & Kakuta, N. Catalytic effects of metal-loaded

membrane-like alumina tubes on ammonia synthesis in atmospheric pressure plasma by di-

electric barrier discharge. Plasma Chemistry and Plasma Processing 27, 1–11 (2006). URL

http://dx.doi.org/10.1007/s11090-006-9034-2.

[7] Kim, H.-H., Teramoto, Y., Ogata, A., Takagi, H. & Nanba, T. Atmospheric-pressure nonther-

mal plasma synthesis of ammonia over ruthenium catalysts. Plasma Processes and Polymers

14, 1600157 (2016). URL https://doi.org/10.1002/ppap.201600157.

[8] Xie, D. et al. Ammonia synthesis and by-product formation from H2O, H2 and N2 by dielectric

barrier discharge combined with an ru/Al2O3 catalyst. RSC Advances 6, 105338–105346

(2016). URL https://doi.org/10.1039/c6ra21351k.

[9] Peng, P. et al. Atmospheric pressure ammonia synthesis using non-thermal plasma assisted

catalysis. Plasma Chemistry and Plasma Processing 36, 1201–1210 (2016). URL https:

//doi.org/10.1007/s11090-016-9713-6.

[10] Hong, J. et al. Plasma catalytic synthesis of ammonia using functionalized-carbon coatings in

an atmospheric-pressure non-equilibrium discharge. Plasma Chemistry and Plasma Processing

36, 917–940 (2016). URL https://doi.org/10.1007/s11090-016-9711-8.

[11] Aihara, K. et al. Remarkable catalysis of a wool-like copper electrode for NH3 synthesis from

N2 and H2 in non-thermal atmospheric plasma. Chemical Communications 52, 13560–13563

(2016). URL https://doi.org/10.1039/c6cc06752b.

[12] Peng, P. et al. Ru-based multifunctional mesoporous catalyst for low-pressure and non-thermal

plasma synthesis of ammonia. International Journal of Hydrogen Energy 42, 19056–19066

(2017). URL https://doi.org/10.1016/j.ijhydene.2017.06.118.

[13] Akay, G. & Zhang, K. Process intensification in ammonia synthesis using novel coassembled

supported microporous catalysts promoted by nonthermal plasma. Industrial & Engineer-

18

ing Chemistry Research 56, 457–468 (2017). URL https://doi.org/10.1021/acs.iecr.

6b02053.

[14] Akay, G. Sustainable ammonia and advanced symbiotic fertilizer production using catalytic

multi-reaction-zone reactors with nonthermal plasma and simultaneous reactive separation.

ACS Sustainable Chemistry & Engineering 5, 11588–11606 (2017). URL https://doi.org/

10.1021/acssuschemeng.7b02962.

[15] 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 7, 6924–6929 (2017). URL https://doi.org/10.1021/acscatal.7b01624.

[16] Mehta, P. et al. Overcoming ammonia synthesis scaling relations with plasma-enabled

catalysis. Nature Catalysis 1, 269–275 (2018). URL https://doi.org/10.1038/

s41929-018-0045-1.

[17] Herrera, F. A. et al. The impact of transition metal catalysts on macroscopic dielectric barrier

discharge (dbd) characteristics in an ammonia synthesis plasma catalysis reactor. Journal of

Physics D: Applied Physics 52, 224002 (2019). URL https://doi.org/10.1088/1361-6463/

ab0c58.

[18] Barboun, P. et al. Distinguishing plasma contributions to catalyst performance in plasma-

assisted ammonia synthesis. ACS Sustainable Chemistry & Engineering 7, 8621–8630 (2019).

URL https://doi.org/10.1021/acssuschemeng.9b00406.

[19] Rouwenhorst, K. H. R., Kim, H.-H. & Lefferts, L. Vibrationally excited activation of N2 in

plasma-enhanced catalytic ammonia synthesis: A kinetic analysis. ACS Sustainable Chemistry

& Engineering 7, 17515–17522 (2019).

[20] Fridman, A. Plasma Chemistry (Cambridge University Press, New York, NY, 2008).

[21] Neyts, E. C., Ostrikov, K. K., Sunkara, M. K. & Bogaerts, A. Plasma catalysis: Synergistic

effects at the nanoscale. Chemical Reviews 115, 13408–13446 (2015). URL http://dx.doi.

org/10.1021/acs.chemrev.5b00362.

[22] Snoeckx, R. & Bogaerts, A. Plasma technology - a novel solution for CO2 conversion? Chem-

ical Society Reviews 46, 5805–5863 (2017). URL https://doi.org/10.1039/c6cs00066e.

[23] 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 4, 1115–1133

(2019). URL https://doi.org/10.1021/acsenergylett.9b00263.

19

[24] Rettner, C. T. & Stein, H. Effect of vibrational energy on the dissociative chemisorption of

N2 on fe(111). The Journal of Chemical Physics 87, 770–771 (1987). URL https://doi.

org/10.1063/1.453575.

[25] Romm, L., Katz, G., Kosloff, R. & Asscher, M. Dissociative chemisorption of N2 on ru(001)

enhanced by vibrational and kinetic energy: Molecular beam experiments and quantum me-

chanical calculations. The Journal of Physical Chemistry B 101, 2213–2217 (1997). URL

https://doi.org/10.1021/jp962599o.

[26] Murphy, M. J., Skelly, J. F., Hodgson, A. & Hammer, B. Inverted vibrational distributions

from N2 recombination at ru(001): Evidence for a metastable molecular chemisorption well.

The Journal of Chemical Physics 110, 6954–6962 (1999). URL https://doi.org/10.1063/

1.478601.

[27] Diekhoner, L. et al. N2 dissociative adsorption on ru(0001): The role of energy loss. The

Journal of Chemical Physics 115, 9028–9035 (2001). URL https://doi.org/10.1063/1.

1413746.

[28] Smith, R. R. Preference for vibrational over translational energy in a gas-surface reaction.

Science 304, 992–995 (2004). URL https://doi.org/10.1126/science.1096309.

[29] Hu, C. et al. Surface plasmon enabling nitrogen fixation in pure water through a dissociative

mechanism under mild conditions. Journal of the American Chemical Society 141, 7807–7814

(2019). URL https://doi.org/10.1021/jacs.9b01375.

[30] Thimsen, E. Beyond equilibrium thermodynamics in the low temperature plasma processor.

Journal of Vacuum Science Technology B 36, 048501 (2018). URL https://doi.org/10.

1116/1.5022470.

[31] Uytdenhouwen, Y. et al. How process parameters and packing materials tune chemical equi-

librium and kinetics in plasma-based co2 conversion. Chemical Engineering Journal 372,

1253–1264 (2019). URL https://doi.org/10.1016/j.cej.2019.05.008.

[32] Zhou, L. M., Xue, B., Kogelschatz, U. & Eliasson, B. Nonequilibrium plasma reforming

of greenhouse gases to synthesis gas. Energy Fuels 12, 1191–1199 (1998). URL https:

//doi.org/10.1021/ef980044h.

[33] Eliasson, B., jun Liu, C. & Kogelschatz, U. Direct conversion of methane and carbon dioxide

to higher hydrocarbons using catalytic dielectric-barrier discharges with zeolites. Industrial

& Engineering Chemistry Research 39, 1221–1227 (2000). URL https://doi.org/10.1021/

20

ie990804r.

[34] Wang, Q., Yan, B.-H., Jin, Y. & Cheng, Y. Investigation of dry reforming of methane in a

dielectric barrier discharge reactor. Plasma Chemistry and Plasma Processing 29, 217–228

(2009). URL https://doi.org/10.1007/s11090-009-9173-3.

[35] Wang, Q., Yan, B.-H., Jin, Y. & Cheng, Y. Dry reforming of methane in a dielectric barrier

discharge reactor with Ni/Al2O3 catalyst: Interaction of catalyst and plasma. Energy Fuels

23, 4196–4201 (2009). URL https://doi.org/10.1021/ef900286j.

[36] Sentek, J. et al. Plasma-catalytic methane conversion with carbon dioxide in dielectric barrier

discharges. Applied Catalysis B: Environmental 94, 19–26 (2010). URL https://doi.org/

10.1016/j.apcatb.2009.10.016.

[37] 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 1, 94–99 (2016).

URL http://dx.doi.org/10.1021/acsenergylett.6b00051.

[38] Gallon, H. J., Tu, X. & Whitehead, J. C. Effects of reactor packing materials on H2 production

by CO2 reforming of CH4 in a dielectric barrier discharge. Plasma Processes and Polymers 9,

90–97 (2011). URL https://doi.org/10.1002/ppap.201100130.

[39] Tu, X., Gallon, H. J., Twigg, M. V., Gorry, P. A. & Whitehead, J. C. Dry reforming of

methane over a Ni/Al2O3 catalyst in a coaxial dielectric barrier discharge reactor. Journal of

Physics D: Applied Physics 44, 274007 (2011). URL https://doi.org/10.1088/0022-3727/

44/27/274007.

[40] Tu, X. & Whitehead, J. Plasma-catalytic dry reforming of methane in an atmospheric dielec-

tric barrier discharge: Understanding the synergistic effect at low temperature. Applied Catal-

ysis B: Environmental 125, 439–448 (2012). URL https://doi.org/10.1016/j.apcatb.

2012.06.006.

[41] Mahammadunnisa, S., Reddy, P. M. K., Ramaraju, B. & Subrahmanyam, C. Catalytic non-

thermal plasma reactor for dry reforming of methane. Energy Fuels 27, 4441–4447 (2013).

URL https://doi.org/10.1021/ef302193e.

[42] Scapinello, M., Martini, L. M. & Tosi, P. CO2 hydrogenation by CH4 in a dielectric barrier

discharge: Catalytic effects of nickel and copper. Plasma Processes and Polymers 11, 624–628

(2014). URL https://doi.org/10.1002/ppap.201400023.

21

[43] Zeng, Y., Zhu, X., Mei, D., Ashford, B. & Tu, X. Plasma-catalytic dry reforming of methane

over γ-Al2O3 supported metal catalysts. Catalysis Today 256, 80–87 (2015). URL https:

//doi.org/10.1016/j.cattod.2015.02.007.

[44] Wang, L., Yi, Y., Wu, C., Guo, H. & Tu, X. One-step reforming of CO2 and CH4 into high-

value liquid chemicals and fuels at room temperature by plasma-driven catalysis. Angewandte

Chemie International Edition 56, 13679–13683 (2017). URL https://doi.org/10.1002/

anie.201707131.

[45] Michielsen, I., Uytdenhouwen, Y., Bogaerts, A. & Meynen, V. Altering conversion and prod-

uct selectivity of dry reforming of methane in a dielectric barrier discharge by changing

the dielectric packing material. Catalysts 9, 51 (2019). URL https://doi.org/10.3390/

catal9010051.

[46] Pan, K. L., Chung, W. C. & Chang, M. B. Dry reforming of CH4 with CO2 to generate syngas

by combined plasma catalysis. IEEE Transactions on Plasma Science 42, 3809–3818 (2014).

URL https://doi.org/10.1109/tps.2014.2360238.

[47] Kraus, M., Eliasson, B., Kogelschatz, U. & Wokaun, A. CO2 reforming of methane by the com-

bination of dielectric-barrier discharges and catalysis. Physical Chemistry Chemical Physics

3, 294–300 (2001). URL https://doi.org/10.1039/b007015g.

[48] Goujard, V., Tatibouet, J.-M. & Batiot-Dupeyrat, C. Influence of the plasma power supply

nature on the plasma-catalyst synergism for the carbon dioxide reforming of methane. IEEE

Transactions on Plasma Science 37, 2342–2346 (2009). URL https://doi.org/10.1109/

tps.2009.2033695.

[49] Nguyen, H. H. & Kim, K.-S. Combination of plasmas and catalytic reactions for CO2 reforming

of CH4 by dielectric barrier discharge process. Catalysis Today 256, 88–95 (2015). URL

https://doi.org/10.1016/j.cattod.2015.04.034.

[50] Gibson, E. K. et al. Probing the role of a non-thermal plasma (ntp) in the hybrid ntp catalytic

oxidation of methane. Angewandte Chemie International Edition 56, 9351–9355 (2017). URL

https://doi.org/10.1002/anie.201703550.

[51] Hansen, F. Y., Henriksen, N. E., Billing, G. D. & Guldberg, A. Catalytic synthesis of ammonia

using vibrationally excited nitrogen molecules: Theoretical calculation of equilibrium and

rate constants. Surface Science 264, 225–234 (1992). URL https://doi.org/10.1016/

0039-6028(92)90180-e.

22

[52] Evans, M. G. & Polanyi, M. Inertia and driving force of chemical reactions. Trans. Faraday

Soc. 34, 11 (1938). URL http://dx.doi.org/10.1039/TF9383400011.

[53] Bligaard, T. et al. The brønsted-evans-polanyi relation and the volcano curve in heterogeneous

catalysis. Journal of Catalysis 224, 206–217 (2004). URL http://dx.doi.org/10.1016/j.

jcat.2004.02.034.

[54] Grabow, L. C. CHAPTER 1. Computational Catalyst Screening, 1–58. RSC Catalysis Series

(Royal Society of Chemistry, 2013). URL https://doi.org/10.1039/9781849734905-00001.

[55] Dahl, S., Logadottir, A., Jacobsen, C. J. & Nørskov, J. K. Electronic factors in catalysis: the

volcano curve and the effect of promotion in catalytic ammonia synthesis. Applied Catalysis

A: General 222, 19–29 (2001). URL https://doi.org/10.1016/s0926-860x(01)00826-2.

[56] Snoeckx, R., Aerts, R., Tu, X. & Bogaerts, A. Plasma-based dry reforming: A computational

study ranging from the nanoseconds to seconds time scale. The Journal of Physical Chemistry

C 117, 4957–4970 (2013). URL https://doi.org/10.1021/jp311912b.

[57] Boisen, A., Dahl, S., Nørskov, J. & Christensen, C. Why the optimal ammonia synthesis

catalyst is not the optimal ammonia decomposition catalyst. Journal of Catalysis 230, 309–

312 (2005). URL https://doi.org/10.1016/j.jcat.2004.12.013.

[58] ichi Aika, K., Yamaguchi, J. & Ozaki, A. Ammonia synthesis over rhodium, iridium and

platinum promoted by potassium. Chemistry Letters 2, 161–164 (1973). URL https://doi.

org/10.1246/cl.1973.161.

[59] White, A. H. & Melville, W. The decomposition of ammonia at high temperatures. Journal

of the American Chemical Society 27, 373–386 (1905). URL https://doi.org/10.1021/

ja01982a005.

[60] Hagelaar, G. J. M. & Pitchford, L. C. Solving the boltzmann equation to obtain electron trans-

port coefficients and rate coefficients for fluid models. Plasma Sources Science and Technology

14, 722–733 (2005). URL https://doi.org/10.1088/0963-0252/14/4/011.

[61] Kim, J., Go, D. B. & Hicks, J. C. Synergistic effects of plasma-catalyst interactions for

CH4 activation. Physical Chemistry Chemical Physics 19, 13010–13021 (2017). URL https:

//doi.org/10.1039/c7cp01322a.

[62] Mehta, P. et al. wfschneidergroup/SI-mehta-beyond-eq (2019). URL https://doi.org/10.

5281/zenodo.3566833.

23

ACKNOWLEDGMENTS

This work was supported by the U.S. Department of Energy, Office of Science, Basic En-

ergy Sciences, Sustainable Ammonia Synthesis Program, under Award No. DE-SC-0016543

and by the the U.S. Air Force Office of Scientific Research, under Award No. FA9550-18-1-

0157. P.M. acknowledges support through the Eilers Graduate Fellowship for Energy Related

Research from the University of Notre Dame. Computational resources were provided by the

Notre Dame Center for Research Computing. We thank the Notre Dame Energy Materials

Characterization Facility and the Notre Dame Integrated Imaging Facility for the use of the

X-ray diffractometer and the transmission electron microscope respectively.

AUTHOR CONTRIBUTIONS

P.M. and P.M.B. contributed equally to this work. P.M. developed the kinetic model.

P.M.B. performed ammonia synthesis experiments. All authors discussed the results and

co-wrote the manuscript.

COMPETING FINANCIAL INTERESTS

The authors declare no competing financial interests.

24

download fileview on ChemRxivChemrxiv-manuscript.pdf (504.40 KiB)

Supplementary Information: Plasma-Catalytic Ammonia

Synthesis Beyond the Equilibrium Limit

Prateek Mehta,1 Patrick Barboun,1 Yannick Engelmann,2 David B. Go,1, 3

Annemie Bogaerts,2, ∗ William F. Schneider,1, † and Jason C. Hicks1, ‡

1Department of Chemical and Biomolecular Engineering,

University of Notre Dame, Notre Dame, Indiana 46556, United States

2Department of Chemistry, Antwerp University,

Campus Drie Eiken, Universiteitsplein 1, 2610 Wilrijk

3Department of Aerospace and Mechanical Engineering,

University of Notre Dame, Notre Dame, Indiana 46556, United States

(Dated: January 27, 2020)

∗ annemie.bogaerts@uantwerpen.be† wschneider@nd.edu‡ jhicks3@nd.edu

SUPPLEMENTARY METHODS

We provide below additional information about the methods used, along with supple-

mentary figures and results. Source code and raw data is provided in an external Zenodo

repository [1]. The python class simpleMkm in simplemkm.py contains the core functions

necessary to perform the microkinetic calculations, while the class mkmRunner in mkmu-

tils.py contains utility functions to run the calculations in an automated fashion on our

computing cluster. The input (*.mkminp) and output (*.mkmout) files for the microkinetic

model are included in the Zenodo directory. Raw experimental data is also provided as an

Excel spreadsheet (high-T-expts.xlsx ). Example python scripts to perform the calculations

in this work, and to create the figures in the main text are provided below. These scripts

were executed within an Emacs org-mode document (supporting-data.org), which was then

exported to create this supplementary data pdf file.

Description of thermal reactions

N2 dissociative chemisorption was modeled using the same scaling relation as previous

work [2, 3]:

EactN2+∗−−⇀↽−−2N∗ = 1.57EN + 1.56 (1)

For this step, we assumed that the transition state was product-like, and the entropies of

both assumed to be zero, i.e. they were treated in the frozen-adsorbate limit.

NH3 formation from N · and H2 was modeled in a lumped fashion according to the

approach of Grabow [4] (see reaction (2) in the main text). The rate of this step was written

as,

r2 = k2P32H2θN (2)

The activation energy for the step was calculated using the scaling relation [4],

EactN∗+ 3

2H2−−⇀↽−−NH3+∗ = −0.39EN + 1.24 (3)

The transition state entropy for this step was assumed to be the same as the entropy of the

reactants.

The simplified treatment of the reaction kinetics described above is sufficient to model the

directional trends in ammonia synthesis rates as a function of the nitrogen binding energy

[4]. However, we note that the predicted reaction rates are overestimated compared to more

detailed kinetic models [2, 3].

Effect of degree of N2 excitation on NH3 yields

In the main text, we assumed that non-thermal activation of N2 reduces its dissocia-

tion barrier on the catalyst surface by 1 eV (see Figure 3). Here, we examine how the

temperature-yield profiles change when the barrier is assumed to be lowered by a greater or

lesser extent. Supplementary Figures 1 and 2 plot the NH3 yields for the three characteristic

catalysts discussed in the main text, but the N2 activation energy is assumed to be reduced

by 1.2 eV and 0.6 eV respectively. In both Supplementary Figures 1 and 2, we find that

the general features of Figure 3 are retained for the different catalysts. As observed in the

main text, no departures from thermal NH3 yields are observed for the catalyst that bind

N strongly. For materials that bind N with intermediate or low strength, ammonia yields

exceeding thermal yields in both the kinetic and equilibrium-limited regimes are observed

when reaction rates along the non-thermal pathway are sufficiently greater than those along

the thermal pathway. Differences in the temperature-yield profiles are most pronounced

for the material that binds N weakly (compare panel (c) in Figure 3 in the main text and

Supplementary Figures 1 and 2). The light-off temperature is a function of the extent to

which the N2 dissociation barrier is assumed to be reduced. When the barrier is assumed

to be reduced by only 0.6 eV, N2 activation remains strongly rate-limiting for this material,

and NH3 yields only light-off at high temperatures (Supplementary Figure 2c). In contrast,

when the barrier is assumed to be reduced by 1.2 eV, a temperature-independent plateau in

NH3 yields is observed, indicating a regime where the plasma-on reaction rate is limited by

the electron-impact activation of N2 (reaction (3) in the main text).

Supplementary Figure 1. Modeled steady-state NH3 partial pressures in plasma-enhanced catalytic

operation as a function of bulk temperature and rate of plasma-excitation when plasma-excitation

is assumed to reduce the barrier for N2 dissociation by 1.2 eV. The subpanels show catalysts

characterized by (a) strong, (b) intermediate and (c) weak nitrogen binding energies (EN). For

each catalyst steady state thermal (plasma-off), and plasma-on (for three values of kene) partial

pressures of NH3 are plotted.

Supplementary Figure 2. Modeled steady-state NH3 partial pressures in plasma-enhanced catalytic

operation as a function of bulk temperature and rate of plasma-excitation when plasma-excitation

is assumed to reduce the barrier for N2 dissociation by 0.6 eV. The subpanels show catalysts

characterized by (a) strong, (b) intermediate and (c) weak nitrogen binding energies (EN). For

each catalyst steady state thermal (plasma-off), and plasma-on (for three values of kene) partial

pressures of NH3 are plotted.

Catalyst Characterization Data

Supplementary Table 1. CO uptake, surface area, and average particle size of the catalysts inves-

tigated in this study.

Material CO uptake (µmol/g) Surface area (m2/g) Avg. particle size (nm)

5% Ni/Al2O3 20.8 191.9 15

5% Pt/Al2O3 84.4 183.4 2.5

Ammonia yields in unpacked and Al2O3 packed DBD reactor

Supplementary Figure 3. Plasma background ammonia yields in an unpacked DBD reactor and

a DBD reactor packed with γ-Al2O3 as a function of bulk reactor temperature at 10 W plasma

power. Reaction conditions: total flow rate = 40 SSCM, inlet N2:H2 = 2:1. The dashed line shows

the bulk thermodynamic equilibrium conversion at the specified reaction conditions.

Using the kinetic model: Plasma-off calculations

1 from utils import cd

2 from mkmutils import mkmRunner as runner

3

4 EAs = [-1.2, -0.6, 0.0]

5

6 plasma_on = False

7 rxn2_barrier = True

8

9 for EA in EAs:

10 mod = runner(EA,

11 plasma_on=plasma_on,

12 rxn2_barrier=rxn2_barrier,

13 npts=100,

14 concentration_based=True)

15

16 with cd('mkm-calcs/{0}'.format(mod.prefix)):

17 mod.write_input()

18 mod.run_job()

Using the kinetic model: Plasma-on calculations

1 from utils import cd

2 from mkmutils import mkmRunner as runner

3

4 EAs = [-1.2, -0.6, 0.0]

5

6 kei = range(-15, -12, 1)

7 ks_eimpact=[float('1.0e{0}'.format(i)) for i in kei] # cm3 / s

8

9 # Excitation energy, eV

10 Evib_A2 = 1.0

11

12 for EA in EAs:

13 for k_eimpact in ks_eimpact:

14 mod = runner(EA,

15 plasma_on=True,

16 k_eimpact_A2=k_eimpact,

17 npts=200,

18 rxn2_barrier=True,

19 excite_type='vib',

20 Evib_A2=Evib_A2)

21

22 with cd('mkm-calcs/{0}'.format(mod.prefix)):

23 mod.write_input(ncores=2)

24 mod.run_job()

Figure 2: Modeled plasma-off NH3 yields

1 from simplemkm import simpleMkm as mkm

2 import numpy as np

3 import matplotlib.pyplot as plt

4 from mkmutils import *

5

6 plt.style.use('seaborn-paper')

7 plt.rcParams["font.family"] = "Helvetica"

8

9 fig = plt.figure(figsize = (3.5, 3.25), dpi=200)

10

11 rates = []

12

13 EAmetal = [-1.2, -0.6, 0.0]

14

15

16 for EA in EAmetal:

17 mod = mkmRunner(EA, plasma_on=False, rxn2_barrier=True)

18 prefix = mod.prefix

19 d = load_variables('mkm-calcs/{0}/{0}.mkmout'.format(prefix))

20

21 allpressures = d['pressures']

22 pABs = []

23

24 for p in allpressures:

25 # N2, H2, and NH3 pressures

26 pA2, pB, pAB = p

27 pABs.append(np.float(pAB))

28

29 plt.plot(d['T'], pABs, '-',

30 label='$E_{{\mathrm{{N}}}} = {0}$ eV'.format(EA))

31

32 Xeq = [float(x) for x in d['Xeq']]

33 pABeq = [float(x[-1]) for x in d['eq_pressures']]

34

35 plt.plot(d['T'], pABeq, c='k', ls='--', label='Eqb. limit')

36

37 plt.ylim(-0.001, 0.1)

38 plt.xlim(350, 1000)

39 plt.legend(frameon=False, fontsize=8)

40 plt.xlabel('Temperature (K)')

41 plt.ylabel('NH$_{3}$ pressure (atm)')

42

43 # Inset

44 ax2 = fig.add_axes([0.65, 0.4, 0.25, 0.25])

45

46 EAs = np.linspace(1., -1.5, 150)

47 T = 473.

48

49 theta = 0.0

50

51 X = 0.05

52

53 rates = []

54 for i, EA in enumerate(EAs):

55 mod = mkm(T, EA, rxn2_barrier=True)

56

57 kf, kr = mod.get_rate_constants()

58 K2 = kf[1] / kr[1]

59 pA2, pB, pAB = mod.get_pressures(X)

60

61 theta = mod.integrate_odes(theta0=theta, X=X)[0]

62

63 try:

64 theta = mod.find_steady_state_roots(theta0=[theta], X=X)

65 except:

66 theta = mod.integrate_odes(theta0=theta, X=X)[0]

67 try:

68 theta = mod.find_steady_state_roots(theta0=[theta], X=X)

69 except:

70 pass

71

72 r = mod.get_rates(theta, mod.get_pressures(X))

73

74 if r[0] > 0:

75 ls = '-'

76 else:

77 ls = '--'

78

79 rates.append(abs(r[0]))

80 ax2.plot(EAs,

81 np.log10(rates),

82 ls,

83 label='$p_{{\mathrm{{AB}}}} = {0:1.3f}$ atm'.format(pAB), c='C7')

84

85 EAmetal = [-1.2, -0.6, 0.0]

86

87 for EA in EAmetal:

88 mod = mkm(T, EA, rxn2_barrier=True)

89

90 kf, kr = mod.get_rate_constants()

91 K2 = kf[1] / kr[1]

92 pA2, pB, pAB = mod.get_pressures(X)

93

94 theta = mod.integrate_odes(theta0=theta, X=X)[0]

95

96 try:

97 theta = mod.find_steady_state_roots(theta0=[theta], X=X)

98 except:

99 theta = mod.integrate_odes(theta0=theta, X=X)[0]

100 try:

101 theta = mod.find_steady_state_roots(theta0=[theta], X=X)

102 except:

103 pass

104

105 r = mod.get_rates(theta, mod.get_pressures(X))

106

107 if r[0] > 0:

108 ls = '-'

109 else:

110 ls = '--'

111

112 ax2.plot(EA, np.log10(abs(r[0])), 'o')

113

114 plt.ylim(-16, -4)

115 plt.xlim(-1.5, 0.5)

116

117 plt.yticks(np.arange(-15, 0, 5))

118

119 plt.xlabel('$E_{\mathrm{N}}$ (eV)')

120 plt.ylabel('log$_{10}$(TOF [s$^{-1}$])')

121

122 plt.tight_layout()

123

124 for ext in ['eps', 'pdf', 'png']:

125 plt.savefig('figures/thermal-pNH3.{0}'.format(ext), dpi=200)

126 plt.show()

Figure 3: Modeled plasma-on NH3 yields

1 from mkmutils import *

2 import matplotlib.pyplot as plt

3 from mkmutils import mkmRunner as runner

4

5 plt.style.use('seaborn-paper')

6 plt.rcParams["font.family"] = "Helvetica"

7

8 # Barrier reduced by,

9 Evib_A2 = 1.0

10

11 plt.figure(figsize=(3, 4), dpi=200)

12

13 grid = plt.GridSpec(3, 1, hspace=0)

14 ax1 = plt.subplot(grid[0, 0])

15 ax2 = plt.subplot(grid[1, 0], sharex=ax1)

16 ax3 = plt.subplot(grid[2, 0], sharex=ax1)

17

18 axes = [ax1, ax2, ax3]

19

20 EAs = [-1.2, -0.6, 0.0]

21 kei = range(-15, -12, 1)

22 ks_eimpact=[float('1.0e{0}'.format(i)) for i in kei] # cm3 / s

23

24 plasma_on = True

25 rxn2_barrier = True

26

27 for ax, EA in zip(axes, EAs):

28

29 for c, k_eimpact in zip(['tan', 'palevioletred', 'seagreen'], ks_eimpact):

30 mod = runner(EA,

31 plasma_on=plasma_on,

32 k_eimpact_A2=k_eimpact,

33 rxn2_barrier=rxn2_barrier,

34 excite_type='vib',

35 Evib_A2=Evib_A2)

36

37 prefix = mod.prefix

38 try:

39 d = load_variables('mkm-calcs/{0}/{0}.mkmout'.format(prefix))

40

41 if plasma_on:

42 inp = mod.input_params

43 kene = inp['k_eimpact_A2'] * inp['n_e']

44 allpressures = d['pressures']

45 pABs = []

46

47 for p in allpressures:

48 pA2, pA2prime, pB, pAB, pABprime = p

49 pABs.append(np.float(pAB))

50

51 label = '$k_{{e}} n_{{e}} = 10^{{{0:1.0f}}}$ s$^{{-1}}$'.format(np.log10(kene)))

52 line, = ax.plot(d['T'],

53 pABs,

54 '-',

55 c=c,

56 label=label)

57

58 except Exception as E:

59 print prefix, E

60

61 modoff = runner(EA,

62 plasma_on=False,

63 rxn2_barrier=rxn2_barrier)

64

65 prefix = modoff.prefix

66 doff = load_variables('mkm-calcs/{0}/{0}.mkmout'.format(prefix))

67

68 allpressures = doff['pressures']

69 pABsoff = []

70

71 for p in allpressures:

72 pA2, pB, pAB = p

73 pABsoff.append(np.float(pAB))

74

75 pABsoff2plot = [pAB for i, pAB in enumerate(pABsoff) if i % 4. == 0]

76 T2plot = [T for i, T in enumerate(doff['T']) if i % 4. == 0]

77

78 ax.plot(T2plot, pABsoff2plot, 'o', ms=4, mew=1, mfc='None', mec='C5', label='plasma off')

79

80 pABeq = [float(x[-1]) for x in d['eq_pressures']]

81

82 ax.plot(d['T'], pABeq, c='k', ls='--', label='Eqb. limit')

83

84 ax.set_ylim(-0.01, 0.3)

85 ax.set_yticks([0.0, 0.1, 0.2])

86 ax.set_yticklabels([0.0, 0.1, 0.2], fontsize=7.5)

87 ax.set_ylabel('NH$_{3}$ pressure (atm)', fontsize=7.5)

88

89 ax1, ax2, ax3 = axes

90

91 ax1.legend(fontsize=6, frameon=False, ncol=2, columnspacing=1, handlelength=1.5)

92

93 plt.setp(ax1.get_xticklabels(), visible=False)

94 plt.setp(ax2.get_xticklabels(), visible=False)

95

96 ax3.set_xlim(350, 950)

97 ax3.set_xlabel('Temperature (K)', fontsize=7.5)

98

99 plt.figtext(0.22, 0.93, 'a', fontsize=8, fontweight='bold')

100 plt.figtext(0.22, 0.65, 'b', fontsize=8, fontweight='bold')

101 plt.figtext(0.22, 0.37, 'c', fontsize=8, fontweight='bold')

102

103 plt.figtext(0.7, 0.75, '$E_{N} = -1.2$ eV', fontsize=7.5)

104 plt.figtext(0.7, 0.5, '$E_{N} = -0.6$ eV', fontsize=7.5)

105 plt.figtext(0.74, 0.32, '$E_{N} = 0.0$ eV', fontsize=7.5)

106

107 plt.tight_layout()

108 for ext in ['eps', 'png', 'pdf']:

109 plt.savefig('figures/plasma-on-NH3-syn.{0}'.format(ext), dpi=200)

110

111 plt.show()

Figure 4: Experimental plasma-only NH3 yields

1 import pandas as pd

2 import matplotlib.pyplot as plt

3

4 plt.style.use('seaborn-paper')

5 plt.rcParams["font.family"] = "Helvetica"

6

7 plt.figure(figsize = (3.25, 3), dpi=200)

8

9 data = pd.read_excel('high-T-expts.xlsx',

10 'eqb-data',

11 header=0)

12

13 plt.plot(data['T(K)'], data['X'] * 100, 'k--', label='Eqb. limit')

14

15 data = pd.read_excel('high-T-expts.xlsx', 'plasma-sweep', header=1)

16

17 plt.errorbar(data['Temperature.1'] + 273.15,

18 data['Conversion.1'],

19 data['Conversion Error.1'],

20 capthick=1.5,

21 fmt=None,

22 label=None)

23

24 plt.plot(data['Temperature.1'] + 273.15,

25 data['Conversion.1'],

26 'o',

27 ms=6,

28 mew=1.5,

29 mfc='w',

30 c='C0',

31 label='10 W')

32

33 plt.errorbar(data['Temperature'] + 273.15,

34 data['Conversion'],

35 data['Conversion Error'],

36 c='tan',

37 fmt=None,

38 capthick=1.5,

39 label=None)

40

41 plt.plot(data['Temperature'] + 273.15,

42 data['Conversion'],

43 'o',

44 ms=6,

45 mew=1.5,

46 mfc='w',

47 c='tan',

48 label='5 W')

49

50 plt.errorbar(data['Temperature.2'] + 273.15,

51 data['Conversion.2'],

52 data['Conversion Error.2'],

53 c='C3', capthick=1.5,

54 fmt=None,

55 label=None)

56 plt.plot(data['Temperature.2'] + 273.15,

57 data['Conversion.2'], 'o',

58 ms=6, mew=1.5,

59 mfc='w',

60 c='C3',

61 label='15 W')

62

63 plt.ylim(0, 4)

64 plt.xlim(400, 1173)

65 plt.xlabel('Temperature (K)')

66 plt.ylabel('NH$_{3}$ Yield (%)')

67 plt.legend(frameon=False, fontsize=8, ncol=2)

68 plt.tight_layout()

69

70

71 for ext in ['eps', 'pdf', 'png']:

72 plt.savefig('figures/NH3-power-expts.{0}'.format(ext), dpi=200)

73 plt.show()

Figure 5: Experimental plasma-catalytic NH3 yields

1 import pandas as pd

2 import matplotlib.pyplot as plt

3

4 plt.style.use('seaborn-paper')

5 plt.rcParams["font.family"] = "Helvetica"

6

7 plt.figure(figsize = (4, 5), dpi=200)

8

9

10 ax2 = plt.subplot(212)

11

12 data = pd.read_excel('high-T-expts.xlsx', 'NH3-decomposition', header=1)

13

14 plt.plot(data['Temperature.2'] + 273.15,

15 data['Conversion.2'],

16 'o',

17 ms=6,

18 mew=1.5,

19 mfc='w',

20 c='C0',

21 label='Al$_{2}$O$_{3}$')

22 plt.errorbar(data['Temperature.1'] + 273.15,

23 data['Conversion.1'],

24 data['Error.1'],

25 capthick=1.5, c='C1',

26 fmt=None,

27 label=None)

28

29 plt.plot(data['Temperature.1'] + 273.15,

30 data['Conversion.1'],

31 '^', ms=6, mew=1.5,

32 mfc='w',

33 c='C1',

34 label='Ni/Al$_{2}$O$_{3}$')

35

36 plt.errorbar(data['Temperature'] + 273.15,

37 data['Conversion'],

38 data['Error'],

39 fmt=None,

40 c='C2',

41 capthick=1.5,

42 label=None)

43

44 plt.plot(data['Temperature'] + 273.15,

45 data['Conversion'],

46 's',

47 ms=6,

48 mew=1.5,

49 mfc='w',

50 c='C2',

51 label='Pt/Al$_{2}$O$_{3}$')

52

53 plt.xlim(400, 1173)

54 plt.ylim(-5, 105)

55 plt.legend(frameon=False, fontsize=8)

56 plt.xlabel('Temperature (K)', fontsize=10)

57 plt.xticks(fontsize=10)

58 plt.yticks(fontsize=10)

59 plt.ylabel('NH$_{3}$ Conversion (%)', fontsize=10)

60

61 ax1 = plt.subplot(211, sharex=ax2)

62

63 data = pd.read_excel('high-T-expts.xlsx', 'eqb-data', header=0)

64 plt.plot(data['T(K)'], data['X'] * 100, 'k--', label='Eqb. limit')

65

66 data = pd.read_excel('high-T-expts.xlsx', 'metal-sweep', header=1)

67

68 plt.errorbar(data['Temperature.2'] + 273.15,

69 data['Conversion.2'],

70 data['Conversion Error.2'],

71 capthick=1.5,

72 fmt=None,

73 label=None)

74 plt.plot(data['Temperature.2'] + 273.15,

75 data['Conversion.2'],

76 'o',

77 ms=6,

78 mew=1.5,

79 mfc='w',

80 c='C0',

81 label='Al$_{2}$O$_{3}$')

82

83 plt.errorbar(data['Temperature.1'] + 273.15,

84 data['Conversion.1'],

85 data['Conversion Error.1'],

86 capthick=1.5,

87 fmt=None,

88 label=None)

89 plt.plot(data['Temperature.1'] + 273.15,

90 data['Conversion.1'],

91 '^',

92 ms=6,

93 mew=1.5,

94 mfc='w',

95 c='C1',

96 label='Ni/Al$_{2}$O$_{3}$')

97

98 plt.errorbar(data['Temperature'] + 273.15,

99 data['Conversion'],

100 data['Conversion Error'],

101 fmt=None,

102 capthick=1.5,

103 label=None)

104 plt.plot(data['Temperature'] + 273.15,

105 data['Conversion'],

106 's',

107 ms=6,

108 mew=1.5,

109 mfc='w',

110 c='C2',

111 label='Pt/Al$_{2}$O$_{3}$')

112

113 plt.ylim(0, 2.5)

114 plt.xlim(400, 1173)

115 plt.yticks(fontsize=10)

116 plt.setp(ax1.get_xticklabels(), visible=False)

117

118 plt.ylabel('NH$_{3}$ Yield (%)', fontsize=10)

119 plt.legend(frameon=False, fontsize=8)

120 plt.tight_layout()

121

122 plt.figtext(0.03, 0.96, 'a', fontsize=12, fontweight='bold')

123 plt.figtext(0.03, 0.49, 'b', fontsize=12, fontweight='bold')

124

125

126 for ext in ['eps', 'pdf', 'png']:

127 plt.savefig('figures/NH3-plasma-metal.{0}'.format(ext), dpi=200)

128

129 plt.show()

SUPPLEMENTARY REFERENCES

[1] Mehta, P. et al. wfschneidergroup/SI-mehta-beyond-eq (2019). URL https://doi.org/10.

5281/zenodo.3566833.

[2] Vojvodic, A. et al. Exploring the limits: A low-pressure, low-temperature haber-bosch process.

Chemical Physics Letters 598, 108–112 (2014). URL https://doi.org/10.1016/j.cplett.

2014.03.003.

[3] Mehta, P. et al. Overcoming ammonia synthesis scaling relations with plasma-enabled catalysis.

Nature Catalysis 1, 269–275 (2018). URL https://doi.org/10.1038/s41929-018-0045-1.

[4] Grabow, L. C. CHAPTER 1. Computational Catalyst Screening, 1–58. RSC Catalysis Series

(Royal Society of Chemistry, 2013). URL https://doi.org/10.1039/9781849734905-00001.

download fileview on ChemRxivsupporting-info.pdf (396.20 KiB)