principles of dynamic heterogeneous catalysis: …heterogeneous catalysis is an integral component...
TRANSCRIPT
doi.org/10.26434/chemrxiv.7790009.v1
Principles of Dynamic Heterogeneous Catalysis: Surface Resonance andTurnover Frequency ResponseM. Alexander Ardagh, Omar A. Abdelrahman, Paul Dauenhauer
Submitted date: 01/03/2019 • Posted date: 01/03/2019Licence: CC BY 4.0Citation information: Ardagh, M. Alexander; Abdelrahman, Omar A.; Dauenhauer, Paul (2019): Principles ofDynamic Heterogeneous Catalysis: Surface Resonance and Turnover Frequency Response. ChemRxiv.Preprint.
Acceleration of catalytic transformation of molecules via heterogeneous materials occurs through design ofactive binding sites to optimally balance the requirements of all steps in a catalytic cycle. In accordance withthe Sabatier principle, the characteristics of a single binding site are balanced between at least two transientphenomena, leading to maximum possible catalytic activity at a single, static condition (i.e., a ‘volcano curve’peak). In this work, a dynamic heterogeneous catalyst oscillating between two electronic states was evaluatedto demonstrate catalytic activity as much as three-to-four orders of magnitude (1,000-10,000x) above theSabatier maximum. Surface substrate binding energies were varied by a given amplitude (0.1 < ΔU < 3.0 eV)over a broad range of frequencies (10-4 < f < 1011 s-1) in square, sinusoidal, sawtooth, and triangularwaveforms to characterize surface dynamics impact on average catalytic turnover frequency. Catalyticsystems were shown to exhibit order-of-magnitude dynamic rate enhancement at ‘surface resonance’ definedas the band of frequencies (e.g., 103-107 s-1) where the applied surface waveform kinetics were comparableto kinetics of individual microkinetic chemical reaction steps. Key dynamic performance parameters arediscussed regarding industrial catalytic chemistries and implementation in physical dynamic systemsoperating above kilohertz frequencies.
File list (2)
download fileview on ChemRxivManuscript_Dauenhauer.pdf (2.03 MiB)
download fileview on ChemRxivSupporting_Information_Dauenhauer.pdf (2.05 MiB)
____________________________________________________________________________ Ardagh, et al. Page 1
Principles of Dynamic Heterogeneous Catalysis: Surface Resonance and Turnover Frequency Response
M. Alexander Ardagh1,2, Omar A. Abdelrahman2,3, Paul J. Dauenhauer1,2* 1University of Minnesota, Department of Chemical Engineering and Materials Science, 421 Washington
Ave SE, Minneapolis, MN 55455 2Catalysis Center for Energy Innovation, University of Delaware, 150 Academy Street, Newark, DE
19716 3University of Massachusetts Amherst, Department of Chemical Engineering, 159 Goessmann
Laboratory, 686 North Pleasant Street, Amherst, MA 01003.
*Corresponding author: [email protected]
Introduction. Heterogeneous catalysis is an
integral component of many industrial processes
that manufacture food, materials, and energy for a
high quality of life[1]. Catalytic surfaces, pores, and
single-atom sites are responsible for accelerating
the rates of reactions that protect and provide the
most important resources including fuels[2],
fertilizers[3], medicines[4], basic chemicals[5,6], CO2
feedstock[7], clean air[8], and polymers[9,10] that
would be inaccessible absent catalytic acceleration.
In the past century, enhancement of these processes
has progressed via increased catalyst selectivity and
stability along with increased overall activity for
smaller reactors that operate at higher efficiency
and lower temperature[11].
Catalytic rate enhancement occurs primarily
through catalyst design to tune the binding
characteristics of surface species and transition
states for maximum catalytic turnover
frequency[12,13,14]. In the past two decades, advances
in nanostructured materials led to detailed synthesis
of atomic-scale active sites that precisely balance
the surface substrate binding energies[15]. The limit
of this approach is characterized by the Sabatier
principle, which states that the binding of substrates
must be neither too strong nor too weak[16,17].
Quantitative description of the Sabatier principle
was captured in Balandin-Sabatier volcano-shaped
curves (“volcano curves” for the remainder of the
manuscript), which depicted a metric of catalyst
activity relative to a descriptor of substrate
binding[18,19]. Balandin depicted volcano-shaped
curves in 1960 and 1964 for the dehydration and
dehydrogenation of alcohols, with the catalytic
Abstract. Acceleration of catalytic transformation of molecules via heterogeneous materials occurs
through design of active binding sites to optimally balance the requirements of all steps in a catalytic
cycle. In accordance with the Sabatier principle, the characteristics of a single binding site are balanced
between at least two transient phenomena, leading to maximum possible catalytic activity at a single,
static condition (i.e., a ‘volcano curve’ peak). In this work, a dynamic heterogeneous catalyst oscillating
between two electronic states was evaluated to demonstrate catalytic activity as much as three-to-four
orders of magnitude (1,000-10,000x) above the Sabatier maximum. Surface substrate binding energies
were varied by a given amplitude (0.1 < ΔU < 3.0 eV) over a broad range of frequencies (10-4 < f < 1011
s-1) in square, sinusoidal, sawtooth, and triangular waveforms to characterize the impact of surface
dynamics on average catalytic turnover frequency. Catalytic systems were shown to exhibit order-of-
magnitude dynamic rate enhancement at ‘surface resonance’ defined as the band of frequencies (e.g.,
103-107 s-1) where the applied surface waveform kinetics were comparable to kinetics of individual
microkinetic chemical reaction steps. Key dynamic performance parameters are discussed regarding
industrial catalytic chemistries and implementation in physical dynamic systems operating above
kilohertz frequencies.
____________________________________________________________________________ Ardagh, et al. Page 2
activity dependent on the bond energies between
the alcohol oxygen and the metal oxide
catalysts[18,19]. Since that time, volcano curves have
been generated for numerous catalytic chemistries
including NOx decomposition[20], propylene
oxidation[21], hydrodesulfurization[22,23], ammonia
synthesis[24,25], CO oxidation[26], and oxygenate
decomposition[27], among many other reactions[28].
The simplest surface catalytic mechanism of
species A reacting to species B depicted in Figure
1A obeys the Sabatier principle regarding
adsorption enthalpy (ΔHA, ΔHB) and surface
reaction activation energy, (Ea). Reactant molecule
A adsorbs to the surface as A*, undergoes surface
reaction to B*, and then desorbs to gas-phase
product B; the overall turnover rate can potentially
be limited by any one of these three steps. Reactant
adsorption is a fast, barrierless step unless it is
combined with surface reactions; a combined step
of dissociative adsorption (e.g., N2) is commonly
rate limiting on some catalytic materials[29]. The
volcano curve therefore results from the sequential
kinetics of surface reaction(s) and product
desorption, as presented originally by Balandin and
depicted in Figure 1B, the transition between
surface reaction- and desorption-control exhibits
the characteristic ‘volcano’ two-kinetic-regime
plot. As depicted in Figure 1C, surface adsorbates
desorb slowly on strong-binding materials, while
surface reactions occur slowly on weak-binding
materials; the kinetic balance of these two steps
forms the optimum turnover frequency of the
system (i.e., volcano peak) characteristic to
materials only exhibiting the optimal binding
energy[30].
The asymmetry of some volcano curves
depicted in Figure 1B arise from the relationship
between the surface binding energy and the surface
reaction activation energy. As described in the
Brønsted-Evans-Polyani (BEP) principle[31,32], the
activation energy of a catalytic reaction linearly
correlates with the surface reaction enthalpy by a
linear-scaling parameter, α, and offset of E0
associated with a reaction class[33].
Ea = ∝∗ ∆Hsr + E0 (1)
As depicted in Figure 1B, α ~ 0 (purple) indicates
negligible relationship between the enthalpy of
surface reaction and the surface activation energy
resulting in a ‘flat’ volcano, while a completely
proportional relationship, α ~ 1.0 (red), forms a
more symmetric volcano curve; values between
zero and one form the interspersed curves: blue of
α ~ 0.8, green of α ~ 0.6, yellow of α ~ 0.4, and
orange of α ~ 0.2 all with an offset E0 of 102
kJ/mole. The linear scaling relationships can be
universal and consistent across different
molecules[34,35,36,37], while other chemicals such as
formic acid exhibit different linear scaling
relationships across different transition metal
groups of the periodic table[38]. The Figure 1B set of
volcano curves are also defined by the condition
that the surface energy of B* changes at twice the
rate of the surface energy of A*, which is a ratio
A
A*
B
B*
ΔHA ΔHB
EA
1E-5
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E+3
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
Tu
rno
ve
r F
req
ue
ncy [s
-1]
Relative Binding Energy of B* [eV]
Surface
Reaction
ControlledDesorption
Controlled
10-5
10-4
10-3
10-2
10-1
1
101
102
103
-160
-120
-80
-40
0
40
En
thalp
y [
kJ m
ol-1
]
A
B
ΔHBEAA*
B*
ΔHA
Tu
rno
ver
Fre
qu
en
cy
[s-1
]
En
thalp
y [k
J m
ol-1
]
A B C
Figure 1. Catalyst Optimization via the Sabatier Principle. A. Surface reaction of A converting to B via transition
state with forward activation energy, EA. B. Volcano plots for the turnover frequency of A-to-B with variable binding
energy of A* and B* and variable Brønsted-Evans-Polyani relationships of Ea to ΔHsr (0 < α < 1.0: 0.2 increments,
purple to red). Conditions: perfectly mixed reactor at 150 °C, YB ~ 1%. C. Three conditions of surface intermediate
binding energy: black (+0.4 eV), red (-0.1 eV, α = 1.0), blue (-0.5 eV, α = 0.2).
____________________________________________________________________________ Ardagh, et al. Page 3
that can vary between surface chemistries and
materials.
Catalyst activity optimization within the
context of volcano curves has focused on catalyst
design to achieve optimal turnover at the volcano
curve apex. Of the existing catalysts and multi-
metal combinations, computational screening of the
relevant surface-binding descriptors aims to
identify single- or multi-descriptor optima from
databases of catalytic materials[39,40,41]. Other
strategies have aimed to create and tune the
properties of new materials including physical and
electronic descriptors such as metal spacing and
coordination, d-band center and fermi level, and
electronic interaction with supports, solvents, and
co-adsorbents via multi-metal mixing and/or
nanostructured synthesis[42,43,44,45,46,47]; all of these
approaches have achieved success in creating new
materials near the maximum theoretical turnover
frequency of a static catalyst.
The limitation of the Balandin-Sabatier
maximum arises from the multi-purpose catalyst,
which must balance the kinetics of competing
reaction steps (activation, desorption, etc.). One
strategy to exceed this maximum is to decouple the
reaction phenomena and physically disassociate
sequential chemistries. The physical reaction
dissociation approach was recently demonstrated
with ammonia synthesis, whereby gas-phase
activation of N2 occurred via plasma, and metals
catalyzed hydrogenation of nitrogen to ammonia
with rapid desorption[48,49]. Deconvoluted control of
individual reactions (activation with plasma and
hydrogenation with a metal catalyst) permits
independent system tuning to yield an overall rate
potentially in excess of the Balandin-Sabatier
maximum in conventional catalytic reactors.
In this work, we will focus on the kinetics of
temporally decoupling surface reaction steps via
oscillation of the catalytic surface binding energy.
As depicted in Figure 2A, the volcano curve can be
depicted with its independent slopes extended
above the apex (dashed black lines); these represent
the potential rates of surface reaction and
desorption absent other limitations. In a dynamic
system, the surface energy could oscillate between
two or more binding energy states, with the
oscillation amplitude identified as the total distance
in traversed binding energy (ΔU [=] eV) at the
frequency of oscillation (f [=] s-1); in Figure 2A, the
volcano plot has a BEP of moderate slope, α ~ 0.8,
and amplitude (-0.10 to +0.50 eV, ΔU = 0.60 eV).
1E-6
1E-5
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E+3
1E+4
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
Tu
rno
ve
r F
requ
en
cy [
s-1
]
Relative Binding Energy of B* [eV]
Optimum Dynamics at
Resonant Frequency
Amplitude
Static
Optimum
Minimum
Dynamics far
From Resonant
Frequency
Dyn
am
ic P
erfo
rma
nce
Ra
ng
e
10-6
10-5
10-4
10-3
10-2
10-1
100
101
102
103
104
A B
-0.3
-0.1
0.1
0.3
0.5
0.7
0
5
10
15
20
25
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 0.1 0.2 0.3 0.4 0.5 0.6
Rela
tive B
indin
g
Energ
y o
f B
[eV
]
Turn
over
Fre
quency,
TO
FB
[s-1
]
Surf
ace C
overa
ge
of A
, θ
A[-
]
Time [s]
C
D
Figure 2. Dynamic Catalysis – Conditions and Surface Response. A. Transient variation of the catalyst surface (α
= 0.8) binding between a maximum and minimum binding energy comprises the overall surface amplitude resulting
in dynamic performance with optimum (purple) turnover frequency at the reaction resonance frequency. B. The
catalyst binding energy changes as a square wave below the resonant frequency (f = 10 Hz), resulting in maximum
and minimum surface coverage of surface intermediate B* and A*; loading and unloading of B from the surface
produces transient B production rates.
____________________________________________________________________________ Ardagh, et al. Page 4
The optimal turnover frequencies are depicted for
each state at the given amplitude as purple points,
while the minimum turnover frequencies below the
static optimum are identified as green points.
The response of the substrate on the catalyst
surface depends on the relative dynamics of the
system to the kinetics of the surface steps (i.e.,
reactions, desorption). For a catalyst oscillating
between two states as a square waveform with
amplitude of ΔU and frequency (f ~ τ-1), the
optimum time-averaged turnover frequency will
occur when the time scale of each state is
approximately the same as the time scale of the
individual surface steps. Referred to here as
“surface resonance”, the resonant frequencies
depicted in Figure 2A permit the surface coverage
of B* to vary (θmin < θB < θmax) without stabilization
before switching surface states.
In this work, the concept of ‘dynamic catalysis’
illustrated in Figure 2A is explored for a broad
range of catalyst and dynamic applied conditions to
understand the connection between catalyst-system
design combinations and catalytic turnover
frequency. For BEP relations identified in Figure
1B (0 < α < 1.0), we present the averaged turnover
frequency for a broad range of conditions including
applied frequency (f), surface energy amplitude
(ΔU), and wave shape (e.g. square versus
sinusoidal). Optimal performance is then identified
with the constraints of practical implementation in
mind.
Results and Discussion. The following results
were generated computationally using Matlab (see
Methods section for more details). Turnover
frequencies (TOFs) were calculated at 1% yield of
product (B) in all cases for a CSTR reactor setup.
TOFB can be defined as follows,
TOFB = [B]q̇
# of active sites [=] s−1 (2)
where q̇ is the volumetric flow rate through the
reactor and [B] is the concentration of component
B in the reactor effluent.
The impact of oscillating the surface binding
energy of B* with time is depicted in Figures 2B-
2D for a square waveform of amplitude ΔU ~ 0.6
eV and frequency of f ~ 10 Hz. The square
waveform of surface binding energies of B*
depicted in Fig. 2B was simulated for a perfectly
mixed reactor operating at 1% yield of B, 150 °C,
and 100 bar A inlet. These conditions produce the
instantaneous turnover frequency depicted in Fig.
2C which ranges from 3-19 s-1 in a complex
oscillating form; the TOFB achieves a maximum of
19 s-1 soon after the binding energy of B* switches
to relatively low energy (BEB ~ 0.9 eV), while the
minimum TOFB of 3 s-1 occurs just before BEB
switches from 1.5 to 0.9 eV. These turnover
frequencies of B are below the predicted resonance
frequencies identified in purple in Fig. 2A (~100 s-
1). An explanation for the lower-than-expected
TOFB is provided by the surface coverages of Fig.
2C. At 10 Hz, the surface coverage of A* achieves
complete oscillation between θA ~ 0 and θA ~ 1;
moreover, the surface coverage of A* stabilizes for
a significant fraction of the period of oscillation,
indicating a period (~ 0.02 s) with negligible change
in the surface composition of the catalyst. In other
words, faster oscillation above 10 Hz of the surface
binding energy of B* should more efficiently utilize
the catalyst.
The TOFB of Fig. 2C indicate that the highest
rates occur when surface states flip from high to
low binding energy of B*. The energetic path
leading to this unloading of the surface of B* is
depicted in Figure 3A. In the initial strong binding
state 1, A adsorbs to the surface as A* and forms a
thermodynamic distribution with state B*. When
the surface flips to weaker-binding state 2, B*
readily desorbs with lower activation energy to
form product B. By these two states, the complete
cycle can be interpreted as filling of the surface
sites (state 1) followed by forced desorption (state
2), the overall rate of which is determined by the
surface frequency and amplitude associated with
the surface binding energies of the two states.
The impact of the surface state-flipping
frequency on the time-averaged turnover frequency
is depicted in Figure 3B for fixed square waveform
amplitude (ΔU = 0.6 eV). At low frequencies (10-4
< f < 10-2 Hz), the average TOFB is an average of
the static conditions of the two states (i.e., a slow
catalyst). At the corner frequency (fC1) of ~0.02 Hz,
the average turnover frequency begins to increase
until the dynamic system eventually matches the
optimal turnover frequency of the static system at
the volcano apex (depicted in red). Further
increasing the surface waveform frequency
increases the average turnover frequency until
maximizing over a range of dynamic resonance
(~103 < f < ~107), identified in Figure 3B in purple.
Above the resonance frequency band, the average
____________________________________________________________________________ Ardagh, et al. Page 5
turnover frequency decreases before stabilizing at
2.6•10-1 s-1 at a waveform frequency of ~1011 Hz,
the TOFB associated with optimal conditions of the
static system at the volcano curve optimum.
For the volcano curve system depicted in Fig.
2B with amplitude of 0.6 eV square waveform, the
instantaneous TOFB is depicted in Figure 3C for
four frequencies: 0.001 Hz in green, 0.25 Hz in red,
10 Hz in blue, and 1,000 Hz in purple. At low
frequency (0.001 Hz), the surface coverages of A*
and B* (in SI Fig. S2) rapidly respond to the change
in surface state, with static operation occurring in
either of the two states. Low frequency below fC1
results in TOFB response comparable to a mix of the
two low activity states identified in green in Fig.
2A. The unique behavior to the general TOFB
response exists only at the condition of flipping
surface states from strong to weak binding of B*;
as noted in the highlighted region of Fig. 3C for
0.001 Hz, the TOFB overshoots, resulting from the
unloading of surface B* species into the gas phase
as product B. As the waveform frequency increases
to 0.25 and 10 Hz, the unloading of B* species from
the surface becomes the dominant mechanism
leading to catalyst activity. For these two
frequencies, the TOFB and the surface coverages of
A* and B* (Fig. S3-S4) are transient for most of the
waveform period. At 1,000 Hz in Fig. 3C, the
0.00001
0.0001
0.001
0.01
0.1
1
10
100
Insta
nta
neo
us T
urn
ove
r F
req
uen
cy
to B
[s
-1]
102
101
1
10-1
10-2
10-310-3
10-4
10-5
0.001 Hz
0.25 Hz
10 Hz
1000 Hz
1000
sec
4.0 s 0.1 s
Time (Different Scales)
Insta
nta
neous T
urn
over
Fre
quency to B
[s
-1]
-160
-140
-120
-100
-80
-60
-40
-20
0
20
40
En
tha
lpy [
kJ
mo
l-1]
A
B
A*
State 2
B*
State 1
Osc
illa
tio
n
Fre
qu
en
cy
B*
TS
En
thalp
y [k
J m
ol-1
]A B
D C
0.01
0.10
1.00
10.00
100.00
-4.00 -2.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00
Ave
rage T
urn
ove
r F
requ
en
cy,
B [
s-1
]
Frequency [s-1]
Activity
Response
Static
Optimum
10-2 1 102 104 106 108 1010 101210-2
10-4
10-1
1
101
102
fC1 fC2
Ave
rag
e T
urn
ove
r F
req
uen
cy,
B [s
-1]
Waveform Amplitude [eV]
Wavefo
rm F
req
uen
cy [
s-1
]
0 0.2 0.4 0.6 0.8 1.0
10-2 10-1 100 101 102 103
Average Turnover Frequency, B [s-1]
1010
108
106
104
102
1
10-2
10-4
Figure 3. Activity response of applied oscillating surface binding energy - square waveform. A. Oscillating state
energy diagram for A reacting on a catalytic surface to B product (-0.1 to 0.5 eV of B*). B. Average catalytic turnover
frequency to product B at waveform amplitude of 0.6 eV at 150 °C and 100 bar; resonance frequencies identified in
purple. C. Instantaneous turnover frequency to B for four frequencies at ΔU of 0.6 eV at 150 °C and 100 bar. D.
Average turnover frequency to B at 150 °C and 100 bar for variable square waveform amplitude and frequency.
____________________________________________________________________________ Ardagh, et al. Page 6
TOFB and surface coverages of A* and B* (Fig. S5)
are always transient; under these conditions TOFB
and surface coverages only minimally oscillate in
value (e.g., 0.29 < θB < 0.32).
An interpretation of catalytic surface resonance
comes from evaluating the TOFB response of each
condition independently, as shown in Fig. 3C. The
rate of production for surface species B* is defined
by the forward surface reaction rate constant and
surface coverage of A* (which is in equilibrium
with gas phase A). Similarly, the rate of desorbing
B* to gas product B is defined by the desorption
rate constant and surface coverage of B*. In this
case, the time scales of these two processes sum to
the total time scale, which is comparable to the
applied square waveform time scale at resonance.
This concept is visually observed in Fig. 2A, where
TOFB for the two purple points predict 60 s-1 for
each independent process, while the actual TOFB
predicted by simulation is exactly half of that value,
29 s-1 (Fig. 3B and 3C). To put it simply, catalytic
surface resonance occurs when the frequency of the
applied surface state-switching waveform matches
the natural frequency of the catalytic kinetics.
Variation of the surface square waveform
amplitude will change the kinetics of the surface
chemistry, resulting in a shift of the resonance
frequency band. As depicted in the heat map of Fig.
3D, a range of amplitudes (0 < ΔU < 1.0 eV) was
evaluated for the volcano curve of Fig. 2A for
frequencies varying over 15 orders of magnitude
(10-4 < f < 101 s-1) to determine the average steady
state turnover frequency to B, TOFB. The steady
state system response is oscillatory, so the average
TOF was calculated via integration over a range of
time after the system reached stable oscillation. For
each value of the oscillation amplitude ΔU, the two
extreme values of U [eV] corresponding to the two
states of the square surface waveform were selected
to yield two conditions of equal rate; put simply,
each value of ΔU should produce a horizontal line
connecting two purple points as in Figure 2A. The
variation in surface kinetics with square waveform
frequency and amplitude is visually apparent in
Figure 3D, where low frequencies below 0.1 Hz are
1.00E-14
1.00E-13
1.00E-12
1.00E-11
1.00E-10
1.00E-09
1.00E-08
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1.00E-03
1.00E-02
1.00E-01
1.00E+00
1.00E+01
1.00E+02
1.00E+03
1.00E+04
1.00E+05
1.00E+06
1.00E+07
1.00E+08
1.00E+09
1.00E+10
-1.0 -0.5 0.0 0.5 1.0 1.5
Relative Binding Energy, B [eV]
Turn
over
Fre
quency,
B [s
-1]
α = 1.0
100
10-3
10-6
10-9
10-12
103
106
109
0.6 eV1.0 eV
1.5 eV
A
Avera
ge T
urn
over
Fre
quency,
B [s
-1]
0
10
20
30
40
50
60
70
80
90
100
1.E-04 1.E-02 1.E+00 1.E+02 1.E+04 1.E+06 1.E+08 1.E+10
0.6 eV
Avera
ge T
urn
over
Fre
quency,
B [s
-1]
Square Wave Frequency [s-1]
10-4 10-2 100 102 104 106 108 1010
α = 1.0
1.00E+001.00E+021.00E+041.00E+06
1
2
31.5 eV
1.0 eV
0.6 eV
106104102100
Resonance Turnover Frequency [s-1]
B
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1.00E-03
1.00E-02
1.00E-01
1.00E+00
1.00E+01
1.00E+02
1.00E+03
1.00E+04
1.00E+05
1.00E+06
1.00E+07
-0.5 0.0 0.5 1.0 1.5
Relative Binding Energy, B [eV]
100
10-2
10-4
10-6
102
104
106
Turn
over
Fre
quency,
B [s
-1]
0.6 eV1.0 eV
1.5 eV
α = 0.4C
0
2
4
6
8
10
12
1E-04 1E-02 1E+00 1E+02 1E+04 1E+06 1E+08 1E+10
Square Wave Frequency [s-1]
α = 0.4
1010108106104102110-210-4
0.6 eV
1.00E+001.00E+021.00E+041.00E+06
1
2
31.5 eV
1.0 eV
0.6 eV
106104102100
Resonance Turnover Frequency [s-1]
D
1.00E-15
1.00E-14
1.00E-13
1.00E-12
1.00E-11
1.00E-10
1.00E-09
1.00E-08
1.00E-07
1.00E-06
1.00E-05
1.00E-04
1.00E-03
1.00E-02
1.00E-01
1.00E+00
1.00E+01
1.00E+02
1.00E+03
1.00E+04
1.00E+05
1.00E+06
1.00E+07
1.00E+08
-1.0 -0.5 0.0 0.5 1.0 1.5
Turn
over
Fre
quency,
B [s
-1]
Relative Binding Energy, B [eV]
α = 0
0.6 eV1.5 eV
108
104
1
10-4
10-8
10-12
E
1.00E+001.00E+021.00E+041.00E+06
1
2
3-0.43 to +0.17 eV
-0.43 to +1.07 eV
-0.75 to +0.75 eV
106104102100
Maximum Turnover Frequency [s-1]
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
1E-04 1E-01 1E+02 1E+05 1E+08 1E+11
α = 0
Avera
ge T
urn
over
Fre
quency,
B [s
-1]
Square Wave Frequency [s-1]
0.6 eV
10-4 10-1 10+2 10+5 10+8 10+11
F
Figure 4. Dynamic Catalytic Response of Three Linear-Scaling Relationships (α of 1.0, 0.4, 0.0). All panels
comprised of an A-to-B reaction at 150 °C, 100 bar, perfectly mixed system operating at 1 % yield of product B with
BEP relationship parameter α as the sole variable. A. Volcano plot for α of 1.0 system. B. Catalytic turnover frequency
response of α of 1.0 system. C. Volcano plot for the α of 0.4 system. D. Catalytic turnover frequency response of the
α of 0.4 system. E. Volcano plot for the α of zero system. F. Catalytic turnover frequency response of the α of zero
system.
____________________________________________________________________________ Ardagh, et al. Page 7
slower (dark blue) than static catalysis for
amplitudes greater than 0.1 eV. Alternatively,
above ~1 Hz, the average turnover frequency
increases dramatically to 10 and 1,000 s-1 per
catalytic site for amplitudes above 0.3 eV.
The ability to dynamically accelerate catalytic
turnover depends on the energetics of the
obtainable states defined by the shape of the
volcano curve. Of the many parameters that define
the volcano shape, the linear-scaling relationship
parameter, α, relating the surface reaction enthalpy
to the surface reaction activation energy can
dramatically shift the slope of the volcano plot.
While Figures 2 and 3 describe a system with α of
0.8, three volcano plots for α of 1.0, 0.4. and zero
are shown in Figure 4. For steep volcano plots such
as Fig. 4A, extension of the slopes as dashed lines
above the volcano apex indicate rapid increase in
the turnover frequency for amplitudes of 0.6, 1.0
and 1.5 eV at resonance conditions. This is
supported by the catalytic reactor simulation
kinetics of Fig. 4B, which considered the catalyst
system of Figure 4A at variable applied square
waveform frequency (10-4 < f < 1011). For a square
waveform at an amplitude of 0.6 eV, the resonance
frequencies of 103 to 107 s-1 yield an average
turnover frequency to B of about 52 s-1 per catalytic
site. At higher amplitudes of 1.0 and 1.5 eV, the
average turnover frequency per catalytic site at
resonance achieves 2,074 and 2.0•105 s-1.
A broader volcano of α of 0.4 in Fig. 4C limits
the overall speed achievable for a given amplitude;
the purple points above the curve are further apart
and at lower turnover frequencies. This translates to
lower overall reaction rates at resonance conditions
as shown in Fig. 4D. At the extreme case where the
activation energy of the surface reaction does not
change with the binding energy of an adsorbate
such as the volcano curve of Fig. 4E, the potential
of dynamic operation is limited as shown in Fig. 4F.
There exist at least two cases where the slope of the
volcano curve is horizontal on one side as drawn:
(1) catalytic systems where the surface reaction
enthalpy does not change with the binding energies
of the descriptor component (e.g., B*), thus leading
to constant surface activation energy of reaction,
and (2) systems with α of zero. In these rare cases,
the rate of the surface reaction can never be
accelerated to match a fast rate of desorption, and
the average overall turnover frequency is limited to
the rate of the surface reaction.
Applying dynamic operation to heterogeneous
catalytic applications will require identifying the
conditions of optimal performance in addition to
new design variables such as surface waveform
shape that can be implemented in real reactor
technology. As depicted in Fig. 5A, the sinusoidal
surface binding waveform varying from -0.1 to
+0.5 eV at frequency of 10 Hz applied to the
catalyst system characterized by the volcano plot of
Fig. 2A yields oscillatory turnover frequency (Fig.
5B) and surface coverage of A* (Fig. 5C) at 100 bar
A inlet, 150 °C, and 1% yield of B. Similar to the
case with the square waveform, the turnover
frequency of B increases when the applied
waveform changes from strong binding of B* to
weak binding. At the same time, the surface
coverage of A* increases to take the place of the B*
that was removed from the surface as desorbed
product B. Other considered waveform types
including triangle and sawtooth are depicted in Fig.
5D. For all conditions, the square waveform
exhibits superior activity at surface resonance
conditions. At higher frequencies of 10 and 1,000
Hz, the sinusoidal waveform outperforms the
triangle and sawtooth shapes.
Implementation of dynamic operation of
heterogeneous catalysts requires the capability to
modify the binding energy of surface intermediates
with time. Based on the simulations of Figs. 2-5,
catalyst and system parameters should be selected
to achieve average turnover frequencies above the
optimum of static conditions and preferably as high
as resonance conditions. This implies that a
physical catalyst system must achieve surface
waveform amplitudes of at least 0.3 eV (and
preferably above 0.5 eV) and operating frequencies
above 10 Hz (and preferably 100-1,000 Hz). These
performance targets change with the selected
surface chemistry, which will likely have more than
two surface intermediates exhibiting linear scaling
relationships over a broad range (0.2 < α < 0.8). The
complexity of each catalytic chemistry combined
with the large number of dynamic catalysis
parameters indicates that each system can be guided
by the principles proposed here but will require
detailed microkinetic modeling for design and
optimization.
Device construction for tuning of the surface
intermediate adsorbate binding energy can be
interpreted via the electronic state of the catalyst
material; surface intermediates such as adsorbed
____________________________________________________________________________ Ardagh, et al. Page 8
nitrogen, N*, correlate linearly with the d-band
edge/center when compared across a broad range of
metals[17,50,51,52]. Temporal variation of metal d-
bands exists in at least two categories including
electronic and physical (and even electro-
mechanical) manipulation. Straining of surfaces
has been shown to shift the d-band centers of
metals, metal alloys, and other 2D materials[53],
which alters the binding energy of adsorbates such
as carbon monoxide[54]. When combined with
dynamic approaches such as sound waves or
piezoelectrics capable of 1% strain oscillation
exceeding kilohertz frequencies, this approach can
potentially achieve the frequencies and amplitudes
required for resonant dynamic catalytic
acceleration. Other methods electronically
manipulate a catalyst surface including field effect
modulation[55,56,57] or non-Faradaic electrochemical
modification[58,59,60,61,62], both of which can
potentially achieve the frequency and amplitude
targets necessary for surface catalytic resonance.
Future work will expand the principles developed
here to more complex surface chemistries in
parallel with implementation in physical catalytic
reactors.
Methods. Continuously-stirred tank reactor
(CSTR – perfect mixing assumed) models were
implemented in Matlab 2017b and Matlab 2018b,
using GPU computing resources at the Minnesota
Supercomputing Institute (MSI) at the University
of Minnesota. See SI section S1 for the full shell
and reactor codes, which can be used and modified
to reproduce these simulations. The shell code set
reactor parameters included the temperature (T),
inlet volumetric flow rate (q̇), catalyst weight (w),
and active site loading. Reactor time-on-stream
data was generated using the Matlab ODE15s
differential equation solver. This solver was
selected based on its performance; see SI section S2
for a comparison with Matlab ODE45, ODE23s,
ODE23t, and ODE23tb. The set of differential
equations consisted of forward and reverse rates for
the consumption of gas phase (A, B) and surface
species (A*, B*). This general reaction system, A
↔ B, was modeled using three reversible
elementary steps: (i) adsorption of A, (ii)
conversion of A* to B*, and (iii) desorption of B
with a site balance involving *, A*, and B*.
A ⇌ A* (3)
A* ⇌ B* (4)
B* ⇌ B (5)
0.01
0.1
1
10
100
0.001 0.25 10 1000
Ave
rag
e T
urn
ove
r F
req
ue
ncy,
B [s
-1]
Waveform Frequency [s-1]
A
B
C
D
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0
2
4
6
8
10
12
14
16
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
Rela
tive B
ind
ing
En
erg
y o
f B
[eV
]
Tu
rno
ver
Fre
qu
en
cy
B [s
-1]
Su
rface C
ove
rag
e
of A
* [-
]
Time [s]
Figure 5. Surface B* Binding Energy Waveform. Volcano curves comprised of ΔU ~ 0.6 eV, 150 °C, perfectly
mixed reactor at 1% yield of B product. A. Relative binding energy of B* varying in sinusoidal waveform at f~10 Hz.
B. Turnover frequency to B response to a 10 Hz sinusoidal waveform. C. Surface coverage of A* response to a 10
Hz sinusoidal waveform. D. Comparison of average turnover frequency to B for an applied B* surface binding energy
oscillation of four waveform types: square (blue), sinusoidal (orange), triangle (green), and sawtooth (red).
____________________________________________________________________________ Ardagh, et al. Page 9
[Sites] = [∗] + [A]∗ + [B]∗ (6)
Generalized forms of the differential equation used
for each gas phase and surface species are:
d[A]
dt=
q̇
V([𝐴]feed − [A]) − r1,forward + r1,reverse
(7)
d[A]∗
dt= r1,forward − r1,reverse + r2,forward −
r2,reverse (8)
Reaction rate equations consisted of rate constants
and species concentrations, with each elementary
step assumed to be 1st order in all participating
reactants. Since this was modeled as a gas phase
reaction, adsorption steps were expressed in terms
of A and B pressures (bar).
𝑟𝑎𝑑𝑠,𝑖 = 𝑘𝑎𝑑𝑠,𝑖𝑃𝑖[∗] (9)
𝑟𝑑𝑒𝑠,𝑖 = 𝑘𝑑𝑒𝑠,𝑖𝜃𝑖[𝑠𝑖𝑡𝑒𝑠] (10)
𝑟𝑠𝑢𝑟𝑓 𝑟𝑥𝑛 𝑓𝑜𝑟𝑤𝑎𝑟𝑑 = 𝑘𝑠𝑢𝑟𝑓 𝑓𝑥𝑛 𝑓𝑜𝑟𝑤𝑎𝑟𝑑𝜃𝐴[𝑠𝑖𝑡𝑒𝑠]
(11)
Rate constants were constructed as Arrhenius
expressions using pre-exponential factors and
activation energies for adsorption, desorption, and
surface reactions[63]. Pre-exponential factors were
set to 106 (bar-s)-1 for adsorption steps and 1013 s-1
for surface reaction and desorption steps. These
values are typical starting points when fitting
microkinetic models to experimental data.
Activation energy was set to 0 kJ/mol for
adsorption and to the binding energies (BEs) of A
and B for their respective desorption steps. The
binding energies for A and B, the surface reaction
activation energy (Ea), and the surface enthalpy of
reaction were selected; the base conditions were
BEA = 1.3 eV, BEB = 1.0 eV, Ea = 102 kJ mol-1, ΔHsr
= 0 kJ/mol, and ΔHovr = -20 kJ/mol.
Brønsted-Evans-Polyani relationships between
Ea and BEs were held at a constant offset (E0) of
102 kJ/mol and the slope of the relationship, α, was
varied (0 < α < 1.0). These values fall within
previously observed ranges for experimentally
derived BEP parameters and resulted in volcano
peak TOFs between 0.1-10 s-1. Thus, the activation
energy was expressed as a linear function of the
surface enthalpy of reaction, ΔHsr (i.e., the
difference in binding energies between A* and B*
plus the overall enthalpy of reaction from A B):
Ea = α*ΔHsr + E0 (12)
Volcano plots were generated by varying ΔHsr and
measuring the time-averaged turnover frequency
(TOF) at 1.0 % overall yield of B. Turnover
frequency was defined as 𝑇𝑂𝐹 = [𝐵]q̇
# 𝑜𝑓 𝑠𝑖𝑡𝑒𝑠 for
the CSTR design equation, so in practice q̇ (the gas
flowrate [=] L/s) was adjusted until the outlet yield
of component B was 1.0 %. Variation in the BEP
slope (0 < α < 1.0) resulted in surface reaction
activation energies (25 < Ea < 170 kJ/mol) between
binding energies of 0.5 and 2.0 eV, which are
comparable with a broad range of reactions
described in the literature.
Dynamic catalysis was simulated by running
ODE15s for a system where BEs varied with time
on stream as square, sinusoidal, triangle, or
sawtooth waves. Binding energy changes for A*
and B* were specified in the shell code, and these
shifts resulted in varying ΔHsr and Ea. Oscillation
period/frequency was set by specifying the time
duration spent at each condition. Reported TOFs
were calculated when the system oscillation was
centered on 1.0 % yield of B and after the reactor
had achieved oscillatory steady state, defined as a
steady time-averaged turnover frequency. Example
time on stream data for dynamic catalysis can be
found in the supporting information section S4.
Plots of the average turnover frequency as a
function of surface binding energy oscillation
amplitude and frequency (i.e., heat maps) were
generated in Matlab 2018b using the jet color
scheme to indicate low and high TOF, ranging from
dark blue to dark red. The heat map data consists of
a modified Akima cubic Hermite fit through
discrete data points calculated at 0-1.0 eV ΔBE.
Data points are tabulated in supporting information
section S5. This data was obtained for symmetric
dynamic catalysis starting at the volcano peak (ΔBE
~ -0.05 to 0.05 eV) and oscillating the same
amplitude in each direction (from 0-0.75 eV). Data
was also obtained for asymmetric dynamic catalysis
where the end points were chosen based on
extrapolated linear fits of each side of the volcano
curve. These lines were set equal with a specified
____________________________________________________________________________ Ardagh, et al. Page 10
oscillation amplitude between 0-1.5 eV, and the end
points were chosen by drawing a vertical line down
to the volcano plot. Frequency response figures
were generated for scenarios with varying BEP
relationships where the BEP slope ranged from zero
to one.
Acknowledgements. We acknowledge financial
support of the Catalysis Center for Energy
Innovation, a U.S. Department of Energy - Energy
Frontier Research Center under Grant DE-
SC0001004. The authors acknowledge the
Minnesota Supercomputing Institute (MSI) at the
University of Minnesota for providing resources
1 C.R. Catlow, M. Davidson, C. Hardacre, G.J.
Hutchings, “Catalysis is making the world a better
place,” Philosophical Transactions of the Royal
Society A, 2016, 374(2061), 1-2. 2 K. Narsimhan, K. Iyoki, K. Dinh, Y. Roman-Leshkov,
“Catalytic Oxidation of Methane into Methanol over
Copper-Exchanged Zeolites with Oxygen at Low
Temperature,” ACS Central Science 2016, 2(6), 424-
429. 3 A.J. Medford, J. Wellendorff, A. Vojvodic, F. Studt,
F. Abild-Pedersen, K.W. Jacobsen, T. Bligaard, J.K.
Norskov, “Assessing the reliability of calculated
catalytic ammonia synthesis rates,” Science 2014,
345(6193), 197-200. 4 C.A. Busacca, D.R. Fandrick, J.J. Song, C.H.
Senanayake, “The Growing Impact of Catalysis in
the Pharmaceutical Industry,” Advanced Synthesis &
Catalysis 2011, 353(11-12), 1825-1864. 5 Arora, S. S.; Nieskens, D. L. S.; Malek, A.; and Bhan,
A.; "Lifetime improvement in methanol-to-olefins
catalysis over chabazite materials by high-pressure
H2 co-feeds", Nature Catalysis 2018, 1, 666-672 6 R.E. Patet, N. Nikbin, C.L. Williams, S.K. Green,
C.C. Chang, W. Fan, S. Caratzoulas, P.J.
Dauenhauer, D.G. Vlachos, "Kinetic Regime Change
in the Tandem Dehydrative Aromatization of Furan
Diels-Alder Products," ACS Catalysis 2015, 5(4),
2367-2375. 7 P. Szuromi, “Upgrading CO2 with methane,” Science
2016, 354(6311), 429. 8 C. Paolucci, I. Khurana, A. A. Parekh, S. Li, A. J.
Shih, H. Li, J. R. Di Iorio, J. D. Albarracin-Caballero,
A. Yezerets, J. T. Miller, W. N. Delgass, F. H.
Ribeiro, W. F. Schneider*, R. Gounder, "Dynamic
Multinuclear Sites Formed by Mobilized Copper Ions
in NOx Selective Catalytic Reduction", Science,
2017, 357, 898-903. 9 M.A. Hillmyer, “The Promise of Plastics from
Plants,” Science 2017, 358(6365), 868-870.
that contributed to the research results reported
within this paper. URL: http://www.msi.umn.edu
We acknowledge helpful discussions with
Professors Turon Birol, Dan Frisbie, Michael
Tsapatsis, and Dionisios Vlachos.
Keywords. Catalysis, Sabatier, Dynamics,
Frequency, Resonance, Volcano
Supporting Information. Additional information
including computer code, time-on-stream data, and
simulation methods are included in the supporting
information.
10 Omar Abdelrahman, D.S. Park, K. Vinter,
C.Spanjers, L. Ren, H.J. Cho, D.G. Vlachos, W. Fan,
M. Tsapatsis, P.J. Dauenhauer, "Biomass-Derived
Butadiene by Dehydra-Decyclization of
Tetrahydrofuran," ACS Sustainable Chemistry and
Engineering. 2017, 5(5), 3732-3736. 11 U.S. Department of Energy. “Basic Research Needs
for Catalysis Science to Transform Energy
Technologies: Report from the U.S. Department of
Energy, Office of Basic Energy Sciences Workshop
May 8-10, 2017, in Gaithersburg, Maryland”.
Retreived Feb. 3, 2017. Link:
https://science.energy.gov/~/media/bes/pdf/reports/20
17/BRN_CatalysisScience_rpt.pdf 12 J.F. Chen, Y. Mao, H.F. Wang, P. Hu, “A simple
method to locate the optimal adsorption energy for
the best catalysts directly,” ACS Catalysis 2019, DOI:
10.1021/acscatal.8b04896 13 J. Kari, J.P. Olsen, K. Jensen, S.F. Badino, K.B.R.M.
Krogh, K. Borch, P. Westh, “Sabatier Principle for
Interfacial (Heterogeneous) Enzyme Catalysis,” ACS
Catalysis 2018, 8(12), 11966-11972. 14 A. Vojvodic, J.K. Norskov, “New design paradigm
for heterogeneous catalysts,” National Science
Review 2015, 2, 140-149. 15 A.T. Bell, “The Impact of Nanoscience on
Heterogeneous Catalysis,” Science 2003, 299(5613),
1688-1691. 16 P. Sabatier, La catalyse en chimie organique,
Librairie Polytechnique, Paris et Liège, 1920. 17 A.J. Medford, A. Vojvodic, J.S. Hummelshoj, J.
Voss, F. Abild-Pedersen, F. Studt, T. Bligaard, A.
Nilsson, J.K. Norskov, “From the Sabatier principle
to a predictive theory of transition-metal
heterogeneous catalysis,” Journal of Catalysis 2015,
328, 36-42. 18 A.A. Balandin, V.A. Ferapontov, A.A.
Tolstopyatova, “The activity of cadmium oxide as a
catalyst for hydrogen dehydrogenation,” Bulletin of
____________________________________________________________________________ Ardagh, et al. Page 11
the Academy of Sciences of the USSR, Division of
Chemical Science. 1960, 9(10), 1630-1636. 19 A.A. Balandin, “The multiplet theory of catalysis.
Energy factors in catalysis,” 1964, 33(5), 549-579. 20 A.K. Vijh, “Sabatier-Balandin interpretation of the
catalytic decomposition of nitrous oxide on metal-
oxide semiconductors,” Journal of Catalysis 1973,
31, 51-54. 21 A. Vijh, “Interpretation of some aspects of catalytic
oxidation of ammonia and propylene by
semiconducting oxides in terms of volcano-shaped
plots,” Journal de Chimie Physique et de Physico-
Chimie Biologique 1973, 70(10), 1444-1447. 22 H. Toulhoat, P. Raybaud, S. Kasztelan, G. Kresse, J.
Hafner, “Transition metals to sulfur binding energies
relationship to catalytic activities in HDS: back to
Sabatier with first principle calculations,” Catalysis
Today 1999, 50, 629-636. 23 R.R. Chianelli, G. Berhault, P. Raybaud, S.
Kasztelan, J. Hafner, H. Toulhoat, “Periodic trends in
hydrodesulfurization: in support of the Sabatier
principle,” Applied Catalysis A: General, 2002, 227,
83-96. 24 S. Ichikawa, “Volcano-shaped curves in
heterogeneous catalysis,” Chemical Engineering
Science 1990, 45(2), 529-535. 25 A. Logadottir, T.H. Rod, J.K. Norskov, B. Hammer,
S. Dahl, C.J.H. Jacobsen, “The Bronsted-Evans-
Polyanyi relation and the volcano plot for ammonia
synthesis over transition metal catalysts,” Journal of
Catalysis 2001, 197, 229-231. 26 A. Vijh, “Volcano relationships in catalytic reactions
on oxides,” Journal of Catalysis, 1974, 33, 385-391. 27 M.A. Barteau, “Linear free energy relationships for
C1-oxygenate decomposition on transition metal
surfaces,” Catalysis Letters, 1991, 8, 175-184. 28 A.J. Medford, A. Vojvodic, J.S. Hummelshoj, J.
Voss, F. Abild-Pedersen, F. Studt, T. Bligaard, A.
Nilsson, J.K. Norskov, “From the Sabatier principle
to a predictive theory of transition-metal
heterogeneous catalysis,” Journal of Catalysis 2015,
328, 36-42. 29 T.R. Munter, T. Bligaard, C.H. Christensen, J.K.
Norskov, “BEP relations for N2 dissociation over
stepped transition metal and alloy surfaces,” Physical
Chemistry Chemical Physics 2008, 10, 5202-5206. 30 M. Salciccioli, M. Stamatakis, S. Caratzoulas, D.G.
Vlachos, “A review of multi-scale modeling of
metal-catalyzed reactions: Mechanism development
for complexity and emergent behavior,” Chemical
Engineering Science 2011, 66, 4319-4355. 31 R.P. Bell, “The theory of reactions involving proton
transfers,” Proceedings of the Royal Society A, 1936,
154(882), 414. 32 M.G., Evans, M.J. Polanyi, “Further considerations
on the thermodynamics of chemical equilibria and
reaction rates,” Transactions of the Faraday Society,
1936, 32, 1333. 33 J.E. Sutton, D.G. Vlachos, “A theoretical and
computational analysis of linear free energy relations
for the estimation of activation energies,” ACS
Catalysis 2012, 2(8), 1624-1634. 34 S. Wang, V. Petzold, V. Tripkovic, J. Kleis, J.G.
Howalt, E. Skulason, E.M. Fernandez, B. Hvolbaek,
G. Jones, A. Toftelund, H. Falsig, M. Bjorketun, F.
Studt, F. Abild-Pedersoen, J. Rossmeisl, J.K.
Norskov, T. Bligaard, “Universal transition state
scaling relations for (de)hydrogenation over
transition metals,” Phys. Chem. Chem. Phys. 2011,
13, 20760-20765. 35 S. Wang, B. Temel, J. Shen, G. Jones, L.C. Grabow,
F. Studt, T. Bligaard, F. Abild-Pedersen, C.H.
Christensen, J.K. Norskov, “Universal Brønsted-
Evans-Polyani relations for C-C, C-O, C-N, N-O, N-
N, and O-O dissociation reactions,” Catalysis Letters,
2011, 141(3), 370-373. 36 A. Vojvodic, F. Calle-Vallejo, W. Guo, S. Wang, A.
Toftelund, F. Studt, J.I. Martinez, J. Shen, I.C. Man,
J. Rossmeisl, T. Bligaard, J.K. Norskov, F. Abild-
Pedersen, “On the behavior of Bronsted-Evans-
Polanyi relations for transition metal oxides,” Journal
of Chemical Physics 2011, 134, 244509. 37 D. Loffreda, F. Delbecq, F. Vigne, P. Sautet, “Fast
prediction of selectivity in heterogeneous catalysis
from extended Brønsted-Evans-Polanyi relations: A
theoretical insight,” Angewandte Chemie 2009, 121,
9140-9142. 38 M.A. Barteau, “Linear free energy relationships for
C1-oxygenate decomposition on transition metal
surfaces,” Catalysis Letters 1991, 8, 175-184. 39 K.T. Winther, M.J. Hoffmann, O. Mamun, J.R. Boes,
J.K. Norskov, M. Bajdich, T. Bligaard, “Catalysis-
Hub.org: An open electronic structure database for
surface reactions,” Version 2. ChemRxiv. Preprint.
https://doi.org/10.26434/chemrxiv.7336094.v2 40 G. Mpourmpakis, D.G. Vlachos, “Computational-
based catalyst design for thermochemical
transformations,” MRS Bulletin, 2011, 36(3), 211-
215. 41 Carles Bo, Feliu Maseras, and Núria López. “The
role of computational results databases in
accelerating the discovery of catalysts” Nature
Catalysis, 2018, 1, 809-810. 42 D.A. Hansgen, D.G. Vlachos, J.G. Chen, “Using first
principles to predict bimetallic catalysts for the
ammonia decomposition reaction,” Nature Chemistry
2010, 2, 484-489. 43 W.D. Williams, J.P. Greeley, W.N. Delgass, F.H.
Ribeiro, “Water activation and carbon monoxide
coverage effects on maximum rates for low
temperature water-gas shift catalysis,” Journal of
Catalysis, 2017, 347, 197-204.
____________________________________________________________________________ Ardagh, et al. Page 12
44 J.A. Herron, M. Mavrikakis, C.T. Maravelias,
“Optimization methods for catalyst design,”
Computer Aided Chemical Engineering, 2016, 38,
295-300. 45 T.Z.H. Gani, H.J. Kulik, “Understanding and
breaking scaling relations in single site catalysis:
Methane to methanol conversion by FeIV=O” ACS
Catalysis 2018, 8(2), 975-986. 46 M.T. Darby, M. Stamatakis, A. Michaelides, E.C.H.
Sykes, “Lonely atoms with special gifts: Breaking
linear scaling relationships in heterogeneous
catalysis with single-atom alloys,” J. Phys. Chem.
Lett. 2018, 9, 5636-5646. 47 G. Giannakakis, M. Flytzani-Stephanopoulos, E.C.H.
Sykes, “Single-atom alloys as a reductionist
approach to the rational design of heterogeneous
catalysts,” Accounts of Chemical Research, 2019.
DOI: 10.1021/acs.accounts.8b00490 48 P. Peng, Y. Li, Y. Cheng, S. Deng, P. Chen, R. Ruan,
“Atmospheric pressure ammonia synthesis using non-
thermal plasma-assisted catalysis,” Plasma Chemistry
and Plasma Processing 2016, 36(5), 1201-1210. 49 P. Mehta, P. Barboun, F.A. Herrera, J. Kim, P.
Rumbach, D.B. Go, J.C. Hicks, W.F. Schneider,
“Overcoming ammonia synthesis scaling relations
with plasma-enabled catalysis,” Nature Catalysis
2018, 1, 269-275. 50 L.G.M. Pettersson, A. Nilsson, “A molecular
perspective on the d-band model: Synergy between
experiment and theory,” Topics in Catalysis 2014,
57(1-4), 2-13. 51 B. Hammer, J.K. Norskov, “Why gold is the noblest
of all metals,” Nature 1995, 376, 238-240. 52 A. Nilsson, L.G.M. Pettersson, B. Hammer, T.
Bligaard, C.H. Christensen, J.K. Norskov, “The
electronic structure effect in heterogeneous
catalysis,” Catalysis Letters, 2005, 100(3-4), 111-
114. 53 Lei Wang, Zhenhua Zeng, Wenpei Gao, Tristan
Maxson, David Raciti, Michael Giroux, Xiaoqing
Pan, Chao Wang, Jeffrey Greeley, “Tunable
intrinsic strain in two-dimensional transition metal
electrocatalysts,” Science 2019, 363(6429), 870-
874.
54 M.F. Francis, W.A. Curtin, “Mechanical stress
combined with alloying may allow continuous
control over reactivity: Strain effects on CO
dissociation and subsequent methanation catalysis
over Ni(211), Ni3Fe(211), and NiFe(112),” Journal
of Physical Chemistry C 2017, 121, 6113-6119. 55 Y. Wang, C.H. Kim, Y. Yoo, J.E. Johns, C.D.
Frisbie, “Field effect modulation of heterogeneous
charge transfer kinetics at back-gated two-
dimensional MoS2 electrodes,” Nano Letters 2017,
17, 7586-7592. 56 C.H. Kim, C.D. Frisbie, “Field effect modulation of
outer-sphere electrochemistry at back-gated, ultrathin
ZnO electrodes,” Journal of the American Chemical
Society 2016, 138, 7220-7223. 57 C.H. Kim, Y. Wang, C.D. Frisbie, “Continuous and
reversible tuning of electrochemical reaction kinetics
on back-gated 2D semiconductor electrodes: steady-
state analysis using a hydrodynamic method,”
Analytical Chemistry 2019, 91, 1627-1635. 58 Imbihl, R. “Electrochemical promotion of catalytic
reactions” Prog. Surf. Sci. 2010, 85, 241 59 Neophytides, S. G.; Vayenas, C. G. “TPD and cyclic
voltammetric investigation of the origin of
electrochemical promotion in catalysis” J. Phys.
Chem. 1995, 99, 17063 60 Ladas, S.; Kennou, S.; Bebelis, S.; Vayenas, C. G.
“Origin of non-faradaic electrochemical modification
of catalytic activity,” J. Phys. Chem. 1993, 97, 8845 61 Basini, L.; Cavalca, C. A.; Haller, G. L. J.
“Electrochemical promotion of oxygen atom back-
spillover from yttria-stabilized zirconia onto a porous
platinum electrode: Detection of SERS signals” J.
Phys. Chem. 1994, 98, 10853. 62 Ladas, S.; Bebelis, S.; Vayenas, C. G. “Work
function measurements on catalyst films subject to in
situ electrochemical promoation,” Surf. Sci. 1991,
251–252, 1062 63 A.B. Mhadeshwar, H. Wang, D.G. Vlachos,
“Thermodynamic consistency in microkinetic
development of surface reaction mechanisms,”
Journal of Physical Chemistry C, 2003, 107, 46,
12721-12733.
download fileview on ChemRxivManuscript_Dauenhauer.pdf (2.03 MiB)
Page S1
SUPPORTING INFORMATION
Principles of Dynamic Heterogeneous Catalysis:
Surface Resonance and Turnover Frequency Response
M. Alexander Ardagh1,2, Omar A. Abdelrahman2,3, Paul J. Dauenhauer1,2*
1University of Minnesota, Department of Chemical Engineering and Materials Science, 421 Washington
Ave SE, Minneapolis, MN 55455 2Catalysis Center for Energy Innovation, University of Delaware, 150 Academy Street, Newark, DE
19716 3University of Massachusetts Amherst, Department of Chemical Engineering, 159 Goessmann
Laboratory, 686 North Pleasant Street, Amherst, MA 01003.
*Corresponding author: [email protected]
# of Figures: 9
# of Tables: 1
# of Equations: 0
Table of Contents:
Section S1. Matlab 2017b and 2018b Code
CSTR Model
Volcano Plot Generator
Dynamic Catalysis Shell Code for Square, Triangle, and Sawtooth Waveforms
Dynamic Catalysis Shell Code for Sinusoidal Waveform
Section S2. Matlab ODE Solver Performance
ODE45
ODE15s
ODE23s
ODE23t
ODE23tb
Section S3. Static Catalysis Time on Stream Data
Example at 1 % Yield of B
Section S4. Dynamic Catalysis Time on Stream Data
Square Waveform:
o 0.001 Hz
o 0.25 Hz
o 10 Hz
o 1000 Hz
10 Hz Oscillation Frequency:
o Square Waveform
o Sinusoidal Waveform
o Triangle Waveform
o Sawtooth Waveform
Section S5. Figure 3 Heatmap Data
Page S2
Section S1. Matlab 2017b and 2018b Code
CSTR Model
% CSTR model
% Description:
% Continuously Stirred Tank Reactor with overall reaction A -> B.
% The number of catalytic active sites is the control.
function xdot=cstr1(t,x)
global T Rg Caf Cbf V u deltE0 q
% Volcano parameters
% Binding energies
BEAeV0 = 1.3; % eV
BEA0 = BEAeV0*96.485e3; % J/gmol
BEBeV0 = 1.0; % eV
BEB0 = BEBeV0*96.485e3; % J/gmol
% Correlation between BE A and BE B
dCorrA = 0.5;
dCorrB = 1;
% Overall reaction enthalpy
dHovr = -20e3; % J/gmol
% BEP relationship parameters
alpha = 0.4;
beta = 102e3; % J/gmol
% Input (1):
% Number of catalytic active sites (gmol)
N = u;
% States (2):
% Concentration of A in CSTR (M)
Ca = x(1,1);
% Concentration of B in CSTR (M)
Cb = x(2,1);
% Number of A* in CSTR (gmol)
Ca_star = x(3,1);
% Number of B* in CSTR (gmol)
Cb_star = x(4,1);
% Parameters:
% Binding energy (J/gmol)
BEa = BEA0 + dCorrA*deltE0;
BEb = BEB0 + dCorrB*deltE0;
% Heat of reaction (J/gmol)
delH1 = -BEa; % A --> A*
delH2 = dHovr + BEa - BEb; % A* --> B*
delH3 = BEb; % B* --> B
% E - Activation energy in the Arrhenius Equation (J/gmol)
R = 8.314459848; % J/gmol-K
Page S3
Eaf1 = 0e3;
Eaf2 = beta + alpha*delH2;
Eaf3 = delH3;
% Pre-exponential factor
Af1 = 1e6; % 1/bar-sec
Af2 = 1e13; % 1/sec
Af3 = 1e13; % 1/sec
% Equilibrium constants
K1 = 1e-7*exp(-delH1/(R*T)); % 1/bar
K2 = 1*exp(-delH2/(R*T)); % unitless
K3 = 1e7*exp(-delH3/(R*T)); % bar
% Rate constants
kf1 = Af1*exp(-Eaf1/(R*T)); % 1/bar-sec
kr1 = kf1/K1; % 1/sec
kf2 = Af2*exp(-Eaf2/(R*T)); % 1/sec
kr2 = kf2/K2; % 1/sec
kf3 = Af3*exp(-Eaf3/(R*T)); % 1/sec
kr3 = kf3/K3; % 1/bar-sec
% Reactant pressure (bar)
Pa = Ca*Rg*T;
Pb = Cb*Rg*T;
% Site balance (gmol)
star = N - Ca_star - Cb_star;
% Surface coverages (unitless)
Theta_a_star = Ca_star/N;
Theta_b_star = Cb_star/N;
Theta_star = star/N;
% Compute xdot:
xdot(1,1) = (q/V*(Caf - Ca) - kf1*Pa*Theta_star*(N/V) + kr1*Theta_a_star*(N/V));
xdot(2,1) = (q/V*(Cbf - Cb) + kf3*Theta_b_star*(N/V) - kr3*Pb*Theta_star*(N/V));
xdot(3,1) = kf1*Pa*Theta_star*N - kr1*Theta_a_star*N - kf2*Theta_a_star*N + kr2*Theta_b_star*N;
xdot(4,1) = - kf3*Theta_b_star*N + kr3*Pb*Theta_star*N + kf2*Theta_a_star*N - kr2*Theta_b_star*N;
Volcano Plot Generator
% Remove size discrepancies
clear
% Step test for Model 1 - CSTR
global T Rg Caf Cbf V u BEA0 BEB0 deltE0 dCorrA dCorrB dHovr alpha beta q
% Input (1):
% Temperature
Tc = 150; % deg C
T = Tc + 273.15; % K
Page S4
% Feed Pressure (bar)
Paf = 100;
Pbf = 0;
% Ideal Gas Constant (L-bar/K-gmol)
Rg = 8.314459848*10^-2;
% Feed Concentration (M)
Caf = Paf/(Rg*T);
Cbf = Pbf/(Rg*T);
% CSTR Volume
w = 200; % mg-cat
rho = 3.578; % g-cat/mL-cat
V = (w/1000)*(1/rho)*(1/1000); % L-reactor
% Volumetric Flowrate
qs = 160; % mL/min
% Steady State Initial Condition for the Control
u_ss = 0; % gmol sites
% Open Loop Step Change
srho = 20e-6; % gmol sites/g-cat
u = (w/1000)*srho; % gmol sites
% Steady State Initial Conditions for the States
Ca_ss = Caf; % M
Cb_ss = Cbf; % M
Ca_star_ss = 0; % gmol
Cb_star_ss = 0; % gmol
x_ss = [Ca_ss;Cb_ss;Ca_star_ss;Cb_star_ss];
% Startup Time (sec)
ts = 5e13;
% Create TOF matrix
storx = zeros(201,1);
story = zeros(201,1);
% Volcano parameters
% Binding energies
BEAeV0 = 1.3; % eV
BEA0 = BEAeV0*96.485e3; % J/gmol
BEBeV0 = 1.0; % eV
BEB0 = BEBeV0*96.485e3; % J/gmol
% Correlation between BE A and BE B
dCorrA = 0.5;
dCorrB = 1;
% Overall reaction enthalpy
dHovr = -20e3; % J/gmol
% BEP relationship parameters
alpha = 0.4;
beta = 102e3; % J/gmol
Page S5
% Iterate i
for i = -1:0.01:1
% Binding energy shift
delteV0 = i; % eV
deltE0 = delteV0*96.485e3; % J/gmol
% Iterate at i
for n = 1:inf
% Volumetric Flowrate
q = qs*1.66667e-5; % L/sec
% Spacetime (sec)
taur = V/q;
% Run ODE solver
[t,x] = ode15s('cstr1',[0 ts],x_ss);
% Parse out the state values
Ca = x(:,1); % M
Cb = x(:,2); % M
% Calculate rates
TOF = Cb*q/u; % 1/sec
Effluent = Cb./(Ca + Cb)*100; % Mole percent
% Constrain to 1 percent yield
if abs(1-Effluent(end)) > 0.01
qs = qs*Effluent(end);
clear Ca Cb TOF Effluent
else
j = (i+1)/0.01 + 1;
storx(round(j)) = i;
story(round(j)) = TOF(end);
clear Ca Cb TOF Effluent
break
end
end
end
% Plot volcano
figure(1)
semilogy(storx,story)
Dynamic Catalysis with Square, Triangle, or Sawtooth Waveforms
% Remove size discrepancies
clear
% Step test for Models 1, 2, and 3 - CSTRs
Page S6
global T Rg Caf Cbf V u deltE0 deltE1 deltE2 q
% Temperature
Tc = 150; % deg C
T = Tc + 273.15; % K
% Feed Pressure (bar)
Paf = 100;
Pbf = 0;
% Ideal Gas Constant (L-bar/K-gmol)
Rg = 8.314459848*10^-2;
% Feed Concentration (M)
Caf = Paf/(Rg*T);
Cbf = Pbf/(Rg*T);
% CSTR Volume
w = 200; % mg-cat
rho = 3.578; % g-cat/mL-cat
V = (w/1000)*(1/rho)*(1/1000); % L-reactor
% Steady State Initial Condition for the Control
u_ss = 0; % gmol sites
% Open Loop Step Change
srho = 20e-6; % gmol sites/g-cat
u = (w/1000)*srho; % gmol sites
% Startup Time (sec)
ts = 5e1;
% Dynamic catalysis parameters
delteV0 = 0.04; % eV
deltE0 = delteV0*96.485e3; % J/gmol
delteV1 = delteV0 + 1.04; % eV
deltE1 = delteV1*96.485e3; % J/gmol
delteV2 = delteV0 - 0.46; % eV
deltE2 = delteV2*96.485e3; % J/gmol
tau1 = 5000; % sec
tau2 = tau1; % sec
chkr = tau1/tau2;
Nosc = 11; % Number of oscillations
fosc = 1/(tau1 + tau2); % 1/sec
% Volumetric Flowrate
qs = 160; % mL/min
q = qs*1.66667e-5; % L/sec
% Create empty matrices
x_ss = zeros(4,Nosc);
Ca_ss = zeros(1,Nosc);
Cb_ss = zeros(1,Nosc);
Page S7
Ca_star_ss = zeros(1,Nosc);
Cb_star_ss = zeros(1,Nosc);
% Steady State Initial Conditions for the States
Ca_ss(1) = Caf; % M
Cb_ss(1) = Cbf; % M
Ca_star_ss(1) = 0; % gmol
Cb_star_ss(1) = 0; % gmol
x_ss(:,1) = [Ca_ss(1);Cb_ss(1);Ca_star_ss(1);Cb_star_ss(1)];
% Run ODE Solver
[t,x] = ode15s('cstr1',[0 ts],x_ss(:,1));
% Parse out the state values
Ca(:,1) = x(:,1); % M
Cb(:,1) = x(:,2); % M
Ca_star(:,1) = x(:,3); % gmol
Cb_star(:,1) = x(:,4); % gmol
star = u - Ca_star(:,1) - Cb_star(:,1); % gmol vacant sites
% Calculate rates
TOF = zeros(size(Cb,1),Nosc);
TOF(:,1) = Cb(:,1)*q/u; % 1/sec
Effluent = zeros(size(Cb,1),Nosc);
Effluent(:,1) = Cb(:,1)./(Ca(:,1) + Cb(:,1))*100; % Mole percent
% Surface coverage (unitless)
Theta_a_star(:,1) = Ca_star(:,1)/u;
Theta_b_star(:,1) = Cb_star(:,1)/u;
Theta_star(:,1) = star(:,1)/u;
% Create time matrix
tshift = zeros(size(t,1),Nosc);
tshift(:,1) = t;
% Create temporary storage
store = zeros(Nosc,5);
store(1,:) = [x(end,:),t(end)];
% Begin dynamic catalysis
for k = 2:Nosc
% Steady State Initial Conditions for the States
Ca_ss(k) = store(k-1,1); % M
Cb_ss(k) = store(k-1,2); % M
Ca_star_ss(k) = store(k-1,3); % gmol
Cb_star_ss(k) = store(k-1,4); % gmol
x_ss(:,k) = [Ca_ss(k);Cb_ss(k);Ca_star_ss(k);Cb_star_ss(k)];
% CSTR Selection
Page S8
if mod(k,2) == 0 % even step #
% Run ODE solver
[t,x] = ode15s('cstr2',[0 tau1],x_ss(:,k));
% Shift time span
t = t + store(k-1,5);
store(k,5) = t(end);
else % odd step #
% Run ODE solver
[t,x] = ode15s('cstr3',[0 tau2],x_ss(:,k));
% Shift time span
t = t + store(k-1,5);
store(k,5) = t(end);
end
% Resolving size discrepancies 1
if size(Ca,1) > size(x,1)
store(k,1:4) = x(end,:);
x = [x;repmat(zeros,size(Ca,1)-size(x,1),size(x,2))];
else
store(k,1:4) = x(end,:);
Ca = [Ca;repmat(zeros,size(x,1)-size(Ca,1),size(Ca,2))];
Cb = [Cb;repmat(zeros,size(x,1)-size(Cb,1),size(Cb,2))];
Ca_star = [Ca_star;repmat(zeros,size(x,1)-size(Ca_star,1),size(Ca_star,2))];
Cb_star = [Cb_star;repmat(zeros,size(x,1)-size(Cb_star,1),size(Cb_star,2))];
star = [star;repmat(zeros,size(x,1)-size(star,1),size(star,2))];
end
% Parse out the state values
Ca(:,k) = x(:,1); % M
Cb(:,k) = x(:,2); % M
Ca_star(:,k) = x(:,3); % gmol
Cb_star(:,k) = x(:,4); % gmol
star(:,k) = u - Ca_star(:,k) - Cb_star(:,k); % gmol vacant sites
% Resolving size discrepancies 2
if size(TOF,1) > size(Cb,1)
Ca = [Ca;repmat(zeros,size(TOF,1)-size(Ca,1),size(Ca,2))];
Cb = [Cb;repmat(zeros,size(TOF,1)-size(Cb,1),size(Cb,2))];
else
Page S9
TOF = [TOF;repmat(zeros,size(Cb,1)-size(TOF,1),size(TOF,2))];
Effluent = [Effluent;repmat(zeros,size(Cb,1)-size(Effluent,1),size(Effluent,2))];
end
% Calculate rates
TOF(:,k) = Cb(:,k)*q/u; % 1/sec
Effluent(:,k) = Cb(:,k)./(Ca(:,k) + Cb(:,k))*100; % Mole percent
% Resolving size discrepancies 3
if size(Theta_a_star,1) > size(Ca_star,1)
Ca_star = [Ca_star;repmat(zeros,size(Theta_a_star,1)-size(Ca_star,1),size(Ca_star,2))];
Cb_star = [Cb_star;repmat(zeros,size(Theta_a_star,1)-size(Cb_star,1),size(Cb_star,2))];
star = [star;repmat(zeros,size(Theta_a_star,1)-size(star,1),size(star,2))];
else
Theta_a_star = [Theta_a_star;repmat(zeros,size(Ca_star,1)-size(Theta_a_star,1),size(Theta_a_star,2))];
Theta_b_star = [Theta_b_star;repmat(zeros,size(Ca_star,1)-size(Theta_b_star,1),size(Theta_b_star,2))];
Theta_star = [Theta_star;repmat(zeros,size(Ca_star,1)-size(Theta_star,1),size(Theta_star,2))];
end
% Surface coverage (unitless)
Theta_a_star(:,k) = Ca_star(:,k)/u;
Theta_b_star(:,k) = Cb_star(:,k)/u;
Theta_star(:,k) = star(:,k)/u;
% Correcting time matrix
% Resolving size discrepancies 4
if size(tshift,1) > size(t,1)
t = [t;repmat(zeros,size(tshift,1)-size(t,1),size(t,2))];
else
tshift = [tshift;repmat(zeros,size(t,1)-size(tshift,1),size(tshift,2))];
end
tshift(:,k) = t;
end
% End dynamic catalysis
% Clean up data
Page S10
txls = nonzeros(tshift);
TOFxls = nonzeros(TOF);
Theta_A_starxls = nonzeros(Theta_a_star);
Theta_B_starxls = nonzeros(Theta_b_star);
for p = 1:size(tshift,1)
for r = 1:size(tshift,2)
if tshift(p,r) == 0
Theta_star(p,r) = 0;
else
if tshift(p,r) > 0 && Theta_star(p,r) == 0
Theta_star(p,r) = 1;
else
continue
end
end
end
end
Theta_starxls = nonzeros(Theta_star);
Theta_starxls(Theta_starxls > 0.99) = 0;
% Generate Time Matrix
tsh = tshift - tshift(1,:);
tsh(tsh < 0) = 0;
tshxls = nonzeros(tsh);
% Generate BE shift wave
BEshift = zeros(size(TOF));
% Cleanup Errors
for v = 1:size(TOF,1)
if tshift(v,1)>0
BEshift(v,1) = delteV0;
else
BEshift(v,1) = 0;
end
end
for m = 2:size(TOF,2)
if mod(m,2) == 0
for n = 1:size(TOF,1)
if tshift(n,m)>0
BEshift(n,m) = delteV1;
else
BEshift(n,m) = 0;
end
end
else
for n = 1:size(TOF,1)
if tshift(n,m)>0
BEshift(n,m) = delteV2;
else
BEshift(n,m) = 0;
end
Page S11
end
end
end
% Clean up wave
BEshiftxls = nonzeros(BEshift);
Dynamic Catalysis with Sinusoidal Waveform
% Remove size discrepancies
clear
% Step test for Models 1 and 2 - CSTRs
global T Rg Caf Cbf V u deltE0 deltE1 deltE2 tau q
% Temperature
Tc = 150; % deg C
T = Tc + 273.15; % K
% Feed Pressure (bar)
Paf = 100;
Pbf = 0;
% Ideal Gas Constant (L-bar/K-gmol)
Rg = 8.314459848*10^-2;
% Feed Concentration (M)
Caf = Paf/(Rg*T);
Cbf = Pbf/(Rg*T);
% CSTR Volume
w = 200; % mg-cat
rho = 3.578; % g-cat/mL-cat
V = (w/1000)*(1/rho)*(1/1000); % L-reactor
% Steady State Initial Condition for the Control
u_ss = 0; % gmol sites
% Open Loop Step Change
srho = 20e-6; % gmol sites/g-cat
u = (w/1000)*srho; % gmol sites
% Startup Time (sec)
ts = 5e1;
% Dynamic catalysis parameters
delteV0 = 0.02; % eV
deltE0 = delteV0*96.485e3; % J/gmol
delteV1 = delteV0 + 0.48; % eV
deltE1 = delteV1*96.485e3; % J/gmol
delteV2 = delteV0 - 0.12; % eV
deltE2 = delteV2*96.485e3; % J/gmol
tau = 5000; % sec
Page S12
fosc = 1/tau; % 1/sec
tspan = 10*tau; % sec
Nosc = tspan/tau; % Unitless
% Volumetric Flowrate
qs = 160; % mL/min
q = qs*1.66667e-5; % L/sec
% Steady State Initial Conditions for the States
Ca_ss = Caf; % M
Cb_ss = Cbf; % M
Ca_star_ss = 0; % gmol
Cb_star_ss = 0; % gmol
x_ss = [Ca_ss;Cb_ss;Ca_star_ss;Cb_star_ss];
% Run ODE Solver
[t,x] = ode15s('cstr1',[0 ts],x_ss);
% Parse out the state values
Ca = x(:,1); % M
Cb = x(:,2); % M
Ca_star = x(:,3); % gmol
Cb_star = x(:,4); % gmol
star = u - Ca_star - Cb_star; % gmol vacant sites
% Calculate rates
TOF = Cb*q/u; % 1/sec
Effluent = Cb./(Ca + Cb)*100; % Mole percent
% Surface coverage (unitless)
Theta_a_star = Ca_star/u;
Theta_b_star = Cb_star/u;
Theta_star = star/u;
% Begin dynamic catalysis
% Steady State Initial Conditions for the States
Ca_ss2 = Ca(end); % M
Cb_ss2 = Cb(end); % M
Ca_star_ss2 = Ca_star(end); % gmol
Cb_star_ss2 = Cb_star(end); % gmol
x_ss2 = [Ca_ss2;Cb_ss2;Ca_star_ss2;Cb_star_ss2];
% Run ODE solver
[t2,x2] = ode15s('sin_cstr2',[0 tspan],x_ss2);
% Parse out the state values
Ca2 = x2(:,1); % M
Cb2 = x2(:,2); % M
Ca_star2 = x2(:,3); % gmol
Cb_star2 = x2(:,4); % gmol
Page S13
star2 = u - Ca_star2 - Cb_star2; % gmol vacant sites
% Calculate rates
TOF2 = Cb2*q/u; % 1/sec
Effluent2 = Cb2./(Ca2 + Cb2)*100; % Mole percent
% Surface coverage (unitless)
Theta_a_star2 = Ca_star2/u;
Theta_b_star2 = Cb_star2/u;
Theta_star2 = star2/u;
% Shift time matrix
t2s = t2 + t(end);
% End dynamic catalysis
% Clean up data
txls = [nonzeros(t);nonzeros(t2s)];
TOFxls = [nonzeros(TOF);nonzeros(TOF2)];
Theta_A_starxls = [nonzeros(Theta_a_star);nonzeros(Theta_a_star2)];
Theta_B_starxls = [nonzeros(Theta_b_star);nonzeros(Theta_b_star2)];
Theta_starxls = [nonzeros(Theta_star);nonzeros(Theta_star2)];
% Generate BE shift wave
BEshift1 = zeros(size(t));
BEshift1(2:end,1) = delteV0;
BEshift2 = ((delteV2-delteV1)/2)*cos(2*pi*t2/tau) + mean([delteV1,delteV2]);
BEshiftxls = [nonzeros(BEshift1);nonzeros(BEshift2)];
Page S14
Section S2. ODE Solver Selection and Justification
Matlab ODE solvers were screened using a sinusoidal dynamic catalysis waveform. ODE45 is the
recommended solver for general use in Matlab, however, this solver failed to generate a solution after 8 h
of continuous computation.
Stats for ODE15s:
3179 successful steps
312 failed attempts
4727 function evaluations
1 partial derivatives
631 LU decompositions
4721 solutions of linear systems
Elapsed time is 1.449973 seconds.
Stats for ODE23s:
10561 successful steps
44 failed attempts
74017 function evaluations
10561 partial derivatives
10605 LU decompositions
31815 solutions of linear systems
Elapsed time is 10.730825 seconds.
Stats for ODE23t:
14214 successful steps
125 failed attempts
21663 function evaluations
45 partial derivatives
1015 LU decompositions
21437 solutions of linear systems
Elapsed time is 4.710374 seconds.
Stats for ODE23tb:
11161 successful steps
92 failed attempts
30658 function evaluations
5 partial derivatives
897 LU decompositions
41881 solutions of linear systems
Elapsed time is 4.704730 seconds.
ODE23s has the least number of failed attempts, but ODE15s performed most efficiently in terms of
number of steps and amount of time to generate the solution. Therefore, ODE15s was used throughout
this manuscript to solve CSTR equations under static and dynamic catalysis conditions.
Page S15
Section S3. Time on Stream Data for Static Catalysis
Figure S1. Example of time on stream data generated with ODE15s for a CSTR operating at 1 % yield of
B. Reaction conditions: 150 oC, 100 bar A inlet, and a volumetric spacetime of 1 s.
Page S16
Section S4. Time on Stream Data for Dynamic Catalysis
Section S4.1. Varying Oscillation Frequency with a Square Waveform
Figure S2. Dynamics of 0.001 Hz System. A. Waveform amplitude of -0.1 to +0.5 eV. B. Catalytic
turnover frequency of B with time. C. Surface coverage of A* with time. D. Surface coverage of B* with
time. E. Surface coverage of open sites with time. Conditions: 150 °C, 100 bar A inlet, ΔU = 0.6 eV, 1%
yield of B, CSTR reactor, oscillation frequency 0.001 Hz, Volcano Curve Parameters: α of 0.8 and E0
of 102 kJ/mol, B* shifts twice as much as A*
-0.40
-0.20
0.00
0.20
0.40
0.60
0.80
0 500 1000 1500 2000
Wa
vefo
rm A
mp
litu
de
[e
V]
Time [Seconds]
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0 500 1000 1500 2000Tu
rno
ver
Fre
qu
en
cy o
f B
[s
-1]
Time [Seconds]
-0.20
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0 500 1000 1500 2000Surf
ace
Cove
rage
of A
, θ
A[-
]
Time [Seconds]
-0.20
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0 500 1000 1500 2000Surf
ace C
overa
ge o
f B
, θ
B[-
]
Time [Seconds]
0.00E+00
5.00E-07
1.00E-06
1.50E-06
2.00E-06
2.50E-06
3.00E-06
0 500 1000 1500 2000
Surf
ace
Cove
rage
of O
pen
, θ
*[-
]
Time [Seconds]
A B
C D
E
Page S17
Figure S3. Dynamics of 0.25 Hz System. A. Waveform amplitude of -0.1 to +0.5 eV. B. Catalytic
turnover frequency of B with time. C. Surface coverage of A* with time. D. Surface coverage of B* with
time. E. Surface coverage of open sites with time. Conditions: 150 °C, 100 bar A inlet, ΔU = 0.6 eV, 1%
yield of B, CSTR reactor, oscillation frequency 0.25 Hz, Volcano Curve Parameters: α of 0.8 and E0 of
102 kJ/mol, B* shifts twice as much as A*
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
0 5 10 15 20
Wa
vefo
rm A
mp
litu
de
[e
V]
Time [Seconds]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 5 10 15 20
Tu
rno
ve
r F
req
ue
ncy,
B
[s-1
]
Time [Seconds]
-0.5
0.0
0.5
1.0
1.5
0 5 10 15 20Su
rfa
ce
Co
ve
rag
e o
f A
, θ
A[s
-1]
Time [Seconds]
-0.5
0.0
0.5
1.0
1.5
0 5 10 15 20Su
rfa
ce
Co
ve
rag
e o
f B
, θ
B[s
-1]
Time [Seconds]
0.0E+00
5.0E-07
1.0E-06
1.5E-06
2.0E-06
2.5E-06
3.0E-06
0 5 10 15 20Surf
ace C
ove
rage o
f O
pen, θ
*[s
-1]
Time [Seconds]
A B
C D
E
Page S18
Figure S4. Dynamics of 10 Hz System. A. Waveform amplitude of -0.1 to +0.5 eV. B. Catalytic
turnover frequency of B with time. C. Surface coverage of A* with time. D. Surface coverage of B* with
time. E. Surface coverage of open sites with time. Conditions: 150 °C, 100 bar A inlet, ΔU = 0.6 eV, 1%
yield of B, CSTR reactor, oscillation frequency 10 Hz, Volcano Curve Parameters: α of 0.8 and E0 of
102 kJ/mol, B* shifts twice as much as A*
0
5
10
15
20
25
0 0.1 0.2 0.3 0.4 0.5 0.6
Tu
rno
ve
r F
req
ue
ncy,
B
[s-1
]
Time [seconds]
-0.4
-0.2
0
0.2
0.4
0.6
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6
Wa
vefo
rm A
mp
litu
de
[e
V]
Time [seconds]
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 0.1 0.2 0.3 0.4 0.5 0.6Surf
ace
Cove
rage
of A
, θ
A[-
]
Time [seconds]
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 0.1 0.2 0.3 0.4 0.5 0.6Surf
ace
Cove
rage
of B
, θ
B[-
]
Time [seconds]
0.0E+00
5.0E-07
1.0E-06
1.5E-06
2.0E-06
2.5E-06
3.0E-06
0 0.2 0.4 0.6
Surf
ace C
ove
rage o
f O
pen S
ites, θ
*[-
]
Time [seconds]
A B
C D
E
Page S19
Figure S5. Dynamics of 1,000 Hz System. A. Waveform amplitude of -0.1 to +0.5 eV. B. Catalytic
turnover frequency of B with time. C. Surface coverage of A* with time. D. Surface coverage of B* with
time. E. Surface coverage of open sites with time. Conditions: 150 °C, 100 bar inlet A, ΔU = 0.6 eV,
1% yield of B, CSTR reactor, oscillation frequency 1,000 Hz, Volcano Curve Parameters: α of 0.8 and
E0 of 102 kJ/mol, B* shifts twice as much as A*
-0.4
-0.2
0
0.2
0.4
0.6
0.8
0 1 2 3 4 5
Wavefo
rm A
mplit
ude
[eV
]
Time [Milliseconds]
28.0
28.5
29.0
29.5
30.0
30.5
31.0
0 1 2 3 4 5
Turn
over
Fre
quency,
B
[s-1
]
Time [Milliseconds]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1 2 3 4 5Surf
ace C
overa
ge o
f A
, θ
A[-
]
Time [Milliseconds]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 1 2 3 4 5Surf
ace C
overa
ge o
f B
, θ
B[-
]Time [Milliseconds]
0.0E+00
2.0E-07
4.0E-07
6.0E-07
8.0E-07
1.0E-06
0 1 2 3 4 5
Surf
ace C
overa
ge o
f O
pen, θ
*[-
]
Time [Milliseconds]
A B
C D
E
Page S20
Section S4.2. 10 Hz Dynamic Catalysis with Varying Waveforms
Figure S6. Square waveform at 10 Hz. Conditions: A CSTR reactor for A converting to B operating at
1% yield, 150 oC, and 100 bar A inlet. Volcano plot: ΔU ~ 0.6 eV, α ~ 0.8, E0 of 102 kJ/mol, and B*
moves twice as much as A*
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.40 0.50 0.60 0.70 0.80 0.90 1.00
Rela
tive
Bin
din
g E
nerg
y, B
[eV
]
Time [seconds]
0
2
4
6
8
10
12
14
16
18
20
0.40 0.50 0.60 0.70 0.80 0.90 1.00
Tu
rnove
r F
req
uen
cy
to B
[s
-1]
Time [seconds]
-0.1
0.1
0.3
0.5
0.7
0.9
1.1
0.40 0.50 0.60 0.70 0.80 0.90 1.00
Su
rface c
ove
rag
e A
[-]
Time [seconds]
-0.1
0.1
0.3
0.5
0.7
0.9
1.1
0.40 0.50 0.60 0.70 0.80 0.90 1.00
Su
rface c
ove
rag
e B
[-]
Time [seconds]
0.0E+00
2.0E-06
4.0E-06
6.0E-06
8.0E-06
1.0E-05
0.40 0.50 0.60 0.70 0.80 0.90 1.00
Su
rface c
ove
rag
e:
Op
en
Sites
[-]
Time [seconds]
Page S21
Figure S7. Triangle waveform at 10 Hz. Conditions: A CSTR reactor for A converting to B operating at
1% yield, 150 oC, and 100 bar A inlet. Volcano plot: ΔU ~ 0.6 eV, α ~ 0.8, E0 of 102 kJ/mol, and B*
moves twice as much as A*
-1.00
-0.80
-0.60
-0.40
-0.20
0.00
0.20
0.40
0.60
0.80
1.00
51.50 51.60 51.70 51.80 51.90 52.00Bin
din
g E
ne
rgy o
f B
(e
V)
Time (seconds)
0
1
2
3
4
5
6
7
8
9
10
51.50 51.60 51.70 51.80 51.90 52.00
Turn
ove
r F
requency t
o B
[s
-1)
Time (seconds)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
51.50 51.60 51.70 51.80 51.90 52.00
Su
rfa
ce
Co
vera
ge
A [-]
Time (seconds)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
51.50 51.60 51.70 51.80 51.90 52.00
Su
rfa
ce
Co
vera
ge
B [
-]
Time (seconds)
A B
C D
Page S22
Figure S8. Sawtooth waveform at 10 Hz. Conditions: A CSTR reactor for A converting to B operating
at 1% yield, 150 oC, and 100 bar inlet A. Volcano plot: ΔU ~ 0.6 eV, α ~ 0.8, E0 of 102 kJ/mol, and B*
moves twice as much as A*
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.40 0.50 0.60 0.70
Rela
tive
Bin
din
g E
nerg
y, B
[eV
]
Time [seconds]
0
1
2
3
4
5
6
7
8
9
10
0.40 0.50 0.60 0.70
Tu
rnove
r F
req
uen
cy
to B
[s
-1]
Time [seconds]
-0.1
0.1
0.3
0.5
0.7
0.9
1.1
0.40 0.50 0.60 0.70
Su
rface c
ove
rag
e A
[-]
Time [seconds]
-0.1
0.1
0.3
0.5
0.7
0.9
1.1
0.40 0.50 0.60 0.70
Su
rface c
ove
rag
e B
[-]
Time [seconds]
Page S23
Figure S9. Sinusoidal waveform at 10 Hz. Conditions: A CSTR reactor for A converting to B operating
at 1% yield, 150 oC, and 100 bar inlet A. Volcano plot: ΔU ~ 0.6 eV, α ~ 0.8, E0 of 102 kJ/mol, and B*
moves twice as much as A*
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.0 0.2 0.4 0.6 0.8 1.0
Bin
din
g E
nerg
y B
[eV
]
Time (s)
0
2
4
6
8
10
12
14
16
0.0 0.2 0.4 0.6 0.8 1.0
Tu
rnover
Fre
qu
en
cy B
[s-1
]
Time (s)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.2 0.4 0.6 0.8 1.0
Su
rface C
overa
ge A
[-]
Time (s)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0.2 0.4 0.6 0.8 1.0
Su
rface C
overa
ge A
[-]
Time (s)
Page S24
Section S5. Figure 3 Heatmap Data
Table 1. Heatmap data (TOFs [=] 1/s) for oscillation amplitudes between 0.0-1.0 eV and frequencies
between 10-4-1010 Hz. Volcano parameters: α of 0.8, E0 of 102 kJ/mol, 150 oC, 100 bar, 1 % yield of
product B, and B* shifts twice as much as A*
0.0 eV 0.1 eV 0.2 eV 0.3 eV 0.4 eV 0.5 eV 0.6 eV 0.7 eV 0.8 eV 0.9 eV 1.0 eV
10-4 Hz 0.2609 0.2106 0.1397 0.1219 0.1021 0.0779 0.0621 0.0465 0.0289 0.0216 0.0174
10-3 Hz 0.2609 0.2106 0.1442 0.1219 0.1021 0.0808 0.0621 0.0403 0.0289 0.0216 0.0181
10-2 Hz 0.2609 0.2056 0.1407 0.1219 0.1021 0.0808 0.0617 0.0556 0.0364 0.0325 0.0293
10-1 Hz 0.2609 0.2263 0.1989 0.1926 0.1907 0.1820 0.1695 0.1501 0.1334 0.1232 0.1113
100 Hz 0.2609 0.2972 0.4505 0.7468 0.8541 0.9143 1.001 1.139 1.288 1.541 2.366
101 Hz 0.2609 0.4149 0.9343 2.231 4.321 7.425 8.030 8.721 9.325 13.26 28.00
102 Hz 0.2609 0.4202 0.9620 2.371 5.581 12.81 26.88 49.68 90.94 120.1 152.9
103 Hz 0.2609 0.4203 0.9632 2.385 5.611 13.04 29.08 65.08 142.4 296.3 542.5
104 Hz 0.2609 0.4203 0.9634 2.383 5.618 13.05 29.43 65.38 144.0 315.9 690.5
105 Hz 0.2609 0.3990 0.9550 2.383 5.616 13.05 29.43 65.39 144.2 316.6 693.1
106 Hz 0.2609 0.2750 0.3364 0.8856 4.947 13.01 29.43 65.39 144.2 316.7 693.9
107 Hz 0.2609 0.2609 0.2613 0.2667 0.2833 0.3799 3.078 59.92 144.0 316.7 694.0
108 Hz 0.2609 0.2609 0.2609 0.2611 0.2617 0.2637 0.2700 0.2917 0.4478 17.81 40.14
109 Hz 0.2609 0.2609 0.2609 0.2609 0.2609 0.2611 0.2614 0.2622 0.2645 0.2719 0.3063
1010 Hz 0.2609 0.2609 0.2609 0.2609 0.2609 0.2609 0.2609 0.2610 0.2611 0.2615 0.2625
download fileview on ChemRxivSupporting_Information_Dauenhauer.pdf (2.05 MiB)