High-throughput screening of nanoparticle catalysts made by flame spray pyrolysis as hydrocarbon/NO oxidation catalysts

Article abstract of DOI:10.1021/ja809134s

We describe here the use of liquid-feed flame spray pyrolysis (LF-FSP) to produce high surface area, nonporous, mixed-metal oxide nanopowders that were subsequently subjected to high-throughput screening to assess a set of materials for deNO_x c

Full text of DOI:10.1021/ja809134s

Published on Web 06/15/2009
High-Throughput Screening of Nanoparticle Catalysts Made
by Flame Spray Pyrolysis as Hydrocarbon/NO Oxidation
Catalysts
†
†
,†
†
‡
‡
B. Weidenhof, M. Reiser, K. St o¨ we,* W. F. Maier, M. Kim, J. Azurdia,
§
§
‡
,‡
E. Gulari, E. Seker, A. Barks, and R. M. Laine*
Lehrstuhl f u¨ r Technische Chemie, UniVersit a¨ t des Saarlandes, 66123 Saarbr u¨ cken, Germany,
and Departments of Materials Science and Engineering and Chemical Engineering, UniVersity of
Michigan, Ann Arbor, Michigan 48109-2136
Received November 21, 2008; E-mail: k.stoewe@mx.uni-saarland.de; talsdad@umich.edu
Abstract: We describe here the use of liquid-feed flame spray pyrolysis (LF-FSP) to produce high surface
area, nonporous, mixed-metal oxide nanopowders that were subsequently subjected to high-throughput
screening to assess a set of materials for deNO catalysis and hydrocarbon combustion. We were able to
x
easily screen some 40 LF-FSP produced materials. LF-FSP produces nanopowders that very often consist
of kinetic rather than thermodynamic phases. Such materials are difficult to access or are completely
inaccessible via traditional catalyst preparation methods. Indeed, our studies identified a set of Ce1-xZr
and Al nanopowders that offer surprisingly good activities for both NO reduction and
propane/propene oxidation both in high-throughput screening and in continuous flow catalytic studies. All
of these catalysts offer activities comparable to traditional Pt/Al catalysts but without Pt. Thus, although
x 2
O
2
O
3
-Ce1-xZr
x
O
2
x
2
O
3
Pt-free, they are quite active for several extremely important emission control reactions, especially
considering that these are only first generation materials. Indeed, efforts to dope the active catalysts with
Pt actually led to lower catalytic activities. Thus the potential exists to completely change the materials
used in emission control devices, especially for high-temperature reactions as these materials have already
been exposed to 1500 °C; however, much research must be done before this potential is verified.
Introduction
catalytic reduction of NO to N by hydrocarbons offers the
x
2
advantage that the unburnt hydrocarbons from combustion,
possibly combined with purposely added hydrocarbons, can be
used to reduce NO continuously without drastic modification
x
In view of the issues of global warming and the energy crisis,
the development of new, fuel economic vehicles that also offer
low CO
mobile society. In addition, rising public concern about envi-
ronmental hazards, and in particular NO pollutants and diesel
2
emissions has become a major challenge of the modern
of the engine management system.
This fact provides significant motivation for intense research
x
4
,5
1
in this area. Different catalyst systems have been studied
soot, has led to strict automotive exhaust emission regulations.
Prompted by these concerns, the automobile industry has
developed two engine technologies, the lean-burn direct injection
gasoline and the common turbo-diesel engines. Both engine
types offer excellent fuel economy compared to the classical
gasoline engine but operate with higher air-to-fuel ratios (λ >
extensively for HC-SCR, targeting development of new catalyst
materials that provide the desired NO control under lean-burn
x
conditions. Such catalysts are also expected to display enhanced
selectivity, improved low-temperature activity (a wider
temperature range of operation), and high stability (H O or SO
N
2
2
2
1
). Under these so-called lean-burn conditions (excess oxygen),
conventional three-way auto exhaust catalysts (TWCs) are
incapable of completely reducing NO
To overcome this problem, several promising techniques
including NO storage reduction (NSR), selective catalytic
reduction of NO by NH or hydrocarbons (NH -SCR,
poisoning, thermal degradation) under realistic exhaust gas
6
conditions. Among the systems investigated, several groups
x
.
of metal oxide catalysts have attracted attention due to their
high activity for NO reduction and good hydrothermal stability.
x
Previous studies have emphasized the use of Co, Cu, Ag, Ga,
In, and Sn containing oxides on alumina and other supports as
x
x
3
3
2
HC-SCR), and nonthermal plasma have been proposed for
abating NO
active HC-SRC catalysts. In addition, rare-earth-based materi-
2,3
x
emissions. Among these techniques, the selective
als, especially CeO
used widely in deNO
dynamic oxygen exchange capacity (OEC) involving the Ce /
2
and CeO
2
2
-ZrO solid solutions, have been
x
and soot combustion systems due to their
†
4+
Universit a¨ t des Saarlandes.
Department of Materials Science and Engineering, University of
‡
Michigan.
§
Department of Chemical Engineering, University of Michigan.
1) Klingstedt, F.; Arve, K.; Eranen, K.; Murzin, D. Y. Acc. Chem. Res.
(4) Burch, R.; Breen, J. P.; Meunier, F. C. Appl. Catal. B 2002, 39, 283–
303.
(
2
006, 39, 273–282.
(5) Burch, R. Catal. ReV. 2004, 46, 271–334.
(6) Houel, V.; James, D.; Millington, P.; Pollington, S.; Poulston, S.;
Rajaram, R.; Torbati, R. J. Catal. 2005, 230, 150–157.
(
(
2) Liu, Z.; Woo, S. I. Catal. ReV. 2006, 48, 43–89.
3) Twigg, M. V. Appl. Catal. B 2007, 70, 2–15.
10.1021/ja809134s CCC: $40.75  2009 American Chemical Society
J. AM. CHEM. SOC. 2009, 131, 9207–9219 9 9207

A R T I C L E S
Weidenhof et al.
Ce3 redox couple as well as spillover of active oxygen.
+
7-14
useless waste. Flame aerosol synthesis and in particular liquid-
feed flame spray pyrolysis (LF-FSP) offers an attractive alterna-
tive to traditional preparative techniques because it provides easy
access to a wide variety of mixed-metal oxide solid solution
1
5
16
Furthermore, for instance, Baiker et al. and Ka sˇ par et al.
showed that the thermal stability as well as the OEC of these
solid solutions is improved by doping with small amounts of
silica.
24-29
and phase-separated nanopowders using one-step processing.
In contrast, these researchers report that doping with small
amounts of alumina has little effect on OEC values. These doped
nanopowders were made using LF-FSP processing. Baiker et
al. do suggest formation of solid solutions with the silica or
alumina dopants but do not further characterize silica or alumina
species present. In the case of the alumina, this is understandable
Flame aerosol synthesis is used widely to produce carbon-blacks,
fumed silica, and titania pigments on a large scale with good
30
cost-benefit ratios. The flame-made metal-oxide nanopowders
are usually characterized by high specific surface areas,
1
5
nonporous structures, and improved resistance to sintering,
properties that appear to be very promising for multiple
2
9,31-37
as the OEC activities are not comparable with the SiO
materials.
2
-doped
applications in heterogeneous catalysis.
In LF-FSP processing, mixtures of metalloorganic precursors
(metal carboxylates and/or alkoxides) dissolved in alcohol
solvent are aerosolized with oxygen and afterward ignited via
methane torches within a quartz chamber. Subsequent combus-
tion of the aerosol at temperatures ranging from 1500 to 2000
°C followed by rapid quenching produces dispersible nanosized
oxide powders with element ratios identical to those in the
precursor solutions and occasionally with unknown phase
Several additional factors including support materials, catalyst
composition, metal loading, calcination temperatures, prepara-
tion methods, and reaction conditions must be considered for
the successful design of any new, high-activity catalyst systems.
However, a vast number of possible elemental combinations
with potential catalytic activity for the desired reactions remain
to be explored. Unfortunately, the complexity of the variables
involved limits the total number of systems that can be and
have been investigated on various supports in the recent
3
8,39
compositions.
LF-FSP allows the production of up to five
different oxide compositions within a single week and, therefore,
may be regarded as a combinatorial method for synthesizing
2
,17-21
past.
Some of the reasons for this are as follows.
2
4,40
metal oxide powders.
Heterogeneous catalysts are most often prepared using a set
of standard methods including wet precipitation, sol-gel
processing, solid-state reactions, and impregnation techniques
High-throughput experimentation (HTE) has recently emerged
as valuable approach to discover new materials, especially
catalysts, and is rapidly becoming a widely accepted tool in
2
2,23
(
e.g., incipient wetness).
All of these methods have been
4
1-53
industrial as well as academic research.
In the field of
investigated extensively as a means to improve surface area,
porosity, microstructure, oxidation states, compositions, and
catalyst stability. Despite remarkable progress, there remains a
general demand for optimized synthetic methods to produce high
surface area, sinter-resistant, multicomponent catalysts that are
smoothly tailorable for the required catalytic application.
In addition, conventional catalyst preparation methods (es-
pecially wet-chemical synthesis) often entail multiple process
steps that are time-consuming, difficult to control, and to some
extent accompanied by the production of large amounts of
heterogeneous catalysis, high-throughput approaches normally
(
(
24) Kim, M.; Laine, R. M. J. Ceram. Process. Res. 2007, 8, 129–136.
25) Sutorik, A. C.; Neo, S. S.; Treadwell, D. R.; Laine, R. M. J. Am.
Ceram. Soc. 1998, 81, 1477–1486.
(26) Baranwal, R.; Villar, M. P.; Garcia, R.; Laine, R. M. J. Am. Ceram.
Soc. 2001, 84, 951–961.
(27) Marchal, J.; John, T.; Baranwal, R.; Hinklin, T.; Laine, R. M. Chem.
Mater. 2004, 16, 822–831.
(28) Kim, S.; Gislason, J. J.; Morton, R. W.; Pan, X. Q.; Sun, H. P.; Laine,
R. M. Chem. Mater. 2004, 16, 2336–2343.
(
29) Azurdia, J. A.; Marchal, J.; Shea, P.; Sun, H.; Pan, X. Q.; Laine,
R. M. Chem. Mater. 2006, 18, 731–739.
(
7) Machida, M.; Murata, Y.; Kishikawa, K.; Zhang, D.; Ikeue, K. Chem.
(30) Oswald, M.; Deller, K. Elements Degussa Science Newsletter 2004,
9, 16–20.
Mater. 2008, 20, 4489–4494.
(
(
8) Stark, W. J.; Maciejewski, M.; M a¨ dler, L.; Pratsinis, S. E.; Baiker,
A. J. Catal. 2003, 220, 35–43.
(31) Piacentini, M.; Strobel, R.; Maciejewski, M.; Pratsinis, S. E.; Baiker,
A. J. Catal. 2006, 243, 43–56.
9) Krishna, K.; Bueno-Lopez, A.; Makkee, M.; Moulijn, J. A. Appl.
Catal. B 2007, 75, 189–200.
(32) Strobel, R.; Krumeich, F.; Pratsinis, S. E.; Baiker, A. J. Catal. 2006,
243, 229–238.
(
(
(
(
(
(
(
(
(
10) Krishna, K.; Bueno-Lopez, A.; Makkee, M.; Moulijn, J. A. Appl.
(33) Strobel, R.; Krumeich, F.; Stark, W. J.; Pratsinis, S. E.; Baiker, A.
J. Catal. 2004, 222, 307–314.
Catal. B 2007, 75, 201–209.
11) Krishna, K.; Bueno-Lopez, A.; Makkee, M.; Moulijn, J. A. Appl.
Catal. B 2007, 75, 210–220.
(34) Strobel, R.; Stark, W. J.; M a¨ dler, L.; Pratsinis, S. E.; Baiker, A. J.
Catal. 2003, 213, 296–304.
12) Rao, G. R.; Ka sˇ par, J.; Meriani, S.; Monte, R.; Graziani, M. Catal.
Lett. 1994, 24, 107–112.
(35) Stark, W. J.; Wegner, K.; Pratsinis, S. E.; Baiker, A. J. Catal. 2001,
197, 182–191.
13) Fornasiero, P.; Dimonte, R.; Rao, G. R.; Kaspar, J.; Meriani, S.;
Trovarelli, A.; Graziani, M. J. Catal. 1995, 151, 168–177.
14) Trovarelli, A. Catalysis by Ceria and Related Materials; Imperial
College Press: London, 2002; Vol. 2.
(36) Stark, W. J.; Pratsinis, S. E.; Baiker, A. J. Catal. 2001, 203, 516–
524.
(37) Vormberg, R. Elements Degussa Science Newsletter 2004, 9, 21–
23.
15) Schulz, H.; Stark, W. J.; Maciejewski, M.; Pratsinis, S. E.; Baiker,
A. J. Mater. Chem. 2003, 13, 2979–2984.
(38) Hinklin, T. R.; Azurdia, J.; Kim, M.; Marchal, J. C.; Kumar, S.; Laine,
R. M. AdV. Mater. 2008, 20, 1373–1375.
16) Di Monte, R.; Fornasiero, P.; Kapar, J.; Graziani, M.; Gatica, J. M.;
Bernal, S.; G o´ mez-Herrero, A. Chem. Commun. 2000, 2167–2168.
17) Gobin, O. C.; Martinez Joaristi, A.; Sch u¨ th, F. J. Catal. 2007, 252,
(39) Laine, R. M.; Bickmore, C. R.; Treadwell, D. R.; Waldner, K. F.
J. Am. Ceram. Soc. 1996, 79, 1419–1423.
(40) Azurdia, J.; Marchal, J.; Laine, R. M. J. Am. Ceram. Soc. 2006, 89,
2749–2756.
2
05–214.
18) Busch, O. M.; Hoffmann, C.; Johann, T. R. F.; Schmidt, H. W.;
Strehlau, W.; Schuth, F. J. Am. Chem. Soc. 2002, 124, 13527–13532.
19) Ozturk, S.; Senkan, S. Appl. Catal. B 2002, 38, 243–248.
20) Richter, M.; Langpape, M.; Kolf, S.; Grubert, G.; Eckelt, R.; Radnik,
J.; Schneider, M.; Pohl, M. M.; Fricke, R. Appl. Catal. B 2002, 36,
(41) Senkan, S. M.; Ozturk, S. Angew. Chem., Int. Ed. 1999, 38, 791–
795.
(
(
(42) Bein, T. Angew. Chem., Int. Ed. 1999, 38, 323–326.
(43) Maier, W. F. Angew. Chem., Int. Ed. 1999, 38, 1216–1218.
(44) Jandeleit, B.; Schaefer, D. J.; Powers, T. S.; Turner, H. W.; Weinberg,
W. H. Angew. Chem., Int. Ed. 1999, 38, 2494–2532.
(45) Hagemeyer, A.; Jandeleit, B.; Liu, Y.; Poojary, D. M.; Turner, H. W.;
Volpe, A. F.; Henry Weinberg, W. Appl. Catal. A 2001, 221, 23–
43.
2
61–277.
(
(
(
21) Krantz, K.; Ozturk, S.; Senkan, S. Catal. Today 2000, 62, 281–289.
22) Delmon, B. J. Therm. Anal. Calorim. 2007, 90, 49–65.
23) Ertl, G.; Kn o¨ zinger, H.; Weitkamp, J. Handbook of Heterogeneous
Catalysis; Wiley-VCH: Weinheim, Germany, 1997; Vol. 1.
(46) Senkan, S. Angew. Chem., Int. Ed. 2001, 40, 312–329.
9
208 J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009

High-Throughput Screening of Nanoparticle Catalysts
A R T I C L E S
consist of four closely connected elements: design of experiment,
automated synthesis of catalyst libraries, high-throughput library
screening, and data mining of the results. With reference to the
automated catalyst synthesis, numerous different preparation
methods have been successfully automated with the aid of
robots. In addition to several sputtering techniques, the wet-
achieving combustion temperatures >1500 °C. The products are
3
carried downstream by a radial pressure exhaust (19.8 m /min) and
collected in electrostatic precipitators (ESPs) that are maintained
at a 10 kV dc potential. The nanopowders are recovered manually
from the ESP tubes after the apparatus has cooled down. Typical
collection amounts for each of the 46 samples synthesized were
3
0-50 g. As stated earlier, one to three different compositions were
5
4
55
chemical methods (e.g., precipitation, impregnation, and
produced per day due to the facile collection and large production
rates per batch.
5
6,57
sol-gel preparation methods
) in particular have proved
suitable for the high-throughput synthesis. Nevertheless, these
approaches can turn into more complex tasks if multicomponent
systems (e.g., quaternary catalyst composition spreads or
supported multimetallic catalysts) have to be synthesized and
varied in their compositions. High-throughput screening
methods also have another valuable advantage, all materials are
studied under identical conditions, which allows direct com-
parison of the data.
Consequently, the first objective of the present investigation
was to combine the unique features of LF-FSP for synthesis of
multicomponent oxides with high-throughput technology (HTT)
to accelerate the discovery of new lead catalyst compositions
for the selective catalytic reduction of NO with propane as well
as for the combustion of propane (propene).
2
. Powder Compacts. These were prepared by weighing ∼50
mg of powder and pressing it into a pellet (3.0 mm diameter) using
a dual action hand press.
B. Analytical Methods. 1. Specific Surface Area Analyses
(SSAs). SSAs were obtained using a Micromeritics ASAP 2010
sorption analyzer (Norcross, GA) for all nanopowders. Samples
were loaded (350 mg) and degassed at 350 °C until a degas rate of
5
8
<
2
5 mTorr was achieved, followed by analysis at 77 K with N as
the adsorbate gas. The specific surface areas were calculated using
the BET multipoint (10 points) method. The average particle
diameter or size (APS) was determined by
6
APSBET
)
(1)
Fth × SSA
Here we report results from the high-throughput screening
of 46 different LF-FSP produced pure as well as mixed-metal
oxide nanopowders by means of emissivity-corrected IR ther-
where Fth is the theoretical density of the nanopowders, and SSA
is the specific surface area.
For the reference catalysts (Hopcalite from Dr a¨ ger Safety and 5
wt % Pt on Al O from STREM Chemicals), nitrogen physisorption
5
9-62
mography (ecIRT).
A further objective of this work was
2
3
to evaluate the general applicability of the presented flame-
made materials for exhaust gas catalysis in comparison to
commercial benchmark catalysts.
measurements were preformed on a Carlo Erba Sorptomatic 1990
at the temperature of liquid nitrogen (77 K). The samples were
outgassed for 2 h under vacuum at 200 °C before adsorption. The
specific surface areas were calculated using the BET multipoint
method.
Experimental Section
2
. Scanning Electron Microscopy (SEM). Micrographs were
A. Synthesis. 1. LF-FSP. The liquid-feed flame spray pyrolysis
LF-FSP) system has been described in detail elsewhere.
Ethanol precursor solutions were prepared by diluting the sample
precursors to 2-5 wt % ceramic yield, typically 3 wt %. These
2
solutions are then atomized with O through a nozzle. The fine
mist generated is ignited with methane/oxygen pilot torches,
taken using a Nova 600 Nanolab FIB/SEM (FEI Company,
Hillsboro, OR). About 1 mg of nanopowder sample was first
dispersed in 5 mL of DI water using an ultrasonic horn (Vibra-
cell, Sonics and Materials Inc., Newton, CT) for 10 min. A drop
of the dispersion was placed on a SEM sample stub, which was
heated on a covered hot plate and the water allowed to evaporate.
The stubs were then sputter coated with ∼180 Å of Au-Pd using
a Technics Hummer VI sputtering system (Anatech, Ltd., Alexan-
dria, VA) to improve resolution.
2
8,29,40
(
(
47) Mallouk, T. E.; Smotkin, E. S. Combinatorial Catalyst Development
Methods. In Handbook of Fuel Cells: Fundamentals, Technology and
Applications; Vielstich, W., Lamm, A., Gasteiner, H. A., Eds.; John
Wiley & Sons: Chichester, UK, 2003; Vol. 2, pp 334-347.
3
. X-ray Diffraction Analyses (XRD). Analyses were per-
formed on a Rigaku Miniflex (Rigaku, The Woodlands, TX). The
diffractometer is equipped with a Cu X-ray tube (Cu KR , λ )
(
48) Bricker, M. L.; Sachtler, J. W. A.; Gillespie, R. D.; McGonegal, C. P.;
Vega, H.; Bem, D. S.; Holmgren, J. S. Appl. Surf. Sci. 2004, 223,
1
1
09–117.
1.54059 Å) with operating voltage of 30 kV and current of 15 mA.
Scans were performed continuously from 20 to 80° 2θ in 0.03°
increments at 2°/min. The nanopowder samples were prepared by
packing ∼100 mg into an amorphous silica holder and loaded into
the diffractometer. Scan data were analyzed using Jade 7.5 software
(Materials Data, Inc., Livermore, CA) to determine average particle
sizes (APSs) and phases present.
(
49) Hendershot, R. J.; Snively, C. M.; Lauterbach, J. Chem.sEur. J. 2005,
1, 806–814.
50) Maier, W. F.; St o¨ we, K.; Sieg, S. Angew. Chem., Int. Ed. 2007, 46,
016–6067.
1
(
6
(
(
(
(
51) Reetz, M. T. Angew. Chem., Int. Ed. 2001, 40, 284–310.
52) Newsam, J. M.; Sch u¨ th, F. Biotechnol. Bioeng. 1998, 61, 203–216.
53) Gennari, F.; Seneci, P.; Miertus, S. Catal. ReV. 2000, 42, 385–402.
54) Hoffmann, C.; Wolf, A.; Sch u¨ th, F. Angew. Chem., Int. Ed. 1999,
X-ray powder diffraction patterns for the reference catalysts
3
8, 2800–2803.
(
Hopcalite from Dr a¨ ger Safety and 5 wt % Pt on Al
STREM Chemicals) were obtained using a Huber G670 Guinier
image plate system with Cu KR radiation (λ ) 1.54059 Å). For
2 3
O from
(
(
55) Rodemerck, U.; Wolf, D.; Buyevskaya, O. V.; Claus, P.; Senkan,
S.; Baerns, M. Chem. Eng. J. 2001, 82, 3–11.
56) Scheidtmann, J.; Weiss, P. A.; Maier, W. F. Appl. Catal. A 2001,
1
2
22, 79–89.
sample preparation, a few milligrams of powder was applied on
adhesive tape in a sample holder. The data were accumulated
continuously from 4 to 100° 2θ and analyzed using TOPAS
(
(
57) Kim, D. K.; Maier, W. F. J. Catal. 2006, 238, 142–152.
58) Hannemann, S.; Grunwaldt, J.-D.; Lienemann, P.; G u¨ nther, D.;
Krumeich, F.; Pratsinis, S. E.; Baiker, A. Appl. Catal. A 2007, 316,
63
software to determine crystal data as lattice constants as well as
2
26–239.
average particle sizes (APSXRD) by Rietveld refinement.
(
59) Holzwarth, A.; Schmidt, H.-W.; Maier, W. F. Angew. Chem., Int.
Ed. 1998, 37, 2644–2647.
4
. Dynamic Light Scattering (DLS). DLS was also used to
(
(
(
60) Holzwarth, A.; Maier, W. F. Platinum Met. ReV. 2000, 44, 16–21.
61) Kirsten, G.; Maier, W. F. Appl. Surf. Sci. 2004, 223, 87–101.
estimate APSs of LF-FSP nanopowders. Suspensions of each
nanopowder in DI water (∼1 vol %) were loaded in compact
goniometer system (ALV, Langen, Germany) equipped with a
62) Kirsten, G.; Maier, W. F. Relative Quantification of Catalytic Activity
in Combinatorial Libraries by Emissivity-Corrected Infrared Ther-
mography. In High Throughput Screening in Chemical Catalysis;
Hagemeyer, A., Strasser, V., Volpe, A. F., Eds.; Wiley-VCH:
Weinheim, Germany, 2004.
(63) TOPAS, version 2.1 (General profile and structure analysis software
for powder diffraction data); Bruker AXS: Karlsruhe, Germany, 2000.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009 9209

A R T I C L E S
Weidenhof et al.
Table 1. Catalysts in Slate Library Plate
catalyst
short description
composition
source
count
MO (M ) Ni)
M
x
x ) 100
this work
3
MO
2
(M ) Ce, Zr)
(M ) Ce, Ni, Zr)
Ce
(
CeO
2
)
2
x/100(Al O
3
)
1-(x/100)
x
Al100-x
x ) 0.10, 0.50, 1.00, 1.50, 2.00, 3.00, 5.00, 6.00,
.50, 15.0, 30.0, 50.0, 75.0, 95.0
this work
14
7
(
(
CeO
CeO
2
)
)
x/100(ZrO
x/100(ZrO2)y/100(Al
2
)
1-(x/100)
Ce
Ce
x
Zr100-x
x ) 10.0, 30.0, 50.0, 70.0, 90.0
this work
this work
5
3
2
2
O
)
3 1-((x+y)/100)
x y
Zr Al100-(x+y)
x ) 35.0, 49.0, 63.0
y ) 15.0, 21.0, 27.0
(
(
(
(
CoO)x/100(Al
NiO)x/100(Al
NiO)x/100(CuO)1-(x/100)
x/100(Al
2
O
3
)
1-(x/100)
Co
x
Al100-x
x ) 5.00, 10.0, 25.0, 70.0, 85.0
x ) 5.00, 10.0, 22.0, 25.0, 50.0, 80.0
x ) 50.0, 75.0
this work
this work
this work
this work
Strem
5
6
2
8
1
O
2 3
)
1-(x/100)
Ni
Ni
x
Al100-x
Cu100-x
Al100-x
x
ZrO
2
)
O
2
O
3
)
1-(x/100)
Zr
Pt/Al
Hopcalite
x
x ) 2.40, 4.10, 4.60, 5.90, 8.40, 13.6, 49.7, 80.6
Pt/Al
2
3
(reference)
2 3
O
Hopcalite (reference)
Dr a¨ ger Safety
1
Table 2. Compositions of Reaction Gas Mixtures
multitau digital correlator (ALV-5000E, Langen, Germany). The
laser source had a wavelength of λ
Coherent Inc., Santa Clara, CA).
0
) 0.488 µm (Innova 70C,
c(C
[vol %]
3
H
8
)
c(NO)
[vol %]
c(O
[vol %]
2
)
2
c(N )
[vol %]
mixture no.
brief description
C. Notation for Catalysts (Short Description). A simple
notation was used to name the examined oxides, whereby, for
example, a ternary oxide mixture is indicated by the contraction
A B C100-(x+y) generally. The upper case A, B, and C stand for a
x y
particular metal oxide, the lower case x and y describe the share of
the respective oxide in mol %.
1
2
3
4
5
C H 20.0% O
0.40
0.40
0.40
0.40
0.40
0.00
0.20
0.20
0.20
0.20
20.0
20.0
5.00
1.00
0.00
79.6
79.4
94.4
98.4
99.4
3
3
3
3
3
8
8
8
8
8
2
C
C
C
C
H
H
H
H
NO 20.0% O
2
2
2
NO 5.00% O
NO 1.00% O
NO
D. High-Throughput Screening of Nanoparticle Catalysts.
The catalyst pellets were crushed, and the resulting powders (see
Table 1) were manually transferred into hexagonally positioned
wells of a slate library plate (see Figure 1). Emissivity-corrected
IR thermography (ecIRT) was used for the parallel detection of
heat production on the surface of the catalyst beds, which is
considered to be proportional to catalytic activity. The setup for
parallel screening by means of ecIRT, as well as its measurement
x
mainly varied in oxygen as well as NO content, contained nitrogen
as carrier gas, and the reaction sequence started from the mixture
with highest O content (no. 1, Table 2).
2
A series of four IR images were recorded at certain times for
each reaction gas atmosphere, after 1, 5, 10, and 30 min of reaction
time as test on steady-state conditions. The images after 30 min
were always used later to evaluate the relative catalytic activities.
Each series of images was followed by the recording of a
background image at the reaction temperature under inert nitrogen
gas atmosphere for the subsequent emissivity correction.
5
9-62
principles, has been described in detail elsewhere.
For the IR
thermographic investigation, the library plate was inserted in a
gastight, gas flow reactor with an IR transparent sapphire window
on the top, allowing IR imaging of the library surface with an IR
camera (PtSi crystal detector with 256 × 256 pixel, Thermosensorik
Corp.). The whole experimental setup, including gas dosing,
temperature, and IR camera control, was managed automatically
2
E. Conventional Screening Experiments. Samples of 6 m /g
2
of Ce Zr O and 20 m /g Ce Zr O nanopowders were tested
0
.5 0.5
2
0.7 0.3
2
for catalytic activity toward NO reduction and propene oxidation
x
in a conventional gas-phase flow reactor. Thus, 0.070 g of the oxide
was loaded into a quartz U-tube flow reactor (i.d. 3.5 mm) and
subjected to an inlet gas composition of 688 ppm NO, 20 ppm
6
4
with the IRTestRig software.
Prior to the measurement, the library was pretreated at 380 °C
for 30 min in synthetic air (50 mL/min) to evaporate water and
oxidize volatile organic combustion residues (if present). After that,
a six-point temperature calibration was carried out in a range of
NO
2
, 711 ppm propene, 10 vol % O
2
2
, 2.5-3.0 vol % H O, and the
remainder He. The total gas flow rate was 74 mL/min. The reactor
was heated to selected temperatures, and the amounts of conversion
were obtained at steady state at each temperature (i.e., the sample
was kept under this inlet condition at a given temperature until the
outlet concentration of NO and NO
analysis was performed using an online Mattson Galaxy Series FTIR
(to determine N O, CO, and CO concentration) and a Thermo
Environmental 42CHL NO chemoluminescence analyzer (to
measure NO and NO concentration).
1
0 K in the area of -4 to +6 K around each specific reaction
temperature under nitrogen gas atmosphere. To examine the
catalytic activity of the library samples (for propane oxidation and
2
were constant). Product
NO
x
reduction), the reactor was held at 300 °C and five different
2
2
reaction gas mixtures (see Table 2) passed successively through
the system with a constant flow rate of 50 mL/min. The mixtures
x
2
Figure 1. Images of (a) a slate library plate filled with 48 different catalyst powders and (b) layout of the library plate.
210 J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009
9

High-Throughput Screening of Nanoparticle Catalysts
A R T I C L E S
x 2 3 x 2 3 0.9
Figure 2. SEMs of (a) (CeO )0.5(Al O )0.5 and (b) (CeO )0.1(Al O ) .
x 2 3 x 2 3
Figure 3. TEMs of (a) (CoO )0.08(Al O )0.92 and (b) (CoO )0.80(Al O )0.20 samples. Note change in particle morphology.
1
. Pt Impregnation. Pt (0.5 wt %) was deposited onto the
individual particles but does provide a view of the general
particle population. These SEMs indicate that the particle
populations produced here do not include any obvious microme-
ter size particles.
2
surface of 20 m /g of Ce0.7Zr0.3
O
2
using standard methods described
The material was calcined at 500 °C for 5 h and
then reduced with H at 450 °C for 2 h.
6
5-67
elsewhere.
2
Results and Discussion
TEM images were used to assess particle morphologies and
sizes of as-prepared powders. Discussions of actual size/size
distributions are not appropriate if based solely on TEM
micrographs, unless combined with XRD line broadening results
and/or specific surface area measurements.
In the following sections, we first provide some characteriza-
tion data for selected catalyst powders that were used to produce
the powder compacts used in the HTE studies. Thereafter, we
discuss HTE studies and identify some of the more unusual
results of these studies.
A. Catalyst Characterization. A variety of LF-FSP powders
have been made and discussed in the literature.
we provide some selected examples of powder characterization
to provide the basis for selected comments to be made in the
HTE studies.
In Figure 3, most of the particles are spherical and well below
8
0 nm in diameter, with the vast majority <30 nm. Note the
change in morphology, which is mostly faceted and spherical
for the (CoO 0.92 sample (Figure 3a), while nearly
all particles in the (CoO 0.20 sample in Figure 3b
24,29,38-40
Below
x 2 3
)0.08(Al O )
x 2 3
)0.80(Al O )
appear rhombic. This is attributed to the formation of the highly
crystalline spinel phase coincident with increases in Co content.
Particle necks, while occasionally visible in the 8 mol % sample,
are not a major morphological feature.
SEM was used to demonstrate powder uniformity. Figure 2
shows samples of (CeO
x 2 3 x 2 3 0.9
)0.5(Al O )0.5 and (CeO )0.1(Al O ) ,
demonstrating that SEM resolution is insufficient to reveal
Figure 4 images are high-resolution TEM micrographs of
(
2 x 2 3
ZrO ) (Al O )1-x nanopowders. The particle sizes here are
(
(
64) Scheidtmann, J. Ph.D. Thesis, Universit a¨ t des Saarlandes, Saar-
br u¨ cken, 2003.
typically <30 nm in diameter with most <20 nm. Figure 4b
shows clearly visible lattice planes indicating a high degree of
crystallinity. The lattice planes in a single particle are unidi-
rectional, suggesting single crystal particles formed during rapid
quench from the gas phase.
65) Seker, E.; Cavataio, J.; Gulari, E.; Lorpongpaiboon, P.; Osuwan, S.
Appl. Catal. A 1999, 183, 121–134.
66) Seker, E.; Gulari, E. Appl. Catal. A 2002, 232, 203–217.
67) Seker, E.; Yasyerli, N.; Gulari, E.; Lambert, C.; Hammerle, R. H.
Appl. Catal. B 2002, 37, 27–35.
(
(
J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009 9211

A R T I C L E S
Weidenhof et al.
x 2 3 0.20
Figure 4. (a) TEM and (b) HR-TEM images of (ZrO )0.80(Al O ) .
2 2 3 x 2 3
Figure 5. Particle size distribution of (a) (ZrO )0.54(Al O )0.46 and (b) (CeO )0.50(Al O )0.50 by dynamic light scattering.
Besides XRD and high-resolution microscopes, dynamic light
scattering (DLS) was used to estimate average particle sizes
one distinguishes between the external mass diffusion (film
diffusion) and the internal mass diffusion (pore diffusion) of
reactants or products. The latter should not affect the reaction
rate of the examined nanocatalysts since they are not porous.
On the other hand, the agglomeration of the particles forms voids
within the agglomerates, which represent diffusion barriers
similar to micro- and mesopores. In view of this, a direct
comparison between the catalytic performances of the examined
mixed-metal oxide nanopowders and the used benchmark
(
2
APSs). According to Figure 5, the APSs from DLS are about
0 nm higher than previous XRD results. These results suggest
that we are measuring the hydrodynamic radius of the water
layer on the nanoparticles in the DLS experiments.
The catalytic activity of an oxide is usually related to its
textural properties, in particular, to its specific surface area.
Therefore, in addition to particle size analyses based on DLS
and X-ray line broadening, the specific surface area of selected
catalysts was determined by BET methods to corroborate APSs.
The results are reported in Table 3. The catalysts are listed
according to their chemical compositions in order to allow a
direct comparison between different classes of mixed oxides.
2 3
materials (Pt on Al O and Hopcalite) appears only conditionally
favorable since their physical properties (porosity, surface area)
are fundamentally different. Particularly, the much higher
surface areas of the two reference materials should be taken
into account when they are ranked according to their activity
against the nanopowders. Furthermore, the degree of Pt metal
dispersion in the commercial catalysts is not known sufficiently
accurately to allow a direct comparison of catalyst activities.
B. High-Throughput Experiments. 1. Discussion. The ap-
plication of high-throughput and combinatorial techniques in
the development of new materials allows one to significantly
enhance the number of experiments carried out during a given
time. In contrast to conventional one-at-a-time approaches, high-
throughput research methods make use of materials libraries
containing numerous different components which are synthe-
sized and tested for the desired application in a highly parallel
or fast sequential manner with a high degree of automation and
2
-1
The materials have surface areas ranging from 26 to 66 m g .
In some groups, the increase or decrease in surface area
correlates with the changing content of one mixture component
(
x
e.g., Co Al100-x), whereas in other cases, the surface area is
not affected by this parameter.
Note that, due to the very small particle sizes of these flame
made materials, they are completely nonporous but have a
tendency to agglomerate (see Figure 2); otherwise, they are
3
8
easily dispersed. These characteristics have essential conse-
quences for the catalytic properties of these materials because
the kinetics of a heterogeneously catalyzed reaction are generally
influenced by mass transport processes. In connection with this,
9
212 J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009

High-Throughput Screening of Nanoparticle Catalysts
A R T I C L E S
Table 3. Main Physical and Structural Properties of Selected Catalyst Powders
SSABET
2
APSBET
[nm] ( 5%
APSXRD
[nm] ( 5%
catalyst
short description
Al100
[m g-1] ( 5%
structure
Al
CeO
CeO
2
O
2
3
66
45
60
63
57
60
57
50
45
45
32
32
32
33
36
45
42
53
53
51
38
30
26
69
58
45
54
57
57
47
53
51
103
174
29
20
24
22
22
20
20
20
20
20
28
28
28
26
22
22
20
31
30
30
38
35
40
18
19
23
28
26
26
30
22
22
29
20
14
17
14
15
15
16
18
20
16
16
16
16
16
13
16
18
21
20
22
21
22
20
22
20
14
18
15
13
12
12
Ce100
cubic
magnetoplumbite
cubic
cubic
cubic
cubic
cubic
cubic
cubic
tetragonal
tetragonal/cubic
cubic
cubic
cubic
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
(
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
)
)
)
)
)
)
)
)
)
)
)
)
)
)
)
0.05(Al
0.075(Al
0.10 (Al
2
O
3
)
0.95
Ce5.00Al95.0
Ce7.50Al92.5
Ce10.0Al90.0
Ce15.0Al85.0
Ce30.0Al70.0
Ce50.0Al50.0
Ce75.0Al25.0
Ce95.0Al5.00
Ce10.0Zr90.0
Ce30.0Zr70.0
Ce50.0Zr50.0
Ce70.0Zr30.0
Ce90.0Zr10.0
Ce49.0Zr21.0Al30.0
Ce63.0Zr27.0Al10.0
Co4.00Al96.0
Co8.00Al92.0
Co21.0Al79.0
Co50.0Al50.0
Co87.0Al13.0
Co94.0Al6.00
Ni43.0Al57.0
Ni63.0Al47.0
Ni78.0Al22.0
Zr2.40Al97.6
Zr4.10Al95.9
Zr5.90Al94.1
Zr13.6Al86.4
Zr49.7Al50.3
Zr79.4Al20.6
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CeO
CoO)0.04(Al
CoO)0.08(Al
CoO)0.21(Al
CoO)0.50(Al
CoO)0.87(Al
CoO)0.94(Al
2
O
O
3
)
)
0.935
2
3
0.90
0.15(Al
0.30(Al
0.50(Al
0.75(Al
0.95(Al
2
2
2
2
2
O
3
O
3
O
3
O
3
O
3
)
0.85
0.70
0.50
0.25
0.05
)
)
)
)
0.1(ZrO
0.3(ZrO
0.5(ZrO
0.7(ZrO
0.9(ZrO
2
)
2
)
2
)
2
)
2
)
2
2
0.9
0.7
0.5
0.3
0.1
0.49(ZrO
0.63(ZrO
)
)
0.21(Al
0.27(Al
2
O
O
3
)
)
0.3
0.1
cubic
2
3
2
2
2
2
2
2
O
3
)
0.96
0.92
0.79
0.50
0.13
0.06
O
3
O
3
O
3
O
3
O
3
)
)
)
)
)
NiO)0.43(Al
NiO)0.63(Al
NiO)0.78(Al
2
O
O
O
3
)
)
)
0.57
0.47
0.22
spinel
spinel
NiO/spinel
delta
delta
2
2
3
3
ZrO
ZrO
ZrO
ZrO
ZrO
ZrO
2
2
2
2
2
2
)
)
)
)
)
)
0.024(Al
2
2
2
2
2
2
O
O
O
O
O
O
3
)
3
)
3
)
3
)
3
)
3
)
0.976
0.959
0.941
0.864
0.503
0.206
0.041(Al
0.059(Al
0.136(Al
0.497(Al
0.794(Al
delta
tetragonal/ delta
tetragonal/ delta
tetragonal/ delta
nearly amorphous
amorphous
Pt/Al
2
O
3
(reference)
Pt/Al
Hopcalite
2 3
O
Hopcalite (reference)
5
0
miniaturization. Hence, these methods provide a perfect tool
to accelerate the discovery and optimization of new heteroge-
neous catalysts. However, in order to seek out new and, in
particular, formally unknown lead catalyst compositions, a
nearly infinite chemical parameter space has to be explored.
Even with high-throughput methods, it is impossible to sys-
tematically search through such a huge space. Therefore, HTE
has to be combined with suitable discovery and/or optimization
strategies in order to effectively search for new catalysts in a
light of the numerous exclusive advantages provided by LF-
FSP (e.g., one-step preparation of multicomponent nanosized
mixed-metal oxides), this restriction appears to be just a small
hindrance, which can easily be overcome by a supplementary
automation and parallelization of the method. Consequently, this
study is regarded as an initial step for further investigations in
this field.
The prepared metal oxides as well as the commercially
available references (see Table 1) were used to generate a
starting library of 48 materials. Emissivity-corrected infrared
thermography (ecIRT) was used for the rapid parallel screening
of the library to monitor the heat production on the surface of
the materials under different reaction gas atmospheres, followed
by quantification of the individual heat spots with appropriate
software.
6
1
reasonable time-to-benefit ratio.
A high-throughput study often starts with an initial screening
of several highly diverse starting libraries with the intention to
cover the selected parameter space. The results of these
experiments (identification of active materials) provide the basis
for the selection of promising materials for the next generations
of libraries with reference to the chosen development strategy.
The synthesis of these libraries normally comprises a compo-
sitional variation and/or an additional doping with different
elements to obtain further improvements of the catalytic activity.
The results presented in this work are considered as part of
a prescreening sequence of different binary and ternary flame-
made metal oxides with the main intention to evaluate the
general applicability of these materials for the selective catalytic
reduction of NO with propane (propene) as well as for propane
In general, the exposure of a catalyst surface to a reaction
gas mixture leads either to heat production, if exothermic
reactions take place, or to heat consumption due to endothermic
reactions. The total sum of heat emitted is proportional to the
linear combination of catalytic activity and the enthalpies of
62
all occurring reactions. Therefore, the visualization of reaction
heats by means of ecIRT provides no detailed information about
the selectivity of the analyzed catalysts. This is a critical fact
particularly for reactions with parallel and sequential side
reactions because the recorded heat increase may be generated
not only by higher catalytic activity but also by an increase in
side reactions. The measurement conditions chosen in this study
(see Experimental Section) enable several possible reaction
pathways for propane and NO (see eqs 2-9 in Table 4). The
(
propene) combustion in comparison to commercial benchmark
catalysts. As already described in the prior section, all examined
oxides were prepared by LF-FSP with an average production
rate of up to 10 materials per week. Compared to other well-
proven high-throughput synthesis techniques,
to be a major bottleneck in our approach. Nevertheless, in the
5
4-57
this seems
2
desired selective reduction of NO with propane to N (eq 5)
J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009 9213

A R T I C L E S
Weidenhof et al.
Table 4. Chemical Equations and Calculated Reaction
to produce organo-nitrogen species, which hydrolyze to form
68,69
Enthalpies
NH
3
, R-NCO, or R-CN (g or ads). The hydrolyzed intermediates
/NO or surface nitrates to produce N . Accord-
0
chemical equation
∆
R
H
298 [kJ/mol]
react with NO
2
2
ingly, the adsorbed oxygen appears in this mechanism as an
activator for NO as well as for hydrocarbons and furthermore
the remaining gas-phase oxygen is able to oxidize NO to more
C H (g) + 5O (g) f 3CO (g) + 4H O(g)
-2219.9
3
8
2
2
2
(2)
reactive NO
for the whole process because it rapidly reacts with some other
key intermediates such as R-CN to form N
Apart from oxygen’s beneficial properties, it also has an
undesired effect. At higher temperatures, the unselective
combustion of hydrocarbons becomes much faster than selective
2 2
. NO is also proven to have an important function
7
C H (g) + / O (g) f CO (g) + CO(g) +
-
1372.1
2
.
3
8
2
2
2
C(s) + 4H O(g) (3)
2
1
catalytic reduction of NO. Therefore, the conversion of NO
starts to decrease as the temperature increases.
-57.2
x
NO(g) + / O (g) h NO (g)
(4)
2
2
2
Independent of the true reaction mechanism, the adsorption
of NO and hydrocarbon combined with their activation on
catalytic sites has a significant influence on reaction initiation.
Furthermore, the relevance and rate of each mechanistic step
depends strongly on the nature of the active site(s) and the
experimental conditions. Note that both mechanisms were
postulated to occur based on studies using unsaturated hydro-
carbons (e.g., propene) as reducing agent and are therefore just
conditionally applicable to the results discussed below. More
specifically, already in 1997 Burch et al. pointed out that there
is a remarkable difference between alkanes and alkenes in the
reduction of NO by hydrocarbons over Pt catalysts under lean-
burn conditions.
4
1
NO(g) + C H (g) + 3O (g) f 2N (g) +
-2581.1
-3122.9
-2318.5
3
8
2
2
3
CO (g) + 4H O(g) (5)
2 2
0NO(g) + C H (g) f 5N (g) + 3CO (g) +
3
8
2
2
4
H O(g) (6)
2
73
9
2
NO(g) + C H (g) + / O (g) f N O(g) +
3 8 2 2 2
3
CO (g) + 4H O(g) (7)
2 2
However, the two proposed mechanisms at least allow
classification of the reactions in Table 4 into promotional or
obstructive side reactions with regard to propane-SCR. Obvi-
C H (g) f C H (g) + H (g)
(8)
(9)
+124.2
-90.3
3
8
3
6
2
ously, the pure combustion of propane to CO
2
and water (eq 2)
1
1
NO(g) f / N (g) + / O (g)
as well as the conversion of NO to N O (eq 7) must be
2
2
2
2
2
considered as undesired side reactions because both are strongly
exothermic and therefore give misleading activity results in the
ecIRT screening. As already mentioned, at higher temperatures,
propane combustion is the favored reaction, whereas the
and nearly all other reactions (with the exception of reaction 8)
are exothermic and might occur either alone or simultaneously
and hence are recorded altogether during the ecIRT measurement.
According to literature, the mechanisms for selective catalytic
2
conversion of NO to N O decreases with rising temperature after
going through a maximum (in particular over noble metal
73
catalysts).
A complete suppression of these side reactions by modifying
the reaction conditions (oxygen exclusion from feed gas,
decrease of reaction temperature) is not possible because the
desired test reaction (eq 5) requires a specific and exclusive
reaction path in the energy hyper-surface. However, the prob-
ability of finding this situation without additional local minima
for catalysts is very low. Also, the influence on parameters (e.g.,
temperature) is limited because very low heat emissions are
difficult to record by ecIRT. Consequently, the results reported
here allow no general conclusions about the catalytic selectivity
of a particular active sample. These initial studies are useful
both for excluding inactive materials and for identifying new
catalyst systems not expected based on the literature and
therefore represent successful discoveries of new points of
departure for further studies.
x
reduction of NO by hydrocarbons can be divided into two
classes. The so-called “adsorption/dissociation mechanism” is
regarded as applicable to describe reactions on noble metal and
4
,70,71
Cu-ZSM5 catalysts.
This mechanism involves adsorption
of NO on active metal sites, which then dissociates into N(ads)
and O(ads). Thereafter, combination of two N(ads) produces
N
2
, whereas O(ads) reacts with hydrocarbons to form CO
Remaining undissociated NO reacts with N(ads) to form N O.
The second, “oxidation-reduction mechanism” seems to be
2
.
2
4
,72
suitable for metal-oxide catalysts.
anism (for illustration see literature reference), NO reacts first
with adsorbed oxygen to form reactive intermediates like NO
or NO (ads). Simultaneously, the hydrocarbon is oxidized at
According to this mech-
2
3
the catalyst surface to hydrocarbon oxygenates (e.g., acetate).
These oxygenates undergo further reactions with surface nitrates
2
. Results. For the high-throughput screening, a reaction
temperature of 300 °C was chosen and the oxygen content of
the feed gas was varied between 0 and 20 vol %. This intentional
variation was used to discover materials that exhibit heat
production (catalytic activity) at low oxygen concentrations.
Furthermore, in one part of the applied screening program, NO
was omitted from the feed gas to explore the catalytic activity
(
(
(
68) Binnewies, M.; Milke, E., Thermochemical Data of Elements and
Compounds, 2nd ed.; Wiley-VCH: Weinheim, Germany, 1999.
69) Barin, I.; Knacke, O.; Kubaschewski, O., Thermochemical Properties
of Inorganic Substances; Springer-Verlag: Berlin, 1977.
70) Burch, R.; Millington, P. J.; Walker, A. P. Appl. Catal. B 1994, 4,
6
5–94.
(
71) Walker, A. P. Catal. Today 1995, 26, 107–128.
72) Meunier, F. C.; Zuzaniuk, V.; Breen, J. P.; Olsson, M.; Ross, J. R. H.
Catal. Today 2000, 59, 287–304.
(
(73) Burch, R.; Watling, T. C. Catal. Lett. 1997, 43, 19–23.
9
214 J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009

High-Throughput Screening of Nanoparticle Catalysts
A R T I C L E S
Figure 6. Emissivity-corrected IR thermographic images of catalyst library after 30 min under reaction gas atmosphere containing 20 vol % O
propane, 0.2 vol % NO, and N as balance at 300 °C (a) or 0.4 vol % propane, 0.2 vol % NO, and N as balance at 300 °C (b). 1: Hopcalite; 2: 5 wt % Pt
on Al
2
, 0.4 vol %
2
2
O .
2 3
of all library samples for pure combustion of propane. Prior to
the measurement, the library was pretreated at 380 °C for 30
min in a flow of synthetic air for 30 min to clean the catalysts’
surfaces from water and organic combustion residues (if
present).
by the combustion of propane, which is inhibited when oxygen is
excluded from the feed gas.
The second reference Pt supported on Al O is characterized
2 3
by a lower heat emission under oxygen-rich conditions com-
pared to Hopcalite, but in contrast, it seems to retain some
activity when oxygen is excluded from the feed gas. It is long
proven that this noble metal catalyst has a certain activity for
Note that the following discussion of the high-throughput
screening results is based on a ranking of the different examined
catalysts according to their heat production on the surface under
reaction conditions, which is considered to be proportional to
catalytic activity. In this context, the concept of catalytic activity
is therefore not used in terms of a specific rate of reaction
because it is not possible to record specific reaction rates by
means of ecIRT. The identification of a certain catalyst as highly
active has to be understood in terms of a higher heat emission
of this material in comparison to all other library components.
Figure 6 shows two emissivity-corrected IR thermographic
images of the examined library under oxygen-rich as well as
under oxygen poor-reaction conditions. The yellow to red
colored dots in the images mark materials that show heat
production after exposing the library for 30 min to a selected
reaction gas mixture at 300 °C. In general, the appearance of
numerous heat spots beside the expected ones from the
2
the reduction of NO to N with hydrocarbons in the presence
7
0
of excess oxygen (eq 5). On the other hand, this catalyst is
also characterized by a comparatively poor selectivity toward
2 2
N (due to the formation of N O) and a narrow activity
2
temperature window. The recorded heat emission over this
reference is therefore attributed to several parallel reactions. It
is obvious that under excess oxygen, reactions 2, 5, and 7 of
Table 4 are the primary contributors to the total heat. Which of
these reactions take place preferentially under the selected
conditions is not derivable from the measurement results.
The fact that this catalyst additionally shows slight heat
emission in the absence of oxygen suggests a potential catalytic
activity for reactions involving just propane and/or NO. Possible
reaction pathways that have to be mentioned in this relationship
are the dehydrogenation of propane (eq 8), the direct conversion
of propane with NO (eq 6), and the decomposition of NO (eq
9). Pt and, in particular, Pt-Sn supported on alumina are used
widely in industry as catalyst for the selective dehydrogenation
2 3
commercial benchmarks (Hopcalite and Pt on Al O ) indicates
a potential capability of the analyzed flame-made mixed oxides
for catalytic oxidation.
77,78
of propane to propene.
Since this is an endothermic reaction,
From a simple comparison of the two images, it becomes
obvious that all samples demonstrate higher heat emission under
oxygen-containing reaction conditions. This result is not really
surprising but confirms the already discussed crucial role of oxygen
to function as an activator for several chemical processes.
One of the most notable findings is the high activity of the
reference Hopcalite under oxygen excess, whereas the same catalyst
becomes nearly inactive when oxygen is excluded from the feed
gas. This result is in good agreement with literature since Hopcalite,
a mixture of copper(II) and manganese(IV) oxides, is well-known
for its high catalytic activity in low-temperature oxidation of CO
which requires relatively high temperatures (ca. 500-600 °C),
even in the presence of a catalyst, it cannot proceed under the
conditions used here.
Furthermore, a reaction between NO and propane without
oxygen as an additional reaction partner (eq 6) seems to be
kinetically unfavorable because propane (as saturated hydro-
carbon) is comparatively chemically inert, and a large excess
of NO is required to convert 1 equiv of propane to CO
O. The decomposition of NO into N and O is slightly
exothermic and known to be catalyzed by La -, Ba/MgO-,
3
and LaCoO -based perovskite oxides. However, this reaction
2
and
H
2
2
2
2
O
3
7
9
7
4
to CO
2
and in the catalytic combustion of volatile organic
75,76
proceeds, like propane dehydrogenation, only at high temper-
atures (ca. 800 °C) and therefore cannot be responsible for the
recorded heat emission.
compounds (VOC).
Carbon monoxide is not present in the
initially injected reaction gas mixture. This fact suggests that the
visualized heat production over this catalyst is primarily induced
(
77) Kogan, S. B.; Schramm, H.; Herskowitz, M. Appl. Catal. A 2001,
208, 185–191.
(
74) Lamb, A. B.; Bray, W. C.; Frazer, J. C. W. J. Ind. Eng. Chem. 1920,
1
2, 213–221.
(78) Yu, C.; Ge, Q.; Xu, H.; Li, W. Catal. Lett. 2006, 112, 197–201.
(79) Ishihara, T.; Ando, M.; Sada, K.; Takiishi, K.; Yamada, K.;
Nishiguchi, H.; Takita, Y. J. Catal. 2003, 220, 104–114.
(
(
75) Lintz, H. G.; Wittstock, K. Appl. Catal. A 2001, 216, 217–225.
76) Musick, J. K.; Williams, F. W. Product R&D 1975, 14, 284–286.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009 9215

A R T I C L E S
Weidenhof et al.
Figure 7. Emissivity-corrected IR thermography screening results for seven groups of nanoparticle catalysts under five different reaction gas atmospheres.
Ce Zr Cu100-x (e), and Co
x
Al100-x (a), Ce
x
Zr100-x (b), Zr
x
Al100-x (c), Ce
x
y
Al100-(x+y) (d), Ni
x
Al100-x/Ni
x
x
Al100-x (f).
The most reasonable explanation for the apparent warming
under oxygen exclusion is presumably due to emissivity changes
and not by any type of catalytic reaction. Only in the case of Pt
by bar diagrams in Figure 7. Most of the materials exhibit the
highest heat radiation under pure combustion conditions, meaning
in excess O and in the absence of NO in the feed gas. The addition
2
on Al
2
O
3
a certain degree of NO reduction cannot be ruled out.
of NO to the reaction gas atmosphere depresses heat emission from
the different materials in most cases. Heat reduction is also observed
with decreasing O contents in the feed gas, resulting in the absence
2
This conclusion is valid in the same way for all other catalysts
that exhibit heat emission under oxygen exclusion.
With reference to chemical composition, the different library
of oxygen in the low values already discussed. This activity decline
is very uniform for some of the samples, while other candidates
fluctuate.
With the exception of the oxygen-free test segment, the
Hopcalite benchmark is always located at the top of the activity
samples are classified as follows: (I) Ce
III) Zr Zr
Ni Al100-x, and (VII) Ni Cu100-x. The recorded temperature
changes for every sample within these seven groups are illustrated
x x
Al100-x, (II) Ce Zr100-x,
(
x
Al100-x, (IV) Ce
x
y
Al100-x, (V) Co Al100-x, (VI)
x
x
x
9
216 J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009

High-Throughput Screening of Nanoparticle Catalysts
A R T I C L E S
ranking lists, whereas several samples from groups (I) to (IV)
offer activities superior to the Pt supported on Al O reference
2 3
catalyst. The mostly higher activities of both reference catalysts
are expected since their specific surface areas are very large in
comparison to all other materials in the library. This makes it
all the more surprising that some nanopowders offer similar or
alteration in oxidation states induced by a partial surface
reduction of these materials with NO and propane, the applied
emissivity correction is no longer correct, leading to “false
negatives” in the obtained IR delta image.
The most interesting behaviors are those of materials from
groups (I) to (IV), in particular the different mixtures of CeO2
slightly higher heat emissions than Pt/Al
lower surface areas.
2
O
3
given their much
with ZrO (group (II)), which provide the highest heat emissions
2
of all examined oxides with the exception of the reference
When all diagrams are compared, it becomes obvious that
the mixed-metal oxides from group (V) Co
Ni Al100-x, and (VII) Ni Cu100-x demonstrated low activity under
catalysts. However, a direct correlation of composition with
catalytic activity is not reflected by the data, although higher
ZrO contents seem to have a positive effect. Apart from the
2
x
Al100-x, (VI)
x
x
all reaction conditions used here. In addition, the mixed-metal
oxides containing higher Co contents or mixed-metal oxides of
Ni and Cu demonstrated a surprising cooling behavior under
ceria-zirconia solid solutions, the single oxides of Ce and Zr
also exhibit comparatively high heat emissions.
In view of the already quite substantial research efforts in
the application of ceria and related materials for catalysis, in
reaction atmospheres with low O
2
concentrations. Previous
studies already showed that pure or supported oxides of
particular for automobile exhaust gas conversion, these findings
8
0-82
83-85
14,91
cobalt
reduce NO with propane or propene in the presence of excess
. However, these earlier results are not confirmed by our
and copper
exhibit an appreciable capacity to
are not completely new.
For many years, ceria has been
the chief oxygen storage component for three-way catalysts,
and the mechanisms by which it works have been the subject
O
2
92
investigation. These results can presumably be explained by
dissimilar test conditions and essential differences with regard
to the catalyst preparation methods used resulting in completely
different catalytic properties.
of many studies. Another important catalytic property of ceria
is its ability to completely oxidize hydrocarbons along with CO
and H . For these reasons particularly, the catalytic combustion
2
of hydrocarbons on ceria has been studied for a long time. For
Note that cobalt (in particular Co
be useful in improving sulfur poisoning resistance in NO
storage and reduction (NSR) catalysts. Furthermore, Co has
been suggested as a promoter for platinum (oxidizing agent) in
3
O
4
) and copper appear to
example, total oxidation of parafins as well as unsaturated
hydrocarbons on ceria can be achieved between 300 and 500
°C.
x
8
6
14
A main disadvantage of ceria is its poor resistance to sintering.
This is a major problem particularly for catalysts where specific
surface areas should be as high as possible. Therefore, many
attempts have been made to improve the sintering resistance of
ceria-based catalysts by doping or mixing ceria with other
oxides. An important breakthrough in this context was the use
of zirconia as a dopant because zirconium ions can enter the
ceria network without too much stress on the lattice, and
therefore, solid solutions of ceria and zirconia with many
different compositions have been synthesized successfully and
characterized with regard to thermal stability, oxygen storage
NSR catalysts because it improves NO
x
storage and conversion
The storage of NO is an
endothermic process which comprises the oxidation of NO to
NO over the oxidation agent followed by a subsequent
adsorption of NO by means of disproportion to form nitrates
at the surface of a alkaline-earth metal-containing material
8
7,88
efficiencies in these catalysts.
x
2
2
8
9,90
(
typically Ba).
Taking this into account, it is unlikely that
storage processes are responsible for the cooling seen for
Cu100-x and Co containing catalysts as an NO storage
NO
x
the Ni
x
x
component like barium is absent in these catalysts. Additionally,
the apparent cooling increases with reducing oxygen content
in the feed gas, but the oxidation of NO (eq 4) usually proceeds
under lean conditions, meaning under excess oxygen.
However, a more realistic explanation could be a reversible
change in the emissivity characteristics of these compositions
during the course of the measurement. For example, due to an
9
3
capacity, sintering resistance, and catalytic properties.
Numerous studies concerning the catalytic combustion of
1
4,94-96
volatile organic compounds
NO
x
and the decomposition of
9
7,98
over ceria-zirconia catalysts have been reported in
the catalysis literature, which emphasizes the general utility of
these materials for environmental catalysis. Taking this into
account, the results found in our investigation can be regarded
as additional verification of previous studies, but it also
demonstrates the applicability of liquid-feed flame spray py-
rolysis as an appropriate synthesis method for ceria-zirconia
catalysts.
(
(
(
(
(
(
80) Yan, J.; Kung, M. C.; Sachtler, W. M. H.; Kung, H. H. J. Catal.
1
997, 172, 178–186.
81) Liotta, L. F.; Pantaleo, G.; Macaluso, A.; Di Carlo, G.; Deganello,
G. Appl. Catal. A 2003, 245, 167–177.
82) Horiuchi, T.; Fujiwara, T.; Chen, L.; Suzuki, K.; Mori, T. Catal.
Lett. 2002, 78, 319–323.
As shown in the diagrams a, b, and c in Figure 7, the
combination of CeO or ZrO with Al O does not lead to drastic
2 2 2 3
83) Bennici, S.; Gervasini, A.; Ravasio, N.; Zaccheria, F. J. Phys. Chem.
B 2003, 107, 5168–5176.
84) Bennici, S.; Carniti, P.; Gervasini, A. Catal. Lett. 2004, 98, 187–
1
94.
(91) Gonz a´ lez-Velasco, J. R.; Guti e´ rrez-Ortiz, M. A.; Marc, J.-L.; Botas,
J. A.; Gonz a´ lez-Marcos, M. P.; Blanchard, G. Appl. Catal. B 1999,
22, 167–178.
85) Metelkina, O. V.; Lunin, V. V.; Sadykov, V. A.; Alikina, G. M.;
Bunina, R. V.; Paukshtis, E. A.; Fenelonov, V. B.; Derevyankin,
A. Y.; Zaikovskii, V. I.; Schubert, U.; Ross, J. R. H. Catal. Lett.
(92) Trovarelli, A. Catal. ReV. 1996, 38, 439–520.
2
002, 78, 111–114.
(93) Di Monte, R.; Kaspar, J. J. Mater. Chem. 2005, 15, 633–648.
(94) de Rivas, B.; Guti e´ rrez-Ortiz, J. I.; L o´ pez-Fonseca, R.; Gonz a´ lez-
Velasco, J. R. Appl. Catal. A 2006, 314, 54–63.
(
(
(
86) Clacens, J.-M.; Montiel, R.; Kochkar, H.; Figueras, F.; Guyon, M.;
Beziat, J. C. Appl. Catal. B 2004, 53, 21–27.
87) Vijay, R.; Snively, C. M.; Lauterbach, J. J. Catal. 2006, 243, 368–
(95) D ´ı az, E.; de Rivas, B.; L o´ pez-Fonseca, R.; Ord o´ n˜ ez, S.; Guti e´ rrez-
Ortiz, J. I. J. Chromatogr. A 2006, 1116, 230–239.
3
75.
88) Vijay, R.; Hendershot, R. J.; Rivera-Jim e´ nez, S. M.; Rogers, W. B.;
(96) Guti e´ rrez-Ortiz, J. I.; de Rivas, B.; L o´ pez-Fonseca, R.; Gonz a´ lez-
Velasco, J. R. Appl. Catal. B 2006, 65, 191–200.
Feist, B. J.; Snively, C. M.; Lauterbach, J. Catal. Commun. 2005, 6,
1
67–171.
(97) Adamowska, M.; Muller, S.; Da Costa, P.; Krzton, A.; Burg, P. Appl.
Catal. B 2007, 74, 278–289.
(
(
89) Fridell, E.; Skoglundh, M.; Westerberg, B.; Johansson, S.; Smedler,
G. J. Catal. 1999, 183, 196–209.
(98) Silveira, E. B.; Perez, C. A. C.; Baldanza, M. A. S.; Schmal, M.
Catal. Today 2008, 133-135, 555–559.
90) Hess, C.; Lunsford, J. H. J. Phys. Chem. B 2002, 106, 6358–6360.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009 9217

A R T I C L E S
Weidenhof et al.
2 3
increases in activity. As might be expected, increasing Al O
contents greatly reduces catalytic activities. In particular, for
compositions with g50 mol % Al , the activities fall to very
low levels compared to pure CeO and pure ZrO . Given that
LF-FSP produced materials along the CeO /Al and ZrO
Al tielines lead to core-shell materials where the alumina
2
O
3
2
2
x
2
O
3
2
/
2
O
3
is coating crystals of ceria or zirconia or solid solutions thereof,
we can easily imagine that gaseous access to the active
components is limited. In the view of this result, it is not really
surprising that combinations of all three oxides do not result in
supplementary improvements in catalytic activity. Nevertheless,
2 3
the ternary oxides with low Al O contents demonstrate activities
comparable to those provided by the binary Ce/Zr oxides.
9
9
An accompanying paper suggests that the higher ZrO
2
contents actually contain significant amounts of the partially
reduced species generally called zirconium suboxides (e.g.,
3
+
2+
ZrO2-x with some Zr /Zr ions). Furthermore, we recently
reported that LF-FSP of compositions along the CeO -Al
tieline produced solid-phase solutions at up to 5 mol % CeO
and at compositions up to 10 mol %, the materials consist
Figure 8. NO reduction/propene oxidation on 20 m
2
/g of Ce0.7Zr0.3O as
x
2
O
3
2 x
a function of temperature. Propene oxidizes (>90% to CO ) NO , reduces
2
2
6
5,66
to N (>99%) selectively.
2
2
,
1
00
primarily of the magnetoplumbite phase, CeAl11
after, at higher CeO contents, core-shell particles are observed
with CeO cores but with the shells containing considerable
amounts of Ce3 species probably as CeAl11
therefore to envision that, at low Al contents, the resulting
O
18
.
There-
2
2
+
O18. It is possible
2
O
3
3+
3+
shells contain considerable quantities of Ce and possibly Zr ,
which may contribute to the overall activity. Finally, given that
the LF-FSP often produces kinetic rather than thermodynamic
phases, it is also possible that even the “pure” Ce
solutions actually may be better considered to exist as
Ce 2-z partially reduced materials.
. Catalyst Testing. Complementary to the above test runs,
x 2
Zr1-xO solid
x
Zr1-xO
3
a set of experiments was run briefly to characterize the catalytic
properties on stream by using propene instead of propane for
reasons discussed below.
As noted above, numerous catalysts have been tested for
reduction of NO with hydrocarbons. As of now, there are no
catalysts that are robust and active enough to reduce the NO
emissions of diesel engines with the onboard hydrocarbon fuel,
which is the simplest and lowest cost solution to NO reduction.
The ideal catalyst must operate over a wide temperature range,
be resistant to poisoning by small amounts of SO in the exhaust,
x
x
Figure 9. NO reduction activities of three of the most active and studied
catalysts.
6
5,66
x
reached only above 300 °C. Furthermore, Cu-ZSM5 catalyst is
not stable in a humid atmosphere, while silver is deactivated
2
be cheap, and have a long service life under high-temperature
conditions. With these requirements in mind, and based on our
by water as well as sulfur dioxide. Ce0.7Zr0.3
stable and has a very steady conversion rate from 250 °C
actually the window extends to below 200 °C) to 550 °C,
2
O is thermally very
initial survey, the catalytic activity of two Ce
tions toward simultaneous NO reduction and propene oxidation
was investigated using conditions described in the Experimental
x 2
Zr1-xO composi-
(
x
ideally matching the diesel and lean-burn gasoline engine
exhaust emission temperature windows. Sensitivity to water
appears minimal, and conversion rates do not change signifi-
cantly with temperature.
Section. Figure 8 shows the results of a selective NO
x
reduction/
powder.
2
propene oxidation test made on 20 m /g of Ce0.7Zr0.3
O
2
These tests were run with propene for comparison with previous
Of special interest is that Ce0.7Zr0.3
a better catalyst than a Pt/Ce0.7Zr0.3
O
2
by itself appears to be
catalyst prepared from
studies on three traditional sol-gel catalysts, made in the Gulari
O
2
6
5
group at UM, as shown in Figure 9.
the same material (Figure 10). Finally, in addition to the
catalyst’s excellent behavior with NO , all of the propene used
While the specific peak activities of the catalysts shown in
Figure 9 are higher than that of Ce0.7Zr0.3 shown in Figure 8,
x
O
2
in the Figure 9 studies is oxidized at g430 °C, indicating a
high degree of catalyst activity for control of hydrocarbon
emissions, as well. Thus, it is likely that use of this same type
of catalytic material in standard TWCs will also work well.
all of the known catalysts in Figure 9 have significant limitations.
For example, the activity window of the Pt/alumina catalyst runs
from 150 to 300 °C and as such is too narrow for commercial
use (diesel engines need catalysts that are active at 200-600
°C). The peak activities of silver and Cu-ZSM5 catalysts are
Conclusions/Outlook
The objective of the work described here was to couple the
rapid sample output of LF-FSP for the production of high
surface area, nonporous mixed-metal oxide nanopowders to
high-throughput screening to demonstrate the utility of coupling
(
99) Kim, M.; Laine, R. M. J. Am. Chem. Soc. 2009, 131, http://dx.doi.org/
0.1021/ja9017545.
100) Kim, M.; Hinklin, T. R.; Laine, R. M. Chem. Mater. 2008, 20, 5154–
162.
1
(
5
9
218 J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009

High-Throughput Screening of Nanoparticle Catalysts
A R T I C L E S
Improvements will no doubt come with further efforts to
increase surface areas. Experiments will also be necessary to
understand the long-term aging of these catalysts plus additional
studies of compositional variations to optimize performance
under conditions which are different than those of interest for
the TWC systems. Finally, there is the potential opportunity to
develop these catalysts for gas turbine engines which currently
operate at temperatures of 1100-1300 °C, where simple noble
6
5-67
metal catalysts react with typical support materials.
With regard to compositional variation, we intend to combine
the most active representatives of the two systems with variable
contents of selected elements (single- and multielement doping)
in order to seek out new catalyst lead compositions by means
of combinatorial optimization circles. An additional interesting
question is related to the impact of different morphologies within
these systems as well as different particle sizes on the catalytic
activity. In this context, the catalytic properties of the mixed-
Figure 10. Selective NO
x
reduction and propene oxidation activity of 0.5
2
wt % Pt on 20 m /g of Ce0.7Zr0.3 nanopowder as a function of
O
2
6
5-67
temperature.
phase core-shell CeO
of Al will be further characterized, in particular, with regard
to the influence of the CeO species in the Al shell.
2 2 2 3
/ZrO /Al O particles with lower contents
2 3
O
these two methods toward the preparation and screening of well-
known as well as novel catalyst systems. We chose to study
x
2 3
O
Apart from that, the direct inkjet printing of ceramic suspen-
sions has become a well-established method to prepare libraries
deNO
materials.
We find that this approach works extremely well to screen a
x
catalysis as a means of screening a set of LF-FSP
101
of nanosized ceramic powders in the recent years. Therefore,
we are planning to incorporate this auspicious technique into
our library generation procedures to accelerate our research.
wide variety of LF-FSP produced materials. Because LF-FSP
produced nanopowders consist very often of kinetic rather than
thermodynamic phases, it is possible to produce materials that
are difficult to access or are completely inaccessible via
traditional catalyst preparation methods. Indeed, our studies have
Acknowledgment. The University of Michigan researchers
would like to thank AFOSR (Contract No. F49620-03-1-0389) and
a subcontract from UES Inc. (F074-009-0041/AF07-T009), Maxit
Inc., Diamond Innovations Inc., and NSF for support of the LF-
FSP synthesis work. We would also like to thank Dr. George W.
Graham for helpful discussions on TWC catalysts in general and
the results presented here.
identified a series of nano-Ce1-xZr
materials that offer surprisingly good activities for both NO
x
O
2
and Al
2
O
3
x 2
-Ce1-xZr O
x
reduction and propane/propene oxidation both in high-through-
put screening studies and in continuous flow catalytic studies.
Under the reaction conditions and with the model reductants,
such as propane and propene, used in this paper, LF-FSP
produced oxide catalysts offer activities comparable to traditional
Supporting Information Available: Additional tables. This
material is available free of charge via the Internet at http://
pubs.acs.org.
2 3
Pt/Al O catalysts but without the addition of Pt. Thus, they
are Pt-free, yet still active for extremely important emission
control reactions, especially considering these are only first
generation materials. For these oxide catalysts to be viable
alternatives, further testing is needed using the real exhaust of
a diesel engine and road conditions.
JA809134S
(
101) Chen, L.; Zhang, Y.; Yang, S.; Evans, J. R. G. ReV. Sci. Instrum.
2007, 78, 072210–072216.
J. AM. CHEM. SOC. 9 VOL. 131, NO. 26, 2009 9219