the superior performance of the Pt 1 Sn 1 /SiO 2
catalyst. Increasing the Sn:Pt atomic ratio on
SiO 2 -supported catalysts to the levels used in the
commercial Al 2 O 3 -supported catalysts (Pt:Sn =
1:5) led to poor performance (fig. S3A) ( 15 ).
Similarly, decreasing the Sn:Pt atomic ratio
on Pt-Sn/g-Al 2 O 3 to Pt 1 Sn 1 led to rapid catalyst
deactivation (fig. S3B).
The data in Fig. 1, A to C, were obtained with
16 vol % propane in helium. Propane streams
are diluted in commercial operations to in-
crease the limiting equilibrium conversion, which
is higher at lower propane partial pressure
(fig. S2), and to limit catalyst deactivation rates
( 7 , 16 , 17 ). This approach requires additional
downstream separation units and limits pro-
cess efficiency. In an undiluted propane stream,
the Pt 1 Sn 1 /SiO 2 catalyst reached the thermody-
namic conversion limit of 40% (Fig. 1D and fig.
S2), with >99% selectivity to propylene. Even
under these harsh, high-conversion and undiluted
propane stream conditions, the Pt 1 Sn 1 /SiO 2 cata-
lyst was stable and operated at the thermody-
namic conversion limit and >99% selectivity for at
least 30 hours on stream, whereas the PtSn/
g-Al 2 O 3 (commercial mimic) deactivated rapidly
under these conditions (Fig. 1D).
We also compared the performance of the
Pt 1 Sn 1 /SiO 2 catalyst to various catalysts previ-
ously tested in propane dehydrogenation ( 18 – 53 ).
Data in Fig. 2A and table S1 show the comparison
of the initial (fresh catalyst) selectivity-conversion
results reported for different catalysts. The
slanted lines in the figure quantify the prox-
imity of the measured conversion-selectivity
data points to the highest possible conversion-
selectivity lines for different conditions at which
these catalysts were tested. The highest possible
conversion-selectivity line has the value of 100%
when the catalyst operates at equilibrium con-
version and 100% propylene selectivity. By this
metric, the Pt 1 Sn 1 /SiO 2 catalyst outperformedother catalysts, reaching very close to the highest
possible conversion-selectivity line (100% selec-
tivity to propylene at equilibrium conversion),
whereas other tested catalysts were mostly
<80% of the highest possible performance. We
emphasize that on Pt 1 Sn 1 /SiO 2 , the highest
possible yield can also be achieved with un-
diluted propane feed (point labeled as 1 in Fig. 2A),
whichhasnotbeendemonstratedpreviously.
For example, PtSn/Al 2 O 3 nanosheets ( 45 ) reach
near-equilibrium conversion (point labeled as
42 in Fig. 2A); however, the propane is diluted
with hydrogen (C 3 H 8 :H 2 = 0.8) and an inert
gas, leading to low propane partial pressure.
The overall productivity of a PDH catalyst is
affected by the conversion-selectivity data as
well as the catalyst stability and the inherent
reaction rates. We developed a figure of merit
that combined these two metrics and quantified
the relative productivities of different catalysts.
To quantify the stability of the catalysts, the218 9JULY2021•VOL 373 ISSUE 6551 sciencemag.org SCIENCE
Fig. 1. Catalyst performance in propane dehydrogenation.(A) Propane
conversion, (B) propylene selectivity, and (C) propylene yield as function of time
on stream for supported PtSn catalysts with and without hydrogen added to
the feed. Reaction temperature (T) = 580°C, propane partial pressurePC 3 H 8
= 0.16,and WHSV = 4.7 hours−^1 .(D) Performance of Pt 1 Sn 1 /SiO 2 and PtSn/g-Al 2 O 3
(commercial mimic) for PDH with an undiluted propane stream.T= 580°C,
PC 3 H 8 ¼ 1 :0, and WHSV = 4.7 hours−^1. For Sn-promoted catalysts on an Al 2 O 3
support, the atomic ratio of Sn:Pt is 5; on a SiO 2 support, the atomic ratio is 1.RESEARCH | REPORTS