¨
ZALC, SOKOLOVSKII, AND LOFFLER
170
catalyst particles ranging 0.42–0.50 mm in size and mixed culated from experiments conducted for catalyst bed tem-
with glass beads was located inside a pipe with inert glass peratures of 250, 300, 350, and 400◦C.
beadsectionsupstreamanddownstream. Thestandardfeed
consisted of N2, CO, CO2, H2, and H2O. The gases (Scott
Specialty Gases) were fed to the reactor by calibrated mass
RESULTS AND DISCUSSION
flow controllers, while a syringe pump was implemented
Typical results for a 1.5% Pt/ceria system at 250◦C are
to feed deionized water to an evaporator unit upstream
shown in Fig. 1. CO conversion falls rapidly during the first
of the reactor configuration. A backpressure regulator
few hours to reach a constant rate of decay. Ignoring the
first two data points, α values in the figure can be fit with a
straight line. Based on the slope and intercept of the best-fit
line, which had a correlation coefficient of 0.96, the initial
controlled the vessel pressure. For the results presented
here, the catalyst was prepared by adsorption of diammine-
dinitritoplatinum(II) (Johnson Matthey) from nitric acid
solution onto a ceria–zirconia based support. The catalyst
activity and the deactivation coefficient were computed to
was dried at 110◦C overnight followed by calcination at
be 0.25 (mol/(gcat h−1 atm−2)) and 0.009 h−1, respectively.
500◦C for 4 h. The platinum loading determined by induc-
In Ref. 10, Bunluesin et al. had investigated water–gas
shift rates for ceria-supported noble metals using a feed
tively coupled plasma analysis was 1.5 wt%. For all experi-
mental conditions considered, diffusional resistances were
consisting of CO, water, and N2 under differential condi-
negligible, asindicatedbyhighlylinearArrheniusplotswith
tions. Rates were observed to be stable over several hours
initial activation energies around 70 kJ/mol. Effluent gases
anddiffusionallimitationswereabsent. Basedontheirdata,
were analyzed via gas chromatography and during each run,
turnover frequencies can be estimated to be about 0.8 s−1 at
the outlet carbon monoxide conversion was recorded as a
393◦C, 0.2 s−1 at 300◦C, and 0.04 s−1 at 200◦C. The turnover
function of time-on-stream.
frequencies computed for our Pt/ceria formulations were
The rate expression used for comparison of catalyst per-
very similar to these values. It should be noted here that
formance is given by Eq. [1]:
turnover frequencies far below 1 are considered to be too
low for practical applications (12). Indeed, rough estimates
rwgs = kꢁ(pCO pH O − pCO pH /Kwgs).
[1]
of catalyst requirements based on initial activity reveal that
approximately 50 kg of 1.5 wt% Pt catalyst would be needed
for a 50-kW system. The reactor volume needed to contain
such an amount of catalyst would be impractical in an au-
tomotive environment; the use of 750 g of platinum per
reactor also raises serious cost issues.
In addition to exhibiting insufficient activity for trans-
portation applications, Pt/ceria WGS catalysts deactivate
rapidly. In fact, deactivation rates around 0.008 h−1 were
observed for all the Pt-based WGS catalysts tested in our
laboratory. In all cases, those deactivation rates were largely
independent of temperature. A deactivation coefficient of
2
2
2
Here, rwgs [=] (mol of CO/(g catalyst h)) is the local rate of
reaction of carbon monoxide. It should be noted that this
kinetic equation is not based on a mechanistic approach;
rather it provides a reasonable model for comparing the
performance of different catalysts. Equation [1], along with
a first-order deactivation model that has a deactivation con-
stant of kd [=] h−1, can be substituted into the isothermal,
plug flow fixed-bed reactor design equation to obtain
x
ꢀ
dx
ln
(1−x)(ꢀH O − x) − Kw−g1s(ꢀCO + x)(ꢀH +x)
2
2
2
-1.5
-2
0
ꢁ
ꢄ
ꢂ
ꢃ2
W
FC0O
= ln
pC0 O kꢁ − kdt.
[2]
-2.5
-3
Here, x is the conversion of carbon monoxide, Kwgs is the
water–gas shift equilibrium constant, W is the mass of cata-
lyst, FC0O is the inlet molar flow rate of CO, pC0 O is the inlet
partial pressure of CO, t is the time-on-stream, and each
ꢀi is defined as p0/pC0 O. For brevity, let us denote the left-
hand-side of Eq. i[2] by α in the remainder of this docu-
ment. For our standard operating conditions, the absolute
pressure was 3 atm, ꢀCO = 0.3, ꢀH = 1.5, ꢀH O = 3, and
-3.5
0
10
20
30
40
50
60
70
80
t (hrs)
2
2
2
FIG. 1. A plot of α versus time-on-stream is shown for a 1.5% Pt/ceria
catalyst. Correlation results reveal a deactivation coefficient of 0.009 h−1
andaninitialactivityof0.25(mol/(gcat hatm))withacorrelationcoefficient
of 0.95.
ꢀ
N2
= 1.5. The catalyst space velocity was 2.5 (g catalyst
h/mol of CO) in order to yield kinetic data for outlet conver-
sions far from equilibrium. Values for α, kꢁ, and kd were cal-
Zalc
