5022 J . Phys. Chem.

1990, 94, 5022-5028

measured complexation equilibrium constants by a variety of Experimental Section

techniques and a certain consistency of results emerges. Certain
small molecules (such as ethanol and acetonitrile have significant
constants while others like dimethyl sulfoxide have relatively small
constant^.^ Cyclohexanol and phenol were noteworthy for their
particularly large constants. Coinclusion phenomena have been
noted not only with cyclodextrins but also with
Noteworthy in several studies of cyclodextrin complexes is the
sensitivity of the structure of the complex as a function of the
structure of the cyclodextrin and the included molecules.6,1’,25
(20) Veno, A,; Takahashi, K.; Osa, T J . Chem. Soc., Chem Commun.
1980, 1921-1922.
(21) Lindner, K.: Saenger, W. Angew. Chem.. Int. E d . EngI. 1978. 17.
694-695.
(22) Russell. J . C.; Wild, U . P.; Whitten. D.G. J . Phys. Chem. 1986, 90,
1 3 I 9- I 323.
(23) Thomas, J. K . Chem. Reo. 1980, 8. 283-299.
(24) Jones 11, G.; Jackson, W. R.; Choi, C.; Bergmark. W. R J . Phys.
Chem. 1985, 89. 294-300.
The coumarins were commercially available laser-grade ma-
terials obtained from Eastman Kodak Co. or Exciton Chemical
Co. and used in most cases as received following TLC analysis
for impurities. CI F was synthesized and purified as described
p r e v i ~ u s l y . The
~ ~ cyclodextrins were obtained from Aldrich
Chemical Co. and used as received. Emission spectra were re-
corded on a Perkin-Elmer MPF-44A fluorescence spectropho-
tometer equipped with a differential spectrum correction unit or
a Perkin-Elmer MPF-2A instrument. Spectra were recorded at
room temperature and integrated intensities were obtained by
cut-and-weigh.
Acknowledgment. Support for this work by The Office of Naval
Research, Research Corporation, and the Charles A. Dana
Foundation is gratefully acknowledged.
(25) Turro, N J.; Okubo, T.; Chung, C. J Am. Chem Sot. 1982. 104,
3954-3957

Effect .of Structure in Selective Oxide Catalysis: Oxidation Reactions of Ethanol and
Ethane on Vanadium Oxide

S. Ted Oyama*
Department of Chemical Engineering, Clarkson University, Potsdam, New York I3676

and Cabor A . Somorjai

Department of Chemistry, University of California, Berkeley, Berkeley, California 94720
(Received: August 31, 1989; In Final Form: November 7 , 1989)

The catalytic activity of well-characterized samples of vanadium oxide supported on silica were studied for the partial oxidation
of ethanol and ethane. Surface vanadium atoms were counted by a new oxygen chemisorption technique which allowed
measurements of sample dispersions and turnover rates. Ethanol oxidation was found to be a structure-insensitive reaction
that produced mostly acetaldehyde. Ethane oxidation was different; while its overall turnover rate changed little with vanadium
oxide dispersion, its selectivity did vary with catalyst structure. The products of this reaction were mostly ethylene and
acetaldehyde, but at low temperatures on the highest dispersion catalysts there was excess carbon dioxide formation.

Introduction employ the results of a recently developed method for counting

For catalytic reactions on metals the structure of the surface
can exert a profound influence on catalytic activity and selectivity.
For catalytic reactions on oxides the effects of structure are known
to a much lesser extent but have been receiving increasing at-
tention. Recent studies have investigated the structure sensitivity
of stoichiometric reactions on available oxide single crystals like
Zn0,’-3 Ti02,4-5 or oxide thin films like Mg063’ or V 2 0 5 . s Other
studies have relied on preparation of samples with different ratios
of exposed crystal facesg-13 for investigations of the effect of
structure. The subject area has been reviewed.14-”
A common approach for investigating the effect of structure
in metal catalysis is to study the response of reactivity to dispersion,
the ratio of surface to total atoms in a small metal particle.’* In
the critical region where particle diameters grow from atomic sizes
to I O nm. the surface structure evolves from isolated atoms to large
multiatomic and multilayered ensembles as found in the exposed
planes of macroscopic single crystal^.'^,^^ In this region the
dispersion decreases from 100% to very small values, and the effect
of structure on reactivity can be assessed.
Ilnfortunately, the effect of dispersion has rarely been used in
the study of oxide catalysts because of the difficulty in measuring
the number of surface atoms in oxides. In this investigation w e
____
*To whoin correspondence should be addressed.
surface sites on supported and unsupported vanadium oxide. The
effect of dispersion in the critical range is investigated for two
reactions: ethanol and ethane oxidation. We find that the oxi-
( I ) Berlowitz, P.; Kung, H. H. J . A m . Chem. SOC.1986, 108, 3532.
(2) Vohs, J. M.; Barteau, M. A. J . Phys. Chem. 1987, 91, 4766.
(3) Vohs, J. M.; Barteau, M. A. Surf.Sci. 1988, 197, 109.
(4) Onishi, H.; Aruga, T.; Egawa, C.; Iwasawa, Y . Surf. Sci. 1988, 193,
33.
( 5 ) Kim, K. S.; Barteau, M. A. Surf. Sci., in press.
(6) Martinez, R.; Barteau, M. A. Langmuir 1985, 1, 684.
(7) Barteau, M. A.; Peng, X. D. Mater. Chem. Phys. 1988, 18, 425.
(8) Lewis, K. B.; Oyama, S. T.; Somorjai, G. A. Surf. Sci., in press.
(9) Volta, J. C.; Desquesnes, W.; Moraweck, B.; Coudurier, G. React.
Kinet. Catal. Lett. 1979, 12, 241.
( I O ) Volta, J. C.;Forissier, M.; Theobald, F.; Pham, T. P. Faraday Dis-
cuss. Chem. SOC.1981, 72, 225.
( 1 I ) Tatibouet, J . M.; Germain, J . E. C. R . Seances Acad. Sci. 1980, 290,
321.
(12) Tatibouet, J. M.; Germain, J. E. J . Catal. 1981, 72, 375.
(13) Oyama, S. T. Bull. Chem. Soc. Jpn. 1988, 61, 2585.
(14) Henrich, V. E. f r o g . SurJ. Sci. 1979, 9, 143.
(15) Volta, J . C.; Portefaix, J . L. Appl. Catal. 1985, 18, 1.
(16) Kung, H . H. Ind. Eng. Chem. Prod. Res. Deo. 1986, 25, 171.
(17) Barteau, M. A,; Vohs, J. M. In Successful Design of Catalysts; Inui.
T.. Ed.; Elsevier: Amsterdam, 1988; p 89.
( 1 8 ) Boudart, M. Adc. C a t d 1969, 20, 153.
(19) Van Hardeveld, R.; Hartog, F. Surf. Sci. 1969, 15, 189.
(20) Ptrez. 0.-L.; Romeu, D.: YacamBn, M. J . J. Catal. 1983, 79, 240.

0022-3654/90/2094-S022$02.50/0 1990 American Chemical Society

Reactions of Ethanol and Ethane on Vanadium Oxide
TABLE I: Oxygen Uptake on Silica-Supported Vanadium Oxide
before reaction“
0, uptake,c
sample SA,” m2 g-I wmol g-l
Si02 99 0.6
0.3% V205/S!O2 15.3
1.4% v20s/s102 92 67.0
3.5% V205/Si02 95 133
5.6% V205/Si0, 82 142
7.7% V205/Si02 78 191
9.8% V20,/Si02 82 258
100% v20s 2.8 7.3
” Reference 21. bSurface area. COxygenmolecule uptake. dDispersion = fraction of V atoms a t the surface, assuming Oadr/Vaurl
dation of ethanol is structure insensitive. Ethane oxidation,
however, exhibits a large change in selectivity (ethylene vs carbon
dioxide production) with dispersion while the overall reaction rate
is independent of surface structure.
Experimental Section
The supported V2Os catalysts were prepared by impregnating
Si02( C a b - O M , 90 m2 8-l) to incipient wetness with vanadium
oxalate solutions formed from N H 4 V 0 3 (Aldrich, 99.99%) and
oxalic acid (Adrich, 99.99%). Vanadium oxide concentrations
are reported as weight percent throughout the paper. The solids
were dried for 2 h at 573 K and calcined for 12 h at 773 K. They
were used as powders of mesh size < 120 (particle size, L < 0.012
cm), except in diffusion limitation tests where they were pressed
to 10 MPa and sieved to 6C-120 mesh ( L = 0.025 cm) and 30-60
mesh ( L = 0.040 cm). Pure unsupported V,Os (Aldrich, Gold
Label, 99.999%) in the form of a fine powder ( L < 0.012 cm)
was used as received.
Surface areas (BET) and oxygen uptakes were measured with
a flow apparatus equipped with a thermal conductivity (TC)
detector. BET areas were obtained by flashing samples equili-
brated in a 30% N 2 / H e stream. Oxygen uptakes were measured
at 641 K on samples reduced in H 2 at the same temperature by
adsorbing calibrated pulses of 02.
Catalytic activity was obtained in a 16-mm-o.d./14-mm-i.d.
quartz packed-bed flow reactor. A sealed quartz capillary tube
held axially in the center allowed a thermocouple to probe tem-
perature gradients along the reactor length. Axial temperature
differences of a few kelvin were detected only at the highest
conversions. I n all experiments, the total flow rate was 73-74
Kmol s-I and the total pressure was 101 kPa. (Flow rates in Mmol
s-l may be converted to cm3 (NTP) min-’ by multiplying by 1.5.)
In the ethanol oxidation studies the partial pressures of reactants
were PEtOH = 1.6 kPa, Po, = 28 kPa, PHZO = 10 kPa, and PHe
= 61 kPa. I n the ethane oxidation studies these were PCHtCH, --
13 kPa, Po, = 28 kPa, PHlO= I O kPa, and PHe= 50 kPa. The
ethane contained a 0.5% ethylene impurity whose presence was
taken into consideration in the rate calculations. The ethane was
purified from possible sulfur impurities by passage through a
reduced N i / S i 0 2 bed. All other gases were research grade
(Matheson, >99.995%) and were employed as received with flow
rates adjusted by rotometers. The liquids, including the unde-
natured ethanol, were also research grade (Aldrich, 99.99%). They
were loaded into a syringe pump and metered into an on-line
evaporator. All inlet and outlet lines were made of stainless steel
and were heat-traced.
The ethanol partial pressure was chosen to be below the lower
flammability limit (5 kPa). The ethane partial pressure was chosen
to be above the upper flammability limit (10 kPa) because the
lower limit ( 3 kPa) made the analysis for products at low con-
versions difficult. Although these choices caused the ethanol and
ethane feeds to be oxygen-rich and oxygen-lean with respect to
combustion stoichiometry, the conversions were such that there
was always excess oxygen. For the ethanol and ethane studies
the maximum O2conversions were 17% and 33% at the maximum
reactant conversions of 100% and 20%, respectively.
The H 2 0 partial pressure was kept high in order to reduce
variations due to conversion level. H 2 0 is a product of reaction
The Journal of Physical Chemistry, Vol. 94, No. 12, 1990 5023
after ethane oxidation
0, uptake:
dispd SA,b m2 g-I wmol g-l dispd
58
0.93 61 9.2 0.56
0.87 50 26.2 0.34
0.69 56 57.4 0.30
0.46 59 63.4 0.21
0.45 63 126 0.29
0.48 59 180 0.33
0.001 4.9 13.0 0.002
= I.
and an inhibitor in oxidation reactions.
In all cases amounts of catalyst corresponding to 7.0 mg of V205
were loaded in the reactor. Si02diluent was also added so that
the total sample weight was 500 mg, corresponding to a constant
bed volume of 3.1 cm3. The only exceptions were for the 0.3%
V 2 0 5 sample for which 7.0 mg of V2Os resulted in 2.33 g of total
sample and for the pure unsupported V 2 0 5in which case 5.4 g
was used. For analysis a gas chromatograph with a FID (Poropak
QS column) and a TC detector (Carbosphere column) was used.
Rates are reported as turnover rates based on surface vanadium
atoms counted by oxygen chemisorption:
0, =
QYX
-
vw2s
The validity of the chemisorption technique has been established
elsewhere.2’ In the equation above ut is the total ethanol or ethane
turnover rate, Q is the total volumetric flow rate, y is the mole
fraction of ethanol or ethane, Vis the molar volume at the con-
dition of flow measurement, w is the weight of catalyst, and 2 s
is the oxygen uptake value. The factor of 2 accounts for the
stoichiometry of one chemisorbed oxygen molecule per two surface
vanadium atoms. The conversion, x, is defined as the fraction
of ethanol or ethane that reacts. The other turnover rates are
defined as
ci = u,s,
where S,,the selectivity of product i, represents the fraction of
ethanol or ethane that gets converted to i and takes into consid-
eration the number of carbon atoms in the product. (Thus, CO
or C 0 2 production is divided by 2.) The turnover rates are given
in units of ks-I equivalent to IO-) s-l.
Particular care was used to ensure that the catalysts were stable
at the reaction conditions employed. Before every run the catalysts
were pretreated in the 02-H20-He reactant flow mixture for 1
h at the maximum temperature of reaction. The temperature was
then lowered and the hydrocarbon added to the feed. Reactivity
measurements were then taken at increasng temperatures while
verifying that steady state was reached. Each point typically
required 1-2 h, and runs of 16-24 h were carried out. Importantly,
after the highest temperature was attained, a lower temperature
measurement was always repeated to verify catalyst stability and
lack of deactivation during the rate data acquisition. These points
are indicated by arrows in Figures 1,3, and 4. Carbon and oxygen
mass balances closed to 100 f 5%.
Results
Oxygen Chemisorption. The chemisorbed oxygen uptake values
of the fresh catalysts are reported in Table I. It was found that
the catalysts employed in the ethanol oxidation reaction had
essentially the same uptakes and surface areas after reaction.
However, the catalysts employed in the higher temperature ethane
oxidation lost surface area. This loss occurred in the initial
hydrothermal activation period and not during reaction, and thus
the rate measurements remain accurate. However, for this reason,
(21) Oyama, S. T.; Went, G.T.; Lewis, K. B.; Bell, A. T.; Somorjai, G.
A . J . Phys. Chem. 1989. 93, 6786.
5024 The Journal of Physical Chemistry, Vol. 94, No. 12, I990 Oyama and Somorjai

TABLE 11: Representative Conversion and Selectivity Values in Ethanol Oxidation

sample conversion, % %
SCH,CHO- S C H ~%C H ~ SA~OH,
% SCO$ % ut, ks-'
T = 490 K
Si02 0.4 100
I .4% V20s/Si02 2.0 84 4 6 6 0.32
3.5% V20s/Si02 3.5 77 2 11 IO 0.97
5.6% V2OS/S1O2 5.5 77 1 II 11 2.2
7.7% V205/Si02 3.9 65 1 17 17 1.3
9.8% V205/Si02 3.8 74 1 14 11 1.6
T = 530 K
SiO, 1.5 98 2
I .4% v ~ o ~ / s~o~ 5.0 77 13 5 5 0.97
3.5% V205/Si02 12 66 8 12 14 27
5.6% V205/Sj02 15 66 6 14 14 5.8
1.7% V205/S102 13 59 4 17 20 4.1
9.8% V205/Si02 12 66 4 15 15 4.2
T = 570 K
Si02 6.8 87 3 IO
1.4% V20s/Sj02 15 62 30 4 4 3.0
3.5% V20s/S102 30 56 26 9 9 7.3
5.6% V20s/Si02 35 61 I5 12 12 14
7.7% V205/S102 35 59 11 15 15 13
9.8% V205/S102 30 62 11 14 13 12
p--€--
__ -_
b ) I'

7 10 t 0 9 8%
I" x 77%
.
v)
Y 5 6 %

I
A

0
c

2 1 1
L
m
5 t
0
E i
2 01 I

Too
Temperature / K
Figure 1. Conversion and selectivity in ethanol oxidation: X, conversion;
0, acetaldehyde; 0,ethylene; 0, acetic acid; V, carbon monoxide; A,
carbon dioxide. (a) SO,;(b) 1.4% V205/Si02;(c) 3.5% V20s/Si02;(d)
5.6% V205/Si0,;(e) 7.7% V20s/Si0,; (f) 9.8% V20s/Si02.
for the ethane oxidation reaction the turnover rates were based
on oxygen chemisorption at 641 K on the catalysts after reaction
(Table I).
Table I also reports vanadium oxide dispersions, defined as the
fraction of vanadium atoms on the surface as measured by oxygen
chemisorption. These range from 0.93 to 0.45 for the catalysts
used in ethanol oxidation and from 0.56 to 0.21 for those used
in ethane oxidation.
Ethanol Oxidation. Ethanol oxidation was studied between
400 and 700 K on the silica-supported samples. The results are
summarized in Figure 1 and Table 11. All samples showed similar
behavior regardless of dispersion; at low temperatures the main
product was acetaldehyde while at high temperaures it was
ethylene. Acetic acid was produced at intermediate temperatures,
and CO and COz were produced at the most elevated tempera-
tures. At these highest temperatures traces of methane were
occassionally detected, probably from the homogeneous decom-
position of the acetaldehyde product. No formaldehyde, ethyl
i
t
1
h
h
, Ethylene Acetic Acid
0.01 I )L2 1 -,
1.6 1.8 2 . 0 2 . 2 2 . 4 1.6 1.8 2.0 2.2 2 . 4
103 T - ~ I K - '
Figure 2. Arrhenius plots in ethanol oxidation: V, 1.4% V20s/Si02;0,
3.5% V205/Si02;A, 5.6% V20s/Si02;X , 7.7% V20s/Si02;0,9.8%
V20s/Si02. (a) total turnover rate; (b) acetaldehyde rate; (c) ethylene
rate; (d) acetic acid rate.
ether, or higher coupling products were ever observed, although
these were detectable. Silica had a considerably smaller activity
and differed from vanadium oxide in producing mostly acet-
aldehyde and no ethylene at low conversions. All conclusions in
this study will be drawn from data at low conversions where the
silica remains chemically inert, as shown in Figure 1 and Table
11, and where product decomposition does not occur.
Arrhenius plot for the various products summarize the effect
of dispersion on the turnover rate as shown in Figure 2 and Table
111. The activation energies do not vary drastically from sample
to sample, and there are no trends. The average values for ac-
etaldehyde, ethylene, acetic acid, and COXare 5 1, 120, 64, and
66 kJ mol-'. The Arrhenius plots are linear except for the acetic
acid product (Figure 2d) because it is probably an intermediate
Reactions of Ethanol and Ethane on Vanadium Oxide The Journal of Physical Chemistry, Vol. 94, No. 12, 1990 5025

60 X Conrersion
0 CH3CHO 1 0 CHJCHO
0 CH2CH2 0 CHzCH2

-.
> .
40

I
1 V
A
CO
COz
I
s
.
401 V CO
A Cop

5 0
20- , ,, /&
fn 700 800 900 % 700 800 900
Lo

Temperature / K Temperature / K

Figure 3. Conversion and selectivity in ethane oxidation:

conversion; X,
Figure 4. Conversion and selectivity in ethane oxidation:
X, conversion;
0,acetaldehyde; 0.ethylene; V, carbon monoxide; A, carbon dioxide. 0,acetaldehyde; 0,ethylene; V, carbon monoxide; A, carbon dioxide.
(a) S O 2 ; (b) 0.3% V20s/Si02;(c) 1.4% v205/siO& (d) 3 3 % V2OS/ (a) 5.6% V20s/Si02; (b) 7.7% V20s/Si02; (c) 9.8% V20s/Si02; (d)
Si02. 100% V20s.

TABLE 111: Summary of Activation Energies (in kJ mol-’) for Ethane Oxidation. As expected, the oxidation of ethane re-
Ethanol Oxidation quired much higher temperatures than the oxidation of ethanol
(Figures 3 and 4), conversions becoming appreciable only above
sample ECH~CHO ECH~CH EA~OH
~ Eco, Et 800 K. The products of the reaction were acetaldehyde, ethylene,
1.4% V 2 0 5 / S i 0 2 53 1 I8 58 67 CO, and COz. Traces of methane at high temperatures but no
3.5% V 2 0 5 / S i 0 2 48 121 63 62 59
5.6% V20s/Si02 45 1IO 67 72 51
formaldehyde, acetic acid, ethyl ether, or higher coupling products
7.7% V20s/Si02 63 123 71 68 65 were obtained.
9.8% V205/Si02 47 1 I4 60 66 53 Unlike ethanol oxidation, where small amounts of CO and COz
were observed only at the highest temperatures (>630 K), for
which is consumed at high temperature. The data tend to cluster, ethane oxidation large amounts of CO and COz were produced
although there is some spread for the acetic acid. There is a small at all temperatures (Table IV). For the supported samples
but consistent trend for the low concentration samples (1.4% and combined CO and C 0 2 selectivity tended to increase with tem-
3.5%) to have lower rates. perature to just over 50% at 870 K (Table IV). For the unsup-

TABLE IV: Conversion and Selectivitv Values in Ethane Oxidation

sample conversion, % SCH,CHO? % S C H ~%C H ~ ~ sco, % SCO*, 7% ut, ks-I
T = 800 K
Si02 0
0.3% V205/Si02 1.o 7 30 63 2.4
1.4% V205/Si02 1.5 3 19 5 73 5.1
3.5% V205/S102 0.6 12 63 IO 15 2.2
5.6% V205/Si02 0.5 7 73 9 11 3.3
7.7% V205/Si02 0.8 9 65 7 19 3.8
9.8% V 2 0 s / S i 0 2 0.5 8 59 5 28 2.7
100% V2O5 8.0 0.2 26 39 35 6.2
T = 825 K
Si02 0.1 100
0.3% V205/Si02 2.1 9 36 2 52 9.3
I .4% V 2 0 5 / S i 0 2 2.2 4 28 8 60 16
3.5% V205/Si02 1.8 11 57 10 22 15
5.6% V205/Si02 1.9 6 66 11 17 22
7.7% V205/Sj02 2.5 9 59 5 21 21
9.8% V205/S102 2.8 8 58 8 29 21
100% V 2 0 s 10 0.2 24 43 33 8.6
T = 870 K
Si02 2.9 81 4 15
0.3% V205/Si02 14 8 42 15 35 21
1.4% V205/Si02 12 5 39 13 43 30
3.5% V 2 0 5 / S i 0 2 13 5 38 22 35 63
5.6% V205/S102 12 3 46 22 29 69
7.7% V205/Si0, 14 5 46 24 30 57
9.8% V , 0 5 / S i 0 2 16 5 44 17 34 49
100% v20s 18 0.2 21 49 30 15
5026 The Journal of Physical Chemistry, Vol. 94, No. 12, 1990 Oyama and Somorjai

j a) ' b)
10 1 i

t I

. L
a!
> 1 c)
0
E
t-
10

L
1.1 1.2
L
1.3
1 -1

1.1
IO3 T - ' i K - '
1.2 1.3
r
i tL i \ , &
,
1.1
,
1.2 1.3
,

lo3 T-'/ K - '

L

1.1 1.2 1.3

Figure 5. Arrhenius plots in ethane oxidation: 0, acetaldehyde; 0, Figure 6. Arrhenius plots in ethane oxidation: 0, acetaldehyde; 0,
ethylene; V, carbon monoxide; A,carbon dioxide. (a) 0.3% V205/Si02; ethylene; V,carbon monoxide; A,carbon dioxide. (a) 5.6% V205/Si02;
(b) 1.4% V205/Si02;(c) 3.5% V205/Si02. (b) 7.7% V205/Si02;(c) 9.8% V,0s/Si02; (d) 100% V 2 0 5 .
TABLE V Summary of Activation Energies (in kJ mol-') for Ethane
Oxidation
- Eco2 - E, t .' T = 870 K
sample ECHf2H0 ECH2CH2 ECO LT HT LT HT
0.3%V20s/Si02 207 223 307 78 162 122 337
1.4% V20s/Si02 229 248 238 96 167 87 217
3.5% V205/Si02 184 224 354 323 294
5.6% V205/Si02 181 220 346 344 253
7.7% V20s/Si02 192 216 304 219 224
9.8% V20s/Si02 176 210 311 252 210
100% V2OS 70 55 93 56 72

ported 100% V 2 0 s sample this figure rose to 80%. In fact, the

100% V2Os sample had much higher selectivity to the deep ox- Ethylene
idation products throughout the entire temperature range. It also 0.1LdL- LA ~

displayed much higher conversions and less temperature depen-

dence (Figure 4d).
10
1
0.3 1.4 3 5 5.6 7.7 9 8 100
c)
I
I
t T = 870 K
The most dramatic difference between ethanol and ethane
oxidation was the sudden shift in selectivity in ethane oxidation
in going from high to low dispersion samples (Figure 3). For the
0.3% and 1.4% samples at low conversion, the main product was
C 0 2 ,and this decreased in favor of ethylene as the temperature
was raised (Figure 3b,c). For the 3.5% and higher concentration
samples this behavior was entirely reversed, with ethylene being
the main product at low conversion and carbon dioxide increasing
in its place as the temperature was raised (Figures 3d and 4a-c).
I n all cases acetaldehyde was a minority product (5-10%).
Arrhenius plots of these data reflect the same peculiar difference
between low and high vanadium concentration samples (Figures
5 and 6 ) . For the low-concentration samples the C 0 2 production
rates are high and the Arrhenius plots have curvature. For the
high concentration samples the C 0 2production rates are low and
u Acetaldehyde
0'01 0.3 1.4 3.5 5.6 7.7 9.8 100
Vanadium Loading / %
Figure 7. Effect of vanadium concentration on ethane oxidation: (a) X
the Arrhenius plots are linear. Activation energies obtained from or +, total rate; (b) 0,ethylene rate; (c) 0,acetaldehyde rate.
these plots by linear regression are summarized in Table V. There
is a clear trend toward decreasing activation energies with in- The effects of V 2 0 5 Concentration on the turnover rates are
creasing vanadium concentration. There is a very large drop in displayed at three temperatures in Figure 7. The overall rate
activation energies for the 100% V 2 0 5 sample. For the low- is seen to be relatively constant at all temperatures. However,
concentration samples values of activation energy are given at low the rates for ethylene and acetaldehyde formation increase with
temperature (LT) and high temperature (HT) because of the concentration, although the increases are modest and tend to be
curvature. leveled at higher temperatures.
Reactions of Ethanol and Ethane on Vanadium Oxide
100
100% v*o5
t study of ethanol oxidation was motivated in part by two studies
40i
v C a r b o n monoxide
A C a r b o n dioxide
0 1000 2000 3000 4000 5000
Time I s
Figure 8. Conversion and selectivity in ethane oxidation in the absence
of oxygen: catalyst, 100%V205,4.58 g: X, conversion: 0 , ethylene; V,
carbon monoxide: A, carbon dioxide.
The behavior of the unsupported 100% VzOs was investigated
further by cutting off the gas-phase oxygen supply and following
the products of ethane oxidation in time (Figure 8). At zero time
the helium flow rate was increased so as to maintain the total flow
rate constant at the usual 7 3 pmol s-I. The selectivity to the
various products stabilized in -2000 s. The conversion decreased
slightly during this time and then gradually rose. The temperature
remained constant at 7 5 0 K throughout the experiment.
Transport Effects. Lack of mass- and heat-transfer effects were
checked by varying the particle size of the catalysts in both the
ethane and ethanol oxidations (Figure 9). The particle size refers
to the macroscopic catalyst powder agglomefates obtained by
sieving the samples, not the supported crystallites of V20s. Plots
of conversion and selectivity for catalysts employed in this study
( L < 0.012 cm) are shown on the left panels (Figure 9a,c) and
reproduced on the right panels for the pressed and sieved samples
(Figure 9b,d) of larger particle size ( L = 0.025 and 0.040 cm).
The points for the larger sized particles are seen to fall very close
to the plots for the fine particle sizes.
Discussion
Oxygen Chemisorption. The oxygen chemisorption technique
employed here has been described in a recent publication in this
journal.2' It was clearly shown by using X-ray diffraction, laser
Raman spectroscopy, and isotopic labeling that the method counts
surface atoms and not bulk atoms. The technique relies on finding
an optimal temperature where surface but no bulk reduction
occurs. The finding that ethanol oxidation is a structure-insensitive
reaction is an internal check that also confirms the validity of the
method. This will be discussed in the next section.
From the oxygen uptake values, vanadium dispersions can be
calculated (Table I), and these are in excellent agreement with
the Raman spectra of the same samples reported earlier.21 Thus,
where oxygen chemisorption shows high dispersion, the Raman
shows monatomic vanadium species on the surface; where low
dispersions are measured, small crystallites of V2Os of diameter
<4 nm are observed. At high loadings the dispersions tend to reach
constant values, although somewhat scattered, especially for the
catalysts examined after the high-temperature ethane oxidation.
This is due in part to the limited concentration range employed
but also to the sintering of the finer vanadium oxide crystallites.
Nevertheless, the ethanol and ethane oxidation studies are con-
ducted at two ranges of dispersion, 0.93-0.45and 0.56-0.21, where
surface structure is changing appreciably and, thus, the samples
are ideally suited to examine the effect of structure on catalytic
activity.
I t is possible that the loss in surface area of the catalysts used
in the ethane oxidation reaction may have been due to a melting
of the V 2 0 5crystallites or the formation of very mobile vanadium
oxide clusters. The leveling of rate at the highest temperatures
(discussed below) may also be due to such phenomena. However,
the lack of discontinuities in the conversion and selectivity data
The Journal of Physical Chemistry, Vol. 94, No. 12, 1990 5027
and the absence of hysteresis indicate that no phase transitions
occur.
Ethanol Oxidation, A Structure-Insensitive Reaction. Our
of methanol oxidation on vanadium oxide based catalysts that
indicated that the reaction is structure i n s e n s i t i ~ e . ~ ~Our
,~~
confirmation of structure insensitivity would validate the oxygen
chemisorption method. The first study by Baiker and Monti
examined the methanol oxidation activity of two unsupported
vanadium oxide samples of greatly differing external morphology
and concluded from lack of differences that the reaction was
structure insensitive.z2 A related study by Sorensen and Weber
found the same rate per vanadium atom for different hetero-
polytungstates substituted with various numbers of vanadium
atoms and reached the same conclusion.z3 Our study finds that
ethanol oxidation is structure insensitive, and therefore it appears
that the alcohol oxidation reaction is indeed structure insensitive
over V 2 0 5 .
There is one study of methanol oxidation that suggests structure
sensitivity for this reaction.24 In this work the selectivity of a
number of well-crystallized, low surface area samples of V 2 0 5
varied in proportion to the ratios of different surface planes, and
the authors concluded that formaldehyde was preferentially formed
on the [OOl] planes. However, the study using low surface area
materials did not report the rates per unit surface area, only
selectivities, and these can differ with conversion. The relative
amounts of surface planes are difficult to obtain by microscopy
because of the difficulty of measuring crystallite sizes in three
dimensions.
Examination of the data (Figure 2, Table 11) reveals that the
low-concentrationsamples of highest dispersion have turnover rates
that are consistently smaller and have selectivities that favor
acetaldehyde over ethylene. However, these differences are minute
and are accounted for by factors of 2 or 3 in rates. Such variations
in rate are small and are not uncharacteristic for structure-in-
sensitive reactions.l* The changes in selectivity (Figure 1 ) vary
smoothly and only to a small degree and display only a subtle effect
of surface structure. Similarly, the activation energies for the
overall reaction and the various products show no distinct trends
(Table 111). The exact border line in the classification of struc-
ture-sensitive and -insensitive reactions is not clearly marked,
although rate differences exceeding lo3have been associated with
the former.2s Ethanol oxidation on supported vanadium oxide
should be considered a structure-insensitive reaction in the classical
sense.
We form our conclusions from the lower conversion region of
the data (Table 11). It should be noted in that table that the
highest dispersion catalyst (1.4%) has lower conversions by factors
of about 2. Its higher selectivity to the selective oxidation products,
acetaldehyde and ethylene, can be understood on that basis. At
lower conversions the effect of secondary reactions is reduced.
Thus, the data consistently point to the uniform behavior of all
the catalysts.
Ethane Oxidation, a Structure-Directed Reaction. The oxi-
dation of ethane on the supported samples contrasts profoundly
with the oxidation of ethanol. Whereas in the oxidation of ethanol
selectivity differences were small and varied gradually with
concentration, in ethane oxidation a considerable jump occurred
in selectivity between loadings of 1.4% and 3.5% (Figure 3c,d).
With a concentration of 1.4% and below, C 0 2 was the primary
product at low temperatures, but for the 3.5% and higher con-
centration samples, ethylene was the predominant product. Such
marked effects of concentration are characteristic of structure-
sensitive reactions. Activation energy values in ethane oxidation
(22) Baiker, A.; Monti, D. J . Coral. 1985, 91, 361.
(23) Sorensen, P.; Weber, R. S. Presented at the 2nd China-US.-Japan
Seminar on Heterogeneous Catalysis, Berkeley, CA, July 1985.
(24) T a t i W t , J.-M.; Germain, J.-E. C. R. Seances Acad. Sci., Ser. 2 1983,
296, 613.
(25) Boudart, M . In Proceedings of the 6th Inrernational Congress on
Catalysis, London, 1976; Bond, G. C.; Wells, P. B., Tompkins, F. C., Eds.;
The Chemical Society: London, 1977; p 1 .
5028 The Journal of Physical Chemistry, Vol. 94, No. 12, 1990
100
~
C
n time amount to 3900 pmol, which is 32 times the amount of
\
0.12 c m 4 ;
20
0-
700 aoo 900 700 800 900
Temperature / K
Figure 9. Particle size effect on conversion and selectivity: X, conversion;
0, acetaldehyde; 0,ethylene; V, carbon monoxide; A, carbon dioxide.
(a) ethanol oxidation, 9.8% V205/Si02,L = 0.012 cm; (b) ethanol ox-
idation, 9.8% V205/Si02,L = 0.025 cm; (c) ethane oxidation, 1.4%
V20s/Si02.L = 0.012 cm; (d) ethane oxidation, 1.4% V,05/Si02.L =
0.040 cm.
changed systematically with concentration (Table V), indicating
that catalyst structure affected the various reaction pathways
available. This again contrasts with the findings in the ethanol
oxidation reaction where activation energies did not vary sys-
tematically with concentration. Thus, activation energy changes
also point to ethane oxidation being structure sensitive.
We base our conclusions on data at 800 and 825 K where
conversions are below 3% (Table IV). At these conversion levels
the influence of secondary reactions is minimized and so is the
concern, noted for the ethanol studies, of comparing catalysts at
different conversions. At higher temperatures and conversions
(870 K, 14%) the selectivity and rate differences between catalysts
of different V 2 0 5concentration are largely eliminated (Table IV,
Figure 7).
While there are modest changes in reaction selectivity with
disperson at low conversions, the overall rate of the reaction is
not altered by changes in dispersion. Figure 7a shows that for
the supported samples the magnitude of the overall rate is relatively
invariant with concentration, which is usually considered as in-
dicating structure insensitivity. The unsupported 100% V 2 0 5
departs markedly from the supported samples and will be discussed
separately. The rates of production of the various products do
vary with Concentration. Clearly, C O and C 0 2 production rates
are balancing the ethylene and acetaldehyde rates so that the total
rate remains constant. These variations are small and disappear
at higher temperatures (Figure 7b,c). Overall, an interesting
reaction behavior is found for ethane oxidation: although variation
in rates are small, changes in selectivity and activation energies
indicate that catalyst structure does have an effect on reactivity.
All conclusions drawn here are free from effects due to the support,
secondary reactions, or transport limitations.
This type of chemical behavior has been observed for another
system, propylene oxidation on MOO,. Again in that system, rates
did not change appreciably with surface structure while selectivities
did. Although some have called the reaction structure sensitive,
others have called for caution.13 Structure appears to influence
the direction the intermediates take among various reactive
pathways available, but rates are not affected greatly.
I t was noted earlier that 100% V 2 0 , behaves differently from
the supported samples. Its activation energy values are signifi-
cantly smaller than those on the supported samples (Table V),
and hence it has a substantial conversion over a much larger
temperature range (Figure 4d). For this reason at most tem-
peratures its rate is also substantially different from that of the
Oyama and Somorjai
supported samples (Figure 7 ) . It is likely that a different
mechanism is operative for the 100% V 2 0 ssample. Bulk oxides
are known to possess a vast reservoir of oxygen which becomes
highly mobile when the oxide is in a partially reduced state.
Normally, this mobile lattice oxygen is capable of selective oxi-
dation, but at the very high temperatures used in ethane oxidation
they cause overoxidation. Indeed, Figure 8 shows that when the
oxygen supply is cut off, the selectivity to C O increases and that
to ethylene decreases. Although COz formation decreases also,
the overall selectivity shifts toward less selective product formation.
The oxygen atoms used in forming the products integrated over
surface oxygen but only 0.08 times the available reducible bulk
oxygen. It is interesting to note that no acetaldehyde is formed
in the absence of gas-phase oxygen.
Table V shows that there is a consistent trend for the activation
energies of the supported samples to decrease with concentration,
approaching the values for the 100% V 2 0 5sample. This is due
to the growth of the crystallites of the supported samples.
The Arrhenius plots for CO, rates for the catalysts prepared
with low vanadium concentrations showed curvature (Figure 5a,b),
and activation energy values at low temperature (LT) and high
temperature (HT) are reported (Table V). The curvature reflects
the production of excess C 0 2 at low temperatures and may be
due to the involvement in the reaction of some particularly active
form of adsorbed oxygen. Again the result points to the existence
of structural effects in the reaction.
Lack of Mass- and Heat-Transfer Limitations. The excellent
agreement between the conversion and selectivity for the larger
sized particles and the fine-sized particles (Figure 9) indicates
that there are no external transport effects. A detailed discussion
of the internal (intraparticle) transfer effects is given in a pre-
liminary report of part of these results.26 It is shown there that
the Anderson criterion2' indicates that there are no heat-transfer
limitations, and a Carberry analysis for combined heat and mass
transfer2* yields an effectiveness factor of unity. Thus, at the
conditions employed in this study there are no internal or external
mass- or heat-transfer effects for both ethanol and ethane oxi-
dation.
Conclusions
1. This study confirms the validity of the oxygen chemisorption
method developed earlier2' for counting the active surface vana-
dium atoms on supported vanadium oxide.
2 . In ethanol oxidation a series of supported catalysts have
turnover rates and selectivity that do not vary significantly with
dispersion, indicating that the reaction is structure insensitive.
3. In ethane oxidation at low conversions overall turnover rates
do not vary, but selectivities change dramatically with vanadium
concentration indicating that surface structure affects reactivity.
4. For ethane oxidation the supported V,05 samples behave
substantially different from 100% V20s, in which bulk lattice
oxygen appears to participate in the reaction. However, as con-
centration increases, the supported sample behavior tends to ap-
proach that of the 100% V205.
Acknowledgment. The experimental work was carried out at
U.C. Berkeley with support from the Director, Office of Energy
Research, Office of Basic Energy Sciences, Materials Sciences
Division of the US. Department of Energy, under Contract
DE-AC03-76SF00098. The paper was written at Clarkson
University with support from the Director, Division for Chemical
and Thermal Systems of the National Science Foundation, under
Grant CTS-890998 1. The authors thank A. M. Carr for assistance
with the ethanol oxidation experiments.
Registry No. V 2 0 5 , 1314-62-1; EtOH, 64-17-5; CH3CH3,74-84-0;
0 2 , 7782-44-7.
(26) Oyama, S. T.;Lewis, K. B.; Carr, A. M.; Somorjai, G . A. In Pro-
ceedings of the 9th International Congress on Catalysis, Calgary, 1988
Phillips, M. J., Ternan, M . , Eds.; Chemical Institute of Canada: Ottawa,
1988; Vol. 111, p 1489.
(27) Anderson, J. B. Chem. Eng. Sei. 1963, 18, 147.
(28) Carberry, J. J. AIChE J . 1961, 7 , 350.