Transient FTIR studies of the reaction pathway for n-butane selective oxidiation over vanadyl pyrophosphate

Article abstract of DOI:10.1006/jcat.1999.2452

New information has been provided about the reaction pathway for n-butane partial oxidation to maleic anhydride over vanadyl pyrophosphate (VPO) catalysts using FTIR spectroscopy under transient conditions. Adsorption studies of n-butane, 1,3-butadiene, and related oxygenates were performed to gain information about reaction intermediates. n-Butane was found to adsorb on the VPO catalyst to form olefinic species at low temperatures. Unsaturated, non-cyclic carbonyl species were determined to be precursors to maleic anhydride.

Full text of DOI:10.1006/jcat.1999.2452

Journal of Catalysis 184, 87–104 (1999)
Article ID jcat.1999.2452, available online at http://www.idealibrary.com on
Transient FTIR Studies of the Reaction Pathway for n-Butane
Selective Oxidiation over Vanadyl Pyrophosphate
Zhi-Yang Xue and G. L. Schrader
Department of Chemical Engineering and Ames Laboratory-USDOE, Iowa State University, Ames, Iowa 50011
Received July 30, 1998; revised February 12, 1999; accepted February 12, 1999
ring generates 2(5H)-furanone; additional oxygen insertion
New information has been provided about the reaction pathway at the �-position of 2(5H)-furanone results in production
for n-butane partial oxidation to maleic anhydride over vanadyl py- of maleic anhydride. Noncyclic oxygenates, such as methyl
rophosphate (VPO) catalysts using FTIR spectroscopy under tran-
sient conditions. Adsorption studies of n-butane, 1,3-butadiene, and
related oxygenates were performed to gain information about reac-
tion intermediates. n-Butane was found to adsorb on the VPO cata-
lyst to form olefinic species at low temperatures. Unsaturated, non-
also been proposed which do not involve the formation of
cyclic carbonyl species were determined to be precursors to maleic
vinyl ketone and crotonaldehyde, can also be formed but
are believed to undergo complete oxidation (2–4).
This reaction pathway postulates the formation of cyclic
oxygenates; however, other reaction intermediates have
furan-related cyclic species. Zhang-Lin et al. (6, 7) stud-
anhydride.
�c 1999 Academic Press
ied the oxidation of n-butane, 1,3-butadiene, furan, and
maleic anhydride on various VPO phases. Maleic anhy-
dride was observed to be formed from n-butane, while only
a small amount of maleic anhydride was obtained from
of 1,3-butadiene and furan molecular intermediates (6).
Zhang-Lin et al. proposed an “alkoxide route” for maleic
Key Words: n-butane oxidation; VPO catalysts; transient FTIR;
adsorbed intermediates.
INTRODUCTION
The selective oxidation of n-butane to maleic anhydride anhydride formation, in which adsorbed noncyclic species
over vanadium phosphorus oxide catalysts (VPO) has been were the precursors to maleic anhydride (7). Zi o´ lkowski
studied for more than 20 years (1–4). Commercial tech- et al. (8) examined the energetics and geometries of fa-
nologies using fixed bed reactors, fluidized bed reactors, vorable pathways for n-butane and 1-butene oxidation on
and more recently a circulating fluidized bed reactor have the (100) face of (VO) P O . Analysis of the heats of reac-
2
2
7
been developed (2). Despite the commercial success of this tion for each possible reaction step on the surface suggested
processing route, many aspects of the reaction mechanism that the intermediates in n-butane oxidation were noncyclic
are not well understood.
species anchored on the surface through Cterminal–Osurface
Only very little byproducts other than carbon dioxide bonds.
are produced under commercial reaction conditions (3). In
In situ spectroscopic investigations of surface species
laboratory studies, however, butene, 1,3-butadiene, furan, have provided important insights into the reaction inter-
crotonaldehyde, and methyl vinyl ketone, etc. have been mediates for these reaction pathways. Fourier transform
reported to be present in reactor effluents (4, 5). Several infrared (FTIR) spectroscopy is one of a few techniques
general reaction pathways have been proposed based on capable of detecting surface compounds in heterogeneous
these observations (3). A frequently cited reaction scheme reaction systems under reaction conditions (9). Several
involves the following steps (2–4):
groups have performed in situ FTIR studies to investi-
gate n-butane oxidation on vanadium oxide-based catalysts
n-butane → butenes → 1,3-butadiene → 2,5-dihydrofuran
(
10–19). Due to the complexity of the reaction on VPO
catalysts, however, it is unlikely that surface intermediates
can be identified by FTIR studies using only n-butane. The
→
furan → 2(5H)-furanone → maleic anhydride.
Typically, n-butane oxidative dehydrogenation to butenes general approach has been to perform FTIR studies for the
has been suggested to be the rate-limiting step. Oxidation of adsorption of compounds expected to be related to surface
the butene isomers generates 1,3-butadiene which cyclizes species, such as 1-butene, 1,3-butadiene, furan, maleic an-
upon oxygen insertion to form 2,5-dihydrofuran. According hydride, etc. Previous studies have also emphasized steady-
to this reaction pathway, furan is produced by further allylic state adsorption or reaction conditions. These approaches
oxidation. Oxygen insertion at the �-position of the furan have come under criticism in recent years, and with current
87
0021-9517/99 $30.00
Copyright �c 1999 by Academic Press
All rights of reproduction in any form reserved.

8
8
XUE AND SCHRADER
TABLE 1
Common IR Bands (cm 1) Observed in C4 Hydrocarbon Oxidation on VPO and Their Assignments
�
Position
Assignment
Remarks
1
850
775
�s C=O of maleic anhydride
Weak shoulder band; reported in the study of n-butane (10), 1-butene (10, 18), 1,3-butadiene (19), furan
(
16, 18), 2,5-dihydrofuran (18), 2(5H)-furanone (16).
�
1
� 1
1
1
�as C=O of maleic anhydride Strong band, reported in most of the studies; associated bands at 1795 cm and 1765 cm found in furan
adsorption (15, 17) and 2(5H)-furanone adsorption (16) were attributed to 2(5H)-furanone.
715 � 20 �C=O stretching vibration
Reported in the study of n-butane (10, 18), 1-butene (11, 18), 1,3-butadiene (19), 2,5-dihydrofuran (18),
furan (16, 18), 2(5H)-furanone (16), maleic anhydride (12, 16), and maleic acid (12); assigned to carbonyl
stretching of 2(5H)-furanone (18, 19), maleic acid (10), aldehyde species (16), or furan (15).
1
1
620 � 30 �C=C stretching vibration
Reported in most of the adsorption studies (10–19); attributed to unsaturated surface species; adsorbed
�
1
water (13) and carbon oxide (14) also have bands around 1620 cm
.
560 � 20
�
�
Reported in 1,3-butadiene (15, 19), furan (17), and maleic anhydride (15, 16), acetic anhydride (14)
as COO
�
1
oxidation; attributed to surface carboxylate species together with the band near 1430 cm (14, 17, 19).
Reported in 1,3-butadiene oxidation (18) and furan adsorption (16); attributed to furan.
Attributed to surface-bound furan (16, 18).
1
1
1
490
Ring vibration of furan
Ring vibration of furan
460
�
1
430 � 20
�
s COO
�
Attributed to surface carboxylate species together with the band at 1560 cm (14, 17, 19).
advances in instrumentation, transient FTIR studies have major intermediates. Later, Busca et al. (19) studied 1,3-
now become possible. butadiene adsorption of V2O5–TiO2, and the results sug-
In our previous studies (10, 11), in situ FTIR studies of gested the existence of similar reaction intermediates.
n-butane, 1-butene, 1,3-butadiene oxidation on VPO cata- Table 1 summaries the observed IR bands and their as-
lysts were conducted under steady state reaction conditions signments as reported in these studies (10–19). Bands at
�
1
(
1.5% hydrocarbon in air at various temperatures). In all 1850 and 1775 cm have been used to confirm maleic an-
�
1
cases, bands at 1845 and 1775 cm (due to maleic anhy- hydride formation by oxidation of n-butane and other hy-
dride) and a band near 1720 cm were observed. The later drocarbons. Detection of bands near 1715 and 1620 cm
�
1
� 1
was attributed to maleic acid. Using additional information has also been reported, but their assignment has been less
gained in adsorption studies of crotyl alcohol, maleic acid, clear. Several possible intermediates, such as surface-bound
crotonic acid, and maleic anhydride, a reaction scheme for furan or lactones, have been suggested. In general, the ob-
1
,3-butadiene oxidation was proposed which involved an servation and/or interpretation of the spectra has not been
unspecified peroxide intermediate species (12). consistent among the researchers working in this area, and a
Puttock and Rochester (13) demonstrated the presence convincing reaction pathway has not been developed from
of both Lewis and Brønsted acid sites on the catalyst sur- the FTIR studies.
face by adsorption of water and pyridine. They further con-
In our current research, transient operation techniques
ducted infrared studies of the adsorption of carbon monox- including pulse reaction studies and reactant feed cycling
ide, carbon dioxide, acetic acid, acetic anhydride, 1-butene, have been utilized to induce a departure from steady state
1
,3-butadiene, furan, and maleic anhydride. Many spectral operation. These techniques provide an opportunity to ex-
features were attributed to the interactions between acidic plore the reactivity of intermediates and the surface species.
sites and adsorbates (14, 15). Computer simulations have demonstrated that the con-
Baerns and colleagues (16, 17) investigated the adsorp- centration of surface species under transient operation
tion of 1-butene, 1,3-butadiene, furan, 2(5H)-furanone, and conditions can be significantly different from that under
maleic anhydride on supported VPO catalysts. An adsorp- steady state conditions (20). Therefore, transient IR stud-
tion mode for each adsorbate was proposed. Their analysis ies may reveal surface species that are difficult to detect
favored the argument that furan and 2(5H)-furanone re- under steady state conditions, such as reaction intermedi-
tained their ring structure and were directly oxidized to ates which are transformed very rapidly.
maleic anhydride.
EXPERIMENTAL PROCEDURE
Busca and Centi (18) studied n-butane adsorption on the
surface of (VO)2P2O7. In addition to major bands at 1780
and 1620 cm , a shoulder band at 1720–1710 cm was re-
Catalyst Preparation and Characterization
�
1
� 1
ported which was attributed to a 2(5H)-furanone-like (lac-
The current study used an activated vanadyl pyrophos-
tone) species. Studies of the adsorption of 1-butene, furan, phate (VPO) catalyst prepared in an organic medium
and dihydrofuran were also performed. A reaction scheme by DuPont (21). The catalyst was prepared by reacting
was proposed in which furan and 2(5H)-furanone were the vanadium pentoxide and anhydrous phosphoric acid in

n-BUTANE SELECTIVE OXIDATION
89
isobutanol and benzyl alcohol. Following the reaction, quisitions was 1 min. For the adsorption studies, the time
�
VO(HPO4) � 0.5H2O was precipitated and dried at 100 C. intervals were usually longer. Unless indicated, the spec-
�
Calcination was performed at 400 C in a N2–O2 flow, and trum of the pretreated catalyst at the given temperature has
�
the catalyst wasactivated in a reaction atmosphere at 450 C. been subtracted for spectra presented in this paper. Since
Laser Raman, XRD, and infrared spectroscopic characteri- the transient studies did not alter the structure of the cata-
zation revealed that only the vanadyl pyrophosphate phase, lyst, spectral bands were due to the adsorbed or gas-phase
(
VO)2P2O7, was present. The surface area of the catalyst as species.
2
measured by BET methods was 23 m /g.
FTIR Cell
Catalyst Wafer Preparation and Pretreatment
The design of FTIR cell is shown in Fig. 1. The cell body
was constructed from a stainless steel cylinder 40 mm in
Polished stainless steel dies were used to press the cata-
lyst powder under a load of 15,000 lb. The resulting cata-
lyst wafer had a diameter of 20 mm, a thickness of 0.1 mm
00
�
length with four Swagelok 1/8 fittings welded 90 apart.
The top and bottom fittings were the gas inlet and outlet,
respectively; the third fitting was connected to a vacuum
(
or less), and a weight of about 50 mg. The catalyst wafer
0
0
line. The fourth fitting allowed the insertion of a 0.040
was transferred to a sample holder and placed in the FTIR
cell (Fig. 1). Unless indicated by the catalyst pretreatment
procedure, wafers were exposed to a nitrogen flow (50 ml/s)
thermocouple (Omega Engineering) close to the catalyst
surface. At the center of the cell chamber, a ring groove
tightly held the stainless steel catalyst sample holder. Two
38 mm � 6 mm KBr windows (International Crystal Labo-
ratories) were used as primary windows. Two 25.2 mm �
�
at 300 C for 4 h in the FTIR cell. Bands due to adsorbed
�
1
� 1
water (near 1620 cm and 3600–3100 cm ) were largely
removed by this procedure.
5
mm KBr windows were used as secondary windows.
Brass end caps pressed the KBr windows against the win-
dow mounts which were constructed from Macor machin-
Fourier Transform Infrared Spectrometer
A Nicolet Model 60-SX FTIR spectrometer was used in able glass ceramic (Technical Products). Kalrez O-rings
�
1
these studies. Typically 100 scans were collected at 4 cm
(DuPount Dow Elastormers) were used to seal the win-
resolution in about 40 s. For the reaction studies reported dows. The optical path length between inner surfaces of
here, the time that elapsed between individual spectral ac- the secondary windows was 20 mm. The cell was heated by
FIG. 1. Assembly view of the FTIR cell.

9
0
XUE AND SCHRADER
resistance heating wire (Omega Engineering) which was a stretching vibrational band at 1715 cm 1. A variety of fac-
�
embedded in electric heater cement (Sauereisen). The cell tors will cause the band position of carbonyl groups to shift
�
� 2
could be heated to 400 C and evacuated to 10 Torr.
(22, 23). For the studies reported here, the following guide-
lines can be used. A neighboring electron-withdrawing
group or a strained ring will cause carbonyl stretching bands
to be present at higher frequencies; a neighboring electron-
donating group will make this band appear at lower fre-
quencies. For example, the bands near 1715 cm are typ-
ically due to the stretching vibration modes of carbonyl
groups for saturated aliphatic ketones. A shift of the band
from this position could result from interaction with the
catalyst or could be due to formation of a strained ring.
In regard to other surface species, alkenes may form �-
complexes with the surface cations, and the band for �C == C
may be at lower positions than typical gas-phase species
Gas and Liquid Delivery System
A gas delivery system was used to direct ultrapure N2 car-
rier gas or a 10% oxygen–90% N2 gas mixture into the cell.
Research purity n-butane and 1,3-butadiene (Matheson)
were used as hydrocarbon feeds. Moisture traps, oxygen
traps, and CO2 traps were present. Tylan mass flow con-
trollers were used to regulate the flow. The system could
deliver a continuous flow of C4Hn/O2/N2, C4Hn/N2, O2/N2,
or N2; or, step changes between C4Hn/N2 and O2/N2 could
be introduced. The flow rate of O2/N2 or pure N2 was 50 ml/s
for all studies. Liquid samples (20 �l) were introduced into
the system by syringe injection into a Swagelok “T” in-
stalled on the gas line. Direct injection into the cell was also
used. Solid samples were vaporized and introduced into the
cell. All chemicals used in adsorption studies (except for n-
butane and 1,3-butadiene) were purchased from Aldrich
�
1
�
1
(
1670–1640 cm ). The magnitude of the shift can be
�
1
in the range 10–100 cm (24). Conjugation of unsaturated
C == C bonds also lowers the position of this band. For ex-
ample, �C == C for 2-butene is at 1660 cm (25) while that for
�
1
�
1
1
,3-butadiene is at 1587 cm (26).
(
purity typically above 95% ).
n-Butane
Reaction studies of n-butane involving step changes and
Reaction and Adsorption Studies
continuous flow were performed at several temperatures
from room temperature to 350 C. Figure 2 shows the
Reaction and adsorption studies differed in the method
of feeding the reactant. Prior to the transient studies, the
FTIR cell was operated as a continuous flow microreac-
tor. The concentration of n-butane and 1,3-butadiene was
�
5
mol% in N2, and the concentration ofoxygen was10mol%
in N2. Step changes consisting of C4Hn/N2 vs O2/N2 were
performed using the above-mentioned gas delivery system.
For the adsorption studies, compounds were injected into
the system using a syringe. The behavior of adsorbates in
the presence of N2 or O2/N2 was then observed. For a gas
flow rate of 50 ml/s, the residence time in the empty cell was
less than 1 min.
RESULTS
A large number of spectra have been collected in our cur-
rent studies, and the results were reproducible; only repre-
sentative spectra have been presented in this paper. Unless
otherwise indicated, the spectra displayed in each figure
have the same intensity scale. Before analyzing the results
of these studies, some general observations about the inter-
pretation of the spectra can be made.
For FTIR studies using (VO)2P2O7, only bands above
�
1
1
300 cm are observable due to strong adsorption by the
�
1
catalyst in the region below 1300 cm . In previous studies,
bands present in the region of 2000–1300 cm have been
examined most extensively. Bands near 1850 and 1775 cm
�
1
�
1
are characteristic of the symmetric and asymmetric stretch-
ing vibrational modes of carbonyl groups for cyclic dicar-
bonyl compounds (22). The carbonyl groups of saturated
�
FIG. 2. n-Butane step change reaction at 50 C. Spectra a–c for the
C4H10/N2 step, d–f for the O2/N2 step. (a) 0 min, (b) 1 min, (c) 2 min,
aliphatic ketones (e.g., acetone, ethyl methyl ketone) have (d) 3 min, (e) 4 min, (f) 5 min.

n-BUTANE SELECTIVE OXIDATION
91
�
spectra collected at 50 C under step change conditions priate, as discussed previously. (As reported in the follow-
Figs. 2a–2c in C4H10/N2; Figs. 2d–2f in O2/N2). Each step ing discussion the adsorption of noncyclic carbonyl species
lasted about 3 min, and the time for each collection was less such as ethyl methyl ketone, butyraldehyde, methyl vinyl
(
�
1
than 1 min. Strong bands near 1466 and 1392 cm were ketone, and crotonaldehyde all resulted in major carbonyl
�
1
due to gas-phase n-butane (asymmetrical and symmetri- stretching bands near 1720 cm .)
cal CH3 deformations, respectively). Bands near 1720 and Figure 3 provides spectra obtained at 200 C, with
620 cm were prominent during the experiment. Bands Figs. 3a–3c collected in the C4H10/N2 step and Figs. 3d–3f
at 1637, 1620, and 1608 cm were apparent in Fig. 2a. The collected in the O2/N2 step. Figures 3g–3l were obtained in
608 cm band became more intense during the C4H10/N2 a second similar C4H10/N2–O2/N2 cycle. In the first C4H10/
�
�
1
1
�
1
�
1
1
�
1
� 1
N2 step (Figs. 3a–3c), bands at 1465 and 1396 cm were due
step and obscured neighboring bands at 1637 and 1620 cm
(
Figs. 2b, 2c). The maximum for these broad bandsshifted to to gas-phase n-butane; the presence of gas-phase 1,3-buta-
�
1
� 1
near 1615 cm in the O2/N2 step (Figs. 2d–2f). These bands diene was revealed by the doublets at 1830 and 1808 cm
can be attributed to �C == C of adsorbed olefinic species. The and at 1605 and 1585 cm (27). Bands around 1777, 1723,
difference in their positions may due to the presence of and in the region of 1680–1660 cm were apparent after
single C == C bonds (e.g., butene) or conjugated (e.g., 1,3- 2 min in the reaction (Figs. 3c–3l). In the first O2/N2 step
butadiene) C == C bonds. Adsorbed water also exhibits a (Figs. 3d–3f), bandsdue to gas-phase n-butane and 1,3-buta-
band centered at about 1620 cm (13), and water is a prod- diene diminished, while the bands near 1777 cm intensi-
uct of hydrocarbon oxidation. The bands around 1620 cm
can be partially attributed to adsorbed water. This assign- 1777, 1723, 1658, and 1598 cm became more intense. In
ment has also been supported by the observation of –OH vi- the O2/N2 step of the second cycle (Figs. 3j–3l), the band at
brations near 3500 cm . Broad bands in the region of 1750– 1854 cm became evident. The associated bands at around
�
1
�
1
�
1
� 1
�
1
fied significantly. In the second cycle (Figs. 3g–3l), bands at
�
1
�
1
� 1
�
1
� 1
� 1
1
690 cm (peaks approximately at 1728 and 1713 cm
were also observed. These features were present in spectra hydride.
obtained in the C4H10/N2 step, but clearly became more in- The results of additional studies performed at 300 C are
)
1854 and 1777 cm indicate the generation of maleic an-
�
tense in the O2/N2 step. Assignment of these bands to the illustrated in Fig. 4, in which the catalyst was first treated
carbonyl stretching vibration of noncyclic species is appro- with C4H10/N2 for 1min and then wasexposed to O2/N2. The
�
FIG. 3. n-Butane step change reaction at 200 C. (Left) First C4H10/N2–O2/N2 cycle: spectra a–c for the C4H10/N2 step, d–f for the O2/N2 step, g–i
for the C4H10/N2 step, and j–l for the O2/N2 step. (a) 0 min, (b) 1 min, (c) 2 min, (d) 3 min, (e) 4 min, (f) 5 min. (Right) Second C4H10/N2–O2/N2 cycle:
spectra g–i for the C4H10/N2 step, and j–l for the O2/N2 step. (g) 6 min, (h) 7 min, (i) 8 min, ( j) 9 min, (k) 10 min, (l) 11 min.

9
2
XUE AND SCHRADER
1
,3-Butadiene
Figure 5 shows the spectra collected at 100 C during 1,3-
butadiene step-change reaction studies. Doublets at 1829
�
�
1
� 1
and 1810 cm and at 1605 and 1587 cm were due to gas-
phase 1,3-butadiene. The broad shoulder band in the region
�
1
1
1
390–1370 cm was also partially attributed to gas-phase
,3-butadiene which has a doublet at 1390 and 1371 cm
�
1
(
in-plane bending vibration of terminal vinyl group � == CH ).
2
A broad band with three maxima at 1550, 1487, and
�
1
1
466 cm could be observed. The relative intensity be-
�
1
tween the 1487 and 1466 cm bands changed as the reac-
tion proceeded. This phenomenon was persistent in studies
at temperatures of 250 C and below. The band at 1550 cm
may be ascribed to the asymmetric double bond stretching
mode of furan (�14). The band at 1487 cm was due to
the ring “breathing” mode (�3) of furan, and the 1466 cm
�
� 1
�
1
�
1
band was attributed to a combination tone of ring bend-
ing modes of furan (�9 + �21) (29, 30). It should be noticed
that n-butane adsorption also exhibited bands at 1466 and
�
1
1
392 cm (Fig. 2), but they were due to methyl deformation
�
modes of n-butane. At 100 C, adsorption of 1,3-butadiene
did not result in the formation of maleic anhydride.
�
Figure 6 shows the spectra collected at 300 C during step
change reaction studies. Figures 6a–6f were collected in the
first C4H6/N2–O2/N2 cycle (Figs. 6a–6c in the C4H6/N2 step
FIG. 4. Evolution of surface species in flowing O2/N2 after n-butane
adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 7 min, (d) 14 min, (e) 17 min,
�
(
f) 25 min, (g) 32 min.
spectra collected immediately after the adsorption (Fig. 4a)
�
1
showed two major bands at around 1718 and 1607 cm with
�
1
a shoulder present at 1777 cm . The relative intensity be-
tween the 1718 cm band and the 1777 cm band reversed
as a function of time. A band at 1685 cm became appar-
ent after about 7 min (Fig. 4c); the 1777 cm band and its
shoulder band at 1845 cm clearly were more intense in
the last spectrum (Fig. 4g). The shift in the position of the
607 cm band to 1595 cm may be explained by the pres-
�
1
� 1
�
1
�
1
�
1
�
1
� 1
1
ence of more maleic anhydride on the surface which has a
�
1
�
C == C in the region of 1600–1590 cm (28).
Wenigand Schrader (10) reported spectra for steadystate
reaction of n-butane on VPO catalysts with P/V ratios of
0
1
.9, 1.0, 1.1 at various temperatures. Bands near 1720 and
�
1
� 1
620 cm were observed in all studies. The 1720 cm
�
band was already present at 100 and 200 C, but bands in-
dicating formation of maleic anhydride were not observed
under these conditions. The 1720 cm band was a major
band in most of the spectra presented. FTIR spectra of
n-butane and 1-butene adsorption on (VO)2P2O7 reported
by Busca and Centi (18) were similar. Bands at 1720–
�
1
�
1
� 1
1
710 cm and 1620 cm were observed upon hydrocar-
bon adsorption. Bands characteristic of maleic anhydride
at 1850 and 1780 cm were detected and became more in-
�
FIG. 5. 1,3-Butadiene step change reaction at 100 C. Spectra a–c for
�
1
the C4H6/N2 step, d–f for the O2/N2 step. (a) 0 min, (b) 1 min, (c) 2 min,
tense after oxygen was introduced at higher temperatures. (d) 3 min, (e) 4 min, (f) min.

n-BUTANE SELECTIVE OXIDATION
93
bands at 1540 and 1440 cm 1 (attributed to carboxylate
ions). Aside from bands due to carboxylate ions, the band
positions which were detected were at similar positions to
those for our current study.
�
2
,5-Dihydrofuran and 2,3-Dihydrofuran
Spectra for the adsorption of 2,5-dihydrofuran at 100 C
�
are shown in Fig. 7. According to Klots and Collier (31),
liquid-phase 2,5-dihydrofuran exhibits bands at 1777 (vw),
�
1
1
694 (vw), 1618 (w), 1587 (w), 1483 (w), and 1345 cm (m);
liquid-phase 2,3-dihydrofuran has bands at 1837 (w), 1766
w), 1735 (w), 1619 (s), 1590 (sh), 1453 (w), and 1375 (mw).
Most of the spectral features in Fig. 7a—the bands at 1844,
(
�
1
1
776, 1739, 1621, 1580, 1479, and 1350 cm —were at-
�
1
tributed to 2,3-dihydrofuran. The band at 1726 cm , which
increased while other bands due to the dihydrofurans de-
creased, was characteristic of a noncyclic carbonyl species.
This indicated that dihydrofuran experienced ring cleavage
to form carbonyl species after adsorption.
�
The adsorption of 2,5-dihydrofuran at 300 C rapidly gen-
erated bands near 1785 and 1617 cm with two shoulder
�
1
�
1
bands around 1705 and 1487 cm (Fig. 8). As observed in n-
�
butane and 1,3-butadiene adsorption studies at 300 C, the
�
1
1
784 cm band became more intense as the reaction con-
tinued while other bands diminished in intensity. The band
�
FIG. 6. 1,3-Butadiene step change reaction at 300 C. Spectra a–f were
collected for the first C4H6/N2–O2/N2 cycle: a–c for the C4H6/N2 step, d–f
for the O2/N2 step. Spectra g–i were collected for the O2/N2 step of the third
C4H6/N2–O2/N2 cycle. (a) 0 min, (b) 1 min, (c) 2 min, (d) 3 min, (e) 4 min,
(
f) 5 min, (g) 15 min, (h) 16 min, (i) 17 min.
and Figs. 6d–6f in the O2/N2 step). Figures 6g–6i were col-
lected in the O2/N2 step of the third C4H6/N2–O2/N2 cy-
�
1
cle. Bands at 1830, 1808, 1604, and 1584 cm were again
due to gas-phase 1,3-butadiene; a shoulder band at 1720–
�
1
1
(
700 cm intensified quickly in the first C4H6/N2 step
�
1
Figs. 6a–6c). The band near 1773cm wasdetectable in the
following O2/N2 step (Figs. 6d–6f). After three C4H6/N2–
�
1
O2/N2 cycles, the bands near 1773 and 1712 cm were
clearly prominent (Figs. 6g–6i).
1
,3-Butadiene oxidation on VPO catalysts of different
�
P/V ratios at 300 C was reported by Wenig and Schrader
(
1
11). An intense band with two maxima at 1720 and
690 cm , which was apparently due to carbonyl stretch-
�
1
ing vibrations of noncyclic species, was observed to coexist
with bands attributed to maleic anhydride. Ramstetter and
�
1
Baerns (16) reported a band in this region at 1680 cm
.
For 1,3-butadiene adsorption and oxidation on V2O5–TiO2
studied by Busca et al. (19), the following spectral features
�
1
were observed: bands at 1490 and 1455 cm (attributed
�
1
to molecular furan), a strong band at 1560 cm (assigned
to perturbed furan), a band at 1720 cm (VC == O), bands at
870 and 1790 cm (assigned to maleic anhydride), and (e) 7 min, (f) 10 min.
FIG. 7. Evolution of surface species in flowing O2/N2 after 2,5-
dihydrofuran adsorption at 100 C. (a) 0 min, (b) 1 min, (c) 3 min, (d) 5 min,
�
1
�
�
1
1

9
4
XUE AND SCHRADER
lected directly after injection of vapor phase furan. The
�
1
strong band at 1484 cm was due to the furan ring “breath-
�
1
ing” mode (�3), and the band near 1577 cm was assigned
to the combination of the bending vibrations for furan
ring (�19 + �20) (29). The weak bands at 1871, 1851, and
�
1
1
774 cm could also be associated with molecular adsorp-
�
1
tion of furan. The broad band centered at 1714 cm in the
bottom spectrum (Fig. 10a) was ascribed to overtone and
combination bands of the bending vibrations of the ring
and C–H bonds (29, 30). Upon exposure to O2/N2, bands
due to gas-phase furan at 1577 and 1484 cm disappeared
quickly; weak bands around 1559, 1494, and 1458 cm in
the later spectrum were due to surface-bound furan species.
However, the broad band around 1717 cm persisted and
became the major feature. This band was not attributable to
molecular furan since it should have decreased in intensity
together with the 1577 and 1484 cm bands. Assignment
to a noncyclic carbonyl species was more appropriate.
�
1
�
1
�
1
�
1
�
The adsorption of furan at 300 C resulted in bands at
�
1
around 1786, 1723, 1550, 1488, and 1453 cm (Fig. 11). The
�
1
1
550, 1488, and 1453 cm bands were due to surface-bound
�
1
furan molecules. After 3 h, the 1858 cm band was more
�
1
distinct, the 1786 cm band was stronger, and the bands
FIG. 8. Evolution of surface species in flowing O2/N2 after 2,5-
�
dihydrofuran adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 4 min, (d) 7 min,
(
e) 20 min, (f) 55 min.
near 1705 cm� 1 was more apparent in the first spectrum
�
1
and became obscured as the 1784 cm band became more
prominent. The spectra of 2,5-dihydrofuran on (VO)2P2O7
presented by Busca and Centi (18) also exhibited a band at
�
1
� 1
1
715 cm together with bands at 1840, 1780, and 1620 cm
.
�
The adsorption of 2,3-dihydrofuran at 150 C also re-
sulted in the formation of the band at 1725 cm which in-
creased in intensity with time (Figs. 9a–9d). Bands at 1772,
�
1
�
1
1
617, 1453, and 1357 cm in the early spectra were due
to molecular 2,3-dihydrofuran. These bands decreased in
�
1
intensity compared to the bands near 1725 and 1695 cm
which were indicative of noncyclic species formed by ring
�
cleavage. After the sample was held at 150 C for 30 min, the
�
reaction temperature was slowly raised to 300 C over 1 h
[
the generation of water resulted in increased “noise” in the
�
1
region of 1650–1400 cm (Figs. 9e–9h)]. At higher temper-
atures, bands due to maleic anhydride (1843 and 1778 cm
�
1
)
emerged and became more intense. The apparent shift of
�
1
� 1
the 1725 cm band into the region 1740–1730 cm may be
partially attributable to the formation of maleic anhydride
�
1
which has a shoulder band near 1750 cm
.
Furan
FIG. 9. Evolution of surface species in flowing O2/N2 after 2,3-
�
dihydrofuran adsorption: a–d collected at 150 C, e–h collected during and
�
�
The adsorption of furan at 100 C on vanadyl pyrophos- after the temperature was raised to 300 C. (a) 0 min, (b) 1 min, (c) 9 min,
phate is shown in Fig. 10. The bottom spectrum was col- (d) 33 min, (e) 48 min, (f) 72 min, (g) 90 min, (h) 105 min.

n-BUTANE SELECTIVE OXIDATION
95
due to furan (1550, 1488, and 1453 cm 1) decreased. Two
�
�
1
new maxima at 1611 and 1428 cm were formed which
may be due to adsorbed species with C == C double bonds
and carboxylate species, respectively.
Ramstetter and Baerns (16) studied the adsorption of
furan on an alumina-supported VPO catalyst. Because of
�
1
changes in the relative intensities of the 1490 and 1455 cm
bands, an addition complex of the furan ring at the �-
position to the catalyst was believed to exist on the catalyst
�
�
1
surface. A broad band in the region of 1600–1550 cm was
1
also prominent. More importantly, a band at 1690 cm was
formed upon adsorption which was likely due to a carbonyl
species. In another study performed by Do and Baerns (17),
�
1
the 1690 cm band was also observed and was assigned to
an �,�-unsaturated aldehyde.
2
(5H)-Furanone
Spectra for the adsorption of 2(5H)-furanone at 100 C
are shown in Fig. 12. The split carbonyl stretching bands at
�
�
1
1
777 and 1741 cm are observed typically for unsaturated
five-membered ring lactones in which the double bond is
conjugated with the carbonyl group (32). The treatment in
�
O2/N2 at 100 C eventually resulted in the disappearance
of these two bands. A broad band in the region of 1740–
1690 cm was evident which was characteristic of noncyclic
carbonyl compounds.
FIG. 10. Evolution of surface species in flowing O2/N2 after furan
adsorption at 100 C. (a) 0 min, (b) 1 min, (c) 3 min, (d) 7 min, (e) 10 min.
�
1
�
FIG. 11. Evolution of surface species in flowing O2/N2 after furan
adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 5 min, (d) 23 min, (e) 36 min,
FIG. 12. Evolution of surface species in flowing O2/N2 after 2(5H)-
furanone adsorption at 100 C. (a) 0 min, (b) 5 min, (c) 30 min, (d) 55 min,
�
�
(
f) 60 min, (g) 104 min, (h) 180 min.
(e) 105 min.

9
6
XUE AND SCHRADER
in the later spectra. There wasno evidence for the formation
�
1
of maleic anhydride (bands at 1850 and 1780 cm ).
5
-Hydroxy-2(5H)-Furanone
The result of 5-hydroxy-2(5H)-furanone adsorption at
�
2
00 C is shown in Fig. 15. Three major bands were ob-
�
1
served at 1800, 1756, and 1707 cm with a shoulder band at
729 cm : all of these bands are in the carbonyl stretching
�
1
1
vibration region. The first two bands are likely due to car-
bonyl stretching modes of 5-hydroxy-2(5H)-furanone. The
�
1
bands near 1729 and 1707 cm were due to the noncyclic
carbonyl compounds generated in the reaction. The 1800
�
1
and 1756 cm bands diminished quickly, and the band at
1
�
1
707 cm was very intense in the beginning but slowly be-
came less apparent. Similar behavior was observed for the
�
1
bands at around 1564 and 1405 cm which were attributed
�
to carboxylate species. For adsorption studies at 100–250 C,
prolonged adsorption of5-hydroxy-2(5H)-furanone did not
produce a detectable amount of maleic anhydride.
Maleic Anhydride
The adsorption of maleic anhydride on the catalyst at
�
1
00 C is shown in Fig. 16. The bands near 1851, 1795–1773,
�
1
and a shoulder band at 1754 cm were due to maleic anhy-
dride (28). The bands at around 1683 (shoulder), 1565, and
1451 cm were the result of transformation of the adsorbed
FIG. 13. Evolution of surface species in flowing O2/N2 after 2(5H)-
�
1
�
furanone adsorption at 300 C. (a) 0 min, (b) 3 min, (c) 11 min, (d) 44 min,
(
e) 67 min.
�
The adsorption of 2(5H)-furanone at 300 C instantly
�
1
generated an intense peak at 1793 cm with a shoulder
band near 1716 cm (Fig. 13). Maxima near 1622 (�C == C),
�
1
�
1
1
559 (�as,COO� ), and 1421 cm
(�as,COO� ) were also ob-
served. At longer time periods, the intensity of the
�
1
1
1
793 cm peak decreased while the intensities of the
�
1
� 1
783 cm band and its 1850 cm shoulder band increased.
Formation of maleic anhydride was indicated.
Ramstetter and Baerns (16) also studied 2(5H)-fura-
none adsorption on a supported VPO catalyst. Other than
bandsdue to molecular and surface-bound 2(5H)-furanone,
�
1
a broad band (maxima at 1705 and 1690 cm ) was detected
�
at 100 C which was assigned to aldehyde species.
�
-Butyrolactone
Figure 14 shows the evolution of spectra following the
adsorption of � -butyrolactone. The bands at 1813 and
�
1
1
�
749 cm were due to the carbonyl stretching mode of
�
1
-butyrolactone. The band at 1699 cm was attributed to
the formation of a noncyclic carbonyl species. The band
�
1
in the region of 1640–1620 cm was likely due to gen-
eration of C == C bonds or water. Bands due to molecular
FIG. 14. Evolution of surface species in flowing O2/N2 after � -
butyrolactone adsorption at 300 C. (a) 0 min, (b) 3 min, (c) 5 min, (d) 9min,
�
-butyrolactone decreased with time while the noncyclic
�
�
1
carbonyl stretching band (� 1699 cm ) became dominant (e) 16 min, (f) 21 min, (g) 28 min.

n-BUTANE SELECTIVE OXIDATION
97
FIG. 15. Evolution of surface species in flowing O2/N2 after 5-
�
hydroxy-2(5H)-furanone adsorption at 200 C. (a) 0 min, (b) 2 min,
c) 5 min, (d) 10 min, (e) 13 min, (f) 20 min, (g) 34 min.
(
FIG. 16. Evolution of surface species in flowing O /N after maleic
anhydride adsorption at 100 C. (a) 0 min, (b) 2 min, (c) 6 min, (d) 13 min,
2
2
�
(
e) 19 min.
species. The shoulder band near 1683 cm 1 was more dis-
�
�
1
cernable in the later spectra when the 1780 cm band had
substantially decreased in intensity. The 1683 cm band
was likely due to a surface-bound aldehyde complex. The
�
1
�
1
1
565 and 1451 cm bands corresponded to the asymmetric
and symmetric vibration of carboxylate ions (33). At higher
�
temperatures (e.g., 200 C), the decomposition of maleic an-
hydride was more rapid.
Adsorption studies of maleic anhydride by Puttock et al.
(
15) and Wenig et al. (12) provided little information due
to the interference from the strong adsorption bands of
the VPO catalyst. However, bands due to maleic anhydride
and carboxylate ions could be detected. In the studies by
�
1
Baerns et al. (16, 17), the 1720 cm band and a shoulder at
�
1
1
680 cm were observed.
Ethyl Methyl Ketone and Butyraldehyde (C4H8O)
�
The adsorption of ethyl methyl ketone at 300 C (Fig. 17)
resulted in the rapid observation of a carbonyl stretching
�
1
� 1
band at 1724 cm , a C == C stretching band at 1620 cm
,
�
1
and a band in the region of 1790–1770 cm (characteris-
tic of maleic anhydride). As reaction continued, the bands
�
1
near 1780 and 1850 cm were became dominant while the
�
1
1
1
724 and 1620 cm bands decreased accordingly. The band
FIG. 17. Evolution of surface species in flowing O2/N2 after ethyl
�
around 1427 cm indicated the generation of carboxylate
species.
�
methyl ketone adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 7 min,
(d) 19 min, (e) 39 min, (f) 99 min, (g) 154 min.

9
8
XUE AND SCHRADER
�
1
the intensity of bands near 1711 cm (�C == O of noncyclic
�
1
species) and 1618 cm (�C == C) decreased while the inten-
sity of the 1778 cm band increased. A shoulder band at
853 cm behaved similarly (Fig. 19).
�
1
�
1
1
FTIR spectra for crotonaldehyde adsorption are shown
�
1
in Fig. 20. The doublet at 1722 and 1707 cm was due to
�
1
gas-phase crotonaldehyde. A shoulder band at 1674 cm
was also present in the region for carbonyl stretching
modes. The adsorption again exhibited similar results in
which maleic anhydride (indicated by the band around
�
1
1
779 cm ) was produced while the intensities of bands for
�
1
open chain carbonyl compound (� 1701 cm ) and C == C
bonds (� 1620 cm� ) diminished.
1
Methyl Acrylate, Vinyl Acetate, Vinyl Acetic Acid,
Crotonic Acid (C4H6O2)
Each of these compounds contains two oxygen atoms,
one of which is incorporated in a carbonyl group. An in-
�
1
tense band near 1720 cm was observed in all cases when
these compounds were adsorbed on the catalyst surface.
However, adsorption at the same conditions involved in
the previous studies did not produce detectable amounts of
maleic anhydride.
FIG. 18. Evolution of surface species in flowing O2/N2 butyraldehyde
�
adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 6 min, (d) 8 min, (e) 18 min,
(
f) 33 min, (g) 44 min.
Figure 18 provides spectra for butyraldehyde adsorption
�
� 1
on the catalyst at 300 C. The 1752 cm band originated
from butyraldehyde whose carbonyl-stretching band was at
a higher position than ketones (22, 23). The band at around
�
1
1
620 cm was ascribed to the C == C stretching mode. The
�
1
band near 1712 cm likely resulted form a carbonyl group
conjugated with the C == C bond. In the earlier spectra, the
�
1
�
1
1
752 cm band decreased while the 1712 cm band grew
which indicated the formation of an unsaturated carbonyl
compound. Later in the reaction, the growth of the max-
�
1
imum at 1773 cm indicated the formation maleic anhy-
dride.
Ethyl methyl ketone and butyraldehyde are saturated
compounds. In both cases spectral features due to the C == C
double bonds were generated quickly as indicated by the
�
1
emergence of the characteristic band at 1620 cm . As
the reaction proceeded, this band decreased in intensity
�
1
with the carbonyl stretching band (� 1720 cm ). This indi-
cated that hydrogen abstraction occurred upon adsorption
to form an unsaturated noncyclic species which was then
converted to maleic anhydride.
Methyl Vinyl Ketone and Crotonaldehyde (C4H6O)
FIG. 19. Evolution of surface species in flowing O2/N2 after methyl
vinyl ketone adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 6 min, (d) 19 min,
e) 32 min, (f) 39 min, (g) 43 min, (h) 48 min, (i) 56 min, ( j) 62 min,
�
The nature of methyl vinyl ketone adsorption was ob-
(
served to be nearly identical to that of ethyl methyl ketone: (k) 76 min, (l) 90 min.

n-BUTANE SELECTIVE OXIDATION
99
FIG. 20. Evolution of surface species in flowing O2/N2 after croton-
aldehyde adsorption at 300 C. (a) 0 min, (b) 3 min, (c) 4 min, (d) 8 min,
e) 12 min, (f) 25 min, (g) 47 min.
FIG. 21. Evolution of surface species in flowing O2/N2 after 2-butene-
,4-diol adsorption at 300 C. (a) 0 min, (b) 2 min, (c) 4 min, (d) 8 min,
e) 14 min, (f) 34 min, (g) 49 min, (h) 90 min.
�
�
1
(
(
2
-Butene-1,4-diol (C4H8O2)
Although 2-butene-1,4-diol showed no signs of reaction
�
�
at 150 C in the empty cell, this compound reacted at 300 C
in the gas-phase to produce crotonaldehyde. Adsorption of
2
-butene-1,4-diol on the VPO catalyst (Fig. 21) also exhib-
ited behavior the same as that for crotonaldehyde adsorp-
tion (Fig. 20).
Maleic Acid and Fumaric Acid (C4H4O4)
Maleic acid can be readily dehydrated to maleic an-
hydride upon heating in the gas-phase. The adsorption
�
of maleic acid on the catalyst at 150 C is shown in
Fig. 22. Bands near 1850, 1775, 1722, 1682, 1635, 1567, and
�
1
1
409 cm were detected. The two carbonyl bands at 1850
�
1
and 1775 cm (maleic anhydride) were the result of gas-
phase dehydration. The adsorbed species showed split car-
bonyl stretching bands at 1722 and 1682 cm . The bands at
�
1
�
1
1
567and 1409cm were due to carboxylate species. The ad-
sorption pattern resembles the results for maleic anhydride
�
adsorption (Fig. 16). At 250 C, surface species decomposed
to carbon oxides very quickly.
Fumaric acid adsorption on the catalyst at 200 C (Fig. 23)
�
exhibited a pattern similar to that for maleic acid adsorp-
�
1
tion: a carbonyl stretching band at 1720 cm , a C == C dou-
FIG. 22. Evolution of surface species in flowing N2 after maleic acid
�
1
�
ble bond stretching band at 1643 cm , and carboxylate ion adsorption at 150 C. (a) 0 min, (b) 1 min, (c) 3 min, (d) 6 min, (e) 25 min.

1
00
XUE AND SCHRADER
n+
carbonyl oxygen (C=O: → M ); �C == O would therefore be
lowered. This effect has been elaborated by Davydov using
the example of acrolein oxidation (24). Maleic acid gas-
phase dehydration produced a significant amount of maleic
anhydride, which can adsorb on the catalyst, perhaps dis-
sociatively, to form aldehyde functionalities. The aldehyde
groups can coordinate with Lewis acid sites, and a carbonyl-
�
1
stretching band at 1682 cm results. Fumaric acid dehydra-
tion to maleic anhydride did not occur at low temperatures
since its hydroxyl groups are trans to each other. Therefore,
no surface aldehyde functionalities were formed upon ad-
sorption and the only carbonyl stretching band observed
�
1
was at 1720 cm because of the –COOH groups.
These observations suggest that (a) fumaric acid and
maleic acid can be converted to maleic anhydride in gas-
phase, (b) the acids adsorb to form carboxylate species
which decompose upon heating.
DISCUSSION
In the current study, n-butane adsorption on (VO)2P2O7
was observed to produce unsaturated compounds at tem-
�
peratures below 100 C. Gas-phase 1,3-butadiene could be
detected, and the formation of other unsaturated com-
�
1
pounds were indicated by a band near 1720 cm and by
FIG. 23. Evolution of surface species in flowing N2 after fumaric acid
adsorption at 200 C. (a) 0 min, (b) 7 min, (c) 15 min, (d) 20 min, (e) 25 min.
�
1
�
a weaker band in the region of 1680–1660 cm (Figs. 2
and 3). At higher temperatures, IR bands due to maleic an-
�
1 1
�
1
hydride (1850 and 1775 cm ) became significant. These
species were detected at short time intervals under tran-
sient reaction conditions.
bands at 1565 and 1404 cm were observed. The surface
species also decomposed quickly at higher temperatures
�
(
T > 250 C). The gas-phase dehydration of fumaric acid to
�
Important intermediates were associated with the ap-
maleic anhydride can occur at 300 C.
�
1
pearance of a broad band near 1720 cm . This band
emerged in the n-C4H10/N2 step and persisted in the O2/N2
step when bands due to gaseous compounds (e.g., 1,3-
butadiene) disappeared (Fig. 3). The surface species asso-
ciated with this band was formed before the detection of
maleic anhydride, and it decreased in intensity as maleic
anhydride was produced (Figs. 2–4). Therefore, the band
near 1720 cm was likely related to precursors to maleic
anhydride. Busca and Centi (18) ascribed a band around
715 cm and a band at 1780 cm to 2(5H)-furanone.
The shifts in carbonyl stretching bands were interest-
ing: adsorbed maleic acid exhibited shifts in band positions
�
1
[
bands at 1720 and 1680 cm compared to its typical posi-
�
1
tion at 1705 cm (33)]. The carbonyl band of furamic acid
shifted from its normal position at 1680 to 1720 cm , and
spectra did not show a distinguishable band at 1680 cm
�
1
�
1
at temperatures where maleic anhydride was not produced
�
1
�
(
150–250 C). Maleic and fumaric acid are rotational iso-
mers. At higher temperatures, their adsorbed forms should
form an identical surface complex. Therefore, the band at
�
1
� 1
1
�
1
However, the IR spectrum of 2(5H)-furanone has two
1
682 cm in Fig. 22 was likely a result of chemical conver-
�
1
�
1
bands at different positions, 1780 and 1740 cm , and there-
fore this assignment is unlikely. Our studies of 2(5H)-
furanone (Fig. 12) and � -butyrolactone (Fig. 14) adsorp-
tion also indicated that these cyclic species were not stable
on the catalyst surface: both compounds underwent ring
cleavage upon adsorption. Wenig and Schrader assigned
sion of maleic acid. The �C == O at 1720 cm in both cases
could be readily attributed to the carbonyl stretching of the
COOH groups. The interaction between the surface and
the carbon atom of the –COOH groups increased �C == O by
–
withdrawing electron density from the carbon atom. The
�
1
1
680 cm band [also observed in n-butane oxidation and
adsorption (Fig. 3 and Fig. 4), maleic anhydride adsorption
�
�
at 100 C (Fig. 16), and crotonaldehyde adsorption at 300 C
Fig. 20)] was due to the interaction between the aldehyde
1
� 1
� 1
The 1850 cm band is less intense than 1780 cm band because the
(
symmetric vibration of two carbonyl groups of maleic anhydride reduces
the dipole moment (22). Based on our observation, this effect is more
significant at higher temperatures: a decrease in the temperature increases
group with coordinately unsaturated metal cations (Lewis
acid sites). The metal cation could form a �-complex with
the carbonyl through sharing the lone electron pair of the the intensity of the 1850 cm band.
�
1

n-BUTANE SELECTIVE OXIDATION
101
the 1720 cm 1 band to maleic acid (10–12). The observa- 2(5H)-furanone and cyclopentanone can exhibit split car-
tion in our transient studies that this band emerged at room bonyl bands as a result of Fermi resonance interactions.
temperature before the formation of maleic anhydride dis- Based on this information about carbonyl stretching vibra-
�
�
1
counts this assignment also.
The IR bands of carbonyl stretching modes for most non- signed to noncyclic carbonyl species.
tions, therefore, the band near 1720 cm can best be as-
�
1
cyclic compounds such as ketones and aldehydes are in the
Although not as intense as the 1720 cm band, other
region of 1750–1700 cm (22–24, 27). The interaction of bands near 1680 cm were also observed for n-butane
�
1
� 1
carbonyl groups with the catalyst may shift the positions of oxidation (Fig. 3). These bands indicate the presence of
�
1
these bands, according to the different modes of adsorp- two types of surface complexes: the 1720 cm band may
tion. For instance, the carbonyl oxygen may form a hy- be due to a species whose carbonyl carbon atom interacts
drogen bond with surface hydroxyl groups (C == O ���HO), with the catalyst surface (O == C ���surface); the bands near
�
1
or it may form a coordination bond with a surface cation 1680 cm may be due to a species whose carbonyl oxy-
n+
4+
(
C == O ���M ). This inductive effect increases the polarity gen interacts with Lewis acid sites (C == O ���V ) or with
of the C == O bond and makes it more ionic; a frequency shift surface hydroxyl groups (C == O ���OH).
to lower wavenumbers results (22, 24). On the other hand,
The adsorption studies of 2,5-dihydrofuran, furan, and
if electron-withdrawing interactions occur through the 2(5H)-furanone verified that they can be oxidized to maleic
carbon atom (electron-withdrawing group � C == O), the anhydride on the VPO catalyst, as indicated by the pres-
�
1
polarity of the C == O bond is reduced, and the carbonyl fre- ence of peaks at 1850 and 1780 cm . However, in do-
quencies will increase (22). The 1�C == O caused by such inter- ing so, these compounds likely undergo ring cleavage first.
actions is relatively small, and the �C == O of many aldehydes Even at room temperature, the adsorption of these com-
�
1
and ketones remains close to 1700 cm . For instance, the pounds on the catalyst resulted in ring opening: this was ev-
carbonyl stretching band of surface formaldehyde species idenced by the disappearance of ring vibration bands and
(
24), adsorbed acrolein (24), products of propane partial the emergence of the noncyclic carbonyl stretching band
�
1
oxidation (34), and the aldehydes and ketones involved in near 1720 cm . In contrast, ethyl methyl ketone, methyl
the current study are all in the region of 1750–1650 cm
�
1
.
vinyl ketone, butyraldehyde, and crotonaldehyde oxida-
The carbonyl frequencies of strained ring compounds tion resulted in formation of maleic anhydride (1850 and
�
1
have unique features such as their occurrence at higher 1775 cm ). No evidence for cyclic intermediates such as
wavenumber positions and/or the presence of double car- furan or 2(5H)-furanone was observed.
bonyl bands; these features distinguish these compounds
The aldehydes and ketones used in these adsorption stud-
from open chain species. For instance, Bellamy (22) showed ies still require two oxygen atoms to form maleic anhy-
that carbonyl frequencies of four or five-membered rings dride. However, 2(5H)-furanone is only “one oxygen away”
�
1
are usually higher than 1750 cm . Also, 1,3-dicarbonyl from maleic anhydride and could be converted to maleic
compounds exhibit two carbonyl bands due to the symmet- anhydride through an open chain intermediate. This sur-
ric and asymmetric vibration modes of the two carbonyl face species could therefore be involved in the reaction step
groups, as exemplified by maleic anhydride (bands at 1855 just prior to the formation of maleic anhydride. Figure 24
�
1
and 1784 cm ) and succinic anhydride (bands at 1876 and provides several possible products of 2(5H)-furanone ring
1
�
1
799 cm ). Some cyclic monocarbonyl compounds such as cleavage. Formation of many open chain compounds is
FIG. 24. Possible pathways of 2(5H)-furanone ring cleavage and resulting compounds.

1
02
XUE AND SCHRADER
possible: crotonic acid, methyl acrylate, vinyl acetate, vinyl adsorbing 1-butene on an alumina supported VPO catalyst,
acetic acid, 4-hydroxy-crotonaldehyde, 2-butene-1,4-dial, Ramstetter and Baerns (16) observed the carbonyl stretch-
�
1
or formic acid allyl ester.
ing band at 1680 cm . Puttock and Rochester (15) reported
Adsorption of crotonic acid, vinyl acetic acid, vinyl ac- the isomerization of 1-butene to trans and cis-2-butenes on
etate, and methyl acrylate on vanadyl pyrophosphate at the surface of (VO)2P2O7.
�
3
00 C did not produce bands for maleic anhydride. The
The adsorption of 1,3-butadiene on the catalyst at low
0
� 1
RCO-OR bond is relatively reactive. Breaking of this bond temperatures did not show a band near 1720 cm . Instead,
will produce 4-hydroxy-crotonaldehyde or 2-butene-1,4- it exhibited bands at 1550, 1490, and 1460 cm which can be
dial. The carbonyl stretching bands of these compounds are ascribed to adsorbed furan. However, the change in the rel-
�
1
�
1
� 1
allnear 1700cm (35, 36). Because ofthe reactivityofthese ative intensities between the 1490 and 1460 cm band sug-
aldehydes, it was not possible to purchase these compounds. gested the formation of a furan complex whose �-position
The adsorption of their structural analog, 2-butene-1,4-diol, was attacked. A possible surface complex was proposed by
was performed, but dehydration occurred in the gas-phase Crew and Madix in their studies of furan oxidation (39).
to form crotonaldehyde. However, additional experimen- This complex may lead to ring cleavage to form 2-butene-
tal evidence for this intermediate has been reported in the 1,4-dial. Oxidation of 1,3-butadiene at higher temperatures
literature. H o¨ nicke (37, 38) detected substantial amount clearly exhibited a band at 1712 cm which is consistent
of 2-butene-1,4-dial when investigating 1,3-butadiene par- with the eventual conversion of this intermediate to maleic
tial oxidation on supported V2O5 catalysts, and it was sug- anhydride (Fig. 6).
�
1
gested to be a precursor for maleic anhydride. In our re-
The oxidation of n-butane and 1,3-butadiene on VPO
actor studies using mass spectrometry, a peak at m/e = 84 catalysts are known to result in very different product dis-
was detected for n-butane and 1,3-butadiene oxidation. tributions (4). The current study suggests that reason for
Kubias et al. (5) also detected a peak at m/e = 84 in n-butane this difference may be attributed the initial adsorption of
oxidation on VPO catalysts. They further showed that the these compounds. n-Butane adsorption and activation is
compound corresponding to this peak contains two oxygen likely a transformation largely related to hydrogen abstrac-
18
atoms since peaks at m/e = 86, 88 emerged upon O iso- tion. It is possible that only active sites capable of hydrogen
topic labeling. Additional evidence for the existence of such abstraction from n-butane are geometrically and energet-
surface species has come from Crew and Madix (39) who ically favorable for selective conversion of the adsorbed
studied furan oxidation to maleic anhydride on Ag(110). species to maleic anhydride. In contrast, 1,3-butadiene, due
Using O18 labeling and temperature-programmed reaction to its possession of double bonds, adsorbs strongly on the
spectroscopy (TPRS) experiments, it was found that the catalyst surface. Species adsorbed on selective sites will be
furan ring oxygen became one of the carbonyl oxygens of converted to desired oxygenated products; other adsorbed
maleic anhydride product. They postulated that the furan species could be involved in side reactions such as oligomer-
ring opened to form an intermediate, specifically 2-butene- ization, isomerization, etc. Other workershave reported the
1
,4-dial, and re-closed to form maleic anhydride. Based on
these previous studies and the results of our new research,
a noncyclic dialdehyde surface species structurally related
to 2-butene-1,4-dial is proposed to be the direct precursor
to maleic anhydride.
It should further be emphasized that the precursors to
maleic anhydride are unsaturated compounds. The spectral
�
1
bands 1620 cm for C == C double bond near were always
observed upon adsorption of the hydrocarbons, even if the
adsorbates were saturated compounds (e.g., n-butane, ethyl
methyl ketone, butyraldehyde). Moreover, the evolution of
�
1
�
C == C always paralleled that of the 1720 cm band for �C == O.
The results of IR studies of 1-butene oxidation on VPO
catalysts agree well with these observations. The study con-
ducted by Wenig and Schrader (11) showed that a band
�
1
near 1725 cm was in present together with bands at 1850,
1
�
1
775, and 1615 cm . Busca and Centi (18) observed the
same bands in their study. Moreover, they showed that the
�
1
1
715 cm band was intense early in the adsorption when
�
1
the 1850 and 1775 cm were weak, and diminished after
being heated in oxygen when the later bands dominated. By ene/N2 step in reaction studies.
FIG. 25. Mass spectra of gas phase products during the 1,3-butadi-

n-BUTANE SELECTIVE OXIDATION
103
FIG. 26. Proposed reaction pathway for n-butane oxidation to maleic anhydride.
formation of several products for 1,3-butadiene oxidation.
CONCLUSION
Centi and Trifir o` (40) have observed maleic anhydride, fu-
ran, and crotonaldehyde for 1,3-butadiene oxidation on
VPO. H o¨ nicke (37) identified as many as 33 oxidation
products in partial oxidation of 1,3-butadiene on supported
V2O5 catalysts. The major products discovered in his study
included maleic anhydride, phthalic anhydride, crotonalde-
hyde, furan, dihydrofurans, 2-butene-1,4-dial, and benzene
derivatives.
(
VO)2P2O7 issufficientlyactive to activate saturated C–H
�
bonds at low temperatures, e.g. 50 C. Although formation
of furan from 1,3-butadiene oxidation was observable and
conversion of cyclic compounds to maleic anhydride can-
not be completelyexcluded, the evidence from our transient
IR and reactor studies indicates that: (a) the predominant
surface species for n-butane oxidation are unsaturated non-
cyclic carbonyl compounds, and (b) these reactive surface
species are the precursors to maleic anhydride. Based on
these results and comparison with the literature, the reac-
tion pathway shown in Fig. 26 can be proposed for n-butane
partial oxidation to maleic anhydride.
This pathway suggests that for the main route for n-
butane conversion to maleic anhydride, hydrogen abstrac-
tion occurs first to generate surface-bound olefinic com-
pounds. Subsequent oxidation of these compounds forms
unsaturated noncyclic carbonyl intermediates; formation of
a noncyclic dialdehyde surface species appears likely. This
surface species undergoes oxygen insertion and cyclization
to form maleic anhydride. Complete combustion of inter-
mediates or maleic anhydride involves carboxylate surface
species which can be transformed to carbon oxides.
Benzene derivatives can be produced form 1,3-butadiene
dimerization (41). In our work, we monitored the reactor
effluent for 1,3-butadiene oxidation over VPO catalysts at
�
3
00 C with mass spectrometry using a microreactor system.
Intense peaks were observed at m/e = 68, 78, 91, and weak
peaks were detected at m/e = 82, 84, 98, 106 (see Fig. 25).
The peak at m/e = 68 is due to furan, m/e = 98 is due to
maleic anhydride, m/e = 78 is due to benzene, and m/e = 84
may be attributed to 2-butene-1,4-dial. The peak at m/e =
9
1 is likely due to methylbenezene molecular ions, while
the signal at m/e = 106 might be ascribed to ethylbenzene.
Comparing the spectra collected at the first, second, and
third minutes of a 1,3-butadiene/N2 step (Fig. 25), signals
from furan (m/e = 68), benzene (m/e = 78), and methyl-
benzene (m/e = 91) were intense at the beginning of the 1,3-
butadiene step and decreased as the reaction proceeded;
peaks due to maleic anhydride (m/e = 98) remained consis-
tently low. This observation does not support the proposal
that furan is a direct precursor to maleic anhydride.
ACKNOWLEDGMENTS
The authors are grateful to K. Kourtakis at DuPont Central Research
and Development for synthesis of the catalyst used in these studies. This
work was conducted through the Ames Laboratory which is operated
through the U.S. Department of Energy by Iowa State University under
The adsorption of maleic anhydride, maleic acid, and fu-
maric acid provided insights to the mechanism of maleic an-
hydride oxidation. The adsorption of all these species pro- Contract W-7405-Eng-82. Support from the Office of Basic Energy Sci-
ences, Chemical Sciences Division is also acknowledged.
duced carboxylate species, which lead to the formation of
carbon oxides (complete oxidation). The band at 1682 cm
�
1
evidenced the presence of aldehyde groups bound to Lewis
acid sites corresponding to maleic anhydride oxidation.
Lewis acid sites may play a role in maleic anhydride de-
composition, for instance, by strongly anchoring the maleic
anhydride molecules on the surface through the coordina-
REFERENCES
1
2
. Hodnett, B. K., Catal. Rev.-Sci. Eng. 27, 373 (1985).
. Cavani, F., and Trifir o` , F., in “Catalysis, Vol. 11,” The Royal Society
of Chemistry, London, 1994.
3
. Albonetti, S., Cavani, F., and Trifir o` , F., Catal. Rev.-Sci. Eng. 38, 413
4+
tion bond: C == O���V
.
(1996).

1
04
XUE AND SCHRADER
4
5
6
7
. Centi, G., Trifir o` , F., Enber, J. R., and Franchetti, V. M., Chem. Rev.
8, 55 (1988).
. Kubias, B., Rodemerck, U., Zanthoff, H.-W., and Meisel, M., Catal.
Today 32, 243 (1996).
. Zhang-Lin, Y., Forissier, M., Sneeden, R. P., V e` drine, J. C., and Volta,
J. C., J. Catal. 145, 256 (1994).
. Zhing-Lin, Y., Forissier, M., V e` drine, J. C., and Volta, J. C., J. Catal.
24. Davydov, A. A., in “Infrared Spectroscopy of Adsorbed Species on the
8
Surface of Transition Metal Oxides” (C. H. Rochester, Eds.), Wiley,
New York, 1984.
25. Mckean, D. C., Mackenzie, M. W., Morrisson, A. R., Lavalley, J. C.,
Janin, A., Fawcett, V., and Edwards, H. G. M., Spectrochim. Acta. A
41, 435 (1985).
26. Bondybey, V. E., and Nibler, J. W., Spectrochim. Acta. A 29, 645 (1973).
27. Pouchert, C. J., “The Aldrich Library of FT-IR Spectra, Edition I,
Vol 3,” Aldrich Chemical, Co., 1989.
28. Mirone, P., and Chiorboli, P., Spectrochim. Acta 18, 1425 (1962).
29. Klots, T. D., Chirico, R. D., and Steele, W. V., Spectrochim. Acta. A 50,
765 (1994).
1
45, 267 (1994).
8
9
. Zi o´ lkowski, J., Bordes, E., and Courtine, P., J. Catal. 122, 126 (1990).
. Busca, G., Catal. Today 27, 457 (1996).
1
1
1
1
0. Wenig, R. W., and Schrader, G. L., J. Phys. Chem. 90, 6480 (1986).
1. Wenig, R. W., and Schrader, G. L., J. Phys. Chem. 91, 1911 (1987).
2. Wenig, R. W., and Schrader, G. L., J. Phys. Chem. 91, 5674 (1987).
3. Puttock, S. J., and Rochester, C. H., J. Chem. Soc., Faraday Trans. 1
30. Rico, M., Barrachina, M., and Orza, J. M., J. Mol. Spectrosc. 24, 133
(1967).
8
2, 2773 (1986).
31. Klots, T. D., and Collier, W. B., Spectrochim. Acta. A 50, 1725 (1994).
1
1
4. Puttock, S. J., and Rochester, C. H., J. Chem. Soc., Faraday Trans. 1 32. Jones, R. N., Angell, C. L., Ito, t., and Smith, R. J. D., Can. J. Chem.
2, 3013 (1986). 37, 2007 (1959).
5. Puttock, S. J., and Rochester, C. H., J. Chem. Soc., Faraday Trans. 1 33. Hargreaves, M. K., and Stevinson, E. A., Spectrochim. Acta. 21, 1681
8
8
2, 3033 (1986).
(1965).
16. Ramstetter, A., and Baerns, M., J. Catal. 109, 303 (1988).
17. Do, N. T., and Baerns, M., Appl. Catal. 45, 9 (1988).
18. Busca, G., and Centi, G., J. Am. Chem. Soc. 111, 46 (1986).
19. Busca, G., Ramis, G., and Lorenzelli, V., J. Mol. Catal. 55, 1 (1989).
20. Pekar, M., and Koubek, J., Chem. Eng. Sci. 52, 2291 (1997).
21. Horowitz, H. S., Blackstone, C. M., Sleight, A. W., and Teufer, G.,
Appl. Catal. 38, 193 (1988).
34. Centi, G., Marchi, F., Perathoner, S., J. Chem. Soc., Faraday Trans.
92(24), 5141 (1996).
35. Cyr, A., Wermeckes, B., and Beck, F., J. Prakt. Chem. 336, 602
(1994).
36. Hufford, D. L., Tarbell, D. S., and Koszalka, T. R., J. Amer. Chem. Soc.
74, 3014 (1952).
37. H o¨ nicke, D., J. Catal. 105, 10 (1987).
2
2. Bellamy, L. J., “The Infrared Spectra of Complex Molecules, 2nd ed. 38. H o¨ nicke, D., J. Catal. 105, 19 (1987).
Vol. 2: Advances in Infrared Group Frequencies,” Chapman & Hall,
London, 1980.
3. Nakanishi, K., and Solomon, P. H., “Infrared Absorption Spec-
troscopy, Second Edition” Holden–Day, Oakland, CA, 1977.
39. Crew, W. W., and Madix, R. J., J. Am. Chem. Soc. 115, 729 (1993).
40. Centi, G., and Trifir o` , F., J. Mol. Catal. 35, 255 (1986).
41. Diesen, R. W., U.S. Patent: 5,276,257, assigned to The Dow Chemical
Company, 1994.
2