Homogeneous catalysis by evaporated solids

Article abstract of DOI:10.1016/S1381-1169(99)00088-6

About 10 years ago, sodium aluminate has been found as suitable catalyst for the pure dehydrogenation of methanol to anhydrous formaldehyde. At first, the reaction was supposed to follow a heterogeneous mechanism. Later on, experimental results with a special set-up were at variance. Recent investigations revealed the loss of catalytically active species from the solid aluminate into the vapour phase where the entire reaction is likely to take place. Furthermore, evaporated elemental sodium catalyses the dehydrogenation of methanol in a homogeneous vapour-phase reaction. With respect to additional investigations carried out lately, a conclusive reaction mechanism is proposed which explains both the reaction with sodium aluminate and evaporated elemental sodium as catalysts in a proper way.

Full text of DOI:10.1016/S1381-1169(99)00088-6

Ž
.
Journal of Molecular Catalysis A: Chemical 146 1999 271–278
www.elsevier.comrlocatermolcata
Homogeneous catalysis by evaporated solids
1
)
Steffen Ruf , Gerhard Emig
Lehrstuhl fur Technische Chemie I, UniÕersitat Erlangen-Nurnberg, Egerlandstr. 3, 91058 Erlangen, Germany
¨
¨
¨
Abstract
About 10 years ago, sodium aluminate has been found as suitable catalyst for the pure dehydrogenation of methanol to
anhydrous formaldehyde. At first, the reaction was supposed to follow a heterogeneous mechanism. Later on, experimental
results with a special set-up were at variance. Recent investigations revealed the loss of catalytically active species from the
solid aluminate into the vapour phase where the entire reaction is likely to take place. Furthermore, evaporated elemental
sodium catalyses the dehydrogenation of methanol in a homogeneous vapour-phase reaction. With respect to additional
investigations carried out lately, a conclusive reaction mechanism is proposed which explains both the reaction with sodium
aluminate and evaporated elemental sodium as catalysts in a proper way. q 1999 Elsevier Science B.V. All rights reserved.
Keywords: Homogeneous vapour phase catalysis; Formaldehyde; Methanol; Sodium
1. Introduction
Unfortunately, both catalysts suffer from deacti-
vation phenomena as soon as the methanol frac-
tion in the feed stream exceeds 10%. Further-
more, the two aluminates reach their activity at
relatively high temperatures, i.e., not until 1150
K.
At these temperatures methanol is also con-
verted homogeneously in absence of any cata-
lyst resulting in remarkable lower selectivities
of formaldehyde than catalytically obtained. The
uncatalysed reaction is therefore superimposed
to the dehydrogenation catalysed by lithium or
sodium aluminate.
Formaldehyde as an aqueous solution is com-
monly produced by the silver catalyst or formox
process. As an alternative route, the pure dehy-
drogenation of methanol leads to anhydrous for-
maldehyde, required for the production of poly-
oxymethylenes. For the last 10 years, we have
intensively studied this endothermic reaction us-
ing lithium–sodium aluminate Li1yx Nax AlO2
and particularly sodium aluminate NaAlO2 as
catalyst yielding formaldehyde, hydrogen and
Ž
.
Ž
.
w x
carbon monoxide 1 :
CH3OH ™ CH2OqH2 D HR ,973 K s92.7 kJrmol
w x
In previous experiments 2 , a special experi-
CH2O ™ COqH2
D HR ,973 K s12,4 kJrmol
mental set-up gave rise to the assumption that
sodium aluminate acts not as heterogeneous cat-
alyst but that it is decomposed during the reac-
tion emitting fragments into the vapour phase.
Thermodynamic calculations identified these
)
Corresponding author. Tel.: q49-9131-8527420; Fax: q49-
9131-8527421; E-mail: emig@tc.uni-erlangen.de
1 Present address: Aventis Research and Technologies, Indus-
triepark Hochst, G 811, D-65926 Frankfurt, Germany. E-mail:
¨
rufs@aventis.com.
w x
catalyst fragments as elemental sodium 2 .
1381-1169r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
Ž
.
PII: S1381-1169 99 00088-6

(
)
272
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
Therefore, we supposed sodium to catalyse the
reaction homogeneously in the vapour phase.
According to additional examinations recently
performed, these catalyst fragments could be
w x
identified by mass spectroscopy 3 . Further-
more, elemental sodium was evaporated in or-
der to scrutinise the thesis whether sodium
catalyses the dehydrogenation of methanol or
w x
not 4 .
For the uncatalysed conversion a reaction
scheme is presented consisting of only 12 ele-
mentary free-radical reactions and 13 different
species. It was derived from kinetic simulation
in order to describe experimental data. After-
wards three reactions considering the influence
of the catalyst were added to this scheme. With
respect to further examinations studying the re-
action mechanism, a catalytic cycle with the
formation of sodium methanolate and its re-
peated decomposition is proposed which is likely
to be valid for both the solid sodium aluminate
catalyst and the evaporated sodium.
Fig. 1. Experimental set-up with two splitted feed streams.
.
Ž
.
1 . A second stream feed stream 2 , alterna-
tively consisting of nitrogen, nitrogenrwater or
nitrogenrhydrogen, was passed through the cat-
Ž
.
alyst layer Li0.5Na0.5AlO2 and mixed with
feed stream 1 afterwards. Varying the composi-
tion of feed stream 2, the degree of methanol
conversion and the yield of formaldehyde
changes dramatically. The reactor temperature
was kept at 1173 K, the mole fraction of
methanol at 10%.
2. Experimental
The reaction was carried out in a corundum
Ž
.
tube reactor 12 mm=8 mm=1000 mm with
a reaction zone centred in the axis of the tube.
The product stream was analysed by on-line gas
chromatography. Further details are given in
In order to study whether sodium catalyses
the dehydrogenation of methanol a tee carrying
a flat crucible with elemental sodium was at-
w
x
Ž
.
Refs. 2,4 .
tached to an empty tube reactor Fig. 2 . A
separate nitrogen stream was passed over the
crucible to transport evaporated sodium to the
reaction zone. The amount of sodium in the
reaction zone was varied by changing the cru-
cible temperature. A second, smaller inner tube
While studying the uncatalysed reaction, an
empty tube was used heating the reaction zone
to 1123 and 1173 K, respectively. The mole
fraction of methanol in the feed stream was set
to 10 or 20% with nitrogen as inert gas for
balance varying the flow velocity between 0.75
and 1.5 mrs. The feed stream was pre-heated to
473 K.
Ž
.
6 mm=4 mm enabled the feed of methanol
diluted with nitrogen to be supplied directly
Ž
.
into the reaction zone without contacting ele-
mental sodium placed in the crucible. Mean-
while, the reactor temperature was kept con-
stantly at 973 K as the mole fraction of methanol
in the feed was held at 10%.
Fig. 1 shows the special experimental set-up
which has led to the assumption that catalyst
fragments leave the solid aluminate. The mix-
ture of methanol and nitrogen is fed to the
reaction zone through an inner tube without
For the additional experiments with the aim
to examine more details of the reaction mecha-
Ž
contacting the aluminate packing feed stream

(
)
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
273
the degree of methanol conversion is increasing
while the selectivity to formaldehyde decreases.
The maximum yield is approximately 19%. Car-
bon monoxide appears as by-product while
methane and water can only be detected in
small traces. At 1173 K reactor temperature,
10% methanol mole fraction in the feed stream
and 1.0 mrs flow velocity, the degree of
methanol conversion and the formaldehyde se-
lectivity are approximately 30 and 48%, respec-
tively.
Looking at the reaction scheme of the un-
catalysed dehydrogenation, the formation of
carbon monoxide is discussed in the literature as
parallel reaction from both formaldehyde and
w
x
Ž
methanol 5,6 in a triangular parallel-consecu-
.
tive scheme. However according to our investi-
gations, the exclusive formation of carbon
monoxide from formaldehyde in a consecutive
reaction and not directly from methanol seems
to be more likely taking elementary reaction
steps into account. Varying the residence time,
the decision, whether the reaction follows a
consecutive reaction or triangular scheme, can
be made by plotting reactant concentrations vs.
residence time. In case of a consecutive reaction
the slope of carbon monoxide vs. time curve
must be zero in the origin showing simultane-
ously a maximum of the formaldehyde curve.
Otherwise, in case of a parallel reaction, the
slopes of both formaldehyde and carbon monox-
ide curves are larger than zero in the origin.
The variation of residence time reveals exper-
imentally that carbon monoxide is formed from
Fig. 2. Experimental set-up for evaporating sodium.
nism, first methanol was substituted by methane
and second sodium as catalyst was substituted
by sodium methanolate.
A
commercial
Ž
.
methanolic solution 30%, Merck was diluted
to 0.05% by mixing with the methanol feed
stream. Changing the direction of flow in the
reactor, it was dripped directly from the top to
the reaction zone realising a methanol mole
fraction of 10% at 973 K reactor temperature.
3. Results
Ž
.
The dehydrogenation of methanol to formal-
dehyde was performed non-catalytically in an
empty tube as well as catalytically with sodium
aluminate as solid and elemental sodium as
vaporised catalyst. Supplementary experiments
gave special insights to the reaction mechanism.
formaldehyde in a consecutive reaction Fig. 3 .
As expected, the mole fraction of formaldehyde
shows a maximum while the slope of the carbon
monoxide curve is zero in the origin and the
curve increases monotonically with longer resi-
dence times. Carbon monoxide is therefore ex-
clusively formed from formaldehyde and not
directly from methanol. The reaction was simu-
lated with a consecutive reaction scheme under
isothermal conditions; the corresponding curves
show good agreement with the experimental
data.
3.1. Uncatalysed reaction
Methanol is converted homogeneously with-
out catalyst as soon as the reaction temperature
exceeds 1073 K. With increasing temperature

(
)
274
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
formaldehyde selectivity of 48%, known as typ-
ical values of the uncatalysed reaction.
The addition of small amounts of water to
feed stream 2 afterwards influenced neither the
degree of methanol conversion nor the selectiv-
ity to formaldehyde. After adding hydrogen to
feed stream 2, the degree of methanol conver-
sion however jumped to 98% while the formal-
dehyde selectivity increased from 48% to ap-
proximately 70%. Interrupting the supply of
hydrogen, a breakdown of methanol conversion
to the value of the uncatalysed reaction of 30%
is observed. Likewise, the selectivity of formal-
dehyde decreased to the initial value of 48%.
The procedure of adding hydrogen and inter-
rupting the supply leads always to the same
result. As long as hydrogen or, alternatively,
methanol as reducing gases were added in small
amounts, a similar conversion and selectivity is
obtained as in the fixed bed or tube wall reactor
with sodium aluminate as catalyst-although, in
this case, the direct contact between methanol
and catalyst is excluded.
Fig. 3. Reactants concentration under variation of residence time.
3.2. Aluminate catalyst
With sodium aluminate as catalyst in both
fixed bed and catalytic tube wall reactor the
maximum formaldehyde yield of 65% is
achieved at nearly total methanol conversion at
1173 K reactor temperature and 10% methanol
mole fraction in the feed stream. Outside this
range at only 50 K lower reactor temperature,
the catalyst loses 60% of its full activity within
only 6 h. Therefore, the catalyst must be kept at
temperatures of 1153 or higher for stationary
operation.
Fig. 4 shows experimental results applying
the special set-up of Fig. 1 with different com-
positions of the second feed stream which passed
through the catalyst bed. During the first 150
min the feed consisted exclusively of nitrogen.
After a short induction period, the degree of
methanol conversion reached 30% with a
In another experiment, the thermal behaviour
of sodium aluminate under inert atmosphere
w x
was examined by mass spectroscopy 3 . Fig. 5
shows a temperature-dependent increase of the
Ž
.
ion stream of the mass mres23 sodium .
Ž
.
Sodium aluminate loses sodium Na, as shown
and simultaneously sodium oxide NaO, not
shown at elevated temperatures by decomposi-
Ž
.
tion.
wŽ .
Fig. 4. Results of set-up with splitted feed streams a 100% N2;
Ž . Ž . Ž .
b 99.3% N2 , 0.7% H2O; c 98.7% N2 , 1.3% H2O; d 87% N2 ,
Ž . Ž .
Fig. 5. Emission of sodium from sodium aluminate plotted vs.
x
w x
13% H2; e 100% N2; f 87% N2 , 13% H2
.
temperature 3 .

(
)
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
275
lower than using sodium aluminate as solid
‘catalyst’.
In order to decide whether sodium is con-
sumed stoichiometrically or acts as catalyst the
turnover number is determined. It is calculated
as ratio of the molar consumption of methanol
to sodium and resulted in values between 500
and 700 at 973 K reactor temperature. With one
sodium atom between 500 and 700 methanol
molecules were converted. The influence of
sodium is therefore definitively a phenomenon
of catalysis, the dehydrogenation of methanol to
formaldehyde is therefore a homogeneously
catalysed vapour-phase reaction.
Fig. 6. Conversion of methanol in dependency on temperature of
Ž
the crucible at a reaction temperature of 973 K 0–150 min: no
.
sodium present, 150–600 min with sodium .
3.3. Elemental sodium as catalyst
3.4. InÕestigations of the reaction mechanism
Fig. 6 depicts the results of the basic experi-
ment which obviously demonstrates the influ-
ence of evaporated sodium on the dehydrogena-
tion reaction at a reaction temperature of 973 K.
During the first 130 min no methanol conver-
sion was detected according to the fact that no
sodium was placed in the crucible. Under these
conditions no homogeneous reaction is taking
place, considering the experimental set-up as
inert. Placing sodium in the crucible and in-
creasing its temperature to 583 K, the degree of
methanol conversion reached 25% with formal-
dehyde as the main product. Hydrogen and car-
bon monoxide were detected as the other reac-
tion products. After lowering the crucible tem-
perature to 423 K, no further methanol conver-
sion was observed. As long as the temperature,
or more precisely the partial pressure of sodium
in the vapour phase is high enough, methanol is
converted within the vapour phase to formal-
dehyde.
Additional experiments in order to elucidate
the reaction mechanism of the sodium catalysed
methanol dehydrogenation were performed first
by substituting sodium by sodium methanolate
and second by the simultaneous supply of
Ž
.
methane instead of methanol and sodium.
Discussing the mechanism, the product
formed in the reaction between methanol and
sodium is of great importance. A special experi-
ment should reveal whether there is a reaction
between sodium and the hydrogen atoms of the
w x
methyl group in methanol or not. Sauer 7
proposed the reaction between methanol and
sodium to sodium hydride and hydroxy-methyl
radicals.
Although the reactor temperature was in-
creased to 1223 K no conversion of methane
was observed. Under these reaction conditions
hydrogen atoms of methane are inactive against
an attack of sodium. Hence, a reaction of sodium
Ž .
The repeated cycles of heating and cooling of
the crucible and the appearance of formal-
dehyde show that sodium is essentially needed
for the dehydrogenation reaction. Only as long
as sodium is present in the vapour phase,
methanol is converted to formaldehyde. The
reaction is likely to take place entirely in the
vapour phase. The conversion of methanol can
be successfully conducted at 973 K, 200 K
and methanol yielding sodium hydride NaH
Ž
.
and hydroxy-methyl radicals CH2OH is un-
likely, without regarding this test as final proof
but as a strong indication. In accordance with
the known reaction in the liquid phase, the
reaction to sodium methanolate and atomic hy-
drogen seems very likely.
Finally, the substitution of sodium by sodium
methanolate as catalyst should reveal the precise

(
)
276
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
mechanistic role of sodium, which is discussed
later. With sodium methanolate as catalyst simi-
lar methanol conversions and turnover numbers
Ž
.
450–800 are achieved under comparable reac-
tion conditions, especially concerning catalyst
Ž
.
concentrations and reactor temperature 973 K .
Fig. 8. Reaction cycle of the catalysed reaction with Na or
NaAlO2 as catalyst.
4. Discussion
4.1. Homogeneous reaction
of the reaction of sodium and methanol is cru-
cial, on the other hand the precise role of sodium
is of great importance. As the methane experi-
ment revealed, there is no reaction between
methane and sodium although the reactor tem-
perature was elevated to 1223 K. Moreover, in
solution under normal conditions the product is
definitively sodium methanolate. Therefore, we
suppose the latter to be formed also in the
vapour phase. The second question refers to the
role of sodium, precisely whether there is a
repeated decomposition of sodium methanolate
during the reaction or not. The latter would
mean that the only function of sodium is to
generate H-atoms which could act as chain car-
rier analogous to the uncatalysed reaction.
The experiments using sodium methanolate
as catalyst show two substantial clarifying re-
sults. First, sodium methanolate is decomposed
under reaction conditions and acts also as cata-
lyst for the pure dehydrogenation. Second, simi-
lar turnover numbers are calculated in compari-
son to elemental sodium as catalyst. The first
result proves sodium methanolate not to be inert
which means that the function of sodium is
definitively not only to produce H-atoms; there
would be no methanol conversion, if sodium
methanolate was inert. The second result shows
that it is irrelevant for the reaction mechanism
whether the catalytic cycle is started by supply-
ing sodium or sodium methanolate as catalyst
The uncatalysed reaction as well as the
sodium-catalysed reaction are taking place ho-
mogeneously in the vapour phase. Kinetic simu-
lations turned out H-atoms to act as chain carri-
ers while at the same time 12 elementary reac-
tions with 13 different species are sufficient to
w x
describe the experimental data 8 The simpli-
fied reaction scheme is shown in Fig. 7.
The uncatalysed methanol decomposition is
induced by homolytic C–O cleavage forming
first methyl and hydroxyl radicals and then by
reacting with methanol to methane, water and
particularly hydroxy-methyl radicals. From these
hydroxy-methyl radicals formaldehyde is formed
which is further decomposed to carbon monox-
ide via formyl radicals by reaction with H-Atoms
or by collision with an inert partner M.
Looking at the sodium-catalysed reaction, the
determination of the turnover number revealed a
catalytic effect of sodium on the dehydrogena-
tion of methanol in the vapour phase. Concer-
ning the reaction mechanism, two questions are
of major interest. On the one hand, the product
Ž
.
Fig. 8 ; offering sodium, the actual catalyst is
supplied, offering sodium methanolate the inter-
mediate is supplied.
The reaction between methanol and sodium
leads to H-atoms and sodium methanolate. The
Fig. 7. Scheme of uncatalysed dehydrogenation reaction.

(
)
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
277
latter decomposes to methoxy radicals and
sodium which is able to react repeatedly with
methanol. Formaldehyde is formed from
methoxy radicals. Unfortunately, H-atoms can
also trigger chain reactions mentioned in the
uncatalysed cycle explaining the occurrence of
carbon monoxide. Obviously, the formaldehyde
selectivity is therefore lower than 100%.
until 1150 K. On the other hand a temperature
of 973 K is enough for the chemical reaction as
the results with elemental sodium show. The
additional difference of 200 K is only required
for the activation of sodium aluminate.
As the experiments with the two splitted feed
Ž
.
streams Fig. 4 revealed, the degree of methanol
conversion could only be increased when reduc-
ing gases were led over the aluminate. On the
other hand, sodium aluminate also emits cata-
lyst fragments under non-reducing conditions.
The divergent results can be explained by the
4.2. Effect of sodium aluminate on methanol
dehydrogenation
Sodium aluminate is decomposed at elevated
Ž
.
design of the experimental set-up Fig. 1 . A
fixed bed of aluminate with a following inert
bed has been used as sodium source. We sup-
Ž
.
temperatures leading to a loss of sodium Na
Ž
.
and sodium oxide NaO and the formation of
alumina:
Ž
.
Ž
.
pose sodium Na and sodium oxide NaO to
recombine within the bed forming gaseous Na2O
before getting in contact with methanol. Na2O
is very likely to be catalytically inert under
reaction conditions. There is neither a thermal
decomposition forming sodium nor a reaction
with methanol to form sodium methanolate and
therefore no enhanced degree of methanol con-
version. Under reducing atmosphere the recom-
bination is probably hindered. Sodium can reach
the reaction zone and catalyses the dehydro-
genation reaction.
2 NaAlO2 ™ NaqNaOqAl2O3
The temperature of 1150 K, when decomposi-
tion starts to be significant, coincides well with
the temperature when sodium aluminate obtains
its activity for the dehydrogenation reaction. At
1123 K, only in the very first beginning of the
reaction enough sodium is emitted into the
vapour phase. Later on, the temperature of the
aluminate is too low to emit sufficient sodium
for total conversion causing the decrease of
catalyst activity within only 6 h. This may be
caused by the slow process of diffusion in solid
state. Furthermore, the plots of the ion stream of
sodium and the degree of methanol conversion
vs. temperature show a parallel behaviour.
5. Conclusions
The catalyst does not act heterogeneously but
emits catalytically active sodium into the vapour
phase. Within the vapour phase, the reaction is
taking place homogeneously catalysed by emit-
ted sodium. Therefore, sodium aluminate is con-
sumed during the reaction and acts only as
reservoir for the catalytically active sodium. In
this case, deactivation of the solid ‘catalyst’ by
loss of active species is not undesirable but
imperative for the activity of sodium aluminate
for the dehydrogenation of methanol.
On a first glance, lithium and sodium alumi-
nate as solid catalysts on the one side and
evaporated sodium on the other side seem to be
two different catalyst systems acting differently.
However, our experimental results revealed
sodium as active species in both cases. The
solid catalyst acts not classically by adsorbing
methanol and lowering thus the activation en-
ergy of the dehydrogenation reaction but by
emitting active fragments into the vapour phase
where the reaction is taking place. The ‘hetero-
geneous’ aluminate catalyst as well as the eva-
porated sodium turned out to catalyse the dehy-
drogenation in a homogeneously catalysed
vapour phase reaction.
The high reactor temperature of 1173 K is
necessary for the production of formaldehyde
when sodium aluminate is used as catalyst be-
cause decomposition of aluminate begins not

(
)
278
S. Ruf, G. EmigrJournal of Molecular Catalysis A: Chemical 146 1999 271–278
The dehydrogenation of methanol follows the
References
same mechanism, independent whether sodium
or sodium aluminate is applied. Sodium
methanolate is repeatedly decomposed forming
sodium and closing the catalytic cycle to form
formaldehyde from methanol.
w x
Ž .
G. Wiesgickl, H.P. Beck, G. Emig, Appl. Catal. 59 1990 L1.
1
w x
2
Ž .
J. Sauer, G. Emig, Chem. Eng. Technol. 18 1995 284.
w x
3
M. Bender, Doctoral Thesis, Universitat Bochum, 1997.
¨
w x
Ž
.
4
S. Ruf, G. Emig, Appl. Catal. A 161 1997 L19.
w x
5
P. Zaza, H. Randall, R. Doepper, A. Renken, Catal. Today 20
Ž
.
1994 325.
w x
Ž .
A. Musie, J. Batista, J. Levec, Appl. Catal. A 165 1997 115.
ˇ `
6
w x
7
J. Sauer, Doctoral Thesis, Universitat Karlsruhe, 1994.
¨
Acknowledgements
w x
8
S. Ruf, Doctoral Thesis, Universitat Erlangen-Nurnberg, 1998.
¨ ¨
The work was supported by the BMBF
Ž
.
Bundesministerium fur Bildung und Forschung
under the number 03D0012B9.
¨