248 J. Keijsper et al.

/ “Green” MMA;an environmentally benign and economically attractive process

Recl. Trav. Chim. Pays-Bas 115,248-255 (1996) 0165-0513/96/04248-08$12.00

SSDI 0 1 6 5 - 05 1 3 (96 ) O 000 2 - 6

“Green” MMA, an environmentally benign and economically attractive process

for the production of methyl methacrylate

J. Keusper, P. Amoldy, M.J. Doyle, E. Drent

Koninklijke /Shell Laboratium (Shell Research B. K), P.O. Box 38000, 1030 BN Amsterdam, The
Netherlands
(Received October 6, 1995)

Abstract. Over the last seven years Shell has developed a novel, economically attractive, and
environmentally benign process for the production of methyl methacrylate (MMA). The process is
based on the highly efficient Palladium-catalysed methoxycarbonylation of propyne (methyl-
acetylene or MA) under mild, non-corrosive reaction conditions (temperature ca. 50°C, C O
pressure ca. 10 bar) according to the equation:

MeOH + CO + HCIC-Me( MA) --* H,C=C( Me)-C( O)OMe( MMA)

The activity (> 20 000 mol-MMA/mol-Pd.h) and selectivity ( > 99%mol) of the homogeneous Pd
catalyst are very high making a once-through process with minimal waste possible.
The MA feedstock for the process, together with propadiene (PD), is separated from a steam-
cracker C, stream by extractive distillation with DMF. Subsequently, P D is removed from this
MA + PD/DMF mixture and isomerized over a heterogeneous K,O/Al,O, catalyst to yield
additional MA.
Crude MMA obtained from the methoxycarbonylation of MA can easily be purified by a ‘topping
and tailing’ procedure to yield MMA of very high purity (> 99.99 w%).
This article describes some of the R& D work carried out at the ‘Koninklijke/Shell-Laboratorium,
Amsterdam’ which has enabled the design of a plant, based on this novel technology.

1. Introduction a
Nowadays, in addition to economic motives, environmen-
tal attractiveness is also a major driving force for techno-
logical innovation in the chemical industry. As a conse-
a Nomenclature and abbreviations quence, important incentives are: cheap feedstocks, sim-
ple processes, waste minimization, high feedstock utiliza-
acrylic acid = propenoic acid
ACH process = acetone-cyanohydrine process tion and low utility consumption. In this article, a novel
crotonic acid = (EI-but-2-enoic acid route of production of methyl methacrylate (MMA) is
CSTR = continuously stirred tank reactor described which fully meets all of the above criteria.
DMF = N,N-dimethylformamide MMA is a large-volume industrial monomer [ca. 1800
FID = flame-ionization detection kilo-ton per annum; (kta)] finding major outlets in ho-
FPU = feed-preparation unit mopolymer poly-MMA (main application: artificial glass)
GLC = gas-liquid chromatography
HE stream = heavy-end stream and in a variety of copolymers’. The two most important
isobutene = methylpropene production processes for MMA are:
kta = kilo-ton per annum = 1000000 kg/year (1) The acetone-cyanohydrin (ACH) route, accounting
LE stream = light-end stream for ca. 80% of worldwide MMA production and oper-
LHSV = liquid hourly space velocity (in l/l/h) ated by many companies of which ICI is the largest.
MA = methylacetyleen = propyne Starting from acetone and HCN, a base-catalysed
MAA = methacrylic acid = 2-methyl propenoic acid
MAP D= MA+ PD reaction yields the intermediate ACH, which in a
MEHQ = methylether of hydroquinone = 4-methoxyphenol subsequent step, using concentrated sulfuric acid, is
methacrolein = 2-methylpropenal converted to 2-hydroxy-2-methyl propionamide and its
MMA = methyl methacrylate = methyl 2-methylpropenoate sulfate. A final acid-catalysed esterification with
MSA = methanesulfonic acid methanol (MeOH) yields MMA. This process, the
MTBE = methyl tert-butyl ether = 2-methoxy-2-methylpropane origin of which dates from the 1930s, suffers from
PD = propadiene
PN = (pyrid-2-y1)diphenylphosphine environmental drawbacks such as HCN handling,
tetrene = welding gas containing ca. 36 w% of MAPD acidic waste (usually > 1.5 t/t MMA) and/or capital-
wppm = weight ppm and energy-intensive acid recovery and regeneration.
Recueil des Travaux Chimiques des Pays-Bas, 115/04, April 1996
(2) The isobutene- o r tert-butanol-oxidation route was
commercialized by Mitsubishi-Rayon in Japan in the
1980s. In this process the C, feedstock is catalytically
oxidized by air in a two-stage vapour phase reaction
yielding methacrylic acid (MAA) which is esterified
with MeOH to yield MMA. Drawbacks of this process
are the low conversion per pass, and the limited
availibility and high cost of the feedstock which has an
alternative outlet in the gasoline pool, e.g., as MTBE.
In addition to these two major processes, BASF has
commercialized a production route to MMA based on the
homogeneously rhodium-catalysed hydroformylation of
ethene, yielding propanal. This in turn is converted to
methacrolein using formaldehyde. Subsequent heteroge-
neously catalysed oxidation to h4AA followed by esterifi-
cation with MeOH yields MMA. Moderate yields ( < 85
mol% based on propanal) and limited availability of cheap
propanal are the major drawbacks of this process. The
different routes are schematically presented in Figure 1.
This paper describes a novel production process for MMA
based on the homogeneous, palladium-catalysed methoxy-
carbonylation of propyne (or methyl acetylene, MA) ac-
cording to the following equation (see Figure 1):
MeOH + CO + H C I C - M e ( MA)
-,H2C=C( M e ) - C ( O)OMe(MMA)
Examples will be given of the R & D work carried out at
the Shell Laboratory in Amsterdam, which has led to the
present process description. More fundamental aspects of
the palladium-catalysed methoxycarbonylation of propyne
have already been reported2, and will be mentioned only
briefly here.
2. Experimental
2.1. Feedstocks
Appropriate feedstocks for the isomerization were obtained from Air
Liquide and Fina-Borealis (sold as welding gas or tetrene) and
contained ca. 36 w% of MAPD. From Shell naphtha crackers, more
representative feedstocks were obtained. A typical composition of
the feed as used in Figure 2 is (in w%): ethane < 0.1, MA 22.2, PD
14.1, other C,s 54.3, C,s 9.4. Before use, trace impurities such as
water were removed by a zeolitic pre-bed.
Figure I . Exkting and novel routes to methylmethacrylate (MMA). A ) ACH B) Isobutene oxidation C) BASF D ) Novel propyne route.
249
CO was obtained from Air Products and was purified before use on
active carbon and BTS pre-beds.
Propyne (MA) was obtained from Scott Speciality Gasses and puri-
fied by a 'topping and tailing' procedure before use to remove traces
of propadiene and oligomeric components.
Methanol (MeOH) p.a. was obtained from Aldrich; commercial
MeOH from MCN.
Catalyst ingredients: Pd(OAc),, 4-methoxyphenol (MEHQ), and
methanesulfonic acid (MSA) were obtained from Aldrich; the (pyrid-
2-y1)diphenylphosphine ligand (PN ligand) was synthesized following
procedures developed in house3. Pre-commercial batches were ob-
tained from Technochemie and other toll processors.
2.2. lsomerization
Isomerization experiments in the liquid phase (PD MA) were
conducted between -20 and +2WC and at 10-15 bar, in a Hoke
microflow reactor (50 ml) with an outside oven attached for in-silu
activation and regeneration of the catalyst extrudates. The heteroge-
neous K,O/AI,O, catalysts were prepared by standard 'dry' im-
pregnation techniques of K,CO, on y-alumina, then sequentially
dried and calcined according to known procedures4. Next, 7.5 g was
mixed with 52.5 g of SIC followed by 18-h in-situ activation under an
inert atmosphere. Using the MAPD-containing tetrene feedstock,
catalyst activity was determined at high space velocity (LHSV = 30
I/l-catalyst/h), whereas catalyst stability experiments were carried
out at low LHSV (ca. 2 l/l/h). Conversion and selectivity were
measured using standard GLC techniques (PLOT fused silica col-
umn, coated with AI,O,/KCI, 50 mX0.32 mm ex Chrompack;
injection T = lOWC, carrier gas He, split ratio 1:100, T program:
40-175°C; FID detection at 190°C).
2.3. Carbonylation of propyne
Propyne was carbonylated in a continuously stirred tank reactor
(CSTR, 220 ml liquid volume) at 30-80°C and pressurized by a
constant supply of CO. A low-partial-pressure CO experiment at the
same overall pressure was conducted by introducing a N2 /CO
mixture. Liquid propyne was fed to the CSTR via two ISCO-pumps
and the MeOH/catalyst solution was continuously fed via an HPLC
pump. The gas-cap pressure was kept constant by a constant pres-
sure valve and a constant liquid level was maintained. A hollow shaft
stirrer was used to ensure adequate gas/liquid mixing. The catalyst
was prepared by adding under an argon atmosphere appropriate
amounts of MSA, MEHQ, the PN ligand, and Pd(OAc), to the
MeOH. Conversion and selectivity were derived from the feed rate
of propyne and the production rate of MMA plus methyl crotonate
as measured in the liquid product using off-line GLC techniques (DB
wax column, 60 m, 0.324 mm, 0.5 pm, ex Chrompack; carrier gas He,
split ratio 1:100,injection T 150"C, T program 45-210"C, FID at
225°C).
250 J. Keijsper et al. / "Green" MMA; an environmentally benign and economically attractive process

100
LHSV = 3 0 L A / h

[MA+PDI in feed : 36.3 %w

40

20

I I
0 ' I I

0 300 600 900 1200 1500

--> mo/ MA+PD processed per mol K

Figure 2. PD --t MA komerization on activated 20 w% K,CO, /alumina.

2.4. Continuous distillation of the crude MMA mixture obtained during
carbonylation
This process was carried out off-line using two tray columns
(Oldershaw) operated in series (20 and 40 trays, respectively) at
slightly reduced pressure. The temperature in both reboilers was
90-100°C; reflux ratios were varied between 1.5 and 4; feed streams
were varied between 100-150 g/h. The bottom stream of the first
stripping column was directly fed to the second, rectifying column
from which pure MMA was obtained as a top stream.
3. Results and discussion
3.I . Zsornerization of propadiene to propyne
From the l i t e r a t ~ r e ~
it .is~ known that many basic as well
as acidic heterogeneous catalysts promote the isomeriza-
tion of propadiene (PD) to propyne (MA). Since the
reaction is equilibrium-limited and since at higher tem-
peratures this equilibrium shifts towards unwanted PD, a
preferred catalyst should be active at low temperatures
( e . g . , at 20°C where the equilibrium constant [MA]/[PD]
for 40 w% [MA + PD] in a liquid C, mixture is approxi-
mately 10). After having screened various catalysts, acti-
vated K2C03 on y-alumina has been targeted for further
process development since this catalyst combines a satis-
factory activity at low temperature (between -20 and
+2o"C) with a suitable stability. This is indicated in
Figure 2, showing for a typical accelerated life-test experi-
ment the rate constant k (in l / h ) of the forward reaction
PD -+ MA as a function of the amount of M A + PD
processed for a catalyst derived from 20 w% K,CO,/y-
Al,0;.
The initial activity, expressed as the rate constant k (ca.
80 l / h in this experiment), of the catalyst and its stability
(estimated life 4000 mol(MA + PD).mol-'K for this ex-
periment) are both important design parameters deter-
mining, e.g., the amount of capital and catalyst needed.
Further research has shown that the apparent deactiva-
tion is most probably due to trace impurities in the feed

100
LHSV = 30 LAIh: T = 20 degl

[MA+PDI In feed 36.3 %w

0
a0 average cfeclrne = 3 % per cycl

60

40

20

0
0 2 4 6 a 10

- -> Number of Regenerations

Figure 3. PD --t MA isomerization; effect of regeneration on activity.
Recueil des Travaux Chimiques des Pays-Bas, 1 IS / 04, April 1996
used and not to either thermal ageing o r to main compo-
nents in the feed. This is because the rate of deactivation
is not affected by the time on stream but is only a function
of the amount of feed processed. Figure 3 shows that even
after repetitive runs with concurrent deactivation, the
catalytic activity is almost fully restored by calcination
each time, thus vastly extending the useful life of the
catalyst.
Based on the above experience and similar results ob-
tained with various MAPD-containing feeds, plus addi-
tional work comprising modelling, optimization, and pro-
cess engineering, a complete process description for the
isomerization part of the novel MMA process can be
given (uide infra).
3.2. Carbonylation: MA + CO + MeOH + MMA
Some relevant carbonylation results obtained in our CSTR
are collated in Table I; unless otherwise mentioned, the
selectivity to MMA is around 99.2 mol%, the remaining
part of propyne converted yielding almost completely
methyl crotonate’. In Ref. 2, a detailed mechanism for
the carbonylation is proposed, highlighting the roles of
the various catalyst components in the catalytic cycle.
During the present set of experiments, the rate of stirring
has not affected catalyst performance, pointing to the
absence of gas/liquid-mass-transfer limitations.
The conversion as measured in experiment 1 corresponds
to spectacular MMA-formation rates: at residence times
of I f and 43 h, > 40000 and > 20000 mol/mol-Pd/h,
respectively. Assuming first-order kinetics in MeOH and
Pd, the conversions at these different residence times
indicate a kinetic order in propyne close to 1, as reported
earlier’. However, at longer residence time (see e.g., exp.
2) catalyst performance is slightly affected and the conver-
sion starts to drop. We attribute this behaviour to PN
ligand being consumed during the reaction to below a
critical level. This is in line with other experiments (e.g., 1
and 3 in Table I) from which it emerges that, under the
present conditions, excess PN ligand with respect to Pd is
required in the catalyst solution for superior catalyst per-
formance. At too low PN/Pd molar ratios (exp. 3), con-
version levels are adversely affected. This excess PN lig-
and is probably needed for various reasons such as: (i) the
PN ligand stabilising the active Pd-catalyst through a
complexation equilibrium; (ii) PN being required as a
proton messenger in the final protonation step of the
catalytic cycle’; (iii; uide supra) part of the PN inventory
being consumed during the reaction, as indeed has been
the general observation. Note that at a low PN/Pd molar
ratio (exp. 3) the conversions at If h and 4; h residence
Table I MA carbonylation; the effect of various variables on catalyst Performance.
Exp. Conditions a MA conversion
no. at residence
time I$ h
(% mol)
1 standard 83
2 residence time 7.9 h - -
3 cat. ratio = 1/15/10 43
4 cat. ratio = 1/80/60 78
5 exp. 4 & T = 75°C 95
6 p ( C 0 ) = 1.4 bar - 71
7 p(C0) = 40 bar 72
8 [PD] = 215 ppmw 18
9 IH,Ol= 1.2 w% 74
10 5v%H,inCO 78
25 1
time are almost equal, which is in line with the above
reasoning. Probable PN-ligand consumption reactions are
oxidation and quarternization, which are discussed below.
Phosphine ligands are easily oxidized by various inorganic
and organic sources especially in solution, yielding phos-
phine oxides. Since Pd is a known oxidation catalyst, these
reactions may even be catalysed by the present carbonyla-
tion catalyst. Because the resulting oxides are poorly
complexating ligands, it is clear that oxidation of the
present PN ligand may in principle affect catalyst perfor-
mance. However, since the concentration of PN oxide in
the reactor effluent is negligible (< 5 mol% conversion to
PN oxide on PN intake), it is concluded that PN oxidation
in the present set of experiments has not affected catalyst
performance.
A second potential consumption reaction for the PN
ligand is a quaternization reaction, e.g., according to the
following Michael-addition type of reaction, yielding a
phosphonium salt:
PN ligand MKA
OMe
I
Phosphonium salt
Because the above phosphonium salt is indeed detected in
the effluent of the reactor, we suggest that quaternization
reactions, like the above, are responsible for the PN-ligand
consumption obsemed during the reaction. Further stud-
ies have revealed that the rate of these quaternization
reactions depends on, amongst others, the temperature
and acid concentration and are Pd-catalysed.
Interestingly, very high PN/Pd molar ratios (see exp. 4)
also affect catalyst performance negatively. This is most
probably due to inhibition of the active catalyst by the
strong coordination of the PN ligand to sites required in
the catalytic cycle’. From experiment 5, it can be seen
that for such high PN/Pd catalyst systems, superior activ-
ity can be restored by raising the temperature. From this
and analogous experiments, however, it has become clear
that increasing the temperature slightly affects the selec-
MA conversion Remarks
at residence
time 45 h
(% mol)
92 < 400 wppm MAA in product
MA conversion 88%mol
41
92
99 selectivity = 98.7 %mol
overall pressure = 10 bar
86
45
95 0.9 w% MAA in product
96
252 J. Keijsper el al. / “Green” MMA; an environmentally benign and economicalty attractive process

MMA MeOH Methyl- Acid Water polyMMA Pd

(W%) (wppm) crotonate (wpprn) (wpprn) (wppm) (wpprn)
(wppm)
FEED 84.2 14.3 w% 0.9 w% 525 540 940 9
MMA product > 99.99 < 30 < 20 < 10 < 200 n.d. a n.d. ‘
Typical
industry > 99.9 < 30 < 20 < 200 < 200 <I absent
specification

a n.d. not detected.

tivity. The results of these and other studies on the effect
of PN-ligand concentration on catalyst performance have
been important for the economic fine-tuning of the final
process design, since catalyst costs and investment capital
are a complex function of the carbonylation temperature.
As may be expected, the reaction rate is CO-pressure
dependent at a low C O pressure ( < 2 bar; cf. exps. 1, 6,
7). At a very high CO concentration, a slight inhibiting
effect of C O is indicated, probably due to C O coordinat-
ing too strongly to catalytic sites necessary in the catalytic
cycle2, i.e., analogously to PN-ligand inhibition discussed
above.
The effect of some feedstock impurities on catalyst per-
formance is shown in Table I (exps 8, 9, 10). Evidently,
propadiene (PD) inhibits the present catalyst. This may be
due to PD reacting with the catalyst, or a precursor, to
form a relatively inert complex, e . g . , comprising a stable
Pd-ally1 moiety. The sensitivity of the catalyst towards PD
is an important design parameter since the feed prepara-
tion unit (uide infra) should deliver propyne of sufficient
purity to the carbonylation reactor. However, not only
design measures can be taken to deal with this P D sensi-
tivity; the catalyst can also be slightly modified, yielding
an increased P D tolerance.
As exemplified by experiments 9 and 10, the catalyst is
very robust towards many other potential impurities. Wa-
ter in the methanolic catalyst solution reacts to yield
methacrylic acid (MAA) without affecting the catalyst;
rates and selectivities for the main propyne methoxycar-
bonylation reaction remain very similar. Hydrogen in the
C O feed acts as an inert component, affecting neither
catalyst performance nor by-products. Both findings result
in ‘relaxed’ design feedstock specifications for these com-
ponents which, obviously, is economically attractive.
The main conclusions from the carbonylation part are
that the activity and selectivity of the catalyst is high
under a broad range of conditions. Inhibition may occur
at too high CO, PN, or P D levels, while during a run the
activity is slightly affected by PN-ligand consumption re-
actions. However, the catalyst shows a surprising stability
towards many potential feedstock contaminants.
3.3. Distillation: crude MMA + purified MMA
In a 2-1 CSTR bench-scale unit, 60 kg of crude MMA has
been produced in a series of carbonylation experiments
analogous to those described above. This MMA, still
containing catalyst residues, has been processed in a con

hrncrisation c4 rrm0v.l Abr0l-k PD Slrippcr MA slrippcr

i
i
I
-4 i,
4

T
PP S p l i m bonocn product
‘Y
i
‘3-
Figure 4. Schematic overview of feed preparation unit (FPU).
L
7-
f

Solvm1
I
 Recueil des T r a v a u Chimiques des Pays-Bas, 115/04, April 1996
tinuous, two-step distillation unit for ‘topping and tailing’,
respectively. The composition of the feed and of the final,
purified MMA product coming from the top of the second
column, is given in Table 11. MeOH present in the feed
has been recovered as top stream of the first (or stripping)
column as a MeOH/MMA azeotrope. Methyl crotonate
and catalyst residues present in the feed have been recov-
ered from the bottom stream of the second (or rectifying)
column. From Table 11, it is evident that a very high purity
MMA (> 99.99 w%; cf. standard industrial specification:
> 99.9 w%) can easily be obtained from a relatively
simple distillation procedure. This is a direct consequence
of the high selectivity of the carbonylation reaction. More-
over, it is clear that the purity is not affected by the
presence in the feed of catalyst residues originating from
the carbonylation. When comparing the purity of MMA
as prepared according to our novel route against typical
industrial specifications it should be realized that trace
impurities may be present in the latter ( e . g . , methyl
propionate, methyl isobutyrate, methyl hydroxyisobu-
tyrate) which do not show up in MMA prepared via the
propyne route.
4. Process description
The above results, together with extensive additional work
(engineering, safety design, basic data collection, mod-
elling, corrosion studies, etc.) have made i t possible to
describe a commercial production process for MMA,
based on Shell’s carbonylation technology lo. Some rele-
vant elements are given below.
3
I
I Figure
methanol MA
5. Schematic overview of MA carbonyiation
co
and MMA purification.
253
4.1. MAPD production
C, acetylenes (MAPD) are by-products formed in ethene
or steam crackers where feedstocks heavier than ethane
are thermally cracked. As a rule of thumb, these crackers
produce enough MAPD to make 5 kta of MMA for every
100 kta ethene produced. At present, most of this MAPD
is selectively hydrogenated over a heterogeneous catalyst
to yield propene/propane. At some locations part of this
MAPD containing stream is being concentrated to ca. 40
mol% and sold as welding gas (tetrene).
To recover MAPD from a steam cracker as a feedstock
for the Shell MMA process, the stream containing the
MAPD has to by-pass the hydrogenation reactor (see
above). Next, the easily recoverable bottom product of a
downstream propene/propane fractionating column (i.e.,
a ‘PP-splitter’ present at many cracker locations), contains
MAPD in a 20-50 mole% concentration.
Although no other efficient production routes towards
MAPD exist commercially, in principle, the pyrolysis of
various chemicals (propene, isobutene, acetone) may also
be used to yield MAPD.
4.2. Feed-preparation unit (FPU; see Figure 4 )
Objectives for the FPU are: ( i ) To provide MA of high
purity to the carbonylation section, PD especially has to
be removed from the feed since it drastically reduces the
activity of the Pd-catalyst (uide supra). ( i i ) To isomerize
PD to MA, since this improves overall feedstock utilisa-
tion. (iii) To remove propane/propene from the feed in
order to minimize bleed rates in the carbonylation sec-
tion.
I,
1
HE MMA
I
LE
254 J. Keijsper et al. / “Green” MMA; an environmentally benign and economically attractive process
In the resulting design, these objectives are met by an
extractive distillation, i.e., a procedure very similar to
commercial practices for acetylene and butadiene recov-
ery. Thus the MAPD-containing stream is combined with
the isomerization product and tailed in a C:-removal
column. In a next column, MAPD is adsorbed using an
extraction solvent like DMF. The loaded solvent is sent to
a column (PD stripper) where essentially all non-MA
material is selectively stripped from the solvent and sent
to the isomerization unit. This enables, in a second strip-
ping column (MA stripper), the production of pure MA as
a feedstock for the carbonylation. Finally, the top stream
of the PD stripper, being a mixture of predominantly
propane, propene, MA, and PD, is isomerized over fixed
bed isomerization reactors operated in a swing or merry-
go-round mode, thus enabling in-situ activation and re-
generation of the catalyst. Since the heat of the isomeriza-
tion reaction is only small (6 kJ/mol PD)6 the reactor can
be operated adiabatically. Furthermore, a relatively low
space velocity of 2 l/l/h is applied so that almost full
equilibrium is reached up to the start of regeneration.
4.3. M A carbonylation (see Figure 5 )
The methoxycarbonylation of MA is the heart of the
process:
MeOH + CO + H C X - M e ( M A )
+ H,C=C( Me)-C( O)OMe(MMA)
The reaction takes place in one or more continuously
stirred tank reactors operating at high conversion (low
MA concentration) with an excess of MeOH to enhance
the rate. The heat of reaction (200 Id/mole MA) is
removed in an external cooling loop which also enables
the use of a jet stirrer. This type of stirrer combines good
liquid mixing with optimum gas/liquid-mass-transfer
properties. T h e catalyst contains an excess of PN ligand in
order to stabilize the Pd catalyst complex and to make up
for PN-ligand consumption during the reaction (vide
supra). Because of its high activity and its robustness
against many impurities, the concentration of the Pd
catalyst in the reactor can be kept very low (5-50 ppmw).
Before purification, the crude reactor product is stripped
with fresh CO thus enabling recycling of unconverted
MA. Overall MA conversion is > 95 mole%.
4.4. MMA distillation (see Figure 5)
The detailed design of the distillation section is based on
experimental work as described above, in combination
with extensive spiking, modelling, and optimization stud-
ies. Thus, the reactor product is sent to two vacuum-distil-
lation columns for ‘topping and tailing’, respectively. The
light-end stream (LE) of the first column, being mainly a
MeOH/MMA azeotrope, is recycled to the reactor with a
small bleed to prevent build-up of minor by-products.
Finally, in the second column, methyl crotonate,
methacrylic acid, and the catalyst residues (heavy ends,
HE) are removed from the crude MMA and very high
purity MMA ( > 99.99 mol%) is obtained. During distilla-
tion radical-induced poly-MMA formation can be con-
trolled by carrying out the distillation in the presence of a
phenolic polymerization inhibitor and molecular oxygen.
4.5. Safety
Analogous to C, acetylene, C, acetylenes (MAPD) are
susceptible to exothermic decomposition, although the
restrictions with respect to operating conditions are less
severe. Critical factors are pressure, temperature, and
MAPD concentration of the C, mixture. To develop an
intrinsically safe design, Shell has cooperated with both
industry and governmental bodies knowledgeable in this
field. The resulting design with respect to those process
parts where MAPD is handled is based on the following
rules: ( i ) MAPD-containing streams will be maintained
below their ‘Stability-Limit Pressure’, i.e., the maximum
pressure at which a decomposition of the mixture can not
be initiated by a given ignition source. ( i i ) Equipment
containing high MAPD concentrations is designed to
withstand the pressure rise of a decomposition (deflagra-
tion). ( i i i ) Guidelines f o i T , acetylene handling will be
followed.
4.6. Waste
Compared to commercial MMA processes, Shell’s MMA
process can be considered very environmentally benign as
is probably best illustrated by the low amount of waste
produced (less than 5 w% on MMA; i.e., including uncon-
verted propyne and non-recovered MMA). This is a direct
consequence of the high activity and selectivity of the
homogeneous Pd catalyst applied in the carbonylation
section making a once-through, relatively simple, process
feasible. Moreover, in contrast to, e . g . , the ACH process,
the waste (i.e., heavy ends from the feed preparation unit
and MMA distillation column plus light ends and bleeds
from the carbonylation reactor, MeOH/MMA azeotropic
recycle, and distillation columns) is of organic nature so
that its energy content can be recovered by combustion.
4.7. Economic evaluation
In recent SRI and Chem-Systems reports’ the calculated
manufacturing costs for the various MMA technologies
have been compared on an equal basis; These costs in-
clude capital (fixed and variable) elements. From this
report, the following order of manufacturing costs for the
three processes ACH, isobutene, and propyne (see Intro-
duction) emerges: propyne < isobutene < ACH.
Although sometimes different premises have been used,
an analogous in-house evaluation has yielded very similar
results (see Figure 6), highlighting the fact that an envi-
ronmentally benign or ‘green’ process can very well be
economically competitive.
5. Conclusions
A novel, low-cost production route for methyl methacry-
late (MMA) has been described. Feedstocks for this pro-
cess are CO, MeOH, and an MAPD-containing C, stream
2 1.00
h
3
1.60
$j 1.00
e
$ 0.60
0.00
ACH lsobutene Propyne
-> process /see introductionl.
Figure 6. Comparison of M M A manufacturing costs.
Recueil des Trauau Chimiques des Pays-Bas, 115/04, April 1996
which is readily available from many ethene crackers,
thereby upgrading their output. Conversion to MMA takes
place in only three, relatively simple units. In the feed
preparation unit, MA is separated from the C, stream
and the PD is isomerized over a heterogeneous, basic
catalyst to MA. In the reaction section, methoxycarbony-
lation of MA yields crude MMA. In the final purification
section, pure MMA (> 99.99 mol%) is obtained. The
major break-through, making this process viable, has been
the Pd-based homogeneous carbonylation catalyst. Its high
activity (> 20,000 mole MMA per mole Pd per h) and
selectivity ( > 99 mol%) makes a once-through operation
with minimal waste production possible. This environmen-
tally benign process is now ready for commercialization.
Acknowledgements
The authors thank W.W. Jager, E. Kragtwijk, D.H.L.
Pello, A.A.J. Geeraert, A.P.M. Kremers, A . Nahuysen,
P.J.A.M. Cilray and F.C. Sfruyt for their skillful technical
assistance. Ir. P.B. de Blank, Ir. D.C. uan der Beek, Ir.
H.J. Scheffer, Ir. J. uan Gogh, Dr. J.A.M. uan Broekhouen,
and Drs. T.J. Simons are thanked for making numerous
valuable contributions to this work. Support given by Dr.
J.M. uan der Eijk, Dr. T.A.B.M. Bolsman, Dr. N. McFar-
lane, Dr. J.C. uan Rauenswaaij Claasen, Dr. P. Koelewijn,
Drs. J. Hengeueld, C. Smith, and Ir. I. Lindhout through-
out the years has been highly appreciated.
255
References
I SRI-PEP report No. 11D: “Methacrylic Acid and Esters” (1993).
Chem-Systems PERPreport 90-10: “Methacrylic Acid/Methacry-
lates” (1991).
2 E. Drent, P. Arnoidy, and P.H.M. Budze[aar, J. Organomet. Chem.
475, 57 (1994). E. Drent, W.W. Jager, 1. Ketjsper, F.J. Niele,
“Applied Homogeneous Catalysis by Organornetallic Complexes”,
Eds. B. Cornils and W.A. Herrmann; VCH; in preparation.
E. Drent and P.H.M. Budzelaar, Eur. Pat. Appl. EP-A-386834
(1990).
J. Hengeveld and P.B. de Blank, Eur. Pat. Appl. EP-A-0539628
(1993).
See e.g., J.F Cordes and H. Guenzler, Chem. Ber. 92, 1055 (1958);
C.C. Chang and R.J. Kokes, J. Catal. 28, 92 (1973) and J. Am.
Chem. SOC.92, 7517 (1970); L.I. Lafer, Kinet. and Catal., 15, 1423
(1974); F.D. Zeiseler and G. Zimmerman Patent DD-112975
(1975); J.H. Parmentier, J. Catal. 22, 213 (1971); W.N. Smith,
Patent Appl. US-3671605 (1972).
J.H. Kieffer, S.S. Kumaran, and P.S. Mudipalli, Chem. Phys. Lett.
224, 51 (1994) and refs therein.
The rate constant k (l/h) has been derived from: k = LHSV.
( K / ( K + l)).ln((X,-X)/X,-Xi)]; where LHSV is liquid hourly
space velocity (I-feed/l-catalyst/h), K is equilibrium constant
[MA]/[PD], X , is equilibrium conversion, Xi is conversion at
inlet, X = [MA]/([MA] + [PD]); see Ref. 8.
0. Leuenspiel, “Chemical Reaction Engineering”, J. Wiley & Sons
inc. 1972.
E. Drent, W.W. Jager, J.C.L.J. Suykerbuyk, Pat. Appl. W O
95/05357 (1995).
I0
See various patents granted to Shell: Eur. Pat. Applications:
EP-A-533291 (19931, EP-A-571044 (19931, EP-A-521578 (1993),
EP-A-552846 (1993).