Synthesis of Dimethyl Carbonate with CuClz by Gregg Andrew Logan A thesis submitted to the department of Chernical Engineering in conformity with the requirernents for the degree of Master of Science (Engineering) Queen's University Kingston, Ontario, Canada July, 1997 Copyright @ Gregg Andrew Logan, 1997
157
Embed
Synthesis of Dimethyl Carbonate with CuClz - nlc-bnc.ca · Synthesis of Dimethyl Carbonate with CuClz by ... the operating parameters of the experimental apparatus to examine the
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Synthesis of Dimethyl Carbonate with CuClz
by
Gregg Andrew Logan
A thesis submitted to the department of Chernical Engineering in conformity with the requirernents for the degree
of Master of Science (Engineering)
Queen's University Kingston, Ontario, Canada
July, 1997
Copyright @ Gregg Andrew Logan, 1997
National Library I*I of Canada Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliographie Services services bibliographiques
395 Wellington Street 395, rue Wellington OttawaON KIAON4 OttawaON K1A ON4 Canada Canada
The author has granted a non- exclusive licence aUowing the National Library of Canada to reproduce, ban, distriiute or sell copies of this thesis in microform, paper or electronic formats.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts kom it may be printed or othewise reproduced without the author's permission.
L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la fome de microfiche/film, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent ê e imprimés ou autrement reproduits sans son autorisation.
Abstract
The synthesis of dimethyl carbonate (DMC) with cupnc chloride based catalysts was investigated in a 300 ml ParrTM batch reactor. DMC was synthesized via the direct oxidative carbonylation of carbon monoxide with oxygea in methanol. Ex- periments were first conducted with cupric chloride exchangecl onto &ous zeolites. Unfortunately, batch reactor runs using t hese cat alysts were unsuccessfd as insignifi- cant quantities of DMC were synthesized. In the few cases that DMC was synthesized, the results could not be reproduced. Attempts were then made to synthesize DMC with a combined PdC12/CuC12 catalytic system- Runç involving this system were unsuccessful as DMC was not synthesized under various reaction conditions and sig- nifiant quantities of palladium oxide, PdO, covered the wetted intemal surfaces of the reactor and blocked the liquid sampling lines. FolIowing these unsuccessN nuis,
the synthesis of DMC was continued with CuCll as the sole catalyst. Three sets of experiments were successfully designed and implemented within
the operating parameters of the experimental apparatus to examine the effects on the synthesis of DMC. The first set of experiments examined the effect of catalyst concentration , the second set examined the effect of temperature, and the third set was a 23 factorial design examining the effects of catalyst concentration, temperature and ratio of C0/o2.
The results of the experiments indicated that the greatest quantities of DMC were produced with a temperature of 125"C, a CUCI* concentration of 47 mmol/L and a CO/Oz ratio of 2. These operating conditions produced the highest concentration of DMC (0.31 mol/L) and also helped to minimize the CO2 formation. Significantly Lower quantities of DMC were produced at 175OC when compared to the other tem- perat ures.
Gibbs reactor simulations confirmed and further helped to explain the previous experimental results. Between 120 and 160°C, temperature had a negligible effect on the production of DMC. This lead to the conclusion that differences observed in previous experiments at 125 and 150°C can be attributed to the catalyst, which was not considered in the simulations. The simulations did indicate that DMC was not t hermociynamically favoured a t temperat ures above 170°C. This decrease in DMC formation was evident in runs performed at 175°C. The quantities of DMC produced increased with higher ratios of CO/02; indicating the importance of significant CO partial pressure in the synthesis.
Acknowledgments
With the completion of this thesis, my time here a t Queen's has finally corne to an end. It has truly been one of the greatest rides of my Me. The experiences and fnends 1 have gained on this journey are invaluable and I will always remember them fondly.
Firstly, 1 would like to thank my supervisors Dr. Tom Harris and Barrie Jackson for their guidance and support throughout the completion of this project. I would also like to thank Steve Hodgson, Lisa Pnor and Martin York for their patience and technical assistance as 1 struggled with my research equipment.
1 am indebted to m y office mates throughout the years. In particular, Neil Milier and Chris Seppala for their computing knowledge and conversations when 1 was frus trated with my work. And of course, I have to go back to the very beginning and thank Jon Rose and Steve Asprey for introducing me to graduate student life in the department. I cannot forget Marc d'Anjou, Farzad Bakhtiar and Sean Burns, with- out whom, 1 probably would have finished this thesis months ago Cjust joking guys!) . Thanks to the three of you for hours of enjoyment and procrastination. I am indebted to Patrick Kehoe, Dave Ueno, Bill Krywko and Kevin T m in the woods' Robertson for hours of conversation and frustration on the golf cornes in the Kingston area. My thanks also goes out to Julie Ldamme and Janani Swamy for providing me with many meaningfd conversations and an ear to listen to my ramblings. 1 would &O
like to thank Harry and Katherine Iordanou and Claudio and Liliana Neagu for their support and their abiliw to invite me out for a break whenever I needed it the most.
1 would like to thank Che Wojtyk, Chris Evans and Chris Hilborn who have always managed to drag me away from my work. I am sure we wi11 still manage to get together fcr Our marathon weekends despite the fact that we are al1 heading in different directions.
Lastly, and most importantly, 1 would like to thank my parents, John and Irene Logan, and my sister Stacey. 1 am now finished and it could not have been completed without your support and encouragement.
4.3 Experimental Program For CuC12 Catalyst . . . . . . . . . . . . . . . 75 4.3.1 Planned Runs to E d u a t e Effect of CuC12 Concentration . . . 75 4.3.2 Planned Runs to Evaluate the Effect of Temperature . . . . . 75 4.3.3 z3 Factorial Design to Evaluate Effect of CuClz Concentration,
4.6.1 Effects of CuCl2 Concentration on DMC Produced . . . . . . 85 4.6.2 Effect of Catalyst Concentration on Byproduct Formation . . 87 4.6.3 Effect of Catdyst Concentration on Reactant Conversion . . . 92 4.6.4 Summary of the Effects of Catalyst Concentration . . . . . . . 95
4.7 Analysis of Temperature Effects with CuClz . . . . . . . . . . . . . . 97 4.7.1 Effect of Temperature on the Synthesis of DMC . . . . . . . . 97
. . . . . . . . 4.7.2 Effects of Temperature on Byproduct Formation 100 . . . 4.7.3 Effect of Temperature on Instantaneous DMC Selectivity 101
4.7.4 Effect of Temperature on Reactant Conversion . . . . . . . . 104 4.7.5 Arrhenius Expression Parameter Calculation . . . . . . . . . . 106 4.7.6 Sumrnaq of the Effect of Temperature on the Synthesis of DMCllO
dicated emission standards for different engine ages . (Horrocks, 1994) 2.4 A List of oxygenates with their associated rnolecular formula. molecular
weight and percentage breakdown of oxygen . . . . . . . . . . . . . . . 2.5 Oxygenates used in Liotta's study along with the physical properties
of the resulting blends (Liotta and Montalvo, 1993) . . . . . . . . . . . 2.6 Emissions results bom fuel blends tested by Liotta and Montalvo . . . 2.7 A list of oxygenates with their physicai properties and cost estimates .
. . . The costs were estimated from 1996 Chernical Marketing Reports 2.8 Physical Properties of DMC . . . . . . . . . . . . . . . . . . . . . . .
Legend for Process flow diagam . . . . . . . . . . . . . . . . . . . . . Valve settings for experiments . . . . . . . . . . . . . . . . . . . . . . Thermal mass flow controllers with their gas and flow rate range . . . Peak table with retention times and boiling points . . . . . . . . . . . GGMethod operational Parameters . . . . . . . . . . . . . . . . . . . Standards prepared to calibrate GC . Al1 units are in vol% . . . . . . . Statistical data from calibration runs . . . . . . . . . . . . . . . . . . . Retention times for gaseous reactants and products . . . . . . . . . . . Materials used in experiments . . . . . . . . . . . . . . . . . . . . . . .
C.1 Quantities of reactant and byproducts produced in Gibbs Reactor sim- ulations with a CO/02 ratio of 2 and vanous temperatures. Al1 q u a - tities are in Ib-mols. Negative(-) units for reactants indicate quantity of reactaat consumed. . . . . . . . . . . . . . . . . . . . . . . . . . . 140
C.2 Quantities of reactant and byproducts produced in Gibbs Reactor sim- ulations with a CO/Oz ratio of 1 and Mnous temperatures. All quan- t ities are in lbmols. Negat ive(-) units for reactants indicat e quantity of reactant consumed. . . . . . . . . . . . . . . . . . . . . . . . . . . 141
C.3 Quantities of reactant and byproducts produced in Gibbs Reactor sim- ulations with a CO/Oz ratio of 3 and Mnous temperatures. Basis:100 Ib-mollhr, dl quantities are in lb-mols. Negative(-) units for reactants indicate quantity of reactant consumed. . . . . . . . . . . . . . . . . . 141
C.4 Quantities of reactant and byproducts produced in Gibbs Reactor sim- ulations with a CO/02 ratio of 2, various temperatures and no H20 in the sample initially. Al1 quantities are in lbmols. Negative(-) units
. . . . . . . . . for reactants indicate quantity of reactant consumed. 142
viii
List of Figures
2.1 DiEerent synthesis routes to produce DMC . . . . . . . . . . . . . . . 2.2 Experimental setup used by Lee and Park in their investigations . . . . 2.3 Proposed reaction scheme for synthesis of DMC as proposed by Ro-
4.6 Instantaneous selectivity versus time for experiment using 69 mmol/L of CuCl*. . . . . . . . . . . . . . . . . . . . - . . . . . . . . . . . . .
4.7 Conversion of MeOH as a function of varying catalyst concentration 4.8 Conversion of O2 as a function of varying catalyst concentration . . . 4.9 Conversion of CO as a function of varying catalyst concentration . . 4.10 Effects of changing temperature on the production of DMC with 24
mmol/L of CuCl*. . . . . - . . . * . . . . . . . . . . . . . . . . . . . . 4.11 EEects of changing temperature on the production of DMC with 47
mmol/L of CuCl2. . . . . . - . . . . . . . . . . . . . . . . . . . . . . 4.12 Selectivity of products a t different temperatures using 24 mmol/L of
(ULEVs); and zero emission vehicles (ZEVs). Manufacturers selling vehicles in Cal-
ifomia are provided Bexibility in meeting the fleet-average requirements of the new
standards; they can produce any combination of the five low-emission vehicles as long
as they meet that year's fleet-average ernission standards for vehicles sold in California
(Sihota, 1995).
Table 2.1: Heavy duty diesel emission standards as outlined by the 1990 Clean Air Act. AU uni ts are in g/bhphr (Bennethum, 1991).
-
HC 1.3 1.3 1.3 1.3
CO 15.5 15.5 15.5 15.5
NOx 6.0 5.0 5.0 4.0
Particulate 0.6 0.25 0.10 0.10
The California Emission Standards for light duty diesels are outlined in Table 2.3.
Cornparisons between tables 2.2 and 2.3 reveal that California emission standards for
NOx and partieulates axe more stringent than the US federal standards.
While these standards are being implemented in the United States, the European
Union (EU) is also involved in regulating diesel emissions. Unlike the US, the EU has
a higher dependence on diesel fueled vehicles. Predictions for the year 2000 estimate
that 39% of vehicles in the EU will be powered by diesel as opposed to 17% in the
US (Horrocks, 1994). The EU'S emission standards are similar to those mandated by
the EPA. However, the phase in schedules for cornpliance are different due to their
dependence on diesel.
Table 2.2: Light duty diesel emission standards as outlined by the 1990 Clean Air Act. The bracketed numbers indicated emission standards for difïerent engine ages. (Horrocks, 1994)
ethers were not presented in the paper. The glycol ethers were tested in varying
concentrations. Diglyme and methyi soyate were also included in the study as diglyme
is known to be an effective oxygenate and methyl soyate, conventionally known as
biodiesel, is already an accepected alternative to diesel that can also be used as an
oxygenate to reduce harmful emissions. Cetane irnproving additives were added to
the fuel blends to counteract the lowering of the cetane numbers by the oxygenates.
The fuel blends were tested in a Detroit Diesel 1991 Series 60 prototype en-
gine. The blends were evaluated using standard testing procedures that exarnined
the blend's fuel consumption dong with hydrocarbon, carbon monoxide, NOx, and
particulate emissions. The results of the test are listed in Table 2.6. The results
within parentheses in Table 2.6 indicate that the changes between the fuel blends and
the reference fuel were not statistically significant.
Fuel consumption increased in test mixtures employing 1.40 and 1.82 wt% of glycol
ethers B and C, respectively. f i e l consumption was not significantly d e c t e d in the
remaining test fuel samples. The oxygenates had va,rying effects on HC, CO and P M
emissions. The only oxygenate which showed a significant increase in HC ernissions
was the aliphatic alcohol. CO and PM emissions were reduced or not significantly
changed in al1 cases. These results were accompanied with significant increases or
insignificant changes in NOx emissions. The results indicated that oxygenates only
had a significant effect on the increase in NOx if the concentrations were greater
than 5 vol%. At typical use levels, 2 vol% or lower, the oxygenated additives had
Little to no effect on NOx emissions (Liotta and Montalvo, 1993). In addition to
regulated emissions, total aldehyde and ketone emissions decreased by 10 to 25% in
the study for al1 oxygenated fuel blends tested except for the blend containing the
aliphatic alcohol where aldehyde and ketone emissions increased by over 20% (Liotta
and Montalvo, 1993). The effects of employing diglyme and the polyether polyol were
most promising. Both of these compounds significantly decreased HC, CO and PM
ernissions. This was accompanied with no significant increase in NOx ernissions.
Nikanjam (1993) performed a similar study. Table 2.7 lists potential oxygenates
identified by Nikanjam (1993). Listed with each oxygenate is its estimated cost and
physical characteristics. The additional cost per gallon associated with these oxy-
genates ranges from 0.07 to 0.40 $/gai. The necessity to reduce the costs of the
oxygenates to make them more economically viable is evident from these estimates.
Based on his selection criterium, Nikanjam identified ethylene glycol monobutyl
ether acetate and 2-ethylhexyl acetate as suitable diesel oxygenates. Ethylene glycol
monobutyl ether acetate was evaluated in tests due to its higher flash point and higher
oxygen content. The resulting blend in his test contained 10 wt% of the oxygenate
and had an oxygen content of 3% by weight. The test showed promising results with
an 18% reduction in both the particulate and CO emissions. This was accompanied
Table 2.7: A list of oxygenates with their physical properties and cost estimates. The costs were estimated kom 1996 Chernical Marketing Reports.
--
Oxygenate FP 02% Vol Lb/gal Cost Cost Extra
OC % $/Lb $/Gd $/gai h
Sec. butyl-alcohol
Isopropyl Alcohol
Eth-Gly Mono-butyl Ether
Propylene Glycol Methyl Ether
Propylene Glycol Butyl Ether
Diisopropyl Et her
Dimet hyi Carbonate
Ethylene GIycol Monobutyl
Ether Acetate
Diisobutyl Ketone
2-Et hylhexanol
Ethyl Glyme
Diglyme
Butyl Diglyme
with a 3% increase in NOx. Diglyme was not considered due to its cost.
Karas et al. (1994) evaluated oxygenated fuels with P-series glycol ethers. Karas
found that these oxygenates were effective in reducing hydrocarbon, carbon monoxide
and particulate matter emissions. In particular, Karas found that the PM reduction
was proportional to the oxygen content of the ether oxygenate containing fuel (Karas
et al., 1994).
Kanne (1988) and Murrayama (1995) evaluated DMC as a diesel oxygenate.
Kanne indicated that fuel blends with 5% by volume DMC showed a decrease in
particulate emissions between 10% and 30% in cornparison to the reference fuel. Car-
bon monoxide emissions were reduced by 5% to 10% with the use of DMC. Similarly,
Murrayama (1995) reported that particulate matter reduced almost linearly with
DMC concentration. Fuel blends evaluated with 10 vol% DMC indicated that PM
reduction of 3550% were attainable. These results were accompanied with reductions
of HC and CO and a small increase in NOx.
2.6 Producing Economically Viable Oxygenat es
Environmental issues and concems have had a tremendous impact on the chemical
processing industry in recent years. As a result of these growing concerns, companies
are now initiating research into areas that will eventually reduce the environmentai
hazards associated with their processes. The initiatives undertaken can proceed in
one of two directions. The first being a general proces assessrnent whereby the
manufacturer sets out guidelines that will eventually aid the Company in reducing
waste and recovering used chemicalç, materials and solvents. The second approach
is more involved whereby the industry tries to develop new process technologies and
new products which avoid environmental concerns (Cassar, 1989).
Dimethyl carbonate has been identified as a suitable diesel fuel oxygenate. Un-
fortunately, according to 1996 estimates, the costs of producing DMC is $12.46 per
gallon. Table 2.8 outlines the important physical properties for DMC.
Table 2.8: Physical Properties of DMC.
Dimethyl Carbonate C3H6O3 C:H:O by Mass 6: 1:8
Molecular Structure H3C - O - CO - O - CH3 Molecular Weight 90.1
Density at 20°C 1.07 kg/l
Flash Point 18°C
Solubility 13.9g/100g with water
Prior to 1984, DMC was almost exclusively produced via a reaction between
methanol and phosgene. Phosgene is an extremely toxic compound which was used
as a chernical weapon during World War 1. Well founded concerns about the sa&-
handling of phosgene lead researchers to find chernicals which could be used as al-
ternatives in processes requiring phosgene. DMC is an intemediate whose reactivity
makes it a suitable alternative to phosgene and other methylating agents such as
dimethyl sulphate and methyl chloride (Cassar, 1989). In addition to DMC's poten-
tial use as an oxygenate for diesel, DMC could also be used as an alternative for
phosgene in the manufacture of polycarbooates. Traditionally, polycarbonates(PC)
are produced by two methods; direct reaction and rnelt transesterification. In the di-
rect route, phosgene reacts with bisphenol-A(BPA) to produce a polymer in solution.
In transesterification, phosgene first reacts with phenol to produce diphenol carbon-
ate (DPC) , which in turn reacts with BPA to give phenol and a molten solvent-free
polymer (Purvis, 1992). A phosgene free route to produce PC has been developed
using DMC. In this scheme, DMC is synthesized and converted to DPC. The DPC is
then reacted with BPA to produce a polymer. With these potential applications in
mind, research has been initiated to produce both an economical and phosgene-free
route for DMC.
2.7 DMC Proposed Synthesis Routes
Intensified research within the last fifteen years, has culminated in rnany synthesis
routes being developed to produce dimethyl carbonate. The chemistry for these
synthesis routes can be broken down into four areas; indirect oxidative carbonylation,
direct oxidative carbonylation, transesterification and a direct carboxylation. The
following sections will deal with alternative synthesis routes to produce DMC. A
brief review will document current integrated systems to produce DMC along with
the proposed catalysts and operating conditions. Figure 2.1 outlines the different
production routes to synthesize DMC.
Figure 2.1: Different synthesis routes to produce DMC.
The reaction schemes are differentiated based upon the catalyst chosen and the
reaction operating parameters (i.e. temperature and pressure). A few of the pro-
duction routes are commercial processes while others are still in the experimental
stages.
Indirect Oxidat ive Carbonylat ion
Traditionally, DMC was produced by the indirect oxidative carbonylation of methanol
via phosgene (Bhattacharya, 1995). In this reaction scheme, carbon monoxide reacts
with chlorine to produce phosgene (2.1). Phosgene produced from this reaction then
reacts with methanol to produce DMC and hydrochloric acid (2.2).
CO + Cl* + CI(C0)CI (2.1)
CI(C0)Cl + 2CHjOH t (CH30)&0 + 2HC1 (2.2)
Alternatively, one can also produce DMC via the indirect oxidative carbonylation
of methanol via dkyl nitrite (Matuzaki and Simamura, 1993). In this reaction scheme,
methmol reacts with nitric oxide and oxygen to form methyl nitnte and water (2.3).
Methyl nitrite then reacts with carbon monoxide to produce DMC and nitric oxide
(2.4) .
Carbon monoxide and methyl nitrite are pumped into a catalyst packed tubular reac-
tor. The system employs multimetallic (Pd-Cu-Mo-K) halides supported on activated
carbon as a catalyst (Bhattacharya, 1995). The reaction is operated at temperatures
between 80 and 150°C to avoid side reactions. The recommended pressure is 483
kPa. Ube Industries in Japan currently uses this method in a DMC production plant
producing 3,000 metric tons per year (Bhattacharya, 1995).
Direct Oxidative Carbonylation
2.7.2.1 Liquid Phase, Homogeneous, Direct Oxidative Carbonylation
The production of DMC via the direct oxidative carbonylation of methanol in the
presence of carbon monoxide and oxygen is perhaps the most exploited synthesis
route (2.5) (Bhattachaxya, 1995).
The formation of dkyl carbonates by the reduction of transition met al compounds
(ie. palladium and copper) with an alcohoi solution and carbon monoxide was first
investigated by various authors (Mador et al. (1963); Graziani et al. (1971)).
Realizing the advantage in cost of copper halide catalysts over palladium halide
catdysts, Romano e t d.(1980) studied the kinetics of this reaction in the presence
of a cuprous chloride catalyst. Romano et al. (1980) discovered that the reaction
actually occurred in two steps; an oxidation followed by a reduction. During the
first step, cuprous chloride reacts with methanol and oxygen to form cupric methoxy
chloride and water (2.6).
Cupric methoxy chloride is then reduced with carbon monoxide to form dimethyl
carbonate and the cuprous chloride (2.7) is regenerated (Romano, Tesei, Maun and
Rebora, 1980).
Preliminary results indicated that the rate of formation of DMC was a function of
oxygen feed rate and independent of cuprous &onde concentration, temperature and
pressure. Oxygen consumption from the reaction was virtudy complete; therefore
safety concerns were abated since the exit gas from the reactor was not explosive. As
indicated by the stoichiometry of the reaction, water is formed in equimolar amounts
to DMC- The effects of the increased water concentration included a decrease in
selectivity of cabon monoxide to produce DMC, promoting the formation of carbon
dioxide (Romano, Tesei, Mauri and Rebora, 1980).
Enichem Synthesis produces DMC via the direct oxidative carbonylation of methanol
with oxygen and carbon rnonoxide in the presence of cuprous chloride in a stirred tank
reactor. The recommended temperature and pressure operating limits ranged from
90 to 150°C and 2068 to 3102 kPa, respectively. Enichem's DMC capacity a t its plant
in Ravenna, Italy, is 12,000 metric tons per year (Bhattacharya, 1995)
Hallgren et. al. (1982a) investigated the oxidative carbonylation of methanol in
the presence of carbon monoxide and oxygen, at elevated temperatures and pressures.
The process is preferably operated a t temperatures between 180 and 250°C and pres
sures between 4136 and 10341 kPa. A cupric halide catalyst was employed in amounts
ranging from 0.02 wt% to 1.5 tvt% based on the mass of methanoi (Hallgren, 1982a).
Hallgren claims that a reaction time of two hours will result in a 20% conversion
to DMC, based on methanol. Hallgren's results indicated that the reaction did not
proceed below a temperature of lSO°C and that pressures of 10341 kPa or greater
were recommended.
More recently, Bhattacharya (1995) presented a new homogeneous catalytic sys-
tem to produce DMC by the direct oxidative carbonylation of methanol with oxygen
and carbon monoxide. The reaction is carried out in a stirred tank reactor loaded with
methanol and small quantities of N-rnethyl-2-pyrrolidone (NMP) which serves as a
high boiling CO-solvent. The catalyst is copper(I1) methoxy chloride which has already
been identified as an intermediate of the direct oxidative carbonylation reaction by
Romano et al. (1980). NMP and similar compounds are utilized to increase the rate
of the reaction. It has also been suggested that NMP can serve as a catalyst caxrîer
during the rernoval of DMC/water by any flash procedure (Bhattacharya, 1995). Ex-
perirnental data was not presented in the paper, however, conversion to DMC shoulc!
be complete since the catalyst employed is an intemediate outlined by Romano et
al. (1980) and one step of the reaction is removed (i.e. the formation of the cupric
met hoxy chloride intermediate).
2.7.2.2 Liquid Phase, Heterogeneous, Direct Oxidative Carbonylation
Sawicki and Chaftez (1987) suggest the use of solid supported catalysts for the pro-
duction of DMC via the oxidative carbonylation of methanol with carbon monoxide
and oxygen. Suggested supports are aluminum oxide, silicon dioxide, zeolites or other
naturally occurring clays such as mordenite. The catalysts were evaluated in a pres-
sure reactor operated a t temperatures between 50 and 150°C and pressures exceeding
6984 kPa for as long as 8 hours. Based on the copper salt charged, yields of DMC were
as high as 83%. The preparation of the catalysts are labour intensive for industrial
scale applications. No indication on the DMC yield based on the methanol charged
to the reactor is given.
Sawicki et.al. (1987) dealt exclusively with heterogeneous catalysts based upon
oxides of silicon. Lee and Park (1991) took a different approach and evaluated the
synthesis of DMC with NaX, NaY, and HY zeolites exchanged with cupric chloride.
Preliminary results indicated that zeolites exchanged with cupric nitrate failed to
display a significant activity. Evaluation of the zeoiites were conducted in a 1 L
autoclave charged with methanol, car bon monoxide and oxygen. Their experimental
apparatus is illustrated in Figure 2.2.
The operating conditions were ~ a r i e d with temperatures ranging from 90 to 120°C,
reaction time Miied from 1 to 12 hours, and an operating pressure of 5067 kPa (with
varying partial pressures of carbon monoxide and oxygen). The results from the
experiments conducted by Lee and Park indicate the following:
1. Rate of formation of DMC is insignificant after 6 hours of reaction.
2. Production of DMC increases linearly with temperature.
3. Production of DMC is not d e c t e d by exces methanol.
4. Yield of DMC increases lhearly with oxygen partial pressure.
Yields of DMC based on methanol were found to be 36.6% and 17.8% for the
CuNaX and CuHY zeolites, respectively. Unfortunately, t here was no indication of
an experimental design in Lee and Park's analysis. Since the variables were changed
one at a time, interaction effects could not be evaluated.
Further investigations by Hallgren et al. (1982b) suggested that aliphatic carbon-
ates could be produced from an alcohol, carbon monoxide, oxygen, a Bronsted base
(Le. sodium hydroxide), and a Group VIIIB catalyst (Le. palladium, rhodium, etc.)
Experimental results were carried out under atmospheric conditions in Pyrex flasks
with a gas mixture of carbon monoxide and oxygen continually bubbled through the
methanol. Reaction times were often greater than 20 hours. This technique was
successful in producing carbonates from higher alcohols but was not successful in
producing DMC from methanol in significant quantities (1.2% conversion attained).
Smith et al. (1992) present a route to produce DMC via the oxidative carbony-
lation of methanol with oxygen and carbon monoxide but in the presence of a het-
erogeneous halogen free copper catalyst over a fixed bed reactor. The catalyst used
in this system consisted of copper, ion exchanged or impregnated ont0 a clay solid
support such as montmorillonite or a metal pillared interlayered clay such as silica.
Suggested operating conditions for the reaction are temperatures between 70 and
140°C and pressures ranging from 3792 to 4481 kPa. Significant quantities of DMC
were reported to have been produced.
Figure 2.2: Experimental setup used by Lee and Park in their investigations.
2.7.2.3 Vapour Phase, Heterogeneous, Direct Oxidative Carbonylation
Vapour phase oxidative carbonylation reactions to produce DMC have also been in-
vestigated. Curnutt (1986) investigated the vapour phase oxidative carbonylation of
methanol with a mixture of carbon monoxide and oxygen in the presence of a copper
catalyst. Cumutt contends that the b a t results were obtained when using an acti-
vated carbon supported cupric chloride catalyst promoted with potassium chloride
(Curnutt and Haley, 1986).
Experiments were conducted in a Hastelloy C276 plug flow reactor. Pnor to being
charged into the reactor, the rnethanol was vaporized and the gas mixture of carbon
monoxide and oxygen was preheated. As with other oxidative carbonylations, the
major byproduct of the reaction was carbon dioxide. Curnutt found the thermal
oxidation of carbon monoxide to carbon dioxide to be signifiant a t temperatures
greater than 130°C. Since the reaction rate was found to be negligible below 100°C,
the reactor was operated between 110 and 130°C- The optimum operating pressure
was experimentally determined to be 2068 kPa. Although percentage yields of DMC
based upon methanol were not supplied, the results indicated that the catalyst showed
a high selectivity to produce DMC (greater than 80%) at the optimum conditions.
Koyama et al. (1992) investigated the oxidative carbonylation of methanol with
oxygen and carbon monoxide in the vapour phase. The catalyst used for this reaction
was comprised of a copper halide with at least one hydroxide compound selected from
the group consisting of alkaii metal hydroxides exchanged onto activated charcoal
(Koyama et al., 1992). The reaction is carried out in a plug flow reactor with an
operating temperature between 100 and 200°C and an operating pressure between
483 and 1448 kPa. Various catalysts were tested in their system and conversions of
methanol to DMC were as high as 23%.
2.7.3 Transesterification
Tkansesterification is an ester interchange reaction where an ester is reacted with
excess alcohol or acid. It is a mild reaction which is carned out in the liquid phase
without toxic chernicals and corrosiveness (Okada et al., 1995). Two alternative routes
have been proposed to produce dimethyl carbonate via tramesterification. The first
involves ethylene carbonate and the second involves di-tert-bu~l peroxide (DTBP)
(Bhattacharya, 1995).
In the first route, ethylene oxide reacts with carbon dioxide to produce ethylene
carbonate (2.8). The ethylene carbonate then reacts with exces methanol to yield
DMC and ethylene glycol (2.9)-
Using this route, Knifton and Duranleau (1991) evaluated diflerent catalysts to
produce DMC. These catalysts included macroreticular and gel-type exchange resin
with tertiary amine groupings bonded to a polymer backbone, ion exchange resins
with ammonium functional groups, alkali-metal silicates impregnated into silica, am-
monium exchanged zeolites and acidic resins bearing sulfonic acid and carboxylic acid
functional groups (Knifton and Duranleau, 199 1).
Insoluble base catalysts, such as tertiary amine and quaternary ammonium func-
tionalized resins, were effective for the t ransestenfication reaction. These catalysts
are easy to rernove from the final product and are readily available. Product compo-
sitions consisting of 21 wt% DMC were obtained after 2 hours of operation at 100°C
and 690 kPa in a fixed bed reactor. Higher yields would probably be attainable by
operating the reactor at higher temperatures. However, the thermal stability of the
catalysts becomes an issue a t temperatures greater than 120°C.
The second route is more cornpiex. In the first step, isobutane reacts with oxygen
to produce di-tert-butyl peroxide (DTBP) and water (2.10). DTBP then reacts with
carbon rnonoxide to produce di-tert-butyl carbonate (DTBC) (2.11). DTBC is then
transesterified with methanol to produce DMC and tert-butyl dcohol (TBA) (2.12) .
Lucy et- al. (1989) utilized this route using a platinum group metal and a copper
catalyst. The platinum group metal and the copper are added to the methanol as
salts. The reaction is carried out under low pressures, 520 to 1035 kPa, and moderate
ternperatures, 50 to llO°C. With the addition of nitrile promoters (ie. benzonitrile),
yields of DMC based upon the conversion of DTBP ranged from 1% to 50%.
2.7.4 Direct Synthesis - Carboxylation
Fang and Fujimoto (1996) propose the synthesis of DMC via a direct rnethod by the
reaction between carbon dioxide and methanol (2.13).
The reaction was carried out in a 1 L autoclave in the presence of a basic catalyst
and a methyl iodide(CH31) promoter (Fang and Fujimoto, 1996). Suitable bases in-
clude potassium carbonate (K2C03) and potassium phosphate (K3P04). Suggested
operating conditions were 100°C and 5070 kPa for temperature and pressure, respec-
tively. The yield of DMC based on CH31 consumed were between 94 and 228%. It
is believed that HI, an intermediate byproduct of the reaction, reacts with methanol
to produce excess CH31; thus accounting for the greater than 100% yields based on
CH31 consumed (Fang and Fujimoto, 1996).
2.8 Expected By-products and Reactions
The reaction scheme for this reaction is very cornplex. Many of the papers in the
literature review do not even attempt to propose a mechanism. The only mention
of a mechanism concerning the formation of DMC with CO, O2 and methanol was
proposed by Romano et al. (1980). Their scheme, illustrated in Figure 2.3, employed
CuCl and not CuC12 as the catalyst of choice. According to this reaction scheme, CuCl
Figure 2.3: Proposed reaction scheme for synthesis of DMC as proposed by Romano et al. (1980)
reacts with CO to form cupric carboxy chloride (CuCOCI). While this is occurring,
CuCl reacts with CHSOH and O2 to form cupric methoxy chloride (Cu(OCH3)Cl) and
H20. Cupric methoxy chloride is also formed by a reaction between CuCOCl, CH30H
and O2 to produce Cu(OCH3)Cl, H20 and CO. Romano et al. (1980) indicate that
CuCOCl reacts with Cu(OCH3)C1 to form a rnixed valence intermediate complex
which again reacts with Cu(OCH3)Cl to form DMC and regenerate CuCI.
King (1996) indicated that the formation of DMC via the oxidative carbonylation
of methanol with CuC12 c m be broken down into two steps. The first being the
oxidation to form cupric methoxychloride (2.14).
This is followed by a second step whereby the cupric methoxychlonde is reduced with
carbon monoxide to form DMC (2.15).
Besides DMC, a number of other by-products are produced in this reaction. These
include COz, methylal and methyl formate. COÎ is formed from the oxidation of CO
with O2 (2.16).
Studies by Romano et a1.(1980) and King(1996) indicate an induction penod be-
fore DMC formation commences. During this period, King (1996) postulates that
methanol is oxidized to formaldehyde. The formaldehyde than reacts with methanol
to form methylal.
According to King (1996), if the copper forms bigger aggregates during the reaction,
more oxygen complexes with copper to oxidize the methanol to formic acid. This in
tum reacts with methanol to form methyl formate.
CH20 + 0.502 t CHOOH
CHOOH + CHjOH t CHOOCH3 + H 2 0
2.9 Summary
It is clear that a number of synthesis routes are available to produce DMC. The
ultimate goal of the research group is to develop an affordable alternative production
routes for diesel oxygenates with catalytic distillation in mind. Therefore, the purpose
of this thesis is to investigate the process parameters aiffecting the production of DMC.
It was decided that the synthesis of DMC would be pursued with the direct oxidative
carbonylation chemistry. The reasons for this were two fold. Firstly, since there is
already an industrial process in operation using this chemistry, i t seemed like a good
starting point. Secondly, the reaction could be investigated using homogeneous and
heterogeneous catalysts. This now left the task of finding appropriate equipment for
the experimentation.
Chapter 3
Experiment al
The synthesis of alkyl carbonates with metal halides such as PdCh and CuC12 is not
new. Romano et al. (1980) perforxned a kinetic investigation of the synthesis with
cuprous chloride as the main catalyst in the early 80's. Variations of catalytic systems
employing cupric chloride are also reported in the literature. Unfortunately, much of
the pertinent information is found in patents. This chapter will detail the apparatus
used in the experirnent, the materials used and the methods employed to analyze the
sarnples.
Apparat us
The apparatus used in this investigation consisted of a 300 ml parrTM batch reactor.
The gas flow into the reactor was controlled with a ~ r o o k s ~ ~ model 5874A thermal
m a s flow meter. Online sampling was performed with a varianTM 3400 gas chro-
matograph with a thermal conductivity detector (TCD) for the liquid analysis and
~ i s c h e r ~ ~ 1200 gas partitioner equipped with a TCD for the gas analysis. A flow
diagram of the system is illustrated in Figure 3.1. Table 3.1 is a legend for Figure
3.1. Table 3.2 outlines the valve positions before, during and following a run.
ATMOSPHERE 2 V 3 V U
. REACTOR
f v 4 1 &Gl Sample
~ i q u i d Sample
Figure 3.1: Flow diagram of system used to test the synthesis of DMC.
Table 3.1: Legend for Procas flow diagra
[ Designation Description
Gas Line
Check Valve
Needle Valve
Metering Valve
Flow Controuer
Flow Sensor
Bal1 Valve
Pressure Gauge
Back Pressure Gauge
3.1.1 Flow rneters
The gas feed consisting of oxygen(02), carbon monoxide(C0) and nitrogen(N2) was
supplied in the appropriate quantities with a BrooksTM model 5874-A thermal m a s
flow controller. The flow rate of each gas outlet stream was set with a precision po-
tentiometer. The gas flow was adjusted, by the controller, according to the changing
downstream pressure; ensuring that a consistent gas mixture was obtained. A down-
stream pressure drop of 345 kPa was obtained by using a GOTM Inc. Mode1 DP 60
back pressure valve.
The flow meters were powered with a + 15 V DC power supply. The output signal
to the flow meters ranged from O to 5 V DC with O V DC for no flow and 5 V DC for
100% of flow. The signals dictating the flow of the individual gases were determined
according to the desired gas composition and the maximum allowable flow of the flow
meters. Table 3.3 lists the flow meters, their flow range and the metered gas.
Table 3 -2: Valve set t ings for experiments
Valve Number Start-up/Shut-Down Position Run Position
NV1
l W 2
NV3
NV4
NV5
hV6
NV7
NV8
NV9
NVlO
NVl l
NV12 NV13
BVl
BV2
BV4
BV5
BV3
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Open
Closed
Open
Closed
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Closed
Open
Open
Open (for liquid sampling)
Open (for gas sampling)
Closed
Table 3.3: Thermal mass flow controllers with their gas and flow rate range.
3.1.2 Batch Reactor
Flow reactors, autoclaves, cstr's and batch reactors were al1 used in previous inves-
tigations to study this type of reaction. After investigating the equipment available
for this research, it was decided that a batch reactor would best suit our purposes.
The reactor obtained was a PARR batch reactor mode1 4561. The reactor was
rated to a maximum operating temperature and pressure of 350°C and 20700 kPa,
respectively. These limits were well within the desired operating range of the experi-
ments. A cross sectional view of this reactor is illustrated in Figure 3.2.
A proportiond-integral-controller (PID), which was supplied with the unit, was
used to control the temperature of the reactor. The temperature was monitored with
an 8 inch stainless steel J-type thermocouple which was attached to the controller.
The reactor was heated within a heating mantle comprised of an alurninum shell and
fabric liner. The reactor was cooled with cold water which is passed into the reactor
via a 316 stainless steel u-tube. In addition to the flow control valves, the water flow
was adjusted with a needle valve (NV13) to dampen overshoots. The addition of a
needle valve to limit the cold water flow was necessary since the cold water bursts
often cooled the reactor as much as 5°C with large shots. This in turn caused the
reactor temperature to fluctuate rapidly.
The reactor was stirred with a ~ a g n e d r i v e ~ ~ magnetic stirrer. The stirring shaft
and propeilers were constmcted of InconelTM to guard against corrosion. The stirring
motor was capable of rotating to a maximum of 650 rpm.
A dip tube allowed for liquid samples to be removed from the reactor while the
reaction commenced. A gas sample port was also available to analyze the gas contents
of the reactor. Figure 3.3 details the liquid and gas sampling apparatus (Le. the dip
tube and the gas sampling line).
Figure 3.2: Cross-sectional view of parTM mode1 4561 batch reactor. (Reproduced fiom pu rTM instruction manual)
CooIing Tube Stimng Shaft Dip Tube Thermocouple Gas sample Port
Figure 3.3: Liquid and gas sampling apparatus on panTM batch reactor.
3.1.3 Reactor Modifications
Modifications were made to the reactor to aid in analysis and for safety concerns.
These modifications included changing the type of rupture disc to changes in the
head connections of the reactor to aid in reactor sampling.
3.1.3.1 Rupture Discs
Although all the experiments were being conducted at pressures well below the rated
operating limits for the rupture discs, rupture disc failure was experienced in a num-
ber of experirnental runs. Initidy, the cause of the disc rupturing could not be
determined. Possible reasons were a large increase in pressure within the reactor or
faulty rupture discs.
The chemistry did not indicate a possibility of runaway reactions. Other factors
rnay have contributed to a rapid pressure increase within the reactor unit. Firstly,
the components used in the reactions, oxygen, carbon monoxide and methanol are al1
combustible. Ignition of these components could cause the pressure within the reactor
to increase at a high rate and be responsible for a rupture disc breach. Possible
ignition sources include improper grounding of the reactor unit leading to a build up
of a static charge or sparks created from friction between moving parts within the
reactor. The reactor was thoroughly examined and no abnormalities were found.
Cupric chloride was used as the catalyst for al1 these experiments. Unfortunately,
cupric chloride is also very corrosive. Although extreme care was taken to prolong
the life of a rupture disc, continuous contact with a hostile environment could weaken
the discs. This would result in the discs rupturing at pressures well below their rated
bursting pressure. A test was performed to examine the effect of cupric chloride on
the InconelTM rupture discs.
A circular fragment of a blown rupture disc was immersed in a 10 g/L solution of
cupric chloride and methanol. This concentration of cupric chloride was comparable
with the maximum concentration used in the experiments. Following 24 hours of im-
mersion, the rupture disc was removed and examined. The disc was severely corroded
by the solution. Figure 3.4 displays a picture of a corroded and non-corroded circular
section of a rupture disc.
Figure 3.4: A non-corroded circular section of a rupture dise is pictured on the left. A similar section after 24 hows of immersion in a IOg/L cupric chloride solution.
Chlorides are generally more aggressive on stainless steel because of the ability
of the chloride ion to penetrate the passive film and cause pitting (Craig, 1989);
hence the requirement to use resistant alloys for crucial equipment. InconelTM is
nickel-chrornium alloy which are generally resistant to chlondes. Unfortunately, the
InconelTM rupture discs were slowly being corroded with continual contact of the
cupric chloride environment. As a result, they were yielding a t premature pressures.
Upon further consultation with a technical representative from ParrTM, it was decided
that gold plated rupture discs would be used in the experiments. Although more
expensive, the gold plated discs were more suitable for the operating environment and
could offer a considerable increase in resistance to the corrosive chloride environment.
No problems were encountered once the rupture d i sa were changed.
3.1.3.2 Reactor Controls
Al1 experiments were conducted within an explosion proof bunker. The purpose of
the bunker was to reduce or eliminate the risk of persona1 injury while performing the
experiments. The reactor pressure gauge, back pressure controt valve, liquid sampling
line and gas loading valve were al1 installed outside of the bunker. With this set-up,
there was no need for any researcher to enter the bunker during the operation of the
reactor. Figure 3.5 displays the valve and cont roller arrangement outside the bunker.
S tirrer Controller
Temperature Controller
4 Liguid Sarnpling Valve (BV4)
Access to Bunker
@ Cooling water conml Valve (NV 13)
Figure 3.5: Diagram of reactor controller layout.
3.1.3.3 Upper Reactor Assembly
The upper reactor assembly was slightly modified for the purposes of this experi-
ment. Figure 3.6 dispiays the ports available for connections for the upper assembly.
Initially, port E was connected directly to a pressure gauge. This port was modified
A Stirrerport
B U-tube port
C u-tube port
D Thermocouple port
E Gas sarnpting port
F Gas loadingfliquid sarnpiing port G Rupture disc port
Figure 3.6: Ports available for connections in the upper reactor assernbly.
with the addition of a needle valve and a SwagelokTM fitting. The gas sampling line
was connected to the valve. This allowed the gas sample line to be disconnected while
the reactor was pressurized. Additional tubing was required to attach the SwagelokTM
fitting to a pressure gauge outside of the bunker. Port F was used for the product
loading and liquid sampling. A check valve and bail valve were added to this port.
The check valve ensured that the reactor contents did not flow into the gas load-
ing line. The bal1 valve was used to isolate the reactor when pressure testing was
3.2 Liquid Analysis
3.2.1 Analytical Equipment
Liquid samples from the reactor were anaiyzed with a VarianTM 3400 gas chromato-
graph (GC) equipped with a thermal conductivity detector (TCD). A DB-FFAP,
used to analyze the liquid samples. This column was the most effective for the sepa-
ration of the reaction products and reactants.
A TCD was preferred in this analysis due to its ability to detect water; which was
a byproduct of the reaction under study. The TCD consists of a hot tungçten-rhenium
filament over which the eluted gas from the column is passed. The detector produces
a signal which is dependent on the changes of thermal conductiviw of the different
products. The output signal is produced a s the GC compares the signal produced by
the products and the carrier gas to a reference stream of pure carrier gas.
Complete separation of the reactants and products must be obtained to get an
accurate measure of the quantities of each within the sample. Partial peak separation
will result in improper accounting of the different cornponents. Complete separation
of methylal, methyl acetate, and methyl formate was difficult due to their low boiling
points. The temperature profile of the method is dispiayed in Figure 3.7. Table 3.4
outlines the boiling points and retention times of the liquid reactants and products
frorn this reaction.
Once the sample was injected, the column was held a t 40°C for 1 minute. The
temperature of the column was then ramped a t 50°C/min to a final temperature of
110°C. The column was maintained at this temperature for 6 seconds prior to being
cooled to the start-up temperature of 40°C- The total running time of th% method
was 2.5 minutes. On average, samples were analyzed every seven minutes. Table 3.5
lists the other pertinent setting for the GC method used in the analysis.
Table 3.4: Peak table with retention times and boiling points.
Peak Retention Time (min) Boiling Point (OC)
Me thylal 0.658
Methyl Formate 0.687
Methyl Acetate 0.805
Met han01 1.168
DMC 1 .470
Water 1.906
Table 3.5: GC-Method operationd Parameters.
Parameter Setting
Injection Temperature
Ti (Initial Column Temperature)
Ti Hold-tirne
Temp Rarnp Rate
Tf (Final Column Temperature)
Tf Hold Time
Detector Temp.
Attenuation
Range TCD Filament Temp.
130°C
40°C
1 min
50" Cfmin
1lO0C
0.10 min.
130°C
4
0.05
160°C
Figure 3.7: Temperature profile of method used on VarianTM 3400 GC to analyze the liquid sampIes.
3.2.2 Spectrum of Liquid Products
A spectmm of the liquid products produced from the gas chromatograph is displayed
in Figure 3.8. The area underneath each Peak corresponds to the amount of that
particular compound in the sample.
It is possible to have compounds that can have the same retention time within
the column. Product confirmation was perforrned using a combined Gas Chromate-
graph/Mass Spectrometer (GC/MS). The GC/MS breaks each compound into its
constituent groups for identification via electron ionization. In the electron ionization
(EI) mode of operation, sample molecules are introduced into the ion trap, where they
interact with energetic electrons to form positive ions. The ions are sorted according
to their mass to charge ratio (m/z) and the results are reported in a mass spectrum.
The spectrum produced was compared against library spectra and standard spectra
of compounds with similar molecular masses.
It is interesting to note that the library spectrurn for DMC was slightly difkrent
than the spectrum produced by injecting pure DMC into the GC/MS. This is most
likely attributed to the difference in detector used in the analysis and that used to
generate the mass spectra Libraries- A sample of pure DMC was then injected into the
GC/MS. The spectrum produced by the pure sarnple was the same as that obtained
from the reaction sample. Figures 3.9 and 3.10 are GC/MS spectrum obtained from
a pure sarnple of DMC and the library sample of DMC, respectively .
Chart Speed = 7-91 cm/min Attenuation = 3521 Zero O f f s e t = 5 % Start T i m e = 0.000 min End Time = 2.502 min Min / Tick = 1-00
' O . 1 ' 0 . 2 ' 0 . 3 'O - 4 ' 0 . 5 ' O -6 '0.7 I
Methylalf
rnethyl aceta
DMC
Water
Volts
O . 6 5 8
Figure 3.8: Spectrum of Liquid Products.
Background Subtract C :\SATURN\DATA\DECiZ-l Date : 12/11/95 18 : 53 : 82 Comment: tl 1680,T 23, MANIFOLD 228, A f l 2 .5 , EC 12uA DMC PRODUCT Average of: 473 to 477 Hinus: 499 ta 503 l 0 0 * ~ = 10661192
Figure 3.9: GC/MS Spectrum for pure DMC.
Figure 3.10: Library GC/MS Spectrum for DMC.
f
45
31 59
98 ' 1 6 " 1 i " 1 ' I " 1 ' '"'3111rr
-
-
Fflc: N I S T 9 0 Entrg: 838 Peaks: 28 HOI u t : sa
t lasdlnt i l a s d I n t tiassl Int b s s / I n t t i a s d l n t
To ensure the volumetric quantities calculated by the GC were accurate, calibration
runs were performed to properly quanti& the analysis. The samples were prepared
based upon the expected composition from the experimental run.
Standard 1 was prepared to the specifications listed in Table 3.6. The remaining
4 standards were prepared by serial dilution with methanol. Table 3.6 lists the five
calibration standards aiong with the volume percent composition of each compound
in the standard.
Table 3.6: Standards prepared to calibrate GC. AU units are in vol%.
1 Compound STND 1 STND 2 STND 3 STND 4 STND 5
Methylal 10 8 6 4 2
Methyl Formate 5 4 3 2 1
Methyl Acetate 5 4 3 2 1
Met han01 40 52 64 76 88
DMC 20 16 12 8 4
Water 20 16 12 8 4
Each of the five standards samples were injected three times. The varianTM
Star (v4.02) software provided with the GC used the volume breakdown of the stan-
d x d injection and correlated them with the signal, or peak count, produced by the
detector. Calibration curves were constructed with the multiple injections of each
sample. Table 3.7 outlines the statistical data obtained from the calibration data
for the individual cornponents. Calibration sarnples were run periodically to ensure
experimental accuracy. The calibration curves for methylal, methyl formate, methyl
acetate, methanol, DMC and water are displayed in Figures 3.11, 3.12, 3.13, 3.14,
3.15 and 3.16, respectively.
Table 3.7: Statistical data fiorn caiibration runs.
met hy la1 0.930992 ! rnethyi formate 0.986888
methyl acetate 0.992100
met han01 0.985214
dimethyl carbonate 0.992855
water 0.985660
Figure 3.1 1: Met hylal calibration curve.
Replicates 2 3 1 2 2
Figure 3.12: Methyl Formate calibration curve.
Replicates 3 2 3 3 3 1
l
Amount
Figure 3.13: Met hyl Acetate calibration curve.
Figure 3.14: Methanol calibration curve.
Figure 3.15: DMC calibration curve.
Figure 3.16: Water calibration curve.
3.2.4 Sampiing Frequency
Paramount to the investigation of this reaction was the ability to sample and analyze
the liquid contents of the reactor. Ideally, this should be done on line and as often as
possible to provide a true account of the progres of the reaction.
There are currently two systems available to collect samples from the reactor. The
first involves the direct on line sampling of the reactor contents. A liquid sample was
directly injected into the gas chromatograph for analysis with a V A L C O ~ ~ VICI A90
auto-sampling valve. Although useful, injections could only be performed every 7
minutes due to the time required for the GC to restabilize. Samples were therefore
collected in 3 minute intenals. The samples were collected in 2 ml collection vials
and stored in an ice bath to prevent further reactions from occurring. The increased
sampling frequency provided a more accurate account of the reaction as it progressed
with time.
3.3 Gas Analysis
The reactant gas composition was analyzed with a FischerTM 1200 gas partitioner.
The partitioner was equipped with a duat column and a TCD detector. Column 1
was a I lg inch diameter column packed with 80-100 mesh Columnpak PQ. Column 2
was a 3/16 inch diameter column packed with 60-80 mesh molecular sieve 13X. The
column temperature was set to 50aC, the attenuation was set to 2 and the bridge
current to 200 mA. The spectrum was produced with an H P 3390A integrator. The
gases being analyzed by the TCD were CO, 0 2 , CO2 and N2. The retention time for
these gases are listed in Table 3.8. A sample spectrum of the gases is displayed in
Figure 3.17. The results are produced as volume % of the total sample.
Table 3.8: Retention times for gaseous reactants and products.
1 Gas Retention Time (min.) 1
3.4 Mat erials
Table 3.9 outlines the chernicals procured for this aoalysis. In sufficient quantities, dl
the materials used in this investigation can be hazardous to one's health. Strict com-
pliance with safety regdations must be adhered to when operating and analyzing the
results of the experiments. The most significant danger posed by these experiments
is asphyxiation with carbon monoxide. It is extremely important that the ventilation
fan and CO detector be in good working order prior to beginning an experiment. CO
is an odourless and colourless gas that can unknowingly build up in the bloodstream
causing the individual to lose consciousness. It is also important that safety glasses
and chernical resistant gloves be wom when handling the liquid sarnples. Although
the quantities are small, a11 the liquid products c m be imtating to the skin and eyes.
Additional health and çafety information can be found in Appendix A.
Table 3.9: Materials used in experiments.
1 Chernical Supplier Catalog # - - - - -
Methylal BDH 29206
Methyl Formate Aldrich M4683-7
Methyl Acetate BDH 29202
Methanol ~ i s c h e r ~ ~ A408-4
DMC Aldrich D 15,292-7
Cupric Chloride FischerTM C455-500
Figure 3.17: Spectrum produced from gas analyzer.
3.5 Procedures
3.5.1 Reactor Loading and Set-up
1. Weigh the required quantity of catalyst into a clean measuring dish.
2. Transfer the cupric chloride into a g l a s reactor liner.
3. Add 150 ml of rnethanol to the liner. Use a portion of this methanol to rinse the measuring dish and t r a d e r the wash solution to the liner.
4. Insert the g l a s liner into the bottom portion of the reactor.
5. Place upper reactor assernbly on top of the lower reactor assembly.
6. Place Gclamps appropriately around the reactor.
7. Place O-ring around the reactor and secure into position with the hex-nut.
8. Place the reactor unit into the vice and tighten the bolts of the Cclarnps.
9. Position the reactor into the stand with the heating mantle surrounding the lower portion of the reactor and the stirring motor attached to top of the reactor.
10. Attach al1 of required lines to the reactor (gas loading line, gas sampling line, liquid sampling line, cooling water lines and the thermocouple).
11. Turn on cooling water to reactor. Be sure that the needle valve is in the closed position. (Note, although the needle valve is in the closed position, it is not completely closed. Cooling water will still flow into the reactor.)
3.5.2 Pressure Checks 1. Turn on ventilation fan within the bunker.
2. Turn back-pressure valve (BP 1) clockwise 5 turns. This will provide sufficient back-pressure, greater than 3450 kPa, to pressure check the system.
3. Ensure that al1 readings on the Row meters are set to 0.00 and that the blend option is NOT employed.
4. Turn the nitrogen flow to 100 %.
5. Open the gas flow control valve and pressurize the reactor to 3800 kPa (the inlet pressure of the gases).
6. When the reactor is pressurized, close the gas loading valve.
7. Leak test al1 connections.
8. If leaks are detected, depressurize the reactor and tighten loose connections.
9. If leaks persist, fittings may have to be cleaned or replaced.
10. When d l the connections have been sufficiently leak tested, set the flow meter to blend mode.
11. Set flow meter to appropriate settings to provide required gas composition.
12. Adjust back-pressure valve to a point where the reactor pressure reads 2070 kPa.
13. Allow gases to flow through the reactor for 15 minutes to ensure proper gas composition.
14. Simultaneously close the gas flow to the reactor (BV1) and switch the gas outlet from the reactor (BV3) from the back-pressure valve (BPI) to a a stand-alone pressure gauge.
15. Obtain a gas sample from the reactor and confirm that the proper mix has been obtained-
16. If a proper gas mix has been obtained within the reactor, then continue. If not, allow the gases to flow through the reactor for a further 5 minutes and repeat the sampling procedure.
3.5.3 Reactor Sampling Procedure
1. n im on stirrer.
2. Tuni heat controller to desired temperature setting.
3. Start stopwatch
4. Collect liquid sample from the reactor via the liquid sampling line and store in 2 ml sample vials. Samples should be no more than 0.5 ml or roughly one quarter of the sample vial.
5. Store samples on ice until they can be tested on the TCD. Ensure that the tops are secured to prevent water from seeping into the sample.
6 . Liquid sarnpIes are collected in three minute intemals.
7. To collect a gas sample, open the gas sampling line and allow the sample to flow through the GC for 15 seconds. This will ensure that a fresh sample is collected.
8. Record the temperature and pressure within the reactor for every sampling period.
3.5.4 Reactor Shut-Down and Clean-Up Procedure
1. When reaction time is cornplete, set temperature controller to the coldest tem- perature and open the cooling water needle valve (NV13) to the full open posi- tion.
2. Cool reactor to 25OC.
3. Purge gases from reactor. This can be done by switching the gas outlet -dvc from the pressure gauge to the back-pressure valve. 'Iiirn the back-pressure valve counterclockwise until it cannot be tumed any further.
4. Ailow gases to vent from reactor for 15 minutes.
5. Turn off cold water flow.
6. Disconnect al1 line attachments to the reactor.
7. Place reactor into vice and loosen the bolts on the c-clamps.
8. Loosen and remove the O-ring followed by the c-clamps.
9. Separate the upper and lower reactor assemblies.
10. Dispose of reactor contents into the organic waste receptacle.
11. Clean glass liner in ultrasonic bath with a mild sulfuric acid solution.
12. Rinse the wetted reactor surfaces with distilled water and then with methanol to remove any remaining cupric chloride. Use scrub-bmsh to remove solid particles that may have been deposited within the reactor.
13. Dispose of al1 solvents into solvent waste can.
14. Rime rupture disc with distilled water.
15. Rime sarnpling syringe with methanol to remove any residual cupric chloride salts.
3 -5 -5 Maintenance
1. Inspect rupture disc after 10 runs.
2. Thermocouple should be visually inspected regularly. Normal turnover is 3 to 4 weeks.
3. Check al1 connections for leaks daily prior to the commencement of a run.
4. Check for leaks a t regulators weekly and whenever a cylinder is changed.
5. Check CO monitor with calibration gas monthly.
3.6 Summary
For the purposes of this thesis, it was decided that batch reactor experiments would
provide the necessary information to gauge the effects of selected process parameters
on the synthesis of DMC. A significant amount of time was required to ensure that
al1 safety concems were satisfied. With the apparatus assembled and operating pro-
cedures in place, experiments were designed, within the limitation of the apparatus,
and run to evaluate the effect of selected process parameters.
Chapter 4
Results and Discussion
Five sets of experiments were designed to examine the synthesis of dimethyl carbonate
(DMC) under Mnous rezt ion conditions. The first set employed copper exchanged
zeolites as catalysts. The second set used a combined CuC12/PdC12 system. Exper-
imental sets three to five employed CuC12 as the sole catalyst. The effects of cupric
chloride concentration and temperature were examined in sets three and four, respec-
tively. The fifth set utilized a three factor experimentd design exarnining the effects
of catalyst concentration, temperature, and molar ratio of CO to O*.
4.1 Initial Experiments with Zeolites
Initially, the synthesis of DMC was investigated using heterogeneous catalysts. The
motivation behind this lay in the premise that if a suitable heterogeneous catalyst
could be identified in batch reactor runs, the catdyst could also be used in flow
reactor runs; culminating in the use of the catalyst in a catalytic distillation unit. It
was decided from the literature review that zeolites would be a suitable heterogeneous
catalytic support for this reaction. Zeolites are suitable for such a purpose due to
their robustness, large surface area for reactions and relative simple procedures for
exchanging other chernical species ont0 the surface active sites. The use of zeolites for
the synthesis of alkyi carbonates was outlined by King (1996), Lee and Park (1991)
and Smith et al. (1992).
Three types of zeoiites were prepared for testing: an NaY zeolite (Y-zeolite with
sodium on active sites), an KY zeolite (Y-zeolite with hydrogen on active sites), and
an LaX zeolite (X-zeolite with lanthanurn pre-exchanged onto active sites). These
samples were obtained from other researchers in the department who had used zeo-
lites. These zeolites underwent a procedure outlined by Lee and Park (1991) whereby
Cu(I1) was exchanged onto the zeolites with a 0.5 M cupric chloride solution. Al-
though the procedure only required one exchange with a cupric chloride solution,
some of the exchanges were performed a number of times. Table 4.1 outlines the
experiments performed for each catalyst, the reaction conditions under which each
experiment was perfomed and the volume percent DMC produced.
Table 4.1: Summary of experiments performed using copper-exchanged zeolites as catalys ts.
Date Catalyst Quantity(g) Exchanges with CuC12 T("C) ~(6) Vol% DMC 9/13/95 Na-Y 1 .O2 3 120 2400 0.60
O2 were identical for each set. However, one set used 47 mmol/L of CuC12(Set A)
and the other used 24 mmol/L of CuClz (Set B). Liquid sarnples were collected in 3
minute intervals and gas sarnples were coIIected in 6 minute intervais. The runs used
for sets A and B are marked accordingly in Tables 4.3, 4.4, and 4.5.
Each set of replicate runs provided an estimate of the total vaxiance for each
sampling time. Ideally, one would like to pool these variance estimates, s2, to get a
single estimate for the total variance, s;oole,. Statisticd tests were performed on the
sample populations to determine if they could be pooled.
Standard F-tests were performed to analyze the data. The F-test allows one to
test the hypothesis that two variances are equal by calculating the F-ratio (4.8).
In this equation, s&,, is the larger of the two variance estirnates. The F value is
compared to tabulated F-ratio tables for the given confidence level. If the calculated
value exceeds the tabulated value, then the hypothesis is rejected and one must
conclude that given the data and confidence level, the variances are different. If the
null hypothesis is rejected, an estimate of the process variation cannot be calculated
by pooling the results. In some instances, analysis of Mnances indicated that there
were two distinct areas of variation depending on the time of reaction. An example
of this behaviour is illustrated in figure 4.1.
a - - .
O
O IO 20 30 40 50 60 Time (min)
Figure 4.1: The top figure displays the average rate of formation of DMC for the four repiicate nins at 150' C, 47 m l / L of CuCb, and a CO/02 ratio of 2. The bottom figure displays the estimated variance of the results at the Werent time int ervais .
It is apparent from this figure that a higher degree of variability amongst the
results occurs in the region where the rate of DMC formation is the geatest. Since
there appears to be two distinct areas of variation in Figure 4.1, tests were performed
on the data sets to determine if the results were statistically different.
Normally, one would divide the highest population variance by the lowest p o p
ulation va.riance. Since many of the products are not initially formed during the
experiment, variance estimates of O were used. When the highest variance population
was compared to the iowest non-zero population variance, the F-test indicated that
the nuil hypothesis should be accepted and the variances could be pooled.
The variance arnong each group was estimated using the sarnple mean for each
population at a specific time period. This is accomplished by estimating the sum of
squares for the data within each group (4.9).
Where J is the number of data points ivithin each sampling group. Dividing by the
degrees of freedom, J-1, gives an unbiased estimate of the population miance for ce11
i (Hoaglin et al., 1991).
For cases where the nul1 hypothesis was accepted, pooled estimates of the variances
were estimated by formula 4.10.
Where 1 represents the total number of sampling groups used within the estimate
and n is the number of measurements in sample group i. Since many of the products
were not initially formed in the experiment, the initial mns without any of the prod-
ucts were neglected and not considered in the calculations. Estimates for the total
variance, for both replicate sets, are displayed in Table 4.7.
Table 4.7: Total variance estimates for both sets of replicate nuis. Apart fkom the CuC12 concentration, each run was perforrned at 150°C and a 2:l molar ratio of
1 Methylal
( Methyl Formate
1 Methyl Acetate
Set A:47 mmol/L CUCI* Set B:24 mmol/L CuClz 1
J is the nurnber of data points within each sampling group
1 is the number of sampling groups
4.4.3 Process Variation
Although care was taken to ensure that al1 the runs were performed under the same
conditions, a number of factors can affect the process Mnance of a system. Fluctu-
ations in cooling water temperature and heater current are examples of extraneous
variation that can affect experimental results. It is for these reasons that it was neces-
sary to get an accurate rneasure of the process variance. Since the estimates for total
variation are a sum of the measurement and process variation, estimates of a:,,
could be obtained by subtracting the measurement variation(s&) from the total
variation estimate (sTOTAL).
4.4.4 Variation Conclusions
The estimates for al1 the components of variation are tabulated in Table 4.8. IncIuded
in this table is a percentage breakdown of the total Mnation; i.e. the percentage of
the total variation associated to process and measurement error. The components
of vanation for methylal is predominantly skewed towards measurement variation
while there is an even split between process and measurement vaxiation for methyl
formate. The high contribution of measurement variation for these compounds can
be attributed to their similar retention times in the GC. Fortunately, both of these
byproducts were formed in very srnall quantities and were rather insignificant in the
analysis. The significant contribution of variation to the rernaining products and
reactants is attributed to the process variation ( s ~ , ~ ~ ) . The main factor attributing
to the process variation was the time required to heat the reactor to the reaction
temperature. Differences in heating times by as little as one or two minutes would shift
the results. This would translate to the sampling populations for each time period
actually containing data points a t different stages in the reaction. This translates in
process variance contributions greater that 90%, as indicated in Table 4.8. Although
there is evidence of varying heat-up times for the reactor, there was no scientific
method available to accurately account for this. Therefore, the data was used as
collected.
Table 4.8: Components of variation for each compound.
As with dl experimentd work, it was necessary to perform a mass balance on the
system. A rnolar eiemental mass balance was performed on each of the experiemntal
runs. Liquid and gas samples were removed h m the reactor every 3 and 6 minutes,
respectively. Although extreme care was taken to ensure that the same sample sizes
were removed, there was indeed some variation.
Changes in reactor volume were considered in calculating the mass balances.
Changes in sample volume incurred by the dramatic decrease in temperature were
not considered. The samples were collected and stored in a cool environment pnor
to analysis. If the samples were being directly injected into the GC, an adjustment
would have been appropriate. However, the samples collected were sampled a t room
temperature and not the reaction temperature. The maximum calculateci deviation
was &5.4% for oxygen. A majority of the mass balances did not differ by more
than f 3% for hydrogen, carbon, and oxygen. John (1974) considered a differentiai
of W% satisfactory for his study of catalytic cracking. Upon consideration of the
sampling procedure used in this system, the m a s balance deviations encountered in
t hese analyses were considered satisfactory.
4.6 Effects of CuClz Concentration on DMC Syn-
t hesis
4.6.1 Effects of CuC12 Concentration on DMC Produced
Eight experiments were perfonned to gauge the effect of catalyst concentration on
the synthesis of DMC. The four quantities of cupric chloride used were 0.03, 0.61,
1.19 and 1.75 g. These amounts correspond to cupric chloride concentrations of 1,
24, 47 and 69 mmol/L, respectively. These values were selected based on Hallgren's
work which recommended catalyst quantities in the range of 0.02% to 1.5% based
on the m a s of methanol in the system (Hallgren, 1982a). Al1 mm were performed
using 150 ml of methanol, a reaction temperature of 150°C, and a 2:l molar ratio of
CO to 02. These conditions defined the base run which was repeated throughout the
experimental program.
The concentration of DMC obtained per unit time for varying catdyst concentra-
tions are displayed in Figure 4.2. Apart from the lowest concentration, the catalyst
Figure 4.2: Effect of CuClz concentration on DMC production.
concentration did not show any significant effects on the final steady-state DMC con-
centration. Using 1 mmol/L of CuC12, DMC began to appear after approxirnately 25
minutes of reaction time. A first steady state appears to have been obtained at 40
minutes. Higher quantities of DMC were then obtained as a second phase of synthesis
occurred from 60 to 70 minutes before settling on a new steady state.
Under the same conditions with higher catalyst concentrations, the rate of DMC
production was more rapid and higher concentrations were obtained. Figure 4.2 indi-
cates that DMC production began at approximately 16 minutes. The concentration
of DMC increases monotomically in al1 cases to a steady state value. The concentra-
tions obtained for the four catalyst concentrations are Listed in Table 4.9. The results
Table 4.9: Steady state DMC concentration with varying catalyst concentration. 95% con- fidence intervals (CI) are also included in the Iast column of this table to gauge the significance of the results.
[CuCl*] (mmol/L) DMC(mol/L) 95% CI 1 O .O62 z t 0.021
24 0.281 =t 0.021
47 0.262 & 0.021
69 0.275 A= 0.021
for the upper three catalyst concentrations were not significantly different.
4.6.2 Effect of Catalyst Concentration on B yproduct Forma-
tion
The only byproducts detected from this analysis were methylal, methyl formate,
methyl acetate, water and carbon dioxide. Table 4.10 summarizes the moiar quanti-
ties of al1 products formed from this reaction.
Table 4.10: Moles of reaction byproducts formed with varying catalyst concentrations. Al1 units are in moles.
methylal 0.006 0.001 0.007 0.004 ~t0.006
methyl formate 0.005 0.009 Beeting 0.000 i~0.006
methyl acetate 0.000 0.001 Reeting 0.000 ztO.001
DMC 0.008 0.039 0.037 0.039 3~0.012
H2O 0.110 0.132 0.223 0.218 3~0.112
Co2 0.014 0.020 0.015 0.018 &O .O06
The product distribution can be studied by calculating the instantaneous selec-
tivity, p, during the reaction. For a system where many byproducts are produced
from a group of reactants, the instantaneous selectivity c m be calculated for each
component by dividing the molar quantity of a compound, ND, by the total moles of
al1 undesired products, Nu (4.1 1) (Levenspiel, 1972).
Figures 4.3, 4.4, 4.5 and 4.6 display the instantaneous selectivity versus time for
experiments using 1, 24, 47 and 69 mmol/L of CuC12. Although the TCD could
detect water, the selectivities were calculated on a water free basis. A number of
contributing factors for this decision. Firçtly, the reaction did not begin in an anhy-
drous environment. The system was hydrated by the environmental conditions of the
laboratory and the water content of the catalyst (CuC12-2H20) and methanol. The
chemistry of reactions for DMC and the other byproducts did indicate that water
would indeed be formed. However, these increases could not be accurately gauged
due to the amount of water already in the system. Additional increases in the amount
of H20 in the system would be invalidated and considered insignificant due tot he
magnitude of the 95% confidence intervals and the amount of H 2 0 initially in the
system.
It is evident from these figures that the highest selectivity, approaching 100%,
of DMC occurs when using 24 and 47 mmol/L of CuC12. As CO2 forms, the DMC
selectivity for these runs falls to 65% and 63%, respectively. Transition metal com-
plexes have been shown to be suitable for the oxidation of CO to CO2 in aqueous
solutions by Byerley and Peters (1969). Romano et al. (1980) indicated that cuprous
chloride displayed a reduced selectivity towards DMC as water accumulates within
the system; this was accompanied by the formation of COz. The results shown in
. - r
O 1 O 20 30 40 50 60 Time (min)
Figure 4.3: Instantaneous selectivity versus time for experiment using I mmol/L of CuC12.
Figures 4.3, 4.4, 4.5 and 4.6 are consistent with these observations
A higher propensity for CO2 formation is initially displayed using 1 mmol/L of
CuC12. The highest selectivity for DMC is 43% and decreases to 28% following 60
minutes of reaction time. The highest attainable selectivity using 69 mmol/L of CuC12
is 74% which dissipates to 58% following the 60 minutes of reaction time.
No unexpected byproducts were detected dunng the course of the reaction. Apart
from CO2, H20 and DMC, the formation of other byproducts was very small. Of the
more volatile byproducts, only methylal was formed when using d l four catalyst
concentrations. Methyl Formate was formed using 1 and 24 mmol/L of CuCI2, with
higher quantities being formed with the higher catalyst concentration. However,
methyl formate was only detected in fleeting quantities using 47 mmol/L of CuC12
and was not detected in experiments using 69 mmol/L of CuC12. Similarly, methyl
acetate was detected using 24 mmol/L of CuCl2 and it was only found in very small
quantities with 47 mmol/L of CuC12.
r----- 1 1 Acetat /
O 10 20 30 40 50 60 Time (min)
Figure 4.4: Instantaneous selectivity versus tirne for experiment using 24 mmol/L of CuC12.
Figure 4.5: Instantaneous selectivity versus time for experirnent using 47 mmol/L of CuC12.
O 10 20 30 40 50 60 Tirne (min)
Figure 4.6: Instantaneous selectivity versus time for experiment using 69 mmol/L of CuC4.
Effect of Catalyst Concentration on Reactant Conver-
sion
The instantaneous conversion is a measure of the amount of reactant converted to
other products at a particular time in an experiment. The conversion is calculated by
dividing the amount of a reactant at time t by the initial quantity of reactant (4.12).
Where NAO and Na are the molar quantities of the species initially and at the time
period of interest, respectively. Figures 4.7, 4.8 and 4.9 displays the instantaneous
conversion of methanol, oxygen and carbon monoxide for the four catalyst concentra-
tions tested. The points on the figures are the actual data points whereas the lines
were fitted using a cubic spline and a smoothing function.
7 mmo X -- 69 mmo
O 10 20 30 40 50 60 Time (min)
Figure 4.7: Conversion of MeOH as a function of varying cataiyst concentration
Time (min)
Figure 4.8: Conversion of O2 as a function of varying catalyst concentration
+ - - - 47 mmo 69 mmo
O 10 20 30 40 50 60 Time (min)
Figure 4.9: Conversion of CO as a function of varying catalyst concentration
Methanol was not a lirniting reagent in this experiment since i t was supplied in
excess. Previous studies by Romano et al. (1980) indicated that excess methanol did
not inhibit or affect the reaction in any way. The methanol conversion for the four
catalyst concentrations were 26%, 31%~ 31% and 30% for 1, 24, 47 and 69 mmol/L
of CuC12. Following 60 minutes of reaction tirne, the conversion of O2 was 48%, 720,
94% and 96% for 1, 24, 47 and 69 mmol/L of CuC12, respectively. The corresponding
conversions for CO were 27%,72%,77% and 88% for the same concentrations. These
results are surnmarized in Table 4.11.
Table 4.11: Reactant conversion for w i n g catalyst concentration.
1 Species 1 mmol/L 24 mmol/L 47 mmol/L 69 mmol/L 95% CI Methanol 26 31 31 30 2
CO 27 72 77 88 19
O2 48 72 94 96 11
It is interesting to note the similarities in Figures 4.8 and 4.9. Further analysis of
this data indicated that the CO and O2 reacted in a 2 to 1 ratio, as indicated by the
D MC reaction stoichiometry; it decreased in the same manner.
One can also consider the yield of the reaction which is the ratio of the molar
amount of product forrned divided by the amount of specific reactant reacted (4.13).
Table 4.12 outlines the yield of DMC and CO2 based on the conversion of CO. The
yield of methanol to the products was not considered due to the large excess of
methanol in the system. Under these reaction conditions, the yield of DMC based on
CO conversion was only significantly different for the lowest catalyst quantity used
(1 mmol/L). However, the yield of COz based on CO conversion is significantly lower
with 24 as opposed to 47 mmol/L of CuC12.
Table 4.12: Yield of DMC and CO2 based on CO conversion. The 95% confidence intervals (CI) were calculateci ftom replicate r u s .
4.6.4 Summary of the Effects of Catalyst Concentration
Under the conditions in which these experiments were perfomed, there was no signif-
icant difference between the quantities of DMC produced for the upper three catalyst
concentrations. The byproducts produced with varying catalyst concentrations did
differ. However, apart from CO2, the quantities of al1 other byproducts were negligi-
ble.
Runs performed using 1 mmol/L of CuClz produced a significantly reduced amount
of DMC when compared to the other catalyst concentrations. The highest instants-
neous DMC selectivity for the lowest concentration of CuC12 was 43%. However, the
small quantity of CuCl2 was large enough to promote the oxidation of CO to CO2.
A higher propensity for COz formation was displayed in this case. Conversely, a t 69
mmol/L of CuC12, the highest instantaneous DMC selectivity obtained was 74%. The
quantity of DMC synthesized a t this concentration did not differ significantly from
the middle concentrations. Although the selectivity a t 60 minutes was very similar to
that for the other CuClz concentrations, the lower initial DMC selectivity can again
be attributed to the CO2 formation. The conversion of CO and O2 increased with
catalyst concentration. Kowever, considering the 95% CI, the yields of DMC for the
upper three catalyst concentrations were not significantly different at steady-state.
The yield of CO2 a t steady-state was significantly higher with 47 mmol/L of CuC12
as opposed to 24 mmol/L of CuC12.
Studies by Romano et al. (1980) used CuCl to catalyze the same reaction. R e
mano et al. (1980) discovered that cupric methoxy chlonde, Cu(OCH3)C1 was an in-
termediate in the synthesis of DMC. When using different quantities of Cu(OCH3)C1,
Romano found that the quantity of DMC increased with the amount of Cu(OCH3)Cl
used. Cu(OCH3)Cl is unstable and difficult to detect. Romano was able to study its
formation by examining the reduction and the oxidation stages of this reaction s e p
arately, as well as together. In this investigation, we only considered the integrated
system and as a result, we did not detect Cu(OCH3)Cl. In their integrated system,
Romano et al. did not find any advantages of using increasing quantities of CuCl on
the rate of formation of DMC.
CuCl and CuC12 are both corrosive salts. Romano et al. (1980) reported that
corrosion problems associated with these chernicals were not as severe as predicted.
However, observed deterioration of the stirrers and problems associated with the
rupture discs would dispute this claim. Throughout the course of the experiments, the
chloride solution was detrimental to the internal parts of the reactor. Thermocouples
had to be replaced at least every four weeks and there was noticeable deterioration
of the impellers and impeller rods. Dunng the preliminary runs, the copper chloride
solution severely corroded the rupture discs used in the reactor.
Considering the problems associated with the catalyst system, it would be ben-
eficial to reduce the catalyst concentration as much as possible. However, one does
not want to reduce the concentrations so much that the system requires long reac-
tion times. Identifying the threshold catalyst concentration would be beneficial as
this would limit the amount necessary and reduce corrosion problems associated with
halide compounds. From the results of this analysis, and considering the conditions
that these experiments were run, the ideal catalyst concentration will lie sornewhere
between 24 and 47 mmol/L of CuCI2. At 150°C, it is suggested that 24 mmol/L of
CuClz be used since the yield of CO2 based on CO conversion is significantly lower
than with 47 mmol/L of CUCI*. There is no advantage of using 69 mrnol/L of CuClz
as there is not a significaat increase in final steady-state concentration of DMC.
4.7 Analysis of Temperature Effects with CuClz
The goal of the second phase of experiments was to determine the effect of temperature
on the production of dimethyl carbonate. The four temperatures selected were 100,
125,150 and 175°C. Each of these temperatures was evaluated at two different catalyst
concentrations; 24 and 47 mmol/L of Cu&. These catalyst concentrations were
chosen for two reasons. Firstly, results from the first set of experiments indicated
that 1 mmol/L of CuClz did not provide quantities of DMC comparable to the upper
catalyst quantities. Secondly, cupric chloride is very corrosive. Since there was no
apparent advantage of using 69 mmol/L over 24 and 47 mmol/L of Cu&, tests
involving 69 mrnol/L of CuClz were not considered.
4.7.1 Effect of Temperature on the Synthesis of DMC
Figures 4.10 and 4.11 displays the effect of temperature on the production of DMC
with 0.61g and 1.19g of cupric chloride, respectively.
Apart from the run at 150°C, figures 4.10 and 4.11 show that the rate of formation
of DMC is geater, for al1 temperature settings, using 47 mmol/L of CuC12. This is
evident kom the rate of change of DMC concentration to its steady-state value for the
runs using 47 as opposed to 24 mmol/L of CuCL2. In both cases, the production of
DMC at 175°C falls well below the quantities obtained at lower temperatures. Table
4.13 contains the concentrations of DMC obtained for al1 temperatures a t the two
catalyst concentrations.
In the set of experiments using 24 mmol/L of CuC12, there was a significant
Figure 4.10: Effects of changing temperature on the production of DMC w of CUCI*.
difference in the amount of DMC produced after 60 minutes of reaction time under d l
the temperatures, excluding 175°C. The highest concentration of DMC was obtained
at 150°C for this case. In the set of experiments using 47 mmol/L of CuC12, there was
also a significant difference in the amount of DMC produced a t d l temperatures. The
highest quantity of DMC was produced a t 125OC, followed by 150, 100 and 175°C.
When compared on a temperature to temperature b a i s for different catalyst con-
centrations, the results were significantly different in d l cases except a t 150°C. At
100 and 125"C, the amount of DMC produced was greater when using 47 mmol/L of
CuCl2, but insignificantly different a t 175OC. These results are summarized in Table
4.13. An explanation of the behaviour a t 175°C will be eluded to in the upcoming
sections.
Figure 4.11: Effects of changing temperature on the production of DMC with 47 mmol/L of CuC12.
Table 4.13: DMC concentration produced using 24 and 47 of CuC12 and four dinerent tem- peratures. (following 60 minutes of reaction time)
Temp("C) DMC (24 mmol/L CuC12) DMC (47 mmol/L CuC12) 95% C I
100 0.135 0.207 3= 0.021
125 0.107 0.314 z t 0.021
150 0.258 0.245 =t 0.021
175 0.127 0.173 I 0.021
4.7.2 Effects of Temperature on Byproduct Formation
No unexpected byproducts were produced a t the dserent reaction temperatures.
Tables 4.14 and 4.15 outline the quantities of products produced during the reactions
a t different temperatures.
Table 4.14: Moles of byproducts produced using 24 mmol/L of CuC12 and four different temperatures. Note that the results following both 60 and (90) minutes of reaction time are listed.
S pecies iOO(OC) 125(OC) 150("C) 175("C) 95% CI
met hylal O.OOO(O.OOO) O.OOO(0.0023) 0.002 (0.005) 0.002 0.006
Table 4.19 outlines the yield of DMC and COz based on CO conversion for the nins
performed in this set of experiments. It is clearly illustrated from this data that for
both cases, there is a significant decrease in the yield of DMC and a significant increase
in the yield of CO2 based on the CO conversion. As indicated earlier, the quantity
of DMC produced difFered with changes in temperature and catdyst concentration.
However, between each data set, the results obtained for 175°C were similar for both
Table 4.19: Yields of DMC and CO2 based on CO conversion for different cataiyst concen- trations and temperature.
catalyst concentrations. There is no indication that a change in reaction mechanism
accounts for this. Rather, a competing reaction, that being the oxidation of CO to
COz, dominates at this temperature. This assumption is supported by the reduced
quantities of DMC produced at 175OC, regardless of catdyst concentration, and the
signiticantly higher yield of CO2 based on CO conversion.
Temperature
100
24 mmol/L CuCL2 47 rnmol/L CuClz
YDMC Yco2 26.6 1.2
YDMC &oz 41.2 1.6
4.7.5 Arrhenius Expression Parameter Calculation
Romano et al. (1980) concluded that a t a constant temperature, the rate of dirnethyl
carbonate production is proportional to the carbon monoxide pressure. Under the
conditions where this relationship holds, the rate of DMC formation (rDMC) can be
represented by a first order rate expression (4.14).
Where k is the reaction rate constant and Pco is the partial pressure of CO. This
relationship allowed for the temperature dependence of this reaction to be evaluated.
The temperature dependence for a chernical reaction can often be represented by
Arrhenius' law (4.15).
Where ka is the pre-exponential factor and E is the activation energy for the reaction.
This relationship, for experiments perfonned under various temperatures and 47
mmol/L of CuCI2, is illustrated in Figure 4.14. The values of k were calculated by
linear regressing the data points a t the four temperatures. Table 4.20 contains the
values of k for both catalyst concentrations at the four temperatures examhed in this
analy sis.
Table 4.20: Reaction rate constants (k) for both catalyst concentrations at the four tern- peratures of interest. The units for k are in mol/(h L Atm).
T(OC) k (using 24 rnmol/L CuC12) k (using 47 mmol/L CuCl2)
100 0.013 0.065
125 O. 150 O. 129
150 0.617 0.422
175 0.302 0.248
6 a IO Pco (atm)
Figure 4.14: Rate of DMC production versus the partial pressure of carbon monoxide for four dinerent temperatures for 47 mmol/L of Cu&.
Using expression (4.14), Arrhenius plots were generated for both sets of experi-
mental rms. Four runs, each a t different temperatures (100, 125,150 and 175°C) for
two catalyst concentrations (24 and 47 mmoI/L) were performed. The reaction rate
constants (k) were calculated from each data set. The Arrhenius' plots for the data
using 24 and 47 mmol/L of CuClz are illustrated in Figures 4.15 and 4.16, respectively.
The activation energies and preexponential factors for both sets of data are listed
in Tables 4.21 and 4.22, respectively. In both sets of runs, the results indicated poor
DMC synthesis at 17S°C. In light of this behaviour, the Arrhenius expression was fit
to the data both with and without the data obtained at 175°C.
The activation energy required for the runs pedormed with 24 mmol/L of CuCIz
is more than twice that of the activation energy calculated for the runs perforrned
Figure 4.15: Arrhenius' plots for experiments using 24 mmol/L of CuC12.
Table 4.21: Activation energies for experiments at 4 different temperature and 2 catdyst concentrations
[CuC12] (mrnol/L) Comrnents Activation Energy(kcallmo1) 95% CI 24 Al1 Data 17.55 k0.38
5.2.1.2 Effect of Temperature and CO/Oz Ratio on By-Product Produc-
tion
The Gibbs reactor simulation did not require a kinetic data or reaction stoichiometry.
A list of reactants and expected products were al1 that was required. The simulations
predicted that apart from DMC, the only byproducts that would be formed in this
reaction are H 2 0 and COz. Figures 5.3 and 5.4 display the quantities of CO2 and H20
produced for various reaction temperatures and ratio of CO/Oa Figure 5.5 displays
the conversion of CO for various reaction temperatures and ratios of CO/02.
Figure 5.3: Production of COz with varying reaction temperature and ratio of CO/02.
The following conclusions can be drawn frorn these results:
1. For a temperature range of 100 to 120°C, higher quantities of H20 are produced
with a CO/Oz ratio of 1.
2. For a temperature range of 100 to 12U°C, higher quantities of CO2 were obtained
for a CO/02 ratio of 2.
3. There was no significant difference in the production of CO and Oz frorn 125 to
175°C for CO/02 ratios of 1 and 2.
Figure 5.4: Production of H 2 0 with varying reaction temperature and ratio of CO/02.
4. Dramaticdly greater quantities of DMC and H 2 0 were produced for the entire
temperature range for a CO/Oz ratio of 3. The quantity of HzO produced varied
directly with temperature while the quantity of CO2 was inversely proportional
to temperature.
5. For al1 molar ratios of CO/02, CO was produced at 170 and 175°C.
6. On a molar cornparison, the quantities of H20 produced were on a 1:1 basis or
greater when compared to DMC.
7. The quantity of CO2 produced was considerably less than the rnolar quantities
of DMC produced.
These results indicated that the quantity of CO2 was not significantly affected by
reaction temperature. Unfortunately, these simulation results do not compare directly
with the results observed from previous experiments. In the experimental investiga-
tion, the amount of CO2 produced increased with increasing reaction temperature.
One must not forgot to consider that these simulations are based on minimizing the
Gibbs free energy of the system. The effect of catalysts on the system are not a
Figure 5.5: CO conversion with varying reaction temperature and ratio of CO/Oz.
consideration. It has already been documented by Byerley and Peters (1969) that
CuClz is suitable for the oxidation of CO to CO2. Romano et aL(1980) also indicated
that although CuCl2 is initially very selective towards DMC, the DMC selectivity
fails as H20 is produced as a byproduct- This is followed by an increase in the
CuClz selectivity towards CO2. Honrever, the results do show that CO2 formation is
thermodynamically favoured for a reactant gas ratio of 3.
5.3 Chapter Summary
The Gibbs reactor simulation confirmeci and further helped to explain the previous
experimental results. Between 120 and 160°C, temperature had a negligible effect on
the production of DMC. This leads us to the conclusion that differences observed in
previous experiments at 125 and 150°C can be attributed to the catdyst, which was
not considered in the simulations. The simulations did indicate that DMC was not
thermodynamically favoured at temperatures above 170°C. This decrease in DMC
formation was evident in runs perforrned at 175OC. Quantities of DMC produced
increased with an increasing ratio of C0/02; indicating the importance of significant
CO partial pressure in the synthesis- The simulations did indicate that the formation
of CO was thermodynamically favoured at 170 and 175'C. This waç not evident
in the experiments as CO2 and H20 were predominantly synthesized a t these high
temperatures. Mthough useful, analysis of the simulation data stresses the point
that simulation data is just that, simulation data and may not provide an accurate
account of a system. However, as with this system, they did help in explaining the
reaction behaviour .
Chapter 6
Conclusions and Recomrnendat ions
6.1 Conclusions
The synthesis of dirnethyl carbonate (DMC) with cupric chloride based catalysts
was investigated in a 300 ml parrTM batch reactor. Experïments were conducted
with CuC12 exchanged zeolites, a combined PdCI2/CuCl2 system and CuC12 as a sole
catalyst. The results from these experiments are summarized below.
6.1.1 CuC12 Exchanged Zeolites
Batch reactor runs using CuClz exchanged zeolites were unsuccessful as insignificant
quantities of DMC were synthesized. In the few cases that DMC was synthesized, in
quantities l e s than 1 vol% DMC, the results could not be reproduced.
6.1.2 Combined PdC12/CuC12 Catalytic System
The combined system employing PdC12 and C d 2 had been previously examined in
literature by Mador et al. (1963). Unlike Mador et al., who performed the experiments
in the absence of oxygen or air, these experiments were carried out in the presence of
oxygen. Runs involving this system uTere unsuccessful as DMC was not synthesized
under various reaction conditions and significant quantities of a substance believed
to be palladium oxide, PdO, covered the wetted internai surfaces of the reactor and
blocked the liquid sampling lines. The only byproducts detected from this system were
COz and H20; both of which were oxidation byproducts catalyzed by the PdC12.
6.1.3 CuCl2 as Sole Catalyst
Initial scouting runs using CuC12 showed a higher selectivity towards DMC than those
using the cornbined PdC12/CuC12 system. In addition to its ability to synthesize
DMC under the desired operating condition, CuC12 is much more economical than
PdC12 (0.22 $/g as opposed to 23.06 $/g, respectively). Three sets of experiments
were successfully designed and implemented within the operating parameters of the
experimental apparatus to examine the effects on the synthesis of DMC. The results
of these experiments are suwnarized below.
1. The four CuCl2 concentrations examined were 1, 24,47, and 69 mmol/L. Under
the conditions in which these experiments were performed, the yields of DMC
for the upper three catalyst concentrations were not significantly different a t
steady-state. An average DMC concentration of 0.27 moI/L was produced a t
these concentrations.
2. Apart from CO2, the quantities of al1 other byproducts were negligible. The
quantity of COz increased with CuClz concentration. The DMC selectivity of
the system decreased as time progressed and CO2 formation rapidly increased.
3. At 150°C, 24 mmol/L of CuC12 would be a more suitable concentration since the
yield of CO2 based on CO conversion is significantly lower at this concentration
than with 47 mmol/L of CuCi2.
4. Two sets of experiments, each with a different concentration of CuC12, 24 and
47 mmol/L of CuC12, were run at four temperatures; 100, 125, 150 and 175°C.
The highest concentration of DMC, 0.314 rnol/L, was obtained with 47 mrnol/L
of CuC12 and a reaction temperature of 125°C.
5. At 150°C, there was no significant difference in the quantity of DMC produced
(0.25 mol/L) with Mlying catalyst concentrations. Significantly lower quantities
of DMC were produced at 175°C when cornpared to the other temperatures.
6. With the available temperature data, an Arrhenius' expression was fitted and
the appropriate constants were obtained. At 175"C1 the data did not seem to
adhere to the linear relationship proposed. The calculated activation energies
were 17.55 md 6.59 kcal/mol for 24 and 47 mol/L of CuC12. Although a direct
cornparison with literature values was not a d a b l e for this system, a similar
system provided a calculated activation energy of 8 kcal/mol. The results cal-
culated in this experiment were within this magnitude.
The results of the experimental design indicated a reactant gas ratio of 2:l for
CO to O2 should be used. During the reaction, 2:l ratio of CO:02 is maintaineci.
Romano et al. (1980) found that during the synthesis phase of the reaction,
the rate of formation of DMC is proportional to the partial pressure of CO.
Diluting the reactant gas concentration with excess oxygen reduces the CO
partial pressure, which in turn affected the formation of DMC. hcreasing the
reactant gas ratio increases the rate of DMC formation. Unfortunately, this
is accompanied by an increase in the amount of CO2 produced. This can be
partially offset operating at a lower temperature (125°C) and a higher catalyst
concentration (47 mmol/L CuC12).
8. Gibbs reactor simulations showed that between 120 and 160°C, temperature had
a negligible effect on the production of DMC. This lead to the conclusion that
differences observed in previous experiments at 125 and 150°C can be attributed
to the catalyst, which was not considered in the simulations. The simulations
did indicate that DMC was not thermodynamically favoured at temperatures
above 1'70°C. This decrease in DMC formation was evident in runs performed at
175°C. The quantities of DMC produced increased with higher ratios of CO/Oz;
indicating the importance of significant CO partial pressure in the synthesis.
6.2 Recommendations
Considering the physical limitations of the experimental apparatus, the objectives
of the program were met. Future experiments to investigate the synthesis of DMC
should be continued with a heterogeneous catalyst. The motivation behind this work
was based on the eventual production of DMC via catalytic distillation. A suitable
heterogeneous catalyst must be identified to take the project to the next step. Despite
the fact that zeolite based heterogeneous catalysts for this synthesis were prepared by
procedures outlined in literature, DMC could not be produced. Synthesizing catalysts
is time-consuming, tedious and very difficult. Atternpts should be made to obtain
industrially produced copper exchanged catalyst that can be used to investigate their
suitability in this process.
The experiments in this investigation were carried out in a batch reactor. There-
fore, the effect of pressure could not be adequately considered. The pressure fell
as the reactants were consumed. Attempts were made to operate the system as a
flow system. However, problems were encountered as the reactor contents were also
removed from the system. Further investigations should be carried out in a proper
stirred tank reactor or a flow reactor. The effects of operating conditions on the
synthesis of DMC could then be more completely examined. hirther Gibbs reactor
simulations should also be performed to examine the effect of increasing pressure on
the synthesis of DMC.
Appendix A
Health and Safety Concerns
The experimental aspect of this research had many inherent safety concerns which had to be dealt with pnor to any experimentation. The experimentai apparatus was installed within a containment bunker to reduce the possibility of incidents that may threaten the welfare of those performing the experiments. The bunker provided two main purposes; it shielded individuals from debris in the event of a runaway reaction and it provided a barrier to protect individuals from CO gas le&. Apart from CO, the only other reactants and products which could be h a n n N in sufficient quantities were methanol, methyl formate, methylal and dimethyl carbonate. Standard laboratory procedures required the mandatory use of lab coats, safety glasses and rubber gloves whenever handling these materials. Listed below in Table A. l is a list of the chernicals used in the experiments, hazards associated with these chemicals and the suggested safety requirements. The information displayed in Table A S was obtained from the MSDS sheets for the respective chemicals
Table A.1: Chernids used in experiments with associated hazards and safety requirements
Dimethyl Carbonate ( c ~ H ~ & ) -~&and skin irritant Latex Gloves/Eye Protection
Methyl Acetate (CH3COOCH3) Eye and skin irritant Latex Gloves/Eye Protection
Methyl Formate (C2H402) Eye and skin irritant Latex Gloves/Eye Protection
1 O X Y F (02) FlammabIe Use in weil-ventilated area 1 Carbon Monoxide (CO) asphyxiant Respirator
Use in weU-ventilated area - ---
Cupric Chloride (CuC12) Corrosive Latex GIoves
Appendix B
Sample PRO111 Input File
Table B.l contains a sample Ming of a PRO/II keyword input file. The simulations were nui using a windows based version of PRO/II. Temperature and reactant gas ratio were the only factors which were changed between each run.
Table B.1: Sarnple PRO/II Keyword Input File
Generated by PRO/II Keyword Generation System jvenion 2.71 - 02-14-95i
The proceeding tables contain the numerical data obtained fkom the PRO111 Gibbs reactor simdat ions.
Table C.1: Quantities of reactant and byproducts produced in Gibbs Reactor simulations with a CO/02 ratio of 2 and various temperatm. AU quantities are in Ibmols. Negat ive(-) units for reactants indicate quantity of reactant consurned.
1 Temp("C) DMC CHSOH H 2 0 O2 CO con 1 Initial 0.0000 93.0526 4.3374 0.8574 1.7526 0.0000
Table C.2: Quantities of reactant and byproducts produced in Gibbs Reactor simulations with a CO/02 ratio of 1 and various temperatures. Ail quantities are in lbmols. Negative(-) units for reactants indicate quantity of reactant consumed.
Temp("C) DMC CH30H H20 O2 CO Co2
Initial 0.000 93.8931 4.3766 0.8651 0.8651 0.0000
Table C.3: Quantities of reactant and byproducts produced in Gibbs Reactor simulations with a CO/02 ratio of 3 and Mnous temperatures. Basis:100 lb-mol/hr, all quantities are in Ibrnols. Negative(-) units for reactants indicate quantity of reactant consumed.
Table C.4: Quantities of reactant and byproducts produced in Gibbs Reactor simulations with a CO/02 ratio of 2, varions temperatures and no HzO in the sample ini- t idy . Al1 quantities are in Ib-mols. Negative(-) units for reactants indicate quantity of reactant cousumed.
Bennethum, J. E. and Winsor, R. E- (1991). Towards improved diesel fuel, SAE Conference Paper 912325, Society of Automotive Engineers.
Bhattacharya, A. (1995). Fuel oxygenates: Organic carbonate synthesis, Ameriuzn Chemical Society : Division of Fuel Chemistry 40: 119-123.
Byerley, J. and Peters, E. (1969). Kinetics and rnecha.nisms of the reaction between carbon monoxide and copper(ii) in aqueouç solution, Canadian Journal of Chem- istry 47: 313-320.
Cassar, L. (1989). Dimethylcarbonate: A new intermediate for a cleaner future, Chimica Italiana C I X I '89 .
Cowley, L. (1993). The effect of fuel composition on emissions fiom a range of heavy- duty diesel engines, Coordtnating European Council Paper (CEC/93/EF03).
Craig, B. D. (1989). Handbook of Corrosion Data, ASM International, Metals Park, OH.
Cumutt, G. L. and Harley, A. D. (1986). Copper catalyzed oxidative carbonylation of methanol to dimethyl carbonate, Technicol report, Dow Chemical Company.
DeGarmo, J., Paruleka., V. and V-Pinjala (1992). Consider reactive distillation, Chemical Engineering Progress 88 (3) : 43-50.
Durrant, P. J. and Durrant, B. (1962). Introduction To Aduanced Inorganic Chem- Lstry, Longmans, Toronto.
Fang, S. and Fujimoto, K. (1996)- Direct synthesis of dmc from carbon dioxide and methanol catalyzed by base, Applied Catalysis 142(1): 1-3.
Graziani, M., Uguagliati, P. and Carturan, G. (1971). Carbonylation of ethanol promoted by palladium chloride, Journal Of Organometallic Chemistry 27: 275- 278.
Hallgren, J. E. (l982a). US. patent 4,360,477.
Hallgren, J. E. (1982b). U.S. patent 4,361,519.
Hoaglin, D. C., Mosteller, F. and Tukey, J. (1991). Fundamentab of Ezploratory Analysis of Variance, John Wiley & Sons Inc., Toronto.
Horrocks, R. (1994). Light-duty diesels - an update on the emissions challenge, Proc Instn Mech Engrs 208: 289-298.
Kanne, D. (1991). U.S. patent 5,004,480.
Kanne, D. and Iwamoto, R. (1988). A novel approach to the control of heavy-duty diesel particulate emissions, SAE Conference Paper 880634, Society of Automo- tive Engineers.
Karas, L., Kesling, H., Liotta, F. and Nandi, M. (1994). Low emmision oxygenated diesel fuel, Technical report, ARC0 Chemical Company.
King, S. (1996). Reaction mechanism of oxidative carbonylation of methanol to dimethyl carbonate in CU-y zeolite, Journal of Catalysis 161(2): 530-538.
Knifton, J. F. and Duranleau, R. (1991). Ethylene glycol - dimethyl carbonate co- generation, Journal of Molecular Catalysis 67(3): 389-399.
Koyama, T., Tonosaki, M., Yamada, N. and Mori, K. (1992). European patent O 528 498 a2.
Kreucher, W. M. (1994). Clean fuels and clean cars: Will they match?, Preprints of the 208th National Meeting, Arnerican Chemical Society, Washington D. C., August 21-26, Vol. 39, pp. 507-512.
Lee, M.-Y. and Park, D. (1991). Dioxygen Activation and Homogenwus Catalytic &dation, Elsevier Science Publishers, Amsterdam.
Levenspiel, 0. (1972). Chemical Reaction Engineering, John Wiley & Sons, Toronto.
Liotta, F. J. and Montalvo, D. M. (1993). The effect of oxygenated fuels on ernmisions from a modern heavy-duty diesel engine, SAE Conference Paper 932734, Society of Automo tive Engineers.
Mador, 1. 1. and Blackman, A. (1963). U.S. patent 3,114,762.
Mason, R. L., Gunst, R. and Hess, J . (1989). Statistical Design and Analysis of Experirnents, John Wiley & Sons, Toronto.
Matuzaki, T. and Simamura, T. (1993). U.S. patent 5,214,184.
Murayama, T., Zheng, M., Chikahisa, T. and Y.Oh (1995). Simultaneous reductions of smoke and nox from a di diesel engine with egr and dimethyl carbonte, SAE Confernece Paper 95251 8, Society of Automotive Engineers.
Nikanjam, M. (1993). Development of first carb certified alternative diesel fuel, SAE Conference Paper 930728, Society of Automotive Engineers.
Okada, Y., Kondo, T. and Asaoka, S. (1995). Dimethyl carbonate production for fuel additives, Amencan Chernical Society : Division of Fuel Chemistry, Vol. 39, pp. 359-364.
Piel, W., Karas, L. and Kesling, H- (1994). Chernical use in reformulated fuels and potential impact on the petrochemical industry, Preprints of the 208th National Meeting, A men'can Chernical Society, Washington D. C., Augv.et 21 -26, Vol. 39, pp. 520-523.
Purvis, D. (1992). Personal correspondence.
Romano, U., Tesei, R., Cipriani, G. and L-Micucci (1980). US. patent 4,218,391.
Romano, U., Tesei, R., Mauri, M. and Rebora, P. (1980). Synthesis of dimethyl carbonate from methanot, carbon monoxide, and oxygen catalyzed by copper compounds, Ind. Eng. Chem. Prod. Res. Deu. 19: 396-403.
Sawicki, R. A. and Chaftez, H. (1987). US. patent 4,689,422.
Schrauzer, G. (1971). Tkansition Metals an Homogeneous Catalysis, Marcel Dekker, Inc., New York.
Sihota, M. (1995). Environmental protection.
Smith, D. W. and Landau, S. (1992). US. patent 4,171,874.
Spreen, K. B., Ullman, T. and Mason, R. (1995). Effects of cetane number, aromatics, and oxygenates on emissions from a 1994 heavy-duty diesel engine with exhaust catalyst , SAE Confernece Paper 950250, Society of Automotive Engineers.