Dr.Wynne:
This is the final report for our Senior Design project. It covers both semesters and all phases of the project in detail. Also discussed are things we would have liked to accomplish but couldn't, due to time or material constraints. These can serve as recommendations for future Senior Design groups who may work on this project.
Yours respectfully,
______
Michelle Halye
______
Michael Hanks
______
Krunal Patel
A proposed design for the preparation of
3-methyl-3-hydroxymethyl oxetane
Michelle Halye, Michael Hanks,and Krunal Patel
VCU Chemical Engineering
April 13th, 2009
Table of Contents
Executive Summary...... 3
Introduction...... 5
Schedule ...... 8
Procedures...... 9
Design of Experiment...... 12
Data Analysis...... 14
Design...... 16
Design Considerations...... 25
Economic Analysis...... 27
Conclusion and Recommendations...... 34
Acknowledgements...... 35
References...... 35
Appendix A- DSC Graphical Data...... A-1
Appendix B- TGA Graphical Data...... B-1
Executive Summary
The following report concerns itself with the production of 3-methyl-3-hydroxymethyl oxetane, a di-substituted, 4-membered cyclic ether. The initial design proposal centered around the design of a polymerization process, in which 3-methyl-3-hydroxymethyl oxetane would be polymerized into a poly-[AB]-oxetane co-polymer via cationic ring-opening polymerization. This design would be accompanied by a Design of Experiments procedure to determine the key variables in the process which would allow for maximum yield and maximum molecular weight.
Due to a lack of raw materials, however, the project was unable to move forward as planned. Supply of the monomer was limited, and a sufficient quantity to perform the experiments was not available. This being the case, it was decided to undertake the creation of the monomer itself, as the raw materials for the monomer process are readily available in bulk quantities. This would still be accompanied by a Design of Experiment to determine the key variables for the process.
The full suite of designed experiments has been performed, plus a few extra just for good measure. The process has been fraught with mild peril, and the results have not always been straightforward. Nonetheless, the project team feels that they have successfully ascertained the key variables. DSC and GC-MS data has also been obtained, though the GC-MS data is cluttered with ghost readings and residues from previous experiments.
Due to time constraints and a lack of reliable GC-MS data, we were unable to determine the reaction kinetics. A best estimate has been made to approximate the reaction order and reaction constants.
A piping and instrumentation design has been provided, as well as an economic analysis. The analysis suggests that this product is best made with pre-existing, underutilized equipment. Building the process from scratch simply to produce this particular chemical is economically unfavorable.
Michelle Halye
Michael Hanks
Krunal Patel
Introduction
Dr. Kenneth J. Wynne of the VCU Chemical Engineering department specializes in the study of material surface behaviors. Polyoxetane polymers and co-polymers have been a rich area of investigation for surface science, as these polymers possess many interesting properties. Of particular interest are poly-[AB]-oxetane co-polymers, where A and B are functional groups which provide desirable surface properties. When these polymers are used as polymer surface modifiers (PSMs), they can confer hydrophobicity, hydrophilicity, oliophobicity, and even anti-microbial properties.
Figure 1: Co-polyoxetane structure
Initially, the proposed project would have involved creating this polymer in a HEL Simular™ reaction calorimeter. However, the raw materials were unavailable in the bulk quantities required to perform a minimum of eight experiments in the liter-sized reactor. This necessitated a change in plans. Rather than make the polymer, it was decided to work on creating 3-methyl-3-hydroxymethyl oxetane, the monomer from which the polymer is made. This decision was made partly because the raw materials for the production of monomer are cheap and readily available, and partly because Dr. Wynne’s group can use the monomer in further polymer production.
Figure 2: 3-methyl-3-hydroxymethyl oxetane
Attempts to quantify the amount of product needed were based on its projected use as an antifouling paint coating for ships at sea. First, the total paintable surface area of all of the ships in the U.S. Navy was estimated, using public data on ship dimensions and the current makeup of the active Navy ships. This gave us an estimated 3.5 million sq. ft.
The total tonnage of the U.S. Navy is approximately the same tonnage as the next 17 largest navies combined.1 So we can assume that the surface area of those 17 navies is roughly equivalent to our estimation for the surface area of the U.S. Navy. We may further assume that the remaining navies on Earth account for no more than twice this number. As the original estimation for surface area of the ships of the U.S. Navy was ~3.5 million sq. ft., these assumptions would bring the surface area of all the military naval vessels to a total of ~14 million sq. ft.
Using data provided by the list of Merchant Marine capacity by country at Wikipedia2, another estimated ~10 million sq. ft. of surface area can be added to the total, for a final estimate of ~24 million sq. ft. The initial estimate of ~3.5 million sq. ft. of surface area required ~10 gallons of monomer to be made, so ~24 million sq. ft. of surface area would require ~70 gallons of monomer.
The reaction of interest to this project is a variation of one proposed by Dexter B. Pattison in his 1957 paper.3The reaction starts with trimethylolethane and diethyl carbonate. The reactants are heated, which causes the carbonate ester to displace two of the three alcohols on the trimethylolethane. These hydroxyl ions attach to the ethyl groups from the diethyl carbonate. The following images show Pattison's suggested process.
Step 1:
Step 2:
Figure 3: Reaction Mechanism
The intermediate formed at the end of step 1, 5-(hydroxymethyl)-5-methyl-1,3-dioxan-2-one, is suggested by Pattison's paper, but not explicitly stated.
Scheduling
The following tables show the projected schedules for both semesters and then the task list for the second semester.
Table 1: 1st semester Schedule
Task / Duration / Start Date / Completion DateFinish possible research / 5 days / 10/20/2008 8:00 / 10/24/2008 17:00
Calc. Mass Balances / 7 days / 10/24/2008 8:00 / 11/3/2008 17:00
Draw Basic Block Diagram / 2 days / 11/4/2008 8:00 / 11/5/2008 17:00
Size Equipment / 12 days / 11/21/2008 8:00 / 12/8/2008 17:00
DOE / 10 days / 10/27/2008 8:00 / 11/7/2008 17:00
Run Optimization Experiment / 42 days / 11/10/2008 8:00 / 2/6/2009 17:00
Calc. Energy Balances / 5 days / 11/14/2008 8:00 / 11/20/2008 17:00
Draw P&ID / 5 days / 12/9/2008 8:00 / 12/15/2008 17:00
Analyze Experimental results / 15 days / 2/9/2009 8:00 / 2/27/2009 17:00
Optimize Process / 3 days / 3/2/2009 8:00 / 3/4/2009 17:00
Edit P&ID / 2 days / 3/5/2009 8:00 / 3/6/2009 17:00
Write Full report / 10 days / 3/9/2009 8:00 / 3/20/2009 17:00
Run Heat of Reaction Experiments / 10 days / 10/31/2008 8:00 / 11/13/2008 17:00
Table 2: 2nd semester schedule
Task / Duration / Start Date / End DateRun DOE / 26 days / 12/15/08 / 01/19/09
Revise Block diagram / 6 days / 01/26/09 / 02/02/09
Run TGA & DSC for heat of reaction data / 6 days / 02/02/09 / 02/23/09
Design and draw P&ID / 36 days / 02/02/09 / 03/23/09
Run Tests on samples from DOE / 26 days / 02/09/09 / 03/16/09
Analysis of Results / 6 days / 03/17/09 / 03/24/09
Simulate and size equipment / 16 days / 03/24/09 / 04/14/09
Turn in Final report / 1 day / 04/17/09 / 04/17/09
Prepare for Expo / 10 days / 04/13/09 / 04/24/09
Procedure
Originally, the proposed process for creating the monomer involved adding stoichiometric ratios of trimethylolethane and diethyl carbonate, approximately 240g of each, along with 0.1g of KOH and 10 mL ethanol. This slurry would then be heated to reflux at 125°C and allowed to react. Ethanol would be distilled via fractional distillation, leaving behind a cyclic carbonate ester intermediate. The intermediate would be heated further at a higher temperature under near-vacuum conditions to drive off carbon dioxide and complete the reaction, forming the product. This product would be collected via fractional distillation, and would be re-distilled to obtain 99% purity.
In the experiments using the HEL Simular™, the first step of the reaction was completed, but not the second. This was due to the temperature limits for the HEL Simular™; temperatures of ~180°C are needed to complete the reaction, but the heating oil used in the HEL Simular™ has a maximum temperature of 165°C. Upgrading the oil to one that could handle the higher temperature was deemed to be cost prohibitive.
Figures 4: HEL SimularTM P&ID
The initial attempts to create the product via this procedure were unsuccessful. In the first attempt, the wire heating element (the internal heater) inside the calorimeter did not heat the slurry evenly, boiling off some of the trimethylolethane directly without reacting it. Additionally, the heating unit which supplied heated oil to the calorimeter would shut off with an error message whenever the oil temperature would reach 140 °C. The project team later discovered that this was due to a hardware setting, and the setting was changed.
Even after the temperature issues were resolved, the slurry never went fully into solution. It remained a slurry, even at the estimated reaction temperature of 115 °C and even at the boiling temperature of diethyl carbonate, 126 °C. The project team eventually discovered that the mixture needed more ethanol to dissolve the trimethylolethane. Subsequent experiments added 60 mL of ethanol to the reactor, which resulted in approximately a 1:1 molar ratio between ethanol and trimethylolethane. This was sufficient to fully solvate the trimethylolethane.
The contents of the reactor were then heated at a particular temperature, for a particular length of time, under a particular agitation speed. These three parameters were varied according to a Design of Experiments (DOE) scheme, which is described in greater detail below. At the end of each individual experiment, a sample of the intermediate product was obtained from the reactor. A sample of the distillate was also collected. These samples were later analyzed using gas chromatography and mass spectroscopy (GC-MS). Yields were calculated based on the amount of ethanol driven off by the reaction, and were unexpectedly low, on the order of 20%.
A ninth experiment was run as a midpoint for the DOE. For this experiment, the entire intermediate was collected in a 1L 3-neck round-bottom flask. The flask containing the intermediate was placed in a heating mantle and brought to a temperature of ~190°C. A simple distillation apparatus was attached to the flask, under the assumption that the final product, with a normal boiling point of 167°C, would distill over.
As it turns out, this didn't happen. The hypothesis is that the geometry and size of the 1L round-bottom flask caused the final product to simply reflux without distillation. The contents of the flask were heated for several hours before the team concluded that no distillation was likely, and it was assumed that the flask now mostly contained the final product.
After these nine experiments were completed, the distillate and the intermediate product samples were analyzed via GC-MS. The data revealed that the distillate contained much larger amounts of diethyl carbonate than predicted, indicating that it was being boiled off before reacting with the trimethylolethane. This would also explain the unexpectedly low yields.
An additional experiment was run using the parameters of the experiment which generated the best results. Given that the reaction was losing diethyl carbonate before it could react with the trimethylolethane, the project team decided to run the experiment with diethyl carbonate in a 2:1 molar ratio with the trimethylolethane. It was hoped that this excess of diethyl carbonate would provide sufficient material for the reaction to take place before the diethyl carbonate was boiled off.
After the completion of the first stage of the experiment, the intermediate was collected in a 1L round-bottom flask, and a simple distillation apparatus was set up. Based on the earlier experience, the contents of the round-bottom flask (which, presumably, was mostly intermediate product) were allowed to heat up over 200°C. While the reaction itself should take place at ~180°C, it was felt that this temperature did not make the product vapor sufficiently energetic to allow it to reach the distillation apparatus. After attaining a temperature of 236°C, the heat was stabilized at 227°C.
After ~4 hours of distillation, the contents of the flask had boiled away except for ~10 mL of dark brown residue. The collection flask contained a clear fluid (which became slightly cloudy at room temperature), which the project team assumed to be the final product.
Design of Experiment
The first part of our project was to design an experiment to hopefully optimize the process of making 3-methyl-3-hydroxymethyl oxetane. Since it was a two step reaction and our reactor had heat limitations, it was decided that we would run only on the first step. We proposed a 3 factor, 2 level full factorial design of experiments. The factors to be varied were jacket temperature (135 °C and 145 °C), time of reaction (1 hr and 2 hrs), and agitation speed (100 rpm and 150 rpm). The experiments were conducted in the following order:
Table 4:
Factors / Responsespattern / Time (Hours) / Temperature (°C) / Agitation (rpm) / Done? / Yield, ethanol / GC/MS, %
1 / + - - / 2 / 135 / 100 / 8-Dec / 18.6 / 0.45
2 / --+ / 1 / 135 / 150 / 10-Dec / 21.2 / 0.51
3 / + + - / 2 / 145 / 100 / 10-Dec / 24.9 / 0.46
4 / - + + / 1 / 145 / 150 / 10-Dec / 22.2 / 0.44
5 / + + + / 2 / 145 / 150 / 12-Dec / 22.9 / 0.44
6 / - + - / 1 / 145 / 100 / 16-Dec / 8.4 / 0.34
7 / - - - / 1 / 135 / 100 / 15-Jan / 16.7 / 0.47
8 / + - + / 2 / 135 / 150 / 20-Jan / 28.1 / 0.52
8b / + - + / 2 / 135 / 150 / 13-Feb / 80.9 / 0.44
mid / N/A / 1.5 / 140 / 125 / 30-Jan / 15.6 / 0.40
The results indicate that increasing the time or the agitation speed independently resulted in higher yields. Increasing these two parameters simultaneously did not show a synergistic effect, however.
Figure 5: 2-D Graph displaying DOE results
Figure 6: 3-D Graph displaying DOE results
Data Analysis
Differential scanning calorimetry was performed on the reactants, to determine the heat of dissolution of trimethylolethane in ethanol, and to attempt to determine the heat of reaction. The graphs obtained are displayed in Appendix A.
These graphs provide heat of reaction data for modeling purposes. The data is based on grams of trimethylolethane. Aside from a large endothermic event where the ethanol begins to boil and the trimethylolethane begins to dissolve, heating/cooling requirements appear modest, and compensating for these requirements should be well within the ability of standard reactor jackets available in industry.
A TGA was also run on the experiment determine the mass loss throughout the course of the reaction. That graph is displayed in Appendix B. The mass loss was consistent with our predictions, but did not reflect the experimental conditions resulting from the unexpected loss of diethyl carbonate via distillation.
A GC-MS analysis was also performed on the final product and the distillate. Percent reports are displayed in Appendix C. The analysis of the distillate revealed that a substantial amount of diethyl carbonate was being boiled off instead of reacting with the trimethylolethane. As the reactants were added in a 1:1 stoichiometric proportion, this meant that about half of the trimethylolethane was left unreacted (which would explain the semi-solid nature of the intermediate product samples – they contain trimethylolethane in addition to the intermediate).
In addition, the GC-MS analysis of the final product revealed a lot of unidentified "junk" – molecular weights which cannot be accounted for by the chemistry of the desired reaction. The reason for these unidentified chemicals appearing in the final product are not known.
A plan to analyze the final product via nuclear magnetic resonance (NMR) was made. However, the usual solvent for NMR did not solvate the intermediate product. The alternative solvent was prohibitively expensive, and the NMR analysis was not performed.
Design
Block Diagram
The first step in the design of the pilot plant was to make a block flow diagram. This provided a useful overview of the process as we refined it to make an actual piping and instrumentation diagram (P&ID).
Figure 7: Block Diagram
P&ID
All valves labeled FC, are fail close, and all valves labeled as FO are fail open. Specifications for such equipment as valves, pipes, pumps, and the water tower can not be calculated without further knowledge, mostly of the layout of the space that the process will occupy. However the reactor and distillation tower specifications and calculations are in subsequent sections of the design portion of this report.
Figure 8: Final Piping and Instrumentation Diagram
Reactor
The reactor chosen for this project is the Pfaudler RT-14-10 reactor. The reactor is a 10 gallon (38 L) reactor. This is the working volume, as the reactor contains an additional 2.2 gallons (8.4 L) of head space.
Given the paucity of data, certain assumptions have been made regarding the system:
- The density of trimethylolethane in solution is assumed to be the same as the density of the solid. This is almost certainly inaccurate to a non-trivial degree, but no other data is available.
- The volumes of the reactants are assumed to be additive.
- The heat capacity of the reactant mixture is assumed to be a weighted average of the heat capacity of the individual components.
- The heat capacity of the intermediate is assumed to be equal to the heat capacity of the reactant mixture.
- The heat capacities are roughly constant over the range of temperatures of interest.
- The order of the reaction is equal to the molecularity, and the rate constant is the same for both steps.
- The energy input from the agitator can be neglected.
The physical data for diethyl carbonate (DEC) and ethanol (EtOH) was obtained from the National Oceanic and Atmospheric Administration website. The physical data for Trimethylolethane (Tris) was obtained from the Geo Specialty Chemicals website.