Identification and Characterization of Nonflammable Azeotropic Mixtures (Chapter 2)

CHAPTER II Materials and Methods All the developmental solvents and flame retardant solvents used in this study were procured from the following companies and manufacturers: 3M, ChemLogic, Chemours, Daikin, DuPont, Honeywell, Houghton, Kyzen, MicroCare, Nugentec, Reliance, Sherwin-Williams, Sigma Aldrich, Solvay, SolvChem, and Specialty Materials Corporation. Some of the solvents were obtained as research samples, and were provided free of charge. The solvents examined in this study are shown in Table 1with Chemical Abstracts Service (CAS) registry numbers and abbreviations used in this study. Some trade names and more common abbreviations are also shown. Ethylene glycol antifreeze was also used in this study as a bath solution in flash point testing. Table 1. List of chemical compounds examined in this study. ACE acetone, CAS:67-64-1 AE3 1,1,2,2-tetrafluoroethyl-2,2,2-trifluoroethyl ether (AE3000), CAS:40678-0 BAC tert-butyl acetate (tBAc), CAS:540-88-5 BOH tert-butanol (tBOH), CAS:75-65-0 C5F7 1,1,2,2,3,3,4-heptafluorocyclopentane, CAS:15290-77-4 CBTF para-chlorobenzotrifluoride (PCBTF), CAS:98-56-6 DCE trans-1,2-dichloroethene, CAS:156-60-5 EL ethyl lactate, CAS:97-64-3 EOH ethanol (EtOH), CAS:64-17-5 Table continued HEX n-hexane, CAS:110-54-3 HFC3 1,1,1,3,3-pentafluorobutane (HFC365 or Solkane 365), CAS:406-58-6 HFC4 1,1,1,2,2,3,4,5,5,5-decafluoropentane (HFC-4310mee), CAS:13849542-8 HFE0 methoxyperfluoropropane (Novec-7000), CAS:375-03-1 HFE1 methoxyperfluorobutane (Novec-7100), CAS:163702-07-6 HFE3 methoxyperfluorohexane (Novec-7300), CAS: not found HFE5 4-ethoxyperfluoro-5-methylhexane (Novec-7500), CAS:297730-93-9 I16 iodoperfluorohexane* (I-1600), CAS:355-43-1 I32 iodoperfluoropropane* (I-3200), CAS:677-69-0 MBK methyl isobutyl ketone (MIBK), CAS:108-10-1 MEK methyl ethyl ketone (MEK), CAS:78-93-3 MOH methanol (MeOH), CAS:67-56-1 MPK methyl propyl ketone, CAS:107-87-9 nPB or 1-BP n-propyl bromide or 1-bromopropane, CAS:106-94-5 PAc isopropyl acetate (IPAc), CAS:108-21-4 PFBI iodoperfluorobutane*, CAS: 375-51-9 PFC6 perfluorohexane, CAS: 355-42-0 PGE propylene glycol monomethyl ether (PGME), CAS:107-98-2 POH isopropanol (IPA), CAS:67-63-0 Sols trans-1-chloro-3,3,3-trifluoroprop-1-ene & trans-1,3,3,3tetrafluoroprop-1-ene (Solstice PF), CAS:102687-65-0 & 29118-24-9 Table continued Sup 1,2,3,3,4,4,5,5,6,6,7,7,7-tridecafluoro-1-methoxy-1-heptene (cis and trans isomers) (Suprion UP), CAS:69296-04-4 TCE trichloroethylene (TRI), CAS:79-01-6 Note: Some of these abbreviations are the widely accepted abbreviations for these compounds, while others have been shortened for the purpose of figures in this project. The compounds with an “*”, are unknown isomers of the name given. The boiling points and chemical structures of the substances in Table 1 are shown in Table 2. Table 2. Boiling points and chemical structures of the compounds used in the study. Substance Boiling Point (Tb) / °C Chemical Structure O ACE 56 H3C CH3 AE3 CF3 O F2HC 50 F F O H3C BAC H 3C O H3C BOH CH3 96 82 HO CH3 CH3 CH3 F F H C5F7 63 F H F H F F F Table continued Cl CBTF 145 CF3 DCE H Cl Cl H 48 O EL 155 CH3 O H3C HO EOH 78 HEX 69 OH H3C H3C CH3 F HFC3 41 F 3C CH3 F HFC4 F H F 3C 54 CF 3 F H F F F F HFE0 F 3C 34 CH3 O F F F F HFE1 HFE3 60 99 F 3C O CH3 F F F F F F F F F 3C O CH3 F F F F F F Table continued F F F 3C F HFE5 130 F 3C CF3 F F F F I16 113 F I F F F F F F 3C CF 3 40 F I O MBK F F 3C F I32 CH3 O CH3 116 CH3 H3C O MEK 80 CH3 H3C MOH 65 OH H 3C O MPK 102 CH3 H3C nPB or 1-BP 71 PAc 89 H3C Br O H 3C CH3 F PFBI 65 CH3 O F F 3C I F F F F Table continued F F PFC6 F F 3C 57 CF 3 F F H3C PGE F 120 F F OH O CH3 CH3 H3C POH 82 OH H Sols 19 F 3C H Cl and F F F Sup 111 F F F F 3C O F TCE F 3C F F F F Cl Cl Cl H CH3 87 Note: Some of the chemical structures shown are only one isomer in a mixture of the isomers that make up the bulk solvent. Other materials used in this study include: a fractional distillation apparatus, a tag closed-cup flash point tester (Koehler Model K14600), a DuNouy ring tensiometer made in-house, a ball drop viscometer (RGI V-2200 size no. 2), and a Raman spectrometer (Ocean Optics Raman Systems R-3000). Choosing Components to Tune Properties In this study the components of each mixture were carefully chosen. The goal was not to find as many azeotropic mixtures as possible, but to find azeotropic mixtures that were non-flammable and had good solvency against hydrocarbon grease. First, the list of available components was narrowed down from the 22,406 compounds in the Hansen Solubility Parameters in Practice (HSPiP) software database to 28 compounds using the restrictions mentioned in the preface of this document, boiling point restrictions for energy conservation, and market availability. From these 28 compounds, the HSPs helped to predict which compounds would provide good solvency. Generally, solvents that have good solvency against hydrocarbon grease are flammable. These compounds were paired with non-flammable compounds with similar boiling points, because compounds with similar boiling points were believed to have a higher chance to form a low-boiling azeotrope. This would produce mixtures that had the potential to be non-flammable solvents with high solvency against hydrocarbon grease. However, the mixtures need to be azeotropes in order to maintain these properties while boiling. The solvents that made up these pairs were mixed and distilled in order to identify the pairs of solvents that produced low-boiling azeotropic mixtures. The pairs that formed low-boiling azeotropic mixtures were tested to ensure that they were non-flammable using flash point determination, because the percentage of flame retardant in the azeotrope is determined by the intermolecular attractions between the components, and a certain amount is required to quench the flammability of the flammable component. The azeotropes that were found to be non-flammable were tested against hydrocarbon grease in a vial-based solvency study and those with extremely poor solvency were studied no further. The remaining azeotropes were characterized using standard operating procedures (SOPs) that were produced in this study. The SOPs for identifying and characterizing the azeotropes are summarized in the following sections of this chapter and the full SOPs are included in the appendix section of this thesis. The equations used for data analysis are provided in Chapter I. Distillation For the identification of low-boiling azeotropic mixtures, a fractional distillation is the easiest laboratory technique. The distillation process used an ordinary fractional distillation apparatus. The setup of the apparatus is discussed in detail in the appendix section. Either a fractionating column that contains a stainless-steel chain or a Vigreux column with built in deep pointed indentations in the glass was used to maintain the temperature gradient and prevent convective currents in the column. In addition to this, a Teflon sleeve was used on each of the ground glass joints instead of stopcock grease. This ensured that the joints were sealed, and that the distillate was uncontaminated. The fractional distillation was used for two purposes in this study: to identify whether an azeotrope would form, or to purify contaminated solvents so that they may be recycled. When attempting to form a low-boiling azeotropic mixture, the two pure solvents were combined in a 1:1 (v/v) ratio in the round bottom boiling flask. The mixture was brought to boiling and heated until the first drop of distillate reached the receiving flask. At this point, the temperature was recorded. If the temperature reading was below the boiling points of the two components of the mixture, then the distillate was an azeotropic composition of the two components. If this occurred, the distillation was stopped when the temperature began to increase. If the temperature reading after the first drop of distillate was received was at or close to the boiling temperature of one of the components, the distillation was continued until the temperature reading began to increase. This was done for a few different reasons. Sometimes a low-boiling azeotrope has a boiling point that is very close to the boiling point of the lowest boiling component. Near azeotropes may also form between components that are difficult to separate. Also, if the mixture has no azeotropic point, then it is advantageous to continue the distillation so that the pure solvents can be reclaimed. In the case where the temperature is at or close to the boiling point of the lowest boiling component, the next step in determining if the mixture is an azeotrope or near azeotrope was to perform a Raman chemometric analysis on the mixture. This method is outlined later in this chapter. Figure 9 is an example of the distillation setup. Figure 9. Example of a distillation setup. When attempting to separate a soil from a solvent or multiple solvents from each other using fractional distillation, the process started in the same manner as the previous scenario. The solvent was the lowest boiling substance in every case when attempting to clean a solvent, so the distillation could be continued until the boiling temperature of the distillate began to increase or the solvent in the boiling flask was almost exhausted. When separating multiple solvents, the distillate was collected in fractions. The first fraction was collected until the boiling temperature increased, then the next fraction was collected. This process was repeated until the boiling flask had only a small amount of liquid remaining. Raman Chemometric Technique A chemometric technique is a technique that allows the calculation of the amount of certain substances that make up an unknown sample. Unknown, in this context, means that the identities of the sample’s components are known, but the amount of each component is not known. A Raman chemometric technique is used in this study to find the composition of azeotropic mixtures. This technique takes very little work and very little time. In the chemometric analysis, the Raman spectrum of each component was taken and the photon counts versus pixel number data was recorded. Next, the Raman spectrum of the mixture was taken. The diode array of the Raman Spectrometer was calibrated with the IR and Raman spectra of acetonitrile by using the peaks with the following Herzberg designations19: 2ν8, ν4, ν7, ν3, ν6, and ν2. These peaks were chosen because they were in the range of both the IR and the Raman instrument. This calibration allowed the x-axis of the spectrum to be expressed in wavenumbers Raman shift (cm-1). In Microsoft Excel, a weighted average of the spectra of the components was made and the residuals between the combined spectrum and the experimental mixture spectrum were calculated between 200 and 2715 cm-1. This range was chosen to avoid fitting the Rayleigh line. These residuals were used to calculate the sum of the squares of the residuals. Finally, the solver tool was used to minimize the sum of the squares of the residuals by changing the weighting of the different components in the combined spectrum. Two mixtures were made by mixing known amounts of HFE5 and CBTF. The first mixture was made up of 50 % (v/v) of each of the two substances, and the second was 60 % (v/v) HFE5 and 40 % (v/v) CBTF. The Raman chemometric analysis technique was applied to these two mixtures. The results of these analyses were compared to the volume percentages as prepared, the calculated mole percentages, and the calculated mass percentages. Through these comparisons, it was shown that this analysis yielded the volume ratio of the components in the mixture. These results can be seen in Chapter III. Flash Point Flash point determinations in this study were made with a Koehler Model K14600 tag closed-cup flash point apparatus, which can be seen in Figure 10, and a modified ASTM D5620 method. The tag closed-cup apparatus seemed exceptionally appropriate to determine the flash point of these azeotropic mixtures, because the azeotropes were meant for use in a vapor degreaser and this tester mimics the situation that occurs in a vapor degreaser by creating a concentrated vapor zone within the sample cup. The modifications to the method were that the test flame was applied to the sample after every one degree Celsius of temperature rise from the start of the test, and the heating bath was pure ethylene glycol antifreeze. These modifications had to be made because the ASTM D56 method is written to verify flash points of substances whose flash points are known. In this study, the flash points of substances whose flash points were not known were being determined. This means that it is impossible to start applying the flame to the sample after the temperature reaches five degrees Celsius below the expected flash point, because there is no expected flash point. This also means that the bath could not be changed according to the expected flash point. The modified method was used to verify the flash points of decane and methanol. This helped to ensure that the flash point apparatus and the modified method were capable of accurately determining the flash points of substances with known flash points. The flash point tests performed in this study were recorded on video to provide visual evidence. Figure 10. Koehler Model K14600 tag closed-cup flash point apparatus. The detailed setup information for the flash point tester is included in the appendix. When performing a flash point test, the test was performed with fresh sample, under dim lighting, and with a draft shield. To begin the test, the sample was added to the test cup and the test flame was lit. The apparatus was powered on and the control knob was used to adjust the heating rate to approximately one degree Celsius per minute. The test flame was applied to the sample for one second after every one degree Celsius of temperature rise. If the sample exhibited a flash when the test flame was applied, the temperature was recorded as the flash point. If no flash occurred after the sample reached boiling, the sample was said to have no flash point. If the sample flashed within ten degrees Celsius of the starting temperature, the test was performed again with a lower starting temperature. The definition of a flash, as well as other important information, will be provided in the discussion section. Vial-Based Solvency Testing A solvency screening test was developed during this project to compare the solvency of many solvents against hydrocarbon grease. This test also helped decide which solvents should be used in blends as cleaning agents, and which blends had potential as vapor degreasing solvent replacements. The blends that failed this screening were not tested or characterized any further. This test used a gravimetric analysis with a 5-place laboratory balance (Kern model ABT 120-5DM) equipped with an automated data printer. A 4-mL glass vial and lid were weighed, and approximately 0.1 g of marine grade hydrocarbon grease was placed in the bottom of the vial using a large bore syringe. The grease was applied with similar glass contact in all cases. The mass of the vial with grease and the lid was measured to obtain the mass of grease applied to the vial (mo), and approximately 3 mL of the test solvent was added to the vial. The vial was shaken on a laboratory shaker for 1 hour, and then poured into a pre-weighed aluminum pan. Any of the grease that fell out of the vial when it was turned up-side-down was taken as being removed from the vial by the solvent. After emptying the vial, the aluminum pan was placed in a clean laboratory oven to evaporate the solvent. When all of the solvent was evaporated, the pan was weighed again to obtain the mass of grease that was transferred to the pan (mt). The percentage of the mass of grease transferred from the vial by the solvent (Ptransferred) was calculated by taking the ratio of mt and mo (mt / mo) and multiplying by 100. Density and Surface Tension The density and surface tension measurements in this study were performed with a DuNouy ring tensiometer. A diagram of this tensiometer can be seen in Figure 11, and its construction is described elsewhere.15 Figure 11. Diagram of a digital DuNouy tensiometer.15 The tensiometer in Figure 11, consists of a 4-place digital lab balance mounted above a hydraulic platform where the sample was placed. A chain with a hook connector on the end was hung from the under-hook of the balance. Instructions for preparing the tensiometer for measurements are included in the Appendix. A DuNouy ring tensiometer is a very inexpensive, easy, and accurate way to measure hydrostatic density and surface tension. The data collection is outlined in this section, but the theory and analysis equations are described in Chapter I. To perform a hydrostatic density measurement with this tensiometer,15 the data collection software (Hyperterminal) was started and the lab balance was tared. Data collection software was not required for the density measurements, but it was convenient due to the height of the lab balance. A 10.000 mL Pyrex density standard was weighed by placing it on the hook connector. This mass is the dry mass. The wet mass was taken by submerging the sinker in the sample. The temperature of the sample was taken using a thermocouple. This was all of the data needed to calculate the density. This process was repeated at ambient temperature, higher temperature, and lower temperature within a 10 to 20-degree Celsius temperature range. After the calculation of these densities, they could be used to calculate the thermal expansion coefficient for the substance through a regression analysis. The thermal expansion coefficient and its relationship with density are described in Chapter I. The thermal expansion coefficient was used to interpolate the density value so that it may be reported at 25 °C, because this is the temperature at which the density is most commonly reported. This also allowed the density to be calculated at any other temperature within the range for use in any other measurements that required density. The information in the regression analysis allowed a measure of uncertainty in the density to be calculated. The surface tension measurements were performed according to ASTM D133121. For this method, the data collection software was crucial. A platinum iridium tensiometer ring (with a radius of 0.9537 cm from the center of the wire to the center of the ring and a wire radius of 0.01779 cm) was hung from the hook connector and submerged just below the surface of the sample. The sample was in a container whose diameter was large enough for the meniscus to not affect the surface of the liquid in the center of the container. At this point, the data collection software was started and the balance was tared. The hydraulic platform was lowered slowly, allowing the ring to rise out of the surface of the liquid. The ring lifted the meniscus of the liquid and the mass of liquid lifted was measured every second until the ring lost contact with the surface of the sample. Then, the temperature was measured with a thermocouple. Like in the density interpolation, the process is repeated at ambient, higher, and lower temperatures. Using interpolated densities, the surface tension value for each measurement was calculated. A regression analysis revealed a strong inverse linear relationship between temperature and surface tension. This relationship was used to interpolate the surface tension to a new temperature, and calculate the uncertainty if the slope and intercept of the line were found to be statistically significant. This allowed the surface tension to be reported at 25 °C. Viscosity The viscosity measurements were made with a ball drop viscometer also called a falling ball viscometer, which can be seen in Figure 12. The ball drop viscometer is very inexpensive compared to other methods of measuring viscosity, and it is extremely simple to use for data collection. For precision, a more sophisticated modified ASTM D134317 method was used for these measurements. Figure 12. Parts of a ball-drop viscometer. The first modification was to build an apparatus that allowed for video assisted timing of the ball drop, which can be seen in Figure 13. This was done using a laser table and laser table attachments. This apparatus consisted of three stands, a viscometer stand, an optics stand, and a video camera stand. Since this was done on a laser table, all three stands were locked into place. The viscometer stand consisted of a rotating viscometer holder at the top of the stand, a colored piece of paper behind the top set of timing lines to keep the camera in focus, and a catcher at the middle of the stand to ensure that the viscometer is completely vertical. The optics stand consisted of a microscope slide at a 45-degree angle in front of the top set of timing lines on the viscometer, and a frontsilvered mirror at a 45-degree angle in front of the bottom set of timing lines. The video camera stand was only a rod that held the camera so that it was recording the top set of timing lines through the microscope slide on the optics stand. This modification allowed the timing data to be recorded digitally for reviewing later. This also made the timing more accurate, because the camera recorded at 30 frames per second, which gave the timing an uncertainty of ±1 frame or ±0.03 seconds. Figure 13. Viscosity apparatus for video assisted timing. The other modification was to use Type I deionized water and the NIST database of densities and viscosities to calculate a temperature dependent viscometer constant. This was achieved by performing regression analyses on the density versus temperature and viscosity versus temperature data for water. These regression equations for density (ρ) and viscosity (η) in terms of temperature (T) can be seen here as eq 24 and eq 25 respectively. 5.00 2 10 4.30 2 10 5.6 7 10 4.2 1 10 1.000271 9 (24) 1.668(1) (25) After measuring the time of the ball drop for water, the regression equations were used to calculate the density and viscosity of water at the temperatures of each of the ball drops. This allowed the calculation of the viscometer constant for each measurement of water based on the time of the ball drop and the temperature. The temperature was usually slightly higher for each successive measurement, and the calculated viscometer constant was slightly different for measurements with different temperatures. Because of this, the temperature versus viscometer constant data for water allowed a temperature dependent viscometer constant to be calculated based on a regression. This was used to calculate the viscometer constant for the specific temperature of every subsequent viscosity measurement. While the data analysis for the ball drop method was a bit intricate, the operation of the viscometer was very simple. Once the viscometer was filled with the sample and a ball of known density, the viscometer could be placed into the rotating viscometer holder on the homemade apparatus. The temperature was measured by placing the end of a thermocouple into the capillary hole under the cap. The lights in the room were turned off and the viscometer was rotated upside down until the ball was resting at the top of the tube. Then, the video camera was powered on and the recording was started. The name of the sample, the trial number, and the temperature were said aloud for the recording. The tube was rotated right side up, and the ball began to fall through the liquid. Once the ball had passed the first set of timing lines, the bottom set of timing lines were illuminated with a light. The bright reflection in the microscope slide made the camera focus on the bottom set of timing lines. Once the ball passed the bottom timing lines the test was over. This process was repeated until five measurements were made on each sample. The fact that each consecutive measurement had a slightly higher temperature, could be used to extrapolate the viscosity to 25-degrees Celsius. A regression analysis revealed a negative correlation between temperature and viscosity within a small temperature range. This was useful for reporting the viscosity at a temperature of 25 °C, which is the temperature normally used to report viscosity. This was also useful for comparing the measurements of pure substances to the literature values, since the measurements were not temperature controlled. Azeotrope Prediction A method for predicting the formation of binary azeotropic mixtures was produced as an addition to this thesis work. The method is described in this section. Data for each proposed azeotrope component was extracted from the HSPiP database. This data included the molar volume (Vm), the Antoine coefficients (A, B, and C), and the HSPs (D, P, and H). This data was used in an Excel spreadsheet to perform the following calculations. Equation 26, the Antoine equation, was used to calculate the saturated vapor pressure (PSAT) in units of mm Hg at 25 °C using the Antoine coefficients (A, B, and C) in units of mm Hg, mm Hg * °C, and °C respectively. (26) In eq 26, T is the temperature in °C. The Hansen distance (Ra) was calculated using the HSPs and eq 9. A random number between zero and one was used as an initial value for the mole fraction of component A (xA). The mole fraction of component B (xB) was taken as one minus xA, because of the relationship in eq 27. 1 (27) Using an assumption that the volumes of the components are additive when mixed, the volume fractions of component A and B (ϕA, and ϕB) could be calculated using eq 28 and eq 29. (28) 1 (29) In eq 28 and eq 29, the A’s and B’s correspond to component A and component B. Equations to calculate the activity coefficients of component A and B (γA, and γB) were taken from the HSPiP e-book.10 These equations are shown here as eq 30 and eq 31. ln . ln . (30) (31) In eq 30 and eq 31, R is the gas constant (8.31445 Jmol-1K-1), and T is the temperature (298.15 K). With the above information, the separation factor (αAB) could be calculated. The separation factor was calculated using eq 32.8 (32) The separation factor is a measure of how difficult it is to separate two substances using distillation. When the values of the numerator and denominator are far apart, the substances are easily separated using distillation. However, when the values of the numerator and denominator are closer to each other, it will be harder to separate the two substances. When the numerator and denominator are equal, αAB equals one, and the two substances have formed an azeotrope and cannot be separated with a normal distillation. This fact was exploited in order to calculate the composition at which an azeotrope forms. This was easily done in an Excel workbook using the Solver tool. Solver was used to make αAB equal a value of one by changing xA. If solver is unable to find a solution, then no azeotrope is predicted to form between the two substances. If solver is able to find a solution, then the composition that solver used to make the separation factor equal one is the predicted composition of the azeotrope.