Ind. Eng. Chem. Res. 1997, 36, 5023-5027
5023
Effect of Oxygenates on Water Uptake in Hydrocarbon Fuels Michael Golombok*,† and Steven Tierney Shell International Petroleum Company, Shell Research and Technology Centre, P.O. Box 1, Chester CH1 3SH, England
The addition of water to gasoline can boost octane number and decrease NOx emissions. Oxygenatesswhich are already used in gasoline for boosting octane quality (EU) and reducing CO (U.S.)scan boost water uptake in hydrocarbon fuel. At 20% butanol concentration in a paraffinic PRF 60 fuel, up to 1% water can be absorbed, leading to a boost of 1-3 octane points. However, at a more realistic oxygenates (10%) concentration, only about 0.2% water can be absorbed and this is insufficient for significant octane boosting. The effects are independent of the paraffinic or aromatic nature of the base fuel. We compared a range of oxygenates (alcohols and ethers) in a paraffinic refinery base stream by ranking their ability to absorb water. The best oxygenate species for absorbing water have a thin polar tail and a bulky covalent head with a spatially intermediate polarizable element. The relatively low levels of water uptake suggest that surfactants should still be used, although at lower concentrations as the oxygenate can serve as a cosolvent. 1. Introduction Water addition to gasoline is periodically examined in spark ignition engine research because it reduces knock and NOx emissions.1,2 Water is added to the combustion process either by separate injection3 or by prior incorporation into the fuel. The latter is clearly economically more attractive, providing water can be satisfactorily incorporated. The last cycle of studies in the late 1970s came up against the obstacle of the effect of emulsifiers which were used to incorporate water: these had to be present in large quantities (equivalent in weight terms to the amount of water absorbed) and resulted in a slight increase of hydrocarbon (HC) emissions.1,2 Two developments since then indicate possible suggested ways of counteracting these effects: First, the problem of post-light-off HC emissions can be overcome using catalytic convertors.4 Second, oxygenates are now used to reduce CO (U.S.) and to boost octane quality (EU). These two factors could help overcome the problems caused by the organic emulsifier. The range of oxygenates usable as octane boosters includes molecules which may have the ability to solvate water into gasoline.5 Thus, rather than using expensive emulsifiers in gasoline, it may be possible to exploit the polar nature of the oxygenates themselves in order to takeup more water and create a water-in-oil microemulsion. The purpose of this study is to see whether any octaneboosting oxygenates could also enhance water incorporation to promote antiknock properties. The categories of possible oxygenates include ethers, esters, alcohols, and ketones. Previous studies have concentrated on the solubility of the oxygenate in water, whereas we are interested in the solubility of water in the oxygenate, which is always modified by the presence of the bulk gasoline. Recent studies have examined solubility effects at constant temperature for ternary solutions containing water, a pure hydrocarbon, and an oxygenate.6,7 One result was that water, methanol, and synthetic gasoline would be disadvantageous as far as water addition would be concerned as only very small * To whom correspondence should be addressed. Email:
[email protected]. † Present address: Shell International Chemicals, Shell Research and Technology CentresAmsterdam, Postbus 38000, 1030 BN Amsterdam, The Netherlands. S0888-5885(97)00304-7 CCC: $14.00
amounts of water could be incorporated into the organic phase before aqueous/organic separation occurred. Birdi8 and also Garti et al.9 have specified molecular structural features that enhance emulsification in paraffin/alcohol mixtures. Reformulated (i.e., oxygenate-containing) gasoline typically contains 10% v/v methyl tert-butyl ether (MTBE), which is not good at solvating water. However, in the 1980s, Germany and other markets used methanol and butyl alcohol mixtures (particularly tert-butyl alcohol (TBA)) to compensate for lead reduction. The butanols are smaller than the emulsifiers, and by themselves they tend to dissolve rather than disperse water; however, this changes when gasoline is the major component. An additional aspect of the butyl alcohols is that they are found in current manufacturing product streams, particularly the modified methanol synthesis from synthesis gas (CO).10 The product stream contains a mixture of butyl alcohol isomers (along with methanol).11 As the butyl alcohols are themselves precursors to MTBE, they are of intrinsic interest as a cheaper alternative oxygenate. Alcohols in gasoline are also known to pick up water from tank bottoms although as this has always been considered undesirable; the bulk of the effort has been to try to minimize the corrosive effects by use of inhibitors rather than to try to maximize water uptake.12 Our initial experiments thus started with previous octane blending improvers where water uptake had been observed but not optimizedsthe butanols. 2. Experiment 2.1. Fuels. The oxygenate concentration in base fuel was that used in the standard ASTM/API blending octane studies.13 The base fuel is a solution of 60% isooctane and 40% n-heptane, which by definition has a research octane number (RON) and motor octane number (MON) of 60, into which were dissolved the four butanols at 20% v/v. A sample of this fuel/oxygenate blend (“F/O”) was octane rated (see below). The remaining F/O blend was placed in a water bath at 20 °C. Water was then added until a separate aqueous phase became visiblesat this point we know that the fuel is saturated with water (“F/O/W”). This never required more than 2% v/v addition of water. The © 1997 American Chemical Society
5024 Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 Table 1. Base Fuels Used in This Study along with Measured Saturable Water Content fuel
light full-range alkylate platformate
RON
95
102
MON
92
90
max. H2O (ppm)
50
102
fuel % w/w paraffins % w/w aromatics E100a (%v/v)
light full-range alkylate platformate 99 0.5 45
25 73 17
a E100 is the volume fraction of the fuel which has evaporated at 100 °C during ASTM distillation.
organic phase was then collected using a separating funnel and refrigerated at 5 °C for 24 h in order to check for stability. If the mixture is clear, then we have either solvation or a microdispersion with a droplet diameter less than the wavelength of visible light, i.e., < 0.4-0.7 µm. Providing the solution remained clear, octane rating was then carried out. After initial tests with the butanols at 20% in PRF 60, the level of oxygenate in subsequent base fuels was reduced to 10% as this more accurately reflects the levels found in commercial reformulated gasolines. Adding water until two phases are visible changes the concentration of oxygenate in the base fuel due to differential partitioning of oxygenate between the organic and aqueous phases. In order to avoid this, water was titrated into the fuel/oxygenate mixture to assess maximum uptake (see below). Two refinery streams were also used as base fuels. The paraffinic fuel was a refinery light alkylate manufactured using HF alkylation of butene and butane. The aromatic stream was a refinery full-range platformate manufactured by reforming over a platinum catalyst. Details of the fuels are shown in Table 1. For comparing a range of oxygenates we used the paraffinic light alkylate as a base fuel as it has a considerably lower intrinsic water uptake potential than aromatics, as shown in Table 1. This is still a relatively high octane base fuel when one considers that commercial gasoline typically has RON 95. Light alkylate is thus chemically similar to the RON 60 fuel used above but has more branched hydrocarbonssit is basically the
commercial analogue to PRF 94 fuel. The oxygenated gasolines which showed the best uptake in water, i.e., those which had the highest threshold limits, were selected for engine testing to determine their RON and MON values. 2.2. Measurements. Samples (100 mL) were prepared consisting of 90 mL of base fuel (“F”) and 10 mL of oxygenate (“O”). The water content of the fuel mixture (F/O) prior to adding water was measured at 20 °C using an Aquapal II moisture meter. A total of 0.2 mL of the sample was added to a Karl-Fischer reagent and then electrolyzed under computerized control to determine the water content. Water was then titrated into the oxygenated gasoline in increments of 1000 ppm. The fuel blend was kept at 20 °C in a water bath. Successive 0.2 mL samples were tested six times in the moisture meter apparatus. The device titrates slightly past the end point and then backtitrates to obtain the quantity of total charge passed, from which it can calculate how much water was in the known volume of sample which was introduced. The data thus consisted of theoretical water concentration (assuming total miscibility) and the resulting real water uptake as measured experimentally. The results for all oxygenates were compiled on a spreadsheet and plotted. The water uptake of each mixture was then compared to the theoretical value, where it is assumed that 100% of the water added is in the organic phase. Octane rating was carried out using standard CFR engines according to the rating test method for Motor Octane Number (MON) test method ASTM D2700-86. 3. Results and Discussion Figure 1 shows the octane rating results for all the butanol isomers at 20% in PRF 60. The largest octaneboosting oxygenate is isobutyl alcohol, which augments the octane number by 14 when added to PRF 60 but is not very good at absorbing added water. On the other hand, the linear n-butanols were best at incorporating water into the PRF. This can be understood from the fact that they are the ones which are closest to classical emulsifier structuresspossessing a hydrophilic group
Figure 1. Motor octane number for 20% v/v butanol isomers in PRF 60 and the same mixture following saturation with water. The MON of the base fuel is 60. The figure indicates the percent concentration of water in the mixtures.
Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 5025
Figure 2. Concentration of water solubilized in two base fuels (full-range platformate and light alkylate) each containing 10% v/v of 1-butanol. The solid line shows the theoretical response for total solubility of water.
(OH) attached to a small but clearly covalent hydrophobic tail. The concentration of water in percent by volume is shown above the bars for the MON of the fuel-oil-water (F/O/W) blend. Isobutyl alcohol had the smallest uptake of water and also the smallest resulting water absorption-induced MON boost, whereas the n-butanols are much better. These water-induced octane boosts range from 1 to 4 octane points per percent by volume of water. This is higher than previous results for water in fuel mixtures. With emulsions of poly(ethoxy alcohols), typically 0.3-1 ON/% v/v was obtained, but the effect of the knockpromoting emulsifier needs to be taken into account in these studies, so that our results are reasonably in line with previous measurements.1,2 We also need to allow for the fact that the base fuels for the previous studies were somewhat higher, RON ) 801 and 100,2 respectively. This shows that the octane of the base fuel before the oxygenate and water are added is important. Most base fuels used in these standardization tests are paraffinicswe would expect the same amount of
water to be absorbed in the presence of alcohol irrespective of the relative amounts of isooctane and n-heptane in the base fuel because the intermolecular forces are the same. Water uptake is thus independent of paraffinic base fuel octane number. However, two practical fuels of identical octane number might absorb water to differing extents because of their different chemical characteristic internal forces. Aromatic rings and aliphatic chains have different polarizabilitiessthe former are more polar and so might be expected to absorb water more effectively. Indeed we found (Table 1) that fullrange platformate could absorb twice as much water as light alkylate. Figure 2 shows the concentration of water observed as a function of water added to these two refinery streams, which each also contain 10% 1-butanol. The standard deviations on the mean measured water concentrations were all between 4 and 7% of a typical mean result. The solid line shows the theoretical water content based on water in the original fuel/oxygenate mixture. The point at which the difference between ideal theoretical uptake and the observed uptake differs by 10% of the observed uptake was taken as a measure of the water saturation potential. Up to the point of phase separation at 4000 ppm, the solutions remain clear, whereas above this limit, apart from noticeable trace formation of droplets on the bottom, the solutions become slightly clouded. This means that microdroplet sizes have increased to the optical range (∼0.5 µm). The behaviors of the aromatic and paraffinic fuels in the presence of 1-butanol are very similar (Figure 2), which is not what would have been expected from the different solubilities of water in these fuels (see Table 1), as described above.9 It is clear that the 1-butanol acts as a solubilizing agent for water by more than an order of magnitude and that it is this oxygenate which determines the water uptake in a hydrocarbon independently of the chemical nature of the base fuel, i.e., whether it is paraffinic or aromatic. Having established that water uptake can occur (provided enough solubilizing oxygenate is present) and having established that base fuels do not make much difference, we now move
Figure 3. Ranking of maximum water uptakes with molecular formulas for oxygenates at 10% v/v in light alkylate.
5026 Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997
Figure 4. Correlation between maximum water uptake and the separation between the OH and the branching point, for a number of alcohols (including alkoxyalcohols)
on to a comparison of various oxygenates and their ability to augment water uptake in light alkylate. The values of water uptake in light alkylate are ranked in Figure 3, where the molecular formulas of the oxygenates are also shown. These experiments are not the same as assessing water solubility in the pure oxygenatessthe presence of the base fuel has an effect on water uptake: it is the ability of a fuel/oxygenate mixture to absorb water which is of interest. Pure alcohols of the types octane rated above appear in the middle range 1000-2000 ppm. For example, at 20% concentration of 2-butanol in PRF 60, water uptake (Figure 1) is 0.88%, but at 10% in light alkylate, it is only 0.11%. Even allowing for the (slight) difference in base fuel, halving the oxygenate concentration (the difference between Figures 1 and 3) has caused the saturation water level to drop by a lot more. The best performers are relatively short-chain alcohols with a spatially intermediate nonbulky polarizable element such as a double bond or ether linkage. The fact that these molecules are most effective can be understood by comparison with the effect of emulsifiers in the more commonly encountered oil-in-water emulsions. In oil-in-water dispersions, effective emulsifiers have a polar head which attaches to the aqueous continuous phase, with a covalent sterically small tail which is soluble in the dispersed hydrocarbon phase.
However, in our water-in-oil case, the aqueous phase is now the dispersed one, so ideally we want a polar sterically small hydrophilic tail in the water microdroplet with a covalent bulky hydrophobic head lodged in the hydrocarbon continuous phase. Because the oxygenates are quite small molecules, the polar and covalent ends need to be insulated from one another by a double bond or ether linkage which are both polarizable (π orbitals and nonbonding p orbital electrons, respectively). The need for a branched bulky covalent head and a small polar tail can be checked by plotting the water uptake against the number of bonds between the hydroxyl and the branch point on the oxygenates having these characteristics. (This is analogous to the use of paraffin branching: the isoindex is a common measure of octane number used in refinery conversion processes.) The correlation in Figure 4 between the water uptake and the OH branch point separation is reasonable (80%) and certainly no worse than similar structural correlations used for predicting the octane quality of process streams.14,15 Figure 4 confirms that further optimization of oxygenate molecular structure should be along the line we have identified, i.e., a small polar hydrophilic tail, a bulky covalent hydrophobic head, and a mediating polarizable molecular feature. The final test is the octane performance: Figure 5 shows MON results for three relatively good water absorbers in the ranking of Figure 3. In all three cases the oxygenated fuel depletes the MON of the light alkylate, and water only increases the MON for the 1-butanol blend by 0.5. The poor performance is due to the high MON of the base fuel, which at 92 is 7 points higher than the value for standard European unleaded gasoline. The reason for the small effect of water is the relatively small uptake, which never exeeds 0.25%. We showed above that the uptake of water with 20% butanols (as opposed to the 10% oxygenates in the ranking of Figure 3) was much higher and significantly enhanced octane qualities. We are currently designing oxygenates for use at a higher concentration in lower octane base fuels which will enable a greater water
Figure 5. MON for a number of oxygenates which are good at enhancing water uptake in light alkylate: F, fuel only; F/O, fuel with 10% v/v oxygenate; F/O/W, fuel and oxygenate with maximum stable uptake of water.
Ind. Eng. Chem. Res., Vol. 36, No. 11, 1997 5027
uptake. Moreover, it is clear from the levels of water concentration that emulsifiers will need to be used to attain water concentrations which yield significant knock and NOx benefits. Recent work suggests that, with the use of appropriate oxygenates as cosurfactants, the level of emulsifier needed to obtain these fuel benefits will be much less than the concentrations previously used. Literature Cited (1) Peters B. D.; Stebar, R. F. SAE Paper 760547. (2) Dryer F. L. in 16th Symposium (Intl) on Combustion; Combustion Institute: Pittsburgh, PA, 1977. (3) Birch, S. Automot. Eng. 1996, 104 (2), 33. (4) Weaving J. H. In Internal Combustion Engineering; Weaving, J. H., Ed.; Elsevier: London, 1990. (5) Spindelbalker, C.; Schmidt, A. Erdoel, Erdgas, Kohle 1986, 102 (10), 469. (6) Stephenson, R. M. J. Chem. Eng. Data 1993, 38, 134. (7) Stephenson, R. M. J. Chem. Eng. Data 1992, 37, 80.
(8) Birdi, K. S. Colloid Polym. Sci. 1982, 260, 628. (9) Garti, N.; et al. J. Colloid Interface Sci. 1995, 169, 428. (10) Forzatti, P. et al. Catal. Rev. Sci. Eng. 1991, 33, 109. (11) Hermann, R. G.; et al. Stud. Surf. Sci. Catal. 1991, 64, 265. (12) U.K. Patent No. 1,185,801, 1970. (13) API/ASTMS. Knocking characteristics of pure hydrocarbons; API: Philadelphia, 1958; API Research Project 45, ASTMS No. 225. (14) Golombok, M.; de Bruijn, J. N. H.; Morley, C. Trans. Inst. Chem. Eng. 1995, 73A, 849. (15) Nierlich, F. Erdoel, Erdgas, Kohle 1987, 103 (11), 486.
Received for review May 1, 1997 Revised manuscript received August 4, 1997 Accepted August 4, 1997X IE9703043
Abstract published in Advance ACS Abstracts, October 1, 1997. X