ARTICLE pubs.acs.org/est
Sooting Tendencies of Oxygenated Hydrocarbons in Laboratory-Scale Flames Charles S. McEnally* and Lisa D. Pfefferle Department of Chemical Engineering and Center for Combustion Studies, Yale University, New Haven Connecticut 06520-8286, United States
bS Supporting Information ABSTRACT: Sooting tendencies have been measured for 186 oxygenated and 89 regular hydrocarbons under controlled laboratory conditions. The test compounds include alcohols, ethers, aldehydes, ketones, esters, alkanes, alkenes, and cycloalkanes ranging in size from methanol to isododecane. Sooting tendency was characterized with a new method based on measuring particle concentrations in methane/air nonpremixed flames when 1000 ppm of each test compound was added to the fuel. This method offers high precision and high sensitivity to the direct chemical effects of the additive. The results provide a wide-ranging yet detailed quantitative picture of how fuel oxygen affects soot formation, which will be useful for optimizing the soot-reducing benefits of oxygenated renewable fuels. The measured sooting tendencies of 1-alcohols are similar to those of n-alkanes with the same number of carbon atoms, while those of secondary alcohols are slightly higher. Aldehydes and ketones soot the same as n-alkanes with one less carbon atom. The sooting tendencies of esters depend strongly on molecular structure and increase in this order: methyl and ethyl esters < carboxylic acids, propyl esters, and n-alkanes < butyl and pentyl esters. The high sooting tendencies of the secondary alcohols and higher esters suggest that four-center and six-center reaction pathways are important.
’ INTRODUCTION Cellulose—the primary constituent of plants—contains similar amounts of oxygen and carbon. Thus as the world inevitably moves from combustion fuels that have been buried for geologic time scales, such as petroleum and coal, toward renewable fuels that are directly derived from plant matter, the oxygen content of the fuels in use will increase. One important distinction between the combustion behavior of oxygenated and regular hydrocarbons is that oxygenates, which are already partially oxidized, may produce fewer carbonaceous soot particles when they burn.1 Since soot contributes to global warming and to ambient concentrations of toxic fine particles,2,3 any reduction in its emissions would be valuable. In order to maximize this benefit, decision-makers need fundamental data that describes how the rates of soot formation from oxygenated hydrocarbons compare with those of regular hydrocarbons and how they depend on oxygenate structure. This paper reports quantitative sooting tendencies of 275 oxygenated and regular hydrocarbons measured under controlled laboratory conditions. A sooting tendency is a parameter that describes the sooting behavior of a particular fuel.4 In this study sooting tendency is defined as the maximum mass concentration of soot measured in a methane/air nonpremixed jet flame with 1000 ppm of the test compound added to the fuel. We proposed this definition several years ago and have applied it extensively to aromatic hydrocarbons;5,6 the results agree well with other methods for characterizing sooting tendency.7-9 This study greatly expands the number of oxygenates for which sooting tendencies have been measured, from 35 to 194.10-12 The measurements are made in gas-phase steady-state flames; thus they complement studies in engines since they isolate the direct chemical effects of oxygen from indirect effects such as changes in fuel volatility and ignition timing. r 2011 American Chemical Society
Soot formation is an active area of research,13 but the current understanding suffices to explain the chemical significance of sooting tendencies.4,5 Figure 1 depicts this explanation. As the hydrocarbons in the fuel begin to react, either because they are approaching a flame front or are starting to ignite, they are converted into fragments by bimolecular and unimolecular reactions. If the mixture is sufficiently fuel-rich, then these products, instead of oxidizing to CO2, will react with one another to form aromatics and ultimately soot. Sooting tendencies vary because most fuel hydrocarbons react to distinct pools of products that have differing effectiveness as soot precursors. As a simple example, aromatic hydrocarbons (middle portion of the figure) have high sooting tendencies because their products include aromatic species, which circumvents the cyclization steps that are necessary for soot formation from aliphatic hydrocarbons (top portion of the figure). Oxygenates (bottom portion of figure) contain hydrocarbon functional groups, so they also form soot precursors.1,14-17 However, since their products also include oxygen-containing species that do not lead to soot, such as small aldehydes or CO, oxygenates can have smaller sooting tendencies than regular hydrocarbons. Two methods have been used to determine the sooting tendencies of oxygenates. The first is to generate flames of pure oxygenates and measure the smoke height, which is the height of the tallest flame of the test compound that does not emit soot from its tip. This method has been applied to five small alcohols and to seven long-chain alcohols and carboxylic acids.10,12 The second method is to add oxygenates to a heavily sooting base Received: November 5, 2010 Accepted: January 28, 2011 Revised: January 28, 2011 Published: February 17, 2011 2498
dx.doi.org/10.1021/es103733q | Environ. Sci. Technol. 2011, 45, 2498–2503
Environmental Science & Technology
Figure 1. A schematic representation of soot formation from aliphatic, aromatic, and oxygenated hydrocarbons.
flame and measure the resulting reductions in smoke height. The advantage is that the results are more directly related to the issue of reducing soot formation in engines. It has been applied to 24 varied oxygenates,11 with either a heptane/toluene mixture or a commercial diesel fuel as the base fuel. The new method used in this study is to add small amounts of oxygenates to a lightly sooting base flame and measure the resulting increases in soot concentration. The primary advantage is that the additive concentrations—1000 ppm—are too small to affect the soot concentrations through indirect mechanisms such as dilution of the base fuel or changes in flame temperature; thus the measured sooting tendencies depend strongly on the direct chemical effects of oxygenate structure. Another advantage is experimental convenience, which has enabled us to characterize hundreds of test compounds in a reasonable time. We emphasize that all three of these methods quantify the same fundamental property: the effectiveness of a test compound’s combustion products as precursors to soot. Whether oxygenates enhance or reduce the total soot concentration in a particular flame is just a function of whether their products are better or worse soot precursors than those of the base fuel. Indeed, all oxygenated hydrocarbons are soot-enhancing if they are added to a hydrogen flame, and all of them are soot-reducing when added to a graphite flame. In this study we have chosen a base fuel—methane—which is very lightly sooting so that small amounts of the test compounds would produce detectable changes in soot concentration; thus the oxygenates are soot-enhancing.
’ EXPERIMENTAL METHODS Sooting tendencies of oxygenated hydrocarbons were determined by doping 1000 ppm of each test compound into the fuel of a laboratory-scale burner and measuring the maximum mass concentration of soot in the resulting flames with laser-induced incandescence (LII).5,6 The burner generates atmospheric-pressure laminar nonpremixed coflow flames in which a gaseous fuel mixture—CH4, N2, and the test compound—flows out of an 11 mm diameter tube and reacts with air that flows from the annular region between this tube and an outer chimney (see Supporting Information (SI) Figure S1). The nominal reactant flow rates were 330 cm3/min (CH4), 275 cm3/min (N2), 0.61 cm3/min (test compound), and 30 000 cm3/min (air).
ARTICLE
The test compounds, all of which are liquids at room temperature, were injected into the gaseous methane/nitrogen mixture by a syringe pump. The syringe needle entered the fuel line through a septum in a stainless steel tee. Temperaturecontrolled resistive tapes heated the fuel line and the fuel tube of the burner to 145 C; given this heating all of the test compounds vaporized immediately upon injection and then were swept as gases to the flame by the other fuel components. Each test compound flowed for at least 5 min before data acquisition commenced. Time-resolved soot measurements confirmed that all of the test compounds achieved adsorption/desorption equilibrium with the walls of the fuel line and the fuel tube during this 5 min interval. We tested the accuracy of the syringe pump by loading it with a caliper instead of a syringe and measuring the displacement during a given time interval; the results showed that the pump was a negligible source of error. The liquid-phase flow rate that corresponded to 1000 ppm in the gas-phase for each test compound was calculated using the ideal gas law, the test compound’s molecular weight, and its liquidphase density.18,19
’ RESULTS AND DISCUSSION The Sooting Tendency Database. In this study we determined the sooting tendencies of oxygenated hydrocarbons by separately doping 1000 ppm of each test compound into the fuel of a methane/air flame and measuring the resulting maximum mass concentration of soot. Since the raw results depend strongly on our specific experimental parameters, we have followed the standard practice of converting them into a sooting index that is analogous to an octane rating.20 Specifically, we have defined a Yield Sooting Index:
YSIðiÞ ¼ AMðiÞ þ B
ð1aÞ
YSIðn-hexaneÞ ¼ 0:0
ð1bÞ
YSIðbenzeneÞ ¼ 100:0
ð1cÞ
where YSI(i) is the yield sooting index of species i, M(i) is the measured maximum concentration of soot in the flame doped with species i, and A and B are apparatus-dependent constants chosen to satisfy 1b and 1c. N-hexane and benzene were chosen as the end points because they are stable, inexpensive hydrocarbons that span roughly the same range of sooting tendency as the oxygenated test compounds. We used different end points in our studies of aromatic hydrocarbons,5,6 which are much sootier, so the YSI’s are not directly comparable. The procedure for determining YSI’s was as follows. First, we doped n-hexane, benzene, and four test compounds into the base flame and measured M(i) in each case. Second, we used the measured M(n-hexane) and M(benzene) to obtain the values of A and B that satisfied 1b and 1c. Third, we used these values of A and B to calculate YSI(i) for each test compound from its M(i). Fourth, we repeated this entire process, with slightly different values of A and B arising each time, until we had acquired at least three determinations of YSI for each test compound. Finally, we averaged these determinations to arrive at the final values of YSI. YSI’s were measured for 275 test compounds. The full results are listed in SI Table S1. They include 29 alkanes, 16 cycloakanes, 36 alkenes, 4 cycloalkenes, 4 alkynes and alkadienes, 47 alcohols, 13 ethers, 28 alkanones, 18 aldehydes, 5 cyclic ethers, 44 esters and carboxylic acids, and 31 other multiply oxygenated 2499
dx.doi.org/10.1021/es103733q |Environ. Sci. Technol. 2011, 45, 2498–2503
Environmental Science & Technology
Figure 2. Measured sooting tendencies of alkanes as a function of their carbon number. The symbols are the measured data points and the curve is a quadratic least-squares fit to the data for the n-alkanes.
compounds. The primary constraint on the test matrix was that many interesting compounds, such as heavily branched alkanes and alcohols, are not commercially available. Another constraint was that the test compounds had to be liquids at room temperature with boiling points between 34 and 210 °C; species outside this range either spontaneously vaporized out of the syringe or they required excessive time to equilibrate with the walls of the fuel line and fuel tube. Uncertainties. After a careful analysis, we have identified two main sources of uncertainty in the measured YSI’s. The first is random variation caused by the finite precision of the LII diagnostic. In order to assess this uncertainty, we used isooctane as one of the four test compounds in every set of measurements. In the end, YSI(isooctane) had been measured 79 times; the results ranged from 47.4 to 50.1 with an average of 48.6 and a standard deviation of 0.58. Isooctane is a worst-case scenario for random scatter since its YSI is halfway between the end points. The second source of uncertainty is possible error in the literature values of the test compounds’ liquid-phase densities, which were needed to calculate the liquid-phase flow rate corresponding to 1000 ppm in the gas-phase. Overall, we estimate that the uncertainty in the measured sooting tendencies is (2 YSI units. As a test of the uncertainties in the database, we derived quadratic least-squares fits to the YSI’s of 6 homologous series as a function of carbon number. The sooting tendencies for each of these series would be expected to vary in a systematic manner with molecular size, so the deviation of the individual data points from the fits indicates the scatter in the data. The average and maximum deviations are 0.50 and 0.74 YSI units (n-alkanes), 1.25 and 2.59 (1-alkenes), 0.91 and 1.92 (1-alcohols), 0.97 and 2.57 (n-aldehydes), 0.63 and 1.40 (2-alkanones), and 0.39 and 0.68 (methyl esters). These values are consistent with the quoted uncertainty. Sooting Tendencies of Alkanes. Alkanes are the main components of all petroleum-derived fuels; thus they establish the baseline for evaluating the effects of oxygenates on soot formation. Figure 2 shows the sooting tendencies measured for them in this study. The vertical axis is yield sooting index, YSI, so a data point higher in the figure indicates a fuel with a greater sooting tendency. The horizontal axis is carbon number, NC, so the left side corresponds to smaller species and the right side to larger species. The vertical error bars around the n-alkane data points (red circles) indicate the (2 uncertainty in YSI; since they
ARTICLE
Figure 3. Measured sooting tendencies of linear alcohols and of nalkanes as a function of their carbon number. The symbols are the measured data points and the curve is a quadratic least-squares fit to the data for the n-alkanes.
are the same size as the symbols, they are not shown for the other data points or on other figures. The curve is the quadratic leastsquares fit to the n-alkanes data; as discussed above, the scatter of the data points relative to this curve is consistent with the uncertainty. The YSI’s for the n-alkanes increase steadily with increasing NC. The YSI’s also increase as the carbon-chain becomes more branched, but are not very sensitive to the specific arrangement of the branches. For example, the YSI’s of the octane isomers (NC = 8) increase in this order: 18.9 (n-octane), 27.5 to 29.5 (3 methylheptane isomers), 34.8 to 39.9 (4 dimethylhexanes), and 47.3 to 48.6 (2 trimethylpentanes). These trends agree with many other sooting tendency studies.7,8,10 Small n-alkanes predominantly react to ethylene and methyl radical, which are inefficient soot precursors since they require slow addition reactions to form the C3 and C4 species that are necessary precursors to benzene.13 As alkanes become longer or more branched, their combustion products include a greater proportion of higher alkenes such as propene and butene, so their sooting tendencies increase.21 Sooting Tendencies of Alcohols and Ethers. Alcohols and ethers have been used widely as fuel additives and extenders, especially ethanol, butanol, dimethyl ether, and methyl tert-butyl ether (MTBE).1 Figure 3 shows the YSI’s measured for linear alcohols and linear ethers in this study. The data points for the nalkanes from Figure 2 are also shown for comparison. At any given NC, the YSI’s systematically increase in this order: ethers
