Energy & Fuels 1989, 3, 193-199 cluster. The number of methyl groups varies by almost a factor of 3 among the coals studied. The remaining attachments are of two types; either bridges between aromatic clusters or aliphatic loops on an aromatic ring (hydroaromatic structures). Also, included in Table IV is the total molecular weight of an average cluster (calculated from the elemental analysis and f a , ) . The variation in average cluster molecular weight is a combination of cluster size and elemental composition where the molar O/C ratio is a significant factor in the estimated molecular weight. The diversity of structure observed from NMR data on both aliphatic and aromatic carbon types in these coals confirms the expected complexity in the macromolecular structure. In each instance the NMR data are consistent with the structural trends in coals expected for changes in coal rank. The level of detail derived from these data provides a description of the carbon skeletal backbone of these coals and is useful in coal devolatilization It is anticipated that such structural information will also be valuable in assessing liquefaction processes.
193
Acknowledgment. This work was supported by the National Science Foundation under Cooperative Agreement No. CDR 8522618. Also participating in the funding of this effort, in alphabetical order, was a consortium of organizations that includes Advanced Fuel Research, Inc.; Allison Division (General Motors Corp.); Babcock and Wilcox; Chevron Research Co.; Combustion Engineering, Inc.; Conoco, Inc.; Convex Computer Corp.; Corning Glass Works; Dow Chemical USA; Electric Power Research Institute; Empire State Electric Energy Research Corp.; Foster Wheeler Development Corp.; Gas Research Institute; General Electric Co.; Los Alamos National Laboratory; Morgantown Energy Technology Center (US. Department of Energy); Pittsburgh Energy Technology Center (US. Department of Energy); Pyropower Corp.; Questar Development Corp.; Shell Development Co.; Southern California Edison; the State of Utah; Tennessee Valley Authority; and Utah Power and Light Co. Financial support from Brigham Young University and the University of Utah is also acknowledged.
Correlation of Bituminous Coal Hydroliquefaction Activation Energy with Fundamental Coal Chemical Properties S.-C. Shin, R. M. Baldwin,* and R. L. Miller Chemical Engineering and Petroleum Refining Department, Colorado School of Mines, Golden, Colorado 80401 Received May 9, 1988. Revised Manuscript Received November 28, 1988
The rate and extent of direct coal hydroliquefaction for five bituminous coals from the Argonne Premium Sample Bank have been measured. Data were obtained in batch microautoclave tubing bomb reactors at three temperatures (375,400,425 "C) and five residence times (3,5,10,15,40 min) in 1-methylnaphthalene vehicle under a hydrogen blanket. Data on rate of conversion of coal to THF and toluene solubles were modeled with a simple reversible rate expression, and activation energies for conversion to each solvent solubility class were determined. Data on carbon and proton distribution in the coals were obtained by 'H NMR combined rotational and multiple-pulse spectroscopy and 13CNMR CPMAS/dipolar dephasing spectroscopy. A strong correlation of activation energy with the aliphatic hydrogen content of the coal was found for conversion to THF solubles. Activation energies for toluene solubles were found to be highly correlated (R2 > 0.90) with total oxygen and with protonated aliphatic carbon. 13CNMR data indicated that a direct relationship existed between protonated aliphatic carbon and total oxygen for the bituminous coals from the Argonne suite, and this structural feature was found to be very significant in determining coal reactivity. This observation provides direct evidence of the importance of etheric oxygen in cross-linking structures for determination of bituminous coal hydroliquefaction reactivity.
Introduction The relationship between coal chemical and structural features and coal reactivity has been the subject of considerable research for well over 70 years. These studies date from the original observations bf Bergius' regarding the influence of carbon content on hydroliquefaction yield. Early studies of the influence of coal properties on coal reactivity were focused on an attempt to find a single parameter or group of parameters capable of correlating (1) Bergius, F. J . Gasbeleucht. Verw. Beleuchtungsarten Wasseruersorg. 1912,54.
physical, chemical, and geochemical properties with the degree of COnversion to Some solvent-soluble Products under a fixed set of reaction conditions.2-8 Given et al.+l2 (2) Francis, W. Fuel 1932, 11, 171. (3) Petrick, A. J.; Gaigher, B.; Groenewoud, P. J. Chem. Metall. Min. SOC.S. Afr. 19371 38. (4) Wright, C. C.; Sprunk, G. C. Pa. State Colt., Min. Ind. Exp. Stn. Bull. 1939, No, 28. (5) Fisher, C. H.: Srxunk. G. C.: Eisner. A.: Clarke. L.: Fein, M. L.; Storch, H. H. Fuel 19i0, 19, 132. (6) Shatwell, H. G.; Graham, J. L. Fuel 1925, 4 , 25. (7)Horton, L.;Williams, F. A.; King, J. G. G.B. Dept. Sci. Znd. Res., Fuel Res., Tech. Pap 1936, No. 42.
0887-0624/89/2503-Ol93$01.50/00 1989 American Chemical Society
194 Energy & Fuels, Vol. 3, No. 2, 1989
Shin et al.
Table I. Characterization Data of Argonne Premium Coals: Elemental and Proximate Analyses“ proximate anal., wt elemental anal., wt % (daf) (dry) coal seam state rank C H 0 Sh, ash VM Illinois No. 6 IL HVB 77 5.7 10 5.4 14.8 35.4 Pittsburgh No. 8 PA HVB 83 5.8 9 1.6 9.7 34.9 Lewiston-Stockton wv HVB 81 5.5 11 0.6 20.1 28.6 Upper Freeport PA MVB 87 5.5 4 2.8 13.1 25.1 Pocahontas No. 3 VA LVB 91 4.7 3 0.9 5.3 17.1
“ Key:
%
FC 44.7 53.6 49.4 60.7 77.1
VM = volatile matter; FC = fixed carbon.
and ‘HNMR Characterization Data for Argonne Premium Coals 13C NMR,” wt %/MW of ‘H NMR,* wt %/MW of carbon hydrogen state rank c,, Cfuo Cdi H,, H,o Cdi IL HVB 6.42 4.42 2.00 5.7 2.74 2.96 PA HVB 6.92 5.17 1.75 5.8 2.21 3.59 wv HVB 6.75 5.00 1.75 5.5 1.68 3.82 PA MVB 7.25 5.92 1.33 5.5 2.10 3.40 VA LVB 7.58 6.75 0.83 4.7 2.57 2.13
Table 11.
coal seam Illinois No. 6 Pittsburgh No. 8 Lewiston-Stockton Upper Freeport Pocahontas No. 3
“All values on an maf atomic basis. Aromaticity data were furnished by Dr. Ronald Pugmire, University of Utah. bAll values on MAF atomic basis. Proton NMR data were furnished by Dr. Charles Bronnimann, NSF Regional NMR Center, Colorado State University.
developed multiparameter statistical correlations for coal reactivity for an extensive suite of US. coals. These correlations were partitioned according to the geographical location of the coal in order to increase the significance of the relationships. Other researchers have attempted to correlate coal hydroliquefaction reactivity with the mineral matter in coal, most notably pyrite, and other physical and chemical Neavellg and Furlong et alaz0 introduced the concept of using reaction rate as a reactivity parameter while Shin et a1.21attempted to combine reaction rate and reaction extent into a single reactivity parameter that could then be correlated to coal properties. GutmannZ2correlated reaction rate constants with the sulfur content of lignite coals. Other reactivity relationships based on information from instrumental techniques such as pyrolysis/mass spectrometry have been developed by Baldwin et al.23and Marzec et al.24 All of the above attempts to relate coal liquefaction reactivity to coal properties have suffered from two limitations. First, the parameters employed as reactivity definitions have generally fallen into two categories: (1) reactivity as defined by a point-yield definition a t fixed time and temperature (e.g. toluene solubles at 60 min and (8) Fisher, C. H.; Sprunk, G. C.; Eisner, A.; O’Donnell, H. J.; Clarke, L.; Storch, H. H. Technical Paper 642; US.Bureau of Mines: Washington, DC, 1942. (9) Given, P. H.; Cronauer, D. C.; Spackman, W.; Lovell, H. L.; Davis, A.; Biswas, B. Fuel 1975, 54, 34. (10) Fuel 1975, 54, 40. (11) Abdel-Baset, M. B.; Yarzab, R. F.; Given, P. H. Fuel 1978,57,89. (12) Yarzab, R. F.; Given, P. H.; Spackman, W.; Davis, A. Fuel 1980, 59, 81. (13) Mori, K.; Taniuchi, M.; Kawashima, A,; Okuma, 0.;Takahashi, T. Prepr. Pap.-Am. Chem. SOC.Diu. Fuel Chem. 1979,24(2), 28. (14) Durie, R. A. Prepr. Pap-Am. Chem. SOC.Diu. Fuel Chem. 1979, 24(2). (15) Gray, D.; Barrass, G.; Jezko, J.; Kershaw, J. R. Fuel 1980,59,146. (16) Mukherjee, D. K.; Chowdhury, P. B. Fuel 1976,55, 4. (17)Guin, J. A.; Tamer, A. R.; Prather, J. W.; Johnson, D. R. Ind. Eng. Chem. Process Des. Deu. 1978, 17, 118. (18) Gray, D. Fuel 1978, 57. (19) Neavel, R. C. Fuel 1976, 55, 237. (20) Furlong, M. W.; Baldwin, R. M.; Bain, R. L. Fuel 1982, 61. (21) Shin, S.-C.; Baldwin, R. M.; Miller, R. L. Energy Fuels 1987, 1, 371. (22) Gutmann, M.; Radeck, D.; Konig, M.; Keil, G. Coal Sci. Technol. 1987, 11, 187. (23) Baldwin, R. M.; Durfee, S. L.; Voorhees, K. W. Fuel Process. Technol. 1987, 15, 281. (24) Schulten, H. R.; Simmleit, N.; Marzec, A. Fuel 1988, 67, 619.
450 “C); (2) reactivity as defined by a reaction profile with time a t fixed temperature. While both of these “traditional” reactivity definitions can be correlated with liquefaction reactivity, the correlations developed are not truly universal in that they may not hold at different time and/or temperature levels. The second major limitation of prior studies is that the correlations have not been developed in terms of fundamental structural and chemical properties of coal. In many cases, derived coal properties such as volatile matter, fixed carbon, rank, heating value, etc. have been used as the independent variables in multiparameter statistical models. As pointed out by N e a ~ e 1such , ~ ~ correlations are of limited significance. The above discussion highlights the need for a reactivity parameter that is independent of both time and temperature, yet can be correlated with basic coal chemical properties. Further, detailed knowledge of the chemical structure of coal is needed so that the reactivity correlations developed have some significance in terms of the chemistry of the liquefaction process. We have attempted to address both of these problem areas by using activation energy as a reactivity definition and by exploring the use of structural information from two new NMR techniques (‘HNMR/CRAMPS and 13C NMR/CPMAS with dipolar dephasing) that do permit determination of fundamental chemical structural features important in coal liquefaction reactivity.
Experimental Section Five bituminous coals from the Argonne Premium coal collection were liquefied in 1-methylnaphthalene vehicle. Characterization data for the Argonne coals are shown in Tables I and 11. Experimental runs were carried out in batch tubing bomb microautoclave reactors at 375,400, and 425 “C, and at residence times of 3, 5, 10, 15, and 4 0 min by using a fluidized sand bath as the heating medium. Temperature was measured by an internal thermocouple inserted directly into the reaction mass. Temperature control was adequate to maintain the reactor at i l OC of the desired reaction temperature for the duration of the run. T h e reactor typically reached reaction temperature within 90 s and was quenched following reaction by immersion in an ice bath. All runs were performed in a hydrogen atmosphere under an initial (cold) pressure of 6.2 MPa. The reactors employed in these runs were approximately 20 cm3 in volume. A typical run used 1 g of (25) Neavel, R. C. Fuel 1986, 65, 312.
Fundamental Coal Chemical Properties
Energy & Fuels, Vol. 3, No. 2, 1989 195 SOLVENT EFFECT ON REACTIVITY RANKING
Table 111. Aliphatic a n d Aromatic Hydrogen Contents of Argonne Premium Coalsa
%H Goal Illinois No. 6 Pittsburgh No. 8 Stockton Upper Freeport Pocahontas
aromatic 48.0 38.1 30.6 38.2 54.6
TOLUENE CONVERSION
(Yo
DAF)
~
60
aliphatic 52.0 61.9 69.4 61.8 45.4
a Proton spectra were obtained at a proton Larmor frequency of 187 MHz by using a BR-24 pulse sequence. The 90' pulse width was 1.1 ps and the cycle time of 367 was 108 ps. Samples were spun at the magic angle at a nominal rate of 2 KHz. Proton spectra was measured at the NSF Regional NMR Center, Colorado State University.
coal and 1g of vehicle. The reactor contained two small stainless-steel balls to facilitate agitation of the reaction mixture, and the vessel was agitated on a vertical axis a t a rate of approximately 120 cpm while in the sand bath. Data on the rate and extent of coal conversion to the solvent solubility classes were collected by using a standard solvent fractionation procedure involving sequential extraction of the reaction products with THF and toluene. Further details concerning experimental methods have been reported elsewhere.26 The experimental program consisted of two phases. During the first phase, an examination of the effect of vehicle on the measured liquefaction reactivities was carried out by using three aromatic compounds (l-methylnaphthalene, naphthalene, and phenanthrene). The second phase of the project concerned measurement of the data, kinetic modeling, and correlation with coal properties. Proton NMR data for the Argonne coals were obtained from the National Science Foundation Regional Center for NMR a t Colorado State University, utilizing the technique of combined rotational and multiple-pulse spectroscopy (CRAMPS). Basic data on the proton distribution in these coals are shown in Table 111. Information on the carbon distribution via dipolar dephasing of 13C NMR data was obtained from Dr. R. L. Pugmire a t the University of Utah, and these data are presented in Table IV.
Discussion of Results Effect of Vehicle on Liquefaction Reactivity. In order to properly evaluate and compare the inherent reactivities of different coals, a non hydrogen donor liquefaction vehicle was required so that reactivity differences between similar coals within a given rank or subrank could be magnified. The first phase of this experimental program was thus concerned with an evaluation of the effect of three aromatic vehicles on relative reactivities. Liquefaction experiments were performed on the five Argonne bituminous coals in l-methylnaphthalene (1-MN), naphthalene, and phenanthrene, and the relative reactivities of these coals were assessed by toluene conversion at 5 and 40 min. Figure 1presents the results of the 40-min runs. These data clearly indicate that the relative reactivities are independent of the choice of nondonor vehicle and that conversions in l-methylnaphthalene are equivalent to those in the other two aromatic vehicles. Data from the 5-min runs led to similar conclusions. Because of the added ease of operation afforded by 1-MN, this material was utilized as the liquefaction vehicle in all subsequent experiments. Rate Data and Modeling. Rate data for conversion of the Argonne coals to THF and toluene solubles are presented in Tables V and VI. Duplicate runs were made on the Illinois No. 6 coal in order to test for reproducibility. Various kinetic models were investigated for purposes of data fitting, including the following expressions: first-order (26) Shin, S.-C. M.S. Thesis No. T-3267, Arthur Lakes Library, Colorado School of Mins, 1988.
1-MN
lllinots#6
&@8 Upper
Freeport
NAPHTHALENE SOLVENT TYPE
-_-_-. Pittsburgh # 8
PHENANTHRENE
Stockton
Pocahontas
F i g u r e I. Toluene conversion a t 40-min reaction time.
irreversible; first-order reversible; second-order irreversible; second-order reversible; first order (forward)/second order (reverse); Anthony-Howard The model finally chosen for computation of the activation energies was based on an evaluation of the goodness of fit for the rate data to the integrated batch reactor mass balance expressions: first-order irreversible rate expression, dx/dt = k(a - x ) ; integrated form, x = a(1- exp(-kt)); second; order irreversible rate expression, dx/dt = k (a - x ) ~ integrated form, x = ka2t/(kta + 1); first-order reversible rate expression, dx/dt = kf(a - x ) - k p ; integrated form, x = x,(l - exp(-kfat/x,)); second-order reversible rate expression, dx/dt = kf(a - x ) -~ k p 2 ; integrated form, x = a / U - Q2)[1+ &((I- P b p ( R t ) I ) / ( l + P[exp(Rt)l))l; first-order/second-order reversible rate expression, dx/dt = kf(a - x ) - k,x2; integrated form, x = [ax,(l - exp(St))]/[a + (a-x,)(exp(-St))]. In these expressions x = coal converison (% daf), x, = pseudoequilibrium conversion (% daf), a = ultimate conversion (% daf), k = kinetic constant (f = forward, r = reverse), t = reaction time (min),P = a / ( a - Zx,), Q = (a - x e ) / x e , R = 2kfaQ, and S = kf(2a - x,)/x,. Model discrimination for the above expressions was performed by computing the following statistic for each data set: sum = [ ( x - x * ) ~ ] O - ~ / ~ where x = experimental value for coal conversion, x* = calculated value of coal conversion, and n = number of data points. While large differences between the adequacies of the various models in terms of the above statistic did not exist, the first-order/second-orderreversible model was found to be the best overall data descriptor as indicated by the lowest value of "sum" for all of the data sets. The analysis of variance from regression modeling of the Illinois No. 6 coal (the replicated data set) using the first-order/second-order rate expression is presented in Table VII. As can be seen, these data gave very low values €or the pure error mean square, indicating that the data were highly reproducible. These statistics infer that the model provides a satisfactory descriptive expression for the data. Traditional Correlations. As mentioned above, the traditional parameters that have been used for coal liquefaction reactivity correlations are point-yield conversion and a kinetic rate constant. We thus began our correlational efforts by exploring similar types of relationships. Figure 2 presents three correlations using the point-yield (27) Anthony, D. B.; Howard, J. B. AIChE J. 1976,22,625.
Shin et al.
196 Energy & Fuels, Vol. 3, No. 2, 1989 Table IV. Percent Carbon Distribution of Argonne Premium Coals' Illinois No. 6 Pittsburgh No. 8 Stockton Upper Freeport total aliphatic (sp3) 28 28 25 19 8 Hali 18 13 18 11 Nali 10 15 7 (2) (Oad (5) (3) (4) aromatic (sp2) ring,, R-H, R-N,, R-N-P,, R-N-A,, R-N-B,, carbonyl,,
72
72 72 26 46 6 18 22 0
75 75 27 48 5 21 22 0
72 27 45 6 17 22 0
Pocahontas 14 7 7 (1) 86 86 33 53 2 17 34
81 81 28 53 4 20 29 0
0
"Key: Hdi = fraction of total carbon that is sp3-hybridized and CH or CH2; Ndi = sp3-hybridized and CH3 or nonprotonated; 09 = sp3-hybridized and bonded to oxygen; R-H,, = sp2-hybridized and protonated; R-N,, = sp2-hybridized and nonprotonated; R-N-P, = sp2-hybridized and phenolic or phenolic ether; R-N-A, = sp2-hybridized and alkylated; R-N-B,, = sp2-hybridized and at a bridgehead position. Data were obtained on a Bruker CXP-100instrument with a single contact time of 2.5 ms and a pulse delay time of 1.0 ms. Carbon distribution data were furnished by Dr. Ronald Pugmire, University of Utah.
Table V. Kinetic Data for Formation of Toluene Solubles (wt % Daf Basis) kinetic data (wt % ) for reaction times coal seam temp, "C 3 min 5 min 10 min 15 min 40 min Pittsburgh 425 17 23 28 36 52 No. 8 400 10 20 24 27 40 375 8 10 15 18 27 Lewiston425 16 20 24 29 33 Stockton 400 8 9 16 18 27 375 4 7 8 9 12 Upper 425 16 20 22 29 41 Freeport 400 13 16 19 22 30 375 12 13 13 20 26 Pocahontas 425 10 12 13 15 17 No. 3 400 9 11 13 14 14 375 9 9 10 10 11 Illinois No. 6 425 24 36 43 47 52 (double) 22 34 39 42 54 400 18 26 35 40 58 15 27 38 42 58 375 6 14 23 30 44 8 12 24 30 44 Table VI. Kinetic Data for Formation of THF Solubles (wt % Daf Basis) -kinetic data (wt % ) for reaction times coal seam temp, "C 3 min 5 min 10 min 15 min 40 min Pittsburgh 425 42 74 ao 79 79 30 43 55 71 85 No. 8 400 42 45 68 375 24 31 36 45 45 49 48 Lewiston425 39 47 52 Stockton 400 22 29 15 20 25 30 41 375 Upper 425 45 53 61 69 82 43 51 64 76 Freeport 400 36 26 27 36 47 375 23 23 28 31 40 Pocahontas 425 17 26 30 No. 3 400 11 19 24 16 19 21 24 375 10 70 71 76 78 Illinois No. 6 425 63 75 79 (double) 63 74 74 66 78 83 88 400 50 52 63 79 83 88 72 86 375 39 51 66 51 63 70 84 43
reactivity definition. Here, conversion to toluene solubles at 5 min is used as the yield parameter. As shown, a very strong correlation exists between toluene conversion a t 5 min and 425 "C and total oxygen plus total carbon in the coal as indicated by the square of the multiple correlation coefficient (R2)value of 98%. However if this same cor-
F v
9
40-30--
3
$w
20--
W z
3
$ lo-R2 = .98
A
-
I
6.9
7.4
TOTAL CARBON
F v
7.9
+ OXYGEN (WT%/MW)
--I
e
50
40 --
e e
30 -20 --
40 min, 425 OC R2 = .61
e
I
104
6.9
7.4
TOTAL CARBON
7.9
+ OXYGEN (WT%/MW)
14
e v
10 --
e e
8 --
5 min, 375 OC
e
R2 = 0
64 6.9
7.4
TOTAL CARBON
+ OXYGEN (WT%/MW)
I
9
Figure 2. Point-yield reactivity vs atomic carbon + oxygen: (A) 425 "C, 5-min reaction time; (B) 425 "C, 40-min reaction time; (C) 375 "C, 5-min reaction time.
Fundamental Coal Chemical Properties
Energy & Fuels, Vol. 3, No. 2, 1989 197
Table VII. Analysis of Variance for the First-Order/Second-Order Reversible Model for Illinois No. 6 Coal, Duplicated Data Set" toluene solubles THF solubles source DF ss MS std dev DF ss MS std dev regression 1 38213 38 213 1 148 787 148 787 29 323 29 194 7 2.59 11 3.33 error 2.5 1.57 pure error 15 42 2.8 1.67 15 37 lack of fit 14 152 10.9 3.29 14 286 20.4 4.52 tot.
30
38 407
30
149 110
Key: DF degree of freedom; SS, sum of square; MS, mean square; std dev, standard deviation.
different chemical functionalities a t other temperatures and times. This concept of a distribution of reaction pathways was the basis for the development of the Anthony-Howard model. However, when this expression was applied to our rate data, little variation in the derived activation energies between the different coals was found. A strong dependency of the derived activation energies with coal type was found, however, for the other rate expressions listed above. It is important to recognize that the activation energies derived from this exercise have no fundamental significance in terms of the actual chemistry taking place; hence, the absolute values of these parameters are not important. It also should be stressed that the absolute values of these parameters are functions of the vehicle and that different activation energies would be determined in different aromatic or hydroaromatic vehicles. What is important and what is needed for purposes of correlation, however, is a relative indicator of the reactivity of the different coals. An activation energy was calculated from the rate data for each of the six models listed above for conversion to both toluene and T H F solubles. With the exception of the Anthony-Howard model, each of the kinetic expressions gave relative coal reactivity rankings (in terms of the derived activation energies) that were identical. Hence, the conclusions reached regarding the relative reactivities of the coals in this study are not dependent on the rate model employed, and the final choice of an appropriate kinetic expression does not in any way bias the ensuing correlation efforts. The calculated activation energies from the first-order/second-order reversible rate expression are presented in Table VIII. Also presented in this table are the multiple correlation coefficients for the fit of the derived rate constants to an Arrhenius plot, from which the activation energies were ultimately determined. As shown, exceedingly good fits to the expected Arrhenius temperature dependency were found for the toluene and T H F rate constants. The large differences in activation energies found for these five coals suggest that this parameter should be useful as a relative reactivity indicator. Before initiating correlation of reactivity with coal properties, it is important to determine which of the coal characteristics available are independent parameters and which are highly correlated with other characteristics and thus are simply derived properties. Eight coal charac-
Table VIII. Activation Energies for Conversion of Argonne Premium Coalsn toluene solubles T H F solubles coal seam E , kcal/mol R2,% E, kcal/mol R2,% Illinois No. 6 20.9 99.8 17.5 99.5 24.5 96.3 Pittsburgh No. 8 15.5 95.2 29.5 98.3 Lewiston-Stockton 26.8 94.3 20.1 99.3 Upper Freeport 9.9 97.0 7.5 98.3 Pocahontas No. 3 4.3 100
R2values represent squared multiple correlation coefficient of Arrhenius plots. relation is extended to a different time (40 min) a t the same temperature, a much weaker correlation (R2= 61%) results. If the temperature is changed to 375 "C while time is held constant (5 min), essentially no correlation (R2 = 0) exists. Figure 3 shows the same sort of effect, but with a kinetic constant (from the first-order/second-order model) used as the reactivity parameter. A very strong correlation is found a t 425 O C between the rate constant for conversion to THF solubles and the total carbon content of the coal (R2= 94%). Since the rate constant is time independent, this is a valid reactivity parameter as long as the temperature is not altered. However, if the temperature is reduced to 400 and then to 375 "C, the same correlation weakens significantly and then disappears, with R2 falling to 54% and 1296, respectively, at the lower temperatures. This graphically illustrates the inadequacy of both of these "traditional" reactivity parameters, which are either time and/or temperature dependent. The above discussion demonstrates the need for a reactivity parameter that is time and temperature independent. This leads logically to consideration of activation energy as a fundamental reactivity parameter, since: activation energy includes temperature effects; activation energy should reflect more closely the chemical nature of the parent coals; activation energy is obtained by measuring time and temperature effects, but the result is time and temperature independent. Activation Energy Correlation. Intuitively, the behavior noted above might be interpreted as a shift in the controlling mechanism for the liquefaction reaction, with one type of chemical functionality dominating the reaction under one set of conditions. This would be accompanied by a gradual shift to different reaction pathways and hence
ctnt ctnt
Cai Htnt Hero stat
ash VM
Table IX. Interrelationships between Coal Properties" Htat H , , om stat
Cali -92
-49 +75
0 0
0
-79 +75 +27
0 +0.7 0 0
ash -4 1 +34 +8 +49 0
VM -89 +99 +80 +66 +6 +22
Values represent squared multiple correlation coefficients (%) adjusted for degrees of freedom. Positive signs (+) indicate direct relationships while negative signs (-) indicate inverse relationships.
Shin et al.
198 Energy & Fuels, Vol. 3, No. 2, 1989
+ z F Lo
400 - _
0
300 --
z 0
Table X. Correlation of Activation Energy with Coal Properties activation energy correlno coal property toluene solubles THF solubles Cdi 64.8 32.0 -.
cfm
W
2
a n
a
67.9 23.8 32.5 93.2 0.0 72.8 50.7 91.6 27.4
H,,
zoo --
Hdi OtQt
s,,
e loof 04
A
ash
R2 = .94 4250c
6.2
H,t/C,t O,t/Cbt 7.2
6.7
7.7
TOTAL CARBON (WT%/MW)
"""
z +
Ez
0
I
i
p
v
e
0
20-
> c)
e
W E
e
100-
z
W
!
400 OC R2 = .54
IO--
e
$
04
6.7
6.2
7.2
7.7
TOTAL CARBON (WT%/MW)
200
e
R 2 = .93
i
e
B
I
51
200-
s
ze
+ c,t
Values represent squared multiple correlation coefficients ('70 ) adjusted for degrees of freedom.
e
V W
E
otnt
21.9 42.4 95.1 36.2 0.0 37.1 29.2 29.9 0.0
0
5 10 TOTAL OXYGEN (WE%)
15
Figure 4. Toluene solubles activation energy vs total oxygen.
e
IZ
F v)
z
30 --
0 V W
z
100
e
0
*
ze
e
e
20 --
e
375 oc l o --
R2 = .12 0
c '
6.7
7.2
I
TOTAL CARBON (WT%/MW)
Figure 3. Rate constant vs atomic carbon: (A) 425 "C; (B) 400 "C; (C) 375 "C.
teristics, including basic data from proximate and ultimate analyses along with the proton NMR (CRAMPS) information, were first investigated in this manner. Results of single-parameter linear regression analysis on these properties are shown in Table IX. Volatile matter and total carbon were found to be strongly correlated with aliphatic carbon content, while total carbon was found to be highly correlated with carbon aromaticity. Total sulfur, ash, and aromatic (and hence aliphatic) hydrogen content of the coal were found not to be related to any of the other properties of the coal. Table X presents the results of correlation of the liquefaction activation energies for the five bituminous coals with the independent coal properties identified in Table IX. The activation energy for coal conversion to toluene solubles was found to be strongly (R2 = 93%) correlated with the total oxygen content of the coal, while the activation energy for conversion to T H F solubles was found to be highly correlated (R2= 95%) with aliphatic hydrogen content. These correlations are shown graphically in Figures 4 and 5. Activation energies for conversion to
04
2.0
3.0
3
ALIPHATIC HYDROGEN (W)
Figure 5. THF solubles activation energy vs aliphatic hydrogen.
toluene and THF solubles were not found to be highly correlated with any of the other coal properties in Table X, suggesting that the relationships found between reactivity and these two coal properties are unique. Data from 13C NMR spectroscopy were next investigated, and simple single-parameter correlations of these properties with THF and toluene activation energies attempted. Relationships between the various 13C NMR properties and other coal characteristics were first investigated in a manner similar to the one described above. The only intercorrelation found from this data set was for total oxygen and protonated aliphatic carbon (sp3 hybridized and CH or CHJ. The correlation between these two parameters for the six Argonne bituminous coals (including Blind Canyon) was almost 99%; however, when the data for the subbituminous and lignite coals was added, the correlation no longer held. These data are shown graphically in Figure 6. This finding may have strong implications with respect to coal structure and the differences in reactivity found between low-rank and high-
Fundamental Coal Chemical Properties
Energy & Fuels, Vol. 3, No. 2, 1989 199
1.5
d
2-
It
Beulah-Zap
LL
0
Wyoda k
$
1.0
Stockton
lllinois%f
20
v
Blind Ca
v
f I
Stock III#6
6 0 0.5
Pittsburgh Y 8
z
W
Pitt #0
a
E
0
0.0
0.0
Up Free POC
0
0.5
1 .o
O J
1.5
2.0
PROTONATED ALIPHATIC CARBON (W)
Figure 6. Correlation of total oxygen and protonated aliphatic carbon. 30 7
1
R2 = .93
i
20
0
I
--
zi
6
t,
E
10-
EY
0
04 0.4
0.9 PROTONATED ALIPHATIC CARBON (W)
1.4
Figure 7. Toluene solubles activation energy vs protonated aliphatic carbon.
rank coals in general. The activation energy for conversion to toluene solubles was found to be very highly correlated (R2= 93%) with protonated aliphatic carbon (Figure 7). No other correlations of coal properties from the 13CNMR data and toluene or T H F activation energies were found to be statistically significant. The strong relationship found between protonated aliphatic carbon and total oxygen for the six bituminous Argonne coals can be interpreted structurally in terms of etheric groups in the cross-linking structures between aromatic clusters. While based on a rather limited data set, this relationship and the correlation found between total oxygen and toluene activation energy would seem to suggest that the carbon-oxygen bonds in cross-linking structures are of extreme importance in determining coal reactivity. The importance of etheric oxygen and cleavage of ether linkages to coal liquefaction has long been rec~ g n i z e d . ~ & ~Data l from 13C NMR spectroscopy also
(28) Fisher, C.H.;Eisner, A. Znd. Eng. Chem. 1937,29, 1371. (29) Takegami, Y.;Kajiyama, S.; Yokokawa, C. Fuel 1963, 42, 291. (30) Siskin, M.;Aczel, T. Fuel 1983, 62, 1321. (31) Kamiya, Y.;Ogata, E.; Goto, K.; Nomi, T. Fuel 1986, 65, 586.
