Energy & Fuels 1991,5, 179-187
Acknowledgment. This work was supported by the Commonwealth of Kentucky and the US. DOE through Contract No. DE-AC22-84PC70029. Registry No. l-Cb,109-67-1;n-Cs, 109-66-0;l-C6, 592-41-6; n-C6,110-54-3;C-2C6,7688-21-3; 1-C7, 592-76-7;n-C7,142-82-5;
179
l-CB,111-66-0;CO, 630-08-0; Thop, 1314-20-1;Fe, 7439-89-6; ethanol, 64-17-5; l-pentanol, 71-41-0; l-propanol, 71-23-8; 1hexanol, 111-27-3;l-decanol, 112-30-1;ethene, 74-85-1;ethane, 74-84-0;propane, 74-98-6;propene, 115-07-1;i-Butane, 75-28-5; n-butane, 106-97-8;i-butene, 115-11-7;l-butene, 106-98-9;t-2butene, 624-64-6; c-2-butene, 590-18-1.
Interactive Effects between Solvent Components: Possible Chemical Origin of Synergy in Liquefaction and Coprocessing Donald F. McMillen, Ripudaman Malhotra,* and Doris S. Tse Department of Chemical Kinetics] Chemistry Laboratory] SRI International] Menlo Park, California 94025- 3493 Received September 19,1990
The traditional mechanism for coal liquefaction ties liquefaction ability of a solvent with its scavenging ability and cannot explain synergistic effects between solvent components. Such effects are naturally anticipated by mechanisms that involve interaction between components to generate reactive components. This paper discusses three recent papers, in which beneficial effects of adding nondonor species to the donor solvent were reported, in light of the solvent-mediated hydrogenolysis model for coal liquefaction-a model we have previously proposed. The nondonor species, which in these cases are polycyclic aromatic hydrocarbons, are good acceptors of hydrogen and react with hydroaromatic donors to give cyclohexadienyl radicals, which are capable of engendering hydrogenolysis, thereby increasing the overall rates. The PCAH species also help minimize wasteful transfer of hydrogen that would otherwise lead to reduced but uncleaved products. In this case the beneficial effect is reflected in the efficiency of hydrogen utilization. Finally, we extend this line of reasoning to the context of coprocessing coals and heavy oils. We use semiquantitative estimates for the fate of aliphatic radicals in the presence and absence of PCAH components to suggest that the PCAH components provided by coals convert aliphatic radicals, which are ineffective in causing bond scission, into cyclohexadienyl radicals that can do so.
Introduction The question of possible synergistic effects for coal liquefaction has been raised recently in the context of coprocessing and also in that of straight donor-solvent liquefaction. Synergy is generally said to occur when the effect of a combination of components exceeds the sum of the effects of the individual components. The existence of synergy is more difficult to demonstrate in the coprocessing context, because liquefaction of coal alone, that is, in the absence of any other component (e.g., residual oil), is not viable in process terms, and therefore qne of the boundary conditions is not available. For this reason, synergy is often defined as a positive deviation in distillate yield from a linear interpolation between the yields obtained for two processable compositions (e.g., 100%resid and 50%/50% resid/coal). By this or a similar definition, several groups have reported a coal-resid synergy.’” (1) Duddy, J. E.; MacArthur, J. B.; McLean, J. B. Proceedings of Direct Liquefaction Contractors’ Meeting, October 20-22, 2986; Pittsburgh, PA, p 304. (2) Cugini, A. V.; Lett, R. G.; Wender, I. Energy Fuels 1989, 3, 120. (3) Lenz, U.; Wawrzinek, J.; Giehr, A. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1988, 33(1), 27. (4) Fouda, S.A.; Kelly, J. F.; Rahimi, P. M. Energy Fuels 1989,3,154. (5) Muehlenbachs, K.; Steer, J. G.;Hogg, A.; Ohuchi, T.;Beaulieu, G . P r e p . Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1988, 33(1), 122.
0887-0624/91/2505-0179$02.50/0
While it is recognized that one of the benefits of coprocessing is the fact that the unconverted carbon serves as a “getter” for metals in the heavy oil, no chemical explanation has been advanced for synergy among the organic constituents of the coal and heavy oil. The origin of synergy can be more readily assessed in the context of straight liquefaction, where there is always a separate liquefaction medium that can be made up of a number of components. It is instructive to recall that, in straight liquefaction, interactive effects often occur between various solvent components as they act upon the coal; on the other hand, in coprocessing, coal itself is one of the agents. This distinction is arbitrary, because the constituents of coal could just as well interact with solvent components to produce other more effective species for liquefaction. While this paper specifically addresses liquefaction data, we suggest that the conclusions drawn there can be readily extended to coprocessing. Several recent reports of coal liquefaction illustrate the interactive effects of various In each case, ~~~~~
~~~~
~
(6) Caygill, R.; Flynn, T.; Kempe, W.; Steedman, W. Prepn. Pap.Am. Chem. Soc., Diu. Fuel Chem. 1988,33(1), 89. (7) Cassidy, P. J.; Grint, A,;Jackson, W. R.; Larkins, F. P.; Louey, M. B.; Rash, D.; Watkins, I. D. Proceedings of the 1987 International Conference on Coal Science; Elsevier: Amsterdam, 1987; p 223. (8) Mochida, I.; Yufu, A.; Sakanishi, K.; Korai, Y. Fuel 1988,67, 114.
0 1991 American Chemical Society
McMillen et al.
180 Energy & Fuels, Vol. 5, No. 1, 1991 Scheme I H
H
H
H
a beneficial effect was observed upon addition of a nondonor polycyclic aromatic hydrocarbon (PCAH) to the solvent consisting of hydro (or perhydr0)aromatics. In the discussion that follows, we use these results in conjunction with an improved mechanistic model for coal liquefaction to help shed some light on the chemical origin of such beneficial effects. First, it should be noted that the traditional liquefaction mechanism, which ties liquefaction effectiveness to the efficiency with which donor components scavenge fragment radicals formed in the spontaneous thermal scission of the coal structure, cannot easily accommodate interactive effects.'*12 Researchers have therefore been inclined to invoke physical factors such as solvency for the observed synergistic effects.13-14 On the other hand, such effects would be anticipated for mechanisms that involve reaction of one component with another to form an intermediate that then reacts with the coal. More specifically, we will show that the various H-transfer processes that we have hypothesized as leading to bond cleavage by "solventmediated hydrogenolysis" clearly fall in the second category of reaction type. As we previously reported,"*l5most of these H-transfer processes require both a hydrogen "donor" species16 and an "acceptor" species to form the active H-transfer intermediate, a cyclohexadienyl "carrier" radical, which can either give free H atoms or transfer hydrogen directly, to engender cleavage of even strong bonds in the coal structure l7 (Scheme I). In the following paragraphs, we first summarize three recent and striking examples of synergy between the donor and acceptor components of liquefaction solvents, In two (9) Clarke, J. W.; Rantell, T. D.; Snape, C. E. Fuel 1984, 63, 1476.
(IO) It might be argued that positive interactive effects can be explained in terms of the traditional mechanism by invoking the formation of a better scavenger than that provided by any of the separate components. However, it has been already demonstrated that liquefaction effectiveness does not necessarily correlate with scavenging ability."-'* (11) McMillen, D. F.; Malhotra, R.; Hum, G . P.; Chang, S.-J. Energy Fuels 1987, 1, 193. (12) Finseth, D. H.; Bockrath, B. C.; Cillo, D. L.; Illig, E. G.; Sprecher, R. F.; Retcofsky, H. L.; Lett, R. G. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1983,28(5), 17. (13) Shaw, J. M.; Gaikwad, R. P.; Stwoe, D. A. Fuel 1988, 67, 1554. (14) Rincon, J. M.; Angulo, R. Fuel 1986, 65, 889. (15) McMillen, D. F.; Malhotra, R.; Chang, S.-J.;Fleming, R. H.; Ogier, W. C.; Nigenda, S. E. Fuel 1987,66, 1611. (16)In the discussion that follows, we use the word "donor" by itself to mean a hydrocarbon, whether aliphatic, alkyl-aromatic, or hydroaromatic, that can contribute a hydrogen atom to a radical or other acceptor. We make this diatinction because some of these substances, such as fluorene, are good hydrogen atom donors in this Benee but are not good donor solvents for coal liquefaction. (17) Billmers, R.; Brown, R. L.; Stein, S. E. Int. J. Chem. Kinet. 1989, 21. 375.
instances, the interactive effect of the components led to enhanced coal conversion yields. In the third case, which did not lead to enhanced conversion yields, markedly improved selectivity for oils and efficiency of hydrogen utilization were observed. We then discuss, in semiquantitative terms, how a coal-liquefaction picture that includes strong-bond hydrogenolysis mediated by solvent carrier radicals can easily accommodate the liquefaction results. Then, for a "model" hydroaromatic solvent system, we will compare the rate changes predicted by our mechanistic numerical model with the reported liquefaction results. Finally, we extend the mechanistic insight gained in rationalizing the positive interaction among solvent components in coal liquefaction to account for the interactive effects reported for coprocessing. The essential feature of the explanation is that PCAH moieties provided by coals serve to convert aliphatic radicals, which are ineffective in engendering scission of strong bonds, into cyclohexadienyl radicals that can do so.
Experimental Section 1,2'-Dinaphthylmethane was obtained from API Standard Reference Materials Bank. GC analysis showed the sample to be >99% pure and was used without further purification. Anthracene, dihydroanthracene, and other chemicals were purchased from Aldrich Chemical Co. Experiments to determine the efficiency of hydrogen utilization for the cleavage of 1,2'-dinaphthylmethane were performed at 400 OC in sealed, fused-silica ampules enclosed in stainless-steel pressure jackets and heated in a molten-salt bath controlled to within 1 OC. Typically, about 30 mg of the substrate was mixed with 9 times its weight of a mixture of anthracene and dihydroanthracene. Biphenyl was sometimes used as an inert diluent in amounts ranging up to 80 mol % of the reaction mixture. The products were analyzed by GC using an internal standard for quantitation. When necessary, GC-MS was used to identify the components.
Examples of Reported Synergy in Liquefaction 1. Enhanced Conversion in Hydroaromatic Solvents by Addition of Nondonor Aromatic Components. Cassidy and co-workers have recently published results on the liquefaction of an Illinois No. 6 coal in mixtures of decalin (50%), tetralin (50-25%), and various aromatics (O-25%) in a hot-charged, time-sampled autoclave.' The results show substantial increases in oil yield resulting from the replacement of half of the donor (tetralin) in the solvent with various nondonors (aromatics) even at very short reaction times. Of the three different PCAH added to the solvent, pyrene was clearly the most effective additive, increasing the oil yields by some 30% (of daf coal) at very short as well as longer reaction times. Anthracene and phenanthrene are somewhat less effective, in that order. These changes are remarkable, particularly since the PCAH replaced half of the tetralin, such that the donor content was actually lowered from 50% to 25%. The authors recognized the dramatic improvement in liquefaction yield with reduced donor content and evidently took steps to assure themselves that the results were reproducible. These results are parallel to, but more striking than, earlier results of Derbyshire et al., who reported that conversion of an Illinois No. 6 coal (to THF-solubles) in 70% pyrene/30% tetralin was better than conversion in pure tetralin, at two different hydrogen pressures.'* While we intend to focus on possible chemical explanations, we cannot rule out purely physical factors such as solvent potency and physical state; the possible im(18)Derbyshire, F. J.; Varghese, p.; Whitehurst, D. D. Fuel 1982,61, 859.
Interactive Effects between Solvent Components Table I. Effect of Replacement of Part of Tetrahydrofluoranthene (H Donor) by Fluoranthene (Nondonor) on the Efficiency of H Utilization and Selectivitv to Oils"
product yieldb 100% 75%
THFL content ~
gas
~~
oil asphaltene preasphaltene + residue oil yield/THFL consumpn O
18 63 13 6 1.1
13 68 14
Energy & Fuels, Vol. 5, No.
r,
1991 181
100 Perhydrophenanthrene
8
'
Perhydropyrene
80
m
60
W
z
s
0
40
PERHYDRO PUREPCAH MIXTURE
20
s 0
5
1.7
a Data from ref 8. Optimal yield for coal conversion at 450 C/ N,/solvent:coal 3:1.
portance of such physical factors is discussed in a later section of this paper, in conjunction with the presentation of our chemical rationalizations. (In the work of Cassidy et al.? one question has to do with whether certain solvent components are critical or subcritical at the temperatures used. The 425 "C reaction temperature used in that work is -10 "C above the critical temperature of decalin and 20 and 30 "C below the critical temperature of tetralin and naphthalene, respectively, so the effective reduced temperatures and physical states of various mixtures of these three components and pyrene are difficult to specify. However, we note that -2 min were required for the reactor to return to 400 "C after charging. Since improvement was seen even for samples drawn at 1,2, and 3 min? the "interaction" took place well below the critical temperature of the lowest boiling component.) 2. Improved Efficiency of Hydrogen Utilization upon Addition of Nondonor PCAH to Hydroaromatic Solvent. Mochida et al. recently described the liquefaction of an Australian brown coal in tetrahydrofluoranthene (THFL), a donor solvent, containing varying amounts of fluoranthene (FL), a nondonor.8 In this case, the physical properties of the various solvent mixtures are more nearly the same, and specific chemical changes are seen to result from the changes in solvent composition. Therefore, the observed differences in performance can more readily be attributed to chemical factors. Mochida et al. show that, although the conversion rates decrease slightly as FL is added to the reaction mixture, the selectivity to oils (vs gases) goes up, and the efficiency of H utilization for oil production increases substantially. Table I shows the yields reported by Mochida et al. for conversion in two media: 100% THFL and a mixture of 75% THFL and 25% FL. In each case, the values shown are those reported at the time of maximum oil yield. Replacement of 25% of the donor THFL with an equal amount of the nondonor FL in the starting solvent increases the oil yield by five percentage points at the "expense" of the gas yield. Substantially more pronounced is the impact of FL addition on the H-utilization efficiency (the ratio of oil produced to THFL consumed). As Table I shows, this value increases by 60%. 3. Enhanced Conversion in Perhydroaromatic Solvents by the Addition of PCAH. The results of Clarke et al. provide the most dramatic illustration of the importance of H-acceptor solvent component^.^ These workers report that, whereas conversion (to quinoline solubles) of an 84% carbon coal in various perhydro-PCAH was quite poor, it improved slightly when naphthalene was added and markedly when a PCAH such as phenanthrene or pyrene was added. These results are depicted in Figure 1. The conversion levels achieved with the 3- or 4-ring PCAH (good acceptors) are almost as high as that achieved with octahydrophenanthrene, which is known to be an excellent donor solvent.. The authors also note that only
ADDITIVE
Figure 1. Impact of added PCAH on coal liquefaction in perhydroaromatics.
in the presence of the 3- 01'Q-rhg PCAH is there any significant dehydrogenation of the perhydroaromatics, but that in the absence of the PCAH the perhydroaromatics were observed to undergo cis-trans isomerization. Thus, bridgehead radicals are formed by reaction with the coal, but in the absence of PCAH, these radicals merely reabstract hydrogens from the coal.
Rationalization in Terms of Solvent-Mediated Hy drogenolysis The above results are entirely consistent with the mechanistic picture of coal liquefaction that includes hydrogenolysis of strong bonds engendered by the solvent system by transfer of H atoms to ipso positions on aromatic clusters within the coal structure that bear linkages to other c l ~ s t e r s The . ~ ~chemistry ~~~ responsible for increased conversions and improved efficiency of hydrogen utilization is essentially the same. When the base case happens to be only a poor to modest conversion system (such as perhydrophenanthrene or 50% tetralin in decalin), addition of PCAH substantially increases yields. However, if the base case is a good liquefaction system, like tetrahydrofluoranthrene, the improvement is primarily reflected in the efficiency with which the system uses hydrogen to produce oils. Effect of Competing Hydrogen-Transfer Processes on Efficiency of H-Utilization in a Model System. To help understand the origin of this change in efficiency, we will review the modes of H transfeer that lead to hydrogenolysis and hydrogenation (Scheme I). We will see that consideration of the competing H-transfer pathways helps explain the changes in H-utilization efficiency. Specifically, the competition between addition of free H atoms to positions on aromatic clusters bearing linkages (reaction 1 in Scheme I) with some other reaction, such as the recently e l ~ c i d a t e d ~radical ~ * ~ ~hydrogen-transfer *~~*~ process (reaction 2 in Scheme I), strongly influences this efficiency. In previous mechanistic work directed at understanding the modes of bond scission available for relevant coal structures, we observed experimentally that the selectivity and efficiency with which donatable hydrogen is used to cleave aryl-alkyl linkages in coal surrogates is highly dependent on the degree of hydrogenation of the donor.11915*19*20 Data on the cleavage of 1,2'-dinaphthylmethane (DNM) in anthracene/dihydroanthracenemixtures illustrate this point. When DNM was heated at 400 "C in these mixtures, the principal cleavage products, naphthalene and methylnaphthalenes, were accompanied with varying amounts of reduced products like tetralin, and reduced but uncleaved (19) McMillen, D. F.; Malhotra,R.;Chang, S.-J.;Nigenda, S . E. Prepr. Pap.-Am. Chem. SOC.,Diu. Fuel Chem. 1985,30(4), 297. (20) Malhotra, R.; McMillen, D. F. Energy Fuels 1990, 4, 184.
McMillen et al.
182 Energy & Fuels, Vol. 5, No. 1, 1991
-e+&&
but no depolymerization
1.2
.
1.o
0.8 0.6
Experimental - Computed
0.4 0.2
0.0 0
2
4
6
8
10
12
ANTHRACENEDIHYDROANTHRACENE
Figure 2. Experimental and computed cleavage efficiency as a function of aromatic/dihydroaromatic ratio. starting material, namely several isomers of tetrahydroDNM. The latter products were identified by GC-MS. The efficiency of hydrogen utilization, which for this model system we can define as the fractional amount of hydrogen used for cleavage over the total consumption for cleavage and reduction, is shown in Figure 2 to increase dramatically from about 10-25’70 in mixtures with [An]/[AnH2]ratios less than 0.4 to nearly 100%in mixtures with [An]/[AnHz] ratios greater than 2.0. Thus, solvents rich in PCAH utilize hydrogen efficiently for hydrogenolysis of linkages. Conversely, solvents that are very low in the fully aromatic PCAH component tend to transfer hydrogen faster but less discriminately, such that a much smaller fraction of the hydrogen transferred goes to produce hydrogenolysis of linkages, and more goes to simple ring hydrogenation and H2formation. Some of the data shown in Figure 2 are listed in Table I1 where cleavage rates, reaction time, and extent of dilution with an inert additive are also shown. For a given initial level of dihydroanthracene, replacement of the inert diluent with anthracene results in either no change or an increase in the cleavage rate. Data obtained with and without the diluent for the longer reaction times give values close to the single line shown in Figure 2. Data for shorter reaction times exhibit the same trend, but with different absolute values of cleavage efficiency for a given [An]/AnH,] ratio. We have recently reported on a mechanistic numerical model incorporating the various competing H-transfer processes.m The line shown in Figure 2 was computed by using that model; Figure 2 illustrates that the experimentally observed increase in H-utilization efficiency for the cleavage of dinaphthylmethane in anthraceneldihydroanthracene is well reproduced by the model. The model also helped trace the origin of this shift in efficiency to shifts in competition between the various H-transfer processes. In contrast to free H atoms, which readily ab-
Table 11. Rates and Efficiencies for the Cleavage of 13-Dinaphthylmethane in Anthracene-Dihydroanthracene Mixtures at 400 “C [An/AnHzlO, biphenyl, reaction [An/ kc,, b mol % mol 5% time, h AnH2],va cleav efT ~. s-l X 10s 0.74 0 4 0.8 5.6 20170 0 4 1.1 5.4 30160 0.69 0.74 0 4 2.2 4.2 50140 0.50 0 0.5 3.0 5.5 60130 0.93 0 6 5.2 3.3 70/20 0 6 13.4 1.4 1.0 80110 0.093 50 0.08 0.02 6.3 0/40 0.15 50 0.33 0.07 5.9 0/40 0.24 50 1 0.20 9.5 0/40 0.27 50 2.5 0.32 4.1 0/40 0/40 0.30 50 4.0 0.44 3.5 0.17 60 0.5 0.09 6.0 0130 75 7.33 0.51 1.2 0.28 0115 The geometric average of the starting and ending An/AnH2 ratios. The defined first-order rate constant for dinaphthylmethane cleavage by direct ipso displacement, that is, corrected for the 0-3070 of cleavage typically resulting from reduction followed by cleavage. Cleavage efficiency is given by [ZMeNap]/[ZMeNap + P(DNM-HJ + 2(tetralin)].
stract hydrogen from aliphatic structures producing H2, cyclohexadienyl radicals do not readily abstract, and reversion of H-transfer activity to H2 is minimized. Second, when RHT is the dominant transfer process, the recovery of “wastefully”transferred hydrogen is facilitated, as shown by the following discussion. The manner of formation of the ipso-substituted cyclohexadienyl radical (I) itself has no direct bearing on efficiency, because this intermediate, once formed at coal conversion temperatures, has an extremely short lifetime before elimination of virtually any linkage, except a biaryl (Scheme 11). In other words, cleavage will result no matter how I was formed. Efficiency is determined by what happens to all the radical species produced by H transfer to positions not bearing linkages (11). When the concentration (and effectiveness) of polycyclic aromatic hydrocarbons that can act as hydrogen acceptors is low, then the non-ipso radicals are more likely to abstract a second hydrogen (e.g., from hydroaromatic in an “overhydrogenated” solvent), This dihydro intermediate is very reactive and will quickly be reduced to a tetrahydro intermediate. The tetrahydro intermediate can then receive additional hydrogens to open the aliphatic ring and crack off the 4-carbon chain as light hydrocarbons. In the example shown, six or more hydrogens would be consumed and no linkages broken. In contrast, when the concentration of H acceptors is sufficient, the otherwise wastefully transferred hydrogen is recovered from the non-ipso radicals by the bimolecular RHT process, regenerating the hydrogen carrier radicals ArH*, so that the hydrogen-transfer activity is maintained and there is another chance at transfer to an ipso position.
Energy & Fuels, VoZ. 5, No. 1, 1991 183
Interactive Effects between Solvent Components
I 3 w
I
Gas
80 60
1
3
40
I 3 0
20
0 0
10
20
30
40
0
10
20
30
REACTIONTIME (min)
REACTIONTIME (min)
(a)
(b)
40
Figure 3. Product yields from conversion of an Australian Brown Coal in (a) 100% tetrahydrofluoranthene, and (b) a 75/25 mixture of btrahydrofluoranthene and fluoranthene.
In the limit, the only irreversible H transfer would be that to positions bearing linkages. In fact, the observed and computed hydrogen utilization efficienciesshown in Figure 2 indicate that, at least in the anthracene/dihydroanthracene system, a utilization efficiency approaching 9&100% can be achieved by merely adjusting the degree of hydrogenation of the solvent. Besides the concentration of the aromatic acceptors, the two main factors that control the effectiveness with which wastefully transferred hydrogen atoms are retrieved are (1)the nature of the PCAH system (specifically, H transfer to three- and four-ring PCAH is more favorable than that to naphthalene and benzene systems), and (2) the temperature. The absence of good acceptors and/or higher temperatures make the unimolecular loss of H atom from ArH. species more facile.20 Finally, the increase in H-utilization efficiency is achieved at some expense in reaction rate. In the case of the data in Figure 2 and Table I1 showing cleavage efficiency in the anthracene solvent system, when the experimental efficiency increased from 70% to 90%, the cleavage rate declined by about 40%. (Increasing aromatic concentration can result in either a decrease or an increase in cleavage rate, depending on the nature of the solvent system and the substrate.) In the case of Mochida's data for a real coal liquefaction system, as discussed below, substantial increases in efficiency were achieved in return for minor decreases in conversion rates. Impact of Partial Replacement of Donor with Acceptor in Liquefaction. Figure 3, constructed from the data of Mochida,8shows the yields of gas, oil, asphaltene, preasphaltene, and residue as a function of time when (a) the starting solvent contained no fluoranthene (FL) and (b) the starting solvent contained 25% fluoranthene. In the 0% FL case, the oil yield develops and the residue disappears more quickly, but in the 25% FL case, the more slowly developing oil yield reaches an 8% higher value at its maximum (at 20 min rather than at 10 min), and the gas yield is only 67% of that produced when the starting solvent is 100% THFL. The moderate advantage that results from starting with 25% FL becomes quite marked when the ratio of oil yield to THFL consumption is computed, revealing, as shown above in Table I, a 60% increase in H-utilization efficiency. The evolution of the solvent composition for these same two cases also shows the curious trend that equal or greater THFL remains when the starting solvent contained less hydroaromatic. For example, at 20 min the THFL concentration in recovered solvent is 30% when starting with pure THFL and 35% when starting with only 75% THFL.* The extent of crosBover is small and could easily be
assigned to experimental scatter. However, we have also observed a similar trend in our model compound studies. When the original model compound experiments were performed several years ago, we were surprised to observe that not only did the rate of hydroaromatic consumption go down when the solvent initially contained a significant amount of the respective aromatic but the absolute amount of hydroaromatic remaining at various reaction times could actually be larger when starting with less.21 In Mochida's case, the faster THFL decline is associated with a greater rate of production of perhydrofluoranthenes (defined by Mochida as including decahydrofluoranthene). When the initial FL content is zero, the perhydrofluoranthene content at 10,20, and 30 min was 1.5, 1.7, and 2.2 times that observed when the solvent had an initial 25% FL content. Thus, in Mochida's case, as in our earlier model system studies, the absence of a significant initial concentration of PCAH allows multiple H transfers and even full reduction of some hydroaromatic species; one can infer that similar reductions without cleavage, followed by ring opening and gas production, are also occurring within the coal structures themselves, causing the reported higher gas yields. We used a similar explanation22for excessive gas production when high temperature excursions occurred during coal hydropyrolysis, as reported by Gorbaty and Maa.23 In that case, we reasoned that the effectiveness of the PCAH systems as H acceptors decreased as a result of increasing temperature, thereby allowing greater ring reduction and ring opening. We find it gratifying to note that, in the present case, careful analysis by Mochida8 of the recovered liquefaction solvents provides direct evidence for the anticipated wasteful reduction of aromatic systems caused by overhydrogenated solvents. The trend in H-utilization efficiency noted here does not appear to be merely a fortuitous result of scattered behavior. For instance, in an earlier publication, Mochida et al. described in less detail the liquefaction of three subbituminous coals in the fluoranthene solvent system." For two of the coals, a decrease in the starting THFL content from 100% to 67% decreased the oil-plus-asphaltenes yield by only 3% . Although details of THFL consumption were not given, we presume that, since the decreases in THFL consumption found in the more recent studies on brown coals were generally on the order of (21) McMillen, D. F. Unpublished work. (22) McMillen, D. F.; Malhotra, R.; Nigenda, S. E. Fuel 1989,68,380. (23) Gorbaty, M. L.; Maa, P. S. Prepr. Pap.-Am. Chem. SOC.,Diu. Fuel Chem. 1986,31(4), 5. (24) Mochida, I.; Kishino, M.; Sakanishi,K.; Korai, Y.; Takahashi, R. Energy Fuels 1987, 1 , 343.
184 Energy & Fuels, Vol. 5, No. 1, 1991
McMillen et al.
3.0 b
i
0
BASECASE +ADDITIVE
RHT
H-ATOM RRD REACTION TYPE
RHT’
TOTAL
Figure 4. Computed impact of partial replacement of donor with a good acceptor at 400 “C.Substrate: 0.3 M DNM in the following systems. Base case: dihydrophenanthrene 2.6 M and phenanthrene 0.2 M. Additive case: dihydrophenanthrene 2.45 M, phenanthrene 0.2 M, and anthracene 0.15 M.
40-45%, THFL consumption decreases are probably also much more than 3% in the case of subbituminous coals. Thus, a trend in H-utilization efficiency similar to that observed for the brown coal probably also occurred for the subbituminous coals. Other examples undoubtedly exist, despite the fact that concern over conversion rate has tended to overshadow consideration of H-utilization efficiency. For example, Derbyshire et al. earlier reported18 increased conversions in tetralin/pyrene mixtures as compared to pure tetralin. Although they did not report the consumptions of hydroaromatic, consumption very likely decreased upon addition of pyrene. Similarly, the results of Cassidy et al.7 showing a substantial increase in actual oil production rates upon partial replacement of tetralin with pyrene probably also reflect increased H-utilization efficiency. Numerical Modeling of the Impact of Added Nondonor. As mentioned above, the impact of PCAH on the rate of hydrogenolysis is more variable than the impact on efficiency and can lead to either higher or lower rates depending on the nature of the hydroaromatic. Such a range of possibilities has been demonstrated more rigorously with the help of the numerical model for a case similar to that of Cassidy et al.7 For modeling purposes, we used the cleavage of dinaphthylmethane as a surrogate for those structures in coal that cannot cleave by simple thermolysis under the reaction conditions and whose cleavage therefore has to be mediated by the solvent. Using the mechanistic numerical model described elsewhere,20 we computed the hydrogen-transfer-induced cleavage of dinaphtylmethane in a system very rich in the donor hydroaromatics and also for cases in which a portion of the hydroaromatic is replaced by a nondonor species such as pyrene or anthracene. Figure 4 shows the computed rates of cleavage resulting from H transfer (to the ipso position on a naphthalene-X structure) by RHT and free H additions, as well as the total cleavage rate for a 13:l dihydr0phenanthrene:phenanthrene mixture. Also shown (dark bars) are the rates computed for the case where 7% of the dihydrophenanthrene has been replaced by anthracene. The replacement results in a contribution from a step labeled RHT’, which is H transfer from the anthracene-derived carrier radical. Partial replacement of dihydrophenanthrene with anthracene also results in a substantial increase in the concentration of (and therefore transfer from) the hydrophenanthryl radical and in a smaller percentage increase in the free H contribution, such that the overall computed increase in cleavage rate is 80%.
Examination of the various reactions producing and consuming the hydrophenanthryl radical shows that the increase is mainly due to the rapid formation of AnH’, owing to the very good H-acceptor nature of anthracene. The increased AnH’ concentration then results in an increased production of PhenH’ through the metathesis reaction of AnH’ with PhenHa. After 1h of reaction, most of added anthracene has been converted to dihydroanthracene and the computed H transfer by RRD direct M/s) surpasses that from the from this species (8 X higher concentration of dihydrophenanthrene (5 X lo* M/s). However, both rates are still low compared to RHT and free H atoms coming from the hydrophenanthryl radical. In general, the increase in computed cleavage rate is larger or smaller as the system is poorer or richer, respectively, in “native” acceptor (e.g., phenanthrene). In other words, systems that are overhydrogenated, or poorest in acceptors, appear to benefit most from the addition of a good acceptor. The computed increase in PhenH’ production rate is in part analogous to the experimentally measured reduction of anthracene by dihydrophenanthrene, which was reported by Stein and co-workers to be catalyzed by the addition of small amounts of dihydr~anthracene.’~ In that case, as well as the one modeled here, the addition generates a large pool of AnH’ radicals, which in turn produces additional PhenH’ by abstraction of H from PhenH2. In the case described by Stein, the increase in the pool of AnH’ radicals arises because of the very good H-donor quality of the added AnH2. In the present case, the pool increases because of the very good acceptor quality of the added anthracene. “Activation”of Perhydroaromatics by PCAH. The data of Clarke et al.9 on the conversion of coals in perhydroaromatics provides a natural bridge from the liquefaction data discussed above to the coprocessing context. The perhydroaromatics used by Clarke are representative, we suggest, of both the paraffinic and naphthenic alkyl systems in the resid. His results can be rationalized as follows. Any thermally produced coal radicals can, at a modest rate, abstract aliphatic hydrogens from perhydrophenanthrene or other perhydroaromatics. The perhydrophenanthryl radical can undergo the following four reactions: (1)H abstraction from coal-H to regenerate (an isomerized) perhydrophenanthreneand coal; (2) @-scission of a C-C bond; (3) @-scissionof a C-H bond giving a free H atom; and (4) bimolecular transfer of @-hydrogento a PCAH acceptor by RHT. These reactions are shown in Scheme I11 along with the estimated first-order or pseudo-first-order rate constants. The literature basis for the estimated rate constants is given in the Appendix. The precision of these estimates is such that those rates within an order of magnitude of each other may be considered similar, in contrast to those that are several orders of magnitude apart. Assuming an effective concentration of abstractable hydrogens in coal to be ca. 1M, H abstraction from coal-H is the main reaction in the absence of any PCAH molecules. The planar nature of the radical will, of course, result in cis-trans isomerization of the perhydrophenanthrene. @-Scissionof a C-C bond to produce an olefin and another alkyl radical is estimated to be similar to the rate of recapture. However, in perhydroaromatic systems, where the olefin and the new alkyl radical are connected, @-scissionis expected to be substantially reversed, and the observed C-C scission products will be substantially less than those for isomerization. @-Scission
Energy & Fuels, Vol. 5, NO. 1, 1991 185
Interactive Effects between Solvent Components Scheme 111
08 of a C-H bond leading to free H atoms, which could engage in further bond cleavages, is estimated to be about 3 orders of magnitude too slow to compete effectively with either the C-C cleavage or the H abstraction. Thus, the observation of little conversion and little change in the solvent except for cis-trans isomerization is readily understood. On the other hand, when PCAH (in the present case 1 M pyrene) is added to the system, transfer of a hydrogen to a PCAH molecule from the cycloalkyl radicals is estimated to be roughly comparable to the hydrogen recapture (reaction 11). This transfer of a hydrogen to a PCAH molecule produces a cyclic olefin and a cyclohexadienyl radical. These cyclohexadienyl radicals can result in hydrogenolysis of other bonds in the coal structure by RHT (reaction 2). 11~16~17*26 Furthermore, the cyclic olefin has much weakened allylic C-H bonds and the removal of one of these hydrogens ultimately leads to the formation of hydroaromatic structures, making still more hydrogen atoms available for bond cleavage. Thus, the PCAH molecules can channel a number of the hydrogens available in the perhydroaromatic molecules into useful cleavage reactions. Physical Solvency Factors. By focusing attention on the chemical factors that may contribute to interactive effects in coprocessing, we do not wish to exclude beneficial effects of more favorable physical interactionswith solvent components containing PCAH. Phase separations are well-known to occur between paraffinic solvents and coal liquefaction intermediates. Factors affecting these phase separations and their potential impact on coal liquefaction have been recently addressed in the studies of Shaw and co-workers.13 In actual fact, both chemical and physical factors are presumably important: the necessary chemical changes cannot take place unless the potential reactants can access each other, and no degree of mutual solubility will suffice if the chemical reactions are inherently unfavorable. In most of the cases we have examined, the original authors have provided analyses that show specific chemical changes in the solvent that are associated with the observed coal conversions. In the data of Clarke et al., for example, there is no doubt that the opening of a hydrogen-transfer channel was made possible by the ability of the PCAH to accept hydrogen, a chemical ability that is not possessed at all by alkanes. Thus, while physical factors should not be excluded, it is clearly important to consider chemical factors. (25) Bockrath, B. C.; Schroeder,K. T.;Smith, M. R. Prepr. Pap.-Am. Chem. Soc., DIU.Fuel Chem. 1988, 33(3), 325.
Scheme IV
(&
(&
canengender hydrogenolysis
Key Hydrogen-Transfer Reactions in Coprocessing In coprocessing of coals and heavy oils, the heavy oils have a relatively large amount of aliphatic hydrogen (naphthenic or paraffinic) that is potentially useful for cleavage of coal structures. The beneficial effects of adding anthracene oil or other highly aromatic product streams to either coal liquefaction or coprocessing mixtures are ~ e l l - k n o w n . *We ~ ~suggest ~ ~ ~ ~that, in coprocessing, the coal (or anthracene oil) provides PCAH moieties, which can interact with the aliphatic hydrocarbons in a way similar to the one we have described for the case of coal conversion in PCAH and perhydro-PCAH, as reported by Clarke et al.9 We have discussed the latter results because the authors' analysis of the recovered solvent greatly aids chemical interpretation. Reactions of paraffinic radicals, shown in Scheme IV, are analogous to those of the naphthenic radicals. The rate estimates suggest that H transfer from either naphthenic or paraffinic radicals to PCAH should proceed at similar rates. However, two noteworthy differences occur. First, unlike the paraffinic systems, the olefins generated in the naphthenic systems can, by further rapid dehydrogenation, lead to hydroaromatic structures, which are good liquefaction agents. Thus, for every hydrogen initially trans(26) McLean, J. B.; Duddy, J. E. Prepr. Pap.- Am. Chem. SOC.,Diu. Fuel Chem. 1986,31(4), 169. (27) Derbyshire, F. J.; Odoerfer, G. A.; Varghese, P.; Whitehurst, D. D. Fuel 1982,61, 899. (28) Wallace, S.; Bartle, K. D.; Burke, M. P.; Egia, B.; Lu, S.; Taylor, N.; Flynn,T.; Kemp, W.; Steedman, W. Fuel, 1989,68,961.
McMillen et al.
186 Energy & Fuels, Vol. 5, No. 1, 1991
reaction no. -8 9
lo
l1 9'
Table 111. Summary of Rate Constant Estimates est AHo: E&),' analogous reaction" kcal/mol kcal/mol x , kcal/mol t-Bu' + PhCH2-H t-Bu-H + PhCHi -9 9.4 -17.1 +22 34 26
0- .3
-C
X>-).O+H' + py .
11"
py
H
+ pv
11.4 30
log A' 8.5 14.3
log k& 5.0 5.0
38.6
13.8
1.8
-2.3
--L
-c
16
0
0.35
15.2
8.6
3.9
+ 1PyH'
M--+*\
11'
E,(AHo)
+36.6
x> ->13 +
od
0.25
/\\/\
+ 1PyH'
+ 1PyH'
+23
33.9
26.2
0.17
31.2
14.3
4.6
-4.2
16
0
0.35
14.5
8.6
4.1
-4.7
16
0
0.35
14.5
8.6
4.1
8
-26
8
9.6
7.1
-22
"actions that have easily estimatable AH" values and are structurally similar to the indicated reactions in Schemes I and 11. *Enthalpy changes are estimated at 300 K; AHo values at 430 O C (703 K)will typically differ by less than 1 kcal/mol from the room temperature value. The biggest variation will be for reactions involving a mole change (such as reaction 5). Even then, use of the 703 K AH value would mean a rate constant lower by less than a factor of 2 than that indicated by the room temperature value. CActivationenergy measured for a reaction similar to the respective analogue reaction and having an enthalpy change (at 300 K)equal to x . a = AEt/A(AHo). e log (A/&) and log (AIM-' s-l), estimated from similar unimolecular and bimolecular reactions in refs 30, 32, and 33. flog ( k / s - ), Le., for firstorder or pseudo-first-order reactions, based upon an assumed 1 M concentration of reacting partners (Le., pyrene or benzylic coal hydrogen as modeled by PhCH,-H).
ferred from the perhydroaromatic radical to the PCAH, several more hydrogenolytically active hydrogens will be produced in the form of a hydroaromatic. The second major difference between alkane types is that &scission of the paraffinic radicals is not reversed as much as that of the naphthenic radicals. Therefore, the paraffinic radicals are somewhat more difficult to divert from their normal @-scissionalkane-pyrolysis pathways, and naphthenic resids might by more effective in coprocessing. Support for our suggestion that the utilization of paraffinic hydrogen for hydrogenolysis occurs readily when PCAH are available can be found in the differing product distributions obtained upon the pyrolysis of alkylbenzenes and alkylpyrenes. These systems contain no naphthenic or hydroaromatic hydrogens. Pyrolysis of hexadecylbenzene results in the predominant formation of toluene and CI5and lower alkanes and alkenes.B The result is consistent with the cleavage at the benzylic position. In contrast, Javanmardian et a1.30 recently reported that pyrolysis of neat dodecylpyrene at 375-425 OC results primarily in the scission not of the weakest bond, but of the 100 kcal/mol aryl-alkyl linkage. The better H-acceptor nature of pyrene makes possible the transfer of a hydrogen to the ipso position, resulting in the cleavage of the aryl-C1 linkage. This situation is analogous to that depicted in Scheme IV, except that the 1-position of pyrene now bears an alkyl chain, which is displaced directly as a result of the H transfer. In the pyrolysis of dodecylpyrene, the 1-position of pyrene has clearly "accepted" a hydrogen; whether it has come directly from an unconnected aliphatic radical, from some developing char radical, or from a benzylic position in an intact dodecylpyrene (i.e., by reverse radical-disproportionation) is not clear.31 Transfer of hydrogen to
-
(29) Muahrush, G. W.; Hazlett, R. N. Ind. Eng. Chem. Fundam. 1984,
23, 288.
(30)Javanmardian, M.; Smith, P. J.; Savage, P. E. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1988,33(3), 325. (31) The ultimate source of the hydrogen is not certain; the mass balances for the pyrene fragment are good, ranging from 80% to 97%, but those for the alkyl portion are lower.
the non-ipso position would not result in immediate cleavage; however, the cyclohexadienyl radical thus formed is capable of engendering hydrogenolysis, which the aliphatic radicals are unlikely to do. Thus, the presence of PCAH facilitates the formation of H-carrier radicals and the hydrogenolysis of linkages to aromatic systems by making hydrogen available also from paraffinic, as from naphthenic, radicals. Further support for our suggested basis for synergy in coprocessing is provided by recent studies by Bockrath et a1.,% who have used decalin isomerization and methylnaphthalene demethylation as probes for H abstraction and hydrogenolysis in n-CB alkane with added coals and resids to simulate coprocessing conditions. They find that for both probes the combined effect of coal and resid is greater than the sum of the effects of coal and oil acting individually. They also report that the olefins formed from the Czshydrocarbon are predominantly On the basis of Scheme IV, we expect the ratio of terminal to internal olefins to be 1 1 O : l . Within the accuracy of the estimations, the two results are consistent. The chemistry discussed in this paper clearly doea occur in model systems. We believe that the evidence cited here is convincing that it also happens in liquefaction and therfore presumably in coprocessing. Whether it is responsible for the moderate amounts of synergy observed, and if so, whether this synergy in the thermal reaction network is of technological significance (given the heavy catalyst loadings in use in coprocessing development) is less clear. Since some coprocessing approaches currently use catalyst levels resulting in projected catalyst recovery costs that make up a substantial part of the operating it seems worth investigating whether thgse costs might be reduced by the ability to better manipulate the hydrogen-transfer reactions. In the context of straight coal liquefaction, we recognize that the increases in efficiency discussed here may not necessarily translate into advantages in a practical system. (32) Bockrath, B. C. Private communication. (33) Gataie, J. G. Private communication.
Interactive Effects between Solvent Components
For example, the somewhat longer time observed by Mochidas to be required for disappearance of the THF-insoluble residue in the 75% THFL case [Figure 41 could have a detrimental impact: under large-scale continuous processing conditions, persistence of an insoluble phase in the preheater could exacerbate coking on heat-transfer surfaces. Furthermore, if producing mid-boiling-range alkanes is of interest, then ultimately a given amount of hydrogen must be added to the coal structures. On the other hand, if the chosen coal liquefaction approach involves getting the coal as conveniently and cheaply as possible to a liquid product that is suitable for existing petroleum processing technology, then H-utilizaton efficiency could be important. In any case, minimizing the catalyst-intensive use of high-pressure hydrogen (and recycle streams derived therefrom) to convert the coal into C1 to C4 hydrocarbon gases is clearly desirable.
Summary Analysis of synergistic effects of solvent components in coal liquefaction studies indicates that the key chemical features of coprocessing component interaction are (1)coal radicals generate aliphatic radicals from the resid; and (2) aliphatic radicals can undergo 0-scission of C-C bonds to convert the aliphatic portion of the resid or can transfer a hydrogen to the PCAH to form carrier species capable of engendering hydrogenolysis in either the coal or the resid. These reactions allow hydrogen in the aliphatic resid components, which are known to be poor liquefaction solvent components, to be made available for coal and resid conversion. Acknowledgment. We gratefully acknowledge the support of the U.S. Department of Energy (PETC), Contract No. DE-FG22-86PC90908. Appendix The basis for the estimated rate constants cited in this paper is outlined in Table 111. These estimates cannot be expected to be highly accurate, although in most cases the estimated log k430 values should be within half a log
Energy & Fuels, Vol. 5, No. 1, 1991 187
unit of the true value. The value of such estimates generally lies in the ability to project, for instance, whether a reaction will be competitive with another (say within an order of magnitude) or whether one category of reactions will be several orders of magnitude faster than another. The estimates for each of the reactions in Schemes I11 and IV are based on the enthalpy changes for each of the simple analogous reactions shown in Table 111. These reactions were chosen to facilitate the use of measured entropies and heats of f ~ r m a t i o n ,or ~ ~to- facilitate ~~ the estimation3' of these values when no experimental data were available. Kinetic parameters were derived3'?%from the thermochemical values combined with measured Arrhenius parameters for related reactions taken from literature compilation^.^^^ E&) are measured activation energies for reactions of enthalpy change x . The a term is the Evans-Polanyi factor describing the ratio Ma/ A(LVI. It is used to adjust the E, measured at AH = x to the AH estimated for the analogous reactions. The values of a used here are the same as those used in our mechanistic numerical model of cleavage resulting from hydrogen transfer during coal liq~efaction.~~ They reflect the fact that, for H-atom-transfer reactions, values of (Y range from -0.4 for transfers that are near thermoneutral to values that are near -0.1 for reactions that are 40 kcal/mol exothermic or (34) Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermodynamics of Organic Compounds; Chapman and Hall. New York, 1986. (35) Stull,D. R.; Westrum, E. F. Jr. The Chemical Thermodynamics of Organic Compounds; Wiley: New York, 1969. (36) McMillen, D. F.; Golden, D. M. Hydrocarbon Bond Dissociation Energies Annu. Rev. Phys. Chem. 1982,33,493. (37) Benson, S. W. Thermochemical Kinetics, 2nd e& Wiley: New York, 1976. (38) Stein, S. E. A Fundamental Chemical Kinetics Approach to Coal Conversion. In Aduances in Chemistry Series; American Chemical Society: Washington, DC, 1981; No. 169, p 97. (39) Benson, S. W.; ONeal, H. E. "Kinetic Data on Cas-Phase Unimolecular Reactions", National Standard References Data Series, National Bureau of Standards, NSRDS-NBS 21, February 1970. (40) Kerr, J. A.; Moss, S. J. CRC Handbook of Bimolecular and Termolecular Cas Reactions; CRC Press: Boca Raton, FL, 1981; Vol. 1. (41) Herndon, W. C. J. Org. Chem. 1981,46, 2119. (42) Agmon, N. Int. J. Chem. Kinet. 1981,13, 333.
