Impact of Restricted Mass Transport on Pyrolysis Pathways for Aryl

Impact of Restricted Mass Transport on Pyrolysis Pathways for Aryl Ether ... Analytical Sciences Division, Oak Ridge National Laboratory, P.O. Box 200...
0 downloads 0 Views 105KB Size
1314

Energy & Fuels 2000, 14, 1314-1322

Impact of Restricted Mass Transport on Pyrolysis Pathways for Aryl Ether Containing Lignin Model Compounds Phillip F. Britt, A. C. Buchanan, III,* and Elizabeth A. Malcolm Chemical & Analytical Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831-6130 Received July 18, 2000. Revised Manuscript Received September 15, 2000

Pyrolysis studies have been conducted at 375 °C on several silica-immobilized phenethyl phenyl ether (PPE) model compounds, representative of related β-O-4 aryl ether linkages in lignin, to explore the impact of restricted mass transport on reaction pathways. As found previously for fluid-phase PPE, two competitive free-radical decay pathways are operative including a significant rearrangement pathway involving an O,C-phenyl shift for surface-attached PhCH2CH•OPh radicals. The selectivity for the rearrangement pathway is found to be sensitive to substituents and, in particular, to the structure of neighboring spacer molecules on the surface. In contrast to solution-phase behavior, dilution of PPE molecules on the surface with rigid aromatic spacers such as biphenyl or naphthalene hinder the rearrangement path. This phenomenon, attributed to steric constraints that decrease the rate of the 1,2-phenyl shift, is not observed when a more flexible spacer molecule (diphenylmethane) is employed. An improved knowledge of the pathways involved is important since this rearrangement pathway, which was also observed in the pyrolysis of R-aryl ether models, can result in the formation of valuable chemicals (aryl aldehydes and ketones) or undesirable refractory compounds (biphenyls and diphenylmethanes) during the thermochemical processing of lignin.

Introduction Utilization of renewable energy sources that recycle carbon dioxide is one avenue for reducing greenhouse gas emissions. There has been significant interest in biomass feedstocks that could provide alternative sources of electricity and liquid transportation fuels, as well as chemicals and materials.1-6 Lignin is the second most abundant naturally occurring biopolymer and a major byproduct of commercial pulping processes. Given its abundance and chemical structure, lignin is of considerable interest as a source of aromatic chemicals and speciality materials.3-6 The structure of this heterogeneous, three-dimensional biopolymer is quite complex as a consequence of its formation from the enzymeinitiated, dehydrogenative, free-radical copolymerization of trans-p-coumaryl alcohol, trans-coniferyl alcohol, and trans-sinapyl alcohol precursors.3,7-9 The phenylpro* Author to whom correspondence should be addressed. Phone: (865) 576-2168. Fax: (865) 574-4902. E-mail: [email protected]. (1) Developments in the Thermochemical Conversion of Biomass; Bridgwater, A. V., Boocock, D. G. B., Eds.; Blackie Academic: London, 1997. (2) Thermochemical Processing of Biomass; Bridgwater, A. V., Ed.; Butterworth: London, 1984. (3) Lignin: Historical, Biological, and Materials Perspectives; Glasser, W. G., Northey, R. A., Schultz, T. P., Eds.; ACS Symposium Series, No. 742; American Chemical Society: Washington, DC, 2000. (4) Organic Chemicals From Biomass; Goldstein, I. S., Ed.; CRC Press: Boca Raton, FL, 1981; Chapters 5 and 8. (5) Elliott, D. C.; Beckman, D.; Bridgwater, A. V.; Diebold, J.; Gevert, S. B.; Solantausta, Y. Energy Fuels 1991, 5, 399. (6) Bridgwater, A. V.; Cottam, M.-L. Energy Fuels 1992, 6, 113. (7) Alder, E. J. Wood Sci. Technol. 1977, 11, 169. (8) Nimz, H. Angew. Chem., Int. Ed. Engl. 1974, 13, 313.

panoid monomer units in these heteropolymers are connected by a myriad of linkages. However, aryl ether linkages are dominant and, in particular, the β-O-4 aryl ether linkages can account for up to half of the total number.3,7-9 The simplest model compound unadorned by substituents that embodies this aryl ether linkage is phenethyl phenyl ether (C6H5CH2CH2OC6H5 or PPE). Despite the extensive research into the pyrolysis of lignin, the fundamental reactions that lead to the complex array of products remain poorly understood, and necessitate more detailed mechanistic investigations employing model compounds. We have examined in detail the pyrolysis mechanism of PPE in the liquid and vapor phases ((3.8-5.2) × 10-3 M) at 330-425 °C.10 Four major products were found (eq 1), including the unexpected toluene and benzaldehyde product pair (eq 1b). From a thorough kinetic and mechanistic analysis of this reaction,

PhCH2CH2OPh f PhCHdCH2 + PhOH (1a) PhCH3 + PhCHO

(1b)

including the use of isotopic labeling and thermochemical kinetics calculations, we showed that the PPE decomposition was a free-radical chain process (eqs 2-8) with a kinetic chain length of ca. 89 at 330 °C and 25 (9) Dorrestijn, E,; Laarhoven, L. J. J.; Arends, I. W. C. E.; Mulder, P. J. Anal. Appl. Pyrolysis 2000, 54, 153. (10) Britt, P. F.; Buchanan, A. C.,III; Malcolm, E. A. J. Org. Chem. 1995, 60, 6523.

10.1021/ef000160w CCC: $19.00 © 2000 American Chemical Society Published on Web 10/11/2000

Pyrolysis Pathways for Aryl Ether Containing PPE Models

at 400 °C. The known pyrolysis products, phenol and styrene, were formed by a standard free-radical chain sequence involving the propagation steps

PhCH2CH2OPh f PhCH2CH2• + PhO•

(2)

PhO• + PhCH2CH2OPh f PhOH + PhCH•CH2OPh (1) (3a) PhOH + PhCH2CH•OPh (2)

(3b)

1 f PhCHdCH2 + PhO•

(4)

2 f PhCH2CH(Ph)O• (3)

(5)

3 f PhCHO + PhCH2•

(6)

PhCH2• + PhCH2CH2OPh f 1 + PhCH3 (7a) 2 + PhCH3

(7b)

PhCH2CH2• + PhCH2CH2OPh f 1 + PhCH2CH3 (8a) 2 + PhCH2CH3

(8b)

shown in eqs 3a and 4 following the initial C-O homolysis step (eq 2). However, we also discovered a second previously unreported free-radical reaction pathway that produced toluene and benzaldehyde in significant yields. This path (eqs 3b, and 5-7) accounted for ca. 25% of the pyrolysis products at 375 °C over the entire concentration range studied. Two factors are critical to the emergence of this competitive pathway.10 First, the phenoxy radical is electrophilic, and polar effects in hydrogen abstraction at the carbon adjacent to oxygen in PPE (eq 3b) enhance the selectivity for formation of 2 relative to 1 (ca. 1:3) over what is predicted from thermochemical kinetic considerations (ca. 1:18 at 375 °C).10 The second factor is the occurrence of key rearrangement step 5, a 1,2-phenyl shift from oxygen to carbon (neophyl-like rearrangement) in radical intermediate, 2, coupled with β-scission of the rearranged oxy-radical, 3 (eq 6). Kinetic analysis indicated that hydrogen abstraction is slow relative to the β-scission steps and the rearrangement step (e.g., k5/ k3a > 140 M at 375 °C).10 Hence, the product selectivity was governed by the relative rates of hydrogen abstraction at the R- and β-carbons that produce the corresponding radicals 1 and 2. This R,β-selectivity was measured to be 3.1 ( 0.3 at 375 °C independent of the extent of reaction or initial PPE concentration. More recently, we have examined the pyrolysis of PPE and methoxy-substituted PPEs under flash vacuum pyrolysis conditions at 500 °C, which highlights the primary unimolecular reactions that occur under highly dilute (10-8 - 10-9 M), very short contact time conditions (3.1 (the fluid-phase value10) indicating

Pyrolysis Pathways for Aryl Ether Containing PPE Models

Energy & Fuels, Vol. 14, No. 6, 2000 1317

Table 1: Composition and Pyrolysis of Surface-Attached Lignin Models at 375 °C surface composition

surface coveragea (mmol g-1)

no. of pyrolyses

conversion range (%)

rateb (% h-1)

corr. coeff.c (r)

path selectivityd

≈PPE-3 ≈PPE-3-o-OCH3 ≈PPE-1 ≈m-PPE-1 ≈PPE-3/≈p-BP ≈PPE-3/≈m-BP ≈PPE-3/≈NAP ≈PPE-3/≈DPM

0.54 0.55 0.50 0.44 0.050/0.54 0.066/0.51 0.072/0.45 0.059/0.48

6 2 5 5 5 5 6 5

0.9-16.9 17.4-36.1 5.8-26.1 1.4-17.1 0.8-7.7 1.4-9.8 1.0-11.1 2.2-13.7

8.3 e 72.6 11.4 3.6 4.7 4.3 8.0

0.997 e 0.987 0.999 0.997 0.992 0.992 0.973

5.0 ( 0.8 5.9 ( 1.0 3.2 ( 0.2 8.2 ( 1.0 19.9 ( 3.5 10.4 ( 1.8 20.2 ( 1.4 12.6 ( 1.8

a Component surface coverage on a per gram of derivatized silica basis. b Initial rates for total decomposition of PPE derivative from the slopes of linear regressions of conversion versus reaction time (see Figure 2) with reaction extent limited to ca. 17%. For ≈PPE-1, the 26.1% conversion point was not used in the initial rate plots. c Correlation coefficient from linear regression analysis of initial rates. d R/β-path selectivity based on product yields. e Not determined due to high conversions and limited number of runs.

Figure 2. Initial rates for disappearance of ≈PPE models at 375 °C: ([) ≈m-PPE-1 (0.44 mmol g-1), (b) ≈PPE-3 (0.54 mmol g-1), (9) ≈PPE-3/≈NAP (0.072/0.45 mmol g-1), (+) ≈PPE-3/≈p-BP (0.050/0.54 mmol g-1).

a decreased selectivity for the rearrangement path involving the O-C phenyl shift compared with fluidphase PPE. The origins of these effects will be discussed below. Discussion Silica-Attached PPE Models. ≈PPE-3. Pyrolysis of ≈PPE-3 (Figure 1) at high coverage (0.54 mmol g-1) and 375 °C occurred with an initial rate (Figure 2; 8.3% h-1) comparable to that measured for PPE in the gas phase (7-8% h-1).10 Product analysis revealed a simple mixture consisting of the four principal products shown in Scheme 1. These products account for >98% of the products with the remainder being associated with formation of small amounts (ca. 1.6%) of ≈PhCH2CH3 (formed analogous to eq 8). Mass balances were excellent (98 ( 2%) indicating that no significant products remain undetected. In addition, the stoichiometric balances (≈PhCHdCH2 + ≈PhCH2CH3)/PhOH and ≈PhCH3/PhCHO were found to be 1.04 ( 0.06 and 0.97 ( 0.08, respectively, over the conversion range studied. These values are within experimental error of unity, indicating again that all products are being properly accounted for and quantitated. The pyrolysis products are analogous to those detected for PPE in fluid phases, and can be accounted for by a similar free-radical chain mechanism.10 The chain propagation steps are shown in eqs 9-12 in Scheme 2. For ≈PPE-3, the chain-carrying phenoxy

radical is in the gas phase while the chain-carrying benzyl radical is surface immobilized. Clearly, the path involving rearrangement step 12 can also occur under diffusional constraints. The R/β-selectivity was found to be 5.0 (Table 1), which is larger that the value of 3.1 found for fluid-phase PPE. This higher selectivity is a consequence of a substituent effect in which the psilyloxy linkage to the surface provides enhanced stability for the benzylic radical formed competitively in eqs 9a and 10a. This conclusion was confirmed by pyrolysis studies of p-(CH3)3SiOPhCH2CH2OPh in the gas phase at 375 °C, where the trimethylsilyloxy substituent is employed as a model for the silyloxy linkage to the silica surface.13b,c An R/β-selectivity of 4.4 ( 0.5 was obtained consistent with the proposed substituent effect. ≈PPE-3-o-OCH3. Pyrolysis of ≈PPE-3-o-OCH3 has been briefly investigated previously (2-pyrolyses) to examine the effect of the o-methoxy substituent that is prevalent in certain lignins.13c As expected,11 the omethoxy substituent resulted in a much faster PPE decomposition rate (ca. 12-14-fold) as a result of enhanced rate of initiation from O-C bond scission analogous to eq 2, and low conversions could not be examined at 375 °C. This rate enhancement is in good agreement with the FVP studies11 where the rate of conversion for PPE-o-OCH3 was 7.4 times faster than for PPE at 500 °C (corresponding to a factor of 11 at 375 °C). As shown in Scheme 3, a product mixture analogous to that for ≈PPE-3 was obtained, and a comparable R/β-path selectivity was measured (Table 1). ≈PPE-1. Pyrolysis of the ≈PPE-1 isomer provides an interesting comparison to the behavior of ≈PPE-3, since in this case the chain-carrying phenoxy radical will be surface immobilized while the chain-carrying benzyl radical will be in the gas phase. Furthermore, since the benzylic radical is now remote from the silyloxy linkage to the surface, the substituent effect on the R/β-selectivity observed for ≈PPE-3 should be eliminated. Pyrolysis of ≈PPE-1 proceeded analogously to ≈PPE-3 to give the four principal products shown in Scheme 4. In contrast to the ≈PPE-3 isomer (Scheme 1), the phenol and benzaldehyde products are surface attached, while the styrene and toluene products are in the gas phase. The reaction rate for ≈PPE-1 (Table 1) was found to be faster than for ≈PPE-3 (ca. 9-fold). This is most likely the result of a substituent effect (the p-silyloxy linkage) that increases the rate of the initiation reaction (eq 13). The p-methoxy substituent has been shown to weaken the O-CH3 bond in p-methoxyanisole by 4 kcal mol-1,27 and

1318

Energy & Fuels, Vol. 14, No. 6, 2000

Buchanan et al. Scheme 1

Scheme 2

(13)

in Scheme 2. The R/β-selectivity, calculated from the yield of styrene relative to toluene, was found to be 3.2 ( 0.2. This value is comparable to the value measured for fluid-phase PPE, and provides further evidence that the larger selectivity observed for the ≈PPE-3 isomer was the result of a substituent effect (vide supra). The result also indicates that the O-C phenyl shift, which must now occur for ≈PhOCH•CH2Ph involving the phenyl ring attached to the silica surface, is not perturbed by the surface for this para-linked isomer. Mass balances were lower (ca. 91%) compared with ≈PPE-3, and the four main products accounted for about 93% of the products quantified. The additional products detected included PhCH2CH3, ≈PhH, ≈PhPh, and ≈PhCH2Ph. The lower mass balance for ≈PPE-1 pyrolysis was found to be a consequence of nonquantitative recovery of the ≈PhOH product, indicated by a product ratio for PhCHdCH2/≈PhOH of ca. 2 rather than the stoichiometric value of 1. During workup, ≈PhOH is converted to p-HOPhOH, which is very water soluble and inefficiently extracted at the low concentrations involved in this study (see Experimental Section). Interestingly, there is also a deficit of the ≈PhCHO product relative to PhCH3, the magnitude depending on extent of conversion. This is a consequence of the fact

The pyrolysis products formed are consistent with a free-radical chain mechanism analogous to that shown

(27) Suryan, M. M.; Kafafi, S. A.; Stein, S. E. J. Am. Chem. Soc. 1989, 111, 1423.

Scheme 3

Scheme 4

a similar effect is predicted for the p-silyloxy linkage. In support of this conclusion, the rate enhancement is eliminated for ≈m-PPE-1 where the m-silyloxy substituent can no longer stabilize the silica-bound phenoxy radical (vide infra).

≈PhOCH2CH2Ph f ≈PhO• + PhCH2CH2•

Pyrolysis Pathways for Aryl Ether Containing PPE Models

that, for ≈PPE-1, the reactive benzaldehyde product is silica-immobilized, remains in the heated zone, and can undergo radical decomposition analogous to the secondary chemistry observed for liquid-phase PPE.10 Hydrogen abstraction of the C-H bond in benzaldehyde (87 kcal mol-1),28 as shown in eq 14, generates the benzoyl radical that readily decarbonylates at 375 °C (t1/2 ) 14 µs29) to form a surface-attached phenyl radical. Subsequent reaction of ≈Ph• produces ≈PhH (hydrogen abstraction), ≈PhPh (aromatic substitution on ≈PPE1), and ≈PhCH2Ph (coupling with gas-phase benzyl radicals)

Energy & Fuels, Vol. 14, No. 6, 2000 1319 Scheme 5

Scheme 6

≈PhCHO + R• f RH + ≈PhC•O f CO + ≈Ph• ff ≈PhH, ≈PhPh, ≈PhCH2Ph (14) as secondary products. The extent of this secondary chemistry is illustrated by the result that, at 13% conversion of ≈PPE-1, we find 41% of the ≈PhCHO has reacted to form ≈PhH (73%), ≈PhPh (14%), and ≈PhCH2Ph (13%). The ratio of hydrogen abstraction to aromatic substitution for ≈Ph• at 375 °C (5.2) is consistent with the reported selectivity for hydrogen abstraction versus arylation for phenyl radical reacting with toluene in the gas phase at 400 °C (5.2 ( 0.2).30 The significance of this result for the thermal processing of lignin is that aryl substitution reactions by phenyl radicals lead to more thermally robust aryl-aryl linkages (ca. 115 kcal mol-1).29 The degree of formation of the key benzaldehyde precursors will be directly related to the selectivity for the competitive rearrangement path for the β-O-4 aryl ether linkages. We have also found recently that aryl aldehydes can be formed during flash pyrolysis of methoxy-substituted lignin models via related radical rearrangements involving the methoxy groups (ArOCH2• f ArCH2O• f ArCHO),11 but under the low concentration conditions involved, the aldehydes do not undergo subsequent bimolecular reactions. ≈m-PPE-1. Pyrolysis of ≈m-PPE-1 was also examined for comparison with the para-isomer. It was anticipated that the accelerated rate observed for the para-isomer (vide supra) compared to ≈PPE-3 would be largely eliminated if the silyloxy surface linkage were in the meta-position, since it would not be able to directly stabilize the incipient phenoxy radical formed in the initiation step. As shown in Table 1, the pyrolysis rate of ≈m-PPE-1 was indeed found to be about a factor of 6 slower than for the para-isomer, and was comparable to that for ≈PPE-3 consistent with removal of the substituent effect. The pyrolysis products shown in Scheme 5 are analogous to those observed for the paraisomer. An improved PhCHdCH2/≈m-PhOH stoichiometric balance of 1.05 was obtained as a consequence of improved extraction efficiencies of m-HOPhOH compared with p-HOPhOH obtained from the para-isomer of PPE-1. As in the para-isomer case, the surface-bound benzaldehyde product was found to be unstable and to undergo analogous secondary, radical-induced decarbonylation chemistry. (28) McMillen, D. F.; Golden, D. M. Annu. Rev. Phys. Chem. 1982, 33, 493. (29) Poutsma, M. L. J. Anal. Appl. Pyrolysis 2000, 54, 5. (30) (a) Fahr, A.; Stein, S. E. J. Am. Chem. Soc. 1988, 92, 4951. (b) Chen, R. H.; Kafafi, S. A.; Stein, S. E. J. Am. Chem. Soc. 1989, 111, 1418.

Surprisingly, the R/β-selectivity measured for this isomer increased to a value of 8.2 compared with a selectivity of 3.2 for the para-isomer. Our hypothesis is that the rearrangement step involving the O-C phenyl shift experiences steric constraints through interactions with neighboring molecules or un-derivatized silanols on the silica surface. This is illustrated in Scheme 6 for the intermediate oxaspiro[2.5]octadienyl radicals in the rearrangements for the meta- and para-isomers.29,31 The increased selectivity does not appear to be the result of a substituent effect.32 We have observed that, in the preparation of high surface coverages of silica-attached aromatic molecules, saturation surface coverages are typically 10-30% lower for meta- compared to paraisomers, consistent with a higher steric demand. Furthermore, in the decomposition of β-phenylisovaleryl peroxide on silica at 55 °C, it has been demonstrated that the silica surface inhibits the phenyl shift for the archetypical neophyl radical compared with solutions (eq 15).35 The consequence of a decreased rate for the O-C phenyl shift step

PhC(CH3)2CH2• f •C(CH3)2CH2Ph

(15)

(31) In the O-neophyl rearrangement of 1,1-diphenylethoxy radical, a similar intermediate has been detected by transient optical spectroscopy. See: Falvey, D. E.; Khambatta, B. S.; Schuster, G. B. J. Phys. Chem. 1990, 94, 1056. (32) The rate of the neophyl rearrangement is increased by electronwithdrawing substituents and decreased by electron-donating substituents.33 The Hammett substituent constant for p-OCH3 is -0.27 (electron donating) and for m-OCH3 is 0.12 (electron withdrawing).34 Hence, with the reasonable assumption that the silyloxy linkage (see Scheme 6) can be represented by the methoxy group, the rate of the 1,2-phenyl shift for the meta-isomer of PPE-1 is predicted to be similar or slightly faster than for the para-isomer. Likewise, no significant perturbation on hydrogen abstraction selectivities at the R- and β-carbons is expected at this high surface coverage.12c,d (33) Wentrup, C. Reactive Molecules; John Wiley & Sons: New York, 1984; p 94. (34) Isaacs, N. Physical Organic Chemistry, 2nd ed.; Longman Scientific & Technical: Essex, 1995; p 152. (35) Leffler, J. E.; Barbas, J. T. J. Am. Chem. Soc. 1981, 103, 7768.

1320

Energy & Fuels, Vol. 14, No. 6, 2000

at low extents of conversion would be the incursion of competitive bimolecular hydrogen transfer step 16 that converts the β-radical into the R-radical (estimated to be ca. 7.4 kcal mol-1

Buchanan et al. Scheme 7

≈PhOCH•CH2Ph + ≈PhOCH2CH2Ph f ≈PhOCH2CH2Ph + ≈PhOCH2CH•Ph (16) more stable10) and, hence, result in a larger R/βselectivity value. More direct evidence on the role of steric effects in perturbing the selectivity for the rearrangement path for PPE models comes from studies of ≈PPE-3 in the presence of spacer molecules described in the next section. Impact of Spacer Molecules on ≈PPE-3 Pyrolysis. To study the effect of structure of neighboring spacer molecules on the pyrolysis mechanism, ≈PPE-3 was chosen as the lignin model. As described above, pyrolysis of ≈PPE-3 provides a simple product mixture without secondary products in the conversion range studied, and has an R/β-selectivity that is only slightly altered from fluid-phase PPE resulting from an understood substituent effect. The spacer molecules chosen are shown in Figure 1 and include rigid aromatics with no readily abstractable hydrogens (biphenyl,≈p-BP and ≈m-BP; naphthalene, ≈NAP) as well as a more flexible spacer with abstractable benzylic hydrogens (diphenylmethane, ≈DPM). For ready comparison, the samples were prepared with the same surface coverages of ≈PPE-3 and spacer molecules. Pyrolysis of these samples at 375 °C resulted in the same products found for surfaces of ≈PPE-3 alone (Scheme 1) with no new products detected. However, the R/β-selectivity (Table 1) was found to be substantially altered and dependent on the spacer structure. Pyrolysis of ≈PPE-3/≈p-BP resulted in a slower pyrolysis rate as a consequence of dilution of the ≈PPE-3 molecules on the surface, which inhibits the rate-limiting hydrogen transfer steps. More significantly, a dramatic change in the R/β-selectivity occurred with a value of 19.9 obtained compared with 5.0 for surfaces of ≈PPE-3 alone. This 4-fold increase in path selectivity is unexpected based on the behavior of PPE in the fluid-phase. For comparison, pyrolysis of PPE at 375 °C in fluid phases resulted in an R/β-path selectivity (3.1) that was independent of concentration over a wide range (103) in going from the neat liquid to a dilute gas. Furthermore, the R/β-selectivity was unaltered when PPE was diluted in biphenyl as a solvent. Hence, the modified selectivity for ≈PPE-3 in the presence of ≈pBP is a direct result of the restricted mass transport experienced on the surface. The presence of the biphenyl spacer molecules should not affect the selectivity for the hydrogen abstractions from the R- and β-carbons of ≈PPE-3.36 These results suggest that the presence of the rigid biphenyl molecules introduces steric constraints that are not present in high surface coverages of the more flexible ≈PPE molecules. The densely packed ≈BP molecules likely hinder the O-C phenyl shift step, as illustrated in Scheme 7, which is the critical step in the rearrangement path that forms the (36) Pyrolysis studies of ≈Ph(CH2)3Ph showed that biphenyl and naphthalene spacers did not alter the hydrogen abstraction selectivity compared with surfaces of ≈Ph(CH2)3Ph at similar low surface coverage without spacer molecules.12c

≈PhCH3 and PhCHO products. The consequence of the hindered rate for the phenyl shift is an increased R/βselectivity, which arises from the radical interconversion step 16. Similar results were obtained when naphthalene (attached at 2-position as shown in Figure 1) was employed as the spacer. Pyrolysis of ≈PPE-3/≈NAP gave an R/β-selectivity value of 20.2, indistinguishable from the ≈p-BP spacer, and providing additional evidence for aromatic spacers hindering the rearrangement path under diffusional constraints. However, the use of m-biphenyl as a spacer gave a slightly different result. Pyrolysis of ≈PPE-3/≈m-BP resulted in ≈PPE-3 pyrolysis rates comparable to the other aromatic spacers, but the R/β-selectivity value was measured as 10.4. This is still a factor of 2 larger than for ≈PPE-3 alone indicating that steric constraints on the rearrangement path remain, but not as severe as for ≈PPE-3/≈p-BP and ≈PPE-3/≈NAP surfaces. The comparison between the m-BP and p-BP spacers is intriguing and suggests that the surfaces with the m-BP spacers are perhaps more disordered, provide more freedom of motion for ≈PPE-3, and are not as inhibiting to the phenyl shift. Clearly more information is needed to fully understand the dynamics of these multicomponent surfaces, and the subsequent impact on free-radical reaction mechanisms. We have begun transient optical spectroscopy and computational modeling studies to attempt to better understand the dynamics of related silica-immobilized systems. The final spacer examined was diphenylmethane (≈DPM). It was anticipated that this more flexible spacer molecule might relieve some of the steric constraints on the rearrangement path. However, interpretation of the results are complicated by the fact that DPM molecules can become involved in hydrogen transfer steps that determine the proportion of R- and β-carbon radicals formed (eqs 9 and 10 in Scheme 2). To understand the impact of this on the R/β-selectivity in the absence of diffusional effects, we examine the impact of dilution of PPE with DPM reported in fluid phases.10 Pyrolysis of PPE in DPM (mol ratio of 1:8) at 375 °C occurred at a comparable rate to that in biphenyl solvent. However, the R/β-selectivity increased from 3.1 to 8.1 (factor of 2.6). This was shown to be a result of chain transfer chemistry in which the electrophilic phenoxy radical reacts primarily with DPM to give the diphenylmethyl radical (eq 17), which in turn abstracts hydrogen from PPE (eq 18). As discussed in the Introduction and in considerable detail in ref 10, an R/βselectivity value of ca. 18 is predicted in the absence of

Pyrolysis Pathways for Aryl Ether Containing PPE Models

Energy & Fuels, Vol. 14, No. 6, 2000 1321

Scheme 8

polar effects based on the higher thermochemical stability (ca. 7.4 kcal mol-1) of 2 relative to 1. Hence, the presence of DPM leads to a higher R/β-path selectivity by increasing the concentration of nonpolar radicals that increase the selectivity for formation of the R-radical (1) via steps 17 and 18.

PhO• + PhCH2Ph f PhOH + PhCH•Ph

(17)

PhCH•Ph + PhCH2CH2OPh f PhCH•CH2OPh (1) + PhCH2Ph (18a) f PhCH2CH•OPh (2) + PhCH2Ph

(18b)

Using the fluid-phase behavior as a foundation, we studied the pyrolysis of ≈PPE-3/≈DPM at 375 °C at a comparable mole ratio of 1:8 based on the surface coverages. We found that the ≈PPE-3 conversion rate is accelerated compared with the aromatic spacers, and comparable to high coverage surfaces of ≈PPE-3 alone (Table 1). Our previous studies of the pyrolysis of the structurally related, silica-immobilized 1,3-diphenylpropane in the presence of spacer molecules have shown that ≈DPM molecules can participate in facile chain transfer steps at 375 °C during the free-radical chain decomposition.12c This hydrogen transfer, radical relay pathway provides a means for radicals to be translocated on the surface without requiring physical diffusion. One result of this process is accelerated rates compared with non-hydrogen donating spacers such as biphenyl.12c In the case of ≈PPE-3/≈DPM pyrolysis, an additional consequence is the conversion of a significant fraction of the electrophilic phenoxy radical into a surface-bound diphenylmethyl radical (Scheme 8). If one assumes that the R/β-selectivity increase (2.6-fold) found in solution phase is applicable to the surface-immobilized conditions at comparable mole ratios, then an R/βpath selectivity value of 2.6 × 5.0 ) 13.0 is predicted. The measured value of 12.6 ( 1.8 is within experimental error of the predicted value. This suggests that the increased path selectivity for ≈PPE-3/≈DPM compared with ≈PPE-3 alone arises primarily from alterations in the hydrogen abstraction selectivity producing radicals 1 and 2, and is not significantly influenced by steric constraints on the O-C-phenyl shift as observed for the ≈BP and ≈NAP spacers.

Conclusions Pyrolysis studies of lignin model compounds containing the dominant β-O-4 aryl ether linkage have shown that these structures undergo facile decay at 375 °C by two competitive free-radical pathways. In the current study using silica-immobilized models, both pathways were found to be operable under conditions of restricted mass transport. However, the path selectivity was found to be sensitive to substituents on the aromatic rings, the polarity of the dominant hydrogen-abstracting radicals, as well as the structure of neighboring molecules on the surface. In stark contrast to solution phase behavior, pyrolysis of silica-immobilized phenethyl phenyl ether diluted with rigid aromatic spacer molecules such as biphenyl and naphthalene resulted in a reduced selectivity for the rearrangement pathway involving the neophyl-like, O-C phenyl shift in the β-carbon radical, 2. This result has been attributed to steric constraints that reduce the rate of the phenyl shift, and allow competitive hydrogen transfer steps to intervene that convert 2 into the more stable benzylic radical, 1. However, more flexible spacer molecules such as diphenylmethane do not seem to cause significant steric constraints on this rearrangement path. The principal impacts of this spacer are to permit facile decomposition of the diluted PPE as a result of fast hydrogen transferradical relay steps on the surface, and as a chain transfer agent that converts a significant fraction of the electrophilic phenoxy radicals into nonpolar diphenylmethyl radicals that have a different hydrogen abstraction selectivity for PPE. Additional research that provides an improved understanding of the dynamics of molecular interactions in these multicomponent systems would clearly be a valuable in obtaining a more refined understanding of the effects of steric constraints on the rearrangement path. Our recent study of a silica-immobilized model for R-aryl ether linkages in lignin, benzyl phenyl ether (BPE), has provided an additional example for the significance of rearrangement paths involving an O-C phenyl shift. In this case, the rearrangement path for the key radical intermediate, ≈PhOCH•Ph, generated benzophenone and benzyhydrol as major products on the surface.13a In the thermal processing of lignin, these aryl ether rearrangement pathways for PPE- and BPElike structures generate products such as aldehydes and

1322

Energy & Fuels, Vol. 14, No. 6, 2000

ketones that have potential commercial value. On the other hand, restricted radical and substrate diffusion has a significant impact on secondary reactions for aldehyde products (cf. ≈PPE-1). Decomposition of aryl aldehydes generates aryl radicals that can undergo substitution reactions on adjacent aromatic molecules, which is a potential cross-linking reaction that yields refractory biaryl-type products. These arylated products are not detected to any significant extent in solutionphase studies of PPE models. Our increasingly improved knowledge of the formation and decay pathways for aryl aldehyde and ketone products, and the reaction conditions that increase or decrease the selectivity for their formation, could be important in developing advanced processes for recovering these valuable aromatic chemicals. Experimental Section General. GC analysis was performed on a Hewlett-Packard 5890 Series II gas chromatograph employing a J & W Scientific 30 m × 0.25 mm DB-1 methylsilicone column (0.25 µm film thickness) and flame-ionization detection. Product detector response factors were determined relative to cumene (hydrocarbon products) or 2,5-dimethylphenol and p-(2-phenylethyl)phenol (phenolic products) as internal standards. Mass spectra were obtained at 70 eV with a Hewlett-Packard 5972A/5890 Series II GC-MS equipped with a capillary column matched to that used for GC analyses. High-purity acetone, methylene chloride, and water were commercially available and used as received. Benzene was distilled from lithium aluminum hydride under argon before use. Cumene was fractionally distilled (2×), and 2,5-dimethylphenol and p-(2-phenylethyl)phenol were recrystallized (3×) from hexanes and benzene/ hexanes, respectively, prior to use. The syntheses of the precursor phenols for preparation of the PPE surfaces have been previously described.13c The purifications of the precursor phenols for the spacer molecules employed have also been described.12c Preparation of Surface-Attached Materials. Preparation of the surface-attached phenethyl phenyl ether models followed the general methodology described in detail in previous publications.12,13 In general, the precursor phenol (2.25 equivalents relative to surface SiOH) was adsorbed onto the dried (200 °C, 4 h) surface of a fumed silica (Cabosil M-5, Cabot Corp., 200 m2 g-1, ca. 4.5 SiOH nm-2 or 1.5 mmol SiOH g-1) by solvent evaporation from a benzene slurry. Surface-attachment reaction was performed in a sealed, evacuated (2 × 10-6 Torr) Pyrex tube at 225 °C for 0.5 h in a fluidized sand bath. Unattached phenol was removed by either sublimation at 270 °C for 1 h under dynamic vacuum (5 × 10-3 Torr), or Soxhlet extraction with dry benzene. The resulting silica-attached PPE

Buchanan et al. models were free-flowing white powders. Surface coverage analysis was performed on 150-200 mg samples dissolved in 30 mL 1 N NaOH overnight. 4-(2-phenylethyl)phenol in 1 N NaOH was added as an internal standard. The solution was acidified with HCl (pH = 4), and extracted with methylene chloride (3 × 7 mL). The combined organic layer was washed with water (1 × 10 mL), dried over MgSO4, filtered, and the solvent removed under reduced pressure. Silylation with N,Obis(trimethylsilyl)trifluoroacetamide (BSTFA) in pyridine (2.5 M, 0.30 mL) produced the corresponding trimethylsilyl ether. GC analysis of multiple assays provided surface coverages with reproducibility of (2%. Two-component surfaces were prepared by co-attachment of the PPE and spacer phenolic precursors in an analogous manner.12 Pyrolysis Procedure. The pyrolysis apparatus and procedure have been described previously.12,13 In brief, a weighed amount of sample (0.4-0.6 g) was placed in one end of a T-shaped Pyrex tube, evacuated, and sealed at ca. 2 × 10-6 Torr. The sample was inserted into a preheated temperaturecontrolled ((1.0 °C) tube furnace, and the other end placed in a liquid nitrogen bath. The volatile products collected in the trap were dissolved in acetone (0.1-0.3 mL) containing internal standards (vide supra) and analyzed by GC and GCMS. Surface-attached pyrolysis products were similarly analyzed after separation by digestion of the silica in aqueous base and silylation of the resulting phenols to the corresponding trimethylsilyl ethers as described above for the surface coverage analysis. For ≈PPE-1, the extraction efficiency of the hydroquinone product from aqueous solution was increased by conducting the base hydrolysis under an argon atmosphere (to prevent oxidation of hydroquinone to quinone). The solution was then acidified to pH = 4, the solution saturated with LiCl, and extracted with diethyl ether (5 × 5 mL). The combined ether layers were dried over MgSO4 and the solvent removed under reduced pressure. Control experiments showed that, under these conditions, hydroquinone could be recovered with ca. 80% efficiency. For ≈m-PPE-1, resorcinol was also recovered from the aqueous solution by extraction with diethyl ether (5 × 5 mL), and control experiments demonstrated quantitative recovery of resorcinol could be achieved under these conditions.

Acknowledgment. Research sponsored by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, U.S. Department of Energy under contract DE-AC05-00OR22725 with Oak Ridge National Laboratory, managed and operated by UT-Battelle, LLC. The authors appreciate technical support provided by K. B. Thomas, J. Fox, and V. Hitsman. EF000160W