DOI: 10.1021/la101424a 16291Langmuir 2010, 26(21), 16291–16298 Published on Web 06/01/2010

pubs.acs.org/Langmuir

© 2010 American Chemical Society

γ-Valerolactone Ring-Opening and Decarboxylation over

SiO2/Al2O3 in the Presence of Water†

Jesse Q. Bond, David Martin Alonso, Ryan M. West, and James A. Dumesic*

Department of Chemical and Biological Engineering, University of Wisconsin, Madison, Wisconsin 53706

Received April 9, 2010. Revised Manuscript Received May 10, 2010

γ-Valerolactone (GVL) has been identified as a promising, sustainable platformmolecule that can be produced fromlignocellulosic biomass. The chemical flexibility ofGVLhas allowed the development of a variety of processes to preparerenewable fuels and chemicals. In the present work involving a combination of computational and experimental studies,we explore the factors governing the ring-opening ofGVL to produce pentenoic acid isomers, as well as their subsequentdecarboxylation over acid catalysts or hydrogenation over metal catalysts. The ring-opening of GVL has shown to be areversible reaction, while both the decarboxylation and hydrogenation reactions are irreversible and kineticallycontrolled under the conditions studied (temperatures from about 500 to 650 K). The most significant contributor tolactone reactivity toward ring-opening is the size of the ring, with γ- lactones being more stable and less readily openedthan δ- and ε-analogues. We have observed that the presence of either a CdC double bond or a lactone (which opens toform a CdC double bond) is necessary for appreciable rates of decarboxylation to occur. Olefinic acids exhibit higherrates of decarboxylation than the corresponding lactones, suggesting that the decarboxylation of alkene acids provides alower energy pathway to olefin production than the direct decarboxylation of lactones. We observe lower rates ofdecarboxylation as the chain length of alkene acids increases; however, acrylic acid (3-carbon atoms) does not undergodecarboxylation at the conditions tested. These observations suggest that particular double bond configurations yieldthe highest rates of decarboxylation. Specifically, we suggest that the formation of a secondary carbenium ion in theβ position leads to high reactivity for decarboxylation. Such an intermediate can be formed from 2- or 3-alkene acidswhich have at least four carbon atoms.

1. Introduction

Research directed toward the replacement of petroleum withrenewable feedstocks in the production of fuels and chemicals,such as utilization of lignocellulosic biomass, has received in-creased attention in the past decade. In 2004, the U.S. Depart-ment of Energy published a report1 identifying 15 compounds asthose having high potential for application in future biorefineries.In 2010, this list was revised to include more recent advances inbiomass conversion technologies.2 Among the compounds con-sidered in the 2010 review, ethanol and glycerol derivatives arealready being produced at a commercial scale. The production ofother interesting platformmolecules, for example, furans, xylitol,sorbitol, lactic acid, and levulinic acid, have been studied exten-sively at the laboratory scale, and various strategies for theirutilization have beenproposed.3-8 In this respect,γ-valerolactone(GVL) has been identified as a platform molecule with many

applications as a precursor for chemicals and fuels, as illu-strated in Figure 1. GVL can be produced in high yield(>95%) by reduction of levulinic acid,9,10 the latter of whichhas been produced from waste biomass at the pilot scale(Biofine process11,12). GVL synthesis can be achieved, typicallyin the presence of Ru-based catalysts,13 using the formic acidcoproduct (from levulinic acid production) as an in situ sourceof hydrogen.9

GVL can be used directly as a fuel additive in a capacity similarto ethanol14 or as an environmentally benign solvent.15 Alterna-tively, it can be hydrogenated to form 1,4-pentanediol,16 whichreadily undergoes dehydration to formmethyl tetrahydrofuran, aP-series fuel that can be blended to high levels (70% by volume)with gasoline for use in the present transportation infrastructure.17

By reaction with formaldehyde, GVL can be converted intoR-methylene-γ-valerolactone, a precursor of acrylic polymers.13

Through transesterification, GVL can be converted into methyl-pentanoate which can be processed by hydroformylation, hydro-cyanation, or hydroxycarbonylation to yield caprolactone, capro-lactam, or adipic acid, respectively.18

† Part of the Molecular Surface Chemistry and Its Applications specialissue.*To whom correspondence should be addressed. E-mail: dumesic@

engr.wisc.edu. Fax: þ1 608 262 5434.(1) Werpy, T.; Petersen, G. Top Value Added Chemicals from Biomass. Vol. I -

Results of Screening for Potential Candidates from Sugars and Synthesis Gas; NRELReport DOE/GO-102004-1992; U.S. Department of Energy, 2004. http://www1.eere.energy.gov/biomass/pdfs/35523.pdf.(2) Bozell, J. J. Green Chem. 2010, 12, 525–728.(3) Chheda, J. N.; Huber,G.W.; Dumesic, J. A.Angew. Chem., Int. Ed. 2007, 46,

7164–7183.(4) Bozell, J. J.; Moens, L.; Elliott, D. C.; Wang, Y.; Neuenscwander, G. G.;

Fitzpatrick, S. W.; Bilski, R. J.; Jarnefeld, J. L. Resour., Conserv. Recycl. 2000, 28,227–239.(5) Roman-Leshkov,Y.;Chheda, J.N.;Dumesic, J.A.Science 2006,312, 1933–1937.(6) Blommel, P. G.; Keenan, G. R.; Rozmiarek, R. T.; Cortrigth, R. D. Int.

Sugar J. 2008, 110, 672.(7) Kunkes, E. L.; Simonetti, D. A.; West, R. M.; Serrano-Ruiz, J. C.; Gartner,

C. A.; Dumesic, J. A. Science 2008, 322, 417–421.(8) Serrano-Ruiz, J. C.; Dumesic, J. A. ChemSusChem 2009, 2, 581–586.

(9) Deng, L.; Li, J.; Lai, D. M.; Fu, Y.; Guo, Q. X.Angew. Chem., Int. Ed. 2009,48, 6529–6532.

(10) Heeres, H.; Handana, R.; Chunai, D.; Rasrendra, C. B.; Girisuta, B.;Heeres, H. J. Green Chem. 2009, 11, 1247–1255.

(11) U.S. Patent 4897497, 1990(12) U.S. Patent 5608105, 1997(13) Manzer, L. E. Appl. Catal,. A 2004, 272, 249–256.(14) Horvath, I. T.; Mehdi, H.; Fabos, V.; Boda, L.; Mika, L. T. Green Chem.

2008, 10, 238–242.(15) Horvath, I. T. Green Chem. 2008, 10, 1024–1028.(16) Mehdi, H.; Fabos, V.; Tuba, R.; Bodor, A.;Mika, L. T.; Horvath, I. T.Top.

Catal. 2008, 48, 49–54.(17) Huber, G. W.; Iborra, S.; Corma, A. Chem. Rev. 2006, 106, 4044–4098.(18) Lange, J. P.; Vestering, J. Z.; Haan, R. J. Chem. Commun. 2007, 33, 3488–

3490.

16292 DOI: 10.1021/la101424a Langmuir 2010, 26(21), 16291–16298

Article Bond et al.

Our group has recently reported two strategies for the produc-tion of high energy density diesel and jet fuel range hydrocarbonsfrom GVL. In one processing strategy, GVL is converted topentanoic acid, which is used to produce 5-nonanone by ketoni-zation at high yields (92%).19 5-Nonanone can then be deoxyge-nated for use in diesel fuel blends or hydrogenated/dehydrated toproduce nonene, which is an appropriate oligomerization feed-stock for the production of C18 olefins.

20 In a second processingstrategy, GVL can undergo decarboxylation to produce butene,which can be processed directly for the production of branched,jet fuel range olefins by oligomerization.21 The production ofpentanoic acid takes place over bifunctional metal/acid catalysts(e.g., Pd-Nb2O5) at 623 K and 35 bar with a hydrogen cofeed.The production of pentanoic acid occurs through ring-openingover the acidic niobia support to form a mixture of pentenoicacids. Then, in the presence of Pd and a hydrogen cofeed, theCdC double bond is hydrogenated to form pentanoic acid. Atthese conditions, the rate of decarboxylation of the intermediate(pentenoic acid) is slow relative to that of hydrogenation, andonly pentanoic acid and unconverted GVL were observed in theliquid effluent.19 Over an acidic catalyst, such as SiO2/Al2O3, inthe absence of a metallic hydrogenation component, decarboxy-lation of GVL is the primary reaction and high yields of buteneand CO2 are achieved from aqueous solutions (30-60 wt %) ofGVL at 648 K and pressures ranging from 1 to 36 bar.21

In the present paper, we address in greater detail the variouschemical transformations ofGVL in the presence ofmetal or acidcatalysts. Of particular interest is the reversible conversion

between GVL and pentenoic acids, and the mechanism ofdecarboxylation, both of which occur in the presence of solidacid catalysts. We first present a computationally derivedoverview of the thermodynamics governing the production ofpentenoic acid, butene/CO2, and pentanoic acid from GVL.Additionally, we have included experimental studies to probethe extent of equilibrium and kinetic control in the afore-mentioned reactions. The mechanism of pentenoic acid decar-boxylation has not been previously described; as such, we havecarried out experimental studies using various feedstocks, and wepresent a pathway for the decarboxylation of alkene acids andlactones, providing insight regarding molecular structures appro-priate for acid catalyzed decarboxylation. We note that we haveattempted (to the best of our ability) to pattern our approach tothat employed masterfully by Gabor Somorjai throughout hiscareer: to search for fundamental underlying principles towardthe solution of practical problems faced by society.

2. Material and Methods

2.1. Materials. Amorphous SiO2/Al2O3 obtained fromGrace-Davison/Davicat (SIAL 3113) was used as an acid catalystfor the decarboxylation of various lactones and organic acids.Prior to reaction kinetics studies, the SiO2/Al2O3 catalyst wasdried and calcined in situ under flowing air at 723 K. A 1 wt %Pd/C catalyst was prepared by incipient wetness impregnation ofa carbon support (Norit) with a 10wt% tetraamine palladium(II)nitrate solution in 10 wt % nitric acid (Aldrich). After impregna-tion, the catalyst was dried overnight at 393Kand calcined for 3 hin flowing 20% O2 in He gas (250 cm3(STP)/min) at 533 K.

γ-Valerolactone (GVL) (Sigma-Aldrich, >98%), δ-valerolac-tone (DVL) (Sigma-Aldrich, tech. grade), γ-caprolactone (GCL)(Sigma-Aldrich, 98%), ε-caprolactone (ECL) (Sigma-Aldrich,97%), 2-pentenoic acid (2PEA) (Sigma-Aldrich, 97%), 3-pente-noic acid (3PEA) (Sigma-Aldrich, 95%), 4-pentenoic acid (4PEA)(Sigma-Aldrich, 97%), 3-butenoic acid (BEA) (Sigma-Aldrich,

Figure 1. Reaction pathways for the conversion of γ-valerolactone to fuels and chemicals.

(19) Serrano-Ruiz, J. C.; Wang, D.; Dumesic, J. A. Green Chem. 2010, 12, 574–577.(20) Alonso, D. M.; Bond J. Q.; Serrano-Ruiz, J. C.; Dumesic, J. A. Green

Chem. 2010, published online April 26, http://dx.doi.org/10.1039/C001899F.(21) Bond, J. Q.; Alonso, D.M.; Wang, D.; West, R.M.; Dumesic, J. A. Science

2010, 327, 1110–1114.

DOI: 10.1021/la101424a 16293Langmuir 2010, 26(21), 16291–16298

Bond et al. Article

97%), 2-hexenoic acid (2HEA) (Sigma-Aldrich, >98%), penta-noic acid (PAA) (Sigma-Aldrich >99%), propanoic acid (Sigma-Aldrich>99.5%), acrylic acid (Sigma-Aldrich, 99%), 5,6-dihydro-2H-pyran-2-one (DHP) (Sigma-Aldrich, tech. grade), R-angelicalactone (AAL) (Sigma-Aldrich, 98%), and decane (Sigma-aldrich,>99%) were used as purchased without further purification.

2.2. Computational Studies. Gaussian 0322 software wasused for simulation of the thermodynamic properties of themolecules illustrated in Figure 2. Geometry optimizations andsubsequent frequency calculations were performed using B3LYP/6-311þG(2d,p). Frequency calculations provided estimates forstandard changes of enthalpy, entropy, and Gibbs free energy,whichwere used in estimating thermodynamic properties over therange of temperatures relevant to experimental studies.

2.3. GVL Ring-Opening and Hydrogenation/Decarboxy-

lation Studies. Ring-opening of GVL and subsequent hydro-genation or decarboxylation were studied in an up-flow reactor atatmospheric pressure and temperatures from 498 to 623 K. Thecatalyst (SiO2/Al2O3 or a physical mixture of SiO2/Al2O3 and Pd/C) was loaded into a 1/4” tubular stainless steel reactor. Whennecessary, the catalyst was mixed with crushed granules of fusedsilica (Sigma-Aldrich) to fill the reactor volume. The catalyst bedwas held in place by two plugs of quartzwool, and the reactor wasmounted inside an aluminum block within a well insulatedfurnace (Applied Test Systems). Reactor pressure was controlledwith abackpressure regulator (GOBP-60).Reaction temperaturewas monitored at the reactor wall by a Type K thermocouple(Omega) mounted within the aluminum block and controlled byusing a Series 16 temperature controller (Love Controls). Prior tointroduction of feed, the desired reaction temperature and pres-sure were achieved under flowing inert gas (He). Gas flow to the

reactor was controlled using Brooksmass flow controllers (model5850S). Upon reaching the desired reaction conditions, an aqu-eous solution of GVL (10 wt %) was prepared and fed to thepacked-bed tubular reactor using a HPLC pump (Lab AllianceSeries I). The liquid effluentwas collected for quantitative analysisin a separator (Jerguson Gage and Valve) at ambient temp-erature and analyzed by GC (Shimadzu GC-2010 with FIDdetector). Unknown product peaks were identified using GC-MS (ShimadzuGCMS-QP2010S).COandCO2 in the gas effluentwere quantified using a Shimadzu GC-8A instrument equippedwith a TCD detector, and gas phase alkenes and alkanes werequantified using aVarianGC (Star 3400CX) apparatus equippedwith anFIDdetector. Butene carbon yield is reported as a percentof theoretical yield of butene (C4) from GVL (C5). A continuoushydrogen sweep was adjusted to keep the partial pressure of allspecies in the reactor constant equal to 0.016, 0.164, and 0.820 atmfor GVL, H2, and water, respectively, for all temperatures andweight-hourly space velocities (WHSV, defined as the mass ofGVL per mass of catalyst per hour). The hydrogen flow waschosen such that the molar ratio of H2/GVL was equal to 10.Total carbon balances typically closed to within 10%.

2.4. Decarboxylation Studies of Various Lactone/Acid

Feedstocks. The production of alkenes from aqueous solutionsof alcohols, organic acids, and lactones through acid catalysis wasinvestigated in a tubular, fixed bed catalytic reactor identical tothat described above. Commercial SiO2/Al2O3 was used for eachof the experiments, and it was conditioned by calcination (4 h at723K in flowing air at 50 cm3(STP)/min) prior to the introductionof a new feed. In the case of highly water-soluble species (GVL,DVL, ECL,GCL, BEA, acrylic acid, propanoic acid), the feedwasintroduced to the reactor as a 20 wt % solution using an HPLCpump (Lab Alliance, Series I). For sparingly soluble reagents(organic acids, AAL, DHP), the organic feed was delivered usinga syringe pump (Harvard Apparatus) and diluted to 20 wt % bycofeeding deionized water using a HPLC pump (Lab Alliance,Series I). Because hexenoic acid is a solid at room temperature, itwas dissolved in decane and introduced using a syringe pump(Harvard Apparatus) and diluted as described above. The value ofWHSVfor these experiments is definedas themassof feedmoleculeper mass of catalyst per hour. The liquid effluent was collected forquantitative analysis in a separator (Jerguson Gage and Valve) atambient temperature and analyzed by GC (Shimadzu GC-2010with FID detector). Unknown product peaks were identified usingGC-MS (Shimadzu GCMS-QP2010S). Gas phase products werepurged from the separator by flowing He at 50 cm3 (STP) min-1

using a mass flow controller (Brooks model 5850). CO and CO2 in

Figure 2. Reaction pathways for ring-opening of γ-valerolactone to pentenoic acids and subsequent decarboxylation to butene, orhydrogenation to pentanoic acid. ΔG� and ΔH� (in brackets) are included in kJ/mol for each reaction illustrated at standard conditions.

(22) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.;Cheeseman, J. R.; Montgomery, J. A., Jr.; Vreven, T.; Kudin, K. N.; Burant, J. C.;Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.;Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.;Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.;Kitao, O.; Nakai, H.; Klene,M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.;Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev,O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.;Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski,V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.;Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul,A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz,P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng,C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M.W.; Johnson, B.; Chen,W.;Wong,M.W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision C.02; Gaussian, Inc.:Wallingford, CT, 2004.

16294 DOI: 10.1021/la101424a Langmuir 2010, 26(21), 16291–16298

Article Bond et al.

the gas effluent were quantified using a Shimadzu GC-8A instru-ment equipped with a TCD detector, and gas phase alkenesand alkanes were quantified using a Varian GC (Star 3400 CX)apparatus equipped with an FID detector.

3. Results and Discussion

3.1. Ring-Opening Thermodynamics. The initial step inGVL conversion is acid-catalyzed ring-opening to form an iso-meric mixture of pentenoic acids, the latter of which can behydrogenated to form pentanoic acid (metal catalyst) or undergodecarboxylation to form butene and CO2 (acid catalyst). Therelated pathways for the production of pentanoic acid and buteneare shown in Figure 2. Previous studies using a SiO2/Al2O3

catalyst and aqueous solutions of GVL (30-60 wt %) showedhigh conversions of GVL at 648 K and pressures from 1 to 35 barat space velocities near 1 h-1;21 however, low selectivities todecarboxylation products were observed. The primary productsobserved were 2-, 3-, and 4-pentenoic acids. At low spacevelocities (0.18 h-1), nearly quantitiative yields of butene/CO2

were observed without substantial byproduct formation. Theseresults indicate that pentenoic acid is an intermediate in thedecarboxylation reaction (and thus also an appropriate decar-boxylation feedstock), or that it is produced in parallel withbutene from GVL. Additionally, initial studies suggest that theinterconversion between GVL and pentenoic acid isomers isreversible andoccurs rapidly relative to the rate of decarboxylation.

The pathways for which we have carried out thermodynamiccalculations are shown inFigure 2. Computational calculations atstandard conditions indicate that thermodynamics favor theformation of GVL over PEA isomers. The enthalpy change ofreaction for the ring-opening of GVL to 4-pentenoic acid wasestimated to be 36 kJ/mol at 298 K, which compares favorablywith a previously reported value of 38-40 kJ/mol.23 Calculatedenthalpy changes associated with the production of other pente-noic acid isomers are lower than the estimate included for4-pentenoic acid. The enthalpy changes for the production of2-trans, 3-trans, 2-cis, and 3-cis pentenoic acids are estimatedto be 17, 25, 26, 32 kJ/mol, respectively. Similarly, changes inGibbs free energy at 298 K are estimated to be 25, 8, 14, 16, and19 kJ/mol for 4-, 2-trans, 3-trans, 2-cis, and 3-cis pentenoic acids,respectively. At these conditions, the formation of pentenoic acidthrough opening the GVL ring is endothermic and entropicallyfavored; thus, it is expected that the ratio of pentenoic acid toGVL will increase with reaction temperature. Experimentalresults showed that the concentration of the thermodynamicallymost stable isomer, 2-trans pentenoic acid, was in all cases thehighest, while the concentration of the least stable isomer,4-pentenoic acid, was always the lowest. This isomer distributiondiffers from prior results reported for a similar reaction, theproduction ofmethyl pentanoate fromGVL. Lange et al.18 foundthat the distribution of methyl pentenoates was 25-35% pent-4-enoate, 65-75% cis/trans pent-3-enoates, and only 1-5% cis/tras pent-2-enoates. In this case, however, pentenoic acid reactsrapidly with the alcohol and is removed by distillation, and thus,the concentration predicted by thermodynamic equilibrium is notachieved. Thus, in the system we have studied, it is likely that theless stable 4, 3-cis, and 3-trans pentenoic acid isomers are producedby GVL ring-opening and subsequently isomerize in the presenceof acid sites to form the more stable 2-cis and 2-trans isomers.

Using similar estimations for the decarboxylation of pentenoicacids, at standard conditions and considering an average distribu-

tion of isomers, the production of butene and CO2 is predicted toproceed with changes in enthalpy, entropy, andGibbs free energyof-49 kJ/mol, 148 J/mol/K, and-93 kJ/mol, respectively. Thus,the conversion of GVL to butene and CO2 proceeds with overallchanges in enthalpy and Gibbs free energy equal to 22 and-77 kJ/mol, respectively. Because the decarboxylation leads toa large increase in entropy, the production of butene and CO2

from pentenoic acid or GVL is irreversible and favored under allrelevant reaction conditions.

The hydrogenation of pentenoic acid to produce pentanoicacid is exothermic (-111 kJ/mol) and results in a net decrease inentropy (-120 J/mol/K). At 298 K, this reaction leads to a Gibbsfree energy change of-75 kJ/mol and is favorable at all relevantconditions. The overall changes in enthalpy andGibbs free energy

Figure 3. Ratio of outlet pentenoic acid to γ-valerolactone atincreasing WHSV on SiO2/Al2O at different temperatures. (9)623 K; (O) 598 K; (2) 573 K; (b) 548 K); (0) 523 K. Solid lines areincluded to highlight the trend observed in data.

Figure 4. Rate of production of CO2 at increasing WHSV onSiO2/Al2O3 at different temperatures. (9) 623 K; (O) 598 K;(2) 573 K; (b) 548 K; (0) 523 K; (1) 498 K. The thick solid linecorresponds to the inlet flow of γ-valerolactone (μmol min-1

gcat-1). Solid lines are included to highlight the trend observed in

data.

(23) Emel’yanenko, V. N.; Verevkin, S. P.; Burakova, E. N.; Roganov, G. N.;Georgieva, M. K. Russ. J. Phys. Chem. A 2008, 82, 1521–1526.

DOI: 10.1021/la101424a 16295Langmuir 2010, 26(21), 16291–16298

Bond et al. Article

for the production of pentanoic acid from GVL are estimated tobe -84 and -59 kJ/mol, respectively.3.2. Ring-Opening and Decarboxylation of GVL. Experi-

ments to study the ring-opening of GVL were carried out in afixed bed flow reactor over SiO2/Al2O3. The relative outletconcentrations of PEA toGVL at different temperatures between523 and 623 K are plotted against WHSV in Figure 3. Weobserved that, as the temperature increases, the PEA/GVL ratioincreases from an average value of 0.13 at 523K to 0.88 at 623 K,a result which agrees with trends expected based on computa-tional estimates. For each temperature studied, we observe asimilar trend in the ratio of PEA to GVL. Over a range of low tomoderate space velocities, the ratio observed is constant, whichcorresponds to an equilibrium controlled reaction. At high spacevelocities, the observed ratio begins to decrease with increasingWHSV. Most likely, this regime corresponds to a range of spacevelocities in which the rate of ring-opening is slow relative to theinlet feed rate. Thus, pentenoic acid is not produced quicklyenough to achieve equilibrium. At low space velocities and hightemperatures (623 K), the pentenoic acid to GVL ratio (data notincluded in the figure) is lower than the predicted value, which canbe attributed to the onset of decarboxylation. Figure 4 shows therate of CO2 production versus WHSV at the same conditionsas Figure 3. At 498 K, the average rate of CO2 production was0.07 μmol min-1 gcat

-1, while at 623 K and high WHSV the rateof CO2 production was 95 μmol min-1 gcat

-1. The results inFigures 3 and 4 indicate, that under typical operating conditions,the interconversion of GVL and pentenoic acid isomers is quasi-equilibrated. Thus, for a given inlet concentration and tempera-ture, the partial pressures of GVL and pentenoic acid will remainin approximately the same ratio. The decarboxylation step iskinetically controlled and, as thepartial pressure of pentenoic acidis roughly constant, the production rate of CO2 does not reveal aneffect of space velocity. At low space velocities and 623K, the rateof CO2 production is lower because it is controlled by the rate atwhich GVL is introduced to the reactor and not by the equilib-rium partial pressure of pentenoic acid.3.3. Study of the Reactivity of Different Feedstocks.

Experiments similar to those described in the previous sectionwere carried out using a physicalmixture of Pd/C and SiO2/Al2O3

and a H2 cofeed. In these experiments, it was observed that therate of CO2 production decreases by 2 orders of magnitude uponthe introduction of a metal catalyst. At these conditions, pente-noic acid is quantitatively converted topentanoic acid, whichdoesnot readily undergo decarboxylation. Therefore, GVL decarboxy-lation likely proceeds through alkene acid intermediates. It hasbeen previously reported that the decarboxylation of primaryaliphatic acids (e.g., heptanoic acid) using a supportedNi catalystis more difficult than the decarboxylation of tertiary aliphatic andaromatic acids.24 It is thus apparent that the presence of CdCfunctionality greatly decreases the barrier for decarboxylation oforganic acids. To determine the parameters governing facile

decarboxylation, we have considered the catalytic activity fordecarboxylation of other analogous lactones such as δ-valerolac-tone (DVL), ε-caprolactone (ECL), and γ-caprolactone (GCL) aswell as various alkene acids such as acrylic acid, butenoic acid(BEA), pentenoic acid (PEA), and hexenoic acid (HEA). Addi-tionally, based on the initial results of GVL decarboxylation, wehave measured the catalytic activity of broadly related lactonesthat are ostensibly appropriate for decarboxylation such asR-angeli-calactone (AAL) and 5,6-dihydro-2H-pyran-2-one (DHP).3.4. Saturated Lactones. The structural attributes that we

have studied in various lactone feedstocks are (i) the length of thecarbon chain in the lactone and (ii) the size of the ring comprisingthe lactone. Accordingly, we have studied analogous 5- and6- carbon lactones to determine if the length of the carbon chainhas a significant effect on the reactivity of the lactone or the rate ofdecarboxylation. Experimental conditions and production ratesof observed products are summarized in Table 1, while thedifferent structures are illustrated in Figure 5. The results giveninTable 1 suggest that ring size contributes significantly to lactonereactivity. γ-lactones (5-member rings) appear to be the morestable than δ- (6-member rings) and ε- (7-member rings) lactones.We observe that the rate of ring-opening for δ-valerolactone(DVL) is significantly higher (28 μmol min-1 gcat

-1) than that ofeither γ-valerolactone (GVL) (20 μmol min-1 gcat

-1) or γ-capro-lactone (GCL) (19 μmol min-1 gcat

-1). The rate of cleavage ofε-caprolactone (ECL) is comparable to that of δ-valerolactone(DVL). (In fact, the rates of ring-opening for DVL and ECL aresufficiently fast that these lactones undergo complete conversionat the conditions of this study.) We also observe that 40% and23% of δ-valerolactone and ε-caprolactone react to form γ-lac-tones, suggesting that larger, less stable lactones undergo acid-catalyzed ring-opening to form an alkene acid, which can iso-merize and subsequently close, forming themore stable γ-lactone.A significant difference in selectivity toward the formation ofolefins is observed between DVL and GVL, with 25% of DVLconsumed to produce the olefin product, in comparison to 46%ofGVL converted to olefin at the same conditions. Importantly,however, the product distribution in the conversion of DVL islimited exclusively to alkene acids andmore stable lactones, whichwill both ultimately undergo decarboxylation to form olefins.Thus, DVL is consumed by ring-opening to formpentenoic acids,

Table 1. Reaction rates and product selectivities for various lactones (20 wt % in Water) at 648 K, 1 bar over SiO2/Al2O3

product selectivity (%)

feedfeed rate

(μmol min-1 g-1)conversion rate(μmol min-1 g-1)

decarboxylation rate(μmol min-1 g-1) olefin alkene acid γ-lactone other

GVL 32 20 9 46 54 <1DVL 30 28 7 25 35 40 <1GCL 26 19 5 26 38 36ECL 28 28 5 18 20 23 37

Figure 5. Chemical structure of the lactones used in this study.

(24) Maier, W. F.; Roth, W.; Thies, I.; Schleyer, P. V. Chem. Ber./Recl. 1982,115, 808–812.

16296 DOI: 10.1021/la101424a Langmuir 2010, 26(21), 16291–16298

Article Bond et al.

which rapidly react to form GVL before they can undergodecarboxylation. Alternatively, it is possible that the directdecarboxylation of GVL is possible, whereas DVL cannot under-go direct decarboxylation (i.e., without forming the pentenoicacid intermediate).

The above observations are supported by prior studies regard-ing the chemistry of lactones. For example, δ-valerolactone willreadily polymerize while the similarly nonsubstituted 4-carbonγ-butyrolactone will not. This difference indicates that the num-ber of C atoms of the lactone ring has a substantial influence onthe reactivity of lactones,25 and these authors have attributed trendsin activity to the extent of distortion of bond and tetrahedral anglesvicinal to the ester moiety. γ-Butyrolactone (5-member ring) showsless geometric distortion than DVL (6-member ring); consequently,it has less ring strain and is a more stable molecule. Generalizingthese results to the 5- and 6-carbon lactones studied here, it would beexpected that δ- and ε-lactones would be more susceptible to ring-opening than γ-lactones, as observed experimentally in this study.When examining the results of larger, 6-carbon lactones, we observethat the rate of conversion by ring-opening does not change signifi-cantly from GVL to GCL, indicating that they are equally suscep-tible to cleavage by acid hydrolysis despite a difference in chainlength. This result supports the previous claim that the primarydriver for lactone reactivity (toward ring-opening) is the degree ofring strain and lactone instability. Interestingly, although chainlength has little effect on lactone reactivity, we observe a significantloss of selectivity to desired productswhen testing 6-carbon lactones.In the case of GCL and ECL, we observe selectivities of 36% and37%, respectively, to byproducts, primarily cyclohexenone andmethylcyclohexenone which form in approximately equimolarquantities. The increased rate of byproduct formation could accountfor the lower rates of olefin production from larger (6-carbon)lactones when compared to 5-carbon lactones.3.5. Organic Acids. We have outlined above that the de-

carboxylation of GVL can occur through the production ofpentenoic acid intermediates, and that there is a rapid intercon-version between GVL and pentenoic acid. Accordingly, we haveinvestigated the behavior of various organic acids under condi-tions appropriate for decarboxylation. The results are summari-zed in Table 2. In this study, we have considered differences inmolecular structure such as the position of the double bond in thestarting material, the presence of a CdC double bond versus asaturated acid, and the chain length of the acid. The structure ofeach acid studied is illustrated in Figure 6. First, we have assessedthe activity of propanoic and pentanoic acid to discern whetherthe production of alkanes and CO2 is achievable from saturatedorganic acids. With each of these probe molecules, decarboxyla-tionwas not observed to an appreciable extent, which is in accordwith previously discussed results regarding the production ofpentanoic acid from GVL. In contrast, significant quantities of

olefins and CO2 were measured from unsaturated acids. We arethus able to conclude that the barrier to decarboxylation is greatlyreduced by the presence of a CdC double bond. Examining theresults reported in Table 2, we can conclude, based on the resultsobtained for the various isomers of pentenoic acid, that the initialposition of the double bond in the carbon chain has little effect oneither the reactivity of the acid tested or the selectivity toward theproduction of butene. Given the relative ease of alkene isomeriza-tion in the presence of solid acid catalysts,26,27 it is likely that theisomers of pentenoic acid rapidly equilibrate such that there is littledifference in any of the pentenoic acids under reaction conditions.The length of the carbon chain for various organic acids has amore significant effect. In the case of acrylic acid, we observe aninsignificant extent of decarboxylation (<1 μmol min-1 gcat

-1).Most likely, ethylene production is not observed as, in acrylic acid,the CdC double bond is necessarily R- to the acid, and itsposition does not allow for the formation of a stable, protonatedintermediate. In the case of acrylic acid, protonation of theR-carbon would result in the formation of a primary carbeniumion which is not energetically favorable. Beginning with bute-noic acid, we observed significant decarboxylation for all of thefeeds studied, suggesting that, for decarboxylation to occur, thepossibility must exist for a CdC double bond which can beprotonated to form a secondary carbenium ion β- to the acid.We would expect that such an intermediate could be formedfrom either 2- or 3-alkene acids having no fewer than fourcarbon atoms. Additionally, as the length of the unsaturatedacid increases, we observe a decrease in the production rate of(and selectivity to) olefins. At 648 K, we observe that butenoicacid undergoes decarboxylation at a rate of 21 μmol min-1 gcat

-1 .This rate decreases to 12-13 μmol min-1 gcat

-1 for pentenoicacid isomers and decreases again to 6 μmol min-1 gcat

-1 for hexe-noic acid. In addition to a decreased rate of decarboxylation,

Table 2. Reaction Rates and Product Selectivities for Various Organic Acids (20 wt % in Water) at 648 K, 1 bar over SiO2/Al2O3

product selectivity (%)

feedfeed rate

(μmol min-1 g-1)conversion rate(μmol min-1 g-1)

decarboxylation rate(μmol min-1 g-1) olefin γ-lactone other

acrylic acid 45 1 <1 11 893BEA 37 21 21 >99 0 02PEA 30 21 13 61 39 03PEA 30 20 13 62 38 04PEA 28 19 12 68 31 02HEA 26 14 6 42 38 17

Figure 6. Chemical structure of the acids used in this study.

(25) Houk, K. N.; Jabbari, A.; Hall, H. K.; Aleman, C. J. Org. Chem. 2008, 73,2674–2678.

(26) Quann, R. J.; Green, L. A.; Tabak, S. A.; Krambeck, F. J. Ind. Eng. Chem.Res. 1988, 27, 565–570.

(27) Zhou, Y. B.; Woo, L. K.; Angelici, R. J.Appl. Catal., A 2007, 333, 238–244.

DOI: 10.1021/la101424a 16297Langmuir 2010, 26(21), 16291–16298

Bond et al. Article

we observe a loss of selectivity, as larger acids allow for theformation of a broader range of byproduct. For butenoic acid,we observe complete selectivity to propylene. With pentenoicacids, product selectivity is distributed between butene (60-70%) and γ-valerolactone (30-40%). From hexenoic acid,more products are observed. In addition to pentene (42%selectivity), we observe the formation of γ-caprolactone (38%)and various unsaturated, cyclic C6 ketones (17%). As the chainlength increases, it becomes increasingly likely that, in thepresence of an acid catalyst and water, the organic acid willundergo ring-closure to form the lactone, and we observesignificant quantities of themost stable γ-lactone when starting withboth 5- and 6-carbon organic acids. In addition, as the chain lengthincreases from5 to 6 carbon atoms,we begin to see the occurrence ofunsaturated, cyclic C6 ketones (as in the case reported for γ- andε-caprolactones). Importantly, however, the lactones formed from5- and 6-carbon acids exist in equilibrium with the acids. Thus, asthe acid is consumed by the irreversible decarboxylation reaction(i.e., at lower space velocities), the lactones will undergo ring-opening and decarboxylation. Accordingly, we expect that quanti-tative yields of butene are possible from pentenoic acid; however, itappears that byproduct formation will prevent high olefin yields inthe decarboxylation of 6-carbon alkene acids. In addition, asdiscussed below, the lower decarboxylation rate observed for alkeneacids of increasing chain length is also caused by an increasednumber of isomers related to the location of the CdC double bond,decreasing the probability of having the CdC double bond at thehighly reactive 2- or 3-position.3.6. Unsaturated Lactones. In addition to unsaturated

organic acids and lactones, we studied the behavior of relatedfeedstocks based on their chemical similarity to molecules forwhich acid catalyzed decarboxylation had been previously de-monstrated (GVL, alkene acids). In this study, we havemeasuredthe rates of conversion of R-angelicalactone (AAL, a GVLprecursor) and 5,6-dihydro-2H-pyran-2-one (DHP, pyran analo-gue of AAL). The structures of AAL and DHP are illustrated inFigure 5, and catalytic activity toward decarboxylation is sum-marized in Table 3. In the case of DHP, trace amounts of carbonwere detected in the aqueous product; however, even after severalhours on stream, no organic product had accumulated. Theentirety ofmeasurable productswas detected asMVK, butadiene,CO, and CO2 in the gas phase, and we were only able to accountfor 60% of the carbon introduced to the system in the products.Unsaturated lactones have been cited previously to serve as cokeprecursors,28 and it is possible that the loss of carbon in the case ofDHP occurred through polymerization side reactions that re-sulted in coke formation within the catalyst bed. The carbonbalance for AAL closed to within 10%. The two lactones followthe previously observed trend of general reactivity, with DHPbeing completely converted (31 μmolmin-1 gcat

-1) and the rate ofAAL conversion being significantly lower (23 μmolmin-1 gcat

-1).Interestingly, the two unsaturated lactones, which differ only inring size as illustrated in Figure 5, show remarkably different

selectivities toward decarboxylation (suggesting the production ofbutadiene) anddecarbonylation (suggesting the production ofmethyl-vinylketone). It appears thatDHP primarily underegoes decarboxyla-tion, whereas AAL primarily undergoes decarbonylation.3.7. Mechanism of Decarboxylation.We have observed in

this study that lactones, such as GVL and ECL, can undergodecarboxylation to produce olefins and CO2. In addition, un-saturated acids undergo decarboxylation to produce the corre-sponding olefin and CO2, while saturated organic acids, such aspentanoic orpropanoic acid, donot undergo decarboxylation to ameasurable extent. These results suggest that substrates appro-priate for decarboxylation should possess a CdC double bond(unsaturated acids) or exist as cyclic esters (lactones) that can beopened through acid catalysis to produce unsaturated acids. Themechanism that we propose for decarboxylation begins withprotonation of the unsaturated acid/cyclic esters as illustrated inFigure 7. A series of subsequent proton transfer steps follow theinitial protonation and ultimately result in decarboxylation toproduce CO2 and the corresponding alkene. The appearance ofGVL as a product in the study of pentenoic acids supports theclaim of reversibility for the ring-opening of GVL. Because therates of olefin and CO2 production from organic acids are higherthan those from the corresponding lactones, we speculate thatpathway 1 is the lower energy route and that GVL decarboxyla-tion proceeds more readily through an unsaturated acid inter-mediate than directly from the lactone as shown in pathway 2.Additionally, although the position of the CdC double bond inthe probe molecule did not play an important role in the finalreactivity (attributed to rapid pentenoic acid isomerization), it islikely that the preferred configuration for decarboxylation in-volves the location of theCdCdouble bond in the 2- or 3-positionof the en-acid, which could be protonated in the R- or γ- positionsrespectively to form a secondary β-carbenium ion. The lowerdecarboxylation rate observed for C6 acids and lactones canpotentially be attributed to an increased number of isomers,

Table 3. Reaction Rates and Product Selectivities for 5,6-Dihydro-2H-pyran-2-one (DHP) and r-Angelicalactone (AAL) (20 wt % in Water)at 648 K, 1 bar over SiO2/Al2O3

product selectivity (%)

feedfeed rate

(μmol min-1 g-1)conversion rate(μmol min-1 g-1)

decarboxylation rate(μmol min-1 g-1)

decarbonylation rate(μmol min-1 g-1) CO2 CO BALa LAb other

DHP 31 31 16 2 52 6 0 0 41AAL 30 23 1 5 5 22 17 53 2

abeta-angelicalactone (BAL). bLevulinic acid (LA).

Figure 7. Proposed mechanism for decarboxylation of 3-pente-noic acid to 1-butene and CO2 (pathway 1) and ring-opening anddirect decarboxylation of γ-valerolactone to 1-butene and CO2

(pathway 2).

(28) World Patent WO/2008/142127, 2008.

16298 DOI: 10.1021/la101424a Langmuir 2010, 26(21), 16291–16298

Article Bond et al.

decreasing the probability of having the CdC double bond at the2- or 3-position.

4. Conclusions

We have examined in this study the thermodynamics for ring-opening of GVL as well as subsequent reactions such as de-carboxylation and hydrogenation. Experimental results obtainedfor the ring-opening of GVL suggest that rapid interconversionexists between the lactone and pentenoic acids. The production ofeither pentanoic acid or butene/CO2 from GVL is thermodyna-mically favorable and essentially irreversible under our reactionconditions. We have compared the decarboxylation activity oflactones and organic acids of varying chain length. In general, therate of alkene acid decarboxylationwas higher than that observedfor analogous lactones, suggesting either that lactone decarboxy-lation proceeds through an alkene acid intermediate, or thatalkene acids produced by ring-opening of the lactone provide alower energy, parallel pathway for the production of olefins andCO2. The observation of a significant quantity of lactones in theproduct distribution of experiments beginning with pentenoicacids supports the reversibility of GVL ring-opening and rein-forces the involvement of pentenoic acid in the mechanism ofGVL decarboxylation.We have found that lactone ring size (γ, δ,ε) has a significant effect on the stability of the lactone, with theγ-lactone (5member ring) being themost favorable configurationand larger 6- and 7-member rings (δ- and ε-lactones) being lessstable and more susceptible to ring-opening. In general, smallerlactones and acids appear to offer the highest rates of decarboxy-

lation and selectivity to decarboxylation products; however, wehave found that at least four carbon atoms are required forappreciable rates of decarboxylation. The loss of selectivityobserved with larger lactones can be attributed to increasedincidence of byproduct formation, particularly with 6 carbonspecies. We expect that quantitative olefin yields can beachieved from 4- and 5-carbon acids and lactones, as the onlynonalkene products observed are themselves appropriate dec-arboxylation feedstocks. In the case of 6-carbon feeds, theoverall yield to alkenes is limited by the formation of stablebyproducts such as unsaturated, 6-carbon, cyclic ketones.Finally, we have observed that unsaturated lactones do notundergo decarboxylation as readily as saturated lactones;instead, decarbonylation pathways to produce unsaturatedketones become more prevalent.

Acknowledgment. This work was supported through fundingfrom the Defense Advanced Research Projects Agency (Surf-cat:Catalysts for Production of JP-8 range molecules from Ligno-cellulosic Biomass). The views, opinions, and/or findingscontained in this article/presentation are those of the author/presenter and should not be interpreted as representing theofficial views or policies, either expressed or implied, of theDefense Advanced Research Projects Agency or the Departmentof Defense. In addition, this work was supported in part by theU.S. Department of Energy Office of Basic Energy Sciences. Oneof us (J.A.D.) also thanks Professor Gabor Somorjai for con-tinued inspiration and encouragement over the years.