Thermal Decomposition Mechanisms of the Methoxyphenols

Thomas K. OrmondJoshua H. BarabanJessica P. PorterfieldAdam M. .... Adam M. Scheer , Oliver Welz , Darryl Y. Sasaki , David L. Osborn , and Craig A. T...
0 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/JPCA

Thermal Decomposition Mechanisms of the Methoxyphenols: Formation of Phenol, Cyclopentadienone, Vinylacetylene, and Acetylene Adam M. Scheer,†,‡ Calvin Mukarakate,† David J. Robichaud,† Mark R. Nimlos,† and G. Barney Ellison*,‡ † ‡

National Renewable Energy Laboratory 1617 Cole Blvd Golden, Colorado 80401-3393, United States Department of Chemistry and Biochemistry University of Colorado-Boulder Boulder, Colorado 80309-0215, United States ABSTRACT: The pyrolyses of the guaiacols or methoxyphenols (o-, m-, and p-HOC6H4OCH3) have been studied using a heated SiC microtubular (μ-tubular) reactor. The decomposition products are detected by both photoionization time-offlight mass spectroscopy (PIMS) and matrix isolation infrared spectroscopy (IR). Gas exiting the heated SiC μ-tubular reactor is subject to a free expansion after a residence time of approximately 50100 μs. The PIMS reveals that, for all three guaiacols, the initial decomposition step is loss of methyl radical: HOC6H4OCH3 f HOC6H4O + CH3. Decarbonylation of the HOC6H4O radical produces the hydroxycyclopentadienyl radical, C5H4OH. As the temperature of the μ-tubular reactor is raised to 1275 K, the C5H4OH radical loses a H atom to produce cyclopentadienone, C5H4dO. Loss of CO from cyclopentadienone leads to the final products, acetylene and vinylacetylene: C5H4dO f [CO + 2 HCtCH] or [CO + HCtCCHdCH2]. The formation of C5H4dO, HCCH, and CH2CHCCH is confirmed with IR spectroscopy. In separate studies of the (1 + 1) resonance-enhanced multiphoton ionization (REMPI) spectra, ~ ] + λ275.1nm f C6H5OH+. From we observe the presence of C6H5OH in the molecular beam: C6H5OH + λ275.1 nm f [C6H5OH A the REMPI and PIMS signals and previous work on methoxybenzene, we suggest that phenol results from a radical/radical reaction: CH3 + C5H4OH f [CH3C5H4OH]* f C6H5OH + 2H.

’ INTRODUCTION Molecular weight growth and polycyclic aromatic hydrocarbon (PAH) formation is emerging as a topic of importance in diverse fields from interstellar chemistry13 to combustion and gasification processes. Polycyclic aromatic hydrocarbon formation in the gasification of biomass for liquid fuels leads to tars, soot, and other macroscale products that can clog and dirty instrumentation, necessitating expensive cleanup procedures.4 Roughly a third of biomass is comprised of lignin, a complex biopolymer that consists of a network of cross-linked alkyl aryl ethers.5 The thermal decomposition of lignin and the subsequent recombination mechanisms of lignin pyrolysis products are poorly understood.6 Two of the most important monomers that comprise lignin7,8 are the coniferyl and sinapyl alcohols (1).

biopolymer, studying monomeric model compounds is an essential first step in deducing primary decomposition mechanisms. We have used a heated microtubular (μ-tubular) reactor1013 to study the thermal cracking of the methoxyphenols: ortho-, meta-, and para-guaiacol.

Coniferyl and sinapyl alcohols9 are related to the guaiacols (methoxyphenols) and these, in turn, are derivatives of anisole (methoxybenzene). Due to the complexity of the lignin

Received: July 16, 2011 Revised: September 14, 2011 Published: September 19, 2011

r 2011 American Chemical Society

HO-C6 H4 -OCH3 þ Δ f products

ð2Þ

The μ-tubular reactor can be heated to 1700 K and has a characteristic residence time of 50100 μs. Gases exiting the reactor are subject to a free expansion that cools the beam and quenches all chemistry. Photoionization mass spectrometry (PIMS; λ0 = 118.2 nm or 10.487 eV), (1 + 1) resonanceenhanced multiphoton ionization (REMPI), and matrix isolated infrared absorption (IR) spectroscopy have been used to probe14 the pyrolysis products from (2). Because of the short residence time in the μ-tubular reactor and subsequent cooling of the molecular beam, the reaction products from (2), including radicals and other reactive intermediates, can be directly observed. PIMS is an excellent tool to map the primary decomposition channels and to observe major products of the thermal

13381

dx.doi.org/10.1021/jp2068073 | J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

ARTICLE

cracking process (2). The REMPI method is a sensitive technique to detect trace aromatic compounds such as benzene, phenol, or naphthalene. Matrix IR spectroscopy is a good structural tool and is a helpful complementary technique to the PIMS and REMPI ionization measurements. Our results show a common fragmentation pattern for all three guaiacol isomers. This consistent decomposition scheme is predictive and may allow the mapping of the thermal cracking of larger aryl ethers and lignin. The REMPI spectra reveal a bimolecular reaction between the CH3 and C5H4OH radicals to form hydroxymethylcyclopentadiene, which further decomposes to phenol. CH3 þ HOC5 H4 þ Δ f ½HO-C5 H4 -CH3  f H þ ½HO-C5 H3 -CH3  f H þ HOC6 H5

ð3Þ

Reaction 3 is consistent with the observation15 that pyrolysis of anisole generates methyl radicals and cyclopentadienyl radicals, CH3 + C5H5, which subsequently recombine to produce methylcyclopentadiene. The methylcyclopentadiene is then observed to undergo two hydrogen atom losses and a ring expansion to form benzene. Radical recombination reactions have been shown to produce stable aromatic species that are known to be important PAH precursors.1620

’ EXPERIMENTAL SECTION Heated μ-Tubular Reactor. The μ-tubular reactor is based on the original design of Chen.1013,2123 It contains a springloaded solenoid pulsed valve that controls the back and forth motion of a poppet that opens and closes a 1/4 mm diameter circular aperture. A 3.75 cm long, resistively heated silicon carbide tube is mounted immediately after the aperture. The i.d. of the SiC tube is 1.3 mm, and the wall thickness is approximately 0.35 mm. The SiC tube can be heated to 1700 K, and the temperature is monitored by a type C thermocouple attached to the outer wall of the tube. The wall temperature of the SiC tube can be maintained to (10 K. Because of variations in quality of thermocouple contact with the reactor upon replacement, we report absolute temperatures on the outside of the reactor to (100 K. To prevent radiative heat loss, the SiC tube is shielded by a 1 cm diameter alumina cylinder. The aperture faceplate is heat-sinked to a water-cooled flange to avoid damaging the valve and other internal components. Gases exiting the μ-tubular reactor emerge in an under-expanded jet at roughly 105 Torr. The translational, vibrational, and rotational temperatures drop rapidly within a few reactor diameters and all chemistry ceases. Previous studies have demonstrated that there are no detectable wall reactions.15,24 A more detailed description of the μ-tubular reactor was given previously.14 The dynamics of pyrolysis and gas transport through the SiC μ-tubular reactor is poorly characterized. Preliminary computational fluid dynamics24 simulations estimate that the gas pressure in the μ-tubular reactor is about 10% of the stagnation pressure. Inside the reactor, there is a range of temperatures as the gas is heated by the walls. As a result, there is a residence time distribution and not all molecules see the same “temperaturetime” history. However, as the gas approaches the tube exit, it is fairly uniformly heated such that the centerline temperature is within 100200 K of the wall temperature. From simulations24 of the gas velocity, we estimate the residence time within the heated SiC tube to be roughly 50100 μs.

Low vapor pressure samples are loaded in a small quartz tube and inserted before the solenoid pulsed valve. A sample is heated appropriately to achieve an approximate 1% mixture in a given carrier gas. For the PIMS and REMPI experiments, the samples are backed by a reservoir containing 2000 Torr of He. The valve is operated at 30 Hz, creating a pulsed molecular beam. A skimmer is used to select the forward traveling component of the expansion, creating a well-collimated beam. For the matrix IR experiment, the valve is backed by a reservoir containing 600900 Torr of Ar. Samples that have sufficient room temperature vapor pressure are loaded into a gas reservoir attached by vacuum line to the valve and mixed at approximately 1% with 2000 Torr He (PIMS and REMPI) or 900 Torr Ar (IR). In this work, the sample probe setup was used for all experiments except for vinylacetylene (CH2dCH-CtCH) and acetylene, which have sufficient room temperature vapor pressures. Photoionization Mass Spectrometry (PIMS). The molecular beam enters the ionization region of a reflectron time-of-flight mass spectrometer (Jordan). The ninth harmonic of a Nd:YAG laser (λ0 = 118.2 nm or 10.487 eV) is generated by taking the third harmonic of the Nd:YAG (Spectra Physics) through an Ar/ Xe tripling cell25 and is used to ionize the beam. The 118.2 nm light has sufficient energy to ionize most molecules of interest, but lies below the fragmentation threshold for the compounds of interest in this work. Propylene and NO are used to calibrate the mass spectrometer. Full-width-at-half-maximum peak widths of 0.075 amu are typical in our mass spectra. Resonance-Enhanced Multiphoton Ionization Mass Spectrometry (REMPI). The third harmonic (355 nm) of a Nd:YAG laser (Spectra Physics) is used to drive an optical parametric oscillator (OPO; versascan/MB, GWU) with a range of 500 710 nm. The light emitted by the OPO then enters a frequency doubling stage (GWU) resulting in a final, tunable range of 250355 nm (3.504.97 eV) with an instrumental line width of 6.5 cm1. The UV beam intersects the molecular beam in the ionization region of the mass spectrometer as in the PIMS experiment. If a molecule has a resonance in this region it will be excited by one photon. A second photon of the same energy can subsequently ionize the molecule. Aromatic molecules are often particularly sensitive to REMPI ionization in this region because of strong electronic excitations that result in signature resonance structures. Because the (1 + 1) REMPI scheme involves two photons of the same energy, it is limited to the detection of ions via resonances at energies greater than half the ionization potential of the molecule of interest. For both the PIMS and REMPI experiments, data acquisition and signal averaging are performed using Labview. The individual PIMS mass spectra shown in the figures are composite averages of 200010000 individual mass spectra. Each data point comprising the REMPI wavelength scans are composite averages of 150200 individual mass spectra. Step sizes of 0.02 nm were used. The reported wavelength scans show the averaged signal from one particular value of m/z. Matrix Isolation Infrared Spectroscopy. The output of the nozzle passes through a heat shield aperture plate and is deposited on a CsI window cooled to 20 K with a two-stage closed-cycle helium refrigerator. The infrared spectrum of the sample is measured using a Nicolet Magna 550 FTIR with a mercury/cadmium/telluride (MCT-A) detector. The IR beam passes through a pair of CsI side windows that flank the APD cryostat shroud. The FTIR scans are composite averages of 1000 scans with a resolution of 0.5 cm1. Backgrounds are taken 13382

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

Figure 1. PIMS (118.2 nm) resulting from heating a mixture of approximately 1 Torr o-guaiacol in 2000 Torr He (0.05%) in a μ-tubular reactor. The SiC tube is heated to 375 (bottom trace), 1075, 1375, and 1575 K (top trace). Y-axis scaling is done so the most dominant peak in each spectrum extends the length of the plot.

ARTICLE

Figure 3. PIMS (118.2 nm) resulting from heating a mixture of approximately 1 Torr p-guaiacol in 2000 Torr He (0.05%) in a μ-tubular reactor. The SiC tube is heated to 475 (bottom trace), 1175, 1275, and 1575 K (top trace). Y-axis scaling is done so the most dominant peak in each spectrum extends the length of the plot.

of xylene. To ensure a pure sample for the matrix IR experiment, the vinylacetylene/xylene mixture was introduced to a vacuum tube where it was frozen in a liquid nitrogen bath. The nitrogen bath was then removed and the vapor (roughly 2 Torr) from the thawing sample was taken as the portion to be mixed with Ar and deposited on the CsI window. The acetylene was atomic absorption grade obtained from Airgas and was in an acetone stabilizer. No further purification was performed and no signal from acetone was detected.

’ RESULTS AND DISCUSSION i. Unimolecular Decomposition of Methoxyphenols. The pyrolysis of anisole (C6H5OCH3) has been studied several times15,2632 and has been established as a radical process:

C6 H5 OCH3 þ Δ f CH3 þ ½C6 H5 O f CH3 þ CO þ C5 H5

Figure 2. PIMS (118.2 nm) resulting from heating a mixture of approximately 1 Torr m-guaiacol in 2000 Torr He (0.05%) in a μtubular reactor. The SiC tube is heated to 375 (bottom trace), 1175, 1275, and 1575 K (top trace). Y-axis scaling is done so the most dominant peak in each spectrum extends the length of the plot.

approximately 3 h prior to the sample and are also composite averages of 1000 scans with 0.5 cm1 resolution. Sample Preparation. The o-guaiacol was obtained from Fluka at a stated purity of 98%. The m- and p-guaiacols were obtained from Sigma Aldrich at stated purities of 96 and 99%, respectively. No further purification was performed. Vinylacetylene was obtained from GFS Chemicals and was assayed to be a 51.4% mixture in xylene. The PIMS results on the vinylacetylene mixture, both at room temperature and 1475 K, somewhat surprisingly showed only vinylacetylene (m/z 52) with no trace

ð4Þ

We anticipate15 the thermal decomposition of the guaiacol isomers will be similar to the events of (4). The o-guaiacol PIMS spectra from room temperature to 1575 K are shown in Figure 1. In Figures 2 and 3, we present the analogous results for m- and pguaiacol, respectively. Each isomer follows the same decomposition pattern. In each case, the onset of pyrolysis is observed at approximately 1000 K. The thermochemistry of the methoxyphenols is not wellknown. We expect the bond energies of these aromatic ethers to be similar to those of phenol and anisole. Table 1 shows the result of electronic structure calculations of the bond energies of C6H5OH, C6H5OCH3, and the three isomeric guaiacols. The CBS-QB3 calculations do a good job predicting the bond energies of phenol and anisole, and we believe they will provide a reliable guide for understanding the methoxyphenols. The initial products in the thermal decomposition of the methoxyphenols are consistent with reaction 4. The parent ions, 13383

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

ARTICLE

Table 1. Methoxyphenol Thermochemistry CBS-QB3 Electronic Structure Calculations60 DH298

experimental

(kcal mol1)

DH298 (kcal mol1)

ref

C6H5OH o-HOC6H4OCH3

87 ( 1 87 ( 1

87.3 ( 0.5

61,62

m-HOC6H4OCH3

84 ( 1

p-HOC6H4OCH3

82 ( 1

C6H5OCH3

66 ( 1

63.4 ( 0.6

6264

o-HOC6H4OCH3

58 ( 1

57.5

32

m-HOC6H4OCH3

64 ( 1

65.0

32

p-HOC6H4OCH3

61 ( 1

62.2

32

(o-HO-C6H4OCH3+, m-HO-C6H4OCH3+, p-HO-C6H4 OCH3+) appear at m/z 124. Thermal cleavage of the CH3OC6H4OH bond produces CH3 (m/z 15) and the corresponding hydroxyphenoxy radical, HOC6H4O (m/z 109). More HOC6H4O is observed in the o- and p-guaiacol than in the m-guaiacol. This is likely a reflection of a slightly smaller barrier associated with the HOC6H4O exit channel33 and the slightly larger HOC6H4OCH3 bond energy in m-guaiacol.32 Decarbonylation generates the hydroxycyclopentadienyl radical, HOC5H4 (m/z 81), eq 5. These observations are in agreement with Suran et al.32

In earlier studies15,30 of C6H5OCH3 (eq 4), formation of CH3 and C5H5 radicals was observed. At higher temperatures,15 the C5H5 radical was found to decompose to HCCH + HCCCH2 (propargyl radical). However, Figures 13 suggest that the hydroxycyclopentadienyl radical eliminates H atom to produce cyclopentadienone, m/z 80.

Figures 13 indicate that further heating of the guaiacols to 1575 K generates a species with m/z 52, which we assign as HCtCCHdCH2, vinylacetylene.

Figure 4. Matrix IR absorption spectra from 1350 to 1120 cm1 of the 1275 K pyrolysis products resulting from mixing approximately 1 Torr oguaiacol in 900 Torr Ar (0.1%) (black spectrum, top). A background scan of Ar heated to 1275 K is plotted in green (bottom). Known bands of parent guaiacol are marked by (P) and features assigned to cyclpentadienone are marked by (O). The intense >CdO band of cyclopentadienone appears as the insert from 1720 to 1735 cm1.

The matrix IR spectra in Figures 46 confirm the identity of cyclopentadienone, m/z 80, and vinylacetylene, m/z 52. Infrared spectra were collected for o-methoxyphenol decomposition using the μ-tubular reactor at 375, 1075, 1275, and 1475 K. The initial IR spectrum of o-guaiacol at 375 K shows the characteristic OH stretch of the parent o-CH3OC6H4OH at 3576 cm1, in agreement with that of 3572 cm1, also reported in an argon matrix.34 The 1275 K matrix IR spectra shown in Figures 4 and 5 affirm the presence of cyclopentadienone, C5H4dO. Cyclopentadienone has been prepared, at least transiently, by pyrolysis of its DielsAlder adduct.35 The vibrational spectrum of C5H4dO was first reported 30 years ago36 and established the infrared bands of this antiaromatic ketone through a comparative study of the decomposition of nine different precursors.37,38 No vibrational analysis of the IR spectrum of C5H4dO was attempted. Thermal decomposition of o-phenylene sulfite has been used39 to produce molecular beams of cyclopentadienone in a photoelectron spectrometer in order to measure the ionization energy. Figures 4 and 5 show that strong IR transitions of cyclopentadienone are observed at 628, 820, 1135, and 1332 cm1. These bands closely match those reported by Maier37,38. The diagnostic carbonyl stretch that has been reported36 at 1709 cm1 on a sodium chloride window at 77 K has been observed37,38 as a doublet at 1724 and 1727 cm1 in an Ar matrix at 20 K. We also detect this feature as a strong doublet at 1726 and 1729 cm1 (Figure 4). Figures 5 and 6 show authentic IR spectra for HCCH and HCtCCHdCH2 plotted below the experimental scans for thermal cracking of o-CH3OC6H4OH at 1275 K (red traces at the top). In Figure 6, The bands marked by the arrow (V) at 3301 and 3287 cm1 belong to acetylene, ν3(HCCH), and are the absorptions associated with the well-known40 Darling-Dennison mixing of ν3 and ν2 + ν4 + ν5. The features marked by bullets (b) are assigned to ν1 of HCtCCHdCH2. Authentic transitions due to HCtCCHdCH2 and HCCH are also evident in Figure 5. 13384

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

ARTICLE

Table 2. Gas Phase42 and Matrix IR Peak Positions for HCtCCHdCH2 Vibrational Modes (cm1)a o-HOC6H4OCH3

HCtCCHdCH2 Sym

mode

0

ν1

a

a00

a

local mode

gas phase

HCCCHCH2 str

3330

Ar matrix

1275 K decomp

3332

3332

3328

3328

3319

3319

3314

3315

ν2

HCCCHdCH2 asym str

3116

3120

3124

ν3

HCC(CH)CH2 str

3068

3063

3063

ν4

HCCCHdCH2 sym str

3030

3034

3035

ν5 ν6

HCtCCHCH2 str HCCCHdCH2 str

2111 1599

2104

2102 (1575 K)

ν7

HCCCHdCH2 scissors

1415

1414

ν8

HCC(CH)dCH2 bend

1312

ν9

HCCCHdCH2 rock

1096

1097

ν10

HCCCHCH2 str

874

878

878

ν11

HCtCCHCH2 bend

625

637

637

ν12

HCCCHdCH2 bend

539

543

ν14 ν15

HCC(CH)dCH2 bend HCCCHdCH2 wag

974 927

977 927

ν16

HCCCHdCH2 twist

677

675

675

ν17

HCtCCHCH2 bend

618

615

615

978 927

The values given in the last two columns are for an authentic vinylacetylene sample and o-methoxyphenol pyrolysis at 1275 K (Figures 46).

1

Figure 5. Matrix IR absorption spectra from 1000 to 600 cm of authentic samples of HCtCH (bottom) and HCtCCHdCH2 (middle) and the 1275 K pyrolysis products resulting from mixing approximately 1 Torr o-guaiacol in 900 Torr Ar (0.1%; red spectrum, top). The known bands of HCtCCHdCH2 are marked by (b) and the ν5 feature of HCtCH is marked by (V). A pair of bands assigned to cyclpentadienone are marked by (O).

The intense ν5(HCtCH) is observed at 736 cm1 in the decomposition of o-guaiacol (top of Figure 5). Several other vinylacetylene bands are also observed from o-guaiacol pyrolysis in that spectral region. The infrared spectra of amorphous and crystalline vinylacetylenes41 were recently recorded in the range of 7000400 cm1 and have been established in the gas-phase (Table 2) as well.42 These earlier studies did not show a splitting of ν1 of HCtCCHdCH2.

Figure 6. Matrix IR absorption spectra from 3340 to 3265 cm1 of authentic samples of HCtCH (bottom) and HCtCCHdCH2 (middle) and the 1275 K pyrolysis products resulting from approximately 1 Torr of o-guaiacol in 900 Torr Ar (0.1%; red spectrum, top). The known ν1 bands of HCtCCHdCH2 are marked by (b) and the ν3 features of HCtCH are marked by (V).

To investigate whether the splitting observed in our spectra (Figure 6) is due to a matrix effect, the vinylacetylene matrix was annealed at a series of temperatures from 20 through 50 K at which point the matrix began to rapidly evaporate. No significant changes were observed leading us to conclude that the splitting of ν1 is likely not due to matrix effects. This splitting warrants further investigation. Table 2 gives assignments for vinylacetylene IR modes observed in the gas phase compared to our Ar matrix results for both a vinylacetylene standard as well as guaiacol pyrolysis. 13385

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

ARTICLE

Figure 7. Summary chart of the steps in the thermal cracking of o-guaiacol; PIMS signals are observed at m/z 124, 109, 81, 80 and 52.

predicted to be HCtCCHdCH2 with singlet cyclobutadiene, [c-C4H4], 33 kcal mol1 higher in energy. Rearrangement of singlet cyclobutadiene to vinylacetylene via the carbene,45 1[c-C4H4] f 1 [:CdCHCHdCH2] f HCtCCHdCH2 is predicted to occur with the largest barrier being 54 kcal mol1. Cyclobutadiene was calculated to decompose to two HCtCH molecules with a barrier of 46 kcal mol1; however it is easy to imagine that the 1 • [ CHdCHCHdCH•] diradical would collapse to a pair of acetylenes. At temperatures employed in the μ-tubular reactor, all of these barriers could be readily overcome. We note another possibility for o-methoxyphenol decomposition. It is possible that a concerted reaction in which methane is ejected to yield o-quinone is a minor channel.46 The o-quinone would certainly39 lose a CO to produce cyclopentadienone (reaction 7). The PIMS does not detect the o-quinone product at m/z 108. However, the IR spectrum reveals a possible methane47 peak. The ν4(CH4) is an intense transition and a feature is detected in o-methoxyphenol pyrolysis at 1275 K at 1306 cm1 (Figure 4). 1

Figure 8. PIMS (118.2 nm) resulting from heating a mixture of approximately 1 Torr o-guaiacol in 2000 Torr He (0.05%) in a μ-tubular reactor. The SiC tube is heated to 1075 K (bottom trace); the insert shows a set of weak features at m/z 94, 95, and 96.

In Figure 7 we show the general unimolecular fragmentation pattern of the methoxyphenols. Figure 7 is ambiguous about the exact dynamics of decarbonylation of C5H4dO leading to HCtCH and HCtCCHdCH2. Loss of CO could produce singlet cyclobutadiene, 1[c-C4H4], or the 1[•CHdCHCHdCH•] diradical, or other C4H4 compounds.43,44 Mebel et al.41 have completed a computational study of the C4H4 singlet potential energy surface and find singlet cyclobutadiene, 1[c-C4H4], 38 kcal mol1 below 1 • [ CHdCHCHdCH•]. The most stable C4H4 isomer is

ii. Bimolecular Reactions. Figure 8 contains an expanded trace of Figure 1 showing the PIMS of o-guaiacol heated to 13386

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

Figure 9. Resonance enhanced (1 + 1) multiphoton ionization scan of m/z 94 observed in o-guaiacol pyrolysis at 1375 K recorded with a 0.02 nm step size. The (1 + 1) REMPI scan of m/z 94 for an authentic phenol standard recorded with a 0.02 nm step size.

Figure 10. A pathway for the radical/radical recombination to produce phenol is shown: C5H4OH + CH3 f C6H5OH + 2H. PIMS signals are observed for m/z 94, m/z 95, and m/z 96 (Figure 8). Figure 9 is (1 + 1) REMPI that identifies m/z 94 as phenol.

1075 K. A set of weak features at m/z 94, 95, and 96 are magnified in the in-set. If the temperature of the μ-tubular reactor is further increased, the m/z 94 feature grows relative to m/z 95 and 96. We have used REMPI spectroscopy to identify the species at m/z 94 as phenol, C6H5OH.

ARTICLE

At the top of Figure 9, the REMPI wavelength spectrum is scanned for the m/z 94 species observed in o-methoxyphenol pyrolysis at 1375 K. The bottom trace is a spectrum of a prepared phenol standard from 273283 nm. It is well established48,49 ~ f A ~ transition has its (0,0) feature at that the C6H5OH X 275.1 nm. The inset in Figure 9 describes the (1 + 1) REMPI ~ state process; the OPO accesses a resonance for the C6H5OH A at 275.1 nm (4.507 ( 0.001 eV) and a second 275.1 nm photon ionizes the phenol.50 The excellent agreement between the two spectra in Figure 9 is compelling evidence that the m/z 94 species in the thermal cracking of o-guaiacol in Figures 1 and 8 is phenol. The same pattern of peaks at 96, 95, and 94 is observed in the PIMS of m-guaiacol. The p-guaiacol PIMS spectra did not show this pattern. It is likely that the partial pressure of p-guaiacol in the molecular beam was slightly lower than in the o- and mguaiacol experiments, thus, reducing the bimolecular chemistry. One could imagine a unimolecular decomposition of o-guaiacol to phenol, o-HOC6H4OCH3 f HOC6H5 + CH2dO, but all efforts to detect the product formaldehyde have failed. Figure 10 is a conjecture of the bimolecular pathway to phenol resulting from the recombination of methyl radicals with hydroxycyclopentadienyl radicals. Figure 8 clearly shows the expected sequence at m/z 96, 95 and 94, supporting this interpretation. The PIMS in Figure 8 demonstrates the presence of both CH3 (m/z 15) and HOC5H4 (m/z 81). It is observed that as the temperature of the μ-tubular reactor is increased to 1275 K and higher, the PIMS and REMPI signals at m/z 94 fade away. Figure 1 shows that raising the wall temperature destroys m/z 81, a reactant in the production of phenol. The chemistry proposed in Figure 10 is analogous to the earlier findings of benzene formation in anisole.15 Methyl radicals were shown to recombine with C5H5 yielding methylcyclopentadiene, C5H5CH3. Subsequent loss of two H atoms from C5H5CH3 produced benzene. The kinetics and potential energy surfaces of benzene formation from methylcyclopentadiene decomposition were reported by several groups.16,17,51

’ CONCLUSIONS A summary of the unimolecular decomposition pathways for o-methoxyphenol is shown in Figure 7. Because the resonance stabilization of phenoxy radicals is roughly 1 eV, and the OCH3 bond is the weakest in the molecule,15,52 the thermal decomposition of the guaiacols is initiated by cleavage of the HOC6H4OCH3 bond. The initial fragments of o-guaiacol in Figure 1 are CH3 (m/z 15) and the hydroxyphenoxy radical, o-HO-C6H4O, at m/z 109. Phenoxy radicals easily decarbonylate15,30 and indeed we observe loss of CO to generate the hydroxycyclopentadienyl radical, o-HO-C6H4O f HOC5H4 (m/z 81). Loss of H produces the “partially anti-aromatic” cyclopentadienone, C5H4dO (m/z 80). IR spectra in Figures 4 and 5 confirm the identity of the m/z 80 species as C5H4dO. Cyclopentadienone is a highly reactive molecule, and Figure 1 reveals rapid formation of a species at m/z 52 along with m/z 80 and loss of m/z 81. IR spectra in Figures 5 and 6 confirm the carrier of the m/z 52 signal as HCtCCHdCH2. In addition to vinylacetylene, Figures 5 and 6 reveal the presence of HCtCH. The detection of phenol shown in Figures 8 and 9 reveals the presence of bimolecular reactions in the μ-tubular reactor and supports previous studies in which benzene is produced from anisole pyrolysis.1517,51 13387

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A

ARTICLE

Finally, we make a conjecture about the connection of biomass burning and aerosol formation. As demonstrated in this paper, the thermal cracking of guaiacols produces both HCtCH and HCtCCHdCH2 as final products. It is also established14 that thermal decomposition of furan and furfural produces acetylene and propyne. Laboratory studies of OH/O2 oxidation of 2-butyne in a turbulent flow reactor53 observed formation of biacetyl. Atmospheric chemists have demonstrated5456 that oxidation of acetylene by hydroxyl radical and oxygen produces glyoxal and formic acid. The Master Chemical Mechanism55 recommends: HCCH þ OH=O2 f HCO-CHO ð64%Þ þ HCOOH ð36%Þ

ð8Þ

Guaiacols or alkoxy phenols are very common motifs in lignin.5 We believe that biomass burning (“forest fires”) will release large quantities of acetylene and substituted alkynes into the atmosphere. As the alkynes rise into the atmosphere, oxidation by OH during the day or NO3 at night will generate glyoxal or other complex α-dicarbonyls. It is known57 that α-dicarbonyls are rapidly hydrated in the atmosphere to form gem-diols or tetrols. Furthermore, solar overtone pumping58 has the potential to dissociate α-keto-gem-diols to a pair of carboxylic acids. It is believed59 that gem-diols, tetrols, or carboxylic acids can nucleate secondary organic aerosols. There is a possibility that biomass burning releases large quantities of alkynes into the atmosphere to be oxidized to α-dicarbonyls. Hydration of the α-dicarbonyl generates diols and tetrols. Solar photochemistry of α-hydroxycarbonyls transforms them into carboxylic acids. If this pathway can be demonstrated, this would be a molecular picture connecting biomass burning and secondary organic aerosol (SOA) formation.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This research was supported by the DOE’s National Renewal Energy Laboratory (Contract No. 1544759) and by grants from the National Science Foundation (CHE-0848606) and (CHE1548379). We thank AnGayle Vasiliou and Kimberly Urness for technical support, Prof. Anthony Dean (Colorado School Mines) and Dr. Hans-Heinrich Carstensen (Colorado School Mines) for useful discussions, and Dr. Geoffrey Tyndall (NCAR) for kindly supplying an acetylene sample. We have discussed the atmospheric implications of biomass burning with Prof. Veronica Vaida (Univ. Colorado), Prof. John W. Daily (Univ. Colorado), and Prof. John F. Stanton (Univ. Texas). ’ REFERENCES (1) Cook, D. J.; Schlemmer, S.; Balucani, N.; Wagner, D. R.; Steiner, B.; Saykally, R. J. Nature 1996, 380, 227–229. (2) Snow, T. P.; Le Page, V.; Keheyan, Y.; Bierbaum, V. M. Nature 1998, 391, 259–260. (3) Mebel, A. M.; Kislov, V. V.; Kaiser, R. I. J. Am. Chem. Soc. 2008, 130, 13618–13629. (4) Han, J.; Kim, H. Renewable Sustainable Energy Rev. 2008, 12, 397–416. (5) Reale, S.; Di Tullio, A.; Spreti, N.; De Angelis, F. Mass Spectrom. Rev. 2004, 23, 87–126.

(6) Nowakowski, D. J.; Bridgwater, A. V.; Elliott, D. C.; Meier, D.; de Wild, P. J. Anal. Appl. Pyrolysis 2010, 88, 53–72. (7) Carpita, N.; McCann, M. The Cell Wall. In Biochemistry and Molecular Biology of Plants; Buchanan, B. B., Gruissem, W., Jones, R. L., Eds.; American Society of Plant Physiologists: Rockville, MD, 2000; Ch. 2, pp 52109. (8) Croteau, R.; Kutchan, T. M.; Lewis, N. G. Natural Products (Secondary Metabolites). In Biochemistry and Molecular Biology of Plants; Buchanan, B. B., Gruissem, W., Jones, R. L., Eds.; American Society of Plant Physiologists: Rockville, MD, 2000; Ch. 24, pp 1250 1318. (9) Flood, W. E. The Dictionary of Chemical Names; Littlefield, Adams and Co.: Totowa, NJ, 1967. Lignin is derived from the L. lignum (wood). Guaiacol is the monomethyl ether of catechol and is a constituent of beechwood tar. It is also obtained by the dry distillation of guaiacum resin. The genus Guaiacum comprises a group of trees and shrubs native to the West Indies and tropical America. The name, adapted from Sp. Guayaco, guayacan, is of native Haytian origin (1864). Guaiacol is present in wood smoke, resulting from the pyrolysis of lignin. The compound contributes to the flavor of many compounds, e.g., roasted coffee. Coniferyl alcohol is derived from the botanical name Coniferae, which means “the cone bearers”: L. conus, cone, fero, to bear (1867). Sinapyl alcohol and sinapic acid are derived from L. sinapis (mustard seed; 1857). (10) Kohn, D. W.; Clauberg, H.; Chen, P. Rev. Sci. Instrum. 1992, 63, 4003–4005. (11) Blush, J. A.; Clauberg, H.; Kohn, D. W.; Minsek, D. W.; Zhang, X.; Chen, P. Acc. Chem. Res. 1992, 25, 385–392. (12) Rohrs, H. W.; Wickham-Jones, C. T.; Berry, D.; Ellison, G. B.; Argrow, B. M. Rev. Sci. Instrum. 1995, 66, 2430. (13) Zhang, X.; Friderichsen, A. V.; Nandi, S.; Ellison, G. B.; David, D. E.; McKinnon, J. T.; Lindeman, T. G.; Dayton, D. C.; Nimlos, M. R. Rev. Sci. Instrum. 2003, 74, 3077–3086. (14) Vasiliou, A.; Nimlos, M. R.; Daily, J. W.; Ellison, G. B. J. Phys. Chem. A 2009, 113, 8540–8547. (15) Scheer, A. M.; Mukarakate, C.; Robichaud, D. J.; Ellison, G. B.; Nimlos, M. R. J. Phys. Chem. A 2010, 114, 9043–9056. (16) Dubnikova, F.; Lifshitz, A. J. Phys. Chem. A 2002, 106, 8173– 8183. (17) Lifshitz, A.; Tamburu, C.; Suslensky, A.; Dubnikova, F. Proc. Combust. Inst. 2005, 30, 1039–1047. (18) Shukla, B.; Koshi, M. Phys. Chem. Chem. Phys. 2010, 12, 2427–2437. (19) Tranter, R. S.; Klippenstein, S. J.; Harding, L. B.; Giri, B. R.; Yang, X. L.; Kiefer, J. H. J. Phys. Chem. A 2010, 114, 8240–8261. (20) Shukla, B.; Tsuchiya, K.; Koshi, M. J. Phys. Chem. A 2011, 115, 5284–5293. (21) Chen, P.; Colson, S. D.; Chupka, W. A.; Berson, J. A. J. Phys. Chem. 1986, 90, 2319–2321. (22) Chen, P.; Pallix, J. B.; Chupka, W. A.; Colson, S. D. J. Chem. Phys. 1987, 86, 516–521. (23) Chen, P. Supersonic Jets of Organic Radicals. In Unimolecular and Bimolecular IonMolecule Reaction Dynamics; Ng, C. Y., Baer, T., Powis, I., Eds.; John Wiley: Cambridge, U.K., 1994; Ch. 8, pp 371397. (24) Daily, J. W.; Guan, Q.; Vasiliou, A.; Nimlos, M. R.; Ellison, G. B. Int. J. Chem. Kinet. 2011, to be submitted for publication. The dynamics of pyrolysis and gas transport through the SiC μ-tubular reactor is poorly characterized. Preliminary computational fluid dynamics simulations estimate that the gas pressure in the μ-tubular reactor is about 10% of the stagnation pressure. Along the reactor, there is a range of temperatures within the gas as it is heated by the walls. As a result, not all molecules see the same temperature time history. In reactor language, there is a residence time distribution. However, as the gas approaches the tube exit, it is fairly uniformly heated such that the centerline temperature is within 100200 K of the wall temperature. From simulations of the gas velocity, we estimate the residence time within the heated SiC tube to be roughly 50100 μsec. (25) Lockyer, N. P.; Vickerman, J. C. Laser Chem. 1997, 17, 139– 159. 13388

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389

The Journal of Physical Chemistry A (26) Lin, C. Y.; Lin, M. C. Int. J. Chem. Kinet. 1985, 17, 1025–1028. (27) Lin, C. Y.; Lin, M. C. J. Phys. Chem. 1986, 90, 425–431. (28) Lovell, A. B.; Brezinsky, K.; Glassman, I. Int. J. Chem. Kinet. 1989, 21, 547–560. (29) Brezinsky, K.; Pecullan, M.; Glassman, I. J. Phys. Chem. A 1998, 102, 8614–8619. (30) Friderichsen, A. V.; Shin, E.-J.; Evans, R. J.; Nimlos, M. R.; Dayton, D. C.; Ellison, G. B. Fuel 2001, 80, 1747–1755. (31) Khachatryan, L.; Adounkpe, J.; Dellinger, B. J. Phys. Chem. A 2008, 112, 481–487. (32) Suryan, M. M.; Kafafi, S. A.; Stein, S. E. J. Am. Chem. Soc. 1989, 111, 1423–1429 A very low pressure pyrolysis oven (3001275 K) was used to decompose guaiacols. Products were identified by a quadrupole mass filter operating at 20 eV. RRKM simulations of the decomposition kinetics were used to determine the bond energies (kcal mol1) as DH298(o-HOC6H4O-CH3) = 57.5, DH298(m-HOC6H4O-CH3) = 65, DH298(p-HOC6H4O-CH3) = 62.2. (33) Alsoufi, A.; Altarawneh, M.; Dlugogorski, B. Z.; Kennedy, E. M.; Mackie, J. C. J. Mol. Struct.: THEOCHEM 2010, 958, 106–115. (34) Tylli, H.; Konschin, H. J. Mol. Struct. 1988, 176, 245–251. (35) Depuy, C. H.; Lyons, C. E. J. Am. Chem. Soc. 1960, 82, 631–633. (36) Chapman, O. L.; McIntosh, C. L. J. Chem. Soc., Chem. Commun. 1971, 770–771 Pyrolysis o-phenylene sulfite directly onto a sodium chloride window cooled to 77 K produced a transient species. An intense carbonyl absorption at νCO = 1709 cm1 was assigned to cyclopentadienone. (37) Maier, G.; Franz, L. H.; Hartan, H. G.; Lanz, K.; Reisenauer, H. P. Chem. Ber. 1985, 118, 3196–3204. (38) Maier, G. Pure Appl. Chem. 1986, 58, 95–104 Five fundamental vibrational modes for cycopentadienone isolated in an argon matrix have been reported. They are 1727, 1724, 1332, 1136, 822, and 632 cm1. (39) Koenig, T.; Smith, M.; Snell, W. J. Am. Chem. Soc. 1977, 99, 6663–6667 IE(C5H4dO) = 9.49 ( 0.02 eV (2A2); 2B2 (10.01 eV, nO). (40) Shimanouchi, T. Tables of Vibrational Frequencies, 1972; Consolidated Vol. I, NSRDS-NBS 39. Of the five modes of HCCH, only the asymmetric CH stretch, πuν3, and the asymmetric HCCH bend, πuν5, are IR active. In the gas-phase ν3, HCCH is observed at 3294.9 and 3281.9 cm1 and is split by a DarlingDennison resonance (DDR) of 13 cm1. In an Ar matrix, ν3 shifts to 3302 and 3288 cm1; the DDR is 14 cm1 in the cryogenic matrix. The gas phase value for ν5 is 730.3 cm1 and shifts to 736.8 cm1 in an Ar matrix. (41) Mebel, A. M.; Kislov, V. V.; Kaiser, R. I. J. Chem. Phys. 2006, 125. (42) Tørneng, E.; Nielsen, C. J.; Klaeboe, P.; Hopf, H.; Priebe, H. Spectrochim. Acta, Part A 1980, 36, 975–987. (43) Whitman, D. W.; Carpenter, B. K. J. Am. Chem. Soc. 1980, 102, 4272–4274. (44) Voter, A. F.; Goddard, W. A. J. Am. Chem. Soc. 1986, 108, 2830–2837. (45) Gunion, R. F.; Koppel, H.; Leach, G. W.; Lineberger, W. C. J. Chem. Phys. 1995, 103, 1250–1262. (46) Klein, M. T.; Virk, P. S. Abstr. Pap. Am. Chem. Soc., Div. Fuel Chem. 1980, 25, 80. (47) Shimanouchi, T. Tables of Vibrational Frequencies. Consolidated Volume I; 1972; NSRDS-NBS 39. Methane has four vibrational modes but only the degenerate CH stretch, f2ν3, and the degenerate deformation, f2ν4, are IR active. In the gas-phase ν3CH4 is observed at 3019.9 cm1 and ν4CH4 is found at 1306.2 cm1. These values shift in an Ar matrix to ν3 = 3032 cm1 and ν4 = 1305 cm1. The signal from ν4 is very intense and easy to detect in an cryogenic matrix. (48) Matsen, F. A.; Ginsburg, N.; Robertson, W. W. J. Chem. Phys. 1945, 13, 309–316. (49) Bist, H. D.; Brand, J. C. D.; Williams, D. R. J. Mol. Spectrosc. 1967, 24, 413. (50) Lipert, R. J.; Colson, S. D. J. Phys. Chem. 1990, 94, 2358–2361 IE(C6H5OH) = 8.508 ( 0.001 eV. (51) Lamprecht, A.; Atakan, B.; Kohse-Hoinghaus, K. Proc. Comb. Inst. 2000, 28, 1817–1824.

ARTICLE

(52) Blanksby, S. J.; Ellison, G. B. Acc. Chem. Res. 2003, 36, 255–263 DH298(CH3-OH) = 92.1 ( 0.1 kcal mol1 and DH298(CH3-OCH3) = 83.4 ( 0.7 kcal mol1. (53) Yeung, L. Y.; Pennino, M. J.; Miller, A. M.; Elrod, M. J. J. Phys. Chem. A 2005, 109, 1879–1889. (54) Jenkin, M. E.; Saunders, S. M.; Pilling, M. J. Atmos. Environ. 1997, 31, 81–104. (55) Saunders, S. M.; Jenkin, M. E.; Derwent, R. G.; Pilling, M. J. Atmos. Environ. 1997, 31, 1249–1249. (56) Hatakeyama, S.; Washida, N.; Akimoto, H. J. Phys. Chem. 1986, 90, 173–178. (57) Axson, J. L.; Takahashi, K.; De Haan, D. O.; Vaida, V. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 6687–6692. (58) Vaida, V. J. Phys. Chem. A 2009, 113, 5–18. (59) Claeys, M.; Graham, B.; Vas, G.; Wang, W.; Vermeylen, R.; Pashynska, V.; Cafmeyer, J.; Guyon, P.; Andreae, M. O.; Artaxo, P.; Maenhaut, W. Science 2004, 303, 1173–1176. (60) Montgomery, J. A.; Frisch, M. J.; Ochterski, J. W.; Petersson, G. A. J. Chem. Phys. 1999, 110, 2822–2827 The CBS-QB3 model employs pair natural orbital extrapolations of MP2/6-311+G(2df,2p), MP4(SDQ)/6-31+G(d,p), and CCSD(T)/6-31+G(d0 ) calculations to estimate the UCCSD(T) complete basis set (CBS) limit at the B3LYP/ 6-311G(d,p) optimized geometry, and includes B3LYP/6-311G(d,p) zero-point energies and thermochemical corrections. The average, mean absolute, and rms errors for the G2 test set become 0.20, 0.87, and 1.08 kcal mol1, respectively. (61) Angel, L. A.; Ervin, K. M. J. Phys. Chem. A 2006, 110, 10392–10403 DH298(C6H5O-H) = 361 ( 4 kJ mol1 or 86 ( 1 kcal mol1. (62) Nix, M. G. D.; Devine, A. L.; Cronin, B.; Dixon, R. N.; Ashfold, M. N. R. J. Chem. Phys. 2006, 125 D0(C6H5O-H) = 30015 ( 40 cm1 (85.8 ( 0.1 kcal mol1); DH298(PhO-H) = 87.3 ( 0.5 kcal mol1. (63) Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermochemistry of Organic Compounds, 2nd ed.; Chapman and Hall: New York, 1986. ΔfH298(C6H5OCH3) = 23.0 ( 0.2 kcal mol1. Use of ΔfH298(C6H5O) derived from Nix et al. DH298(C6H5O-H) and ΔfH298(CH3) leads to DH298(C6H5O-CH3) = [(35.05 ( 0.07) + (12.2 ( 0.5)  (16.2 ( 0.2)] = 63.4 ( 0.6 kcal mol1. (64) Ruscic, B.; Boggs, J. E.; Burcat, A.; Csaszar, A. G.; Demaison, J.; Janoschek, R.; Martin, J. M. L.; Morton, M. L.; Rossi, M. J.; Stanton, J. F.; Szalay, P. G.; Westmoreland, P. R.; Zabel, F.; Berces, T. J. Phys. Chem. Ref. Data 2005, 34, 573–656.

’ NOTE ADDED AFTER ASAP PUBLICATION This article was published ASAP on October 21, 2011, with an error in the caption of Figure 10. The correct version was reposted on October 31, 2011.

13389

dx.doi.org/10.1021/jp2068073 |J. Phys. Chem. A 2011, 115, 13381–13389