Pretreatment of - ACS Publications - American Chemical Society

Apr 20, 2011 - Mark Davis,. ‡ and Gary E. Maciel*. ,†. †. Department of Chemistry, Colorado State University, Ft. Collins, Colorado 80523, Unite...
0 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/EF

In situ Study of Dilute H2SO4 Pretreatment of 13C-Enriched Poplar Wood, Using 13C NMR Benjamin Kohn,† Mark Davis,‡ and Gary E. Maciel*,† † ‡

Department of Chemistry, Colorado State University, Ft. Collins, Colorado 80523, United States National Renewable Energy Laboratory, Golden, Colorado 80401-3393, United States

bS Supporting Information ABSTRACT: In situ 13C NMR measurements are reported on 13C-enriched powdered poplar wood that is subjected to pretreatment with 0.5 M sulfuric acid as a function of time and at two temperatures. 13C MAS (magic-angle spinning) spectra were obtained in both the DP (direct polarization) and CP (cross-polarization) modes, the contrasts in this combination yielding valuable qualitative information on the effect of pretreatment on local molecular mobilities. T1 values for 13C and for 1H, as well as TCH and T1FH, were measured at various stages of treatment with 0.5 M H2SO4 for lignin peaks and for cellulose peaks in the 13C NMR spectra, as quantitative indicators of the degree of molecular motion for those two structural entities. The results show that a substantial fraction of the solid/semisolid biomass is converted at elevated temperatures to (a) chemically different and more mobile structures and (b) locally similar structures with enhanced atomic-level mobilities and that some fraction of this “mobilized” biomass does not return to the original level of immobility upon cooling the biomass back to room temperature. Analysis of the T1 results by a rather simple model indicates that, for poplar wood in 0.5% H2SO4, the estimated (“global”) motional correlation time (at the multiatom level), τc, is in the range of about 0.71.5 ns at various stages and temperatures of the treatment.

’ INTRODUCTION The production of ethanol, or potentially other organic products, from biomass is an area of vigorous current research.17 One of the key steps in the overall conversion of biomass into useful products is the pretreatment step,815 in which the cellulose contained in the biomass (lignocellulose) is rendered accessible to agents aimed at polysaccharide conversion, e.g., enzyme-based hydrolysis followed by fermentation. One pretreatment approach for the overall conversion of biomass that enjoys a substantial interest and research activity is treatment with dilute sulfuric acid at elevated temperatures.1631 It is important for the scientific understanding and potential for improvement of any pretreatment process to maximize knowledge of the chemicalphysical state of the major components of biomass, i.e., cellulose, lignin, and hemicellulose. This knowledge would include, most centrally, the chemical structure, local mobility, and entanglement of the major chemical components of the biomass under the pretreatment conditions. Figure 1 shows structures of the main constituents of wood. Cellulose (I) and hemicellulose(II), which together are referred to as holocellulose, consist entirely or largely, respectively, of polysaccharides. Lignin is a macromolecular material that consists of substituted phenylpropane (guaiacyl(III) and syringyl(IV)) units that are connected to each other, and probably to the polysaccharides, primarily hemicellulose, by ether and ester linkages. The structure of hemicellulose is more complex than that of cellulose, involving a variety of monosaccharide units, as well as other moieties such as uronic acids. Knowledge regarding the populations of these kinds of structures and their local mobilities in biomass and interactions between components, and how these structural/dynamical characteristic are affected by r 2011 American Chemical Society

Figure 1. Chemically relevant wood components: (I) cellulose structure; (II) structure of 4-O-methyl-glucoronoxylan, a representative example of a hard wood hemicellulose; (III) guaiacyl structure; (IV) syringyl structure. For species II, R can be H or acetyl.

a pretreatment process, are important as a foundation for a logical scientific approach for optimizing a pretreatment process. Received: January 4, 2011 Revised: April 20, 2011 Published: April 20, 2011 2301

dx.doi.org/10.1021/ef2000213 | Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Nuclear magnetic resonance (NMR) is one of the most powerful and versatile approaches for the elucidation of structure and dynamics. The chemical shift parameter of NMR is extremely sensitive to molecular structure at the atomic level,32 and relaxation times are experimentally measurable parameters that provide an entry into characterizing molecular dynamics.3234 NMR is applicable to both liquid32 and solid3339 samples, as well as the semisolid samples that one encounters in the pretreatment processes. A recent study of poplar wood under pretreatment by dilute sulfuric acid employed natural-abundance 13C NMR measurements of samples that had been heated to temperatures ranging up to 150 C and then cooled back down to room temperature for ex situ NMR measurements.40 That study provided valuable hints regarding the structural/dynamical changes that had occurred during pretreatment. The 13C NMR measurements of that natural-abundance study lacked the signal-to-noise (S/N) ratio that would have permitted in situ measurements at the pretreatment temperatures of interest; prohibitively long signal averaging periods would have been required. Accordingly, the study reported here was undertaken, with a first step of growing 13 C-enriched poplar wood; the dramatically improved S/N permitted detailed in situ studies to be undertaken at temperatures up to 150 C. These studies included measurements of a variety of relaxation parameters, as well as the progress of the pretreatment process as a function of time.

’ EXPERIMENTAL SECTION

Figure 2. 13C CP-MAS spectra of (a) 50 mg 13C enriched wood, (b) 500 mg natural-abundance wood, (c) 40 μs dipolar dephasing result on 13 C enriched wood. Number of acquisitions =1800, contact time =1 ms, pulse delay =1 s, MAS rate =3.5 kHz. * denotes spinning side bands.

’ RESULTS

Samples. Young poplar (Populus angustifolia) shoots were grown for 13

a period of about six months in a chamber filled with CO2 (Cambridge Isotopes, 99.5% 13C), yielding an isotopic enrichment of about 70% (as determined by 13C NMR; see the Supporting Information). Approximately 20 mg of 13C enriched poplar sawdust was flame-sealed in a glass ampule (7.0 mm OD, 5.0 mm ID) that also contained 0.2 mL of 0.5 M H2SO4 (or water in the experiments so identified below). The ampule was placed in a 9.5 mm (OD) magic-angle spinning (MAS) rotor, which could be heated up to 150 C, using a modified Chemagnetics variabletemperature probe stack. NMR Measurements. Aside from multifield T1 measurements identified explicitly below, all 13C NMR experiments were carried out at 50.2 MHz, using a modified Chemagnetics CMX II type spectrometer (now Varian-Agilent, Ft. Collins, CO) and a Chemagnetics CP-MAS probe for 9.5 mm MAS rotors. The MAS speed was typically 3.5 kHz, which largely eliminates problems (e.g., spectral overlaps) with chemical-shift-anisotropy (CSA) sidebands36 (4.5 kHz would be better).41 13C spectra were processed by complex Fourier transformation of the sampled complex data and subsequent apodization with 50 Hz Lorentzian line broadening. The 1H and 13C radio frequency field strengths were each about 50 kHz. 13C chemical shifts are reported relative to liquid tetramethylsilane (TMS) at 0.0 ppm, using hexamethylbenzene as a secondary chemical shift reference, and internally referencing each spectrum to position the C1 saccharide peak at 104.5 ppm. 1 H spinlattice relaxation time (T1H) measurements were made by 13 C-detected saturation-recovery.42,46 13C spinlattice relaxation time (T1C) measurements were made by a common CP-based (Torchia) technique.43 The measurement of rotating-frame spinlattice relaxation times for protons (T1FH) and cross-polarization time constants (TCH) were made via 13C detection by the well-known variable-contact-time method.36,42 Unless specifically stated to the contrary below, all 13C NMR measurements were made on the 13C-enriched poplar wood.

Assignments. Figure 2a shows the

13

C CP-MAS NMR spectrum of C-enriched poplar shavings, together with a spectrum of the corresponding natural-abundance wood (Figure 2b; many more scans; different physical configurations than used in the determination of isotopic enrichment, Supporting Information). One can see that the natural-abundance spectrum is slightly sharper than that based on the 13C-enriched sample; this difference is presumably the manifestation of unresolved JCC couplings in the enriched sample, which are not averaged to zero by MAS and 13C13C dipolar couplings that are not effectively averaged by 3.5 kHz MAS. The 13C MAS spectra of wood are reasonably well understood in terms of chemical shift assignments (see the Supporting Information for details).4457 Between about 168 and 200 ppm one finds the chemical shifts of carbonyl groups, including those of carboxyl moieties in the higher-shielding end of the overall carbonyl range. Between about 102 and 165 ppm, one finds the chemical shifts of the aromatic carbons of lignin. From 60 to 105 ppm are the chemical shifts of the carbohydrate carbons of holocellulose, as well as the chemical shifts of some side-chain (propane) units of lignin. Around 56 ppm one finds methoxy carbons of lignin and hemicellulose, as well as certain lignin side-chain carbons, and the acetate methyl carbons of hemicellulose occur at about 22 ppm. More detailed assignments are in the Supporting Information. These assignments should be considered to carry uncertainties of about (2.5 ppm, based on a number of factors, including (a) bulk magnetic susceptibility effects, (b) varying chemical shift referencing methods of different laboratories, and (c) the influences of non-nearest-neighbor structures among the various chemical structures in wood. Figure 2c shows the dipolar-dephasing (interrupted decoupling) 13C CP-MAS spectrum41 of the 13C enriched sample 13

2302

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Figure 3. (left) 13C CP-MAS results on 20 mg wood, 10% in H2O at 150 C, time span of 1 min, separated by 9 min DP segments. (right) 13C DP-MAS results on 20 mg wood, 10% in H2O at 150 C, time span of 9 min, separated by 1 min CP segments.

of Figure 2a. In the dipolar-dephasing technique, an interrupt period of 40 us, during which the 1H decoupler is turned off, is inserted between the generation of 13C magnetization by CP and the detection of 13C magnetization decay during the detection period. During this interrupt period, 13C magnetization from carbons that are directly attached to hydrogen (except for rapidly

rotating methyl groups) is largely dephased by 1H13C dipolar interactions; hence, the remaining, and ultimately detected, 13C magnetization comes from carbon sites that have no directly attached hydrogen atoms (or rapidly rotating methyl groups). In comparing Figure 2a and c, one sees that the intensities of the carbohydrate CH peaks in part c, have been reduced by 90% 2303

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Figure 4. (left) 13C CP-MAS results on 20 mg enriched wood, 10% in 0.5 M H2SO4 at 150 C, time span of 1 min, separated by 9 min DP segments. (right) 13C DP-MAS results on 20 mg wood, 10% in 0.5 M H2SO4 at 150 C, time span of 9 min, separated by 1 min CP segments.

relative to those in a. The lignin, carbonyl, and methyl peaks show a 1020% loss of intensity due at least in part to the disappearance of (i) spinning side bands resultant from the carbohydrate chemical shift anisotropy or (ii) reduced overlap with carbohydrate peaks. The spectra in both Figure 2a and c show a small peak at 30 ppm, due to aliphatic methylene sidechain carbons of lignin; this peak is not seen in Figure 2b due to the low signal-to-noise in the spectra of natural-abundance wood samples. Additional dipolar-dephasing results on other samples can be found in the Supporting Information.

13

C MAS spectra were obtained in both the CP (cross-polarization) and DP (direct polarization) modes as a function of heating time on samples that consisted of 10% slurries of 13C-enriched poplar sawdust (i) with pure water and heated at 150 C (Figure 3), (ii) with 0.5 M H2SO4 at 150 C (Figure 4), and (iii) with 0.5 M H2SO4 at 120 C (Figure 5). Both CP and DP modes were employed because these two modes respond very differently to atomic-level mobility. The CP mode depends upon a static component of 1H13C dipolar interactions, while the DP mode relies entirely on 13C spinlattice relaxation, Pretreatment Transformations.

2304

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Figure 5. (left) 13C CP-MAS results on 20 mg enriched wood, 10% in 0.5 M H2SO4 at 120 C, time span of 1 min, separated by 9 min DP segments. (right) 13C DP-MAS results on 20 mg wood, 10% in 0.5 M H2SO4 at 120 C, time span of 9 min, separated by 1 min CP segments.

hence 13C-spin interactions that have time-dependences with components in the 107108 Hz frequency range. Therefore, highly rigid wood components will yield intensities preferentially in the CP

mode, while highly mobile (liquid-like) components yield intensities only in the DP mode. Components in the intermediate mobility range may generate 13C intensity to some degree in both modes. 2305

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels The experiments leading to Figures 35 were carried out by using CP-MAS and DP-MAS techniques applied in alternating measurements of 1 min CP, 9 min DP, 1 min CP, 9 min DP, etc. Each of the three series of experiments employed one sample over a total of 500 min (including 10 min at 30 C for each sample before and after heating and 10 min of cool-down at the end of heating) and generated 51 separate spectra in each mode. When each sample in the three series has reached the elevated temperature (typically after about 3 min), the spectrometer probe is electronically retuned, so that any observed reduction in 13C NMR intensity is not largely due to probe tuning. Not all the spectra are shown in Figures 35; instead several instances of signal-averaging were employed over time spans during which substantial variations are not seen among the several individual spectra. The series of measurements made on a water/poplar slurry at 150 C (Figure 3) were carried out as a type of “control” experiment, to examine what, if any changes in the sample/ spectra occur in the absence of added acid. The spectra show some interesting features. First, while the main features of the CP-MAS and DP-MAS spectra of the initial, unheated sample (Figure 3A) are very similar, there are some substantial differences, as one might expect from the reasons stated above regarding the relationships between intensity and mobility. As the heating begins, these CP-vs-DP differences increase; most notably the intensities in the CP-MAS spectra decrease with time and new and more highly resolved DP-MAS peaks appear and are more intense with time. These changes are apparent even in the spectra obtained after 1 min at elevated temperature. Changes in the CP-MAS spectra over the period from 10 to 301 min are gradual; most of the overall changes in CP-MAS spectra have occurred within a few minutes. By the time 5.05.5 h of heating at 150 C has occurred, more than 60% of the CP-MAS intensity has been lost (Figure 3G-CP) and only about half of the “lost” intensity returns when the sample is returned to 30 C (Figure 3H-CP). In the corresponding DP-MAS spectra, there is no substantial loss of overall spectral intensity with time; in fact there is an apparent increase in the 5.05.5 h spectrum (Figure 3G-DP), some of which is lost, along with intensity of the sharpest peaks, when the sample is cooled back down to 30 C (Figure 3H-DP). The overall sharpness of the DP-MAS spectrum at 150 C is decreased when the sample is cooled back to 30 C, reflecting an overall loss of liquid-like character in the sample at the lower temperature. There is no large overall loss in DP-MAS intensity from the preheating 30 C spectrum (Figure 3A-DP) to the postheating 30 C spectrum (Figure 3H-DP). Only about 30% of NMR intensity would be lost in the 30150 C change because of the Boltzmann factor and all of that loss should be recovered in the return to 30 C. Thus, the loss of CP-MAS intensity during heating is due to a combination of the conversion of immobile structural components into (a) more mobile forms of the same components and (b) other, more mobile structural components. Some of that enhanced mobility (and resulting reduced CP-MAS intensity) is reversed when the sample is cooled back to room temperature. Detailed examination of Figure 3 (and its 102spectra version) reveals that, in the CP spectra, there is an intensity decrease to about 36% of the initial CP intensity with heating and a return to about 60% of the initial CP signal intensity when the sample is returned to 30 C. The ligninaromaticcarbon intensity between about 100 and 150 ppm is only marginally detected in the CP spectra of samples heated at 150 C, but at

ARTICLE

least partially recovers when the sample is cooled back to 30 C (Figure 3H-CP). Essentially the same kind of behavior is seen for the carbonyl/carboxyl peak at about 173 ppm and for the aliphatic carbon peaks in the 1020 ppm range. In the DP spectra, there is an additional peak at about 111 ppm that becomes very intense at 150 C and reduces to its initial intensity upon return to 30 C. As expected, 150 C heating of the 0.5 M H2SO4-treated sample yields much more dramatic changes in the 13C MAS spectra (Figure 4). The most drastic changes in both the CP-MAS and the DP-MAS spectra occur within about 10 min after the sample reaches 150 C (Figure 4C). The loss of CP-MAS intensity is much more dramatic, about 75%, after about 10 min, and remains essentially constant throughout roughly 3 h of heating (Figure 4C-CPG-CP). Again, some of the lost CPMAS intensity (about half of the initial value) is restored upon cooling to 30 C (Figure 4H-CP). As the 150 C/0.5 M H2SO4 pretreatment progresses, the DP-MAS spectra display even more sharp peaks throughout the spectrum, indicating liquid-like behavior, than in the corresponding 150 C treatment with pure water. And, unlike the 150 C/water case, with 0.5 M H2SO4 treatment at 150 C, the sharp features of the DP-MAS spectra obtained at 150 C between about 10 and 470 min (Figure 4CDP and G-DP) are retained, even enhanced, when the sample is cooled back to 30 C (Figure 4H-DP). In the CP spectra, the peaks at 22 and 173 ppm (due to acetoxy groups) disappear after 20 min and do not reappear when the sample is returned to 30 C; this indicates that the acetate moieties in hemicellulose and lignin have been hydrolyzed essentially irreversibly. However, the same spectral regions in the DP spectra do not disappear but show a decrease in line width, and the carbonyl/carboxyl peak at 173 ppm seems to split into two peaks at 178 and 181 ppm. In addition, several sharp peaks are formed at 167, 125, 97, 93, 50, and 29 ppm. From Figure 4, one sees that most of the NMR-visible chemical/physical transformations that happen in 150 C/0.5 M H2SO4 occur within the first few minutes of heating. It might be advantageous to slow down the transformation enough so that their progress in time can be monitored more successfully. For this reason, the same kinds of 0.5 M H2SO4 experiments were carried out at 120 C; the results are shown in Figure 5. One sees in Figures 4 and 5, and in the 102spectra arrays from which these figures were condensed, that as expected the rate and extent of H2SO4-assisted transformation were greater at 150 C than at 120 C. The main reason why the time resolution of alternated CPMAS/DP-MAS pairs in the experiments represented in Figures 35 is so poor (10 min) is the 9 min required to obtain each DP-MAS spectrum with good S/N. Accordingly, in the interest of improving time resolution in observing the time progression of the pretreatment transformations, two series of CP-MAS experiments, one at 150 C and one at 120 C, were carried out on 0.5 M H2SO4/poplar samples; in these experiments a CP-MAS spectrum was obtained each minute as the heating progresses. A portion of the resulting spectra are shown in Figure 6. One can see in these results the progression/history of relatively immobile structural moieties during even the shorttime portions of what is represented in Figures 4 and 5. Figure 6a shows that the peaks at 173 and 22 ppm disappear (presumably due to hydrolysis of acetate esters) after 3 min at 2306

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Figure 6. 13C CP-MAS experiments on 20 mg enriched wood, 10% in 0.5 M H2SO4. (left) Heating at 120 C. (right) Heating at 150 C. (A) Before heating, at 30 C. Various times elapsed at elevated temperatures (B, C, D, E, F, G, H). (I) After treatment, at 30 C.

150 C, while Figure 6b shows the same peaks disappear after 5 min at 120 C. Relaxation Studies. Variable-contact-time (VCT) experiments were carried out on four samples; the results are summarized in Tables 14 for a sample of dry poplar wood at 30 C

(Table 1), a sample of poplar in 0.5 M H2SO4 at 30 C before treatment at 150 C (Table 2), a sample of poplar in 0.5 M H2SO4 during treatment at 150 oC (Table 3), and a sample of poplar in 0.5 M H2SO4 at 30 C after treatment at 150 C (Table 4). Analysis of the VCT results by fitting the data to the 2307

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Table 1. Variable Contact Time Results on Dry Poplar at 30 C

L

peaka

M¥b

M¥ *c

T1FH (ms)

22

2.1 ( 0.1

2.4 ( 0.1

56 62

5.0 ( 0.1 8.2 ( 0.2

5.8 ( 0.1 9.7 ( 0.2

Table 3. Variable Contact Time Results on 10% Poplar in 0.5 M H2SO4 During Treatment at 150 C

TCH (μs)

peaka

10 ( 1

66 ( 5

d

11 ( 1 10 ( 1

52 ( 5 32 ( 3

22 L

56 62

M¥b

M¥ *c

T1FH (ms)

TCH (μs)

1.0 ( 0.1 1.8 ( 0.1

1.5 ( 0.1 2.7 ( 0.1

24 ( 1 24 ( 2

36 ( 5 37 ( 4

C

66

8.1 ( 0.2

9.7 ( 0.2

11 ( 2

31 ( 3

C

66

1.9 ( 0.1

2.8 ( 0.1

24 ( 2

34 ( 3

C

73

17.8 ( 0.6

21.3 ( 0.6

10 ( 1

36 ( 5

C

73

3.9 ( 0.1

5.8 ( 0.1

26 ( 3

39 ( 4

75

17.5 ( 0.6

20.9 ( 0.6

10 ( 1

36 ( 5

75

3.6 ( 0.1

5.3 ( 0.1

25 ( 3

38 ( 4

84

10.6 ( 0.4

12.6 ( 0.4

10 ( 1

36 ( 5

84

2.4 ( 0.1

3.5 ( 0.1

22 ( 3

44 ( 5

C

89

2.5 ( 0.6

3.0 ( 0.6

11 ( 1

36 ( 4

C

89

0.6 ( 0.1

0.9 ( 0.1

21 ( 3

46 ( 6

C

105

7.5 ( 0.2

9.0 ( 0.2

11 ( 1

41 ( 5

C

105

1.7 ( 0.1

2.5 ( 0.1

24 ( 2

46 ( 4

L L

135 153

4.5 ( 0.2 2.7 ( 0.1

5.2 ( 0.2 3.2 ( 0.1

13 ( 2 14 ( 2

80 ( 10 150 ( 20

L L

135 153

1.3 ( 0.1 0.6 ( 0.1

1.8 ( 0.1 0.8 ( 0.1

24 ( 5 20 ( 3

130 ( 30 160 ( 30

173

2.6 ( 0.1

3.0 ( 0.1

12 ( 1

200 ( 30

a

Peak positions in parts per million. L signifies a lignin peak, C signifies a cellulose peak, and others are from hemicellulose and minor contributions from amorphous cellulose. b M¥ represents the transverse 13C magnetization that would have been generated if the CP transfer had been infinitely fast and if rotating-frame 1H spinlattice relaxation had been infinitely slow. c M¥ * is the corresponding M¥ value that has been corrected by the factor, {1  exp(t/T1H)}1, for incomplete 1H spin lattice relaxation between CP scans.

Table 2. Variable Contact Time Results on 10% Poplar in 0.5 M H2SO4 at 30 C Before Treatment at 150 C a

peak



b

M*¥

c

T1FH (ms)

TCH (μs)

173d a

Peak positions in parts per million. L signifies a lignin peak, C signifies a cellulose peak, and others are from hemicellulose and minor contributions from amorphous cellulose. b M¥ represents the transverse 13C magnetization that would have been generated if the CP transfer had been infinitely fast and if rotating-frame 1H spinlattice relaxation had been infinitely slow. c M¥ * is the corresponding M¥ value that has been corrected by the factor, {1  exp(t/T1H)}1, for incomplete 1H spinlattice relaxation between CP scans. d Not observed under these conditions.

Table 4. Variable Contact Time Results on 10% Poplar in 0.5 M H2SO4 at 30 C After Treatment at 150 C peaka

M¥b

M*¥ c

T1FH (ms)

TCH (μs)

22

0.8 ( 0.1

0.9 ( 0.1

10 ( 1

66 ( 9

56 62

2.2 ( 0.1 4.2 ( 0.2

2.5 ( 0.1 5.2 ( 0.2

7(1 6(1

53 ( 6 33 ( 5

C

66

3.9 ( 0.2

4.8 ( 0.2

8( 1

30 ( 5

C

73

8.5 ( 0.5

10.7 ( 0.5

7(1

34 ( 7

C

66

2.5 ( 0.1

3.2 ( 0.1

23 ( 2

30 ( 3

C

73

5.2 ( 0.2

6.7 ( 0.2

23 ( 3

36 ( 5

L

22d L

56

1.4 ( 0.1

1.7 ( 0.1

24 ( 3

53 ( 3

62

2.2 ( 0.1

2.8 ( 0.1

21 ( 2

33 ( 3

75

7.7 ( 0.5

9.6 ( 0.5

7(1

33 ( 8

84

4.5 ( 0.3

5.5 ( 0.3

9(1

32 ( 8

75

4.6 ( 0.1

5.9 ( 0.1

22 ( 3

35 ( 4

84

3.3 ( 0.1

4.1 ( 0.1

21 ( 3

37 ( 5

89 105

1.0 ( 0.1 2.3 ( 0.1

1.3 ( 0.1 3.0 ( 0.1

21 ( 3 23 ( 3

39 ( 5 40 ( 5

C

89

1.2 ( 0.1

1.5 ( 0.1

10 ( 2

34 ( 8

C

105

3.7 ( 0.3

4.7 ( 0.3

8(1

36 ( 9

L L

135 153

1.6 ( 0.1 0.8 ( 0.1

1.9 ( 0.1 0.9 ( 0.1

18 ( 1 24 ( 4

93 ( 6 90 ( 10

L

135

2.2 ( 0.1

2.6 ( 0.1

30 ( 6

120 ( 20

173

0.6 ( 0.1

0.7 ( 0.1

27 ( 4

110 ( 10

L

153

1.0 ( 0.1

1.2 ( 0.1

28 ( 8

150 ( 40

C C

a

Peak positions in parts per million. L signifies a lignin peak, C signifies a cellulose peak, and others are from hemicellulose and minor contributions from amorphous cellulose. b M¥ represents the transverse 13C magnetization that would have been generated if the CP transfer had been infinitely fast and if rotating-frame 1H spinlattice relaxation had been infinitely slow. c M¥ * is the corresponding M¥ value that has been corrected by the factor, {1  exp(t/T1H)}1, for incomplete 1H spinlattice relaxation between CP scans.

accepted equation36,41 yields values of the parameters, TCH, T1FH, and M¥, which are summarized in Tables 14. M¥ is the value that the CP-generated magnetization would have if the polarization-transfer process were infinitely fast and if rotatingframe proton spinlattice relaxation were infinitely slow. This number, when corrected by a well-established factor due to incomplete 1H spinlattice relaxation, {1  exp(t/T1H)}1, * that should faithfully represent the populayields a number M¥ tion of 13C sites that were available for cross-polarization; these parameters are also collected in Tables 14.

173d a

Peak positions in parts per million. L signifies a lignin peak, C signifies a cellulose peak, and others are from hemicellulose and minor contributions from amorphous cellulose. b M¥ represents the transverse 13C magnetization that would have been generated if the CP transfer had been infinitely fast and if rotating-frame 1H spinlattice relaxation had been infinitely slow. c M¥ * is the corresponding M¥ value that has been corrected by the factor, {1  exp(t/T1H)}1, for incomplete 1H spin lattice relaxation between CP scans. d Not observed under these conditions.

The interpretation of T1FH values is made difficult by the existence of several potentially important mechanisms, even for much simpler chemical systems than wood.58,59 However, one can often usefully make relatively straightforward interpretations of TCH data in terms of atomic-level mobilities. In comparing the TCH results reported in Tables 1 and 2 for dry wood and wood treated with 0.5 M H2SO4, both at 30 C, one sees that the addition of 0.5 M H2SO4 to poplar wood at 30 C causes almost no changes that are detectable via TCH values, with the exception 2308

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Table 5. Summary of T1H Results on Dry Poplar at 30 C, and on 10% Poplar in 0.5 M H2SO4 at 30 C before and after Treatment at 150 C and during Treatment at 150 C TDry 1H (s)

TBefore (s) 1H

22

1.0 ( 0.1

1.1 ( 0.1

TDuring (s) 1H a

TAfter 1H (s) a

56

1.0 ( 0.1

1.0 ( 0.1

1.8 ( 0.1

1.2 ( 0.1

62

1.1 ( 0.1

1.2 ( 0.1

1.8 ( 0.1

1.3 ( 0.1

C

66

1.1 ( 0.1

1.2 ( 0.1

1.8 ( 0.1

1.3 ( 0.1

C

73 75

1.1 ( 0.1 1.1 ( 0.1

1.3 ( 0.1 1.3 ( 0.1

1.8 ( 0.1 1.7 ( 0.1

1.3 ( 0.1 1.3 ( 0.1

84

1.1 ( 0.1

1.2 ( 0.1

1.7 ( 0.1

1.3 ( 0.1

C

89

1.1 ( 0.1

1.2 ( 0.1

1.7 ( 0.1

1.3 ( 0.1

C

C

105

1.1 ( 0.1

1.3 ( 0.1

1.8 ( 0.1

1.3 ( 0.1

L

135

1.0 ( 0.1

1.1 ( 0.1

1.5 ( 0.1

L

153

1.1 ( 0.1

1.0 ( 0.1

173

1.0 ( 0.1

1.0 ( 0.1

L

a

peak

Table 6. Summary of T1C Results on Dry Poplar at 30 C and on 10% Poplar in 0.5 M H2SO4 at 30 C before and after Treatment at 150 C and during Treatment at 150 C peaka

TDry 1C (s)

TBefore (s) 1C

TDuring (s) 1C

TAfter 1C (s)

22

3.9 ( 0.1

4.4 ( 0.3

a

a

56

3.8 ( 0.1

4.3 ( 0.2

M ( 0.6

5.9 ( 0.4

62

3.9 ( 0.1

4.2 ( 0.1

3.9 ( 0.3

5.6 ( 0.3

C

66

4.2 ( 0.1

4.5 ( 0.1

3.7 ( 0.2

5.8 ( 0.2

C

73 75

4.3 ( 0.1 4.2 ( 0.1

4.7 ( 0.1 4.5 ( 0.1

4.1 ( 0.2 4.1 ( 0.2

5.9 ( 0.1 5.9 ( 0.2

84

M ( 0.2

4.3 ( 0.2

4.3 ( 0.2

6.0 ( 0.4

89

4.4 ( 0.1

5.1 ( 0.3

4.4 ( 0.6

6.5 ( 0.5

C

105

4.4 ( 0.1

4.3 ( 0.2

4.1 ( 0.3

6.0 ( 0.3

1.1 ( 0.1

L

135

4.2 ( 0.2

5.0 ( 0.4

4.7 ( 0.4

6.6 ( 0.5

1.6 ( 0.2

1.2 ( 0.1

L

153

4.6 ( 0.2

4.7 ( 0.4

4.1 ( 0.2

a

a

173

M ( 0.2

4.7 ( 0.6

Not observed under these conditions.

of the low-shielding peaks at 153 and 173 ppm. These same peaks are dramatically reduced in intensity in the CP-MAS spectrum in the early stages of heating and are only marginally recovered when the sample is cooled back down to 30 C. This behavior probably reflects substantial hydrolysis of acetate esters and phenolic ethers or esters of lignin, so that the final sample contains a much smaller quantity of unhydrolysed entities as part of the solid-like (CP-detectable) material. And, the unhydrolyzed acetate and phenoxy moieties that remain after 0.5 M H2SO4 addition at 30 C are probably more rigid (smaller TCH values) than their prehydrolysis ancestors. Comparison of the TCH values for the poplar/0.5 M H2SO4 samples summarized in Tables 24 reveals two main patterns: (1) the TCH values are very similar (essentially within experimental error) before and after the sample is heated at 150 C for corresponding 13C chemical shifts and (2) the TCH values for corresponding chemical shifts appear to be larger (implying greater atomic-level mobility) when the sample is at 150 C (Table 3), although the differences are at the edge of experimental uncertainties—with an exception for the methoxy peak (56 ppm), for which TCH is clearly smaller at 150 C than at 30 C. This curious exception may reflect slowed methyl group rotation due to some unknown steric constraint during hydrolysis. Measurements of 1H T1 values were made by a well-established 13C-detection method41 on the same four systems represented in Tables 14. One sees from the results collected in Table 5 that, for any one of these four systems, the 1H T1 values are almost the same, within experimental error, for all the 13C chemical shift ranges sampled. This behavior is typical of solids, as efficient 1H spin diffusion during a recovery period tends to yield an average 1H T1 value for all sites in a sample. Inspection of Table 5 also reveals that the “spin-diffusion-averaged” 1H T1 values of this table differ between samples. Most noteworthy are the facts that (a) the “spin-diffusion-averaged” 1H T1 values of the two 30 C samples in 0.5 M H2SO4, before and after heating at 150 C, are nearly the same, and (b) the “spin-diffusionaveraged” 1H T1 value(s) of the sample in 0.5 M H2SO4 measured during heating at 150 C are roughly 50% larger at 150 C than those measured at 30 C before and after heating. 13 C T1 measurements were made by the Torchia method43 on the same four systems represented in Tables 15. The results are

L

a

a

5.8 ( 0.4 a

Not observed under these conditions.

Table 7. Field Dependence of T1H and T1C on a 10% Poplar/Water Slurry (Natural-Abundance) at Different Magnetic Fields field strength B0 (T)

cellulose region

lignin region

(60110 ppm)

(120160 ppm)

T1H (s)

T1C (s)

T1H (s)

T1C (s)

2.4

0.3 ( 0.1

9(1

0.4 ( 0.1

9(1

4.7

0.6 ( 0.1

16 ( 2

0.7 ( 0.1

18 ( 3

8.5

0.8 ( 0.1

11 ( 3

0.9 ( 0.1

12 ( 4

summarized in Table 6. Unlike what is typically encountered with C in natural abundance, where 13C T1 values can vary dramatically from one type of molecular site to another in a given sample, in this case of high 13C enrichment, the variation among 13C T1 values for the various carbon sites are significant, but small, both within a sample and between samples. In order to interpret T1 variations in terms of atomic-level motion, often characterized by a motional correlation time, τc, it is very helpful to know the relationship between T1 and τc, i.e., the relative positions of τc and the so-called T1 minima in the time domain.36,62,63 For this reason T1 values for both 1H and 13C were determined on a water/poplar slurry sample at three different magnetic fields 2.4 T (100 MHz for 1H), 4.7 T (200 MHz 1 H), and 8.5 T (360 MHz 1H). The results are summarized in Table 7. 13

’ DISCUSSION Interpretation of CP-MAS Spectra As Treatment Progresses. In the 150 C H2O treatment (Figure 3), all the peaks

in the CP spectra decrease in intensity, but roughly half of this intensity returns when the sample is cooled back to 30 C after heating. This intensity decrease is due to a combination of a general loosening of the wood structure (yielding a smaller fraction of the wood structure that is sufficiently immobile to support 1H f 13C cross-polarization) without major changes in local “monomer-level” (molecular building block) structure of the wood. In the DP spectra, the peaks at 22 (acetate methyls), 56 2309

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels (methoxy groups), 112 (uncertain assignment, vide infra), and 173 ppm (acetate carbonyl carbons) become narrow at 150 C and broaden upon return to 30 C; these results suggest that there is no hydrolysis of wood components due to H2O treatment to an extent that is detected by solid state 13C NMR. The above-mentioned narrowing effects are probably also due to the general loosening of the wood structure invoked above for rationalizing the CP-MAS results; this loosening provides some averaging (line-narrowing of broadening due to chemical shift dispersion). The broad lines seen in Figure 3H-DP probably indicate that the level of mobility retained when the sample is brought back to 30 C is sufficient to partially interfere with the CP process but not sufficient to retain the averaging/narrowing effect seen in DP-MAS spectra at 150 C. In the 150 C H2SO4 treatment (Figure 4), the CP-MAS peaks at 22 and 173 ppm (due to acetate esters) are not present after 10 min of treatment and are not present in the CP-MAS spectra after cooling back to 30 C. All other peaks decrease in intensity, about half of which returns after cooling to 30 C. This intensity decrease is in part due to the general “loosening” effect invoked above for the 150 C H2O treatment but can also be identified with a heating-produced decrease in the amount of carbon that has sufficiently low atomic-level mobility to support 1H f 13C CP; this can be seen by comparing the M*¥ values listed for the carbohydrate region in Tables 2 and 4 for the 30 C samples before and after the 150 C treatment, respectively. In the DP spectra of Figure 4, the peak at 22 ppm (methyl of acetate) becomes narrow at 150 C and remains narrow on return to 30 C, suggesting hydrolysis of hemicellulose acetate groups to acetic acid. The peak at 30 ppm is due to aliphatic carbons of lignin. After 10 min of heating, this peak begins to increase in intensity, all of which is retained after treatment. Since the side-chain content of lignin is very unlikely to be increased by the 150 C/H2SO4 treatment, this increased intensity is most likely due to some other kind of chemical transformation, e.g., the formation of levulinic acid (a well-known degradation/dehydration product of simple sugars;60 see Figure 7), whose sp3 carbons may also contribute to the intensity of this peak. The sharp DP peak in Figure 4 at 39 ppm appears after 60 min of heating and also increases in intensity during heating, most of this intensity increase remaining after treatment. This peak is most likely due to the aliphatic carbon atoms in levulinic acid. (see Figure 7).61 The DP peak at 50 ppm in Figure 4 appears after 60 min and is most likely methanol, a result of methoxy ether hydrolysis. The methoxy (ether) peak at 56 ppm becomes very narrow at the beginning of heating and gradually decreases in intensity as treatment progresses, supporting hydrolysis of methoxy carbons from hemicellulose and lignin. The peak at 62 ppm (C6 of hexose and C5 of pentose units) in Figure 4 becomes narrow and increases in intensity at 150 C and remains narrow after cooling back to 30 C. This sharp peak is most likely the C6 carbon of hydrolyzed hexose units and C5 carbon of hydrolyzed pentose units from hemicellulose. The DP peak at 66 ppm (C6 of cellulose) in Figure 4 appears to decrease as heating progresses and does not recover intensity upon the return to 30 C, suggesting a possible decrease in the crystalinity of cellulose. The DP peaks at 73 and 75 in Figure 4 do not show major changes, but these peaks are difficult to interpret in detail because they are most likely the result of overlapping peaks from the C2, C3, and C5 carbons of cellulose, hemicellulose, and the C2 and C3 carbons of monosaccharides. The DP peaks at 84 and 89 ppm (due to carbons of cellulose and hemicellulose) in Figure 4

ARTICLE

Figure 7. Structures of presumed products of hemicellulose hydrolysis during dilute sulfuric acid pretreatment: (1) β-D-glucopyranose; (2) RD-xylopyranose; (3) β-D-xylopyranose; (4) levulinic acid; (5) acetic acid; (6) formic acid; (7) methanol; (8) furfural; (9) hydroxymethylfurfural.

decrease in intensity as treatment progresses, also indicating transformations within the holocellulose fraction. There are two new DP peaks formed at 93 and 97 ppm (Figure 4) most likely due to C1 in the R and β orientations, respectively, of hydrolyzed sugars.60 The DP peak at 97 ppm in Figure 4 has greater intensity than the peak at 93 ppm, consistent with the expectation that the β orientation will be more prevalent than the R orientation in a saccharidic solution.60 The DP-MAS peak at 105 ppm (C1 of cellulose and possibly some hemicelluloses) in Figure 4 decreases in the H2SO4/150 C treatment due to hydrolysis of polysaccharide linkages; this “lost” intensity may be at least partially responsible for increased intensities in the peaks at 93 and 97 ppm. The DP-MAS peak at 112 ppm in Figure 4 becomes very narrow in the H2SO4/150 C treatment and broadens on return to 30 C; this peak is likely due to sp2 carbons of furfural (8), an intermediate in the formation of levulinic acid (4).60 The DPMAS peak at 125 ppm appears after 60 min of heating and increases in intensity during treatment; all of the intensity is recovered on return to 30 C. This 125 ppm peak is most likely due to sp2 carbons of furfural (8). The DP-MAS intensity of the peaks at 135 ppm (aromatic C1 carbon of some lignin structures) and 153 ppm (aromatic C3 and C5 of lignin) appear to show slight intensity changes, but mainly line width changes in the H2SO4/150 C treatment. The DP-MAS peak at 167 ppm appears after 200 min of H2SO4/150 C treatment and is most likely a result of the formation of formic acid, a degredation product formed in the transformation of 5-hydroxymethyl furfural (9) to levulinic acid (4).60 The DP-MAS peak at 173 ppm (carboxyl carbon region), which is rather broad before heating, is 2310

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

Figure 8. High-resolution (liquid sample) 100 MHz 13C NMR spectrum of liquid phase from poplar pretreatment with 0.5 M H2SO4 at 150 C. See the Supporting Information for details.

apparently transformed into one or more sharp peaks in the 150 C heated and postheating 30 C samples. This implies a substantial mobilization and chemical transformation, e.g., hydrolysis, of all detectable carboxy moieties. The DP-MAS peak at 178 ppm begins to form after 20 min of treatment and remains after treatment. This peak is due to carbonyl groups, including those of carboxylic acid moieties, e.g., levulinic acid (4), acetic acid (5), and possibly of hydrolyzed uronic acids of hemicellulose. The DP-MAS peak at 181 ppm appears after 60 min of heating and remains upon return to 30 C; this is suggestive of carbonyl carbons (vide supra) or possibly the aldehyde carbon of furfural (8). Changes in the CP-MAS and DP-MAS spectra obtained before, during, and after the 120 C/H2SO4 treatment (Figure 5) are similar to those changes discussed above for the corresponding 150 C case (Figure 4). The main difference is that the observed changes occur more slowly at 120 C than at 150 C. High-Resolution 13C NMR Data. Additional detail on the liquid phase components produced as a result of the 0.5 M H2SO4/150 C pretreatment is obtained from the high-resolution (HR) (liquid sample) 13C NMR spectrum obtained on the liquid-phase portion of the pretreated sample (Figure 8). The Supporting Information gives a detailed analysis of this spectrum, which provides evidence for the species identified as 17 in Figure 7. The fact that the HR spectrum shows only very weak peaks at 113, 132, and 150 ppm, assigned to furfural species (8, 9) means that, for some reason, the furfural species resulting from the pretreatment is highly mobile but (a) is a very minor contribution of the liquid components in the 0.1 M H2SO4 wash or (b) possibly decomposed during sample storage and/or the separation. The fact that the DP-MAS spectrum of the solid residue resulting from the solid/liquid separation does not show substantial intensity in the 113, 132, 150 ppm regions implies that b is the more likely explanation. No evidence was seen in the HR 13C NMR spectrum for sugar oligomers, which would give

rise to 13C intensity at about 102105 ppm, which is characteristic of glycosidic (C1OC4) linkages. SpinLattice Relaxation. The interpretation of T1 results in terms of atomic-level motion, as represented by the motional correlation time, τc, is usually based on the relationships between spinlattice relaxation rate constants, (T1)1, and the spectral density components, J(ω), of the motion at certain key frequencies, ωk, which depend on relevant Lamor frequencies (expressed here in radians per second). These relationships are given for T1H and T1C by the following equations (for the oversimplified but useful case of isotropic motion):36,62,63 ðT 1H Þ1 ¼ ð1=2ÞJ 1 ðωH Þ þ 2J 2 ð2ωH Þ

ð1Þ

ðT 1C Þ1 ¼ ð1=12ÞJ 0 ðωH  ωC Þ þ ð1=4ÞJ 1 ðωC Þ þ ð1=2ÞJ 2 ðωH þ ωC Þ

ð2Þ

From eq 1, one sees that the key motional frequencies, τc1, that are most effective for proton spinlattice relaxation are the proton Larmor frequency and twice that frequency. These are the motional frequencies that define the shape of a plot of J(ω) vs τc for any specific frequency, ω.36,58,59 For the isotropic rotation model, the spectral densities are viewed in terms of the ubiquitous formula, τ/(1 þ τ2ω2).36,62,63 For 13C spinlattice relaxation, eq 2 tells us that there are three key motional frequencies. These are the 13C Larmor frequency, ωC; the sum frequency, ωH þ ωC; and the difference frequency, ωH  ωC. Since the contribution associated with the difference frequency appears to be much smaller than the other two, it is neglected in the qualitative analysis outlined here. The following four key patterns in the T1 data should be accommodated by any serious interpretation: (1) The T1H values measured at three magnetic fields increase monotonically with increasing magnetic field (Table 7). (2) T1H measured at 150 C is larger than that measured at 30 C before or after heating 2311

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels (Table 5). (3) T1C increases when the magnetic field is increased from 2.4 to 4.7 T but then decreases when the field is increased again to 8.5 T (Table 7). (4) T1C values are smaller at 150 C than at 30 C (measured after heating, Table 6). If one makes the simplifying assumption that the atomic-level motion in any one of these systems is not only isotropic but can also be represented by just one (global) correlation time, then, if one takes these four data patterns into account as constraints on the oversimplified model and in terms of the T1 minima mentioned above, one can conclude that the experimental patterns can be rationalized if the global τc is roughly 1.5 ns at 30 C and is decreased to roughly 0.7 ns at 150 C. In this view of T1H at 30 C, as one increases the magnetic field strength, the ωH relaxation dependence (curve) and/or the 2ωH curve is intersected first (smallest T1H value) for 2.4 T, then the ωH and 2ωH curves are intersected for 4.7 T and finally (largest T1H) the ωH curve is intersected for 8.5 T. The fact that T1H is larger at 150 C than at 30 C is consistent with the interpretation that the two relevant τc values stated above (about 1.5 ns at 30 C and 0.7 ns at 150 C) are on opposite sides of the T1H minimum for ωH at 4.7 T, with the latter T1H value being larger than the former. At 30 C (where τc is roughly 1.5 ns), T1C is larger at 4.7 T than at 2.4 T; this is consistent with intersecting the ωC þ ωH curve for 2.4 T first, as one increases the magnetic field strength, then intersecting the ωC curve for 8.5 T, before intersecting the ωC þ ωH curve for 4.5 T. The fact that T1C for 30 C is larger than at 150 C is due to the fact that the τc values for these two temperatures are both on the high-τc side of the T1Cvsτc minimum for both of these temperatures.

’ SUMMARY AND CONCLUSIONS 13 C NMR spectra, obtained by both DP-MAS and CP-MAS approaches, have allowed us to follow, via the populations of mobile and immobile structural segments, respectively, the progress of transformations that occur in 13C-enriched poplar wood under “pretreatment” with 0.5 M H2SO4 at 150 C (and, less completely, at 120 C). Detailed relaxation experiments allow (a) intensity corrections aimed toward quantitation and (b) an analysis of the degree of mobility of the wood components in terms of a global motional correlation time, τc (0.71.5 ns), in each of these samples under various pretreatment conditions. The time resolution available with the dramatically enhanced signal-to-noise characteristics that are achievable with 13C-enriched samples should render detailed kinetic studies possible; such studies have been initiated. ’ ASSOCIATED CONTENT

bS

Details on (1) 13C chemical shift assignments, (2) the level of 13C enrichment, and (3) the high-resolution 13C NMR analysis of the liquid phase resulting from 0.5 M H2SO4 pretreatment at 150 C. This material is available free of charge via the Internet at http://pubs.acs.org. Supporting Information.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT The authors gratefully acknowledge the support of this research by U.S. Department of Energy, the technical assistance

ARTICLE

of Dr. J. DiVerdi, and the help of S. Skogerboe of the Fort Collins Nursery for providing the young poplar shoots.

’ REFERENCES (1) McMillan, J. D. In Biochemical Conversion of Cellulosic Feedstocks; Conference on Biomass to Chemicals and Fuels: Science, Technology and Public Policy, Houston, TX, 2006. (2) Sun, Y.; Cheng, J. Biores. Technol. 2002, 83, 1–11. (3) Sannigrahi, P.; Ragauskas, A. J. Biofuels Bioprod. Bioref. 2010, 4, 209–226. (4) Kintisch, E. Science 2007, 315, 786. (5) Tullo, A. H. Chem. Eng. News 2007, 85 (48), 28–31. (6) Galbe, M.; Zacchi, G. Appl. Microbiol. Biotechnol. 2002, 59, 618– 628. (7) Lynd, L. R.; Wyman, C. E.; Tillman, U. G. Biotechnol. Prog. 1999, 15, 777–793. (8) McMillan, J. D. In Enzymatic Conversion of Biomass for Fuels Production; Himmel, M. E., Baker, J. O., Overend, R. P., Eds.; ACS Symposium Series; American Chemical Society: Washington, D.C., 1994; Vol. 566, pp 293324. (9) Taherzadeh, M. J.; Karimi, K. Int. J. Mol. Sci. 2008, 9, 1621–1651. (10) Mosier, N.; Wyman, C.; Dale, B.; Elander, R.; Lee, Y. Y.; Holtzapple, M.; Ladisch, M. Bioresour. Technol. 2005, 96, 673–686. (11) Kumar, P.; Barrett, D. M.; Delwiche, M. J.; Stroeve, P. Ind. Eng. Chem. Res. 2009, 48, 3713–3729. (12) Taherzadeh, M. J.; Karimi, K. Bioresources 2007, 2, 707–738. (13) Laurova, M.; Kacik, F.; Kacikova, D.; Sivak, J. Forest Wood Technol. 2008, 64, 14–21. (14) Fan, L. T.; Lee, Y.-H.; Gharpuray, M. M. Adv. Biochem. Eng. 1982, 23, 157–187. (15) Millett, M. A.; Baker, A. J.; Satter, L. D. Biotech. Bioeng. Symp. 1976, 6, 125–153. (16) Cara, C.; Ruiz, E.; Oliva, J. M.; Saez, F.; Castro, E. Bioresour. Technol. 2008, 99, 1869–1876. (17) Sun, Y.; Cheng, J. J. Bioresour. Technol. 2005, 96, 1599–1606. (18) Yang, B.; Wyman, C. E. Biotechnol. Bioeng. 2004, 86, 88–95. (19) Schell, D. J.; Torget, R.; Power, A.; Walter, P. J.; Grohmann, K.; Hinman, N. D. Appl. Biochem. Biotechnol. 1991, 28/29, 87–97. (20) Wyman, C. E. Biomass Ann. Rev. Energy Envir. 1999, 24, 189–226. (21) Torget, R.; Werdene, P.; Himmel, M.; Grohmann, K. Appl. Biochem. Biotechnol. 1990, 24/25, 115–126. (22) Allen, S. G.; Schulman, D.; Lichwa, J.; Antal, M. J., Jr.; Jennings, E.; Elander, R. Ind. Eng. Chem. Res. 2001, 40, 2352–2361. (23) Nguyen, Q. A.; Tucker, M. P.; Keller, F. A.; Eddy, F. P. Appl. Biochem. Biotechnol. 2000, 8486, 561–576. (24) Lee, Y. Y.; Iyer, P.; Torget, R. W. Adv. Biochem. Eng./Biotech. 1999, 65, 93–115. (25) Saha, B. C.; Bothast, R. J. Appl. Biochem. Biotechnol. 1999, 76, 65–77. (26) Nguyen, Q. A.; Tucker, M. P.; Boynton, B. L.; Keller, F. A.; Schell, D. J. Appl. Biochem. Biotechnol. 1998, 7072, 77–87. (27) Lee, K. C. P.; Bulls, M.; Holmes, J.; Barrier, J. W. Appl. Bioch. Biotech. 1997, 66, 1–23. (28) Torget, R.; Hsu, T. Appl. Biochem. Biotechnol. 1994, 45/ 46, 5–22. (29) Grethlein, H. E.; Allen, D. C.; Converse, A. O. Biotechnol. Bioeng. 1984, 26, 1498–1505. (30) Knappert, D.; Grethlein, H.; Converse, A. Biotechnol. Bioeng. 1980, 22, 1449–1463. (31) Knappert, D.; Grethlein, H.; Converse, A. Biotechnol. Bioeng. 1981, 11, 66–77. (32) Abraham, R. J.; Fisher, J.; Loftus, P. Introduction to NMR Spectroscopy; John Wiley & Sons: New York, 1988. (33) Fyfe, C. A. Solid State NMR for Chemists; C.F.C. Press: Guelph, Ontario, 1983. (34) Duer, M. J., Ed. Solid-State NMR Spectroscopy  Principles and Applications; Blackwell Science: Oxford, U.K., 2002. 2312

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313

Energy & Fuels

ARTICLE

(35) Maciel, G. E. Science 1984, 226, 282–288. (36) Mehring, M. Principles of High Resolution NMR in Solids, 2nd ed.; Springer-Verlag: New York, 1983. (37) Stejskal, E. O.; Memory, J. D. High Resolution NMR in the Solid State; Oxford Univ. Press: Oxford, U.K., 1994. (38) Schaefer, J.; Stejskal, E. O. J. Am. Chem. Soc. 1976, 98, 1031– 1032. (39) Stejskal, E. O.; Schaefer, J.; Waugh, J. S. J. Magn. Reson. 1977, 28, 105–1112. (40) Kobayashi, T.; Holmes, L.; Kohn, B.; Faulkner, R.; Davis, M.; Maciel, G. E. Energy Fuels 2011, 25, 1790–1797. (41) Keeler, C.; Maciel, G. E. Anal. Chem. 2003, 75, 2421–2432. (42) Sullivan, M. J.; Maciel, G. E. Anal. Chem. 1982, 54, 1615–1623. (43) Torchia, D. A. J. Magn. Reson. 1978, 30, 613–616. (44) Atalla, R. H.; Gast, J. C.; Sindorf, D. W.; Bartuska, V. J.; Maciel, G. E. J. Am. Chem. Soc. 1980, 102, 3249–3251. (45) Pacchiano, R. A.; Sohn, W.; Chlanda, V. L.; Garbow, J. R.; Stark, R. E. J. Agric. Food Chem. 1993, 41, 78–83. (46) Love, G. D.; Snape, C. E.; Jarvis, M. C. Phytochem. 1998, 49, 1191–1194. (47) Hatfield, G. R.; Maciel, G. E.; Erbatur, O.; Erbatur, G. Anal. Chem. 1987, 59, 172–179. (48) Earl, W. L.; VanderHart, D. H. J. Am. Chem. Soc. 1980, 102, 3251–3152. (49) Earl, W. L.; VanderHart, D. H. Macromolecules 1981, 14, 570– 574. (50) Kolodziejski, W.; Frye, J. S.; Maciel, G. E. Anal. Chem. 1982, 54, 1419–1424. (51) Davis, M. F.; Schroeder, H. R.; Maciel, G. E. Holzforschung 1994, 48, 99–105. (52) Davis, M. F.; Schroeder, H. R.; Maciel, G. E. Holzforschung 1994, 48, 186–192. (53) Davis, M. F.; Schroeder, H. R.; Maciel, G. E. Holzforschung 1994, 48, 301–307. (54) Haw, J. F.; Maciel, G. E.; Schroeder, H. A. Anal. Chem. 1984, 56, 1323–1329. (55) Haw, J. F.; Maciel, G. E.; Biermann, C. J. Holzforschung 1984, 38, 327–331. (56) Leary, G. J.; Morgan, K. R.; Newman, R. H.; Samuelsson, B.; Westermark, U. Holzforschung 1986, 40, 221–224. (57) Freitas, J. C. C.; Bonagamba, T. J.; Emmerich, F. G. Carbon 2001, 39, 535–545. (58) Garroway, A. N. J. Magn. Reson. 1979, 34, 283–293. (59) Schaefer, J.; Stejskal, E. O.; Buchdahl, R. Macromolecules 1975, 8, 291–296. (60) Sj€ostr€om, E. Wood Chemistry Fundamentals and Applications, 2nd ed.; Academic Press: New York, 1993; Chapter 2. (61) Fengel, D.; Wegener, G. Wood Chemistry Ultrastructure Reactions; Gruyter: New York, 1984; Chapter 10. (62) Farrar, T. C.; Becker, E. D. Pulse and Fourier Transform NMR; Academic Press: New York, 1971; pp 5155. (63) Slichter, C. P. Principles of Magnetic Resonance, 2nd Ed.; Springer-Verlag: New York, 1978; pp 164167.

2313

dx.doi.org/10.1021/ef2000213 |Energy Fuels 2011, 25, 2301–2313