Effects of Torrefaction Temperature on Pyrolysis Vapor Products of

Jun 17, 2016 - Effects of Torrefaction Temperature on Pyrolysis Vapor Products of Woody and Herbaceous Feedstocks. Anne K. Starace, Robert J. Evans, D...
4 downloads 13 Views 884KB Size
Article pubs.acs.org/EF

Effects of Torrefaction Temperature on Pyrolysis Vapor Products of Woody and Herbaceous Feedstocks Anne K. Starace,* Robert J. Evans, David D. Lee, and Daniel L. Carpenter National Renewable Energy Laboratory, 15013 Denver West Parkway, Golden, Colorado 80401, United States S Supporting Information *

ABSTRACT: A variety of hardwood, softwood, and herbaceous feedstocks (oak, southern yellow pine mix, loblolly pine, pinyon−juniper mix, and switchgrass) were each torrefied at 200, 250, and 300 °C. Each of the feedstocks was pyrolyzed and the resulting vapors were analyzed with a molecular beam mass spectrometer (py-MBMS). Compositional analysis was used to measure the total lignin content of three of the feedstocks (southern yellow pine, softwood; oak, hardwood; and switchgrass, herbaceous) before and after torrefaction at 300 °C, and large differences in the fraction of lignin lost during torrefaction were found between feedstocks, with oak having the largest decrease in lignin during torrefaction and switchgrass having the least. It is hypothesized that these differences in the thermal degradation are due to, in part, the different ratios of S, G, and H lignins in the feedstocks. Additionally, the torrefaction of kraft lignin was studied using thermogravimetric analysis coupled with Fourier transform infrared spectroscopy (TGA−FTIR) and attenuated total reflectance Fourier transform infrared spectroscopy (ATRFTIR).

1. INTRODUCTION Torrefaction (heating in an inert gas atmosphere between 200 and 300 °C)1 is used to improve storage and handling of biomass as well as to improve conversion to desirable products, such as fuel precursors. In contrast to drying, which simply removes absorbed water, torrefaction is intended to also remove −OH groups, thus creating a more stable, hydrophobic feedstock.2,3 Additionally, torrefaction increases energy density, decreasing transportation costs per Btu,4,5 and results in a feedstock that is easier to grind,6,7pelletize,2 and feed into a reactor8 (pyrolyzer or gasifier). In an integrated process, waste heat from other unit operations can be used to supply the heat necessary for torrefaction.9 In some cases, the volatile gases produced during torrefaction can be combusted to provide or offset the heat required for continued torrefaction or drying the biomass prior to torrefaction.10 Analysis of the oil created from pyrolysis of torrefied material has found that a torrefied feedstock results in a less acidic11−13 but more viscous14 pyrolysis oil than a nontorrefied feedstock. Generally, char yield increases with increasing torrefaction temperature of the feedstock.11,14,15 In addition to studying the physical and bulk chemical changes caused by torrefaction, work has been done to characterize in more detail the chemical changes that take place during torrefaction. During torrefaction, the evolution of water, acetic acid, and furfural have been measured in both woody biomass4 and corn stover.13 Furans, phenols, alcohols, and additional acids have also been detected as condensable products evolved during torrefaction.14 Compositional analysis of torrefied lignocellulosic biomass has found that the proportion of lignin in the biomass increases with increasing torrefaction temperature in pine14−16 and poplar.17 Further study into the chemical changes of lignin wrought by torrefaction by Park et al. using Fourier transform infrared spectroscopy (FTIR) has suggested that during © 2016 American Chemical Society

torrefaction the ether bond in lignin is broken and the lignin fragments recombine via a carbon−carbon bond.18 Work by Ru et al. found that pyrolysis products from lignin containing side branches decreased with increasing torrefaction temperature, suggesting that during torrefaction side branches of lignin are broken.19 Using carbon NMR, Melkior et al. found that in beech wood the etherified lignin signal decreased with increasing torrefaction temperature while the nonetherified lignin signal increased. This suggests, like the work by Park et al., that the ether polymer linker in lignin is broken during torrefaction, but does not necessarily indicate that carbon− carbon bonds are forming between lignin monomers. Melkior et al. also found demethoxylation of the lignin by torrefaction, which occurred at lower temperatures for S lignin than for G lignin,20 and Ru et al. found a decrease in alkane C−H bonds after torrefaction, which could be due, in part, to the demethoxylation of lignin.19 Rousset et al. found that the S/ G ratio of the lignin in beech wood remained almost constant after torrefaction at 220 °C, but decreased significantly after torrefaction at 280 °C.21 Since S lignin has an additional methoxy group, the decrease in S/G lignin could be from S lignin converting to G lignin through demethoxylation, consistent with the findings of Melkior. Alternatively, the decrease in S/G ratio could be due to S lignin as a whole decomposing more rapidly than G lignin during torrefaction at 280 °C. Liu et al. found in pyrolysis of a lignin model polymer that demethoxylation occurred at temperatures above 400 °C,22 suggesting that decomposition is significantly altered by the biomass matrix. In this work, we compare the pyrolysis vapor composition of feedstock samples torrefied at different temperatures. ComposiReceived: February 2, 2016 Revised: May 21, 2016 Published: June 17, 2016 5677

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683

Article

Energy & Fuels

mill. Principal component analysis (PCA) was performed with The Unscrambler X software from CAMO Software. Compositional analysis to determine the amount of lignin was performed following the standard Laboratory Analytical Procedures for Biomass Compositional Analysis.25 To study the gases produced during torrefaction, the feedstocks were heated in a Setaram Setsys Evolution TGA and the gases eluted during the heating were measured via a heated transfer line set to 200 °C to a Nicollette 6700 FTIR with a gas cell attachment heated to 225 °C.

tional analysis of the feedstocks before and after torrefaction at 300 °C was conducted, and large differences in the amount of lignin volatilization from torrefaction between feedstocks was found. We looked further into the changes wrought on the lignin fraction of the biomass by studying kraft lignin using thermogravimetric analysis coupled with Fourier transform infrared spectroscopy (TGA−FTIR) and attenuated total reflectance Fourier transform infrared spectroscopy (ATRFTIR).

3. RESULTS AND DISCUSSION 3.1. Mass Spectrometric Characterization of Pyrolysis Vapors. Mass spectra (10−200 m/z) of the pyrolysis vapors produced at 500 °C from the samples torrefied at 200, 250, and 300 °C were collected. Due to large fluctuations in the low m/z region, the data was normalized to the total ion counts (TIC) from 50 to 200 m/z. Spectra from 50 to 200 m/z were analyzed with principal component analysis (PCA) using The Unscrambler X. The effect of torrefaction temperature on the pyrolysis vapor composition was subtle compared to the effect from the variation in feedstock; thus the data did not cluster with torrefaction temperature. To more easily interpret and visualize the data, the TIC-normalized data was then analyzed by plotting the intensities of peaks previously identified as pyrolysis products of lignocellulosic biomass24 as a function of torrefaction temperature for each feedstock. In the case of 60, 110, 114, 120, 124, 138, 144, 150, 162, 164, 166, and 178 amu peaks, no trend was seen with torrefaction temperature that was consistent between the feedstocks except for m/z 126 and 180. These are shown in Figure 1 and Figure 2, respectively, where

2. MATERIALS AND METHODS The feedstocks studied in this work, each received from Idaho National Laboratory, were loblolly pine (LLP), pinyon−juniper mix (PJ), southern yellow pine (SYP), oak (O) and switchgrass (SG). The species names and harvest locations are shown in Table 1. The oak

Table 1. Species Names and Harvest Locations of Feedstocks Used abbrev

name

species

LLP PJ

loblolly pine pinyon− juniper mix southern yellow pine oak switchgrass

Pinus taeda Juniperus osteosperma and Pinus monophylla various species from Green Circle Bio Energy Quercus alba Panicum virgatum

SYP O SG

harvest location Butler, AL Beaver, UT Jackson County, FL Monroe County, KY Reno County, KS

and pine samples were debarked. The pinyon−juniper and switchgrass were whole biomass samples. Drying was performed in air and torrefaction was performed in nitrogen. The samples were torrefied for 1/2 h under a flow of nitrogen in fixed beds in small batches (hundreds of milligrams in a TGA and a few grams in a tube furnace) so that heat transfer was not a concern. All feedstocks were milled to 2 mm or smaller prior to torrefaction, and thus we expect no significant heat transfer limitations from the outside to the inside of the particle in the torrefaction process.23 Kraft lignin was purchased from SigmaAldrich (part no. 370959). While kraft lignin, which had an S/G ratio of 0.14, is not an ideal representation of the lignin in woody and herbaceous biomass, it still provided insight into transformations occurring in lignin during torrefaction. Generation and analysis of the pyrolysis vapors from the feedstocks were performed using a pyrolyzer coupled with a molecular beam mass spectrometer (MBMS) as described in detail elsewhere.24 Briefly, biomass was placed in a quartz boat that was inserted via a sealing side arm into an inner, annular reactor in a furnace at 500 °C. The inner portion of the reactor had a flow of 400 sccm helium, while the outer portion had a flow of 4 slm helium and 40 sccm argon (tracer gas). The outer gas flow serves two purposes: (1) it puts the system at a positive pressure preventing air from entering the reactor during insertion of the biomass and (2) downstream of the biomass insertion point, the inner flow mixes with the outer flow, thus diluting the pyrolysis products and minimizing subsequent reactions between pyrolysis products, prior to sampling and analysis with the MBMS. Products were ionized using an electron impact ionization source set at 22.5 eV before being mass separated with an Extrel quadrupole and detected with a conversion dynode/electron multiplier detector assembly. Normalization to the total ion counts was used to correct for any changes in the total mass spectrometer signal so that comparisons of the relative amounts of pyrolysis products could be made, even though the absolute amounts of the products were not quantified. Attenuated total reflectance (ATR) Fourier transform infrared (FTIR) spectroscopy measurements were made on the solid material prior to pyrolysis using a Nicollette 6700 spectrometer with a SMART iTR diamond button attachment. Prior to ATR-FTIR measurements, the feedstock samples were made into a powder with a Wig-L-Bug ball

Figure 1. Intensity of 126 amu mass spectrometer signal (normalized to total ion counts) from pyrolysis vapors (pyrolysis temperature of 500 °C) of loblolly pine (LLP), pinyon−juniper (PJ), southern yellow pine (SYP), oak (O), and switchgrass (SG) torrefied at 200 °C (left, blue bar), 250 °C (middle, red bar) and 300 °C (right, green bar). Each bar is the average of two to three feedstock samples. Uncertainty bars are ±1 standard deviation.

the uncertainty bars indicate ±1 standard deviation from the three replicate samples. Since the spectra have been normalized to total ion counts, the graphs indicate change in the proportions of pyrolysis vapor products relative to the total vapor yield, not absolute quantities. In all cases, the 126 amu signal is highest for the feedstocks torrefied at 300 °C, looking at the average values only. Considering the uncertainty in the data, one sees that the 126 amu signal is higher for the feedstock torrefied at 300 °C than the feedstock torrefied at 200 °C in the case of LLP, PJ, and SG. The 180 signal is lowest for the feedstocks torrefied at 300 °C in all cases except for the switchgrass, where the relative amount of 180 signal is the same for each torrefaction temperature within the uncertainty of the measurement. The 126 amu signal is likely from 55678

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683

Article

Energy & Fuels

Figure 3. Estimate of relative total lignin products in pyrolysis vapors as a function of torrefaction temperature for each feedstock. Uncertainty bars are ±1 standard deviation.

Figure 2. Intensity of 180 amu mass spectrometer signal (normalized to total ion counts) from pyrolysis vapors (pyrolysis temperature of 500 °C) of loblolly pine (LLP), pinyon−juniper (PJ), southern yellow pine (SYP), oak (O) and switchgrass (SG) torrefied at 200 °C (left, blue bar), 250 °C (middle, red bar), and 300 °C (right, green bar). Each bar is the average of two to three feedstock samples. Uncertainty bars are ±1 standard deviation.

hydroxymethyl-2-furfural or 2-methyl-3-hydroxy-4-pyrone derived from cellulose.26 The relative increase in the 126 amu signal with increasing torrefaction temperature suggests that the concentration of cellulose-derived pyrolysis products increases with increasing torrefaction temperature. This in turn suggests that hemicellulose and/or lignin is decomposed to a larger extent than cellulose during torrefaction. The 180 signal is associated with the pyrolysis of lignin, likely from coniferyl alcohol or vinylsyringol.26 The decrease in the 180 amu signal suggests that the relative proportion of volatile lignin remaining in the feedstocks decreases with higher torrefaction temperature, due to either decomposition of lignin to vapors during torrefaction or cross-linking of lignin during torrefaction to form a product that is not pyrolyzed at 500 °C, but rather converts to char. While 126 and 180 amu are prominent peaks in the pyrolysis vapors of cellulose and lignin, respectively, other peaks associated with lignin and cellulose were examined for a more detailed analysis. Table 2 shows the peaks associated with lignin and cellulose pyrolysis products as a whole. The division between S and G lignins is taken from the work of Hu and Sykes et al. on switchgrass27 and poplar.26 Figures 3, 4, 5, and 6 were made by summing the TIC-normalized peak intensities of the respective categories listed in Table 2: total lignin, S lignin,

Figure 4. S lignin content as a function of torrefaction temperature for each feedstock. Uncertainty bars are ±1 standard deviation.

Figure 5. G lignin content, as measured by summing peaks indicated in Table 2, as a function of torrefaction temperature for each feedstock. Uncertainty bars are ±1 standard deviation.

Table 2. Prominent Peaks in py-MBMS Spectra Associated with Total Lignin, S Lignin, G Lignin, and Total Cellulosea total lignin

S lignin

G lignin

total cellulose

124 137 138 150 152 154 164 167 168 178 180 182 194

154 167 168 182 194

124 137 138 152 164 178

60 73 85 98 126 144

Figure 6. Cellulose content, as measured by summing peaks indicated in Table 2, as a function of torrefaction temperature for each feedstock.

G lignin, and total cellulose, respectively. Note that the largest lignin signal, 180 amu, has not been included in the plots of S and G lignin signals since it contributes to both.

a

Values given for each from py-MBMS data is the sum of the signals after the entire spectrum has been TIC-normalized. 5679

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683

Article

Energy & Fuels

would. Also included in Table 3 is the total weight percent loss from the torrefaction at 300 °C for 30 min. Oak and southern yellow pine contained roughly the same amount of lignin prior to torrefaction, 33.4 and 34.1 wt %, respectively. However, oak lost 48.5 wt % of its total lignin after torrefaction at 300 °C while southern yellow pine lost 20.9 wt %. Switchgrass had the lowest amount of lignin in its raw form, but also lost the lowest percent of its lignin, 4.5 wt %, during torrefaction at 300 °C. Looking at the total weight loss during torrefaction, we see that the weight loss of oak lignin is a higher percent than the total weight loss, indicating that lignin decomposes proportionately more than cellulose and hemicellulose in oak. In the case of southern yellow pine, the total percent weight loss during torrefaction is about twice the percent weight loss of lignin, indicating that the cellulose/ hemicellulose fraction of SYP volatilizes proportionally more than the lignin fraction. The switchgrass sample exhibits the same trend (lignin volatizing less than hemicellulose/cellulose), but to a much greater extent, with the lignin weight percent loss of almost an order of magnitude less than the total weight loss during torrefaction. This large discrepancy between the percentages of lignin lost during torrefaction could be due to different forms of lignin being more easily volatilized at 300 °C than others. However, it is also important to note that, in addition to differences in lignin composition, there are other differences between the feedstocks, such as ash content. The result that lignin volatilized proportionally less than hemicellulose and cellulose in the pine and switchgrass feedstocks is consistent with findings in the literature where the weight percent of biomass that is lignin increases with increasing torrefaction temperature. However, these studies have been conducted on pine and poplar feedstocks. Work by Pierre et al. measured changes in acid-soluble material (ASM) and acid-insoluble material (AIM) at different torrefaction temperatures for pine and oak.28 In the work by Pierre et al., ASM consisted of mostly hemicellose and cellulose while AIM consisted of mainly lignin. They found that the fraction of AIM in the feedstock increased with increasing torrefaction temperature for both pine and oak, suggesting that in both feedstocks lignin volatilized relatively less than cellulose and hemicellulose during torrefaction. However, the authors also pointed out that previous work by Avat found that reactions between furfural from hemicellulose and phenols from lignin produced acidinsoluble products. Thus, this form of degradation of lignin would increase the apparent concentration of lignin if the concentration of undegraded lignin was measured by the amount of AIM. It is also important to note that while softwoods consistently have high levels of G lignin, the ratio of H:G:S lignin in oak feedstocks is highly variable.29 Pyrolysis molecular beam mass spectrometry (py-MBMS) of the untorrefied feedstocks was measured and can be used to estimate the relative amounts of H, S, and G lignins in the feedstocks, as seen in Figure 7. Spectra were, as before, normalized to total ion counts. Calibrations were not run with these samples, so the results are only semiquantitative. The relative amount of H lignin was measured by the 120 m/z signal. The relative amounts of S and G lignins were measured by summing the representative peaks of S and G lignins, as delineated in Table 2, respectively. Uncertainty bars are ±1 standard deviation of duplicate measurements. The three types of lignin, H, G, and S, contain zero, one, and two methoxy groups, respectively. In this limited data set, the feedstock with the most S lignin lost the most total lignin and

The total lignin content is lowest for the highest torrefaction temperature in the case of PJ and SYP, as shown in Figure 3. In the case of LLP, O, and SG, within the uncertainty of the measurement, the average total lignin content does not change with torrefaction temperature. As expected, Figure 4 shows that oak had a higher relative S lignin content than the other feedstocks, but within the uncertainty of the measurement, there is no change in S lignin content with torrefaction temperature. Considering the G lignin signal alone, Figure 5, we see that the G lignin concentration remains relatively constant, considering the uncertainty of the measurement, with torrefaction temperature. This could indicate that the amounts of S and G lignins are difficult to determine without the contribution from the 180 m/z signal, often the most intense lignin-derived peak, included. The total cellulose signal, shown in Figure 6, is greatest in the sample torrefied at 300 °C for the LLP, SYP, SG, and O feedstocks, though considering the uncertainty of the measurement, it is unclear whether or not the proportion of cellulose pyrolysis products changes with torrefaction temperature. The pinyon−juniper sample shows the lowest amount of total cellulose, perhaps because it is the only feedstock including bark, and within the uncertainty of the measurement, the amount of cellulose products does not change with torrefaction temperature. From these results one sees that the effects of torrefaction temperature are small compared to the effects of the variation in feedstock. Nonetheless, a slight increase in the proportion of cellulose-derived pyolysis vapors and a slight decrease in the lignin-derived vapor pyrolysis products are seen in the initial analysis of the feedstocks, with the exception of SG, which does not show a decrease in the 180 amu lignin signal. This suggests that lignin changes proportionally more than does cellulose with increasing torrefaction temperature. This runs counter to the common finding in the literature that cellulose changes proportionately more than lignin during torrefaction and thus led us to study the changes in lignin during torrefaction in more detail. 3.2. Compositional Analysis. Compositional analysis was performed on the untorrefied feedstocks and those torrefied at 300 °C for one hardwood (oak), one softwood (southern yellow pine), and one herbaceous feedstock (switchgrass), the results of which are shown in Table 3. The amount of lignin is Table 3. Weight Percent of Lignin in Original Feedstocks and Those Torrefied at 300 °C, Measured on an As-Received Basis lignin content per 100 g of untorrefied biomass feedstock

no torrefaction

torrefied at 300 °C

lignin wt % loss

total wt % loss from torrefaction at 300 °C

O SYP SG

33.4 34.1 17.3

17.2 27.0 16.5

48.5 20.9 4.4

46.4 43.2 41.5

presented based on the mass of the feedstock (on an asreceived basis) before torrefaction occurred. That is, the mass loss during torrefaction is taken into account. The method of determining the amount of lignin is a mass-based one;25 thus cross-linking between lignin monomers would not change the value of the measurement, though thermal degradation or cross-linking between lignin monomers and sugar monomers 5680

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683

Article

Energy & Fuels

had a mass loss during torrefaction at 300 °C for 30 min of 22%, which is very similar to the value for southern yellow pine (20.9%) after torrefaction at the same conditions. The gases evolved from kraft lignin at 300 °C under a flow of nitrogen were measured with FTIR and are shown in Figure 8.

Figure 8. FTIR spectra of evolved gases from torrefaction of kraft lignin at 300 °C.

Spectra of the gases evolved from O, SYP, and SG during torrefaction are in the Supporting Information and show features typical of what is already presented in the literature (loss of water, CO2, acidic acid, aliphatic and aromatic hydrocarbons). In the evolved gases from kraft lignin heated under nitrogen at 300 °C, there are signatures of water (4000− 3500 and 1870−1300 cm−1), carbon dioxide (2390−2270 and 667 cm−1), and carbon monoxide (2230−2030 cm−1). In addition, signatures of C−H stretches from alkanes (2990− 2780 cm−1) and alkenes (3100−3000 cm−1), CO stretches (1810−1670 cm−1), and aromatic carbon stretches (1530− 1480 cm−1) are seen. The peaks in the region from 1300 to 1100 cm−1 could be from ethers and/or from carboxylic acids and esters. ATR-FTIR spectra of the kraft lignin taken before and after torrefaction at 300 °C, shown in Figure 9, indicate a loss of

Figure 7. Relative amounts of (A) H, (B) G, and (C) S lignins in untorrefied oak (O), southern yellow pine (SYP), and switchgrass (SG). Uncertainty bars are ±1 standard deviation of duplicate measurements.

the feedstock with the most H lignin lost the least total lignin during torrefaction at 300 °C. While further study is needed to test this, these results suggest a hypothesis that the methoxy groups are key to the degradation mechanism of lignin at 300 °C, and thus the recalcitrance of lignin to thermal degradation from highest to lowest is H > G > S. This is consistent with work by Jakab et al. which found that the volatiles produced from various wood lignins increased with increasing hydroxyl and methoxy groups.30 Further examination of additional feedstocks with varying H:G:S ratios is needed to test this hypothesis. Additionally, the study of various isolated lignin samples would aid in the understanding of any interactions between the biomass components. This work highlights that changes wrought by torrefaction are dependent on the feedstock being torrefied and that the cost-to-benefit analysis of torrefaction as a pretreatment needs to take the feedstock type into account. This work also provides a hypothesis for more systematically predicting the effect torrefaction will have on a feedstock (by analysis of the methoxy content of the lignin) before analysis of each feedstock individually. 3.3. TGA−FTIR and ATR-FTIR of Lignin. To further elucidate the fate of the lignin fraction of the feedstocks during torrefaction, additional measurements were performed on kraft lignin alone. The kraft lignin had an S to G ratio of 0.14, indicating it contains more G lignin than S lignin. Kraft lignin

Figure 9. ATR-FTIR spectra of kraft lignin as received from Alfa Aesar (black, solid line) and torrefied at 300 °C (red, dashed−dotted line).

methyl groups when the lignin is torrefied, evidenced by the decrease in the 1260 cm−1 peak, a strong methyl peak, in the torrefied kraft lignin. The peak at 1510 cm−1, likely from aromatic carbon−carbon bond stretches, decreases with torrefaction. The aromatic ethers signal, around 1220 cm−1, also decreases with torrefaction, confirming the finding of both Park et al. and Melkior et al. that ether linkages of lignin are broken during torrefaction. Like the torrefied whole biomass, 5681

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683

Article

Energy & Fuels

relatively larger CO signal from both aliphatic and aromatic CO bonds than do the other feedstocks. Oak stands out in that while all the other feedstocks have similar spectra from 850 to 710 cm−1 regardless of the torrefaction temperature, the oak feedstock has a higher absorbance from the sample torrefied at 300 °C than samples torrefied at 250 and 200 °C in that region. A decrease in signal from aromatic ethers and methyl groups, which was seen in the ATR-FTIR of lignin with no pretreatment vs lignin torrefied at 300 °C, is not apparent in the spectra of the whole feedstocks. This could be due to the fact that the spectra of whole biomass contains signal from hemicellulose, cellulose, and lignin and thus there are many overlapping peaks and trends in an individual component of biomass that are difficult to observe. ATR-FTIR data of the whole biomass can, however, show how the torrefaction temperature affects the different feedstocks differently. Both pine samples (LLP and SYP) show very little difference in their spectra with different torrefaction temperatures. PJ showed very little difference between samples torrefied at 200 and 250 °C, but a sharp change upon torrefaction at 300 °C. The O and SG samples show more steady changes in the spectra with torrefaction temperature. We see that there is a general trend of decreasing O−H stretch signal (3570−3040 cm−1) with increasing torrefaction temperature for all feedstocks, increasing aliphatic CO stretch signal (1780−1660 cm−1) for the LLP and PJ feedstocks, and increasing aromatic CO stretch signal (1660−1540 cm−1) for the LLP, PJ, SYP, and O feedstocks. The scatter in the duplicate spectra of O and SG, however, alerts us that small changes, such as the difference in the O−H stretch peak between the 200 and 250 °C SG, may be within the uncertainty of the measurement and also that the seeming reproducibility in the pine samples at different torrefaction temperatures could be coincidental.

the O−H stretch signal is smaller in the torrefied lignin than in the nontorrefied lignin. Lastly, below about 1500 wavenumbers, the torrefied lignin sample has less-defined peaks (i.e., shallower troughs between peaks) than the nontorrefied lignin, which may indicate an increase in amorphous character. 3.4. ATR-FTIR Analysis of Torrefied Whole Feedstocks. ATR-FTIR analysis was also performed on the whole torrefied feedstocks, as shown in Figure 10. Spectra of O and SG

4. CONCLUSIONS One hardwood (oak/O), three softwoods (southern yellow pine/SYP, loblolly pine/LLP, and pinyon−juniper/PJ), and one herbaceous (switchgrass/SG) feedstock were torrefied at 200, 250, and 300 °C. These feedstocks were analyzed using FTIR spectroscopy and by MBMS analysis of their pyrolysis vapors. These initial data sets pointed us toward additional studies focusing on the extent of volatilization of lignin during torrefaction and its relationship to the type of lignin (H, G, or S). Here we find that lignin decomposes relatively less than the other biomass components (hemicellulose and cellulose) in softwood (southern yellow pine) and herbaceous (switchgrass) feedstocks, which is consistent with findings in the literature where the weight percent of biomass that is lignin increases with increasing torrefaction temperature for pine and poplar feedstocks. In our work, we have found that oak loses proportionally more lignin than hemicellulose and cellulose combined during torrefaction at 300 °C. We hypothesis that this is due to the much larger proportion of S lignin, which has more methoxy groups than any other type of lignin, in oak compared with the other feedstocks. The methoxy groups may be a key contributor to the decomposition of lignin, perhaps because they can be an electron donating group. Additional analysis on the torrefaction of kraft lignin at 300 °C was also performed.

Figure 10. ATR-FTIR spectra of loblolly pine (LLP), pinyon−juniper (PJ), southern yellow pine (SYP), oak (O), and switchgrass (SG) feedstocks torrefied at 200 °C (red traces), 250 °C (blue traces), and 300 °C (black traces). Spectra underwent SNV transformation. Spectra of O and SG feedstocks were collected in triplicate.

feedstocks were collected in triplicate for each torrefaction temperature. Particle size, sample density, and sample thickness affect the ATR-FTIR signal. To minimize variations caused by the aforementioned effects, and to more easily see the influence of feedstock and torrefaction temperature on the chemical composition, we performed a standard normal variate transformation on the ATR-FTIR spectra.31 The most notable difference between the feedstocks is found in PJ, which has a sharp peak around 780 cm−1 which is not observed in any of the other feedstocks. This could be because PJ is the only feedstock with bark. PJ also had a greater distinction between symmetric and asymmetric C−H stretches than any of the other feedstocks. We also note that PJ has a 5682

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683

Article

Energy & Fuels



(17) Chen, M.; Wang, J.; Min, F.; Liu, S.; Chen, M. 2011 International Conference on Materials for Renewable Energy & Environment; IEEE: 2011, pp 322−326. DOI: 10.1109/ICMREE.2011.5930822. (18) Park, J.; Meng, J.; Lim, K. H.; Rojas, O. J.; Park, S. J. Anal. Appl. Pyrolysis 2013, 100, 199−206. (19) Ru, B.; Wang, S.; Dai, G.; Zhang, L. Energy Fuels 2015, 29 (9), 5865−5874. (20) Melkior, T.; Jacob, S.; Gerbaud, G.; Hediger, S.; Le Pape, L.; Bonnefois, L.; Bardet, M. Fuel 2012, 92 (1), 271−280. (21) Rousset, P.; Lapierre, C.; Pollet, B.; Quirino, W.; Perre, P. Ann. For. Sci. 2009, 66 (1), 110−110. (22) Liu, J.; Wu, S.; Lou, R. BioResources 2011, 6, 1079−1093. (23) Bates, R. B.; Ghoniem, A. F. Fuel 2014, 137, 216−229. (24) Evans, R.; Milne, T. Energy Fuels 1987, 1 (2), 123−137. (25) Sluiter, J.; Sluiter, A. Summative Mass Closure Laboratory Analytical Procedure (LAP) Review and Integration; NREL/TP-51048087; NREL: 2011; Vol. 2011. (26) Sykes, R.; Yung, M.; Novaes, E.; Kirst, M.; Peter, G.; Davis, M. In Biofuels: Methods and Protocols; Humana Press: 2009; pp 169−183. (27) Hu, Z.; Sykes, R.; Davis, M. F.; Charles Brummer, E.; Ragauskas, A. J. Bioresour. Technol. 2010, 101 (9), 3253−3257. (28) Pierre, F.; Almeida, G.; Brito, J. O.; Perré, P. BioResources 2011, 6 (2), 1204−1218. (29) Obst, B. J. R.; Landucci, L. L. Holzforschung 1986, 40, 87−92. (30) Jakab, E.; Faix, O.; Till, F. J. Anal. Appl. Pyrolysis 1997, 40−41, 171−186. (31) Barnes, R. J.; Dhanoa, M. S.; Lister, S. J. Appl. Spectrosc. 1989, 43 (5), 772−777.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.energyfuels.6b00267. FTIR spectra from off-gases of torrefaction of oak, pine, and switchgrass (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: 303-384-6363. Tel.: 303275-4377. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy under Contract No. DE-AC36-08GO28308 with the National Renewable Energy Laboratory. Funding was provided by U.S. DOE Office of Energy Efficiency and Renewable Energy Bioenergy Technologies Office. The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a nonexclusive, paid up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for U.S. Government purposes. We would like to thank Justin Sluiter and Ryan Ness for performing the compositional analysis and Mike Griffin for performing additional ATR-FTIR measurements. We also thank INL for providing the feedstocks.



REFERENCES

(1) While there are some discrepancies in the literature about the temperature range of torrefaction, we find that this is the most common definition. (2) Shankar Tumuluru, J.; Sokhansanj, S.; Hess, J.; Wright, C.; Boardman, R. Ind. Biotechnol. 2011, 7 (5), 384−401. (3) Sadaka, S.; Negi, S. Environ. Prog. Sustainable Energy 2009, 28 (3), 427−434. (4) Medic, D.; Darr, M.; Shah, A.; Potter, B.; Zimmerman, J. Fuel 2012, 91 (1), 147−154. (5) Ren, S.; Lei, H.; Wang, L.; Bu, Q.; Wei, Y.; Liang, J.; Liu, Y.; Julson, J.; Chen, S.; Wu, J.; Ruan, R. Energy Fuels 2012, 26 (9), 5936− 5943. (6) van der Stelt, M. J. C.; Gerhauser, H.; Kiel, J. H. a.; Ptasinski, K. J. Biomass Bioenergy 2011, 35 (9), 3748−3762. (7) Westover, T. L.; Phanphanich, M.; Clark, M. L.; Rowe, S. R.; Egan, S. E.; Zacher, A. H.; Santosa, D. Biofuels 2013, 4 (1), 45−61. (8) Carpenter, D. L.; Westover, T.; Czernik, S.; Jablonski, W. Green Chem. 2014, 16, 384−406. (9) Prins, M. J.; Ptasinski, K. J.; Janssen, F. J. J. G. Energy 2006, 31 (15), 3458−3470. (10) Svoboda, K.; Pohořelý, M.; Hartman, M.; Martinec, J. Fuel Process. Technol. 2009, 90 (5), 629−635. (11) Chen, D.; Zhou, J.; Zhang, Q. Energy Fuels 2014, 28, 5857− 5863. (12) Boateng, A. A.; Mullen, C. A. J. Anal. Appl. Pyrolysis 2013, 100, 95−102. (13) Ren, S.; Lei, H.; Wang, L.; Yadavalli, G.; Liu, Y.; Julson, J. J. Anal. Appl. Pyrolysis 2014, 108, 248−253. (14) Zheng, A.; Zhao, Z.; Chang, S.; Huang, Z.; He, F.; Li, H. Energy Fuels 2012, 26, 2968−2974. (15) Meng, J.; Park, J.; Tilotta, D.; Park, S. Bioresour. Technol. 2012, 111, 439−446. (16) Shankar Tumuluru, J.; Sokhansanj, S.; Hess, J. R.; Wright, C. T.; Boardman, R. D. Ind. Biotechnol. 2011, 7 (5), 384−401. 5683

DOI: 10.1021/acs.energyfuels.6b00267 Energy Fuels 2016, 30, 5677−5683