ARTICLE pubs.acs.org/ac
Expanding the Library of Secondary Ions That Distinguish Lignin and Polysaccharides in Time-of-Flight Secondary Ion Mass Spectrometry Analysis of Wood Robyn E. Goacher, Dragica Jeremic, and Emma R. Master* Department of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto Ontario, Canada M5S 3E5
bS Supporting Information ABSTRACT: Extracted pine (Pinus spp.) wood and the holocellulose and cellulose fractions of pine were analyzed by time-of-flight secondary ion mass spectrometry (ToF-SIMS). The main sources of variation among wood constituents were elucidated by principal component analysis (PCA). Peaks characteristic of lignin or polysaccharides were identified through the combination of high mass resolution analyses of pine fractions and high lateral resolution image analyses distinguishing the lignin-rich middle lamella from the secondary cell wall layers in solid wood cross-sections. A collection of peaks was compiled which (1) extends the library of characteristic lignin and polysaccharide secondary ions in wood, (2) can be applied to both high and nominal mass resolution spectra, and (3) is free from peaks that contraindicate between wood components. The removal of additional peaks to avoid mass interferences with common contaminants was also successful. Many of the characteristic peaks were high-intensity fingerprint ions below m/z 100, which provided for rapid analysis of the lignin and polysaccharide biopolymers in woody samples. The analysis also identified important mass interferences with previously reported wood ions.
S
econdary ion mass spectrometry (SIMS) is a sensitive tool for the mass spectrometric analysis of the surfaces of solid samples. The analysis of organic materials, including biopolymers, is benefited by rapid spectral acquisition over a large mass range using a time-of-flight (ToF) analyzer, as well as the high spatial and mass resolutions, reasonable count rates, and low damage offered by pulsed liquid metal cluster primary ion sources. These attributes, along with minimal sample preparation, make ToF-SIMS using cluster primary ions an attractive technique for distinguishing the spectral signals of lignin and polysaccharides within woody materials. The development of ToF-SIMS as an analytical method to evaluate wood samples could be applied to elucidate the distribution of molecular components within individual wood samples and across different species and facilitate the characterization of engineered wood fiber or the impacts of pretreatment processes. Saito et al. have produced several elegant reports employing dehydrogenation polymers1,2 and dimeric model compounds3 to identify characteristic positive secondary ions that arise from the three main monolignols comprising lignin. In their work, p-hydroxyphenyl (H) units are characterized by peaks at m/z 107 and 121, guaiacyl (G) units are identified by peaks at m/z 137 and 151, and syringyl (S) units are characterized by peaks at m/z 167 and 181 (Table 1).1-3 These ions also appear in the spectra of wood samples and milled wood lignin.1-3 The m/z 137 and 151 ions were further shown to arise from intermolecular ether linkages, and not only from terminal phenolic subunits.3 With the r 2010 American Chemical Society
identification of peaks specific to the lignin monomer units, it has been suggested that the ratio of these peaks can serve as a quick way to identify the S/G ratio of different wood species and regions of a given sample.3 Several ToF-SIMS investigations of wood surfaces have examined the distribution of lignin alongside salt and metal ions.4,5 While resulting images of wood surfaces clearly illustrate the location of various inorganic ions, the S/N ratio is poor for the organic species. Tokareva et al.5 were able to image the different localization of H-lignin (m/z 107 and 121) and G-lignin (m/z 137 and 151) subunits on transverse and longitudinal sections of aspen and spruce, and they also imaged carbohydrates, citing Fardim and Duran6 for the use of m/z 115 as a characteristic ion for xylan (hemicellulose) and m/z 127 and 145 for cellulose. The lateral resolution of ion images in Tokareva et al.'s work was however insufficient to resolve carbohydrates and lignin in particular cell wall layers. Extractives and lignins on the surfaces of pulps have also been studied by ToF-SIMS.6,7 Koljonen et al.7 suggested that the m/z 137 and 151 guaiacyl lignin peaks could be used along with the generic aromatic peak at m/z 77 (C6H5þ) to describe structural changes in lignin and to measure lignin content on wood fiber surfaces. Fardim and Duran6 provide an extensive list of ions for extractives and also list tentative assignments for xylan (hemicellulose) at m/z 115 and 133, cellulose at m/z 127 Received: August 31, 2010 Accepted: December 8, 2010 Published: December 29, 2010 804
dx.doi.org/10.1021/ac1023028 | Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
and Soxhlet extracted in 9:1 (v/v) acetone-water for over 48 h. The holocellulose component was obtained by NaClO2 treatment of extracted sawdust in acetic acid.13 R-Cellulose was purified from holocellulose by successive extractions with 17.5 and 9.45% NaOH according to the TAPPI T203 cm-99 method.14 Approximately 200 mg of air-dried samples were pelletized in a 1 cm diameter die under 22 kg/cm2 pressure using a hydraulic press. Klason lignin was also prepared according to TAPPI standard method T222. However, since preliminary analysis of this sample revealed significant transformation of native lignin, it was excluded from further consideration. Preparation of Pine Cross-Sections. Blocks of lodgepole pine (Pinus contorta Dougl.) approximately 4 13 45 mm3 (radial tangential longitudinal) were Soxhlet extracted in 9:1 (v/v) acetone-water for over 48 h and air-dried. Sections were cut from the block using an acetone-cleaned Xacto knife and were microtomed using a Leica EM UC6 with an acetonecleaned diamond knife. ToF-SIMS Analysis. All measurements were made with a ToF-SIMS IV instrument (Ion-Tof Gmbh, M€unster, Germany) equipped with a bismuth liquid metal ion source and reflectron-type analyzer with multichannel detector. Bi32þ primary ions at 50 keV were incident upon the sample at 45°. High mass resolution spectra (using high-current bunched conditions) were obtained for four spots on each of the pellets with a pulsed current of ∼0.3 pA and a 128 128 pixel random raster pattern covering a 500 500 μm2 area. High lateral resolution ion images with nominal mass resolution were obtained on wood cross-sections with a pulsed current of ∼0.1 pA Bi32þ and a 256 256 pixel random raster pattern. Details of the different resolutions achieved in these modes of analysis may be found in ref 15. A cycle time of 100 μs provided a mass range from 0 to 750 m/z and was sufficient to collect the full spectra of desorbed secondary ions. The pressure during analysis was maintained between 1 10-8 and 1 10-7 mbar. Ion doses were kept below 1 1012 ions/cm2 to limit sample damage. Low-energy electron flooding (20 eV) was used to reduce sample charging. Spectra were calibrated to the CH3þ, H3Oþ, C2H3þ, and C3H5þ ions using IonSpec v.4.1 software. At m/z 91, the mass resolution (M/ΔM) was ∼30005000 in bunched mode and ∼250-300 in burst alignment mode, depending on sample roughness. Principal Component Analysis. Principal component analysis (PCA) is an unsupervised pattern recognition method that is used for objective differentiation of complex spectra. PCA has proven to be a useful method for distinguishing complex samples, including proteins, hydrocarbons, and sugar isomers.10,11 PCA has further been utilized to study the peak purity of mixed samples in liquid chromatography.12 The ability of PCA to provide useful chemical information from relative changes in sample composition, without the need for pure standard samples, makes it an ideal method for the distinction of the complex biopolymers in wood. Accordingly, PCA of pine samples was performed using Matlab software v.7.8.0.347 (The Mathworks, Inc.) with PLS Toolbox v.5.2.1 and MIA Toolbox (Eigenvector Research Inc.). The individual spectra from replicate locations on the sawdust pellets or from pixels in the images were normalized to unit area, and data at each mass were mean centered.
Table 1. Secondary Ions Reported in the Literature as Characteristic of Lignin and Polysaccharides in Wood theoretical mass
ion assignment þ
107.050
C7H7O
121.029
C7H5O2þ C8H9Oþ C8H9O2þ C8H7O3þ C9H11O3þ C9H9O4þ C6H7O3þ C6H9O4þ C5H7O3þ C5H9O4þ
121.065 137.060 151.040 167.071 181.050 127.040 145.050 115.040 133.050
classification [reference] H lignin units2 H lignin units2 G lignin units1-3 G lignin units1-3 S lignin units1-3 S lignin units1-3 cellulose6,8,9 cellulose6,8,9 xylan6,8,9 xylan6,8,9
and 145, and lignin at m/z 137, 151, 167, and 181. In a paper coauthored by Fardim,8 the same ions are used to identify cellulose and xylan with a citation to a fast atom bombardment mass spectrometry (FAB-MS) study of carbohydrates.9 Since FAB-MS and SIMS have similar ionization mechanisms, these are probably good assignments of ions generated from pure carbohydrates in ToF-SIMS. Ions below m/z 100 are often the most intense ions in ToFSIMS spectra generated from wood samples.2,5 Despite this, our survey of wood sample analysis by ToF-SIMS suggests that ions characteristic of polysaccharides and lignin in wood have not been identified in the low-mass region below m/z 100. This is probably because reliable identification of low-mass ions can be complicated by the production of these fragment ions from many parent compounds. However, since the major polymeric components of wood have remarkably different chemical compositions, it is likely that low-mass fragment ions for lignin and carbohydrates can be distinguished and used to define more detailed compositional fingerprints for these biopolymers. Our aims were therefore (1) to expand the library of ToFSIMS secondary ions that distinguish lignin and polysaccharides, including ions in the low-mass region to facilitate faster analysis of woody samples, and (2) to develop a robust peak list for the imaging of wood by identifying mass interferences between ions arising from the individual wood biopolymers and between wood biopolymers and common contaminants. Red pine and lodgepole pine species were chosen for analysis because of their dominance in northern boreal forests and industrial relevance. Two approaches were taken to distinguish ions in the ToF-SIMS spectra arising from lignin and polysaccharide biopolymers in softwood. The first approach was to analyze samples with different relative proportions of lignin, cellulose, and hemicellulose. Samples were prepared using established chemical treatments to substantially delignify extracted pine sawdust to holocellulose and subsequently remove hemicellulose and remaining lignin to prepare a cellulose fraction. The second approach was to generate high-resolution ToFSIMS images of solid pine cross-sections so that the known variation in lignin and polysaccharide concentrations across cell walls could be related to relative peak intensities. The ToF-SIMS data were analyzed by PCA to highlight differences among spectra.
’ MATERIALS AND METHODS
’ RESULTS AND DISCUSSION
Preparation of Pine Fraction Pellets. Red pine (Pinus resinosa Ait.) was ground, sieved through an 80 mesh screen,
Common Surface Contaminants. When building a peak library, it is important to analyze samples that are free from 805
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
Table 2. List of Peaks Grouped According to Whether They Characterized Polysaccharides or Lignin in PCA Models of Pine Fractions and Pine Cross-Sections Images assignments of major
classification by PCA
classification by PCA
contributing ions (deviations in ppm)
of pine fractionsa
of pine images
nominal mass
exact mass
51 63
51.021 63.022
C4H3þ (-50) C5H3þ (-20)
L L
L L
65
65.038
C5H5þ (-20)
L*
L*
67
67.056
C5H7þ (20)
L
L
77
77.037
C6H5þ (-25)
L*
L*
79
79.056
C6H7þ (5)
L
L*
91
91.051
C7H7þ (60)
L*
L*
93
93.073
C7H9þ (50)
L
L
95 105 c
95.092 105.071
C7H11þ (85) C8H9þ (10)
L L
L L
107
107.044
C7H7Oþ (-50)
L
L
115
115.045
C5H7O3þ (30); C9H7þ (10)
L
L*
128
128.051
C6H8O3þ (30)
L
L
131 b
131.049
C5H7O4þ (80); C9H7Oþ (35)
L
L
137
137.063
C8H9O2þ (5)
L*
L*
147 b
147.068
C9H7O2þ (140); C6H11O4þ (5); C10H11Oþ (-110)
L
L
d
151.049 152.050
C8H7O3þ (35); [C5H11O5þ (-100)] C8H8O3þ (20)
L* (H*) L
L* L
153
153.049
C8H9O3þ (-30)
L
L
165
165.059
C9H9O3þ (15)
L
L*
189
189.059
C11H9O3þ (20)
L
L
PS
Group 1: Peaks Confirmed as Characteristic of Lignin
151 152
Group 2: Peaks Confirmed as Characteristic of Polysaccharides 15 b
15.023
CH3þ (-30)
PS
19
19.019
H3Oþ (60)
PS*
PS*
31
31.019
CH3Oþ (15)
PS*
PS*
44 d
44.023 44.053
C13CH3Oþ (-40) C213CH7þ (-140); C2H6Nþ (45)
PS L*
PS
45 b
45.034
C2H5Oþ (-5)
PS* (H)
PS
47
47.013
CH3O2 (-15)
PS
PS
59 b
59.015
C2H3O2þ (15)
PS (H)
PS
60
60.021
C2H4O2þ (-10)
PS*
PS
61.030
C2H5O2þ (10)
PS*
PS*
c
71.014
C3H3O2þ (-5)
PS* (H*)
PS*
73 b 81 d
73.032 81.036
C3H5O2þ (80) C5H5Oþ (0)
PS* (H) PS*
PS* PS*
81.075
C6H9þ (60)
L
83
83.013
C4H3O2 (-10)
PS
PS
85
85.034
C4H5O2þ (60)
PS*
PS*
87
87.051
C4H7O2þ (70)
PS*
PS*
97
97.032
C5H5O2þ (25)
PS*
PS*
99
99.049
C5H7O2þ (90)
PS
PS
101 109
101.029 109.033
C4H5O3þ (50) C6H5O2þ (35)
PS PS
PS PS
113 c
113.026
C5H5O3þ (-5)
PS (H)
PS
127
127.041
C6H7O3þ (60)
PS* (C*)
PS
145 e
145.061
C6H9O4þ (80)
PS (C)
-
61 71
Group 3: Peaks Not Confirmed by PCA Models of Images due to Neutral Loadings
f
30
30.037
CH4Nþ (80)
L*
-
56
56.058
C3H6Nþ (95); C4H8þ (-130)
L
-
58
58.071
C3H8Nþ (60); C4H10þ (-160)
L*
-
806
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
Table 2. Continued 68
68.060
C4H6Nþ (140); C5H8þ (-75)
L
70
70.074
C4H8Nþ (105); C5H10þ (-75)
L*
-
72
72.090
C4H10Nþ (120); C5H12þ (-50)
L
-
86
86.108
C6H14þ (-30)
L
-
-
Group 4: Peaks Not Confirmed by PCA Models of Images due to Conflicting Assignments 27 29 b,d
27.024 29.003
C2H3þ (5) CHOþ (10)
PS PS*
L/PS N/A
29.040
C2H5þ (25)
L
g
39.022
C3H3þ (-40)
PS (H)
N/A
41 g
41.039
C3H5þ (-25)
PS*
N/A
42.009
C2H2Oþ (-45)
PS
-
42.042
C3H6þ (-90)
L
39
42 d 43 b,d
43.019
C2H3Oþ (10)
PS* (H*)
N/A
53
43.055 53.038
C3H7þ (10) C4H5þ (-25)
L* PS
L/PS
55 d
55.019
C3H3Oþ (20)
H*
N/A
55.055
C4H7þ (10)
L*
57 d,c 69 d
103
57.035
C3H5Oþ (15)
PS* (H*)
57.073
C4H9þ (50)
L
68.995
C3HO2 (-35)
PS (H)
69.036
C4H5Oþ (20)
PS*
C5H9þ (45) C4H7O3þ (53); C8H7þ (-95)
L* PS
69.074 103.045
N/A N/A
L
Classification key: L = lignin, PS = polysaccharides, H = holocellulose, C = cellulose, - = neither polysaccharide nor lignin, N/A = not in peak list for PCA (see the Distinguishing Cellulose from Pine Samples section). An asterisk (*) denotes strong loadings. b Possible mass interference with PDMS. c Possible mass interference with phthalate contaminants. d Mass interference at nominal mass resolution between individual peaks within the mass envelope that contraindicated lignin and polysaccharides. e The low signal-to-noise ratio of the m/z 145 ion images probably resulted in the low significance of this peak in the PCA models of the cross-section images. Visual inspection of the m/z 145 ion images revealed that this peak followed the pattern of the other polysaccharide peaks so it was included in group 2 as a confirmed polysaccharide peak. f The peaks in group 3 appear split between nitrogen-containing ions (which might arise from ash in the wood) or saturated hydrocarbon fragments (possibly arising from bonded propanoid side chains linking the benzyl groups of lignin). g Mass interference at nominal mass resolution with Kþ. a
73 distinguished polysaccharides, and m/z 147 distinguished lignin. Again, the absence of high-mass PDMS peaks and the lack of consistent behavior of the low-mass PDMS peaks indicate that this surface contaminant did not influence the results of the PCA modeling. Nevertheless, given the possibility of common contaminants interfering with ToF SIMS analysis of lignocellulose, ions with high potential for mass interference between phthalate or PDMS contaminants and wood-descriptive ions are noted in Table 2. Furthermore, the removal of these peaks from PCA modeling sets is discussed in the context of SIMS imaging in ToF-SIMS Image aAnalysis of Lodgepole Pine Cross-Sections. Spectral Analysis of Pelletized Red Pine and Its Corresponding Holocellulose and Cellulose Fractions. A library comprising 884 peaks was created from the summed spectra of the pine, holocellulose, and cellulose samples by selecting all peaks under m/z 700 that had over 500 integral counts. PCA performed on this peak library resulted in good separation of pine and the holocellulose and cellulose fractions (Figure 1). As expected, the first principal component (PC) separated the three pellets along a gradient from the pure cellulose sample toward the more lignin-rich pine sample. The holocellulose fraction, which had some remaining lignin as indicated by its tan color, scored slightly toward the pine sample on PC1. The second PC revealed the relatively higher hemicellulose content of the holocellulose sample by separating this sample from the pine and cellulose samples.
contamination. ToF-SIMS can be a powerful technique for identifying common surface contaminants because the fingerprint ions of different contaminants are readily distinguished in the mass spectra. By comparison, the contamination of a wood surface is difficult to identify using the complementary technique of X-ray photoelectron spectroscopy because organic contaminants contribute to the elemental (O 1s, C 1s) and chemical shift (C-C/C-H, C-O, CdO) signals used for the quantification of carbohydrates and lignin. In the samples utilized for this work, there was no evidence in the ToF-SIMS spectra of significant contamination from the two most common surface contaminants: phthalates and poly(dimethy siloxane) (PDMS). The positive ion peaks that identify phthalate contamination occur at m/z 57, 71, 105, 113, 149, 261, 279, and 391.16 The high-mass peaks at m/z 261, 279, and 391 had insignificant intensity in the sample spectra, indicating no phthalate contamination. Furthermore, the low-mass peaks had divergent behavior, as peaks at m/z 57 and 149 were not assigned to either polysaccharides or lignin, while peaks at m/z 71 and 113 described polysaccharides and the peak at m/z 105 described lignin. If these peaks were due to phthalate contamination, the set of ions should be distinguished on one principal component by either positive or negative contributions (not both). The positive ion peaks descriptive of PDMS are known to occur at m/z 28, 43, 59, 73, 147, 207, and 221. Of these, the ions at m/z 28, 207, and 221 were of insignificant intensity, m/z 43 was present but not confirmed to distinguish lignin or polysaccharides, m/z 59 and 807
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
Figure 1. (A) Loadings on PC1, (B) loadings on PC2, and (C) scores plot with 95% confidence ellipses for the PCA model describing ToF-SIMS spectra of extracted red pine and its holocellulose and cellulose fractions. Loading plots are truncated at m/z 200 since this range includes all significant loadings; all peaks above 500 integral counts and below m/z 700 were included in the model.
Distinguishing Cellulose from Pine Samples. In agreement with literature assignments, m/z 127 and 145 peaks were characteristic of cellulose and m/z 77, 137, and 151 peaks, characteristic of lignin, distinguished extracted pine (Figure 1A). The predominance of G-type lignin in pine was also confirmed by the insignificant loadings of peaks at m/z 107, 121, 167, and 181 for H and S lignin. In addition to previously reported ion peaks assigned for wood samples, 64 new peaks were classified by their loadings on PC1 as characteristic of polysaccharides or lignin (Table 2). Ions that distinguished pine from cellulose preparations on PC1 likely correspond to lignin (Figure 1A, negative loadings). It is also possible that some of these peaks represent other chemical differences generated during the preparation of the holocellulose and cellulose fractions (e.g., removal of salts, ash, contaminants, or lignin-polysaccharide complexes). Examples of inorganic peaks that strongly characterized pine were m/z 56.964 (CaOHþ) and m/z 110.971 (NaC2O4þ). The identified lignin peaks included the previously reported m/z 137 and 151 ions as well as several well-established aromatic and benzyl ions: m/z 51.021 (C4H3þ), 65.038 (C5H5þ), 77.037 (C6H5þ), and 91.051 (C7H7þ). Notable contributions from other peaks including some at higher mass are tentatively assigned in Table 2. Some difficulty was
encountered in establishing a cutoff for peaks characterizing lignin because most of the higher mass peaks indicated pine. A general threshold of |peak loading| > 0.02 was applied for classification. All peaks characteristic of cellulose corresponded to the formula CxHyOz excepting the H3Oþ ion, which results from water elimination during the fragmentation of carbohydrates in SIMS analysis. Ions at m/z 29.003, 31.019, 57.035, 61.030, 69.036, 73.032, 85.034, 87.051, 103.045, 127.041, and 145.061 correspond to peaks previously identified by Berman et al. using C-13 labeled glucose in their study of monosaccharide stereoisomers.10 Berman et al. also reported significant sugar ions at m/z 91, 111, 115, 163 and 171.10 In the current analysis, m/z 111, 163, and 171 did not distinguish the cellulose sample, and m/z 91 and 115 distinguished samples containing lignin. The C3H7O3þ saccharide peak at m/z 91.040 was visible as a low-intensity shoulder on the C7H7þ lignin peak at m/z 91.053. Likewise, the C5H7O3þ saccharide peak at m/z 115.040 appeared as a low-intensity shoulder on the C9H7þ lignin peak at m/z 115.055. For both m/z 91 and 115, the PCA was dominated by the lignin contribution. These results illustrate the possibility of mass interferences between the polymeric components of wood and raise questions regarding the validity of frequent assignments of m/z 115 to xylan. 808
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
Better mass resolution could theoretically distinguish ions with m/z near 91 and 115 that originate from saccharides or lignin. However, the widespread occurrence of peak broadening caused by surface topography and charge accumulation with electrically insulating wood samples make it challenging to improve mass resolution. In analyses with nominal mass resolution, these mass interferences are unavoidable. Distinguishing Holocellulose from Cellulose and Pine Samples. The fraction of hemicellulose in each sample should be highest in the holocellulose pellet, so the separation of holocellulose from pine and cellulose on PC2 likely describes hemicellulose (Figure 1B). However, the confirmation of these ions requires further validation using isolated hemicelluloses. The peaks that contributed most to the separation of holocellulose from the other samples were m/z 18.040, 43.019, 45.034, 55.019, 57.035, 68.995, 71.014, 73.032, 151.049, and 196.210 (Figure 1B, negative loadings). The origins of the peaks at m/z 18 (NH4þ) and 196 (C14H28þ) are unclear. Since these ions were not intense in the spectra of the original pine fraction or in the images of the pine cross-sections, they were regarded as contaminants in the holocellulose preparation. Aside from the ions at m/z 55.019 and 151.049, which helped to characterize pine on PC1, the remaining peaks describing holocellulose were identified on PC1 as characteristic of cellulose. The m/z 55.019 peak (C3H3Oþ) was detected in the spectra of all of the samples but with slightly higher intensity in the pine and holocellulose samples, indicating that this peak arose from the fragmentation of all three biopolymers (lignin, cellulose, and hemicellulose). The m/z 151.049 peak had minimal intensity in the cellulose spectra and probably arose from a combination of ions originating from lignin (C8H7O3þ; exact mass, 151.0395) as well as quasimolecular ions formed by adduction in the sputter plume of free protons to the hemicellulose monomer xylose ([M þ H]þ, C5H11O5þ; exact mass, 151.0606); the mass resolution of this experiment was not sufficient to resolve these components. The greater intensity in the holocellulose sample of some of the polysaccharide peaks (i.e., C2H3Oþ at m/z 43.019) might be the result of additional ions yielded from acetyl groups found in hemicellulose but not in cellulose. Not all peaks that distinguished pine or cellulose on PC1 contributed to the characterization of these samples on PC2. Instead, PC2 described only those ions that distinguished the pine and the cellulose samples as compared to the holocellulose sample. Peaks attributed to pine were primarily inorganic (m/z 38.936 (Caþ), 43.055, 56.964 (CaOHþ), 91.051, 98.993 (NaCO4þ), 110.971 (NaC2O4þ), and 137.063), while peaks attributed to cellulose included Naþ (m/z 22.992) and a few carbohydrate peaks (27.024, 31.019, 69.036, 87.051, 127.041, and 145.061). Ions previously used to describe hemicellulose at m/z 115 and 133 correlate to ion structures of C5H7O3þ and C5H9O4þ, respectively. These small ions are characteristic of the xylose ring, which is the same fundamental building block in xylan from both hardwoods and softwoods.6,8,9 However, neither of these peaks distinguished holocellulose from the cellulose and pine samples on PC2. This result, along with the dominance of the lignin contribution to m/z 115 on PC1, suggests that while these ions may be used to describe purified carbohydrates,9 they are not unique identifiers for hemicellulose in pine samples. ToF-SIMS Image aAnalysis of Lodgepole Pine CrossSections. High lateral resolution images of lodgepole pine cross-sections were acquired with the goal of distinguishing the lignin-rich middle lamella and cell corners from the cell walls,
which have higher polysaccharide concentrations. The spatial distinction of peaks in the middle lamella and cell walls provides a second method for validating the classification of secondary ions as characteristic of lignin or polysaccharides, which is independent from the analysis of delignified fractions presented in the Spectral Analysis of Pelletized Red Pine and Its Corresponding Holocellulose and Cellulose Fractions section. Benefits of the image analysis are the avoidance of artifacts that might have arisen from the chemical treatments involved in fractionating the pine sawdust. However, the spectra of these high lateral resolution images have nominal mass resolution, preventing the distinction of individual peaks at a given mass. Therefore the results must be interpreted in the light of possible mass interferences. The secondary ion images of the pine cross-sections produced patterns that (1) clearly resembled wood anatomy, with decreased ion intensity in the open lumen and intracellular layers, or middle lamella (Figure 2A-G) or (2) revealed the open lumen but did not have noticeably decreased intensity across the middle lamella (Figure 2H-N). These images reflect the known composition of wood, with polysaccharides concentrated in the cell walls and lignin abundant in the cell walls but present at higher concentration in the middle lamella and cell corners. PCA Modeling of Wood Images. ToF-SIMS spectra of extracted red pine and lodgepole pine species exhibited minor differences, justifying the comparison of lodgepole pine cross-sections with the red pine sawdust fractions. PCA was performed using a peak library consisting of all 297 nominal mass peaks detected under m/z 310. The first 5-7 PCs of the models generated from the cross-section images were dominated by different combinations of a set of seven peaks (m/z 29, 39, 41, 43, 55, 57, and 69); however, corresponding image scores did not clearly resemble the architecture of typical wood cell walls. From the high mass resolution analysis of the pine fractions (Spectral Analysis of Pelletized Red Pine and Its Corresponding Holocellulose and Cellulose Fractions section), it was clear that all of these nominal mass peaks comprised multiple ions that either individually indicated polysaccharides or lignin or interfered with intense Kþ ions (Table 2, group 4, notes C and F). Further within the PCA models (PC 6-8), uniquely assigned lignin and polysaccharide ions identified from analysis of the pine fractions revealed that peaks characteristic of lignin clustered in the middle lamella while peaks characteristic of polysaccharides clustered in the secondary cell wall layers, providing additional evidence needed to support the new peak assignments described in Table 2. Although the seven peaks that dominated the first 5-7 PCs also appeared alongside the nonconflicting ions in the subsequent PCs, these seven peaks did not consistently group with lignin or polysaccharides, and therefore the assignments at these masses could not be confirmed by this analysis. Due to the strong mass interferences and the influence of these peaks on the PCA models, it is recommended that peaks at m/z 29, 39, 41, 43, 55, 57, and 69 be excluded from models built on nominal mass resolution spectra of wood. When the seven contradictory masses were removed from the PCA peak library used for image modeling, the middle lamella and secondary cell walls were distinguished by PC1 (Figure 3); further PCs did not differentiate the middle lamella and secondary cell walls. In general, the higher mass peaks (e.g., m/z 127, 137, 145, and 151) were less significant in the models built on high lateral resolution images than in models describing the pine fractions due to lower S/N ratios in the spectra. 809
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
Figure 2. ToF-SIMS secondary ion images of a representative lodgepole pine cross-section where images A-G represent polysaccharide peaks, images H-N represent lignin peaks, and image O represents total ion counts. Field of view: 68.4 68.4 μm2, tc = total counts in image, and mc = maximum counts per pixel. The acquisition time was 524 s (80 scans).
PCA models built on the imaging data validated 21 nominal mass peaks characteristic of lignin and 23 nominal mass peaks characteristic of polysaccharides. Confirmed lignin peaks were defined as those peaks that distinguished the middle lamella in the cross-section images and also enveloped high mass resolution peaks characteristic of pine (Table 2, group 1). In contrast, confirmed polysaccharide peaks distinguished the secondary cell walls in the images and enveloped high mass resolution peaks characteristic of cellulose or hemicellulose (Table 2, group 2). Image analysis did not identify any significant peaks that had not already been classified as characteristic of lignin or of polysaccharides by the analysis of the pine fractions. Most of the nominal mass peaks validated by the image models enveloped only one peak significant in the separation of pine fractions in the high mass resolution analysis. However, the nominal mass peaks at m/z 44 and 81 characterized polysaccharides in the image models despite their additional enclosure of peaks distinguishing the lignin-rich pine sample in the high mass resolution analysis. Comparison of the relative loadings in the high mass resolution model (Figure 1A) shows that, at m/z 81, the polysaccharide peak influenced the model more strongly than the lignin peak, consistent with the description of the polysaccharide component at nominal mass. By contrast, in the high mass resolution analysis, the lignin peak at m/z 44 influenced the model more strongly than the polysaccharide peak. The lignin peak at m/z 44.053 was one of eight peaks that distinguished pine in the high mass resolution analysis but lacked clear molecular formulas that would be consistent with known lignin structure (Table 2, note E).
The other seven peaks in this group produced noisy ion images and neutral loadings in the PCA models built on the images, preventing their confirmation as lignin characteristic ions (Table 2, group 3). Given the difficulty in assigning ion formulas and the absence of lignin characteristics for the m/z 44.053 and group 3 peaks in the image analysis, it can be postulated that these peaks do not truly characterize lignin but may represent other components (e.g., ash) that were present in the extracted red pine sawdust. Additional research would be required to elucidate the significance and origin of these peaks. Eleven other nominal mass peaks that comprised high mass resolution peaks characteristic of wood fractions could not be confirmed by the PCA models of images (Table 2, group 4). Seven of these peaks were discussed at the start of this section and had obvious mass interferences. Polysaccharide and lignin derived ions near m/z 42 resulted in neutral loading of this peak in PCA models of images. The peak at m/z 103 had marginal loading magnitudes in all PCA models and could not be classified as it was assigned to polysaccharides by pine fraction analysis and to lignin by image analysis. As described below (Peak Exclusion from PCA Modeling of ToF-SIMS Images section), the assignment of peaks at m/z 27 and 53 was influenced by the presence of other peaks that were included in the models. Nevertheless, while PCA analysis of images did not confirm the identity of group 4 ions (Table 2), these ions could distinguish lignin and polysaccharides under high mass resolution conditions. Notably, with the exception of m/z 103, the intensity of group 4 ions was low in the middle lamella (Figure S1 of the Supporting Information), and several peaks (m/z 29.003, 55.019, and 69.036) were 810
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry
ARTICLE
Figure 3. (A) Scores and (B) loadings for PC1 of a PCA model built on a cropped region of the ToF-SIMS imaging data shown in Figure 2. The image scale is in pixels where 50 pixels =13.4 μm. All masses under m/z 310 except for m/z 29, 39, 41, 43, 55, 57, and 69 were included in the model.
similarly assigned to carbohydrates in previous publications.10 Analysis of individual oligosaccharides and native lignin from different wood species would help to confirm the assignments of this group of peaks and will be the subject of future work. Peak Exclusion from PCA Modeling of ToF-SIMS Images. The removal of certain masses from the library of nominal mass peaks supplied for PCA modeling can be an important analytical decision. Removing the seven peaks that had mass interferences between wood biopolymer components greatly improved the ease of PCA model interpretation (Figure 3). Other situations, such as sample contamination or extremely large data sets, may also require trimming of the peak library for PCA. Reduction of the peak library size was easily achieved by removing all of the peaks in the spectrum that did not contribute to the distinction of polysaccharides and lignin (loading magnitude < 0.02). As discussed in the Common Surface Contaminants section, common surface contaminants such as PDMS and phthalates can be identified in ToF-SIMS spectra by the presence of certain characteristic peaks, including peaks at higher mass that do not overlap with wood-indicating peaks. In the event that contamination is evident, it is necessary to remove lower mass peaks that exhibit mass interferences between wood biopolymers and contaminants from the library so that contaminants do not influence the results. Those ions presenting possible overlap with PDMS and/ or phthalate are marked by notes A and B in Table 2. In addition to contaminants, the PCA models at nominal mass resolution can be influenced by nominal mass peaks that include
Figure 4. Summed images for cellulose or polysaccharides (A-C) and lignin (D-F) ions along with their red-blue false color overlays (G-I) and images resulting from the division of the summed lignin and polysaccharide peaks (J-L). The left-hand column (A, D, G, J) shows the cellulose and G-lignin ions previously reported in the literature; the middle column (B, E, H, K) shows the two most intense low-mass ions for polysaccharides and lignin identified in the present work; and the right-hand column (C, F, I, L) shows all confirmed peaks classified in the present work (groups 1 and 2, Table 2). Field of view: 68.4 68.4 μm2, tc = total counts, and mc = maximum counts per pixel. The acquisition time was 459 s (70 scans).
contributions from inorganic species (i.e., alkali metals). The intensity of these peaks can be high and detract from the ability of the PCA models to reveal information about the polymeric components of wood. Inorganic peaks that were influential in the high mass resolution models of the pine fractions (Spectral Analysis of Pelletized Red Pine and Its Corresponding Holocellulose and Cellulose Fractions section) were m/z 23 (Naþ), 39 and 41 (Kþ), 40 (Caþ), 57 (CaOHþ), 63 (Na2OHþ), 81 (Na2Clþ), 98 (Na2C3Oþ), 99 (NaCO4þ), and 111 (NaC2O4þ). PCA of the images was performed again using a trimmed nominal mass resolution peak library generated by removing insignificant peaks and peaks that interfered with PDMS, phthalate, and inorganic species as described above. PCA models of images built using this trimmed peak library distinguished the middle lamella and cell walls on PC1 as clearly as the full model 811
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812
Analytical Chemistry shown in Figure 3. With the exceptions of m/z 27 and 53, which switched from polysaccharide-characteristic with the full peak library to lignin-characteristic with the trimmed library, there were no changes in ion classification upon reduction of the peak list. This result indicated that there was sufficient ability to distinguish lignin and polysaccharides when possible contaminant-interfering peaks were removed from the model library. Utility of Newly Classified Peaks. The expansion of the library of lignin and polysaccharide characteristic peaks allows for much more rapid analysis of lignocellulosic samples (Figure 4). The application of low-mass ions identified through the present analysis clearly improved image clarity and S/N as compared to images that were generated using only previously reported ions. The S/N improvement arises from the addition of more characteristic ions for polysaccharides and lignin as well as the higher intensity of many of the newly characterized low-mass ions. The lack of mass interferences from the peaks identified by the present analysis is also demonstrated by the ability to derive informative images of polysaccharide and lignin distributions using simple sum and division methods (Figure 4I,L).
’ CONCLUSIONS Through the analysis of extracted pine samples by ToF-SIMS and multivariate statistics, the library of characteristic lignin and polysaccharide ions in softwoods was expanded from 6 to 44 ions. Ions in the expanded library were classified with high confidence due to (1) the known reduction in lignin content progressing from samples of pine to holocellulose to cellulose and (2) the localization within the cross-section images of peaks characteristic of lignin concentrated in the middle lamella and cell corners and peaks characteristic of polysaccharides concentrated in the cell walls. The PCA analyses also revealed differences between ion assignments resulting from purified cellulose, xylan, and lignin polymers as compared to whole wood samples. In particular, the previously reported xylan peaks at m/z 115 and 133 did not distinguish hemicellulose and the m/z 115 peak furthermore characterized lignin. The m/z 151 peak attributed to lignin was also enriched in the holocellulose sample, indicating overlap with characteristic xylan ions. Full mass spectra were reduced to a set of ions that were free from major mass interferences that would complicate image analysis under nominal mass resolution. Most of the ions highlighted by the PCA analysis have heretofore not been utilized in the reports of wood analyzed by ToF-SIMS. Many of these were low mass fragment ions with high intensity that could facilitate rapid visualization and relative quantification of lignin and polysaccharides in natural and engineered lignocellulosic materials. The results reported apply to the distinction of lignin and polysaccharides in pine samples, which have been extracted in acetone-water. Caution should be taken when applying these results to unextracted wood as the presence of extractives on the sample surface may introduce mass interferences or change peak proportions, which could affect the classification of the peaks, particularly at nominal mass resolution. Similarly, although the peaks are expected to generally apply to other lignocellulosic materials, it is probable that the proportional intensities of the characteristic peaks will change and/or additional significant peaks will arise in the analysis of different softwoods, hardwoods, and grasses. Further research will elucidate what these differences may be.
ARTICLE
’ ASSOCIATED CONTENT
bS
Supporting Information. Figure S1 showing mages of group 4 ions (Table 2). This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected]. Fax: 416-978-8605.
’ ACKNOWLEDGMENT We thank Professors Elizabeth A. Edwards, Charles A. Mims, and Rana N.S. Sodhi at the University of Toronto for their helpful discussions. The able assistance of Dr. Peter M. Brodersen at Surface-Interface Ontario is also gratefully acknowledged. This work was funded by the Atlantic Innovation Fund and by the Government of Canada through Genome Canada and the Ontario Genomics Institute (Grant 2009-OGI-ABC-1405) as part of the Bioproducts and Enzymes from Environmental Metagenomes (BEEM) project. R.E.G. and D.J. made equal contributions to the research. ’ REFERENCES (1) Saito, K.; Kato, T.; Tsuji, Y.; Fukushima, K. Biomacromolecules 2005, 6, 678–683. (2) Saito, K.; Kato, T.; Takamori, H.; Kishimoto, T.; Yamamoto, A.; Fukushima, K. Appl. Surf. Sci. 2006, 252, 6734–6737. (3) Saito, K.; Kato, T.; Takamori, T.; Fukushima, K. Biomacromolecules 2005, 6, 2688–2696. (4) Saito, K.; Mitsutani, T.; Imai, T.; Matsushita, Y.; Yamamoto, A.; Fukushima, K. Appl. Surf. Sci. 2008, 255, 1088–1091. (5) Tokareva, E. N.; Pranovich, A. V.; Fardim, P.; Daniel, G.; Holmbom, B. Holzforschung 2007, 61, 647–655. (6) Fardim, P.; Duran, N. Colloids Surf., A 2003, 223, 263–276. (7) Koljonen, K.; Oesterberg, M.; Kleen, M.; Fuhrmann, A.; Stenius, P. Cellulose (Dordrecht, Neth.) 2004, 11, 209–224. (8) Freire, C. S. R.; Silvestre, A. J. D.; Pascoal Neto, C.; Gandini, A.; Fardim, P.; Holmbom, B. J. Colloid Interface Sci. 2006, 301, 205–209. (9) Dell, A. Adv. Carbohydr. Chem. Biochem. 1987, 45, 19–72. (10) Berman, E. S. F.; Kulp, K. S.; Knize, M. G.; Wu, L.; Nelson, E. J.; Nelson, D. O.; Wu, K. J. Anal. Chem. 2006, 78, 6497–6503. (11) Graham., D. J.; Wagner, M. S.; Castner, D. G. Appl. Surf. Sci. 2006, 19, 6860–6868. (12) Wiberg, K.; Andersson, M.; Hagman, A.; Jacobsson, S. P. J. Chromatogr., A 2004, 1029, 13–20. (13) Wise, L. E.; Marphy, M.; D'Adieco, A. Pap. Trade J. 1946, 122, 35–43. (14) Technical Association of the Pulp and Paper Industry. TAPPI Standard T203 cm-99 Alpha-, beta- and gamma-cellulose in pulp; TAPPI: Norcross, GA, 2003. (15) Sodhi, R. N. S. Analyst (Cambridge, U.K.) 2004, 129, 483–487. (16) ToF-SIMS: Surface analysis by mass spectrometry; Vickerman, J. C., Briggs, D., Eds.; IMPublications: Chichester, U.K., 2001; 114.
812
dx.doi.org/10.1021/ac1023028 |Anal. Chem. 2011, 83, 804–812