Identifying Softwoods and Hardwoods by Infrared Spectroscopy

derived from fast-growing softwoods, whereas the woods that are so popular with wood ..... Owen, N. L.; Thomas, D. W. Appl. Spectrosc. 1989, 43, 451,...
4 downloads 0 Views 84KB Size
In the Laboratory

Identifying Softwoods and Hardwoods by Infrared Spectroscopy Brady Barker and Noel L. Owen* Department of Chemistry and Biochemistry, Brigham Young University, Provo, UT 84602; *[email protected]

Trees may be classified botanically into two general categories, gymnosperms and angiosperms, on the basis of whether the seeds are enclosed in the ovary of a flower. Plants with naked seeds (gymnosperms) are typically cone-bearing trees with needlelike evergreen leaves, whereas plants with

a

OH

HO

O HO

O

HO O

OH

CH2OH

O

HOOC CH3O HO

O

OH

O

HO O

OH

HO OH

HO

O

O O

O

OH

HO CH3O

O O

O

HO

c

Chemical Composition of Wood

O

O

O O

OH

CH2OH

OH O

O HO

O

HO

O OH HO

O

O

COOH

CH3O

b

CH 2OH

OH

CH 2OH

O

O

O

HO

O HO

COOCH3

HOCH2

O

HO

CHR CHOH

CH CHCHO OCH3 CH2OH O

HC

CH2OH O

CHOH

CH

OCH3 CH3O

CH2OH O

OR

O

CHOH CH2OH

HO

CH

CH HC

O

CHOH O

CH3O OCH3 O

CH3O OCH3 HOCH2 CH O

HOCH2 O

CH

O

The chemical structure of wood is complex and varies from species to species as well as with morphological location within the plant (1). However, the general composition has certain characteristics that appear common to all species. There are three main polymeric components: cellulose (a linear polymer of glucose units), hemicellulose (a polysaccharide polymer composed of several different sugar units, together with various derivatized side groups), and lignin (a threedimensional polymer containing aromatic and other unsaturated groups) (Fig. 1). The percentage of each of these polymers tends to vary a little from wood to wood, and angiosperms and gymnosperms show some significant differences in certain features of the lignin polymer (Fig. 2). Softwood lignins tend to be composed mainly of coniferyl alcohol derivatives (Fig. 2b), with a small component based on coumaryl alcohol (Fig. 2a), whereas hardwoods tend to have lignins composed of coniferyl- and sinapyl-based alcohols (Fig. 2c), with small amounts of coumaryl alcohol derivatives. In addition to these components, most woods contain small quantities of oils, tannins, polyphenolic compounds, and minerals.

CHOH

CH2OH HOCH2

CH3O

enclosed seeds (angiosperms) have broad leaves, which they usually lose in the fall. In forestry, gymnosperms and angiosperms are referred to respectively as softwoods and hardwoods. Much of the timber used by the construction industry is derived from fast-growing softwoods, whereas the woods that are so popular with wood turners and carvers are usually hardwoods. Botanists can often identify softwoods and hardwoods from the color of the wood and from its fibrous structure, but distinguishing between the two classes when different cuts (cross-sectional, tangential, etc.) are compared can create challenges for even the most experienced botanist. In this paper we describe a simple and fairly definitive method based on a quick analysis of the infrared spectrum of the wood sample.

CHOH

OCH3 CH O C O

CH 2OH HC

CH3O

CH

HC CH

CH

OCH3 OH

OH

Figure 1. The main polymer components of wood: (a) cellulose, a linear polymer of 1,4-β-D-glucopyranose units; (b) hemicellulose, a branched-chain polymer composed of different sugars and side groups (the figure shows a 1,4-β-D-xylopyranose fragment with 4-O-methylD-glucopyranosyl uronic acid side chains); (c) lignin, a 3-dimensional phenolic polymer based on hydroxy- and methoxy-substituted phenylpropane units (the figure shown is typical of a softwood).

1706

CH 2OH

CH 2OH HC

H 3CO

H 3CO OH

a

OH

b

OCH 3 OH

c

Figure 2. The three monomeric precursors of lignin: (a) p-coumaryl alcohol, (b) coniferyl alcohol, (c) sinapyl alcohol.

Journal of Chemical Education • Vol. 76 No. 12 December 1999 • JChemEd.chem.wisc.edu

In the Laboratory

Experimental Details In a 1989 publication, it was shown that several features of the infrared spectra of angiosperms and gymnosperms can be used to distinguish between these two types of wood (2). In this paper we describe the use of the simplest of these features to extend the study to include a total of 45 woods. Infrared spectra can be derived from wood samples using experimental techniques such as direct absorbance, diffuse reflectance, or attenuated total reflection (ATR), and analyzing wood samples has become quicker, easier, and more convenient with the introduction of new sampling devices. If wood slivers can be produced with a thickness less than 100 µm, then direct transmittance infrared studies will give good results. For thicker samples a reflectance technique is needed, and diffuse reflectance is very convenient. Diffuse reflectance does have the tendency to distort the shape of some infrared bands (largely owing to the presence of some specular reflectance), but the carbonyl and main aromatic bands of wood are generally unaffected (3). It is possible that the newer ATR devices currently being produced may enable infrared analyses of wood to be conducted even more easily, simply, and reproducibly (4 ).

a

b

Figure 3. Infrared spectra of (a) a typical softwood (cedar), and (b) a typical hardwood (maple).

All our spectra were obtained using a Mattson Polaris spectrometer, with an accumulation of 30 scans at 4 cm{1 resolution. We used two different diffuse reflectance units (The Collector from SpectraTech and The Selector from Specac) in an attempt to reduce the specular reflectance component, and both gave reproducible results with the samples that we studied. Results Although the overall quality of the diffuse reflectance spectra obtained from the many wood samples differed considerably, in virtually every instance it was possible to clearly identify and measure a number of prominent absorptions— in particular, the absorption associated with the carbonyl group (ca. 1740 cm{1) and that associated with the aromatic breathing mode found in lignin (ca. 1510 cm{1). Free carbonyl groups occur abundantly within the polymer components of wood, but they tend to predominate in the branched-chain hemicellulose polymer. Infrared spectra of isolated lignin and holocellulose (cellulose + hemicellulose) confirm this conclusion in that the carbonyl absorption is much stronger and more prominent in the latter (2). Thus, since the ratio of lignin to holocellulose in general tends to be higher in softwoods, it would be expected that the intensity of the carbonyl peak relative to the other fingerprint absorptions might be stronger in the hardwoods. This conclusion is illustrated in Figure 3, which shows the infrared spectra of cedar, a softwood, and maple, a common U.S. hardwood. However, since the overall quality of the infrared spectra of wood samples varies significantly with sample technique and the physical nature of the sample, the relative intensity of the carbonyl absorption is not a foolproof method of distinguishing between softwoods and hardwoods. A far better method is to measure the wavenumber value of the carbonyl absorption. This value tends to depend on the type of carbonyl group giving rise to the absorption. Since the lignin/ holocellulose ratio is different in softwoods and hardwoods, it is instructive to compare the types of carbonyl-containing compounds commonly observed in the two types of polymers. Hemicellulose is largely composed of glucose, mannose, galactose, xylose, arabinose, 4-O-methylglucuronic acid, and galacturonic acid moieties, and the associated carbonyl groups tend to be on acid or ester groups. Esters and monomeric acids tend to absorb over a range of about 50 cm{1, centered at about 1740–1750 cm{1, while the C=O stretch for hydrogenbonded acids tend to be slightly lower. Lignin, on the other hand, is a phenolic polymer consisting of a three-dimensional array of hydroxy- and methoxy-substituted phenylpropane units. The carbonyl groups in lignin tend to be part of aldehyde or ketone groups, and they tend to absorb at wavenumber values lower than 1740 cm{1. The fact that all woods comprise a mixture of the two polymers (together with small quantities of other entities such as terpenes, oils, and waxes that might contain carbonyl groups) will result in a broad absorption, the position of which will be weighted in favor of the major component. We have studied the spectra of 45 woods. Table 1 summarizes our findings regarding the position of the carbonyl absorption for each sample. It can be seen that there is a difference on average of about 10 cm{1 in the position of the carbonyl band for a hardwood and a softwood. This spectral

JChemEd.chem.wisc.edu • Vol. 76 No. 12 December 1999 • Journal of Chemical Education

1707

In the Laboratory Carbonyl

Table 1. Characteristic Infrared Absorptions of Softwoods and Hardwoods Wood Sample

Scientific Name

Infrared Band/cm{1 Carbonyl

Lignin 1513

Pseudotsuga menziesii

1738

Pinus sp.

1734

1510

Ponderosa pine

Pinus ponderosa

1739

1508

Juniper

Juniperus virginiana

1737

1512

Redwood

Sequoia sempervirens

1737

1515

Dawn redwood

Metasequoia glyptostroboides

1743

1508

Alaska cedar

Chamaecyparis nootkatensis

1737

1512

Aromatic red cedar

Calocedrus decurrens

1735

1512

Sitka spruce

Picea sitchensis

1733

1512

Colorado blue spruce

Picea pungens

1739

1512

European larch

Larix decidua

1739

1508

Almaciga

Agathis sp.

1739

1514

HARDWOODS Lignum vitae

Guaiacum sp.

1738

1515

Rosewood

Dalbergia sp.

1736

1512

Tulipwood

Dalbergia frutescens

1744

1511

Pernambuco

Caesalpinia echinata

1742

1512

Pau setim

Euxylophora pargensis

1747

1508

Holly

Ilex opaca

1744

1503

Birch

Betula paperifera

1744

1505

Balsa

Ochroma pyramidale

1743

1504

Sweet gum

Liquidambar styraciflua

1742

1505

Soft maple

Acer sp.

1741

1503

Maple

Acer grandidentatum

1751

1508

Norway maple

Acer platanoides

1748

1506

Sycamore maple

Acer pseudoplatanus

1747

1504

Sycamore

Platanus acerifolia

1749

1504

Quaking aspen

Populus tremuloides

1745

1504

Quaking aspen (attacked by fungus)

P. tremuloides

1733

1508

Myrtle

Umbellularia californica

1741

1504

Butternut

Juglans cinerea

1743

1504

Pear

Pyrus sp.

1745

1503

Koa

Acacia koa

1741

1507

Carpathian elm

Ulmus sp.

1741

1508

White oak

Quercus sp.

1749

1449

English walnut

Juglans regia

1748

1504

Lemon

Calycophyllum candidissimum

1749

1508

Hackberry

Celtis occidentalis

1749

1504

Pacific yew

Taxus brevifolia

1747

1516

Zelkova

Zelkova serrata

1749

1508

1749

1504

Rose leaf mountain ash Sorbus aucuparia Bradford pear

Pyrus caleryana

1753

1506

Padauk

Pterocarpus sp.

1747

1506

Poplar

Populus sp.

1747

1502

Black cherry

Prunus sp.

1743

1504

Red alder

Alnus sp.

1745

1506

distinction holds for virtually all the samples studied, but the few exceptions to this trend (which are discussed later in this paper) are illustrative of some interesting chemistry. The position of the ca. 1510 cm{1 lignin peak is also indicative of the type of wood. Table 1 shows that virtually all softwoods show this absorption above 1510 cm{1, whereas hardwoods tend to absorb below 1510 cm{1. Typically there exists a difference of about 10 cm{1 in the lignin absorption 1708

Lignin

Wavenumber

Douglas fir Pine

Wavenumber

SOFTWOODS

Hardwood

Softwood

Figure 4. Results of statistical analysis of data for softwoods and hardwoods. The median for each set of data is shown by the central line in the rectangles; the rectangles represent the two inner quartiles; the outer quartiles are represented by the solid lines. The spots represent a few outlying values, but they had very little effect on the results and all of them were included in the statistical analysis. In one instance (hardwood, lignin data) one of the outlying values (white oak, at 1449 cm{1) was omitted from the figure. The numbers refer to the following wood samples (Table 1): 2, pine; 5, redwood; 6, dawn redwood; 13, lignum vitae; 14, rosewood; 23, maple; 38, Pacific yew; 41, Bradford pear; 43, poplar.

between the two species. Mono- substituted benzene derivatives tend to show this ring deformation mode in the range 1510–1480 cm{1, whereas the corresponding absorption in para-disubstituted and 1,2,4-trisubstituted derivatives shifts toward higher wavenumber values (1520–1480 cm{1) (5). 1,3,4,5-Tetrasubstituted benzenes, on the other hand, tend to absorb below 1500 cm{1. This trend helps to explain why hardwoods show a lower wavenumber value for this absorption than softwoods, since hardwoods have tetrasubstituted rings as part of their lignin structure. Since wood lignin has a very complex polymer structure, the observed absorptions represent only the average wavenumber values for a range of benzene derivatives, but the spectroscopic trend seen in softwoods and hardwoods appears to correlate with the known differences between them in lignin structure.

Statistical Analysis A two-sample t test was carried out for both the carbonyl band data and the lignin band data. The 12 softwoods gave a mean carbonyl group value of 1737.5 cm{1 with a standard deviation of 2.7 cm{1, while the 32 hardwoods have a mean value of 1745.2 cm{1 with a standard deviation of 3.9 cm{1.

Journal of Chemical Education • Vol. 76 No. 12 December 1999 • JChemEd.chem.wisc.edu

In the Laboratory

The difference in the mean of the observed carbonyl absorption values between hardwoods and softwoods is statistically highly significant (t = {7.49; p < .0001). For the lignin band data, the average softwood wavenumber is 1511.3 cm{1 with a standard deviation of 2.4 cm{1, and for the hardwoods, the average is 1504 cm{1 with a standard deviation of 10.7 cm-1. Here the difference in the mean wavenumber values is also statistically significant, with a t value of 3.35 ( p = .0019). Figure 4 illustrates the differences in the wavenumber values between the two types of wood graphically.

The Exceptions Normal aspen wood shows a carbonyl absorption at 1745 cm{1, typical of a hardwood species, but the aspen sample that has been attacked by a fungus shows the same absorption at 1733 cm{1. This suggest that the fungus has selectively modified the wood polymer composition, possibly preferentially destroying the hemicellulose and thus causing the wood carbonyl component to resemble that of a softwood. Interestingly, the lignin peak of the fungus-attacked sample is typical of a hardwood, again indicating that the fungus has not affected the lignin. This wood sample was not included in the statistical analysis. Rosewood and lignum vitae represent tropical hardwoods, and they have a very waxy or oily appearance. Samples of both woods were chemically analyzed at the Forest Products Laboratory in Madison, Wisconsin. Rosewood had an unusually high extractive content, whereas the lignin content of lignum vitae was significantly higher than that of other hardwoods. Neither of these samples fits the expected pattern for the carbonyl or the lignin absorption. With the exception of the examples described above, only one softwood (dawn redwood) is anomalous in its carbonyl frequency (1743 cm{1), whereas all the hardwoods studied absorb above 1740 cm{1. There are slightly more exceptions to the general pattern shown by the lignin 1510 cm{1 absorption. Three softwoods (ponderosa pine, dawn redwood and European larch) all absorb at 1508 cm{1, whereas three hardwoods (in addition to the two mentioned above) show higher-than-normal wavenumber values.

When exceptions to both carbonyl and lignin absorptions are taken together as indicators, only dawn redwood, lignum vitae, and rosewood remain. The reasons for the latter two have already been outlined, and we do not have the lignin/ holocellulose analysis data for dawn redwood. However, one unusual feature of this cone-bearing softwood is that it sheds its leaves every winter, like as a typical hardwood. When taken in concert, the wavenumber values of the infrared absorptions of the carbonyl and lignin absorptions of wood samples provide a convenient and reasonably definitive method of determining whether the sample represents a softwood or a hardwood (three exceptions were found in 45 samples). Small, relatively inexpensive Fourier transform infrared spectrometers are rapidly becoming ubiquitous, not only in chemical laboratories but in biology departments, and with the current developments in sampling devices, categorizing wood samples by analyzing their infrared spectra could become routine. Acknowledgments We are grateful to Kimball Harper and Blaine Furniss (Botany Department, BYU) for their help in identifying some of the wood samples and in supplying some of the softwood samples, and to Kip Christiansen (Technology Education and Construction Management department, BYU) for supplying us with some of the hardwoods. The help and advice of Dennis Tolley (BYU Statistics Department) is also gratefully acknowledged. Literature Cited 1. Parkham, R. A.; Gray, R. L. In The Chemistry of Solid Wood; Rowell, R., Ed.; Advances in Chemistry 207; American Chemical Society: Washington, DC, 1984. 2. Owen, N. L.; Thomas, D. W. Appl. Spectrosc. 1989, 43, 451, 455. 3. Anderson, T. H.; Weaver, F. W.; Owen, N. L. J. Mol. Struct. 1991, 249, 257–275. 4. Coates, J. Am. Lab. 1997, April, 22C–22J. 5. Colthup, N. B.; Daly, L. H.; Wiberley, S. E. Introduction to Infrared and Raman Spectroscopy; Academic: San Diego, 1990; p 425.

JChemEd.chem.wisc.edu • Vol. 76 No. 12 December 1999 • Journal of Chemical Education

1709