Soil Science Society of America Journal
Mention of trade names or commercial products in this publication is solely for the purpose of
providing specic information and does not impy recommendation or endorsement by the U.S.
Department of Agriculture. USDA is an equal opportunity provider and employer.
Soil Sci. Soc. Am. J. 77:1591–1600
doi:10.2136/sssaj2013.04.0131
Supplemental material is available online for this article.
Received 11 Apr. 2013.
*Corresponding author (francisco.calderon@ars.usda.gov).
© Soil Science Society of America, 5585 Guilford Rd., Madison WI 53711 USA
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by
any means, electronic or mechanical, including photocopying, recording, or any information storage
and retrieval system, without permission in writing from the publisher. Permission for printing and for
reprinting the material contained herein has been obtained by the publisher.
Diffuse-Reectance Fourier-Transform
Mid-Infrared Spectroscopy as a Method of
Characterizing Changes in Soil Organic Matter
Soil Biology & Biochemistry
I
t is challenging to characterize SOM constituents because soils contain a stag-
gering variety of organic C compounds at varied stages of decomposition, from
relatively fresh plant residues to highly processed aromatics and older nonhy-
drolyzable molecules (Stout et al., 1981; Follett et al., 2007). e presence of and in-
teractions with the mineral matrix further complicates interpretation of the results.
Analysis of soil organic C pools involves procedures such as combustion analysis
for CHN, nuclear magnetic resonance spectroscopy (NMR) (Hemminga and Buur-
man, 1997), fumigation–incubation for microbial biomass, wet chemistry extraction
(Steelink, 2002), lignin quantication (evenot et al., 2010), the use of radioiso-
topes for quantication of turnover and C dating (Calderón et al., 2012; Paul et al.,
1997), hydrolysis procedures for quantifying highly recalcitrant C, and incubation–
curve tting for the determination of fast- and slow-turnover C pools (Paul et al.,
Francisco Calderón*
USDA-ARS
Central Great Plains Research Station
40335 County Rd GG
Akron, CO 80520
Michelle Haddix
Richard Conant
Natural Resource Ecology Lab.
Colorado State Univ.
Fort Collins, CO 80523
Kimberly Magrini-Bair
National Renewable Energy Lab.
Golden, CO 80401
Eldor Paul
Natural Resource Ecology Lab.
Colorado State Univ.
Fort Collins, CO 80523
Diffuse-reectance Fourier-transform mid-infrared spectroscopy (MidIR) can
identify the presence of important organic functional groups in soil organic
matter (SOM); however, spectral interpretation needs to be validated to cor-
rectly assess changes in SOM quality and quantity. We amended soils with
known standards, increasing the total C in the sample by 50%, and measured
changes in MidIR spectra. Adenine, casein, cellulose, ergosterol, glucosamine,
glycine, guanine, indole, methionine, palmitic acid, egg protein, chlorophyl-
lin, tannic acid, xylose, urease, and vanillin standards were used. In addition,
corn (Zea mays L.) stalk feedstock and two chars produced at different tem-
peratures were studied. Two soils were used: a Hoytville, OH, soil (2.5% C
and 36% clay) and an Akron, CO, soil (1.5% C and 14% clay). The addition
of standards with >10% N content resulted in increased amide-like absor-
bance at 1670, 1588, and 1513 cm
−1
. Bands at 2970 to 2800, 2200 to 2000,
and 1030 to 1160 cm
−1
were sensitive to added polysaccharide. Protein addi-
tion increased absorption at 2970 to 2800 cm
−1
but also increased the 1691
and 1547 cm
−1
amide bands. Vanillin addition resulted in higher absorbance
at the 1592, 1515, and 1295 cm
−1
aromatic C=C bands. Biochars produced
at 300°C resulted in increased absorbance at carbonyl and aliphatic bands,
while addition of 500°C biochar increased aromatic absorbance. Our results
showed that MidIR is sensitive to relatively small changes in SOM. If assump-
tions about the soil mineralogy are met, specic spectral bands can be used
to follow changes in SOM chemistry.
Abbreviations: MidIR, diffuse-reectance Fourier-transform mid-infrared spectroscopy;
NMR, nuclear magnetic resonance spectroscopy; SOM, soil organic matter.
Published September 20, 2013
1592 Soil Science Society of America Journal
2006). All these techniques have been very useful for expanding
our understanding of SOM chemistry. Recent advances in infra-
red spectroscopy indicate that it will become an important tool for
the study of soils, especially when combined with other technolo-
gies (Davinic et al., 2012).
Spectroscopic techniques like infrared spectroscopy and
NMR have been used to study SOM because both techniques
can detect important organic functional groups. Nuclear mag-
netic resonance spectroscopy can give a semiquantitative under-
standing of carboxylic, aliphatic, and aromatic groups in soils
(Hemminga and Buurman, 1997). Pyrolysis–molecular beam
mass spectrometry (py-MBMS) is another potentially valu-
able way for studying SOM because it can detect high-molec-
ular weight fragments from the nonoxidative breakup of SOM
(Magrini et al., 2007). Infrared and NMR spectra, as well as py-
MBMS data, need to be interpreted judiciously because they all
have the potential for interference or artifacts associated with
soil minerals. e question of how quantitative these techniques
can be has not yet been answered denitively. Nevertheless, the
study of SOM via mid- and near-infrared spectroscopy is under-
going exponential growth due to its convenience, quickness, and
relatively low cost and because of the information-rich nature of
the data (Bellon-Maurel and McBratney, 2011).
Diuse-reectance Fourier-transform mid-infrared spec-
troscopy, also known as DRIFT, can be used quickly and nonde-
structively to quantify total soil C and other soil properties (Mc-
Carty et al., 2002; Leifeld, 2006; Viscarra Rossel et al., 2006).
Calibrations for total C integrate all the information in the Mi-
dIR spectra. e infrared spectrum can “see” important C func-
tional groups in SOM and can complement other techniques,
like NMR, for the quantication of alkyl-C, aromatic-C, and
carboxyl-C (Leifeld, 2006). Spectral interpretation can be used
to dig deeper into the spectral data. However, to identify and
quantify the presence of important organic functional groups
in environmental samples by observing uctuations in specic
spectral bands (Nguyen et al., 1991; Calderón et al., 2011a,
2011b; Demyan et al., 2012). Many of the functional groups to
which MidIR bands are ascribed, such as aromatics, carbonyls,
and aliphatics, cannot be quantied in soils by means of tradi-
tional chemistry. is precludes a side-by-side comparison of
spectral data and chemical data. Because of this, MidIR spectral
analysis should be validated by spiking soils with known organic
standards to correctly interpret changes in the expected spectral
features. Further, soils are mostly made up of mineral rather than
organic material, so tests about how small changes in organic
matter aect MidIR spectra are needed to make more accurate
spectral interpretations. We hypothesized that compounds that
contain functional groups that absorb in the MidIR will cause
predictable spectral dierences on addition to soils.
Previous work on spectral interpretation of soil MidIR data
has used ashing and spectral subtraction to bring out the absor-
bance bands attributable to the soil organics (Cox et al., 2000;
Calderón et al., 2011a, 2011b). Very few studies have tested the
usefulness and limitations of spectral subtraction on soils. e
hypothesis is that subtracting the soil spectrum from the mixture
spectrum results in the absorbance contribution of the added
pure compound. Ashing and subtraction must be used judicious-
ly because of subtraction artifacts brought about by heat-induced
changes in clays that aect the 1500 to 400 cm
−1
region (Reeves,
2012a), yet the regions between 3000 and 2800 cm
−1
and 1750
to 1600 cm
−1
can be subtracted reasonably well in low-carbon-
ate or decalcied soils (Reeves, 2012a).
e purpose of this project was to determine whether MidIR
is sensitive to relatively small additions of organic compounds.
We amended soils with known quantities of standards including
lipids, carbohydrates, amino acids, and proteins. We scanned the
soils alone, the standards alone, and mixtures of soil and standards
using MidIR. One specic question we wanted to answer was: can
we detect small additions of organic C and N? e long-term ob-
jective of this project is to determine if we can use MidIR to inter-
pret soil C dynamics aer changes in agricultural management or
environmental disturbance. We utilized two dierent soils with
varying C and clay contents and added known compounds to the
soils to increase the C content by 50%. We then used multivariate
analysis and spectral subtraction to test the sensitivity of MidIR
to detect changes in the C chemistry within a soil matrix.
MATERIALS AND METHODS
Soil Sampling
Soil samples (0–20-cm depth) from two sites with dierent
C and clay compositions were chosen for the study. e two soils
were: (i) an Aridic Paleustoll from Akron, CO (40°9¢ N, 103°8¢
W) under native grassland with 1.46% C, 0.18% N, 49.6% silt,
and 14.3% smectitic clay; and (ii) a Mollic Ochraqualf from
Hoytville, OH (41°0¢ N, 84°0¢ W) originally under a temperate
forest but now under no-till corn (Zea mays L.) with 2.48% C,
0.28% N, 45.01% silt, and 35.9% illitic clay. e pH of both soils
was slightly acidic (6.7 for Akron and 6.1 for Hoytville) and they
were carbonate free. ree replicates from randomly selected eld
locations were sampled from each site. Aer sampling, the soil
was sieved (2 mm) to exclude any rocks and coarse plant material.
Mixing of Soils and Standards
e standards listed in Table 1 were mixed with the soils to
increase the soil C content by 50%, to 2.19% C (21.9 g kg
−1
) for
the Akron soil and 3.72% (37.2 g kg
−1
) for the Hoytville soil.
We chose this quantity so that the added standard C was within
the range of C changes that could be attained by soil manage-
ment and to allow the presence of predominantly soil-derived
spectral bands in the mixtures. In addition to the 50% mixing
scheme used for all the standards, cellulose was mixed with the
soil to increase the soil C by 25, 50, and 100%. Before mixing,
both the soil and the standards were dried overnight at 60°C.
Temperature-sensitive standards (indole and palmitic acid) were
placed in a desiccator overnight instead of oven dried. Aer mix-
ing, the compounds were stored in sealed vials. Soils and stan-
dards were mixed as dry powders. A mortar and pestle was used
to ensure thorough mixing and grinding. Soils and standards
www.soils.org/publications/sssaj 1593
were analyzed for total C and N with a
Leco CN analyzer. e amount of each
standard added to the soils varied due to
the dierent C content of the standards,
which ranged from 33% in glycine to
>76% in the lipids and indole. e stan-
dards varied in N content from zero in
the lipids and tannic acid to >44% in the
purine nitrogenated bases (Table 1).
Biochar
Corn stalk material was obtained
aer harvest, oven dried (60°C), and
shredded. e biochars were generated
by a procedure developed in our labo-
ratory and detailed in Reeves (2012b).
Briey, the corn material was packed into
125-mL Erlenmeyer asks, and the emp-
ty space at the top was lled with glass
wool. e asks were then topped with
a 50-mL beaker. e biochars were pro-
duced at the prescribed temperatures for
45 min in a mue furnace, aer which time they were brought to
room temperature. e corn stalk material lost 46.3% of its mass
on charring at 300°C and 69.8% at 500°C. e 300°C char had
a pH of 6.0, and the 500°C char had a pH of 7.5. e corn stalks
were 43.3% C, the 300°C char was 57.9% C, and the 500°C char
was 69.8% C.
Infrared Spectroscopy
e soils and soil–standard mixtures were nely ground us-
ing a mortar and pestle before scanning. All soil samples were
scanned in diuse reectance from 4000 to 400 cm
−1
undiluted
with a Digilab FTS 7000 (Varian) with a deuterated triglycine
sulfate (DTGS) detector and a KBr beam splitter; KBr was used
as background. e background spectrum was subtracted from
each recorded spectrum. Each spectrum consisted of 64 co-add-
ed scans at 4 cm
−1
resolution. Each sample was scanned in dupli-
cate and the duplicate spectra were averaged. e spectra for the
reference standards and soils are shown in Supplemental Fig. 1.
Multivariate Analysis and Spectral Subtractions
Principal components analysis (PCA), spectral averages, and
subtractions were calculated using GRAMS/AI and GRAMS/
IQ soware, Version 9.1 (ermo Galactic).
RESULTS
Principal Components Analysis of the Spectra
e PCA shows that the main dierences in the soil and soil–
standard mix spectra are attributable to the soils (Akron vs. Hoyt-
ville), which are separated completely along Component 1 (Fig.
1). e correlation between the spectral data and each component
is indicated by the loadings. Component loadings in Fig. 2 indi-
cate that the Akron soils, with higher Component 1 scores, have
more absorbance near the 3630 cm
−1
clay band, the 3600 to 3400
cm
−1
OH–NH absorbance region, the 2000 to 1790 cm
−1
quartz
bands, and at 1626, 1532 (amide), and 1329 cm
−1
(Fig. 2; Table 2).
Absorbance at 1626 cm
−1
could be due to clay-bound water, but
the lower clay content of the Akron soils suggests that this band
may be at least partially due to amide or aromatics. e Hoytville
soils absorbed more than the Akron soils at 1562 (amide), 1423,
1164, and 1048 cm
−1
(polysaccharide C–O) (Table 2).
e distribution of the standard–soil mixtures along Com-
ponent 2 is very similar between the Akron and Hoytville soils.
e mixtures with urease, tannic acid, and palmitic acid caused
the least spectral dierences and group nearest to their respec-
tive unamended soils (Fig. 1). Adenine, methionine, casein, egg
protein, and ergosterol caused the most spectral dierences com-
pared with the soil-alone spectra. For both Akron and Hoytville,
the unamended soils as well as the tannic acid, egg protein, and
casein mixes have high Component 1 scores, while the adenine
and methionine mixes have low Component 1 scores. Loadings
indicate that high Component 1 scores are partly due to clay and
silicate absorbance (Fig. 2). Low Component 1 scores are due to
high absorbance at 1210 to 1140 and 1070 to 1000 cm
−1
, suggest-
ing organic absorbance and/or blockage and reduction of specular
reection from silicates in the soils by the added standards.
e guanine, adenine, methionine, and egg protein mixtures
have high Component 2 scores due to absorbance at 1670, 1588,
1513, and 1415 cm
−1
, which suggests amide or deformation C–H
absorbance (Table 2). e mixtures with casein, cellulose, and er-
gosterol have low Component 2 scores, indicating higher absor-
bance at 3630 to 3450 and 1250 to 1050 cm
−1
(Fig. 2; Table 2).
Note that the nitrogenous bases adenine and guanine are almost
50% N (Table 1). e PCA shows that soil mixes with N-contain-
ing standards of >10% N content tended to have positive Com-
ponent 2 scores compared with the samples that received zero-N
Table 1. Total C and N contents of the soils and the standards used to prepare the mixtures.
Soil or standard C N Formula Supplier
——— % ———
Akron native soil alone 1.46 0.17
Hoytville no-till soil alone 2.48 0.28
Adenine 45.66 46.45
C
5
H
5
N
5
EMD Biosciences
Casein 49.8 15.11 Calbiochem
Cellulose 45.03 0.00
(C
6
H
10
O
5
)n
Whatman
Ergosterol 85.00 0.00
C
28
H
44
O
MP Biomedicals
Glucosamine 34.08 6.66
C
6
H
13
NO
Sigma-Aldrich
Glycine 33.03 18.17
NH
2
CH
2
COOH
Fisher Chemicals
Guanine 40.69 44.83
C
5
H
5
N
5
O
Alfa Aesar
Indole 85.19 12.86
C
8
H
7
N
Alfa Aesar
Methionine 41.80 9.64
C
5
H
11
NO
2
S
Fisher Scientic
Palmitic acid 76.87 0.00
CH
3
(CH
2
)
14
COOH
TCI America
Egg protein 51.80 13.00
na†
Sigma-Aldrich
Sodium copper chlorophyllin 61.20 5.65
na
TCI America
Tannic acid 52.30 0.00
C
76
H
52
O
46
Mallinckrodt
Xylose 41.32 0.02
C
5
H
10
O
5
Fisher Scientic
Urease 44.87 13.05
na
Fisher Scientic
Vanillin 59.81 0.02
C
8
H
8
O
3
Fisher Scientic
† na, not available.
1594 Soil Science Society of America Journal
standards within each soil type. e casein mix was an exception,
with low Component 2 scores despite being a protein. Loadings
also indicate that the addition of casein, cellulose, and ergosterol is
consistent with increased OH, NH, and carbonyl bands (Table 2).
e mixtures with indole, ergosterol, chlorophyllin, vanil-
lin, and egg protein have high Component 3 scores, indicating
increased absorbance at 1585, 1515, 1436, 1402, and 1297 cm
−1
(Fig. 2). e guanine, casein, methionine, and adenine mixtures
have low Component 3 scores. Low Component 3 loadings are
associated with increased absorption at 3326, 3118, 1701, 1635,
1559, 1478, 1219, 1174, and 1122 cm
−1
(Fig. 2). ese bands
suggest the inuence of N–H, amide-like, aromatic C=C, and
carbonyl absorbance (Table 2).
Spectral Subtraction of Standard–Soil Mixes
We performed spectral subtractions of soil–standard mixes
minus the soil alone for all the standards. Here we discuss the
results from four compounds that are particularly interesting be-
cause of their dierent chemistries: cellulose (lacks N and should
show typical carbohydrate bands on subtraction), methionine
(should show N-containing spectral bands), casein (a protein
that should result in amide absorbance), and vanillin, which
should show aromatic-type bands.
e addition of cellulose resulted in an increase in absorbance
at 3400 cm
−1
, which should be due mostly to OH absorbance and
not NH in this case (Fig. 3). e region between 2970 and 2800
cm
−1
also subtracted well. e rise between 2200 and 2000 cm
−1
observed in the pure compound spectrum corresponds to over-
tones of the –COH stretch (Table 2). is band can also be seen
in the subtracted spectra, although this feature is not as prominent
Fig. 2. Component loadings for the principal components analysis
shown in Fig. 1.
Fig. 1. Principal components analysis of the mid-infrared spectra of the soils and soil–standard mixtures. The Akron soils are the circles and the
Hoytville soils are the squares. The percentage of the total variance explained by each component is shown in parentheses.
www.soils.org/publications/sssaj 1595
as the carbonyl band between at 1030 to 1160 cm
−1
, which we
show to be a particularly sensitive region to cellulose addition.
e methionine subtractions had higher values at 3050 to
2500, 2094, 1670 to 1580, 1530 to 1490, 1450 to 1400, 1360 to
1330, and 1230 to 970 cm
−1
(Fig. 4). is shows that amino acids
increase absorption due to various modes of N–H, C–O, and C–H
vibrational absorption (Table 2). Several peaks between 1180 and
870 cm
−1
present in the pure methionine spectrum did not have
a marked eect on the subtracted spectrum. e casein addition
subtraction data support the results of the PCA analysis (Fig. 1),
with absorption increases near 3400 to 3200, 2970 to 2850, 1700
to 1614, 1560 to 1490, 1460 to 1400, and 1260 to 1170 cm
−1
(Fig.
5), which show that the added protein aected absorption bands
associated with C=O and amide functional groups (Table 2).
Spectral subtraction of the vanillin-amended soils shows
that the 1670 cm
−1
as well as the aromatic C=C bands at 1592,
1515, and 1295 cm
−1
are all amenable to subtraction (Fig. 6).
e spectral subtraction using palmitic acid was included de-
spite this standard being one of the additives that caused the least
spectral dierences within the standard set (Fig. 1). is standard,
however, caused a very marked increase in absorbance at 2920 and
2850 cm
−1
, as expected from a molecule with a large quantity of
aliphatic CH bonds (Supplemental Fig. 1). In addition, palmitic
acid also resulted in less marked increased absorbance at 1691 and
1465 cm
−1
near the carboxylic acid C=O bond stretching and
Table 2. Putative assignments for the bands relevant to this study.
Note that mid-infrared absorption bands occur across a range
and that there are overtone and combination bands from several
different functional groups that may overlap with these frequencies.
Wavenumber Assignment
cm
−1
3630 stretching O–H in clays†
3400 stretching O–H and stretching N–H‡
2930–2870 stretching C–H§¶
2200–2000 overtones of stretching –COH# in aliphatics
2000–1790 quartz overtones†
1625–1630 mineral-bound water
1625–1670
amide I or phenyl ring stretching C=C‡, stretching C=O of
amide groups and nucleic acids, carboxyl¶
1570–1590 ring stretching C=C of phenyl‡, carboxylate stretching C=O
1480–1560
amide II band stretching C–N and bending C–N–H‡§, also
bending CH in phenyl rings
1530 stretching C=N or stretching C=C‡
1400–1450
bending (CH
2
) in polysaccharides and proteins‡, also N–H
and stretching C–N
1330
stretching C–N in amides, bending (CH) in phenyls and
polysaccharides‡
1220–1320 amide III band‡
1060–1170 stretching C–O in carbohydrates, nucleic acids, proteins‡
1050 bending C–O in carbohydrates‡
1020–950 stretching Si–O†
† Nguyen et al. (1991).
‡ Movasaghi et al. (2008).
§ Haberhauer and Gerzabek, (1999).
¶ Leifeld, (2006).
# Janik et al. (2007).
Fig. 3. (a) Spectral subtraction of the cellulose–soil mixes and (b) mid-
infrared spectra of cellulose.
Fig. 4. (a) Spectral subtractions of the methionine–soil mixes and (b)
mid-infrared spectra of methionine.
