Hot Topic Tomrj Contents

The Open Magnetic Resonance Journal

Volume 3, 2010

ISSN: 1874-7698

HOT TOPIC:

Applications and New Developments of Magnetic Resonance

Techniques in Soil Science

Guest editors:

Pellegrino Conte – Università degli Studi di Palermo Anne E. Berns – Forschungszentrum Jülich

Andreas Pohlmeier – Forschungszentrum Jülich Giuseppe Alonzo – Università degli Studi di Palermo

Contents

Editorial 14

Proton Nuclear Magnetic Resonance (NMR) Relaxometry in Soil 15

Science Applications Julia V. Bayer, Fabian Jaeger and Gabriele E. Schaumann

Proton NMR Relaxometry as a Useful Tool to Evaluate Swelling 27

Processes in Peat Soils Fabian Jaeger, Anastasia Shchegolikhina, Henk Van As and Gabriele Ellen Schaumann

Investigation of Iron(III)-Release in the Pore Water of Natural Sands 46

by NMR Relaxometry Ivonne Mitreiter, Sascha E. Oswald and Frank Stallmach

A Possible Difference in the Surface Relaxivity of Costal and Inland 52

Sands Joseph P. Hornak, Gianni Ferrante, Andrew Coy and Evan R. McCarney

Relaxation in a Natural Soil: Comparison of Relaxometric Imaging, 57

T1 – T2 Correlation and Fast-Field Cycling NMR S. Haber-Pohlmeier, S. Stapf, D. van Dusschoten and A. Pohlmeier

Low-Field NMR of Water in Model Soils 63 Oscar Sucre, Federico Casanova, Andreas Pohlmeier and Bernhard Bluemich

MRI in Soils: Determination of Water Content Changes Due to Root 69

Water Uptake by Means of a Multi-Slice-Multi-Echo Sequence (MSME) A. Pohlmeier, F. Vergeldt, E. Gerkema, H. Van As, D. Van Dusschoten and H. Vereecken

Effect of RF Field Inhomogeneity and Sample Restriction on Spectral 75

Resolution of CP/MAS-13C NMR Spectra of Natural Organic Matter Anne E. Berns and Pellegrino Conte

Interaction of a Recombinant Prion Protein with Organo-Mineral 84

Complexes as Assessed by FT-IR and CPMAS 13C NMR Analysis Fabio Russo, Liliana Gianfreda, Pellegrino Conte and Maria A. Rao

CPMAS 13C NMR Characterization of Leaves and Litters from the 89

Reafforestated Area of Mustigarufi in Sicily (Italy) Pellegrino Conte, Claudio De Pasquale, Etelvino H. Novotny, Gianluca Caponetto,

Vito Armando Laudicina, Maurizio Ciofalo, Michele Panno, Eristanna Palazzolo,

Luigi Badalucco and Giuseppe Alonzo

13C-NMR Chemical Shift Databases as a Quick Tool to Evaluate 96

Structural Models of Humic Substances Christian Nyrop Albers and Poul Erik Hansen

14 The Open Magnetic Resonance Journal, 2010, Volume 3 Editorial

Open Access

EDITORIAL

Applications and New Developments of Magnetic Resonance Techniques in Soil Science

In soil and environmental sciences, magnetic resonance techniques are nowadays used for a large variety of applications such as those related to the characterization of natural organic matter, to the analysis of the interactions of pollutants with or-ganic and inorganic moieties in environmental compartments, and to the evaluation of fluid flow in porous media. Recent ad-vances in electronics allowed not only the development of NMR on solids and semi-solid samples, but also the evolution of in situ NMR equipments. In addition to classical NMR applications, magnetic resonance imaging (MRI) of soil-water and tracer transport processes in soil (e.g. spatial and temporal changes of soil-water distributions or the movement of wetting fronts, root growth and water uptake) and NMR relaxometry and diffusometry for pore space exploration in natural porous media are now important fields in soil studies.

This special issue includes the conference proceedings of the session SSS23 titled “Applications and new developments of magnetic resonance techniques in soil science” which was held in April 2009 in Vienna (Austria) within the Soil System Sci-ences program of the European Geoscience Union (EGU) General Assembly (http://meetingorganizer.copernicus.org/ EGU2009/session/947). During this first session in the EGU Assemblies dealing with NMR, all the aspects of NMR spectros-copy were accounted for: from the limits of the classical NMR techniques such as CPMAS

13C NMR spectroscopy applied to

soils and humic substances, to the developments of innovative NMR techniques for soil science investigations such as low field and relaxometry NMR.

The paper selection encompassed in the present special issue of The Open Magnetic Resonance Journal reflects the diver-sity of magnetic resonance applications in the field of soil and environmental sciences. It does of course not claim completeness in these fields, but is meant to give an interesting and hopefully inspiring excerpt of these continuously expanding research fields. The first part of the issue presents new developments and applications of low field NMR and relaxometry, followed by a paper on magnetic resonance imaging applied to the analysis of plant-soil interactions. The last part of the special issue con-cludes with the classical high field solid and liquid state NMR spectroscopy. Namely, a paper on the evaluation of the limits of CPMAS

13C NMR spectroscopy due to imperfect field homogenization along the coil is presented together with a second paper,

where interactions among a prion protein and a soil-like system were investigated. Finally, the last paper deals with the sugges-tion of a quick method for validity evaluation of the humic substances models which were suggested in the literature.

Pellegrino Conte

Università degli Studi di Palermo

Dipartimento di Ingegneria e Tecnologie Agro-Forestali

v.le delle Scienze 13, edificio 4

90128, Palermo

Italy

E-mail: [email protected]

Anne E. Berns

Institute of Chemistry and Dynamics of the Geosphere

ICG-4: Agrosphere

Forschungszentrum Jülich GmbH

52425 Jülich

Germany


Andreas Pohlmeier

Institute of Chemistry and Dynamics of the Geosphere

ICG-4: Agrosphere

Forschungszentrum Jülich GmbH

52425 Jülich

Germany


Giuseppe Alonzo

Università degli Studi di Palermo

Dipartimento di Ingegneria e Tecnologie Agro-Forestali

v.le delle Scienze 13, edificio 4

90128, Palermo

Italy


© Conte et al.; Licensee Bentham Open.

This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided the

work is properly cited.

The Open Magnetic Resonance Journal, 2010, 3, 15-26 15

1874-7698/10 2010 Bentham Open

Open Access

Proton Nuclear Magnetic Resonance (NMR) Relaxometry in Soil Science Applications

Julia V. Bayer*, Fabian Jaeger and Gabriele E. Schaumann

Department of Environmental and Soil Chemistry, Institute of Environmental Sciences, University of Koblenz-Landau,

Fortstraße 7, 76829 Landau, Germany

Abstract: Proton NMR relaxometry is a very powerful tool for investigating porous media and their interaction with wa-

ter or other liquids and the mobility and interaction of organic molecules in solution. It is commonly used in material sci-

ence or earth science. However, it is only scarcely applied in soil science although it shows great potential for helping to

understand water uptake into the soil matrix and processes occurring at the solid-liquid interface at soil particle surfaces.

This review introduces proton NMR relaxometry in the context of soil science and discusses the most important applica-

tions of the method in this field. Relevant results from different applications of NMR relaxometry in soils are described

and research gaps identified. Some original data is presented concerning biofilm formation in soils, which was investi-

gated using proton NMR relaxometry. NMR relaxometry is a sensitive, informative and promising method to study pore

size distribution in soils as well as many kinds of soil physicochemical processes, among which are wetting, swelling or

changes in macromolecular status. It is further a very helpful method to study interactions between molecules in soil or-

ganic matter and can serve to study the state of binding of water or organic chemicals to soil organic matter. Relaxation

times determined by NMR relaxometry are sensitive to various factors that play a role in soil-water interaction which is

both an advantage and shortcoming of the method: NMR relaxometry can be applied to numerous investigations in soil

science, but at the same time interpretation of the results may be very difficult in such complex and heterogeneous sys-

tems like soils.

Keywords: NMR relaxometry, soil, porous media, water, swelling, wetting.

INTRODUCTION

Proton NMR relaxometry is commonly used in geo-sciences, e.g. in oil exploration, and material sciences [1, 2]. It is a powerful tool for non-destructive investigations of pore size distributions of porous media, water content, water uptake and re-distribution as well as molecule mobility and non-covalent binding mechanisms. The technique can be adapted for use in soil investigations, but so far has only been used sparsely. More frequently used for structural analysis is NMR spectroscopy which is able to give insight into the molecular structure of soil organic matter (SOM). It has been used extensively for the determination of humic acid (HA) and fulvic acid (FA) structures and other organic constituents in SOM in either liquid or solid state. Detailed reviews of the state of the art of NMR spectroscopy in natu-ral organic matter NOM, heterogeneous material and poly-mers are given elsewhere [3-5]. More specific details of NMR spectroscopy applications in soil, SOM and biological systems, including several other NMR techniques like mag-netic resonance imaging (MRI) and the investigation of mo-bility of deuterated or fluorinated compounds, can be found e.g. in [6-11].

This review focuses mainly on the application of proton NMR relaxometry in soil science, especially

1H NMR

*Address correspondence to this author at the Department of Environmental

and Soil Chemistry, Institute of Environmental Sciences, University of

Koblenz-Landau, Fortstraße 7, 76829 Landau, Germany; Tel: +49 (0)6341-

280-565; Fax: +49 (0)6341-280-576; E-mail: [email protected]

relaxometry, including some relevant studies on other porous media and magnetic resonance imaging (MRI) studies of soils. Unfortunately, NMR relaxometry as used for studies in geosciences cannot be transferred one to one for studying soils, as e.g. the pore system differs from that found in rock formations. The main challenge for application to soil is, in this context, its huge complexity and heterogeneity and the up to now only scarcely understood soil organic matter [12, 13]. Nevertheless, efforts have been made to use NMR re-laxometry to describe pore size distributions in soils, as well as processes occurring during water uptake, i.e. wetting, swelling of organic matter and re-distribution of water.

Apart from pure NMR relaxometry, MRI studies, based on the same measurement principle as NMR relaxometry, may help to get insight into soil water interactions as they provide spatial resolution additional to the temporal resolu-tion. Many studies employ MRI for understanding water uptake into soils or similar porous media and gain qualitative and quantitative information about local water distribution: Theoretical considerations about formation of preferential flow pathways have been confirmed by MRI; water and hy-drocarbon distribution and displacement have been evaluated and water distribution within the pore system can be ob-served, e.g. [14-22]. However, the resolution of MRI is much lower than that of NMR relaxometry and no detailed infor-mation about the pore size distribution or water properties within the pores can be determined from such measurements. Apart from

1H NMR relaxometry, other nuclei such as

13C or

19F can be used for relaxation studies of liquids in porous

media giving insight into distribution of those liquids [23,

16 The Open Magnetic Resonance Journal, 2010, Volume 3 Bayer et al.

24] and the pore size distribution of the porous media, e.g. [25, 26].

Gaining insight into water distribution in soils is espe-

cially important for nutrient and contaminant distribution,

which is of interest for agricultural applications and in rela-

tion to aquifer contamination [27]. Any substance entering

the soil pore system interacts with the solid surface and,

therefore, depends on solution distribution and interaction

with the matrix. Water uptake in soils is not a simple distri-

bution problem as e.g. preferential pathways form due to the

existence of macropores and different surface wettabilities of

the solid surface influence the wetting process [28]. Also,

water does not only enter pores, but interacts with the or-

ganic matter coatings and organic colloids present in the

pore system. This changes the solid surface and, hence, the

pore system itself. Model calculations are often inadequate

in describing water uptake into and the interaction of water

with the soil matrix [27, 28]. NMR relaxometry, therefore,

offers a great potential for investigating soil water interac-

tions without the need for modelling or sample destruction.

The method can be used in-situ, especially with more recent

developments in mobile NMR techniques that could be used directly in the field [29-37].

Addressing both soil scientists interested in the use of

these techniques for their own purpose and NMR specialists

providing new promising NMR relaxometry tools which

help to obtain further insights into soil processes, the objec-

tive of this contribution is to outline and discuss fields of

application of this technique in soil science. Although, dif-

ferent NMR methods are commonly used in soil science ap-

plications, this review focuses mainly on proton NMR re-laxometry, due to the complexity of the field.

BASICS OF NMR RELAXOMETRY

This section is addressed mainly to the reader unfamiliar

with the field of NMR. Many atomic nuclei posses a non-

zero spin and an intrinsic magnetic moment parallel or anti-

parallel to the spin. The spin is associated with a non-zero

magnetic moment (μ) via the relation μ = J, where is the

gyromagnetic ratio and J the spin angular momentum. is

constant, but assumes different values for different nuclei.

When placed in an outer magnetic field B (conventionally

along the z-axis), the spins orientate and precess about the

external field with the Larmor frequency, which is character-

istic for each nucleus and dependent on the strength of the

outer magnetic field (e.g. hydrogen nucleus 42.6 MHz at 1 Tesla): f= B/2 (e.g., [1]).

A radio frequency (RF) pulse with the characteristic

Larmor frequency is applied and the spins are flipped into an

angle to the external magnetic field (B0) causing a magneti-

sation (M0). In most applications the RF pulse turns the spins

in 90° or 180° direction (in a certain pulse sequence). After

the RF pulse is switched off the spins relax to their equilib-

rium orientation and the apparent magnetisation induced by

the RF pulse decays. The measured signal is called the free

induction decay (FID) [2]. The relaxation process generally

is a first order process. It is characterized by the relaxation

time, which is the reciprocal of the relaxation rate constant.

Two different relaxation mechanisms are involved in the

magnetisation decay, which are the longitudinal or spin-lattice and transverse or spin-spin relaxation [2].

The spin-lattice relaxation time T1 depends mainly on the interaction of the spins with their environment often referred to as the lattice, hence the name. T1 describes how effective interactions between the spin system and the environment are in exchanging magnetic energy. If strong interactions between the spin system and the environment lead to a fast exchange of energy, the equilibrium state is reached fast and T1 is short. Measuring T1 can be very time consuming and is, so far, not often used in soil science applications, although it may be the more appropriate measure than T2 in many cases [1, 2].

The spin-spin relaxation time T2 normally refers to the relaxation due to variable molecular interactions or diffusion in the slightly inhomogeneous magnetic field. The transver-sal relaxation process is not based on energy exchange, but originates from a dephasing of the precessing spins, e.g., due to slight differences in Larmor frequency due to local field inhomogeneity [2]. Variations in the magnetic field caused by neighbouring nuclei are stronger in solids than in liquids where spins can move freely and inhomogeneities due to neighbouring spins are small. As the dephasing of the spins can only take place in the presence of a longitudinal mag-netisation T2 can be smaller than or equal to T1, but it can never be longer [1].

While bulk liquids lacking additional means of interac-tion reveal long proton relaxation times in the range of sec-onds, limitation of mobility can reduce T2. Contrary to T2, T1 can be either increased or reduced by a reduction in mobility, depending on the Larmor frequency and the correlation time for the relaxation-relevant interaction [2]. Molecular diffu-sion in field gradients affects T2 but not T1, because no en-ergy exchange is involved in this relaxation mechanism [38]. The relaxation rate due to diffusion in field gradients is pro-portional to the diffusion coefficient and the square of local field gradients [2]. The local field gradients increase with increasing external field strength. Therefore, measurements in systems like soils, where local field gradients are the rule, are to be carried out preferentially in low fields up to 10-50 MHz. Field cycling NMR explicitly investigates the field and frequency dependency of T1 and T2 at proton Larmor frequencies between 10 kHz and 40 MHz or higher and is, therefore, a promising tool to study dynamic molecular in-teractions and to distinguish between the molecular effects and effects of local field gradients or sample heterogeneity [39]. In traditional high-resolution NMR spectroscopy, where Larmor frequencies are generally above 250 MHz, large T1/T2 ratios are the rule.

Relaxation Times T1 and T2 in Porous Systems

With T1 and T2 of protons of bulk water in the range of 1-3 seconds, bulk relaxation processes are very slow. If con-fined in porous media, relaxation is often controlled by solid-fluid-interactions at the surfaces of the pore space. Water molecules diffuse and eventually reach a pore wall surface where there is a finite probability that their spins are relaxed due to interactions with fixed spins, paramagnetic ions or paramagnetic crystal defects. Further transversal relaxation occurs via diffusion in local field gradients. The total relaxa-

Proton Nuclear Magnetic Resonance (NMR) Relaxometry The Open Magnetic Resonance Journal, 2010, Volume 3 17

tion rate is, therefore, the sum of bulk relaxation (B) and surface relaxation (S) and, for T2, of relaxation due to diffu-sion in field gradients [1]:

FGdiffSBtotal

SBtotal

TTTT

TTT

++=

+=

2222

111

1111

111

(1)

The surface relaxation term contains information of the pore system and is, therefore, further analysed. Relaxation time at the surface is determined by the residence time of the spin at the surface. The longer the residence time the higher the probability for interaction with the surface and, therefore, relaxation. As long as this surface relaxation is slower than the transport of unrelaxed spins to the surface the fast-diffusion or surface-limited regime [40] is fulfilled. Water molecules can transit the pore several times before being relaxed and the magnetization decay in an individual pore is, therefore, spatially uniform and depends on the surface-to-volume ratio. Surface relaxation is then related to the internal surface area S, internal pore volume V and the surface relax-ivity [1] which is strongly influenced by paramagnetic ions on the surface like Mn

2+ or Fe

3+:

surface-limited: rV

S

TS

2,12,1

2,1

1= (2)

where r is the pore radius and is the shape factor (1, 2, 3 for planar, cylindrical and spherical pore geometry) [41].

If, in contrast, the magnetic decay is controlled by the transport of the molecules to the surface the conditions of the slow-diffusion or diffusion-limited regime [40] are met. This may be the case if pores are large or surface relaxation is strong, e.g., due to the presence of effective paramagnetic centres.

diffusion-limited: 2

21

11

r

cD

TTSS

== (3)

where D is the diffusion coefficient and c is a shape-dependent factor. Note that in the case of diffusion limitation T1S and T2S are equal. Relaxation times in the diffusion-limited regime depend on temperature in the same way as the diffusion coefficient. In this case, relaxation times are not spatially uniform, which results in a multiexponential mag-netic decay, even within a single pore, and relaxation is addi-tionally dependent on pore shape [1]:

Converting NMR Signals into Relaxation Time Distribu-tions

For data analysis in NMR relaxometry several different algorithms and software is applied. These are, e.g., the soft-ware ‘WinDXP’ from Resonance Instruments (UK) or ‘Con-tin’ used by Bruker (USA) [42], which are device specific, or UPEN, a software developed at the University of Bologna [43, 44]. Depending on the chosen program and parameters, the analyses may lead to differently strongly separated peaks in the relaxation distribution. One such parameter is the

weight factor used in WinDXP to account for different signal to noise ratios. An example is given in Fig. (1) for a water repellent soil sample using a weight factor of 0.1 (Fig. 1a) and 20 (Fig. 1b). The differences in peak separation and peak number clearly demonstrate the importance to consider the effect of evaluation parameters on the resulting relaxa-tion time distribution [45]. Samples can only be compared to each other on the basis of comparable evaluation parameters using the same software.

WATER IN SOILS AND PEATS

Water uptake and redistribution in soils as well as inter-action of water with SOM is of great importance for, e.g., contaminant sorption or nutrient availability. The availability of water for plants itself is important in agricultural sciences and is determined by the water content, the matric potential ( ) and the water retention curve of a soil. Hereby, a matric potential of -1.5 MPa is defined as the permanent wilting point where plant growths is limited due to water shortage; a matric potential of -0.03 MPa is defined as the field capacity which is the amount of water held in a soil after excess water was drained due to gravitation [28]. Measurement of water retention curves by standard soil science methods is time consuming and cannot be carried out in situ.

A first effort to use low-field NMR relaxometry for analysis of water in soils was made by Prebble and Currie already in 1970 by measuring T1 (at 2.7 MHz) [46]. They used several sands, soils and a vermiculite as sample mate-rial and added different amounts of water. Three states of water in soils were identified: i) At very low water contents water was tightly bound to the clay or sand interface, but no relationship with plant unavailable water was found ( = -1.5 MPa); ii) with increasing water content the water seemed to be independent of the clay lattice and water content calcula-tions resulted in values close to the real amount of water added and iii) further addition of water lead to an incomplete relaxation during the measurement time indicating the pres-ence of bulk water ( = -0.03 MPa). The presence of various states of water was confirmed for peat samples by McBrierty et al. [47]. In a detailed study using high field NMR re-laxometry (300 MHz), differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) the binding of water in peat was investigated and up to four different water states were found, with two forms of loosely bound water, bulk water and tightly bound water that did not freeze at temperatures down to 160 K. The loosely bound water froze around 210 K and bulk water at 273 K, indicating also the temperature range above which each water form became mobile. Drying and re-watering of peat samples did shrink and swell the peat matrix and with that changed the amount of loosely bound water, but the amount of strongly surface associated water was similar after each change in moisture status. Non-freezing water was associated with hydration water, i.e. water in a gel like layer at the solid surface or wa-ter that chemically interacts with the hydrophilic moieties on the surface [47]. Therefore, the amount of non-freezing wa-ter could be an indicator for surface properties. The interac-tions of water molecules at the surface of particles is respon-sible for the thickness of the bound water layer or phy-sisorbed water, which can be up to 3 molecular layers. The relaxation rate increases (linearly) with the amount of solid


surface present, as shown for water clay suspensions, due to the relaxivity offered by the solid surface [48].

The relaxation mechanisms at the solid liquid interface are manifold and paramagnetic substances have an important influence. The coverage of only 0.01% with Fe(III) of a sil-ica surface was enough to increase surface relaxivity by an order of magnitude [49]. However, Mn(II) seems to be an even stronger relaxing agent than Fe(III) with a relaxation acceleration effect of up to three times stronger in solutions [50, 51]. The effect of paramagnetic ions on the surface re-laxivity seems to be restricted to one atomic layer at the sur-face of a particle as shown for Mn(II) on calcite particles [50]. Further increase in manganese concentrations in calcite water systems did not increase surface relaxivity further and also Mn(II) inside calcite particles did not contribute to sur-face relaxivity either [50]. The effect of paramagnetic sub-stances on the relaxation rate was observed to be much stronger when they are adsorbed to the solid surface, due to the restricted molecular motion of the adsorbed species which in turn results in a longer rotational correlation time for the coordinated water molecules. Nevertheless, bulk re-laxation is also accelerated in the presence of dissolved par-amagnetic ions [49-51]. The relaxation acceleration effect of paramagnetic substances in the bulk solution is dependent on the speciation of the ion [49, 51]. It was suggested that the relaxation acceleration decreases from hexa-aqua complexes to aqua complexes with a reduced number of exchangeable protons to organic-complexed ions to dispersed colloids. Therefore, the acceleration of the bulk relaxation rate in comparison to pure water may give additional information on the ion environment in complex soil solutions [51].

Due to the dependency of the relaxation times on the wa-ter binding and distribution NMR relaxometry can be used to describe the water environment: water in small pores or bound water relaxes faster than that in large pores or free water, due to increased accessibility of the solid surface. Gaining information about water uptake and redistribution in soil systems is of high importance, e.g., for agricultural sys-tems or prediction of contaminant distribution. Several stud-ies so far have been carried out investigating water uptake into soils or clays using relaxation time distributions deter-mined by

1H-NMR relaxometry [45, 47, 52-59]. Relaxation

time distributions generally showed three or four separate

peaks representing different water states or water in different pore systems. The boundary conditions vary between publi-cations, but as a general rule one can differentiate between micropores or tightly surface bound water at small relaxation times (e.g. T2: below 60 ms, sometimes separated into sev-eral peaks), mesopores or loosely bound water at medium relaxation times (e.g. T2: 60 – 300 ms) and macropores or bulk water at long relaxation times (e.g. T2 > 300) [47, 52, 60]. Some researchers found several relaxation time peaks at medium relaxation times. The loosely bound water relaxing with these relaxation times was possibly associated with dif-ferent separate environments that did not allow water ex-change at time scales of the relaxation measurements. Over the course of water uptake relaxation time distributions shifted towards smaller relaxation times and peaks at shorter relaxation times increased in size (Fig. 1a and b) [45, 53-55, 60].

The shift of relaxation times towards shorter times indi-cates water movement into smaller pores, which is contrary to the common model of water imbibition into porous media with hydrophilic pore walls where small pores are filled first due to capillary forces. As an explanation it was suggested that pore walls become increasingly hydrophilic with in-creasing soil-water contact time [45, 58] or that micropores that initially collapsed upon drying were reformed during water uptake by formation of water-swollen gels [53]. The latter process was referred to as swelling [53]. However, the definition of swelling is not used consistently in other publications.

Generally, the water uptake and re-distribution were found to be separated into fast and slow processes which can last up to weeks. The activation energies calculated for the fast and slow processes by Todoruk et al. [53] indicate that they are fundamentally different: The fast process had acti-vation energies of ~ 42kJ mol

-1 which is in the upper range

of diffusion associated processes. The slow component, however, had activation energies of > 80kJ mol

-1 indicating

chemical transformations like ester hydrolysis or more com-plex rearrangements of SOM components [53]. During wet-ting of a soil the hydrogen bonds of SOM components and mineral surface, which had been formed previously during drying, have to be broken apart in order to restore hydro-philic surface conditions; this process would be slow and

Fig. (1). Comparison of T2 relaxation time distribution of a water repellent sample directly after water addition and 19 days later using two

smoothing values: a) weight factor 0.1 and b) weight factor 20. Data taken from [45] and adjusted.


energetically unfavourable, leading to high activation ener-gies [53]. Other authors distinguished more clearly between wetting and swelling as two separate more or less independ-ent processes [54]. Wetting was suggested to be considerably faster (indicated by a shorter time constant of relaxation time changes) than swelling in hydrophilic soils and primarily associated with the properties of the solid surface (whether mineral or organic). In order to wet a surface it needs to be hydrophilic, however, prolonged contact of water with an organic hydrophobic soil particle surface could render it wet-table and, hence, would allow further processes like swelling to take place. This is displayed in NMR measurements as a slow change in relaxation times towards shorter relaxation times. Swelling here was defined as the hydration of SOM which increases the thickness of the SOM coating or SOM particle. This in turn would lead to a decrease in interparticu-lar pore size [45, 54]. The process of swelling may be of high relevance for contaminant fixation in soils as it may influence interactions of contaminants with SOM by e.g. increasing the available sorption site, forming new sorption domains or changing its rigidity [54, 61].

As described above it is necessary to have a hydrophilic surface in order to enable instant wetting. However, in sys-tems like soil, surfaces change when in contact with water and originally hydrophobic surfaces become wettable after prolonged contact with water. The breakdown of a hydro-phobic surface during wetting is thought to be fast in com-parison to swelling. It was suggested to exploit that fact and use low-field

1H-NMR relaxometry for soil wettability de-

terminations [56, 57, 62]. In order for a liquid in porous me-dia to be relaxed efficiently it needs to be in contact with the solid surface. Theoretical considerations suggest that relaxa-tion times of hydrophobic samples are longer than that of wettable samples enabling a better proton exchange [56, 57]. T2 of water repellent soil samples and model systems was found to be larger than 1000 ms, but that of wettable samples ~100 ms [56, 57]. As described above water repellency of organic coatings on particle surfaces normally breaks down after contact with water, therefore, relaxation times of water repellent and wettable sample should eventually reach the same equilibrium. The decrease of relaxation time and ap-proach of a similar equilibrium was confirmed in two studies and the time for reaching the equilibrium was dependent on the sample [56, 57].

Proton NMR relaxometry studies of water in soil systems allow to distinguish processes taking place during water up-take. It is also possible to differentiate between water in sev-eral environments, i.e., bound, loosely bound and free bulk water. Furthermore, influences of factors like paramagnetic substances in solution and on the solid surface have been characterised and partly quantified. However, it is still nec-essary to quantitatively describe the processes occurring dur-ing water uptake into soils, such as wetting and swelling and evaluate their environmental impact like their involvement in nutrient or contaminant distribution.

PORE SIZE DISTRIBUTION IN SOILS

It is well established that porosity and pores size distribu-tions can be derived from relaxation time distribution of geo-logical formations, like rocks, sandstones or permafrost and gas hydrate sediments, or materials such as ceramics (e.g. [2,

63-65]). However, even in rocks comparison of pore size distributions from different samples has to be considered carefully. The iron concentrations in rock formations are probably high enough to ensure constant surface relaxivity (compare section “water and porous media”), nevertheless, shifts in relaxation time distributions may not only be due to differences in pore size distributions, but differences in the amount of paramagnetic substances present in the sample [41, 49]. The presence of paramagnetic substances on a coated silica gel reduced the relaxation time of water close to the surface so much that the monomodal relaxation time dis-tributions were changed to bimodal distributions, thereby identifying microporosity of the surface [49].With increasing SOM content the number of identified water compartments increased from three to four suggesting a correlation between pore system and organic matter [52]. An even more detailed relationship between soil components and pore sizes was identified in another study: The relaxation time of soil sam-ples was found to be dependent on sand, silt, clay and SOM content, but the degree of correlation was dependent on the pore system, i.e. micro-or mesopores. The transverse relaxa-tion times of micropores correlate with clay and SOM con-tents, but those of mesopores with silt, sand and SOM [66].

In another study the influence of kaolinite addition to sandy samples was investigated [67] and found an increase in relaxation rate with increasing amount of kaolinite present in the sample. This was ascribed to the increasing surface area (increase in smaller pores) and the higher surface relax-ivity of kaolinite (one reason for this higher surface relaxiv-ity may be the presence of iron in the octahedral layers of kaolinite). However, at a certain amount of kaolinite the re-laxation rate increased less. This was assumed to be the point where all sand surfaces were covered in kaolinite and the surface relaxivity was stable, leaving only decreasing pore size and changing pore geometry responsible for changes in T1.

A slightly different approach to determine pore mobile and immobile fractions in a wetland soil was used by Culli-gan et al. [68]: The sample (a sphagnum peat moss) was saturated with water and T1 was determined (at 122 MHz), then a 1 mM Gd

3+ solution was added and T1 was deter-

mined again. As the Gd3+

solution was added under condi-tions where diffusion is negligible, this second measurement sampled only the mobile pore space. It was found that 43% of the pore space showed a fast relaxation time (T1 = 35 ms), and 56% exhibited a longer relaxation time of T1 = 165 ms. The first was assumed to represent the pore space filled by Gd

3+ solution, whereas the latter only by water, therefore,

confirming the existence of two porosities in the wetland peat.

One main assumption when converting relaxation time

distribution into pore size distributions is that pores are not

interconnected or more specific relaxation starts and ends

within one pore. This may apply to geological formations

which have larger pores than soils, but does not hold true for

soils. Also, the pore drainage in soils can be considered to

not necessarily be total, i.e. some pores drain while others

retain their water [60]. Further assumptions are that the sur-

face relaxation is constant throughout the pore system and

the shape factor of the pores is constant and known [55]. In


most studies assessing pore size distributions the fast diffu-

sion regime is assumed, so that relaxation time is influenced

only by the surface relaxivity of the solid surface and the

relaxation time of the bulk phase [66]. The surface relaxivity

can be determined from volume to surface area ratios which

in turn can be determined from e.g. nitrogen adsorption or

mercury porosity measurements [55, 63, 66].

The application of NMR relaxometry to determine soil

pore size distributions so far has been mainly qualitative.

Several studies agree that relaxation time distributions of soil

samples are related to pore sizes, but do not directly and

quantitatively describe pore size distributions [53, 69]. The

study conducted by Hinedi et al. (1993) was probably the

first one to derive a real pore size distribution from a relaxa-

tion time distribution, but did not verify the outcomes by

comparing them to results from conventionally obtained pore

size distributions [55]. A qualitative comparison of NMR

derived and conventional determined pore size distributions

was undertaken in two later studies, but NMR relaxometry

was recommended only as an additional method to conven-

tional pore size determination to characterize pore connec-

tivity [60]. However, a quantitative comparison between

pore size distributions derived from NMR and conventional

methods so far has been mainly conducted for several rock

types [63]. Pore sizes, determined by NMR relaxation meas-

urements in comparison to mercury porosimetry, were over-

estimated by an order of magnitude. Mercury porosimetry is

based on the Washburn equation (x

rv

l

2

cos= , where v is

the rate of liquid entry into the capillary, r is the capillary

radius, l is the liquid surface tension, is the viscosity of the

liquid, x is the distance penetrated, and is the contact angle

[70]). It, therefore, tends to reflect more pore throats than

pore sizes, leading to an underestimation of the real pore size

[63]. Just recently the application of NMR relaxometry (T2

measurements at 2 MHz) for determination of pore size dis-

tributions by quantitatively comparing it to conventional

pore size distributions derived from water retention curves

was verified for several soil types [66]. In this new approach,

the relaxation time – pore size relation revealed two separate

regions. The condition for the fast-diffusion regime [40] was

fulfilled for T2 < 10 ms. For larger T2 values, a transition

from the fast-diffusion to the intermediate-diffusion regime

[40] for finer textured soil samples, and transition from the

intermediate-diffusion to the slow-diffusion regime [40] for

sandy soil samples was determined. Additionally, the true

bulk relaxation time was used instead of the hypothetical one

of free water commonly assumed for such investigations

[66].Consequently, proton relaxation in larger pores was

governed by surface relaxivity and self diffusion of water.

However, for simplification, the condition for the fast-

diffusion regime was assumed as fulfilled for all pore sizes in

this study. A good consistency (R2 = 0.98) between pore size

distributions determined by conventional soil water retention

measurements and 1H NMR relaxometry was found using

the two different surface relaxivities for micro-and

mesopores (for details on calculations see [66]).

As described above, the determination of conventional soul water retention curves is still necessary in order to be able to calculate surface relaxivities. In order to use the whole time-saving potential of the NMR measurements an independent method for the determination of surface relaxiv-ities is necessary. Additionally, changes in pore sizes during water uptake as often reported have to be investigated further as they may not only be attributed to swelling of organic matter on particle surfaces or water re-distribution into pores previously not available, but also to the formation of new pore systems due to microbial activity.

COMPLEXATION OF PARAMAGNETIC IONS IN SOIL SOLUTIONS

Both relaxation times are greatly reduced in the presence of paramagnetic ions. The strength of the effect depends on the ion environment and specification. The interaction of paramagnetic ions with FA or HA in solution, thus, can be investigated using

1H NMR relaxometry. Variations between

Mn(II), Cu(II) and Fe(III) relaxation times suggested that different complexation mechanisms were at work in several studies [51, 71-73]: No change or only minimal change was found for solutions containing sulphosalicylic acid and Mn(II) in contrast to solutions with only Mn(II), suggesting the formation of outer sphere complexes, as the rotational motion of the ions was not affected [71]. However, Cu(II) and Fe(III) solutions were strongly affected by the presence of sulphosalicyclic acid (reduction of relaxation time with increasing concentration of sulphosalicyclic acid) suggesting the formation of inner sphere complexes [71]. Contrary find-ings were reported by Melton et al. for solutions of Lauren-tian HA [72]: Relaxation times of solutions with Cu(II) de-creased only slightly with increasing concentration of HA [72]. The formation of stable or labile metal complexes, therefore, seems to be very dependent on the organic mate-rial. Interactions of organic compounds and FA or HA in solution were also investigated by changing concentrations and environmental parameters in the solution and their ef-fects on relaxation times were observed. A difference in the interaction of HA and monoaromatic compounds was found depending on the aromaticity and also very strongly on pH [73]. Relaxation acceleration due to interaction with dis-solved and colloidal Fe and Mn species in soil solutions causes a wide range of relaxation times in dependence of the Fe and Mn speciation [51].

FA and HA were shown to form - complexes with hy-drophobic organic compounds like dichlorophenol. FA was less effective in forming such complexes than HA which was attributed to the stronger hydrophobic character of HA [74]. Two NMR relaxometry studies using

13C-labeled acenaph-

tone and fluoro-acenaphtone both found evidence that the mode of interaction of FA and acenaphtone depends strongly on the concentration of FA in solution and the solution pH [75, 76].

Investigations of such interactions may help to under-stand the fate of organic compounds in the aquatic environ-ment and are partly transferable to soil systems; however, the soil matrix is so much more complex and exhibits so much more opportunity for interactions apart from the soil solu-tion, that a direct transfer is not possible.


MOBILITY AND NON-COVALENT BINDING MECHANISMS OF ORGANIC MOLECULES IN THE

SOLID ORGANIC MATTER

NMR relaxometry can be used to probe the spin envi-ronment and, therefore, gain information about binding and association forms of the molecule under investigation. These investigations are indirectly related to soils as they can help predicting behaviour of organic compounds in the environ-ment. Main constituents of SOM are fulvic acids (FA) and humic acids (HA) and several studies investigated the inter-action and non-covalent binding forms of organic molecules or metal ions with HA and FA (e.g. [71, 73-79]). The identi-fication of rigid and flexible structures within organic mate-rials is also possible [79, 80]. The reported temperature de-pendence of rigid and flexible domains within HA correlated well with glass transition temperature determined by several other authors using differential scanning calorimetry (e.g. [81, 82]). Another recent study [61] reported a correlation of decrease in matrix rigidity of a peat sample with an increase of proton relaxation time (T2). After heating a sample in an airtight container its matrix rigidity was reduced and relaxa-tion time increased, indicating a higher mobility of the or-ganic matter involved. After two weeks proton relaxation time had decreased to the original value and matrix rigidity increased. Suggested by earlier studies obtained from de-tailed DSC and TGA analysis [83-85], it was assumed that this may be due to the formation of cross-links between or-ganic molecules via water molecules (physicochemical ma-trix aging). The thermal and moisture history is expected to be linked closely to the mobility of organics and the matrix rigidity [61]. As rigid and flexible domains probably show different sorption towards contaminants the identification and quantification of such domains within SOM is of interest for modelling contaminant sorption behaviour.

MICROBIAL INFLUENCES

Microorganisms can form extended networks, so called biofilms, in order to relieve water stress and use nutrients more efficiently. These biofilms are formed of extended ex-tracellular polymeric substances (EPS) networks which bind water very effectively and form highly hydrated gels [86]. Biofilms or small biofilm-like structural units can also be formed in soils.

The change of the spin environment within such biofilms compared to bulk water was tested by NMR relaxation or MRI [9, 87, 88]. In aqueous solutions the monomodal relaxa-tion time distribution (T2 at 85 MHz) of water became bi-modal in the presence of a biofilms. However, in a porous model system of glass beads the resolution of the peaks was not possible due to the relaxation effects of the solid surface of the pore system. MR images of the same samples con-firmed biofilm distribution true to the optical examination [87] proving the applicability of the methods for such sys-tems.

In soil samples the detection of biofilm growth is not that easy and bacteria do not form free biofilm inside pores, but use EPS to attach themselves to the particle surface and en-hance transport of nutrients [86]. Microorganisms in soils are mainly attached to particle surfaces and primarily found in pores with diameters of 1-30 m [89, 90].

Enhancing microorganism activity in soil samples re-sulted in a stronger shift of the relaxation time (T2) towards shorter T2 in treated (enhanced microbial activity) than un-treated samples over the course of water uptake. This could be due to the increased production of EPS in the treated samples which may have reduced sizes of existing pores or formed a new micropore system. However, the contributions of other processes in reducing relaxation times like swelling of SOM or the change of surface relaxivity due to bacterial growths could not be excluded up to now [88] and further research in this area is needed.

Effects of Biofilm on Proton Relaxation Time Distribu-tions in Model Soil Systems

In a qualitative study, the effects of bacterial biofilm on transverse relaxation time distribution of water in biofilm reactors, used as model soil systems, at 2 MHz (Maran 2, Resonance Instruments, UK) were investigated (not pub-lished). Special designed glass bottles (height x diameter: 12 cm x 5 cm; volume: 160 cm ) with two bottle closures (at the top and bottom) were filled with glass beads of different par-ticle sizes or with natural soil (sandy soil, sieve fraction 63

m to 2 mm). Some of the reactors were inoculated with a biofilm producing isolate (99% sequence identity with Si-norhizobium sp. TB8-10-II, isolated from a waste water sand filter) and relaxation time distributions were measured after incubation time of 5 to 8 days. Optical inspection of the glass bead reactors showed biofilm growth after this time; for the soil reactor a similar growth was assumed. Fig. (2) shows the setup of the reactor system (left hand side) and a sketch of a filled reactor (right hand side). Reactors were filled with up to five layers of glass beads with particle sizes ranging from 5 mm to 150 m (decreasing particle sizes from glass closure to bottle middle) to prevent particle outflow. Layer D in Fig. (2) represents the domain studied in the NMR relaxometer (i.e. filled with the different growing materials). 2.5 L of a 30 g L

-1 Trypticase

TM Soy Broth solution (BD Diagnostic

Systems, Heidelberg, Germany) was used as a culture me-dium and was pumped with 8 mL h

-1 into to a dropping fun-

nel to prevent contamination (Fig. 2). A second pump (circu-late pump with 900 mL h

-1) was responsible for the flow of

culture medium through the reactor. After finishing the ex-periment, the reactors were dried using a pump. However, this was only possible for the 3 mm glass beads as the pres-sure was not high enough to dry the other size fractions.

1H NMR measurements were performed using a CPMG

pulse sequence [91]. The number of 180° pulses ranged be-tween 8192 (soil) and 14336 (3 mm glass beads) with con-stant number of scans of 256. Echo spacing ranged between 150 s (soil) and 300 s (glass beads). The objective was to achieve a signal to noise ratio between 50 and 100. The repe-tition time was set individually for every reactor and chosen based on three to six times the longest T2 and was 3-10 s. Relaxation time distributions were calculated from the decay curves with the WinDXP software (Resonance Instruments, UK) running a zeroth order regularisation to perform a con-tinuous distribution of exponentials applying the BRD (But-ler, Reeds and Dawson) algorithm [92]. The relaxation time distributions consisted of 128 time constants with associated amplitudes. The time constant range was 1-10000 ms, and the weight factor for the regularization was 0.5 for all biofilm reactors.


Fig. (2). Setup of the reactor system (left hand side) and a sketch of a biofilm reactor filled with glass beads (right hand side).

Transverse relaxation time distribution of water in reac-tors filled with 3 mm glass beads (Fig. 3) or 500-350 m glass beads (Fig. 4) consisted of two to three peaks. Peak 0 (3 mm glass beads: T2 = 40-90 ms; 500-350 m glass beads: T2 = 30-40 ms) may be a fitting artefact, because its exis-tence and position was not reproducible in the replicate sam-ples. Furthermore, its intensity was in the range background noise. Position of Peak 1 was T2 = 300 ms for 3 mm glass beads without biofilm and up to 500 ms for 3 mm glass beads with biofilm (Fig. 3). For 500-350 m glass beads, Peak 1 was determined around T2 = 150 ms for reactors with and without biofilm (Fig. 4). This suggests that Peak 1 repre-sents water between the contact areas of the glass particles, because its position decreased and its intensity increased with decreasing particle size in the reactors without biofilm. In the time scale of the NMR experiment, this water is not exchanging with water in larger pores as represented by Peak 2, and, thus, may be represented by an individual peak. In

the inoculated glass bead reactors, it may include water in-side the biofilm matrix, because its intensity tended to in-crease with increasing biofilm dry mass. This suggestion is supported by the finding for agar gels, which were used as model biofilm, were T2 ranged between 1100 ms and 100 ms for agar concentrations of 1.0-10 g L

-1 (not shown). Peak 2

may represent water in large antiparticle pores, because its position decreased with decreasing particle size (3 mm glass beads: T2 = 2300 ms without and ~ 2000 ms with biofilm; 500-350 m glass beads: T2 = 1100 ms without and ~ 800 ms with biofilm).

Fig. (4). Transverse relaxation time distribution of water in reactors

filled with 500-350 m glass beads with and without fresh biofilm.

For both types of glass beads, biofilm growth resulted in a shifting of peak 2 towards smaller T2 values (Fig. 3, 4). This suggests decreasing interparticle pore diameters and/or changes of the surface relaxivity caused by biofilm on the glass bead surfaces. Additionally, the intensity of peak 1 tended to increase in the reactors with fresh biofilm. This observation was not determined for the rewetted biofilm (Fig. 3), suggesting structural changes due to drying.


filled with 3 mm glass beads with and without fresh biofilm, and

with rewetted biofilm after air drying of the reactor with fresh

biofilm.

10 100 10000

3

6

9

Peak 0

Peak 23 mm glass beads without biofilm fresh biofilm rewetted biofilm

Pro

por

tion

of

H2O

tota

l [%

]

T2 [ms]

Peak 1

10 100 10000

3

6

9

500-350 µm glass beads without biofilm fresh biofilm

Pro

por

tion

of

H2O

tota

l [%

]

T2 [ms]

Peak 0 Peak 1

Peak 2


The T2 distribution of water in reactors filled with natural soil (sieve fraction 63-2000 m) consisted of three peaks, representing water in different pore types (Fig. 5). Bacterial inoculation resulted in considerable changes of the T2 distri-bution. The intensity of peak 1 increased and peak 2 and 3 showed a trend to decreasing intensity. This suggests that pore diameters of larger inter-particular pores decreased and that the amount of smaller pores increased due to the forma-tion of biofilm inside the soil matrix. One inoculated soil filled reactor was found to be clogged and also changes in the T2 distribution were very strongly developed (biofilm 2 in Fig. 5), suggesting a very strong biofilm growth. A clog-ging due to small soil particle sizes can most likely be ex-cluded, because particles smaller 63 m were removed by sieving prior to the experiment. Furthermore, the control reactor without inoculation showed no evidence of clogging. Biofouling and pore clogging is also known in technical ap-plications like tubes, membrane filters and sand filters [93]. The results of the soil filled reactors are qualitatively compa-rable to those of Jaeger et al. and support their assumption that biofilm formation affected the T2 distribution of water in soil samples with higher microbial respiratory activity [88].


filled with natural soil (sieve fraction 63-2000 m) with and with-

out fresh biofilm.

A 50 times higher transverse than longitudinal relaxivity was determined for agar gels at 30 MHz. This finding can be interpreted in terms of a reduced rotational mobility of the water molecules due to water structuring of the polymer [94]. Thus, a combination of T1 and T2 measurements can be suggested for a more detailed study of biofilm or other gel phases, e.g. inside the SOM matrix [54]. This may be helpful to determine different water states and to discriminate be-tween the effects of water mobility and pore size distribution in biofilm or gel containing porous media.

OUTLOOK

The potential of proton NMR relaxometry for soil sci-ence is still far from being fully exploited. In order to utilize the full potential of the NMR technique, it is necessary to adapt it to the specific complexity and heterogeneity of soils to gain a more detailed understanding of interaction dynam-ics and soil-specific processes. Related NMR methods for

pore size and water distribution in soils evaluation are stray field STRAFI NMR that uses a strong magnetic field gradi-ent in a high field superconducting magnet, as well as pulsed field gradient PFG NMR measurements for diffusion coeffi-cients determination. The latter is especially promising in combination with NMR relaxometry [60, 64, 95, 96] to relate relaxation time with molecular diffusivity. During the last ten years mobile NMR devices have been developed in order to be able to investigate porosity and water distribution in samples in-situ with devices that promise easy handling. Several different approaches should be mentioned here, in-cluding the NMR MOUSE (mobile universal surface ex-plorer) a unilateral scanner [29, 30, 35-37] and in-side-out NMR devices like the Hallbach Scanner for bore hole appli-cations [2, 31-34]. Field cycling NMR techniques span a wide range of magnetic fields and, therefore, proton Larmor frequencies (10 kHz-40 MHz) within one single instrument. These techniques are a powerful and promising tool to study interaction dynamics [39]. Although the fields of application do not yet span investigation of soil samples, especially field cycling NMR will help to gain valuable complementary de-tailed understanding on interactions between soil and water.

The development of new T1 pulse sequences reducing the overall measurement time may lead to a more frequent use of T1 measurements. This may further improve the understand-ing of soil-water interactions as T1 probes more directly in-teractions of the spin system (i.e. water) and the environment (i.e. the pore surface).

The potential of NMR relaxometry lies in the strong sen-sitivity of relaxation times to numerous factors relevant in soil-water-organics interactions, which is, however, at the same time disadvantageous and hides the danger of severe misinterpretations especially in systems as complex, versa-tile and heterogeneous as soils. It, thus, has to be kept in mind that conclusions on soil processes have to be drawn with care and on the basis on detailed targeted process analy-sis.

ABBREVIATIONS

DSC = differential scanning calorimetry

EPS = extra cellular polymeric substances

FA = fulvic acid

HA = humic acid

MOUSE = mobile universal surface explorer

MRI = magnetic resonance imaging

NMR = nuclear magnetic resonance

NOM = natural organic matter

PFG NMR = pulsed field gradient NMR

RF = radio frequency

SOM = soil organic matter

STRAFI NMR = stray field NMR

ACKNOWLEDGEMENTS

We like to thank Hella Korn, Dr. Uta Böckelmann and Daniel Wicke from the TU Berlin for their help with the biofilm reactors. This study was part of the projects

1 10 100 10000

4

8

12

Peak 3Peak 2Peak 1

Am

plit

ud

e [a

.u.]

T2 [ms]

without biofilm biofilm 1 biofilm 2 (clogged)


SCHA849/5 and SCHA849/8 funded by the German Re-search foundation (DFG).

REFERENCES

[1] Kleinberg RL. Nuclear magnetic resonance. In: Wong P-Z, Ed.

Methods in the physics of porous media. New York: Academic

Press 1999: pp. 337-85.

[2] Dunn K-J, Bergman DJ, Latorraca GA. Handbook of geographic

exploration-seismic exploration: nuclear magnetic resonance-

petrophysical and logging applications. Elsevier Science Ltd.: Ox-

ford 2002.

[3] Bluemich B. Solid-state NMR of heterogeneous materials. Adv

Mater 1991; 3: 237-45.

[4] Bluemler P, Bluemich B. NMR of polymers in the solid-state.

Spectrosc Eur 1995; 7: 8-16.

[5] Conte P, Spaccini R, Piccolo A. State of the art of CPMAS 13C-

NMR spectroscopy applied to natural organic matter. Prog Nucl

Magn Reson Spectrosc 2004; 44: 215-23.

[6] Preston CM. Applications of NMR to soil organic matter analysis:

history and prospects. Soil Sci 1996; 161: 144-66.

[7] Randall EW, Mahieu N, Ivanova GI. NMR studies of soil, soil

organic matter and nutrients: Spectroscopy and imaging. Geoderma

1997; 80: 307-25.

[8] Lens PNL, Hemminga MA. Nuclear magnetic resonance in envi-

ronmental engineering: principles and applications. Biodegradation

1998; 9: 393-409.

[9] Lens PNL, Van AH. Use of 1H NMR to study transport processes

in biofilms. In: P.N.L. Lens, T. Mahony TM, P. Stoodley, O'Fla-

herty V, Eds. Biofilms in medicine, industry and environmental

biotechnology: characteristics, analysis and control IWA Publish-

ing, London, UK 2003; pp. 285-307.

[10] Xiong J, Lock H, Chuang I-S, Keeler C, Maciel GE. Local motions

of organic pollutants in soil components, as studied by 2H NMR.

Environ Sci Technol 1999; 33: 2224-33.

[11] Fomba KW. Investigation of the mobility of organic contaminants

in humic substances. Leipzig: Universität Leipzig 2008.

[12] Schaumann GE. Soil organic matter beyond molecular structure. 1.

Macromolecular and supramolecular characteristics. J Plant Nutr

Soil Sci 2006; 169: 145-56.

[13] Schaumann GE. Soil organic matter beyond molecular structure. 2.

Amorphous nature and physical aging. J Plant Nutr Soil Sci 2006;

169: 157-67.

[14] Amin MHG, Richards KS, Chorley RJ, Gibbs SJ, Carpenter TA,

Hall LD. Studies of soil-water transport by MRI. Magn Reson

Imaging 1996; 14: 879-82.

[15] Amin M, Hall LD, Chorley RJ, Carpenter TA, Richards KS, Bache

BW. Magnetic resonance imaging of soil-water phenomenon.

Magn Reson Imaging 1994; 12: 319-21.

[16] Amin MHG, Chorley RJ, Richards KS, et al. STUDY of infiltration

into a heterogeneous soil using magnetic resonance imaging.

Hydrol Process 1997; 11: 471-83.

[17] Bortolotti V, Camaiti M, Casieri C, De Luca F, Fantazzini P,

Terenzi C. Water absorption kinetics in different wettability condi-

tions studied at pore and sample scale in porous media by NMR

with portable single-sided and laboratory imaging devices. J Magn

Reson 2006; 181: 287-95.

[18] Chen J-D, Dias MM, Patz S, Schwartz LM. Magnetic resonance

imaging of immiscible-fluid displacement in porous media. Phys

Rev Lett 1988; 61: 1489.

[19] Davies S, Hardwick A, Roberts D, Spowage K, Packer KJ. Quanti-

fication of oil and water in preserved rock by NMR spectroscopy

and imaging. Magn Reson Imaging 1994; 12: 349-53.

[20] Baumann T, Petsch R, Niessner R. Direct 3-D measurement of the

flow-velocity in porous media using magnetic resonance tomogra-

phy. Environ Sci Technol 2000; 34: 4242-8.

[21] Hall LD, Gao Amin MH, Dougherty E, et al. MR properties of

water in saturated soils and resulting loss of MRI signal in water

content detection at 2 tesla. Geoderma 1997; 80: 431-48.

[22] Belliveau SM, Henselwood TL, Langford CH. Soil wetting proc-

esses studied by magnetic resonance imaging: correlated study of

contaminant uptake. Environ Sci Technol 2000; 34: 2439 -45.

[23] Nestle N, Wunderlich A, Baumann T. MRI studies of flow and

dislocation of model NAPL in saturated and unsaturated sediments.

Eur J Soil Sci 2008; 59: 559–71.

[24] Cheng Y, MacMillan B, MacGregor RP, Balcom BJ. Direct detec-

tion of hydrocarbon displacement in a model porous soil with mag-

netic resonance imaging. Anal Chem 2005; 77: 1824-30.

[25] Mair RW, Hürlimann MD, Sen PN, Schwartz LM, Patz S,

Walsworth RL. Tortuosity measurement and the effects of finite

pulse widths on xenon gas diffusion NMR studies of porous media.


[26] Lounila J, Telkki V-V, Jokisaari J. Extracting information on po-

rous materials by xenon porometry. Magn Reson Imaging 2007;

25: 569-70.

[27] Rubin H, Narkis N, Carberry JB. Overview of NAPL contamina-

tion and reclamation. In: Rubin H, Narkis N, Carberry JB, Eds. Soil

and aquifer pollution. Heidelberg: Springer-Verlag 1998: pp. 3-17.

[28] Foth HD. Fundamentals of soil science. 8th ed. New York: John

Wiley & Sons, Inc. 1990.

[29] Eidmann G, Savelsberg R, Bluemler P, Bluemich B. The NMR

MOUSE, a mobile universal surface explorer. J Magn Reson, Ser A

1996; 122: 104-9.

[30] Bluemich B, Anferova S, Pechnig R, Pape H, Arnold J, Clauser C.

Mobile NMR for porosity analysis fo drill core sections. J Geophys

Eng 2004; 1: 177-80.

[31] Anferova S, Anferov V, Rata DG, et al. A mobile NMR device for

measurement of porosity and pore size distributions of drilled core

samples. Concepts Magn Reson Part B 2004; 23B: 26-32.

[32] Anferov V, Anferova S, Voda MA, Kupferschlaeger K, Blümich B.

Mobile NMR scanners for nondestructive measurements of poros-

ity of drill cores. Magn Reson Imaging 2007; 25: 547.

[33] Anferova S, Anferov V, Arnold J, et al. Improved Halbach sensor

for NMR scanning of drill cores. Magn Reson Imaging 2007; 25:

474-80.

[34] Anferova S, Talnishnikh E, Anferov V, et al. Determination of

porosity and novel 2D relaxation and diffusion experiments with

mobile NMR halbach scanner. Magn Reson Imaging 2007; 25:

547-8.

[35] Blümich B, Blümler P, Eidmann G, et al. The NMR-mouse: con-

struction, excitation, and applications. Magn Reson Imaging 1998;

16: 479-84.

[36] Blümich B, Casanova F, Perlo J, et al. Advances of unilateral mo-

bile NMR in nondestructive materials testing. Magn Reson Imag-

ing 2005; 23: 197-201.

[37] Kühn H, Klein M, Wiesmath A, et al. The NMR-MOUSE®: qual-

ity control of elastomers[small star, filled]. Magn Reson Imaging

2001; 19: 497-9.

[38] Kleinberg RL, Straley C, Kenyon WE, Akkurt R, Farooqui SA.

Nuclear Magnetic Resonance of Rocks: T1 vs. T2. Society of Petro-

leum Engineers 68th Annual Technical Conference and Exhibition

of the Society of Petroleum Engineers, Houston 1993; pp. 553-63.

[39] Kimmich R, Anoardo E. Field-cycling NMR relaxometry. Prog

Nucl Magn Reson Spectrosc 2004; 44: 257-320.

[40] Brownstein KR, Tarr CE. Importance of classical diffusion in

NMR studies of water in biological cells. Phys Rev A 1979; 19:

2446-53.

[41] Godefroy S, Korb JP, Fleury M, Bryant RG. Surface nuclear mag-

netic relaxation and dynamics of water and oil in macroporous me-

dia. Phys Rev E: Stat Nonlinear Soft Matter Phys 2001; 64:

021605.

[42] Provencher SW. A constrained regularization method for inverting

data represented by linear algebraic or integral equations. Comput

Phys Commun 1982; 27: 213-27.


[43] Borgia GC, Brown RJS, Fantazzini P. Examples of marginal reso-

lution of NMR relaxation peaks using UPEN and diagnostics.


[44] Borgia GC, Brown RJS, Fantazzini P. Uniform-penalty inversion

of multiexponential decay data. J Magn Reson 1998; 132: 65-77.

[45] Schaumann GE, Hobley E, Hurraß J, Rotard W. H-NMR Re-

laxometry to monitor wetting and swelling kinetics in high organic

matter soils. Plant Soil 2005; 275: 1-20.

[46] Prebble RE, Currie JA. Soil water measurement by a low-

resolution nuclear magnetic resonance technique. Eur J Soil Sci

1970; 21: 273-88.

[47] McBrierty VJ, Wardell GE, Keely CM, O'Neill EP, Prasad M. The

characterization of water in peat. Soil Sci Soc Am J 1996; 60: 991-

1000.

[48] Delville A, Letellier M. Structure and dynamics of simple liquids in

heterogeneous condition: an NMR study of the clay-water inter-

face. Langmuir 1995; 11: 1361-7.

[49] Bryar TR, Daughney CJ, Knight RJ. Paramagnetic effects of

iron(III) species on nuclear magnetic relaxation of fluid protons in

porous media. J Magn Reson 2000; 142: 74-85.

[50] Kenyon WE, Kolleeny JA. NMR surface relaxivity of calcite with

adsorbed Mn2+. J Colloid Interface Sci 1995; 170: 502-14.

[51] Jaeger F, Rudolph N, Lang F, Schaumann GE. Effects of soil solu-

tion’s constituents on proton NMR relaxometry of soil samples.

Soil Sci Soc Am J 2008; 72: 1694-707.

[52] Todoruk TR, Langford CH, Kantzas A. Pore-scale redistribution of

water during wetting of air-dried soils as studied by low-field NMR

relaxometry. Environ Sci Technol 2003; 37: 2707-13.

[53] Todoruk TR, Litvina M, Kantzas A, Langford CH. Low-field NMR

relaxometry: a study of interactions of water with water-repellant

soils. Environ Sci Technol 2003; 37: 2878-82.

[54] Schaumann GE, Hurrass J, Müller M, Rotard W. Swelling of or-

ganic matter in soil and peat samples: insights from proton relaxa-

tion, water absorption and PAH extraction. In: Ghabbour EA, Da-

vies G, Eds. Humic Substances: Nature's Most Versatile Materials.

New York: Taylor and Francis, Inc., 2004: pp. 101-17.

[55] Hinedi ZR, Kabala ZJ, Skaggs TH, Borchardt DB, Lee RWK,

Chang AC. Probing soil and aquifer material porosity with nuclear

magnetic resonance. Water Resour Res 1993; 29: 3861-6.

[56] Hum FM, Kantzas A. Using low-field NMR to determine wettabil-

ity of, and monitor fluid uptake in, coated and uncoated sands. J

Canad Petroleum Technol 2006; 45: 23-8.

[57] Manalo FP, Kantzas A, Langford CH. Soil wettability as deter-

mined from using low-field nuclear magnetic resonance. Environ-

SciTechnol 2003; 37: 2701-6.

[58] Hurrass J, Schaumann GE. Hydration kinetics of wettable and

water repellent soils. Soil Sci Soc Am J 2007; 71: 280-8.

[59] Ohkubo T, Kikuchi H, Yamaguchi M. An approach of NMR re-

laxometry for understanding water in saturated compacted bento-

nite. Phys Chem Earth 2008; 33: S169-S76.

[60] Bird NRA, Preston AR, Randall EW, Whalley WR, Whitmore AP.

Measurement of the size distribution of water-filled pores at differ-

ent matric potentials by stray field nuclear magnetic resonance. Eur

J Soil Sci 2005; 56: 135-43.

[61] Schaumann GE, Bertmer M. Do water molecules bridge soil or-

ganic matter molecule segments? Eur J Soil Sci 2008; 59: 423-9.

[62] Howard JJ. Quantitative estimates of porous media wettability from

proton NMR measurements. Magn Reson Imaging 1998; 16: 529-

33.

[63] Liaw H-K, Kulkarni R, Chen S, Watson AT. Characterization of

fluid distributions in porous media by NMR techniques. AIChE J

1996; 42: 538-46.

[64] Kinchesh P, Samoilenko AA, Preston AR, Randall EW. Stray field

nuclear magnetic resonance of soil water: development of a new,

large probe and preliminary results. J Environ Qual 2002; 31: 494-

9.

[65] Kleinberg RL. Nuclear magnetic resonance pore scale investigation

of permafrost and gas hydrate sediments. In: Rothwell G, Ed. New

ways of looking at sediment cores and core data. London: Geologi-

cal Society 2005: pp. 1-31.

[66] Jaeger F, Bowe S, Van As H, Schaumann GE. Evaluation of 1H

NMR relaxometry for the assessment of pore size distribution in

soil samples. Eur J Soil Sci 2009; 60: 1052-64.

[67] Grønras T, Rueslåtten H, Roaldset E, Skjetne T. NMR responses to

kaolinite in sand. Magn Reson Imaging 1996; 14: 961-2.

[68] Culligan PJ, Sinfield JV, Maas WE, Cory DG. Use of NMR relaxa-

tion times to differentiate mobile and immobile pore fractions in a

wetland soil. Water Resour Res 2001; 37: 837-42.

[69] Schaumann GE, Hobley E, Hurrass J, Rotard W. Changes of poros-

ity and soil physical chemistry due to drying and re-wetting cycles.

Mitt Dtsch Bodenkdl Ges 2003; 101: 39-40.

[70] Chibowski E, Holysz L. On the use of Washburn's equation for

contact angle determination. J Adhes Sci Technol 1997; 11: 1289-

301.

[71] Deczky K, Langford CH. Application of water nuclear magnetic

resonance relaxation times to study of metal complexes of the solu-

ble soil organic fraction fulvic acid. Can J Chem 1978; 56: 1947-

51.

[72] Melton JR, Kantzas A, Langford CH. Nuclear magnetic resonance

relaxometry as a spectroscopic probe of the coordination sphere of

a paramagnetic metal bound to a humic acid mixture. Anal Chim

Acta 2007; 605: 46-52.

[73] Nanny MA, Maza JP. Noncovalent iteractions between mnoaro-

matic compounds and dissolved humic acids: a deuterium NMR T1

relaxation study. Environ Sci Technol 2001; 35: 379-84.

[74] Smejkalova D, Piccolo A. Host-guest interactions between 2,4-

dichlorophenol and humic substances as evaluated by 1H NMR re-

laxation and diffusion ordered spectroscopy. Environ Sci Technol

2008; 42: 8440-5.

[75] Dixon AM, Mai MA, Larive CK. NMR investigation of the interac-

tions between 4'-fluoro-1'-acetonaphthone and the suwannee river

fulvic acid. Environ Sci Technol 1999; 33: 958-64.

[76] Nanny MA, Bortiatynski JM, Hatcher PG. Noncovalent interac-

tions between acenaphthenone and dissolved fulvic acid as deter-

mined by 13C NMR T1 relaxation measurements. Environ Sci

Technol 1997; 31: 530-4.

[77] Bryar TR, Knight RJ. Sensitivity of nuclear magnetic resonance

relaxation measurements to changing soil redox conditions. Geo-

phys Res Lett 2002; 29(24): 2197.

[78] Madinelli G. NMR imaging application to teh study of adsorp-

tion/precipitation of chemicals inside porous media. Magn Reson

Imaging 1998; 16: 665-8.

[79] Wang K, Dickinson LC, Ghabbour EA, Davies G, Xing B. Proton

spin-lattice relaxation times of humic acids as determined by solu-

tion NMR. Soil Sci 2003; 168: 128-36.

[80] Gunasekara AS, Simpson MI, Xing B. Identification and charac-

terization of sorption domains in soil organic matter using strucu-

turally modified humic acids. Environ Sci Technol 2003; 37: 852-

8.

[81] LeBoeuf EJ, Weber Jr WJ. Macromolecular characteristics of natu-

ral organic matter. 2: sorption and desorption behavior. Environ Sci

Technol 2000; 34: 3632-40.

[82] Young KD, LeBoeuf EJ. Glass transition behavior in a peat humic

acid and an aquatic fulvic acid. Environ Sci Technol 2000; 34:

4549-53.

[83] Schaumann GE, LeBoeuf EJ. Glass transitions in peat-their rele-

vance and the impact of water. Environ Sci Technol 2005; 39: 800-

6.

[84] Schaumann GE. Matrix relaxation and change of water state during

hydration of peat. Colloids Surf A 2005; 265: 163-70.

[85] Schaumann GE, LeBoeuf EJ, DeLapp RC, Hurrass J. Thermome-

chanical analysis of air-dried whole soil samples. Thermochim

Acta 2005; 436: 83-9.

[86] Flemming H-C, Wingender J. Extracellular polymeric substances

(EPS)-the construction material for biofilms. Vom Wasser 2000;

94: 245-66.


[87] Hoskins BC, Fevang L, Majors PD, Sharma MM, Georgiou G.

Selective imaging of biofilms in porous media by NMR relaxation.

J Magn Reson 1999; 139: 67-73.

[88] Jaeger F, Grohmann E, Schaumann GE. 1H NMR Relaxometry in

natural humous soil samples: insights in microbial effects on re-

laxation time distributions. Plant Soil 2006; 280: 209-22.

[89] Mills AL, Powelson DK. Bacterial interactions with surfaces in

soils. In Fletcher M, Ed. Molecular and ecological diversity of bac-

terial adhesion. New York: Wiley-Liss, Inc. 1996; pp. 25-57.

[90] Kirchmann H, Gerzabek MH. Relationship between soil organic

matter and micropores in a long-term experiment at Ultuna, Swe-

den. J Plant Nutr Soil Sci 1999; 162: 493-8.

[91] Meiboom S, Gill D. Modified spin-echo method for measuring

nuclear relaxation times. Rev Sci Instrum 1958; 29: 688-91.

[92] Butler JP, Reeds JA, Dawson SV. Estimating solutions of first kind

integral equations with nonnegative constraints and optimal

smoothing. SIAM J Numer Anal 1981; 18: 381-97.

[93] Flemming HC, Wingender J. Relevance of microbial extracellular

polymeric substances (EPSs)--Part II: Technical aspects. Water Sci

Technol 2001; 43: 9-16.

[94] Radosta S, Schierbaum F, Yuriev WP. Polymer-water interaction

of maltodextrins. Part II: NMR study of bound water in liquid

maltodextrin-water systems. Starch-Stärke 1989; 41: 428-30.

[95] Preston AR, Bird NRA, Kinchesh P, Randall EW, Whalley WR.

STRAFI-NMR studies of water transport in soil. Magn Reson Im-

aging 2001; 19: 561-3.

[96] Vogt C, Galvosas P, Klitzsch N, Stallmach F. Self-diffusion studies

of pore fluids in unconsolidated sediments by PFG NMR. J Appl

Geophys 2002; 50: 455-67.

Received: May 04, 2009 Revised: November 17, 2009 Accepted: January 25, 2010

© Bayer et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

Proton NMR Relaxometry as a Useful Tool to Evaluate Swelling Processes in Peat Soils

Fabian Jaegera, Anastasia Shchegolikhina

b, Henk Van As

c and Gabriele Ellen Schaumann*

,a

aDepartment of Environmental and Soil Chemistry, Institute of Environmental Sciences, Universität Koblenz-Landau,

Fortstr. 7, 76829 Landau, Germany

bGeographical Institute, Ruhr-University Bochum, Universitaetsstrasse 150, 44801 Bochum, Germany.

cLaboratory of

Biophysics and Wageningen NMR Centre, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands

Abstract: Dramatic physical and physico-chemical changes in soil properties may arise due to temperature and moisture

variations as well as swelling of soil organic matter (SOM) under constant conditions. Soil property variations may influ-

ence sorption/desorption and transport processes of environmental contaminants and nutrients in natural-organic-matter-

rich soils. Notwithstanding the studies reported in literature, a mechanistic model for SOM swelling is unavailable yet.

The objective of the present study was the evaluation of the swelling of peat soils, considered as SOM models, by 1H

NMR relaxometry and differential scanning calorimetry (DSC). Namely, information on the processes governing physical

and physicochemical changes of peat during re-hydration were collected. The basic hypothesis of the present study was

that the changes are slow and may affect water state as well as amounts of different water types into the peats. For this

reason, such changes can be evidenced through the variations of mobility and thermal behaviour of the involved H2O

molecules by using 1H NMR relaxometry and DSC. According to the experimental results, a mechanistic model, describ-

ing the fundamental processes of peat swelling, was obtained. Two different peats re-wetted at three temperatures were

used. The swelling process was monitored by measuring spin-spin relaxation time (T2) over a hydration time of several

months. Moreover, DSC, T1 – T2 and T2 – D correlation measurements were done at the beginning and at the end of the

hydration. Supplementary investigations were also done in order to discriminate between the swelling effects and the con-

tributions from soil solution, internal magnetic field gradients and/or soil microorganisms to proton relaxation. All the re-

sults revealed peat swelling. It was evidenced by pore size distribution changes, volumetric expansion and redistribution

of water, increasing amounts of nonfreezable and loosely bound water, as well as formation of gel phases and reduction of

the translational and rotational mobility of H2O molecules. All the findings implied that changes of the physical and phys-

icochemical properties of peats were obtained. In particular, three different processes having activation energies com-

prised in the interval 5 – 50 kJ mol-1

were revealed. The mechanistic model which was, then, developed included water

reorientation in bound water phases, water diffusion into the peat matrix and reorientation of SOM chains as fundamental

processes governing SOM swelling. This study is of environmental significance in terms of re-naturation and re-watering

of commercially applied peatlands and of sorption/desorption and transport processes of pollutants and nutrients in natural

organic matter rich soils.

Keywords: Soil organic matter, swelling, kinetics, unfreezable water, differential scenning calorimetry, NMR relaxometry, pore water.

INTRODUCTION

Natural soils are exposed to dynamic variations in tem-perature and moisture. Changes in moisture status may affect soil properties like water content or volumetric swelling of soil organic matter (SOM) [1, 2] or sorbent properties [2-5]. Furthermore, the extractability of organic pollutants was found to be affected by the hydration time of soils indicating physical and physicochemical changes of SOM upon swel-ling [2]. Although it is known that SOM swelling includes volumetric expansion [1], water redistribution and re-opening of small-sized pores [6] as well as SOM re-organisation [7, 8] a mechanistic model describing the

*Address correspondence to this author at the Department of Environmental

and Soil Chemistry, Institute of Environmental Sciences, Universität,

Koblenz-Landau, Fortstr. 7, 76829 Landau, Germany; Tel: +49(6341) 280-

571; Fax: +49(6341) 280-576; E-mail: [email protected]

fundamental processes of SOM swelling is not available un-til now.

Peat soils may be considered as concentrated analogues of SOM, and, therefore, they were used as models for envi-ronmental studies of pollutant sorption and transport in SOM [9]. Peats are the product of accumulation and humification of plant material in certain special wet habitats [7, 9], and are commercially used as fuel and horticulture medium [7].

Beside peats, many macromolecular substances, such as synthetic polymers or biopolymers, swell when placed in contact with fluids. The swelling is characterised as sorption of the liquid into the macromolecular solid phase, which in turn, increases its volume. This volumetric swelling, Q, was defined as the ratio of swollen to non-swollen volume of the sorbent or macromolecular material; a value of unity indi-cates no swelling. The amount of swelling depends both on the nature of fluid and sorbent. Strongly cross-linked materi-

28 The Open Magnetic Resonance Journal, 2010, Volume 3 Jaeger et al.

als swell less than weakly cross-linked materials. However, swelling seems to be limited by molecular size exclusion effects to fairly small liquid molecules with molar volumes smaller than about 93 cm

3 mol

1 for most soil organic mate-

rials and 88 cm3 mol

1 for cellulose. The Q values for differ-

ent materials after water sorption ranged between 1.3 and 1.6 for peat, 2.0 for cellulose and 1.6 for chitin. [1]

Unlike synthetic polymers, McBrierty et al. [7, 8] re-ported that the peat matrix re-organises during swelling, thereby permitting access to a greater number of water mole-cules to the hydrophilic sites. Furthermore, up to four differ-ent types of water in hydrated peat samples were identified using differential scanning calorimetry (DSC), thermogra-vimetric analysis (TGA) and proton nuclear magnetic reso-nance (NMR) relaxometry. They differentiated between tightly bound or non-freezable water, up to two types of loosely bound or freezable water and freezable bulk water. Non-freezable water is predominantly hydration water and/or water that interacts chemically with hydrophilic moieties in the matrix showing glassy behaviour with a transition tem-perature around -123°C to -83°C [7]. The character of loosely bound water deviates less dramatically from that of normal water. This water melts with further increase in tem-perature, but at lower temperatures than freezable bulk water that melts around 0°C [7, 10]. In hydrogels, at least three kinds of water were determined: hydrated water, interfacial water, with a certain ordered arrangement, and bulk water [11].

Additionally to the calculation of Q, 1H NMR relaxome-

try [2, 6, 7, 12, 13] and DSC [7, 14] were used to study the hydration of peat and mineral soil samples or to characterise different types of water in peats. DSC was used to study the melting and freezing behaviour of water and to calculate the amounts of non-freezable and freezable water in hydrogels [15, 16] or peat soil samples [7, 14]. The melting of freezable water in moist samples is characterised by endo-thermic water melting peaks in the DSC thermograms, which originate from the additional energy uptake of the melting process. The enthalpies of these peaks are often smaller than it would be expected regarding the amount of water present in the samples, which is due to the existence of non-freezable water. In general, the melting peaks of water in hydrogels and peat differ strongly from those of free pure water or solutions. They are broader and split into two or more overlaying peaks. Radosta and Schierbaum found a splitting of the melting peak of water only in maltodextrin gels, but not in maltodextrin solutions, although the deter-mined amounts of non-freezable water were comparable for both systems [15, 17]. Consequently, they attributed the melting peak splitting to the entrapment of water within the gel matrix. In peat soil samples, the splitting of the melting peak was ascribed to the existence of loosely bound and freezable bulk water [7, 14].

Swelling kinetics experiments of soil samples using 1H

NMR relaxometry have shown that the swelling process is linked to the migration of the peak relaxation times towards smaller relaxation times and the increase of the amount of water protons relaxing at smaller relaxation times [2, 6, 12, 13]. These findings were attributed to a redistribution of wa-ter during swelling. Todoruk et al. [6] identified two proc-

esses with time constants of about 1 d for a fast process and up to 22 d for a slow process for different soils, which were comparable to the values reported by Schaumann et al. [2, 12] and Jaeger et al. [13]. The calculated activation energies of these two processes ranged from 14 kJ mol

-1 to 117

kJ mol-1

suggesting diffusion processes and chemical reac-tions, such as ester hydrolysis [6]. The slow process was attributed to the formation of gel phases and to the slow wa-ter intrusion into micropores, which re-opened during swel-ling.

The changes in the relaxation time distributions during swelling of soils were found to be pronounced stronger in soils with higher microbial respiratory activity [13] and in soils inoculated with a biofilm producing bacteria isolate [18]. Consequently, soil microorganisms may contribute to the swelling of soils due to the production and release of extracellular polymeric substances (EPS) and the formation of biofilm. Furthermore, the proton relaxation in soil solu-tions (bulk relaxation) may significantly contribute to the total proton relaxation of water in soil samples and the con-tribution of the bulk relaxation may increase during hydra-tion due to dissolution of paramagnetic iron and manganese or changes of their chemical speciation [19]. Keating and Knight [20] found for iron-oxide coated sands that some of the studied surface species of iron produced internal field gradients causing additional transverse proton relaxation due to spin diffusion in internal field gradients at 2.2 MHz. Thus, a contribution of this additional transverse proton relaxation mechanism cannot be excluded a priori for hydrated soil samples measured at lower magnetic field strength.

In general, transverse relaxation times (T2) are used to study the swelling of soil samples [6, 12, 13], because of the much faster determination compared to longitudinal relaxa-tion times (T1). However, many swellable materials, such as cellulose [21], hydrogels with different degrees of cross-linking [22], wheat starch [23] and maltodextrin [17] show strong differences between T1 and T2. Their T1/T2 ratios were found to be much larger than for free pure water (T1/T2 ~ 1) or for rock core samples with T1/T2 ratios generally between 2 to 3 [24]. The larger T1/T2ratios in those swellable materi-als were attributed to a water structuring caused by the polymer, which resulted in reduced rotational mobility of water molecules inside the structured water phase [17]. This reduced rotational mobility affects T1 and T2 differently [25]. The structured water is non-freezable water [26, 27] that exchanges very fast with the bulk water phase, which results in only one average relaxation time much smaller than for bulk water [23]. In starch pastes, the T1/T2 ratio of the bound water was found to be 22 and the one determined from the average relaxation times was about 8 [23]. As a conse-quence, both T1 and T2 measurements may be helpful to in-vestigate water mobility and a possible structuring of water caused by the SOM matrix in swollen soils.

We hypothesised that the physical and physicochemical properties of peat soil samples change during volumetric swelling at constant temperature and moisture conditions. These changes are slow and affect the state of water as well as the amounts of different water types inside the peats. Thus, they can be observed through changes in mobility and thermal behaviour of the involved water molecules, which

Swelling of Peat Studied by 1H NMR Relaxometry The Open Magnetic Resonance Journal, 2010, Volume 3 29

can be determined by 1H NMR relaxometry and DSC. Our

objectives were to study the swelling of peat soils via 1H

NMR relaxometry and DSC to characterise the governing processes causing physical and physicochemical changes of peat during re-hydration at constant temperature and mois-ture conditions. For that we have combined T1 and T2 meas-urements together with DSC measurements for the first time to study the swelling of peat soil samples. Swelling kinetics experiments were carried out with two re-wetted peats at three different temperatures. Additional investigations were carried out to distinguish between swelling effects and fur-ther influences on the proton relaxation process caused by the soil solution, internal magnetic field gradients and/or soil microorganisms.

MATERIALS AND METHODOLOGY

Peat Soil Samples

The peat material used in this study was taken from the peat land “Totes Moor” in the nature park “Steinhuder Meer“ near Hannover, Germany. The samples of peat were collected from the drained part of the bog from two different layers: a fibric peat (peat 1) was collected from an upper layer with low degree of decomposition and a well decom-posed sapric peat (peat 2) was taken from 1.1 m depth. After sampling, the peats were air dried at room temperature and characterised (Table 1). For NMR measurements, the air dried peat samples were ground, 2 mm sieved and stored at 19°C until further usage.

Table 1. Description and Some Properties of the Two Studied

Peat Soil Samples. CECeff Effective Cation Exchange

Capacity. DOC Dissolved Organic Carbon. wcair dried

Gravimetric Water Content of the Air Dried Peat

Samples Related to Dry Mass (d.m.) Determined

after Oven Drying for 24 h at 105°C

Fibric Peat Sapric Peat

Label Peat 1 Peat 2

Color Brown Black

Sampling depth /m 0.4 1.1

Ash-content /% 1.25 1.28

pH (CaCl2 0.01 M) 2.7 2.7

Corg /% 45 52

C/N 405 219

DOC /mg L-1 89 56

CECeff /mmolc kg-1 166 123

wcair dried /g g d.m. 0.152 0.161

Sampling coordinates 52° 30' 26.41'' N

9° 21' 14,28'' E

1H NMR Relaxation in Porous Media

In porous materials, the proton relaxation is strongly ac-celerated by interactions between water protons and surfaces

[detailed information can be found in e.g. 12, 28-30]. As a result, the measured relaxation time of water protons is de-termined by the proton relaxation time in the pore space, T1,2B, the surface relaxation time, T1,2S, and the diffusion re-laxation time, T2D, due to spin diffusion in internal magnetic field gradients, which affects the transverse proton relaxation time (T2), but not the longitudinal proton relaxation time (T1) [28, 31].

1

T1=1

T1B+1

T1S=1

T1B+

1S

V (1)

1

T2=1

T2B+1

T2S+1

T2D=1

T2B+

2S

V+1

T2D (2)

With SV 1= r 1 the proton relaxation time is connected

to the pore diameter, dpore = 2r , where r is the pore radius and

= 1, 2, or 3 is the shape factor for planar, cylindrical, and

spherical pore geometry, respectively [32]. SV-1

is the pore

surface to pore volume ratio, 1,2 is the surface relaxivity,

which is strongly affected by paramagnetic ions on the sur-

face like Mn(II) [33] and Fe(III) [34]. T2D is related to the

average internal gradient of the magnetic field, G, and the

self-diffusion coefficient of water, D, by

1

T2D=D

12G tE( )

2, (3)

where is the gyromagnetic ratio, and tE is the echo time, which is the time between two 180° pulses in the Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence [35].

Eq. 1 and 2 are valid if the condition for the fast-diffusion

regime, 2V

SD<< 1 is fulfilled [31]. This means that the relaxa-

tion process is surface-limited and the diffusion of water

protons towards the surface is very fast and can therefore be

neglected. Various rock core samples, e.g. sandstones [28,

36], were assigned to the fast-diffusion regime [31], whereas

some sandy soil samples [37] were in the transition from the

intermediate- to the slow-diffusion regime [31]. However, it

has been shown that Eq. 2 can be also used for the calcula-

tion of pore sizes in soils by assuming the fast-diffusion re-

gime in these soil samples and by using two surface relaxiv-

ities for each soil, one for micro- and one for mesopores,

[37].

Sample Preparation and 1H NMR Measurements

Sample Preparation

1.00 g dry mass (d.m.) of peat 1 and 3.55 g d.m. of peat 2 of the air dried samples were filled in glass vessels and rewetted to their maximal water holding capacity (i.e. maxi-mal water content). The maximal water content was 15.67 g g

-1 d.m. for peat 1 and 4.51 g g

-1 d.m. for peat 2. Thus, the

total amount of water was with 15.67 g (peat 1) and 16.01 g (peat 2) comparable for the two rewetted peats.

The protocol of the rewetting procedure was the same for all studied peat samples: After placing the air dried peat sample into the glass vessel, the water (demineralised water) was added and the peat was carefully mixed with the water


using a thin spatula. After wetting of all surfaces, i.e. the peat surfaces turned dark and no dry areas were observed, the vessel was sealed with a plastic lid and knocked ten times vertically on a solid surface to obtain comparable bulk densities. After centrifugation at 4000 RPM at the beginning and end of swelling (see below), the contact angle of the peat samples was determined using the sessile drop method de-scribed by Diehl and Schaumann [38] to test the wettability of the surfaces. Before starting the rewetting procedure, both the peat sample and the water were adjusted to the respective temperature of 5°C, 19°C or 30°C. In the course of swelling of 5 to 7 months (hydration time), the sealed glass vessels were stored in desiccators with 99.9 % relative humidity in darkness at the respective temperature (± 1.5°C).

Swelling Kinetics at 5°C, 19°C and 30°C by 1H NMR

Measurements

All one-dimensional 1H NMR measurements to deter-

mine the swelling kinetics, the water distribution versus cen-

trifugation speed as well as the amounts of unfrozen water at

-34°C and -5°C were performed at a magnetic field strength

of 0.176 T, i.e. at a proton Larmor frequency of 7.5 MHz

(Minispec 7.5, Bruker, Germany). The temperatures during

the measurements were kept constant (± 0.5°C) using a

XR401 Air-JetTM

sample cooler (FTS Systems, Stone Ridge,

USA) with compressed air as gas stream. The relaxation time

distributions were determined using a MATLAB program

developed by Veevaete [39] applying the BRD (Butler,

Reeds and Dawson) algorithm [40]. The relaxation time dis-

tributions consisted of 200 time constants (TC) with associ-

ated amplitudes. The sum of the amplitudes equals the total

NMR signal intensity at time 0=t ( SIt=0 ), which is a meas-

ure of the total amount of water protons in a sample. All one-

dimensional 1H NMR measurements were performed with

duplicates of the respective peat samples. In this study, we

present the results of the two repetition samples (RS 1 and

RS 2) separately, because the reproducibility between the

duplicates was partly not given.

The two-dimensional T1 - T2 and T2 - D correlation meas-urements were performed at a proton Larmor frequency of 30 MHz at 20 ± 1°C using a home-made NMR spectrometer controlled by a MARAN Ultra console (Resonance In-struments Ltd, Oxfordshire, UK) with an additional pulsed field gradient (PFG) unit. For T1 - T2 correlated measure-ments the inversion recovery method was combined with a CPMG sequence. 25 steps of the inversion recovery time between 750 s and 15 s were used to sample a T1 relaxation decay curve. In the T2 - D correlation measurement a diffu-sion-weighted multi-spin-echo pulse sequence [41] was used to determine the correlation between T2 and the self-diffusion coefficient of water in the peat samples. 25 gradi-ent steps with a maximal gradient strength of 1.2 T m

-1 were

used. Small and big delta were 1 ms and 10 ms, respectively. Two-dimensional T1 - T2 and T2 - D correlation was ana-lyzed by 2D ILT – two-dimensional numerical inverse Laplace transformation [42-44].

The longitudinal relaxation time (T1) distributions of wa-ter in the peat soil samples were determined using an inver-sion recovery (IR) sequence with 25 IR-points between 1.0 ms and 19 s. Because of the long measurement time for T1 of

about 75 min, the T1 distribution was only determined direct after water addition (after 110 - 180 min) and at the end of the swelling experiment after 5 to 7 months. A CPMG pulse sequence [35] with an echo time of 500 s and 50,000 ech-oes was used to obtain transverse relaxation time (T2) distri-butions of water in the peat soil samples in the course of peat swelling. A total of 30 to 36 T2 measurements were per-formed to follow the peat swelling over 5 to 7 months. Dur-ing the first 30 minutes after water addition, T2 measure-ments were performed every five minutes. Within the first three to seven hours after water addition, a total of up to eight T2 measurements were performed to determine fast changes in the T2 distributions. The effect of the additional transverse relaxation due to spin diffusion in internal field gradients was tested according to Keating and Knight [20] determining T2 as a function of increasing echo times (500 to 800 s).

Cryo - NMR Relaxometry: Determination of

Non-Freezable and Loosely Bound Water

1H NMR relaxometry experiments at -34°C and -5°C

were carried out to determine the amounts of water that were left unfrozen in the peats at these temperatures. The first represents the lowest temperature that was reachable in the peats using the Air-Jet

TM sample cooler. The unfrozen water

at -34°C is referred to as tightly bound or non-freezable wa-ter [7] in this study. It was tested by DSC whether this water remained unfrozen between -90°C and -34°C. From DSC it was found that melting of soil solution extracted from the hydrated peats started above -4°C. Thus, the second tempera-ture of -5°C was operationally chosen to measure the amounts of loosely bound water in peat [7]. This kind of NMR measurement at low temperatures is referred to as Cryo - NMR relaxometry hereafter. The amounts of these two types of water in peat were determined from the NMR signal intensity using a T2 Hahn-echo [45] sequence with 25 echo points with increasing echo times between 0.05 ms and 7.5 ms at the beginning and end of peat swelling. T2 of non-freezable and loosely bound water was calculated by mono-exponential fitting to the decay curves determined at -34°C and -5°C, respectively. The wet peat samples were first cooled down to -34°C inside the NMR probe using the Air-Jet

TM sample cooler. The freezing process of water was char-

acterised by a dramatic decrease of the NMR signal inten-sity, because the very short T2 of ice of about 10 s [46] could not be detected due to the dead time of 50 s of the used NMR device. The signal intensity was constant after about 60 min of cooling and the Cryo-NMR measurement was carried out. After that the sample was adjusted to -5°C until constant signal intensity was observed and subse-quently measured. Two additional samples (prepared as de-scribed above) were used to determine the non-freezable and loosely bound water at the beginning of peat swelling. For the determination of both water types at the end of swelling, the original samples from the swelling kinetics at 30°C were used.

Water Distribution Versus Centrifugation Speed

The water distribution in the peat samples versus cen-trifugation speed was determined at the beginning and end of swelling and after 2 days of hydration (no repetition sample) using a Universal 320 centrifuge with the rotor 1494 (Het-


tich, Tuttlingen, Germany). The original samples from the swelling kinetics at 30°C were used to determine the water distribution at the end of swelling. Additional repetition samples were used for the determination of the water distri-bution at the beginning of swelling and after 2 days of hydra-tion. The peat samples were re-wetted as described above and filled into centrifuge tube filters with a 10- m filter in-sert (VectaSpin 20, Whatman). Then they were centrifuged step-wise at nine centrifugation speeds between 500 RPM and 4000 RPM for 15 min at 19°C. The peat samples were desaturated during centrifugation. After every centrifugation step, the gravimetric water contents and the T2 and T1 distri-butions of water in the peat samples at 19°C were deter-mined. Instead of the largest T2 and T1 in the relaxation time distribution, T2 and T1 at 90 % of total sum of amplitudes were determined to reduce variation between replicates [37]. After the last centrifugation step, T1B and T2B were measured in the extracted soil solution, which was gained during cen-trifugation, to determine the contribution of the soil solution to the total proton relaxation in the wet peat samples.

T1/T2 Ratios as a Measure of the Water State in Peat

Samples

T1/T2 ratios were determined to characterise the state of

water in peat samples. For that, we calculated the T1/T2 ra-

tios from the relaxation time distributions of water in peat of

the fully water saturated samples and the step-wise desatu-

rated samples after centrifugation at the beginning and end

of hydration time as a function of the relative water content

(rel. wc). Relative water contents in the fully saturated sam-

ples were calculated from the quotient of the subtotal of the

amplitudes (AT) of the 200 time constants (TC) in a relaxa-

tion time distribution and the total sum of the amplitudes

( SIt=0 )

rel.wc =

ATii=1

200

SIt=0100.

(4)

The T1/T2 ratios in the saturated samples were deter-mined from TC in the T1 and T2 distributions at 25 relative water contents between 7.5 % and 100 %. In the desaturated samples, relative water contents were calculated from the quotient of the gravimetric water content after step-wise cen-trifugation (wccen.) and the total water content at saturation (wctot)

rel.wc =wccentwctot

100. (5)

T1/T2 ratios in these samples were determined from T2 and T1 at 90 % of total sum of amplitudes after every cen-trifugation speed (see above).

DSC Measurements

The freezing/melting behaviour of water inside the wet peat samples was studied between -90°C and 40°C using a TA Instruments Model Q1000 DSC (TA Instruments, Al-zenau, Germany) with nitrogen as purge gas (50 mL min

-1).

Heat flow and temperature calibration were carried out using

Indium. For the DSC measurements ~3 mg - 10 mg of wet and partly desaturated peat soil from the samples used in the centrifugation experiment were weighed into aluminium pans and hermetically sealed using an aluminium lid. The cooling/heating protocol started with an equilibration at 40°C followed by a cooling cycle down to -90°C with cool-ing rates of 3 K min

-1 until -75°C followed by 2 K min

-1 until

-90°C. After equilibration at -90°C, the sample was heated up to 40°C with a heating rate of 5 K min

-1. Data analysis

(determination of peak onset and peak maximum tempera-tures, and melting enthalpy) was performed using the Uni-versal Analysis 2000 software by TA Instruments. The peak maximum temperature is referred to as peak position is this study.

According to [14], the melting peak was investigated at four different heating rates between 0.5 K min

-1 and 10

K min-1

to test whether the melting transition is kinetically controlled. For that, the desaturated peat samples from the centrifugation experiment at 4000 RPM, agar gel (30 g L

-1),

water saturated sand with particle sizes of 150 – 200 m (referred to as sand 150 m) and free pure water were used.

The melting enthalpy of freezable water, H f , in the

peat samples as well as the amount of non-freezable water

were calculated from the DSC thermograms of the desatu-

rated peat samples from the centrifugation experiment. Ac-

cording to McBrierty et al. [7], both were calculated from

the integrated change in enthalpy (per gram of dry peat)

which was plotted as a function of water content. The slope

of a fitted straight line provides H f and the intercept on

the water content axis is the amount of non-freezable water.

Microbial Respiratory Activity

For the incubation experiment the samples were placed in vessels and wetted to 60 % water holding capacity. The rewetted peats (as triplicates) were incubated for 21 days at a temperature of 20°C in a Respicond-apparatus (Nordgren Innovations, Bygdea, Sweden), which determined the CO2-evolution hourly from the changes in electrical conductivity in 10 ml of 0.6 M KOH solution placed inside the incubation vessels [47].

Calculation of Time Constants and of Apparent Activation Energy the Rate-Limiting Step of Peat Swelling

The T2 distributions of water in the peat samples used in

the swelling kinetics experiment were operationally divided

into four T2 ranges with fixed T2 limits (I: 0.1 - 3 ms; II: 3 -

30 ms; III: 30 - 300 ms and IV: > 300 ms). The relative

amplitudes of the four T2 ranges were calculated (i.e. the

sum of amplitudes of the respective T2 range divided by the

total sum of amplitudes or SIt=0 ) and plotted as a function of

hydration time. Previous studies have shown that swelling

kinetics of soil samples can be described by first-order proc-

esses using exponential functions [12, 13]. We tested fittings

with up to four exponential functions to describe the time

dependent change of the relative amplitudes A(t) of the T2

range IV (T2 > 300ms), which represented the intrusion of

water into the peat matrix, and to calculate the time constants

of the rate-limiting processes of peat swelling. The best re-


sults (large R2, small standard errors and good reproducibil-

ity between replicates) were determined using a sum of three

exponentials

A(t) = Ai expt

ii=3

+ A . (6)

Ai (i = 3) is the relative amplitude of each exponential

function, A the relative amplitude for infinite hydration

time, t the hydration time [d], i (i = 3) the time constant of

each first order process [d]. The reciprocal, 1

i

(i = 3) , repre-

sents its rate constant, k, [d-1

].

According to the Arrhenius-Equation, the temperature dependency of the rate constant, k, of the water intrusion allows to calculate the activation energy, EA, which is the minimum energy necessary for a specific process to occur [48]

ln(k) = ln(A*)EAR

1

T. (7)

A* is the pre-exponential factor (commonly written as

A , but changed to A* to avoid confusion with A used for

relative amplitude), R the universal gas constant

( =R 8.314472 J mol1 K

1), T the absolute temperature (K).

The apparent activation energy gives information about the

nature of the rate-limiting step of the investigated process.

Chemical reactions are characterised by EA > 60 kJ mol-1

,

while physically controlled processes require EA < 40 kJ mol-1

[49].

RESULTS AND DISCUSSION

Changes of Relaxation Time Distributions During

Hydration

The T1 and T2 distributions of water protons in the two rewetted peat samples at the beginning and end of the swel-ling experiment are shown in Fig. (1) for peat 1 and Figure 2 for peat 2 for the temperatures 5°C (B+D) and 30°C (A+C). All relaxation time distributions consist of several peaks (up to five peaks for peat 2), which are in some cases hardly dis-tinguishable. However, the relaxation time distributions in Figs. (1 and 2) show three features. Firstly, the shapes of the T1 and T2 distributions at the end of the hydration period are different to those directly after water addition (referred to as start in the Figures). The relative amplitudes at smaller T1 and T2 values increased and those at larger T1 and T2 values decreased for both peat samples. Parallel to this, the T2 val-ues of the peak maxima increased for peaks at smaller T2

Fig. (1). T1 and T2 distributions of water protons in the rewetted peat 1 samples at the beginning and end of the swelling experiment deter-

mined at 5°C (B+D) and 30°C (A+C). The results of the two repetition samples or replicates (RS1 and RS2) are displayed.

5°C

0.1 1 10 100 10000

1

2

3

4

peat 1 RS1 RS2 start end

rel.

amp

litu

de /%

T1 /ms

B

0.1 1 10 100 10000

1

2

3

4


rel.

amp

litu

de /%

T2 /ms

C

0.1 1 10 100 10000

1

2

3

4


rel.

amp

litu

de /%

T2 /ms

D

0.1 1 10 100 10000

1

2

3

4


rel.

amp

litu

de

/%

T1 /ms

A

30°C


(< 20 - 200 ms) and decreased for peaks at larger T2 (> 200 ms). Secondly, the shapes of the T1 distributions differ from the shapes of the T2 distributions. Thirdly, the width of the T1 and T2 distributions for peat 2 are found to be broader than those of peat 1. Furthermore, the relative amplitudes at T1 < 100 ms and at T2 < 10 ms are larger for peat 2.

The sample weights and the total sum of amplitudes

( SIt=0 ) during hydration time of 5 to 7 months was constant

within ± 2 % for all peat samples. After calibration with pure

water at the different temperatures, SIt=0 was used to calcu-

late the NMR detectable amounts of water in the peat sam-

ples, which were comparable to those determined by gra-

vimetric measurements. Hence, all water inside the peat

samples was determined by the 1H NMR relaxometry meas-

urements and no water was lost during the long hydration

times. In the course of hydration, the peat sample volumes

increased and the calculated Q values were 1.5 for peat 1 and

2.0 for peat 2.

The relaxation time distributions of water in the two repetition samples of each peat (referred to as RS 1 and RS 2 in Figs. (1 and 2) were similar with each other for all hydra-tion time points. The only exception was the T1 distribution for the second repetition sample of peat 2 at 30°C (RS 2 in Fig. 2A), which was not comparable to the one for RS 1 (Fig. 2C).

Figs. (1 and 2) also show that the relaxation time distri-

butions were affected by the incubation temperature. An

increase of the peak relaxation times and of peak widths at

higher temperatures was found. Eq. 7 was used to calculate

the activation energies, EA, of the peak relaxation time in-

crease at the beginning and end of hydration time using

T21= k . EA was 4 - 12 kJ mol

-1 for peat 1 and 3 - 16 kJ mol

-1

for peat 2. T2 of pure water increased from 1.5 s at 5°C to 2.8

s at 30°C and EA of this T2 increase was 16.3 ± 0.4 kJ mol-1

.

Thus, the activation energies for the peat samples were

smaller then or comparable to EA of pure water. This sug-

gests that the temperature dependency of the relaxation time

distribution of water in the peat samples can be attributed

mainly to increasing water diffusivity. Furthermore, the posi-

tive relation between peak relaxation time and temperature

suggests that the proton relaxation in the peat samples was

surface limited [50].

The large Q values indicated volumetric swelling [1] of

both peats, but to a larger extent for peat 2. The decrease of

the T2 values for peaks at larger T2 (> 200 ms) is consistent

with the findings reported elsewhere and suggests swelling

of peat [6, 12, 13]. The increase of the T2 values of the peak

maxima for peaks at smaller T2 (< 20 - 200 ms) is contrary to

Jaeger et al. [13] and Todoruk et al. [6]. However, a qualita-

tively comparable peak maxima shifting was also observed

during the swelling of starchy sago beads (data not shown).

The differences between the two repetitions of peat 2 at 30°C

Fig. (2). T1 and T2 distributions of water protons in the rewetted peat 2 samples at the beginning and end of the swelling experiment deter-

mined at 5°C (B+D) and 30°C (A+C). The results of the two repetition samples or replicates (RS1 and RS2) are displayed.

0.1 1 10 100 10000

1

2

3

4 peat 2 RS1 RS2 start end

rel.

amp

litu

de /%

T2 /ms

D

0.1 1 10 100 10000

1

2

3

4


rel.

amp

litu

de

/%

T1 /ms

B

0.1 1 10 100 10000

1

2

3

4


rel.

amp

litu

de /%

T2 /ms

C

0.1 1 10 100 10000

1

2

3

4


rel.

amp

litu

de

/%

T1 /ms

A

5°C30°C


may be due to the two weeks longer hydration time of RS 2

and resulting stronger peat swelling, which may have af-

fected T1 differently in the second repetition than in the first.

Water Distribution Versus Centrifugation Speed

The distribution of water in the two peat samples at the beginning, after 2 days and at the end of hydration time are represented as a function of centrifugation speed in Fig. (3). Direct after water addition (start), centrifugation at 500 RPM resulted in a dramatic decrease of the relative water contents of the peats to about 40 % of the original water contents. No significant changes of the relative water contents were ob-served for the next one to two centrifugation speeds (peat 1 and peat 2, respectively). For centrifugation speeds faster than 800 RPM (peat 1) or 1000 RPM (peat 2), the water con-tents decreased with increasing centrifugation speed, with an end point of ~15 % and ~25 % at 4000 RPM (peat 1 and peat 2, respectively). With increasing hydration time of the peat samples, the relative water contents at all centrifugation speeds increased. This observation was more pronounced for smaller centrifugation speeds, where the relative water con-tents at the end of hydration were almost two times larger than at the beginning. The water contents after two days of hydration for smaller centrifugation speeds were found to be between the values determined at the beginning and the end of hydration. For larger centrifugation speeds, they were comparable to the initial values. The observed changes were more pronounced for peat 2 than for peat 1.

The water distribution of RS 2 of peat 2 at the end of hy-dration (Fig. 3B) was comparable to the one of RS 1 only for centrifugation speeds faster 2000 RPM. For centrifugation speeds slower 2000 RPM, the values of the relative water content were significantly larger. For the other peat samples, the two repetition samples were similar (day 2 was studied without any repetition sample).

With the assumption that every centrifugation speed rep-resents a certain pore size and the represented pore size de-creases with increasing centrifugation speed [51], it can be concluded that the pore size distribution changed signifi-cantly during peat hydration. This change was slow and takes place in two phases. The first phase started very soon

after water addition and lasted for more than two days. It was characterised by the decrease of the number of very large pores (> 50 m) represented by centrifugation speeds of 500RPM and the evolution of medium-sized pores 50 - 10

m [51], represented by centrifugation speeds of up to 1000 RPM. The second phase started anytime after the first two days of hydration and resulted in the ongoing evolution of medium-sized pores 50 - 10 m as well as of small-sized pores 10 - 1 m [51], represented by centrifugation speeds of 1000 - 4000 RPM.

Fig. (4) shows the T1 and T2 relaxation times of water in the desaturated peat samples as a function of centrifugation speed (the same samples as presented in Fig. (3) without the sample of day 2) to test whether the observed changes of the relaxation time distributions in Figs. (1 and 2) were caused by surface relaxivity changes of the peat samples during hy-dration. The T1 and T2 relaxation times decreased with in-creasing centrifugation speeds, but were comparable for the beginning and for the end of hydration.

Again, the only exception was the sample 2 of peat 2 (Fig. 4B+D) for which T1 values were always larger than those of RS 1. In contrast, T2 of RS 2 was larger only for centrifugation speeds slower than 2500 RPM, but compara-ble to those of RS 1 for larger centrifugation speeds. This implies that both the transverse and the longitudinal surface relaxivity decreased in RS 2 of peat 2. However, Fig. (2A+C) show that only T1 but not T2 distributions are differ-ent for the two repetition samples of peat 2 at 30°C. It, there-fore, is more likely that the two repetitions were different in some physical properties which affected the dewatering by centrifugation. For instance, reduced water conductivity due to stronger water binding of entrapped water in a gel phase may result in a less effective dewatering by centrifugation, because smaller speeds cannot overcome the stronger water holding. It was found that at high centrifugation speeds faster 2000 RPM only the T1 but not the T2 and the relative water content values of the two repetition samples of peat 2 were different. Hence, it can be suggested that only the longitudi-nal, but not the transverse surface relaxivity of RS 2 of peat 2 at 30°C increased during hydration. This may be due to a stronger decrease of the water mobility in the vicinity of the peat surface.

Fig. (3). Distributions of water in the two peat samples at the beginning, after 2 days and at the end of hydration time (the 30°C samples at

the end of hydration measured at 19°C) as a function of centrifugation speed. The results of the two repetition samples or replicates (RS1 and

RS2) are displayed (not for day 2).

1000

20

30

40

50

60

708090 peat 1 RS 1 RS 2

start end day 2

40002000

A

rel.

wc

/%

centrifugation speed /RPM500 1000

20

30

40

50

60

708090 peat 2 RS 1 RS 2

start end day 2

40002000

B

rel.

wc

/%

centrifugation speed /RPM500


For peat 1, the T1 and T2 values at 500 RPM were smaller at the beginning than at the end of hydration (Fig. 4A+C). The relative water content at the beginning of hydration was constant between 500 RPM and 650 RPM (Fig. 3). Thus, the smaller T1 and T2 values were due to only pores smaller than represented by centrifugation speeds of 500 - 650 RPM were existent in the desaturated samples direct after water addi-tion. With the exception of RS 2 of peat 2 at 30°C, the longi-tudinal and the transverse surface relaxivities of the two peat samples did not change significantly during hydration time. Surface relaxivity changes may be only to some extent re-sponsible for the observed changes of the T1 distributions, but not for those of the T2 distributions in the course of hy-dration of the two peats soil samples.

Calculation of Time Constants and of Apparent Activation Energy

Fig. (5) shows the changes of relative amplitudes of the four T2 ranges (I: 0.1 - 3 ms; II: 3 - 30 ms; III: 30 - 300 ms and IV: > 300 ms) in the T2 distribution of water in the two rewetted peat soil samples at 30°C (Fig. 5A+C) and 5°C (Fig. 5B+D) as a function of hydration time. Within the first hours after water addition, the relative amplitudes for T2 > 300 ms decreased very fast followed by an ongoing slower decrease, which lasted for several months. This slow de-crease was not fully completed until the end of the available experiment time of 5 to 7 months. Parallel to this decrease the relative amplitudes of the T2 ranges II and III increased

during the course of hydration. For peat 1 mainly the ampli-tudes of the T2 range III (30 - 300 ms) increased. For peat 2 the amplitudes of the T2 range II (3 - 30 ms) increased. Simi-lar to the decrease of the relative amplitudes for T2 > 300 ms, the increases of amplitudes at smaller T2 were very fast in the beginning and much slower after a few days lasting until the end of the experiment.

From the contact angle experiment it can be concluded that the peat surfaces were fully wettable after water addi-tion, because the water drop needed for the contact angle measurement did not rest at the peat surface, but was sucked into the matrix immediately. The T1B and T2B bulk relaxation times in the extracted peat soil solutions measured at 19°C were with 2.0 ± 0.3 s comparable to that of pure water T1,2B = 2.3 s and did not change significantly during the course of hydration for any of the two peats. Thus, the time dependent changes of the relative amplitudes of the T2 ranges repre-sented mainly the intrusion of water into the peat matrix or water re-distribution, but not surface wettability changes or changes of the bulk relaxation times.

For both peats, three time constants in the range of min-utes (fast), hours (medium fast) and several weeks to months (slow) for the rate-limiting processes of water intrusion (de-cease of the relative amplitudes for T2 > 300 ms) were calcu-lated with Eq. 6 (Table 2). The time constants of the three processes were slightly larger for peat 1 and the relative amounts of re-distributed water smaller than for peat 2. In

Fig. (4). T1 and T2 relaxation times of water in the two desaturated peat samples as a function of centrifugation speed (the same samples as

presented in Figure 3 without sample of day 2). The results of the two repetition samples or replicates (RS1 and RS2) are displayed.

1000

10

100

1000peat 2 RS 1 RS 2start end

40002000

D

T2

/ms


10

100


40002000

C

T2

/ms


1000

10

100


40002000

A

T1

/ms

centrifugation speed /RPM500 1000

10

100


40002000

B

T1

/ms



total, about 20 % (peat 1) and 30 - 40 % (peat 2) of the total water in the peat samples took part in the water intrusion. All time constants decreased with increasing temperature, which allowed the calculation of apparent activation energies, EA, of the three rate-limiting processes of water intrusion using Eq. 7. The activation energies (Table 2) were comparable for the two peat samples and ranged between 5 kJ mol

-1 (fast

process), 15 - 25 kJ mol-1

(medium fast process) and 40 - 50 kJ mol

-1 (slow process). Thus, EA of the medium fast process

is comparable to the value determine for pure water (see above), whereas it is smaller for the fast, but larger for the slow process.

The amounts of re-distributed water during hydration of the peats were comparable to those found in mineral soil samples [13]. The water re-distribution was accompanied by a reduction of T2 for 20 - 40 % of the total water, indicating a decrease in mobility of the involved water molecules during hydration [52]. The fast process of water re-distribution was not observed in soil samples until now, which may be due to the faster water addition of two instead of 15 minutes com-pared to similar studies [12], the faster measurement of T2 compared to T1 [2] and the better time resolution within the first minutes after water addition with one T2 measurement every five minutes compared to one measurement per day [2, 12, 13] or every 30 minutes [6]. The medium fast process

was also observed in other studies [6, 12, 13]. The time con-stant of the slow process is up to 10 to 15 times larger than reported elsewhere [2, 6, 12, 13]. The apparent activation energies of the three processes are in the lower range as re-ported by Todoruk et al. [6] or even smaller. As they [6] found a negative relation between SOM content and the value of the apparent activation energy, this may be due to the high organic matter contents of about 99 % of the two peats in this study. However, it cannot be excluded that dis-solution of paramagnetic substances may have additionally increased the values of the apparent activation energy in their study, because the contribution of the soil solution to the proton relaxation was not considered. It was shown in a pre-vious study [19] that the concentration of iron and manga-nese in the solution of peat is generally smaller than in the solution of mineral soil samples and that the iron and man-ganese concentration in soil solution may increase during hydration.

The apparent activation energies calculated for the fast

process are comparable to the energy of 6.3 kJ mol-1

required

to just break the hydrogen bond in a locally symmetric,

strongly H-bonded domain in water [53] leaving the mole-

cules essentially in the same position. EA of the medium fast

process is comparable to the value reported for water self-

diffusion [25], and indicates diffusion processes of water.

Fig. (5). Changes of the relative amplitudes of the four T2 ranges (I: 0.1 - 3 ms; II: 3 - 30 ms; III: 30 - 300 ms and IV: > 300 ms) in the T2

distribution of water in the two rewetted peat soil samples at 30°C (Fig. 5A+C) and 5°C (Fig. 5B+D) as a function of hydration time. The

results of the two repetition samples or replicates (RS1 and RS2) are displayed.

0 30 60 90 120 150 180

0

20

40

60

A

RS 1 RS 2 peat 1 0.1-3 ms 3-30 ms 30-300 ms >300 ms

rel.

amp

litu

de

(T2)

/%

time /days

0 30 60 90 120 150 180

0

20

40

60

C


rel.

amp

litu

de

(T2)

/%

time /days0 30 60 90 120 150 180

0

20

40

60

D


rel.

amp

litu

de

(T2)

/%

time /days

0 30 60 90 120 150 180

0

20

40

60

B


rel.

amp

litu

de

(T2)

/%

time /days

30°C 5°C


The larger activation energies of the slow process suggest

physical or physicochemical controlled processes, such as

water diffusion or reorientation of SOM chains during hydra-tion [6, 38, 49].

T1 - T2 and T2 - D Two-Dimensional NMR Measurements

The results of the T1 - T2 and T2 - D two-dimensional NMR measurements of the two peats at the beginning and end of hydration time at 30 MHz and 20°C are shown in Fig. (6). Direct after water addition, most of the water protons relaxed with T1 and T2 relaxation times comparable to pure water, which is represented by the high intensity areas close to the 1:1 line (white line) in Fig. (6A) for peat 1 and Fig. (6D) for peat 2. However, some water protons at shorter re-laxation times relaxed at smaller T2 than T1 values. This is represented by the deviation of the intensity areas from the 1:1 line. This deviation was more pronounced at the end of the hydration time (Fig. 6B+E), where the amount of water protons relaxing at smaller relaxation times was increased (Figs. 1, 2 and 5). Whereas the deviation from the 1:1 line appeared to be continuously for peat 1 (Fig. 6B), it was more step-wise for peat 2 (Fig. 6E). As a result of this deviation, the T2 values were about ten times smaller than the T1 values at small relaxation times. Furthermore, both figures show that each T1 is related to one T2 value and vice versa. Thus, each T1 population was related to one T2 population in the T1 and T2 distributions in Figs. (1 and 2), which allows a quanti-tative comparison of the T1/T2 ratios of water in the peat samples used in the swelling kinetics experiment (see next section).

The T2 - D correlation spectra of water in the two peat samples (RS 1 only) at the end of hydration at 20°C are shown in Fig. (6C) (peat 1) and Fig. (6F) (peat 2). For larger relaxation times (T2 > 30 ms), the apparent diffusion coeffi-cient of water in the two peat samples was comparable to the one of free water with Dapp = 2.03 x 10

-9 m

2 s

-1 [54]. For

smaller relaxation times (T2 < 30 ms), the apparent diffusion coefficient of water was reduced to Dapp = 1.2 x 10

-9 m

2 s

-1

for peat 1 and Dapp = 0.7 x 10-9

m2 s

-1 for peat 2. Thus, the

translational mobility of 22 % (peat 1) and 35 % (peat 2) of the total water was reduced compared to that of free water. At the beginning of hydration (data not shown), it was found that the translational mobility of only 16 % (peat 1) and 12 % (peat 2) of the total water was reduced. This indicates that the translational mobility of 6 % and 23 % of the total water

in peat 1 and peat 2 was reduced during hydration, respec-tively. The reduced translational water mobility in the peats may be due to the larger surface to volume ratios of smaller pores [55] or due to water entrapment inside a gel phase, where the water mobility is affected by the polymer chains, as shown for e.g. agar gels [56].

The T1 - T2 correlation spectra of the two peats were very different from those of sedimentary rock samples [42]. The main difference is the stronger deviation of the high intensity area from the 1:1 line for the peats. To compare this phe-nomenon found in peat with other materials that are able to swell and form gel phases, we studied additionally different reference materials like agar, starch (1 - 2 mm sago beads) and allophane gel (1.25 SiO2 * Al2O3 * 3.2 H2O). For agar gel (c = 30 g L

-1), the T1 - T2 correlation spectrum showed

one high intensity area in the shape of a circle at T1 = 2000 ms and T2 = 41 ms (data not shown). The resulting T1/T2 ratio was about 50 at 30 MHz. Fig. (7A) shows that the T1/T2 ratios of water in agar gels at 7.5 MHz strongly increase as a function of the agar concentration. The T1/T2 ratio in the agar gel with c = 30 g L

-1 was with 27 smaller than at 30 MHz,

because T1 was with 1240 ms much smaller than at 30 MHz (T1 = 2000 ms). T2 was with 45 ms at 7.5 MHz comparable to the value at higher field strength (T2 = 41 ms). For inor-ganic allophane gel (wc = 5.5 ± 0.5 g g

-1), the T1 - T2 correla-

tion spectrum was qualitatively comparable to the one of the agar gel, but with smaller relaxation time values of T1 = 330 ms and T2 = 10 ms (data not shown). The calculated T1/T2 ratio was 33 at 30 MHz and 25 at 7.5 MHz, because T1 was smaller at 7.5 MHz and T2 similar to that at higher field strength.

Fig. (7) also represents the T1 - T2 (Fig. 7B) and the T2 - D (Fig. 7C) correlation spectrum of water in swollen sago beads. Sago is a starch extracted from the pith of sago palm stems (Metroxylon sagu) and is commonly used in the food industry. In contrast to agar and allophane gel, the T1 - T2 spectrum of swollen sago beads consist of a broad distribu-tion, which is more similar to the one determined for peat 1 at the end of hydration. However, the deviation of the high intensity area from the 1:1 line is stronger and the calculated maximal T1/T2 ratios are with 30 - 40 comparable to those of the agar and allophane gels. The T2 - D spectrum shows that for some water at smaller T2 the water mobility was with Dapp = (0.5 - 1.8) x 10

-9 m

2 s

-1 smaller than for free pure wa-

Table 2. Results of the Peat Swelling Kinetics (i.e. Decrease of the Relative Amplitude of T2 > 300 ms) and Apparent Activation

Energies (EA) of the Three Determined Processes Governing the Swelling of peat. SE Standard Error

Sample Process Time Constant

(30°C to 5°C)

rel. Amount of Total

Water/%

EA ± SE/kJ mol-1

fast 25-35 min 5-7 6 ± 3

medium fast 20-50 h ~5 24 ± 4

peat 1

slow 55-330 d 7-9 49 ± 7

fast 15-20 min 10-17 4 ± 2

medium fast 20-30 h 7-9 16 ± 6

peat 2

slow 45-210 d 11-16 42 ± 5


ter, which is comparable to peat 2 at the end of hydration. Thus, the two-dimensional correlation spectra of the two peats were qualitatively comparable to those of the swollen sago beads suggesting similar properties in terms of proton relaxation and water mobility.

For all peat samples and reference materials, T2 was de-termined as a function of echo time to estimate the contribu-tion of T2D to the total proton relaxation. The T2 distribution did not change significantly with increasing echo times for all samples. Furthermore, T2D was in the range of seconds. Thus, the contribution of T2D to the total proton relaxation was negligible for all studied peat and reference samples. Consequently, the calculated large T1/T2 ratios of water in these samples were not due to diffusion in internal magnetic field gradients, which had to be very small in all studied peat and reference samples.

The findings for water in agar, allophane gels and swol-len sago suggest longer correlation times for the dipole interactions of the bound water protons compared to free water [25] due to water structuring [17, 23]. Thus, swelling of these materials reduced both the rotational and translational water mobility compared to free water, which is

water mobility compared to free water, which is consistent with the findings for wheat starch pastes [23].

Water State Characterization by 1H NMR Relaxometry

and DSC

T1/T2 Ratios as a Measure of the Water State in Peat

Samples

Fig. (8) shows the T1/T2 ratios of the rewetted peat soil samples at the beginning (19°C) and end of hydration time (swollen at 30°C, ratios determined at 19°C) as a function of the relative water content (rel. wc). The rewetted peat sam-ples were measured in a fully water saturated (black) and in a step-wise desaturated state (grey) after centrifugation (i.e. the original peat samples from the swelling kinetics experiment at 30°C shown in Fig. 4). The calculation of the T1/T2 ratio started from 7.5 % rel. wc instead of 0 %, because the num-ber and position of the peaks at very small T1 values (Fig. 1A+B, T1 < 20 ms; Fig. 2A+B, T1 < 2 ms) were not repro-ducible for the repetition samples.

The T1/T2 ratios calculated from the step-wise desatu-rated peat samples were comparable to those from the relaxa-

Fig. (6). Two-dimensional T1 - T2 - and T2 - D - correlation spectra of water in peat 1 (A-C) and peat 2 (D-F) at the beginning (A+D) and

end of hydration time (B+E and C+F) at 30 MHz determined at 20°C.

Fig. (7). T1/T2-ratio of water in agar gels as a function of the agar concentration at 7.5 MHz (A) and two-dimensional T1 - T2 - (B) and T2 -

D - correlation spectra of water swollen starchy sago beads (C) at 30 MHz determined at 20°C.

A B C

D E F

A B C

D E F

A B C

D E F

0

10

20

30

0 10 20 30c (Agar) /g L-1

T1/T

2

A B C

0

10

20

30

0 10 20 30c (Agar) /g L-1

T1/T

2

A B C

0

10

20

30

0 10 20 30c (Agar) /g L-1

T1/T

2

A B C


tion time distributions of the fully water saturated peat sam-ples at the beginning and end of hydration (Fig. 8). The only exception was again RS 2 of peat 2 where the T1/T2 ratios of the desaturated sample were smaller than those of the fully saturated one. This was maybe due to the decanting of the peat samples into the centrifugation tube and the resulting air contact and re-packing of the sample. The T1/T2 ratios of water in the peat samples increased with decreasing rel. wc from 1.1 (at rel. wc = 100 %) to 8 - 9 (peat 1) and 6 - 18 (peat 2) at rel. wc of 10 - 20 %. The amount of water with T1/T2 ratios larger 3 increased from 40 % at the beginning to about 60 % (or 80 % for RS 2 of peat 2) at the end of hydra-tion. The large T1/T2 ratios of RS 2 than RS 1 of peat 2 were mainly due to the different T1 distribution (see Fig. 2A).

It was found for both peat samples that at the beginning

(not shown) and end (Fig. 9) of hydration the amount of wa-

ter with T1/T2 ratios larger 3 increased with increasing tem-

perature. Furthermore, the T1/T2 ratios between 30 % and 80

% rel. wc were always larger at 30°C than at 5°C (Fig. 9).

Eq. 7 was applied to calculate the activation energies, EA, of

the temperature dependent increase of the observed T1/T2

ratio increase by using T1/T2 = k . The calculated EA at the

beginning and end of hydration was 3 - 12 kJ mol-1

for both

peat samples. It, therefore, can be suggested that the tem-

perature dependency of the T1/T2 ratios of water in the peat

samples can be attributed mainly to increasing water mobil-

ity (diffusion and rotation) at higher temperatures. Conse-

quently, at least three types of water can be distinguished by

the different T1/T2 ratios. Type 1 is represented by small re-

laxation times and large T1/T2 ratios (> 6 at rel. wc smaller

30-40 % in Fig. 9). Type 3 is represented by long relaxation

times and small T1/T2 ratios between 1.0 and 1.5. Type 2 can

be considered as the transition from type 1 to type 3 and its

amount of water is depending on the mobility of water (see

Fig. 9). We suggest that water type 1 represents bound or

structured water, type 2 exchange water and type 3 bulk-like

free water.

Some of the water in the peat samples was found to be similar to free pure water over the whole period of hydration

(T1/T2 ratio ~ 1). Another part of the water changed its state after addition to the peat sample as represented by larger T1/T2 ratios, which were much larger than generally found in rock core samples [24]. These large T1/T2 ratios of water in peat were in the same range as determined for agar gel or swollen sago beads (Fig. 7). This suggests that this water was structured by the peat in a comparable way as reported for water in starch [17]. Consequently, the large T1/T2 ratios in the peat samples may be due to a water structuring in the vicinity of the peat surfaces and the resulting longer rota-tional correlation times of the water molecules within this water phase [17]. The amount of structured water increased during hydration, because the amount of water with T1/T2 ratios larger 3 increased by a factor of 1.5. This indicates that about 20 % of the total water significantly changed its state during the swelling, which suggests that the amount of water within a gel phase in the peats increased.

Cryo-NMR Relaxometry: Determination of Non-Freezable and Loosely Bound Water

The amounts of non-freezable water at -34°C of the two peat samples determined by Cryo-NMR ranged between 0.43 g g

-1 d.m. and 0.55 g g

-1 d.m. and increased by 5 % (peat 1)

and 20 % (peat 2) during hydration (Table 3). From the freezing behaviour determined by DSC, it can be concluded that this water was not frozen until -90°C (data not shown). The amounts of loosely bound water at -5°C were about 5 to 10 times smaller than those of the non-freezable water at -34°C and increased only for peat 2 during hydration (Table 3). However, the amount of loosely bound water in peat 2 was 20 % to 50 % smaller than for peat 1.

The relaxation time of water in the air dried peat samples measured at 19°C, where only non-freezable water existed, was 293 ± 3 s for peat 1 and 267 ± 3 s for peat 2. The T2 values determined by Cryo-NMR represent the relaxation times of non-freezable water and of loosely bound water in the frozen peat samples at the maximal water holding capac-ity. Table 3 shows that the T2 values for each peat increased in the following order: non-freezable water in air dried peat, non-freezable water at -34°C and loosely bound water in the

Fig. (8). T1/T2 ratios of the rewetted peat soil samples at the beginning (start) and end of hydration time as a function of the relative water

content (rel. wc) determined from the relaxation time distributions at 19°C. The T1/T2 ratios at the end of hydration were calculated from the

30°C samples measured at 19°C. The rewetted peat samples were measured in a fully water saturated and step-wise desaturated (centrifuga-

tion) state. Relative water contents were calculated with Eq. 4 (saturated samples) and Eq. 5 (desatureated samples). The results of the two

repetition samples or replicates (RS1 and RS2) are displayed.

0 20 40 60 80 100

3

6

9

12

15

18 peat 2 RS 1 RS 2fully saturatedstart end desaturated/centrifugestart end

B

T1/

T2

rel. wc /%0 20 40 60 80 100

3

6

9

12

15

18 peat 1 RS 1 RS 2fully saturatedstart end desaturated/centrifugestart end

A

T1/

T2

rel. wc /%


frozen peat samples. Furthermore, additionally to the in-creasing amounts of non-freezable water at -34°C and of loosely bound water the T2 values of both water types in-creased during hydration.

The amounts of non-freezable water were comparable to those reported for different starch materials [15] or peats [7], but were larger than those reported by Schaumann et al. [14]. Increasing amounts of non-freezable water during hydration were also found for starch gels [52]. The relaxation times of the two types of bound water were about one order of magni-tude larger than determined for peat at 300 MHz [7]. The differences, therefore, may be due to the 10 times higher magnetic field strengths compared to this study.

DSC: Determination of Different Water States and of

Melting Enthalpy of Freezable Water in Peat

The DSC thermograms in Fig. (10) show the endothermic melting peaks of free pure water and soil solutions of the two peat samples extracted at the end of hydration (Fig. 10A+B). Furthermore, they show the melting peaks of the freezable water in the fully saturated (Fig. 10C+D) and in the desatu-rated peat samples after centrifugation at 4000 RPM (Fig.

10E+F) at the beginning and end of hydration time. The number of melting peaks of water in the fully saturated peat samples increased from one at the beginning (broad peak around +2°C in Fig. 10C+D) to two peaks at the end of hy-dration (one sharp peak around 0°C and one broad peak be-tween +2°C and +5°C). For the desaturated samples, a clear splitting of the melting peak into one sharp and one broad peak was observable at the beginning as well as at the end of hydration (Fig. 10E+F). In contrast to the fully saturated peat sample, the position of the sharp peak was around -1°C and the width of the broad peak was significantly smaller. The onset of the melting peak was between -5°C and -10°C for both peats. The DSC thermograms of free pure water and soil solutions consisted of only one melting peak with a very sharp onset at -3°C to -1°C and a maximum between +1°C and +2°C. For agar gel (30 g L

-1), the thermogram consisted

of a sharp peak at -0.3°C and a broad peak at 5°C (not shown).

Fig. (11) shows that the peak temperature in the desatu-rated peat samples after centrifugation at 4000 RPM at the end of hydration time (Fig. 10E+F), agar gel, water saturated sand 150 m and free pure water increased with increasing

Fig. (9). T1/T2 ratios of the rewetted peat soil samples at the end of hydration time as a function of the relative water content (rel. wc) deter-

mined from the relaxation time distributions of water in the fully saturated peat samples at 5°C and 30°C. The results of the two repetition

samples or replicates (RS1 and RS2) are displayed.

Table 3. Results of the Cryo-NMR and DSC Measurements of the Two Peats at the Beginning and End of Hydration Time. SD

Standard Deviation, SE Standard Error, n.s. not Significant. T2 of Water in the Air Dried Samples was 293 ± 3 s for

Peat 1 and 267 ± 3 s for Peat 2. No Enthalpic Peak was Observed in the DSC Thermogram for Both Air Dried Peat

Samples

Cryo-NMR DSC

Non-Freezable Water

at -34°C

Loosely Bound Water

at -5°C

Non-Freezable

Water

Melting

Enthalpy of

Water in Peat

Enthalpy

Sharp Peak

Water

Content

Sharp Peak

Sample

wc ± SD

/g g-1

d.m.

T2 ± SD

/ s

wc ± SD

/g g-1

d.m.

T2 ± SD

/ s

wc ± SE

/g g-1

d.m.

Hf ± SD

/J g-1

Hf ± SD

/J g-1

d.m.

wc ± SD

/g g-1

d.m.

start 0.524 ± 0.001 435 ± 2 0.110 ± 0.001 925 ± 5 0.85 ± 0.11 366 ± 8 199 ± 38 0.54 ± 0.10 peat 1

end 0.553 ± 0.005 469 ± 5 0.106 ± 0.001 973 ± 26 0.77 ± 0.05 347 ± 7 265 ± 42 0.71 ± 0.11

start 0.428 ± 0.007 402 ± 3 0.058 ± 0.001 794 ± 6 0.54 ± 0.24 329 ± 44 95 ± 11 0.28 ± 0.03 peat 2

end 0.522 ± 0.012 568 ± 83 0.083 ± 0.001 1047 ± 92 0.60 ± 0.08 326 ± 9 158 ± 26 0.49 ± 0.06

0 20 40 60 80 100

3

6

9

12

15

18 peat 2 RS 1 RS 2end05°C 30°C

B

T1/

T2

rel. wc /%0 20 40 60 80 100

3

6

9

12

15

18 peat 1 RS 1 RS 2end05°C 30°C

A

T1/

T2

rel. wc /%


heating rate. Thus, the melting of freezable water was kineti-cally controlled in all studied samples. However, the heating rate dependence was pronounced stronger for the broad peak than for the sharp peak. Furthermore, the heating rate de-pendence of the broad peak of the two peats was less pro-nounced than for agar gel, but stronger than for free pure water. For peat 1, it was comparable to sand 150 m.

The amounts of non-freezing water determined from

DSC were ~0.8 g g-1

for peat 1 and ~0.6 g g-1

for peat 2 and

were comparable for the beginning and the end of hydration

(Table 3). The melting enthalpies of freezable water, H f , in

the peats (Table 3) were in the same range as those of free

pure water and of soil solution 338 ± 8 J g-1

and decreased

slightly only for peat 1 during hydration. To test whether the

amount of water, represented by the sharp peak in the DSC

thermogram, increased during hydration, the enthalpy of the

sharp peak, H f , (integral between peak onset at low tem-

perature and minimum between the sharp and broad peak

using the same baseline as for the total peak) was calculated

for the peat sample after centrifugation at 4000 RPM. This

made the observation of the sharp peak at the end of hydra-

tion possible for the full saturated peat samples. Assuming a

melting enthalpy independent of the water binding state, we

estimated the water content from H f / H f (H f per gram of

dry peat), of the sharp peak for the beginning and end of

hydration. Table 3 shows that this calculated water content,

represented by the sharp peak, increased for both peat sam-

ples during hydration. However, this increase was significant

only for peat 2, where the water content increased by about

Fig. (10). DSC thermograms of free pure water and soil solutions of the two peat samples extracted at the end of hydration (A+B) and of the

freezable water in the fully saturated (Fig. 10C+D) and in the desaturated peat samples after centrifugation at 4000 RPM (Fig. 10E+F) at the

beginning and end of hydration time. The results of two of the three repetition samples used in the DSC are displayed as black and grey col-

oured curves.


60 %. This water accounts for 3 - 5 % of the total water in

peat 1 and 6 - 11 % in peat 2.

A melting point depression was found for some water in the peats, which was not caused by dissolved salts in the soil solution. The melting characteristic of the water was strongly affected by the peat matrix and resulted in a splitting into (at least) two to three different types of water represented by the broad onset at low temperatures and the two peaks in the DSC thermograms, respectively. The broad shape of the on-set may be due to the binding forces for water molecules within this boundary phase, which change strongly with in-creasing distance from the peat surfaces. At a certain dis-tance from the surfaces and outside this boundary phase, the water molecules are still affected by the surfaces but in a more uniform way. These uniform conditions resulted in the formation of the sharp melting peak between -2°C and 0°C. We thus assume that the sharp peak mainly represents loosely bound water in peat. The shape of the broad melting peak of water in the fully saturated peat samples is different than for free pure water or for soil solution. However, the temperature range and maxima are comparable to water and soil solution where no loosely bound water exists (Fig. 10). Hence, the broad melting peak may represent freezable bulk water, but additionally water that was still affected by the peat matrix.

According to the findings for maltodextrin gels [15], it can be suggested that the melting peak splitting of water in peat was due to the entrapment of water in a gel phase. By interpreting the energy transformation related with the sharp peak as quantitative measure for the amount of water in gels, the water contents related to the sharp peak at the beginning and end of hydration (Table 3) indicated the formation of new and/or the change of existing gel phases during hydra-tion. This would then indicate that the amount of water in gels increased during hydration and that some water was entrapped in gel phases shortly after water addition. This suggests an initial formation of gel phases in peat after sur-face wetting.

A splitting of the melting peak of water in peat in the DSC thermograms was also found by McBrierty et al. [7] and by Schaumann [14] for peats at water contents below the maximal water holding capacity. For peats with relatively

high water contents, no splitting of the melting peak was observed [7]. The melting enthalpies of freezable water and the amounts of non-freezable water reported by McBrierty et al. [7] and Schaumann [14] were smaller than the values calculated in this study. The peak maxima are located at lower remperatures than reported by McBrierty et al. [7], but at higher temperatures than those reported by Schaumann [14]. In the first study, both peak maxima were detected above 0°C, whereas in the latter both peak maxima were found below 0°C. This may be due to the higher heating rates compared to our study or because different peat sam-ples were used. Schaumann [14] found a heating rate de-pendence only for the broad peak and concluded that the broad peak represents loosely bound water and the sharp peak bulk-like water, which is contrary to the conclusions of this study. These differences may be due to the much higher degradation state and the lower water contents of 0.4 - 0.6 g g

-1 of the peats used by Schaumann [14] and due to the dif-

ferent experimental setups (hydration from the gas phase, higher heating rates). For gelatine gels, Liu and Yao [57] attributed the sharp peak to loosely bound water (referred as “intermediate” water) and the broad peak to free water.

Comparison of the Results Determined by 1H NMR and

Cryo NMR Relaxometry and DSC

The amounts of non-freezable water determined by DSC were larger than those from Cryo-NMR, but were in the range of the sum of non-freezable and loosely bound water determined by Cryo-NMR (Table 3). 3 - 5 % of the total water and 10 - 13 % were determined as non-freezable water in peat 1 and peat 2, respectively. The value for peat 2 agrees roughly with the beginning of the plateau at large T1/T2 ra-tios observed in Fig. (8B). For peat 1, no T1/T2 ratio was calculated at this low relative water content (see above). The sum of non-freezable water and water represented by the sharp peak in the DSC thermogram accounts for 8 - 10 % of the total water in peat 1. This value agrees with the largest T1/T2 ratio in Fig. (8A). In peat 2, this sum of the two water types accounts for 17 - 25 % of the total water, which agrees with the end of the plateau with large T1/T2 ratios in Fig. (8B). This suggests that the large T1/T2 ratios of water in the peat samples were related to non-freezable water and water represented by the sharp peak in the DSC thermograms (loosely bound water and water in gels). Thus, the observed

Fig. (11). Maxima of the DSC melting peaks of the desaturated peat samples after centrifugation at 4000 RPM at the end of hydration time

(Fig. 10E+F), agar gel (30 g L-1

), water saturated sand (particle size 150 m) and free pure water as a function of heating rate.

0 2 4 6 8 10

-1

0

1

2

3

6

9 broad peak maximum peat 1 peat 2 agar gel sand 150 µm water

B

tem

pera

ture

/°C

heating rate /K min-10 2 4 6 8 10

-1

0

1

2

3

6

9 A

sharp peak maximum peat 1 peat 2 agar gel

tem

per

atu

re /°

C

heating rate /K min-1


changes in the T1/T2 ratios during hydration may be due to the increasing of amounts of these two water types.

From the water state characterization by 1H NMR and

Cryo NMR relaxometry and DSC we conclude that a layered bound water phase existed inside the peats. This phase con-sisting of several layers of non-freezable and loosely bound water, whereas the binding forces and, consequently, the water structuring decreases with increasing distance from the peat surfaces resulting in higher mobility of the water mole-cules. This layered bound water phase includes water en-trapped in gel phases. During swelling, the amounts of bound water as well as the mobility of the water molecules inside the bound water phase increased, which indicates a water intrusion into the peat matrix as well as a reorientation of SOM chains into the pore space. This is consistent with the conclusions drawn for mineral soil samples [6] and peat samples [8]. Mikutta et al. [58] studied the hydration of po-lygalacturonate (PGA) coatings on alumina (Al2O3) and con-cluded from their NMR and DSC results that the PGA chains reorient into the pore space upon swelling.

Microbial Activity

Throughout the 21 days incubation the microbial activity in the investigated samples at 60 % of the maximal water holding capacity was constant and extremely low. For peat 1 it varied from 0.06 to 0.07 mg CO2 h

-1 100 g

-1 and for peat 2

from 0.08 to 0.1 mg CO2 h-1

100 g-1

of soil. Additional repli-cates of the peat samples that were fully saturated showed almost no release of CO2 during incubation. The values of the CO2 release from the two peats were about 100 to 1000 times smaller than reported for two mineral soil samples, where relaxation time distribution changes were significantly stronger in samples with high microbial respiratory activity [13]. Consequently, the microbial effects on the changes of relaxation time distribution of water in the two peat samples during hydration may be considered negligible.

Synthesis

The changes of the relaxation time distributions of water in the two peat samples as well as the increase of the Q val-ues during hydration were most probably not caused by in-fluences from the soil solution, internal magnetic field gradi-ents, surface relaxivity changes or soil microorganisms and, therefore, indicate swelling of the two studied peat samples [1, 2, 6, 12, 13]. The swelling of peat was characterised by redistribution of water, by increasing amounts of non-freezable and loosely bound water, by the formation of gel phases as well as by the reduction of the translational and rotational mobility of water molecules in the two peat sam-ples. Furthermore, swelling induced strong changes of the pore size distributions, which resulted in the reduction of number of large pores (> 50 m) and formation of medium-sized pores (50 - 10 m). Some time after two days of hydra-tion the formation of small-sized pores (< 10 m) was also observed.

It was found that the stronger volumetric swelling of peat 2 was linked to stronger changes of the pore size distribution and higher amounts of redistributed water as well as to stronger relative increases of the amounts of non-freezable and loosely bound water. In addition, more water was af-fected by the reduction of the translational mobility of water

molecules. We suggest that this was due to the higher degra-dation state and the more heterogeneous matrix of peat 2. It is very likely that air drying of peat 2, previous to the re-hydration, caused stronger matrix changes, such as closing of small-sized pores and collapsing of gel phases [6, 59].

The physical property of the volumetric swelling was connected to various changes of physical and physicochemi-cal properties of peat during hydration. Stronger volumetric swelling was accompanied by stronger changes of physical and physicochemical properties of peat. The peat swelling was governed by three processes with time constants in the range of minutes (fast process), hours (medium fast process) and weeks/months (slow process) with related apparent acti-vation energies of 5 - 50 kJ mol

-1, indicating the breaking of

hydrogen bonds [53], water diffusion and reorientation of SOM chains during hydration [6, 38, 49].

CONCLUSIONS

The increasing amounts of non-freezable and loosely bound water as well as the reduction of the translational and rotational mobility of water molecules in the two peat sam-ples indicate changes of the physical and physicochemical properties of the peats during swelling at constant tempera-ture and moisture conditions. These changes took place over three time periods between minutes and months and were governed by physical and physicochemical processes. From the results we derived a mechanistic model to describe the fundamental processes of peat swelling. This model is based on results obtained from air dried peat samples with low wa-ter contents. This water exists as non-freezable water in peat. Within the first minutes after wetting of the initially water-accessible peat surfaces water structuring and water reorien-tation in the vicinity of the surfaces took place. This resulted in the increase of the existing non-freezable water phase due to the addition of new layers of structured water. Parallel to this process the formation of loosely bound water and gel phases occurred. Within the next following 20 to 50 hours, the swelling was mainly controlled by the diffusion of water into the peat matrix, which results in volumetric swelling and the formation of medium-sized pores (> 10 m). After this, a very slow reorientation of SOM chains together with an on-going volumetric swelling took place, which lasted for sev-eral months. During this time formation of medium-sized pores (> 10 m) as well as small-sized pores (< 10 m) oc-curred. Parallel to the processes which were observed during the last two time periods, intra-particulate surfaces became water-accessible which in turn resulted in the formation of non-freezable, loosely bound and gel water. The findings of this study are of environmental importance for helping to optimise renaturation and rewatering of commercially used peatlands and to better understand sorption/desorption and transport processes of pollutants and nutrients in natural or-ganic matter rich soils.

ACKNOWLEDGEMENTS

This study was funded by the German Research Founda-tion, DFG, (SCH849/5-3). The T1/T2 - and T2/Diffusion-correlation measurements were supported by the European Community activity Large-Scale Facility Wageningen NMR Center (FP6-2004-026164, 2006-2009). We like to thank Dr. Song from Schlumberger for the ILT programme. A special


“thank you” goes to Julia Bayer (University Koblenz-Landau) for her help in improving the manuscript.

REFERENCES

[1] Lyon WG, Rhodes DE. The swelling properties of soil organic matter and their relation to sorption of non-ionic organic com-

pounds. cincinnati: environmental protection agency (EPA); 1991. Report No.: EPA/600/S2-91/033.

[2] Schaumann GE, Hurraß J, Müller M, Rotard W. Swelling of or-ganic matter in soil and peat samples: insights from proton relaxa-

tion, water absorption and PAH extraction. In: Ghabbour EA, Da-vies G, Eds. Humic substances: nature's most versatile materials.

New York: Taylor and Francis, Inc. 2004: pp. 101-17. [3] Gaillardon P. Influence of soil moisture on long-term sorption of

Diuron and Isoproturon by soil. Pestic Sci 1996; 47(4): 347-54. [4] Johnson AC, Bettinson RJ, Williams RJ. Differentiating between

physical and chemical constraints on pesticide and water movement into and out of soil aggregates. Pestic Sci 1999; 55(5): 524-30.

[5] Altfelder S, Streck T, Richter J. Effect of air-drying on sorption kinetics of the herbicide chlortoluron in soil. J Environ Qual 1999;

28: 1154-61. [6] Todoruk TR, Langford CH, Kantzas A. Pore-scale redistribution of

water during wetting of air-dried soils as studied by low-field NMR relaxometry. Environ Sci Technol 2003; 37(12): 2707-13.

[7] McBrierty VJ, Wardell GE, Keely CM, O'Neill EP, Prasad M. The characterization of water in peat. Soil Sci Soc Am J 1996; 60(4):

991-1000. [8] McBrierty VJ, Martin SJ, Karasz FE. Understanding hydrated

polymers: the perspective of NMR. J Mol Liquid 1999; 80(2,3): 179-205.

[9] Lyon WG. Swelling of peats in liquid methyl, tetramethylene and propyl sulfoxides and in liquid propyl sulfone. Environ Toxicol

Chem 1995; 14(2): 229-36. [10] Ping ZH, Nguyen QT, Chen SM, Zhou JQ, Ding YD. States of

water in different hydrophilic polymers - DSC and FTIR studies. Polymer 2001; 42(20): 8461-7.

[11] Jhon MS, Andrade JD. Water and hydrogels. J Biomed Mater Res 1973; 7(6): 509-22.

[12] Schaumann GE, Hobley E, Hurraß J, Rotard W. H-NMR Re-laxometry to monitor wetting and swelling kinetics in high organic

matter soils. Plant Soil 2005; 275(1-2): 1-20. [13] Jaeger F, Grohmann E, Schaumann GE. 1H NMR Relaxometry in

natural humous soil samples: Insights in microbial effects on re-laxation time distributions. Plant Soil 2006; 280(1-2): 209-22.

[14] Schaumann GE. Matrix relaxation and change of water state during hydration of peat. Colloids Surf A-Physicochem Eng Asp 2005;

265(1-3): 163-70. [15] Radosta S, Schierbaum F. Polymer-water interaction of maltodex-

trins. Part III: non-freezable water in maltodextrin solutions and gels. Starch - Stärke 1990; 42(4): 142-7.

[16] Yoshida H, Hatakeyama T, Hatakeyama H. Characterization of water in polysaccharide hydrogels by DSC. J Thermal Anal 1993;

40(2): 483-9. [17] Radosta S, Schierbaum F, Yuriev WP. Polymer-water interaction

of maltodextrins. Part II: NMR study of bound water in liquid maltodextrin-water systems. Starch - Stärke 1989; 41(11): 428-30.

[18] Bayer JV, Jaeger F, Schaumann GE. Proton Nuclear magnetic resonance (NMR) relaxometry in soil science applications. Open

Magn Reson J 2010 (accepted). [19] Jaeger F, Rudolph N, Lang F, Schaumann GE. Effects of soil solu-

tion’s constituents on proton NMR relaxometry of soil samples. Soil Sci Soc Am J 2008; 72(12): 1694-707.

[20] Keating K, Knight R. A laboratory study to determine the effect of iron oxides on proton NMR measurements. Geophysics 2007;

72(1): E27-32. [21] Fyfe CA, Blazek AI. Investigation of hydrogel formation from

hydroxypropylmethylcellulose (HPMC) by NMR spectroscopy and NMR imaging techniques. Macromolecules 1997; 30(20): 6230-7.

[22] Carenza M, Cojazzi G, Bracci B, et al. The state of water in ther-moresponsive poly(acryloyl-L-proline methyl ester) hydrogels ob-

served by DSC and 1H-NMR relaxometry. Radiat Phys Chem 1999; 55(2): 209-18.

[23] Callaghan PT, Jolley KW, Lelievre J, Wong RBK. Nuclear mag-netic resonance studies of wheat starch pastes. J Colloid Interface

Sci 1983; 92(2): 332-7.

[24] Kleinberg RL, Straley C, Kenyon WE, Akkurt R, Farooqui SA.

Nuclear magnetic resonance of rocks: T1 vs. T2. Society of Petro-leum Engineers 1993; 68th Annual Technical Conference and Exhi-

bition of the Society of Petroleum Engineers, 1993; pp. 553-63. Houston: Texas.

[25] Packer KJ, Dick DAT, Wilkie DR. The dynamics of water in het-erogeneous systems. Philos Trans R Soc Lond B Biol Sci 1977;

278(959): 59-87. [26] Pouliquen D, Gallois Y. Physicochemical properties of structured

water in human albumin and gammaglobulin solutions. Biochimie 2001; 83(9): 891-8.

[27] Pouliquen D, Omnes M-H, Seguin F, Gaignon J-L. Changes in the dynamics of structured water and metabolite contents in early de-

veloping stages of eggs of turbot (Psetta maxima). Comp Biochem Physiol A 1998; 120(4): 715-26.

[28] Dunn K-J, Bergman DJ, Latorraca GA. Handbook of geographical exploration-seismic exploration: nuclear magnetic resonance-

petrophysical and logging applications. 1st ed. Oxford: Pergamon 2002.

[29] Kleinberg RL. Nuclear magnetic resonance. In: Wong P-Z, Ed. Methods in the physics of porous media. San Diego: Academic

Press 1999; pp. 337-85. [30] Bird NRA, Preston AR, Randall EW, Whalley WR, Whitmore AP.

Measurement of the size distribution of water-filled pores at differ-ent matric potentials by stray field nuclear magnetic resonance. Eur

J Soil Sci 2005; 56(1): 135-43. [31] Brownstein KR, Tarr CE. Importance of classical diffusion in

NMR studies of water in biological cells. Phys Rev A 1979; 19(6): 2446-53.

[32] Godefroy S, Korb JP, Fleury M, Bryant RG. Surface nuclear mag-netic relaxation and dynamics of water and oil in macroporous me-

dia. Phys Rev E 2001; 64(2-1): 021605. [33] Kenyon WE, Kolleeny JA. NMR surface relaxivity of calcite with

adsorbed Mn2+. J Colloid Interface Sci 1995; 170(2): 502-14. [34] Bryar TR, Daughney CJ, Knight RJ. Paramagnetic effects of

iron(III) species on nuclear magnetic relaxation of fluid protons in porous media. J Magn Reson 2000; 142(1): 74-85.

[35] Meiboom S, Gill D. Modified spin-echo method for measuring nuclear relaxation times. Rev Sci Instrum 1958; 29: 688-91.

[36] Roberts SP, Macdonald PJ, Tritchard T. A bulk and spatioal re-solved NMR relaxation study of sandstone rock plugs. J Magn

Reson A 1995; 116: 189-95. [37] Jaeger F, Bowe S, Schaumann GE. Evaluation of 1H NMR re-

laxometry for the assessment of pore size distribution in soil sam-ples. Eur J Soil Sci 2009; 60(6): 1052-64.

[38] Diehl D, Schaumann GE. Wetting mechanism assessed from time dependent sessile drop shape. Hydrol Process 2007; 21(17): 2255 -

65. [39] Veevaete M. Applications of Earth’s Field NMR to porous systems

and polymer gels. Bremen, Germany: University of Bremen 2008. [40] Butler JP, Reeds JA, Dawson SV. Estimating solutions of first kind

integral equations with nonnegative constraints and optimal smoothing. SIAM J Numer Anal 1981; 18(3): 381-97.

[41] van Dusschoten D, Moonen CT, de Jager PA, Van As H. Unravel-ing diffusion constants in biological tissue by combining Carr-

Purcell-Meiboom-Gill imaging and pulsed field gradient NMR. Magn Reson Med 1996; 36(6): 907-13.

[42] Song YQ, Venkataramanan L, Hürlimann MD, Flaum M, Frulla P, Straley C. T1-T2 correlation spectra obtained using a fast two-

dimensional laplace inversion. J Magn Reson 2002; 154(2): 261-8. [43] Hurlimann MD, Venkataramanan L, Flaum C. The diffusion-spin

relaxation time distribution function as an experimental probe to characterize fluid mixtures in porous media. J Chem Phys 2002;

117(22): 10223-32. [44] Venkataramanan L, Song YQ, Hürlimann MD. Solving Fredholm

integrals of the first kind with tensor productstructure in 2 and 2.5 dimensions. IEEE Trans Signal Proces 2002; 50: 1017-26.

[45] Hahn EL. Spin echoes. Phys Rev 1950; 80(4): 580-94. [46] Valckenborg R. NMR on technological porous materials.

Eindhoven, The Netherlands: Eindhoven University of Technology 2001.

[47] Nordgren A. Apparatus for the continuous, long-term monitoring of soil respiration rate in large numbers of samples. Soil Biol Bio-

chem 1988; 20(6): 955-7. [48] Wedler G. Lehrbuch der physikalischen Chemie. 3rd ed. Weinheim:

VCH Verlagsgesellschaft mbH 1987.


[49] Sparks DL. Kinetics of ionic reactions in clay minerals and soils.

Adv Agron 1985; 38: 231-66. [50] Delville A, Letellier M. Structure and dynamics of simple liquids in

heterogeneous condition: an NMR study of the clay - water inter-face. Langmuir 1995; 11(4): 1361-7.

[51] Kinniburgh DG, Miles DL. Exraction and chemical analysis of interstitial water from soils and rocks. Environ Sci Technol 1983;

17: 362-8. [52] Wynne-Jones S, Blanshard JMV. Hydration studies of wheat

starch, amylopectin, amylose gels and bread by proton magnetic resonance. Carbohydr Polym 1986; 6(4): 289-306.

[53] Smith JD, Cappa CD, Wilson KR, Messer BM, Cohen RC, Saykally RJ. Energetics of hydrogen bond network rearrangements

in liquid water. Science 2004; 306(5697): 851-3. [54] Holz M, Heila SR, Saccob A. Temperature-dependent self-

diffusion coefficients of water and six selected molecular liquids for calibration in accurate 1H NMR PFG measurements. Phys

Chem Chem Phys 2000; 2(20): 4740-2.

[55] Mitra PP, Sen PN, Schwartz LM. Short-time behavior of the diffu-

sion coefficient as a geometrical probe of porous media. Phys Rev B 1993; 47(14): 8565-74.

[56] Beuling EE, Van Dusschoten D, Lens P, Van Den Heuvel JC, Van As H, Ottengraf SPP. Characterization of the diffusive properties of

biofilms using pulsed field gradient-nuclear magnetic resonance. Biotechnol Bioeng 1998; 60(3): 283-91.

[57] Liu WG, Yao KD. What causes the unfrozen water in polymers: hydrogen bonds between water and polymer chains? Polymer 2001;

42(8): 3943-7. [58] Mikutta C, Krüger J, Schaumann GE, Lang F. Restructuring of

polygalacturonate on alumina upon hydration - effect on phosphate sorption kinetics. Geochim Cosmochim Acta 2006; 70(12): 2957-

69. [59] Todoruk TR, Litvina M, Kantzas A, Langford CH. Low-field NMR

relaxometry: a study of interactions of water with water-repellant soils. Environ Sci Technol 2003; 37(12): 2878-82.

Received: July 26, 2009 Revised: November 30, 2009 Accepted: December 04, 2009

© Jaeger et al.; Licensee Bentham Open.

This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License

(http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided the work is properly cited.

46 The Open Magnetic Resonance Journal, 2010, 3, 46-51

1874-7698/10 2010 Bentham Open

Open Access

Investigation of Iron(III)-Release in the Pore Water of Natural Sands by NMR Relaxometry

Ivonne Mitreiter*,1,2

, Sascha E. Oswald2,3

and Frank Stallmach1

1University of Leipzig, Department of Interface Physics, Linnéstr. 5, 04103 Leipzig, Germany

2UFZ–Helmholtz Centre for Environmental Research, Department of Hydrogeology, Permoserstr. 15, 04318 Leipzig,

Germany

3University of Potsdam, Institute of Earth and Environmental Science, Karl-Liebknecht-Straße 24-25, 14476 Potsdam,

Germany

Abstract: Proton nuclear magnetic resonance (NMR) relaxometry offers a non-invasive and non-destructive measurement

method to observe and to visualise changes in the iron(III)-ion concentration in aqueous solutions. This is possible due to

its paramagnetic influence on the relaxation times. In the context of mineral dissolution processes in natural sediments, the

effect of the presence of dissolved iron(III) on the NMR relaxation times of the pore water was investigated. The decrease

in the relaxation times T1 and T2 corresponding to an increase in the dissolved iron(III) concentration was quantified. This

relation was used to monitor relative changes in the concentration of dissolved iron(III)-ions in natural sands.

Experiments were conducted to calibrate the iron(III) concentration from the measured relaxation times. These were done

in bulk water as well as in iron(III)-mineral bearing sands to take the effect of surface relaxation into account. It was

shown that for relatively coarse-grained sand fractions the effect of iron(III)-ions in solution dominates. This allows the

determination of the dissolved iron(III) concentration in natural sands by NMR. The method also enables us to capture the

changes in the iron(III) concentration with high temporal resolution. This was demonstrated in column experiments, in

which an acid (hydrochloric or sulphuric acid) was applied from the top on the sands to dissolve the mineralogical bound

iron(III) and were the dissolution of iron(III) can be captured with sufficient temporal resolution.

Keywords: Paramagnetic ion, iron(III), relaxation, mineral dissolution.

INTRODUCTION

Acid mine drainage is globally a major environmental is-sue. The groundwater is often heavily polluted near mines of sulphide minerals. This is due to the continuously release of low pH and heavy-metal loaded water from the mines drain-ing into groundwater [1]. The oxidation of pyrite and other metal-sulphide minerals by oxygen plays a key-role in acid mine drainage. It acts as a source of sulphate and iron (Fe) in groundwater, and of heavy metals in general in the environ-ment [1]. Dissolved iron is not detrimental to human health, but high concentrations have a negative impact on the use-fulness of the water. It can cause clogging of well-screen openings and pumps and has an unpleasant metallic taste.

Iron exists either in a ferrous Fe2+

or ferric Fe3+

state. In which form iron is dissolved in water depends on the amount of oxygen and upon its degree of acidity. Fe

2+ is oxidised to

Fe3+

in contact with oxygen or by the action of iron related bacteria. Fe

2+-ion is very soluble, but Fe

3+-ion is only soluble

at low pH values in water. Iron is a major component in acid mine waters. Precise measurements are important in order to understand the processes taking place regarding the dissolu-tion of Fe

3+-ions.

*Address correspondence to this author at the UFZ–Helmholtz Centre for

Environmental Research, Department of Hydrogeology, Permoserstr. 15,

04318 Leipzig, Germany; Tel: +493412351253; Fax: +493412351837;


In most present analytical methods [2], Fe3+

-ion is deter-mined by computing the difference between the total dis-solved Fe and the dissolved Fe

2+, i.e. Fe

3+-ion concentrations

are determined in an indirect way. The determination of Fe

3+-ion concentration in water is based on the determination

of Fe2+

-ion concentrations, followed by a separate determina-tion of the total dissolved Fe concentration after the reduc-tion of Fe

2+. The difference between the concentrations of

total dissolved Fe and Fe2+

is taken as the Fe3+

-ion concen-tration. One major problem with this approach is the overes-timation of Fe

3+-ion concentration at high Fe

2+-ion concen-

trations in the analyzed sample, where the difference be-tween total dissolved Fe and Fe

2+ is comparable to the error

of the determination.

In this study we applied nuclear magnetic resonance (NMR) relaxometry measurements as a method for the direct determination of dissolved Fe

3+-ion concentrations. This is

possible due to the paramagnetic properties of Fe3+

-ions in-fluencing the NMR measurements. Mostly, paramagnetic ions (e.g. Fe

3+, Mn

2+, Ni

2+ and Cu

2+) are seen as a complicat-

ing factor in the NMR interpretation due to their significant influence on the relaxation times. Previous studies concen-trate on the effects of solid paramagnetic species and par-amagnetic ions adsorbed to the solid phases [3, 4]. In many natural samples the iron concentrations are high enough to have a significant influence on the surface relaxivity. Bryar et al. [5] point out to be careful with the interpretation of

Investigation of Iron(III)-Release in the Pore Water The Open Magnetic Resonance Journal, 2010, Volume 3 47

pore size distributions between samples unless the iron con-centration is observed to be the same. Bryar et al. [5, 6] showed that the relaxation rates of Fe

3+-solutions depend

linearly on the Fe3+

-ion concentration.

Present NMR techniques allow a wide range of applica-tions. They can be used to provide information on the tempo-ral and spatial distribution of water and dissolved ions, on flow and transport processes [7-10]. NMR has also been used to investigate several microbial processes and biofilm properties via relaxation time differences [11,12]. NMR ap-plications are used to characterize sediments [13-16] and in well logging [17].

The aim of this study is to apply NMR relaxometry measurements to determine directly the dissolved Fe

3+-ion

concentrations in the sediment pore water. By measuring the relaxation times of water saturated sediment samples and taking surface relaxation at the pore matrix interface into account, the Fe

3+-ion concentration in the pore water solution

can be calculated. Additionally, column experiments are performed which show that this approach allows to monitor the time dependence of the dissolution of naturally occurring Fe

3+ in minerals. Thus, NMR relaxometry is used to gain

non-invasive and non-destructive insights into such proc-esses with high temporal resolution.

BACKGROUND AND THEORY

Relaxation in Solution and in Porous Media

The relaxation times of water decrease with increasing viscosity and increasing amount of solvents, respectively, and especially with increasing amount of dissolved par-amagnetic species, such as dissolved oxygen or Fe

3+-ions.

The theoretical background to these relaxation processes in liquids was founded by the Bloembergen-Purcell-Pound (BPP) theory [18]. Small concentrations of dissolved par-amagnetic metal ions cause huge decrease of the relaxation times. Hence, the bulk relaxation rate is a sum of the par-amagnetic and diamagnetic contributions and consequently proportional to paramagnetic ion concentration as well as the number of water molecules in the hydration sphere of an ion [19, 20]. The relaxation rates in bulk water (1/T1b, 1/T2b) with dissolved paramagnetic ions depend linearly on the ion-concentration c:

1 /T 1b, 2b =1 /T 1b0, 2b0 + R1, 2 * c (1)

where T1b0 and T2b0 are the relaxation times of paramagnetic ion-free water. The relaxivities R1,2 of the paramagnetic ion depends on the electron spin state of the ion and the mag-netic field strength at which the measurement is performed.

The relaxation rate of water in a porous material is larger than in bulk water because of additional mechanisms, which enhance the relaxation. The restriction of the water mole-

cules at the solid pore walls of the pore space and the content of paramagnetic ions at the surface of these particles cause an additional surface relaxation mechanism [21]. So the lon-gitudinal and transverse relaxation rates can be described as a sum of relaxation rates:

1 /T 1, 2 = 1 /T 1b, 2b +1 /T 1s, 2s (2)

where T1,2b and T1,2s are the bulk and surface relaxation times. In the fast diffusion regime (meaning that all protons travel to and relax at the pore surfaces in the time interval of the NMR experiment) the surface relaxation time depends on the volume-to-surface ratio V/S of the pores and is given by

T 2s = 1 / s *V / S (3)

where s is the surface relaxivity. It is influenced by the in-teraction of the pore fluid molecules with the internal pore surface and has been found to increase with the concentra-tion of minerals containing paramagnetic ions on the pore surface [22].

NMR Pulse Sequences

By applying a series of RF pulses, the total nuclear mag-

netization of the 1H nuclei of the (pore) water (i.e. the spin

system) can be manipulated. Such pulse sequences can be

used to determine the relaxation times. The following se-

quences have been implemented in the work reported in this

paper: inversion recovery (IR) sequence [23] and the Carr-

Purcell-Meiboom-Gill (CPMG) echo train [24, 25]. The IR

sequence was used for T1 measurements. In this sequence an

inverting RF pulse is followed, after the time interval t’, by

a /2 pulse (Fig. 1a). During the time interval t’, the mag-

netization is subjected to longitudinal (T1) relaxation only.

The CPMG sequence was used to obtain T2. After an initial

excitation of the magnetization by a /2 RF pulse a spin echo train is generated by RF pulses (Fig. 1b).

Paramagnetic Iron(III)-Ions

Paramagnetic ions are found to have a strong influence

on the NMR relaxation times. A paramagnetic species com-

monly found in groundwater, soils and sediments is Fe3+

.

Fe3+

-ions have the electron spin quantum number S = 2.5.

Compared to this, Fe2+

-ions have a lower electron spin val-

ues of 2. Due to this lower electron spin quantum number

Fe2+

-ions have a smaller effect on the relaxation times.

Hence NMR measurements allow the observation of redox

reactions using the Fe3+

- Fe2+

-redox pair [6]. In another

chemical form, i.e. as complexed Fe2+

, iron has recently been used to characterize biofilms [26].

The most common iron-oxides are goethite ( -FeOOH), hematite ( -Fe2O3), lepidocrocite ( -FeOOH), ferrihydrite (Fe(OH)3*nH2O) and magnetite (Fe3O4). These iron-oxides

Fig. (1). Pulse sequences used in this study, a) IR sequence to obtain T1, b) CPMG for measuring T2.

48 The Open Magnetic Resonance Journal, 2010, Volume 3 Mitreiter et al.

are pure Fe3+

-minerals, with the exception of magnetite which contains both Fe

3+ and Fe

2+. The concentration of total

iron in soils and sediments is normally around 0.2-5%. For porous media Foley et al. [3] and Bryar et al. [5] have shown that the surface relaxivity s and with this the surface relaxa-tion rate 1/T1,2s is linearly proportional to the concentration of paramagnetic ion containing minerals on the pore surface.

In groundwater and soil water, iron exists either in a fer-rous Fe

2+ or ferric Fe

3+ state. Which form iron takes, is con-

trolled by pH and redox potential. Ferric iron is relatively insoluble in water. Only at pH values below 3.0 it is ex-pected to appear in solution, e.g. in acidic water from metal mines and acidic forests soils. The relaxation rates of Fe

3+-

solutions are sensitive to pH-dependent speciation of the ion [5]. In aqueous solution at pH 1.0, Fe

3+ is present mostly

complexed by six molecules of water, [Fe(H2O)6]3+

, but at pH 3.0 hydrolysis reactions change the iron to a mixture of [Fe(H2O)6]

3+, [Fe(H2O)5OH]

2+ and [Fe(H2O)4(OH)2]

+ [5].

The number of exchangeable water molecules in the hydra-tion sphere decreases as pH increases [5].

With increasing pH, the Fe3+

adsorbs to the solid surfaces of surrounding porous media. Bryar et al. [5] showed that at pH 3.0 roughly 20 % of Fe

3+ can be adsorbed to surfaces.

The presence of adsorbed Fe3+

ions on the surface as well as Fe

3+-bearing solid phases present as surface coatings or as

separate mineral grains also significantly increase the surface relaxation.

MATERIALS AND METHODS

Materials and NMR Sample Preparation

The sand samples investigated originate from a sand de-posit about 10 km west of Leipzig, Germany. It was formed as a terminal moraine during the ice ages of the middle Pleis-tocene. Previous analyzes of sand samples from the same location [16] showed that the sand consists of quartz and feldspar and that the averaged elemental composition of the grain surfaces contains iron (Al1Si2.3O9.3Fe0.4(Mg,Ca)0.2

(Na,K)0.1). In the lab the sand was air dried and sieved yield-ing 5 fractions with grain diameters in the following ranges: 63 m – 125 m – 200 m – 500 m – 800 m – 1 mm.

The fluids used in this study were distilled water and so-lutions of Fe

3+. The iron solutions were prepared in distilled

water using ferric chloride (FeCl3 * 6 H2O) with the pH ad-justed at pH 1.0 using hydrochloric acid. Although the pH of pore fluids in natural settings will not often be as low as those used in this experiment, we used solutions with high acidity to ensure a complete dissolution of Fe

3+-ions, which

facilitates the interpretation of the results. Measurements of the relaxation time of the bulk fluids (T1,2b) were made using approximately 2 ml of the equilibrated fluid in an NMR tube.

The water-saturated sands were prepared by mixing 6 g sand with 1.5 ml distilled water in acid-washed glass tubes. Then acid (hydrochloric acid, sulphuric acid) was added from the top. For observing the time depended reactions with acid the measurements started right after sample preparation. All other sand samples prepared with an acid were left for seven days to insure the completeness of the reaction. Before NMR data were collected, excess fluid was removed from the surface of the sample. NMR data for each sample were

collected at room temperature (22°C). Evaporative losses of water from the sample during data collection were avoided.

NMR Experiments

The NMR relaxometry measurements were performed using a PC-controlled NMR console MARAN DRX (Reso-nance Instruments, GB). The home-built permanent magnet permits a magnetic flux density of B0 = 0.2 T, corresponding to a

1H resonance frequency of 9.1 MHz. The NMR samples

have 20 mm outer diameter and the same filling height. In the pulse sequences the length of the /2 and rf pulses was 7 and 14 μs. The T1-weighted time interval t’ of the IR pulse sequence was varied from 5 to 5,000 ms. In order to provide a sufficiently high signal-to-noise ratio, a minimum of 16 scans were acquired for each t’-value. The inter-echo time in the CPMG pulse sequence was varied from 100 to 400 s. A repetition delay time of RD = 5 s was sufficient to allow the nuclear spins to relax to equilibrium after each individual scan.


Fe3+

-Ions in Solution

We measured the longitudinal and transverse relaxation rates (T1

-1 and T2

-1) of solutions containing Fe

3+- ions with

concentrations ranging from 0 g l-1

up to 5 g l-1

at a pH of 1.0. Both measured relaxation rates are shown as a function of the concentration of dissolved Fe

3+-ions in Fig. (2). The

relaxation rates increase linearly with increasing Fe3+

-ion concentration, which corresponds well with the expectation according to the BBP theory. Linear fits for both relaxation rates were generated and are presented in Fig. (2) as solid (T1

-1) and dashed lines (T2

-1), respectively.

Fig. (2). The dependence of the bulk water relaxation rates T1b-1

and

T2b-1

on dissolved Fe3+

-ion concentration at pH 1.0 and room tem-

perature (22°C). The solid line in the graph is the calibration plot

for T1-1

, the dashed line is the one for T2-1

. Please note the logarith-

mic scale, which contort the linear dependency, especially for Fe3+

-

ion concentrations below 10 mg l-1

.

From the slope data of these calibration plots, using Equation 1, the relaxivities of Fe

3+-ions in water, R1 and R2,

were determined to be R1 = 0.1792 ± 0,0040 s-1

mg-1

l and R2 = 0.1393 ± 0,0018 s

-1 mg

-1 l (with correlation coefficients R

2

> 0.997).


These results are used to calibrate relaxation rates meas-ured by NMR to Fe

3+-ion concentrations. In the following

experiments we measured one of the two relaxation rates and then calculated the Fe

3+-ion concentration in solution. This

allows us to determine the Fe3+

-ion concentration in solution, without the necessity of destroying the sample or use any other analytical tool. Therefore, this technique can be used to capture the temporal progress of such dissolution reactions.

Relaxation Times T1,2 in Water-Saturated Sands

Internal surfaces of porous media have a significant in-fluence on the relaxation times. To assess this influence for the samples of natural sands the relaxation times of distilled water in five size fractions of the sand were measured. The results are shown as a function of the grain diameter in Fig. (3). Relaxation times in the sands are reduced compared to the relaxation times of bulk water. The smaller the grain sizes the larger is the decrease of the relaxation times. This is due to the interaction of the water

1H nuclei with the surface

of the sand grains. This effect on relaxation time is repre-sented by the s*S/V term in Equation 2.

Fig. (3). Relaxation times T1 and T2 of bulk water for five sand

fractions. The diameter d is the mean value of the particular sand

fraction.

For the calculation of the surface relaxivity s of water in the sands a porosity of 0.35 is assumed. The S/V ratio of a random packing of spherical grains of diameter d is given as [27].

S /V = 6 *(1 / 1)*1 / d (4)

By substitution Equations 3 and 4 in Equation 2 results

1 /T = 1 /Tb + s *6 * (1 / 1)*1 / d (5)

With this Equation the surface relaxivity s of the sands are calculated to be 0.0056 cm s

-1 for T1

-1 and 0.0123 cm s

-1

for T2-1

.

Relaxation Times of Water-Saturated Sand with Dis-solved Fe

3+-Ions

By adding an acid to a water-saturated sand sample, the Fe

3+-containing minerals dissolve and the Fe

3+-ion concen-

tration in solution increases. Fig. (4) shows the longitudinal relaxation time T1 for the different sand fractions without acid (from Fig. 3, black dots) and after adding H2SO4 (empty

triangles). The addition of the acid (33 mol H+) to the sand

caused a decrease in the T1 relaxation time corresponding to an increase in the Fe

3+-ion concentration in solution. For the

two fine-grained fractions just small changes in the relaxa-tion times could be detected because relaxation is fast al-ready without Fe

3+-ions. This does not imply that no Fe

3+

was dissolved, rather demonstrates the fact that for these fractions the surface relaxation is dominant. The larger the grain size the larger is the relative effect of the dissolved Fe

3+-ions on the total relaxation times.

Fig. (4). Relaxation times T1 for the 5 sand fractions after adding

H2SO4.

After 7 days time the reaction has presumably ended and the resulting relaxation times T1 are in the range from 120 ms to 180 ms for the five sand fractions (Fig. 4). Applying Equations 1, 2 and 3, the Fe

3+-ion concentration can be esti-

mated using the following Equation

1 /T 1 = 1 /T 1b0 + s *S /V + R1* c (6)

This corresponds to a Fe3+

-ion concentration ranging from 3 mg l

-1 for the smallest sand fraction up to 33 mg l

-1

for the largest grain sizes, as seen in Fig. (5). This range is due to the varying surface effects in the different sand frac-tions. The error bars were estimated using the Gaussian error propagation based on the diameter range in each fraction, i.e. the smallest und the largest diameter in each fraction. It is clearly seen that for smaller fraction the error in determining the Fe

3+-ion concentration is almost as large as the estimated

value. This implies that the estimated Fe3+

-ion concentration for the smaller sand fraction is uncertain and therefore, may be misleading. The grain size used in the following experi-ment is the fraction from 200 to 500 m, which is large enough to ensure that the influence of the surface relaxivity on T1 and T2 is low compared to the influence of the Fe

3+-

ions in solution. Another reason for this choice is the fact that this fraction of grain sizes has the largest percentage in our sand sample and generally speaking, this is one of the most important fractions when considering sand and soil samples.

As a next step column experiments were performed, where different concentrations of hydrochloric acid (HCl) were added from the top to one water-saturated grain size fraction to dissolve different amounts of Fe

3+. The added

acid concentrations ranged from 4 mol up to 52 mol H+.

50 The Open Magnetic Resonance Journal, 2010, Volume 3 Mitreiter et al.

Starting from the smallest concentration, the acid concentra-tion was increased in 6 steps till reaching the largest concen-tration. To investigate effects on changes of surface relaxiv-ities due to dissolution of mineral phases by increasing acid-ity the relaxation time distributions were analyzed. To the sand with the highest acid concentration (52 mol H

+) so-

dium hydroxide (NaOH) as base was added in surplus. This leads to the almost complete precipitation of the dissolved Fe

3+. The corresponding relaxation time distribution is

dominated by one peak at short relaxation times. In contrast, the relaxation time distributions before acid addition and after base addition were found to be very similar. Both showed a bimodal distribution with much longer relaxation times than in the state after hydrochloric acid addition and Fe

3+ dissolution (data not shown). Although after base addi-

tion, the iron precipitates not in the same chemical state and physical form as it previously existed, the initial relaxation time distribution can be reproduced, which implies that the surface relaxivity is not substantially affected by the iron minerals present.

For observing the dissolution of Fe3+

-ions with high tem-poral resolution, the directly measured T2 relaxation time is much better suited than the slower T1 measurement. One T2-measurement by the CPMG-method lasts only a few minutes while a quantitative T1 measurement by the IR method re-quires approximately 15 min up to about one hour (depend-ing on the required t’ intervals to capture the relaxation curve, see Fig. 1). Furthermore, the measurements were per-formed at a magnetic field of B0 = 0.2 T, so that influences on the T2 relaxation time due to internal field gradients can be neglected. Thus, we measured T2 and calculated the Fe

3+-

ion concentration using the calibration described by the Equation

c = (1 /T 2 1 /T 2b 1 /T 2s) / R2 (7)

where 1/T2 is the measured relaxation rate and the term (1/T2b+1/T2s), which describes the total relaxation rate due to bulk and surface relaxation, is determined experimentally before the addition of the acid. R2 is known from the slope of the calibration plot described before (compare Fig. 2). At the

beginning, measurements were taken continuously in time intervals of minutes. With time passing, i.e. the reaction slowing down, the time intervals were set larger, up to one hour between two measurements. The duration of the ex-periment was one day (24 hours). The temporal progress can be seen in Fig. (6) for three different acid concentrations. During the first hours a rapid increase in the Fe

3+-ion con-

centration in solution is observed, later the Fe3+

-ion concen-tration converges towards an asymptotic value. These data show that the dissolution of Fe

3+-ions from the minerals is a

fast process. The controlling mechanism to achieve equilib-rium in the sample is the diffusion of Fe

3+- ions as well as

H+-ions.

Fig. (6). Fe3+

-ion concentration in the 200 – 500 m sand fraction

after adding an acid during the first day of reaction.

SUMMARY AND CONCLUSION

The paramagnetic behaviour of Fe3+

-ions was used in this NMR relaxometry study to monitor temporal changes in Fe

3+-ion concentration in aqueous solution. To derive the

relation between both relaxation times (T1 and T2) and the Fe

3+-ion concentration in solution, different solutions were

prepared and the corresponding relaxation times were meas-ured. A linear relationship was found and parameterised. For the subsequent experiments the measured relaxation times could be transformed into Fe

3+-ion concentration using the

calibration plots. In the next step the influence of the sand grain size on relaxivities was investigated. As it is well known from theory, the smaller the particle size the more dominant is the surface relaxation process. This behaviour was reproduced for four different sand fractions. The conclu-sions are that a clear separation between both relaxation mechanisms, via Fe

3+-ions in solution and via surface relaxa-

tion, is not possible for fine-grained sands, where the relaxa-tion times are dominated by the surface relaxation. However, for larger size fraction of natural sand grains, the NMR method was shown to be capable to yield the Fe

3+-ion con-

centration in column experiments, were the increasing dis-solved Fe

3+ concentrations resulted from the penetration of

an acid front through a sand sample. The addition of acid to the sample made Fe

3+-ions dissolve from the iron minerals

present in the sands. The dissolution process was observed at a high temporal resolution and is characterized by a fast dis-

Fig. (5). The estimated Fe

3+-ion concentration for each sand frac-

tion with the error bars based on the diameter range for each frac-

tion.


solution of Fe3+

-ions and diffusion of H+ and Fe

3+-ions

through the pores of the sand.

This quantitative approach offers to monitor changes in dissolved Fe

3+-ion concentrations due to changes of acidity

in sediments and other release or consumption processes affecting dissolved Fe

3+-ions. The experiments proof that the

approach is applicable to natural sediments provided the grain size is not smaller than about 100 m. Via T2 meas-urements the temporal resolution is high enough to follow changes in concentration induced by geochemical and also microbial turn-over processes. Doing so, it should be noted that the formation of metal-organic complexes in sediment- and soil-solutions can reduce the relaxivities of Fe

3+ as com-

pared to those in model solutions [28]. Besides the dissolu-tion of paramagnetic ions the swelling of soil organic matter and the production and release of extracellular polymeric substances (EPS) by microbes can lead to changes in spin relaxation mechanisms and should be considered [29].

Methodological conclusions of the presented study are that for natural sands of medium and coarse size fractions the dissolution of Fe

3+-ions from natural, Fe

3+-containing miner-

als causes a substantial decrease of both, T1 and T2 relaxation times. Future use of the presented approach could make use of systems with low-field permanent magnets, in our case providing B0 = 0.2 T, which are cost-effective or even mo-bile. In general, a number of situations can be assessed in terms of changes in dissolved Fe

3+, for example (i) rate of

acidification and leaching in mine heaps, (ii) the dissolution of Fe

3+ in soil induced by organic acids produced by plant

roots to improve iron uptake from soil or (iii) iron reduction and re-oxidation during microbial degradation of organic contaminants in aquifers and Fe

3+ precipitation from solution

in general.

ACKNOWLEDGEMENTS

Financial support by the DFG (Germany) and the NWO (The Netherlands) via their joint International Research Training Group “Diffusion in Porous Materials” is gratefully acknowledged. The authors thank Dr. J. Kolander (Univer-sity of Leipzig, Germany) for his assistance with NMR measurements.

REFERENCES

[1] Appelo CAJ, Postma D. Geochemistry, groundwater and pollution. 2nd ed. A.A. Leiden: Balkema Publishers 2005.

[2] Skoog DA, Leary JJ. Instrumentelle analytik: grundlagen-geräte-anwendungen. German ed. Berlin: Springer 1996.

[3] Foley I, Farooqui SA, Kleinberg RL. Effects of paramagnetic ions on NMR relaxation of fluids at solid surfaces. J Magn Reson Ser A 1996; 123: 95-104.

[4] Kenyon WE, Kolleeny JA. NMR surface relaxivity of calcite with adsorbed Mn2+. J Colloid Interface Sci 1995; 170: 502-14.

[5] Bryar TR, Daughney CJ, Knight RJ. Paramagnetic effects of iron(III) species on nuclear magnetic relaxation of fluid protons in porous media. J Magn Reson 2000; 142: 74-85.

[6] Bryar TR, Knight RJ. Sensitivity of nuclear magnetic resonance measurements to changing redox conditions. Geophys Res Lett 2002; 29(24): 2197

[7] Van As H, van Dusschoten D. NMR methods for imaging of trans-port processes in micro-porous systems. Geoderma 1997; 80: 389-403.

[8] Nestle N, Wunderlich A, Niessner R, Baumann T. Spatial and temporal observations of adsorption and remobilization of heavy metal ions in a sandy aquifer matrix using magnetic resonance im-aging. Environ Sci Technol 2003; 37(17): 3972-77.

[9] Moradi AB, Oswald SE, Massner JA, Pruessmann KP, Robinson BH, Schulin R. Magnetic resonance imaging methods to reveal the real-time distribution of nickel in porous media. Eur J Soil Sci 2008; 59(3): 476-85.

[10] Oswald SE, Spiegel MA, Kinzelbach W. Three-dimensional salt-water-freshwater fingering in porous media: contrast agent MRI as basis for numerical simulations. Magn Reson Imaging 2007; 25(4): 537-40.

[11] Hoskins BC, Fevang L, Majors PD, Sharma MM, Georgiou G. Selective imaging of biofilms in porous media by NMR relaxation. J Magn Reson 1999; 139(1): 67-73.

[12] Seymour JD, Codd SL, Gjersing EL, Stewart PS. Magnetic reso-nance microscopy of biofilm structure and impact on transport in a capillary bioreactor. J Magn Reson 2004; 167(2): 322-27.

[13] Borgia GC, Fantazzini P, Mesini E. Water 1H spin-lattice relaxation as a fingerprint of porous media. Magn Reson Imaging 1990; 8: 435-47

[14] Mansield P, Issa B. Studies of fluid transport in porous rocks by echo-planar MRI. Magn Reson Imag 1994; 12(2): 275-78.

[15] Muller M, Kooman S, Yaramanci U. Nuclear magnetic resonance (NMR) properties of unconsolidated sediments in field and labora-tory. Near Surf Geophys 2005; 3(4): 275-85.

[16] Stallmach F, Vogt C, Kärger J, Helbig K, Jacobs F. Fractal geome-try of surface areas of sand grains probed by pulsed field gradient NMR. Phys Rev Lett 2002; 88(10): 5505.

[17] Kenyon WE. Petrophysical principles of applications of NMR logging. The Log Analyst 1997; 38(2): 21-43.

[18] Bloembergen N, Purcel EM, Pound RV. Relaxation effects in nu-clear magnetic resonance absorption. Phys Rev 1948; 73: 673-746.

[19] Solomon I. Relaxation processes in a system of two spins. Phys Rev 1955; 99: 559-65.

[20] Bloembergen N, Morgan LO. Proton relaxation times in paramag-netic solutions - Effects of electron spin relaxation. J Chem Phys 1961; 34: 842-50.

[21] Brownstein KR, Tarr CE. Importance of classical diffusion in NMR studies of water in biological cells. Phys Rev A 1979; 19(6): 2446-53.

[22] Kleinberg RL, Kenyon WE, Mitra PP. Mechanism of NMR relaxa-tion of fluids in rocks. J Magn Reson Ser A 1994; 108: 206-14.

[23] Günther, H. NMR-Spektroskopie. Berlin: Wiley-VCH 1992. (German)

[24] Meiboom S, Gill D. Modified spin-echo method for measuring nuclear relaxation times. Rev Sci Instrum 1958; 29: 688-91.

[25] Carr HY, Purcell EM. Effects of diffusion on free precession in nuclear magnetic resonance experiments. Phys Rev 1954; 94: 630-38.

[26] Bartacek J, Vergeldt FJ, Gerkema E, Jenicek P, Lens PNL, Van As H. Magnetic resonance microscopy of iron transport in methano-genic granules. J Magn Reson 2009; 200(2): 303-12.

[27] Vogt C, Galvosas P, Klitzsch N, Stallmach F. Self-diffusion of pore fluids in consolidated sediments by PFG NMR. J Appl Geophys 2002; 50(4): 455-67.

[28] Jaeger F, Rudolph N, Lang F, Schaumann GE. Effects of soil solu-tion's constituents on proton NMR relaxometry of soil samples. Soil Sci Soc Am J 2008; 72(6): 1694-707.

[29] Jaeger F, Grohmann E, Schaumann GE. 1H NMR relaxometry in natural humous soil samples: Insights in microbial effects on re-laxation time distributions. Plant Soil 2006; 280(1-2): 209-22.

Received: May 11, 2009 Revised: December 01, 2009 Accepted: January 26, 2010

© Mitreiter et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

A Possible Difference in the Surface Relaxivity of Costal and Inland Sands

Joseph P. Hornak*,1, Gianni Ferrante

2, Andrew Coy

3 and Evan R. McCarney

3

1Rochester Institute of Technology, Rochester, New York, USA;

2Stelar, Mede (PV), Italy;

3Magritek, Wellington, New

Zealand

Abstract: The 1H nuclear magnetic resonance (NMR) spin-lattice relaxation rate of hydrated sands is related to the sur-

face-to-volume ratio of the voids or pores between the hydrated sand grains and the surface relaxivity of the grains. The

electron spin resonance (ESR) signal is often used to predict the relative surface relaxivity as the surface relaxivity is

thought to be proportional to the concentration of paramagnetic species in the sand grains. We have identified a discrep-

ancy in the surface relaxivity and ESR signal of an ocean beach sand compared to two sands of similar diameter from in-

land deposits. This difference can either be due to more surface weathering of the inland sand or more paramagnetic mate-

rial from seawater adhering to the ocean sand.

Keywords: Hydrated sand, NMR surface relaxivity, spin-lattice relaxation rate, R1.

INTRODUCTION

The relationship between the 1H nuclear magnetic reso-

nance spin-lattice relaxation rate (R1) of a hydrated porous material, the R1 of bulk water filling the pores (R1B), the sur-face relaxivity of the pores ( ), and the surface to volume ratio of the pores (S/V)P is the basis of many geophysical studies related to soil [1-3].

R1 = R1B + (S/V)P (1)

NMR spin relaxation values are used to study hydration, porosity, and soil contamination [4-6]. Magnetic resonance sounding [7] is useful for finding aquifers and the R1 value is often used to infer pore size. Near-surface magnetic reso-nance imaging [8,9] of hydrated soil may one day allow im-aging of buried utilities. The signal from this technique is strongly dependent on the R1 value of the hydrated soil.

We recently reported on this relationship for some fully hydrated unconsolidated natural and synthetic sands [4]. In that study we adopted the convention that for similarly shaped grains, (S/V)P is inversely proportional to the sand grain diameter (d). In doing so, we were able to fit Eq. (1) to R1 values for different diameter sands, thus calculate for the sands. The values used for the fit were proportional to the electron spin resonance signal for the dry sands. This is a reasonable find as the surface relaxivity is caused by par-amagnetic impurities in the surface layer of atoms of the grains. The concentration of paramagnetic impurities in the entire grain volume, represented by the ESR signal, is pre-sumed to be uniform throughout and the same as the concen-tration in the outer surface layer governing .

A preliminary report indicated that this relationship may not be true for all sands [10]. Surface weathering or surface adsorption of paramagnetic metals may cause the concentra-

*Address correspondence to this author at the RIT Magnetic Resonance

laboratory, Center for Imaging Science, Rochester Institute of Technology,

54 Lomb Memorial Drive, Rochester, NY 14623-5604, USA; Tel:

585.475.2904; Fax: 585.475.5988; E-mail: [email protected]

tion of these metals on the surface layer to differ from that within the grain. The comparison made to exemplify this difference is between an active beach sand from Asilomar, CA USA, and two inland sands from Illinois USA. We pre-sent the details of these findings in this paper.

In the previous paper, we reported R1 values at proton resonance frequencies ( ) between 30 MHz and 10 kHz. This data was used to extrapolate a value for R1 and at = 2.5 kHz, the approximate resonance frequency of protons in the magnetic field of the Earth. This extrapolation was risky because the dispersion in R1 with could either continue at < 10 kHz or stop and R1 level off. In the current presentation, we have added a datum point at 1.9 kHz using an Earth’s field NMR spectrometer that eliminated the need to extrapo-late.

BACKGROUND

The R1 of a nuclear spin system is influenced by time-varying magnetic fields. In pure water, R1 results from di-pole-dipole interactions between water protons modulated by the rotational motions. The frequency dependence of these time varying magnetic fields is given by the spectral density function, J( ), for bulk water [11]. J( ) is flat over a broad range of frequencies from zero to a frequency equal to the inverse of the correlation time ( c) for the rotational motion, where J( ) decreases to zero. This change in J( ) is referred to as a dispersion and plots of the R1 value as a function of are referred to as NMR dispersion plots. The R1 of the

1H

spins in pure water depends on the number of time-varying magnetic fields at and 2 experienced by the nucleus [12]. Water molecules experiencing an electrical charge from ions will possess a different c due to the electrostatic attraction between the polar water molecule and the ion.

In the presence of paramagnetic ions, the water R1 is also influenced by electron-nuclear dipolar and contact interac-tions between the nuclear spin of water hydrogens and the electron spin of paramagnetic material [13,14]. The correla-tion times for these interactions are influenced by a rota-tional correlation time for a water molecule in the hydration

A Possible Difference in the Surface Relaxivity of Costal The Open Magnetic Resonance Journal, 2010, Volume 3 53

sphere of the paramagnetic material, the electron spin-lattice relaxation time, the proton-electron hyperfine coupling con-stant, and the electron spin-spin relaxation time.

The R1 of water adsorbed on a surface is also influenced by electron-nuclear dipolar and contact interactions, except the interaction is between the nuclear spin of water and the electron spin of paramagnetic material both in and on the surface. These interactions are influenced by the correlation times described previously for paramagnetic ions in solution, plus ones for diffusion ( m) and desorption ( s) motions of the water on and from the surface of a particle [15].

Randomly packed, unconsolidated, solid sand particles form a network of pores connected by channels [16]. The size of a pore is proportional to the diameter of the particles [15]. When fully hydrated, water fills the pores and channels between the particles. The mechanism of NMR spin relaxa-tion in porous media is well established [1-3], and can be divided into two limiting cases: fast-diffusion or surface-limited, and slow-diffusion or diffusion-limited. In the fast-diffusion case, the magnetization recovery for a single pore is monoexponential and depends on the (S/V)P. In the slow-diffusion regime, the dominant relaxation occurs at the sur-face but the diffusion of spins to the surface is slow. In this case for a single pore, the return of magnetization to equilib-rium is multiexponential and depends on the pore shape. McCall et al. [17] have presented modifications to the theory for coupled-pore systems. Sands similar to the Asilomar sand contain well-coupled pores of similar dimensions [4] and hence we focus on the fast-diffusion case.

In the fast-diffusion case, R1 is influenced by bulk water

and a thickness of water ( ) adsorbed on the surface of the

pore, both with distinctly different R1 values. On the time-

scale of the NMR relaxation measurement, water molecules

readily diffuse between the two environments. R1 is a func-

tion of R1B and the surface relaxation rate ( R1S ).

R1 = R1B + R1SS

V P

(2)

Eqn. 2 is an alternative representation of eqn. 1 as the surface relaxivity is defined as R1S .

Similar to the relaxation in porous materials model of eqns. 1 and 2, water in the presence of solvated diamagnetic ions can be characterized as existing in two different envi-ronments. There is structured water in the solvation shell of the ions, and bulk water not experiencing the effect of the ions. The fraction of structured water associated with the salvation shells of the ions is , and the remaining fraction (1- ) is bulk water. On the NMR timescale, with rapid ex-change between the two environments the measured relaxa-tion rate for an aqueous solution of diamagnetic ions be-comes

R1 = 1( )R1B + R1H (3)

where R1H is the R1 of the structured water in the hydration shell [18,19]. R1H is characterized by a unique c from di-pole-dipole interactions between the water molecules in the hydration shell.

In the fast-diffusion case, the J( ) from all the above in-teractions contribute, although not equally, to the overall frequency dependence of R1 for hydrated sands. Fig. (1) summarizes the frequency dependence of these interactions. For bulk water systems at 20 °C, the rotational c 3.5 10

-12 s

[11], placing the dispersion from these motions at a frequency well beyond the reach of NMR spectrometers. The interaction of water protons with paramagnetic ions in solu-tion or in the lattice can cause dispersions between approxi-mately 10

5 and 10

11 s

-1. The surface diffusion/desorption

interaction is unique in that it is characterized by a large con-stant R1 at < 1/ s, a small constant R1 at > 1/ m, and a logarithmic relation between 1/ s and 1/ m. The value of m is ~4 10

-10 s and s is less than ~1 10

-4 s [15]. The surface dif-

fusion/desorption interaction is also unique in that it pro-duces a dispersion with a linear relationship between R1 and log( ) while the other interactions all produce power law dispersions.

Fig. (1). Plots of the normalized spectral density function (J) as a function of

magnetic field reported as proton NMR frequency for a) proton-electron

hyperfine, b) surface diffusion/desorption, c) proton-electron dipolar, and d)

proton-proton dipolar interactions.

The surface relaxivity is caused by the electron-nuclear dipolar and contact interactions between the

1H nuclear spins

in water and paramagnetic materials on and in the sand grains. These paramagnetic substances can be either ad-sorbed on the surface [20], or randomly distributed through-out the grain and hence exposed at the surface. With the lat-ter, it is possible to assess the surface concentration using ESR as the area of the ESR absorption signal is proportional to the concentration of electron spins in the sample. For par-amagnetic materials adsorbed on the surface, it is only possi-ble to relate the ESR signal to when the concentration of spins in the grains is known separately or zero.


Three natural quartz sands were compared: Asilomar Beach, Ottawa, and Oregon Sands. The first is from the Asi-lomar Beach in Asilomar, CA USA and these sand grains are currently being formed from the erosion of the native bed-rock. Both the Ottawa and Oregon sand grains were formed

0.001 0.1 10 1000 100000

HBo (MHz)

J

a

b c

d

54 The Open Magnetic Resonance Journal, 2010, Volume 3 Hornak et al.

from the same erosive forces associated with a Paleozoic era sea approximately 465 million years ago [21]. Since then, these two sands have experienced varying degrees of surface weathering.

The Asilomar Beach Sand was collected fully hydrated with ocean water and used without sieving. This fully hy-drated sand was used as is for the NMR studies. For ESR and geometric measurements, the sand was rinsed with 18 M ·cm deionized (DI) water and dried at 200 °C. Ottawa Sand (F110, US Silica, Berkeley Springs, WV, USA) and Oregon Sand (7020 Granusil, Unimin Corp., New Canaan, CT, USA) were acquired from the respective suppliers. The Ottawa and Oregon sand samples were cleaned by rinsing with DI water, oven dried at 200 ºC, and sieved using a me-chanical 76 mm sieve shaker (SS-5, Gilson Co. Inc., Wor-thington, OH) to obtain the diameters studied [4]. ESR and geometric measurements were made on the dry samples, while NMR was performed on the sands fully hydrated with DI water.

The geometric properties of all the sand grains were measured using an optical microscope (Eclipse E600PL, Nikon) with image analysis software (analySIS, Olympus). The average minimum and maximum diameters (dmin and dmax) of a random sampling of 100 grains were measured and used to calculate the aspect ratio (RA = dmax/dmin), the average diameter, and uniformity ( d/d) of the grains. All the geo-metric properties are used to estimate similarities in (S/V)P.

The R1 values of the Asilomar Sand were measured by two techniques. R1 values were measured at an Earth’s mag-netic field of 44.6 μT (1.9 kHz) using an Earth’s field NMR spectrometer (Terranova-MRI, Magritek, Wellington, New Zealand). This same instrument was used to measure the R1 in its polarizing field of 18 mT (766 kHz). Both of these measurements were made on 250 ml volume samples. R1 values as a function of field strength were measured between 0.47 mT (20 kHz) and 0.19 T (8 MHz) using a field cycling NMR spectrometer (SMARtracer, Stelar, Mede, Italy). Field cycling measurements were made on samples in 10 mm NMR tubes. The R1 value for the Asilomar sand was interpo-lated at 2.5 kHz from Fig. (2).

Fig. (2).

1H NMR R1 values as a function of resonance frequency between

1.9 kHz and 9 MHz for fully hydrated Asilomar Beach sand, Asilomar

Beach ocean water, and deionized (DI) water as measured by an Earth’s

field (Magritek) and field cycling (Stelar) NMR spectrometers.

R1 values for the Ottawa and Oregon Sands were taken from [4]. This study measured R1 as a function of d and . Because of the high density of data points in d from this study, our R1 values for these two sands were calculated from their fitted plots of R1 vs. d at d = 340 μm, and thus assuring comparison values for similar diameter sand grains.

The paramagnetic metal content of the sand was deter-mined using a low frequency electron spin resonance ESR spectrometer [22] operating at 247 MHz with a magnetic field sweep between 0 and 15 mT. Sand samples were pre-pared by rinsing with 18 M ·cm deionized water and dried. Sand samples were placed in 2 cm diameter glass sample tubes to a height that completely filled the ESR sample probe. Conventional first-derivative ESR spectra were re-corded. The relative number of paramagnetic spins per mass of sand (Nm) was calculated from the double integral of the ESR spectrum (S) minus the same for an empty sample tube (So) all divided by the mass of sand filling the probe (m).

Nm = (S - So)/m (4)

Assuming similar density quartz sand grains, this Nm is pro-portional to the concentration of spins in a grain.


All three sands had similar aspect ratios and mean diame-ters. (See Table 1). Based on this, the sand grains should pack similarly and have similar (S/V)P ratios. However, the distribution of grain diameters in the Asilomar Sand was greater than that for the Ottawa and Oregon Sand samples, because the Asilomar Sand was not sieved. This 2.3 times larger distribution of d will cause a difference in the packing and (S/V)P, which we will address later.

Table 1. Comparison of Hydrated Sands at = 2.5 kHz

Property Asilomar Ottawaa Oregon

a

d (μm) 340 340 340

d/d 0.30 0.13 0.13

RA 1.5 1.5 1.4

Relative ESR Signal 0.9 1.0 1.7

Relative 2.5 1.0 1.6

(R1-R1B) (s-1) 0.80 0.32 0.49

aData at specified diameter were interpolated from [4].

The R1 values for Asilomar Beach sand hydrated with ocean water, Asilomar Beach ocean water, and DI water as a function of proton resonance frequency are presented in Fig. (2). Both Stelar and Magritek instruments produced consis-tent data at their overlapping frequency of 766 kHz. Com-pared to 18 M ·cm deionized water, the ocean water shows a higher R1 at all values. This is attributable the 0.5 M diamagnetic salt ions in sea water acting through eqn. 3, and not the < 30 nM paramagnetic ions [23]. The slight slope of the line for the two bulk water samples is not predicted by theory and may reflect measurement uncertainty.

The R1 of the hydrated sand grains was greater than that for ocean water and increased with decreasing resonance

Asilomar Beach Data

0.1

1

10

0.001 0.01 0.1 1 10

(MHz)

R1 (

s-1

)

Sand (Stelar)

Sand (Magritek)

Ocean Water (Stelar)

Ocean Water (Magritek)

DI Water (Stelar)

DI Water (Magritek)

A Possible Difference in the Surface Relaxivity of Costal The Open Magnetic Resonance Journal, 2010, Volume 3 55

frequency to about 20 kHz. The dispersion in R1 between 20 kHz and 8 MHz is the result of surface relaxation due to in-teractions of the water with paramagnetic impurities in the sand grains and is consistent with theory [15]. It was not possible to determine if the data followed a power law or log( ) relationship as both forms fit the data equally well. Below 20 kHz, R1 appears to level off. The leveling off is significant because it sets a lower limit on the dispersion. Extrapolation of the dispersion to the resonance frequency of 1H in the Earth’s magnetic field may yield a higher R1. It

may be more appropriate to use the R1 value at lowest fre-quency of the field cycling studies, i.e. 10 or 20 kHz, as the extrapolated R1 at 2.5 kHz then to extend the dispersion be-yond the 10 kHz R1 value. This hints that the extrapolation in the previous study may have produced slightly higher values at 2.5 kHz then actual values.

The R1 value of the Asilomar Sand was compared to that of the previously reported Ottawa and Oregon, IL Sands [4]. Fig. (3) shows a plot of (R1–R1B) vs. log(d) for the Ottawa and Oregon Sands. The data was fit with Eq. (1) to obtain the relative values of 1.0 and 1.6 respectively. (See Table 1). These values were proportional to the relative ESR signals of 1.0 and 1.7. The R1 values were measured for several dif-ferent diameter samples; however, these were not the same as the Asilomar Sand grain diameter. Therefore, the (R1–R1B) values of Table 1 are based on the fit to the data in Fig. (3) at d = 340 μm. For comparison purposes, the datum point for the Asilomar Sand in Fig. (3) is displayed with a line seg-ment for a = 2.5. This relative value was determined by fitting eqn. 1 to the point.

Fig. (3). 1H NMR (R1–R1B) values as a function of diameter (d) for fully

hydrated Oregon ( ), Ottawa ( ), and Asilomar Beach ( ) Sands. Solid

lines are best fits of the data using Eq. (1). The line segment through the

Asilomar Sand point is provided as a reference for the value.

The previous study of hydrated Ottawa and Oregon sands showed that the measured (R1-R1B) values correlated well with the ESR signal and could thus be used to predict [4]. In our study, the (R1-R1B) value of hydrated Asilomar Sand is greater than that of inland Ottawa and Oregon Sands. (See Table 1). As per Eq. (1), the difference must be due to (S/V)P or . The similar d and RA values would assure very similar packing of the grains if the d/d values were similar. The

literature suggests that the difference in d must be greater than 6.5 to see a significant change in the porosity [24,25]. Therefore, the 2.3 times larger d may not cause a significant change in (S/V)P and is not the primary cause of the larger (R1-R1B) for Asilomar Sand. This points to as the cause of the larger (R1-R1B) for the Asilomar Sand.

The ESR signal of the Ottawa and Oregon Sands corre-late well with the value needed to fit the data of Fig. (3), but the Asilomar Sand value does not. This observation leads us to believe that the concentration of paramagnetic material on the surface of the grains of sand may be different from the concentration in the center. The ESR signal is pro-portional to the total paramagnetic concentration. Either a higher surface concentration of paramagnetic material in the Asilomar Sand, or a lower surface concentration of par-amagnetic material in the Ottawa and Oregon Sands could account for the difference. The former could be attributed to paramagnetic material from the seawater adhering to the surface of the grains. The rinsing of the sand for ESR meas-urements and the ESR averaging of the signal from the sur-face and interior of the grains would account for the higher than predicted by ESR. The latter cause could come about from repeated weathering of the surface of the inland sand grains. Acid washing the surface of the Asilomar Sands might remove the adsorbed and surface layer paramagnetic materials, however rehydrating with ocean water might cause some ions to be readsorbed. The best way to distin-guish between the two causes is to use an analytical tech-nique that measures only the surface concentration of these trace paramagnetic metals.

CONCLUSIONS

Unlike in a previous study where the ESR signal was found to be proportional to the surface relaxivity of several sands and glass spheres, in this study the ESR signal was found not to be proportional to the surface relaxivity of a specific ocean beach sand hydrated with ocean water. Al-though (S/V)P could not be completely ruled out as the cause, it is believed that a difference between the surface and over-all concentration of paramagnetic material for the sand grains is most probably the cause. The difference may be due to paramagnetic ions found in sea water being adsorbed on the surface of the grains of Asilomar Sand, or the weathering of paramagnetic ions out of the surface layer in the geologi-cally older inland sands. More studies are needed to confirm the cause and determine the best analytical technique for measuring the surface relaxivity.

REFERENCES

[1] Korringa J, Seevers DO, Torrey HC. Theory of spin pumping and

relaxation in systems with low concentration of electron spin reso-nance centers. Phys Rev 1962; 127: 1143-50.

[2] Brownstein KR, Tarr C. Importance of classical diffusion in NMR studies of water in biological cells. Phys Rev A 1979; 19: 2246-

453. [3] Kleinberg RL, Kenyon WE, Mitra PP. Mechanism of NMR relaxa-

tion of fluids in rock. J Magn Reson A 1994; 108: 206-14. [4] Bray CL, Bryant RG, Cox MJ, et al. The 1H nuclear magnetic

resonance spin-lattice relaxation rate of some hydrated synthetic and natural sands. J Environ Eng Geophys 2009; 14: 49-61.

[5] Hertzog RC, White TA, Straley C. Using NMR decay time meas-urements to monitor and characterize DNAPL and moisture in sub-

surface porous media. J Environ Eng Geophys 2007; 4: 293-306.

= 2.5 KHz

0.1

1

10

100 1000d (m)

(R1-R

1B)

(s-1

)

Oregon, IL Sand

Ottawa, IL Sand

Asilomar Beach, CA

56 The Open Magnetic Resonance Journal, 2010, Volume 3 Hornak et al.

[6] Pohlmeier A, Haber-Pohlmeier S, Stapf S. A fast field cycling

nuclear magnetic resonance relaxometry study of natural soils. Vadose Zone J 2009; 8: 735-42.

[7] Legchenko A, Baltassat J-M, Beauce A, Bernard J. Nuclear mag-netic resonance as a geophysical tool for hydrogeologists. J Appl

Geophys 2002; 50: 21-46. [8] Bray CL, Hornak JP. Unilateral MRI using a rastered projection. J

Magn Reson 2007; 188: 151-9. [9] Bray CL, Hornak JP. Underground Variations in BEarth: Implica-

tions for Near-Surface MRI. Concepts Magn Reson 2009; 35B: 153-67.

[10] Hornak JP, Ferrante G, Coy A, McCarney ER. A 1H NMR Spin-Lattice Relaxation Time Study of Asilomar Sands. In: 50th Experi-

mental NMR Conference, 2009; Asilomar, CA, USA. [11] Bloembergen N, Purcell EM, Pound RV. Relaxation effects in

nuclear magnetic resonance absorption. Phys Rev 1948; 73: 679-712.

[12] Abragam A. The principles of nuclear magnetism. Clarendon: Oxford 1961.

[13] Soloman I. Relaxation process in a system of two spins. Phys Rev 1955; 99: 559-65.

[14] Bakhmutov VI. Practical NMR relaxation for chemists. Wiley: Hoboken, NJ 2004.

[15] Godefroy S, Korb J-P, Fleury M, Bryant RG. Surface nuclear mag-netic relaxation and dynamics of water & oil in macroporous me-

dia. Phys Rev E 2001; 64: (021605)1-13.

[16] Hills BP, Belton PS, Quantin VM. Water proton relaxation in het-

erogeneous systems: I. Saturated randomly packed suspensions of impenetrable particles. Mol Phys 1993; 78: 893-908.

[17] McCall KR, Johnson DL, Guyer RA. Magnetization evolution in connected pore systems. Phys Rev B 1991; 44: 7344-55.

[18] Broersma S. Nuclear magnetic relaxation of nonuniform systems. J Chem Phys 1956; 24: 153-60.

[19] Melnichenko NA, Bazhanov AV, Kupriyanov AS. Temperature dependence of NMR relaxation rate in some aqueous electrolytes

of semimolar concentration. J Struct Chem 2003; 44: 397-403. [20] Bryar TR, Daughney CJ, Knight RJ. Paramagnetic effects of

iron(III) species on nuclear magnetic relaxation of fluid protons in porous media. J Magn Reson 2000; 142: 74-85.

[21] Pitman JK, Goldhaber MB, Spöetl C. Regional diagenetic patterns in the St. Peter Sandstone: implications for brine migration in the

illinois basin. U.S. Geological Survey Bulletin 2094–A, United States Government Printing Office, Washington, DC USA 1997.

[22] Hornak JP, Spacher M, Bryant RG. A modular low frequency ESR spectrometer. Meas Sci Technol 1991; 2: 520-22.

[23] Millero FJ. Chemical oceanography. Boca Raton: Taylor & Francis 2006; pp. 91-2.

[24] Cubrinovski M, Ishihara K. Maximum and minimum void ratio characteristics of sands. Soils Found 2002; 42: 65-78.

[25] Westman ER, Hugill HR. The packing of particles. J Am Ceram Soc 1930; 13: 767-79.

Received: July 28, 2009 Revised: October 30, 2009 Accepted: November 04, 2009

© Hornak et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

Relaxation in a Natural Soil: Comparison of Relaxometric Imaging, T1 – T2 Correlation and Fast-Field Cycling NMR

S. Haber-Pohlmeier*,1, S. Stapf

2, D. van Dusschoten

3 and A. Pohlmeier

4

1ITMC, RWTH Aachen University, Worringer Weg 1, Aachen, Germany;

2Institute of Physics, Technical University

Ilmenau, Germany; 3ICG3, Research Centre Jülich, German;

4ICG4, Research Centre Jülich, Germany

Abstract: Longitudinal and transverse relaxation times are used to characterise the pore system of a natural Kalden-

kirchen sandy loam. Here we present new results obtained by relaxometric imaging (MEMS) and two-dimensional T1-T2

correlation relaxometry, and compare these with available T1- relaxation time distributions of water obtained by the analy-

sis of fast field cycling relaxometry (FFC) data. The soil shows relatively broad bimodal distribution functions P(T1) and

P(T2) with a T1/T2 ratio of about 2:1. The average T1 as well as the spatial distribution, which are obtained from the re-

laxometric imaging corresponds well to the relaxometric results. From the analysis of the field dependent FFC data at low

field including T1 data obtained at high field the basic locally averaged relaxation mechanism is derived from the disper-

sion curve, i.e. the dependence of the relaxation rate from the magnetic field strength over five orders of magnitudes.

From this we conclude that two-dimensional diffusion at locally flat surfaces controls the relaxation, i.e. the shapes of the

distribution functions are controlled by surface relaxation.

Keywords: NMR, relaxometry, imaging, pore size distribution, Brownstein Tarr, T1-T2 correlation.

INTRODUCTION

Soils are natural porous media of highest importance for food production and maintenance of water resources. For these functions a prominent property is their ability to retain and transport water which is mainly controlled by pore size distribution. The latter is related to NMR relaxation times of the water molecules filling the pores, for which different measurement methods can be used. Most widespread are Carr Purcell Meiboom Gill (CPMG) for transverse relaxation and inversion or saturation recovery (IR, SR) for longitudi-nal relaxation (T1) [1]. Besides this, T1 can also be deter-mined by Fast Field Cycling (FFC) [2] which allows investi-gations at variable field strength. Relaxometric methods can also be combined among themselves yielding correlation maps [3] or with imaging (MRI) leading to spatially resolved T1 and T2 times, which can also be analyzed statistically [4-9]. Generally, imaging methods contain more information than relaxation methods, since imaging resolves the informa-tion in two or three dimensions which is especially important for heterogeneous samples such as natural soil. But this is often paid by increased noise and a much smaller number of data points.

The aim of this study is to investigate the multimodal na-ture of the relaxation distribution functions observed in re-laxation experiments of macroscopic samples at low fields and to find an answer to the question whether or not they are caused by macroscopic structures like fractures or by the inherent microscopic heterogeneity. For this purpose we measure T1 and T2 maps by relaxometric imaging and two-dimensional T1-T2 correlation maps and compare them with

*Address correspondence to this author at the ITMC, RWTH Aachen Uni-

versity, Worringer Weg 1, Aachen, Germany; Tel: +49(241) 80-264247;

Fax: +49(241) 80-22185; E-mail: [email protected]

available data from fast field cycling relaxometry (FFC) ex-periments of a saturated natural sandy loam from Kalden-kirchen [10]. This will help in the future to better assess the information obtainable from magnetic resonance imaging which aims for understanding flow processes in natural soils.

MATERIAL AND METHODS

Sample Preparation

The Kaldenkirchen soil sample (73% sand, 23% silt, 4% clay, 0.25% Fe, 0.02% Mn, , 1% TOC) was sieved (<2mm) and filled, air-dry, into 9x100mm

2 quartz glass cuvettes

which are closed at the bottom by a porous glass filter plate. The cuvettes were then centrifuged for 30 min at 3000rpm and finally wetted from the bottom for 48 h. The maximum water content was = 0.45. Then the sample was sealed and measured by fast field cycling NMR relaxometry (FFC)[10], IR-CPMG, and imaged by a multislice-multiecho method (IR-MEMS).

Fast Field Cycling Relaxometry (FFC)

This method investigates the relaxation of an excited spin ensemble along the direction of the main magnetic field after a field jump [2]. The timing diagram is shown in Fig. (1). At the beginning the system is in thermal equilibrium corre-sponding to B=Bpol or B=0, respectively. Then the field strength is switched quickly to Brlx, but the spin ensemble does not follow instantaneously, i.e. the restoration of the new thermal equilibrium obeys a first order kinetic law with a characteristic relaxation time T1. The polarization state of the ensemble is probed after varying time intervals , during which the sample is left at the magnetic field strength Brelax. This probing is performed by switching the magnetic field strength to Bacq and subsequently applying a 90° radio-frequency pulse. The free induction decay (FID) following this pulse is recorded and analyzed off-line in order to obtain

58 The Open Magnetic Resonance Journal, 2010, Volume 3 Haber-Pohlmeier et al.

its amplitude, which is proportional to the magnetisation Mz( ). A complete relaxation curve is then obtained by repeating the measurement cycle for different values of . In ideally homogeneous systems Mz( ) is a single exponential function with a unique relaxation time T1, in heterogeneous porous media multiple exponential functions are expected. The whole procedure may be repeated for different values of Brelax. All FFC measurements were performed on a Stelar Spinmaster (Stelar, Mede, Italy). T1 relaxation curves were monitored at relaxation fields corresponding to Larmor fre-quencies of = Brlx/2 = 20, 10.6, 5.6, 3, 1.6, 0.823, 0.435, 0.23, 0.121, 0.064, 0.034, 0.018, 0.009, and 0.005 MHz ( is the gyromagnetic ratio of

1H). The data are analysed by 1D

inverse Laplace transformation [3].

Fig. (1). Fast Field Cycling Method.

IR-MEMS NMR Imaging

The imaging experiments were performed on a 7T verti-cal bore magnet (Oxford Instruments, Oxford, UK), equipped with a Bruker (Karlsruhe, Germany) mini-imaging gradient set (Gmax =300 mT/m), a 38mm rf birdcage coil, and a Maran DRX console. The used imaging sequence was a inversion recovery multiecho multislice sequence (IR-MEMS), which allows spatially resolved simultaneous de-termination of T1 and T2 relaxation times. A simplified tim-ing diagram is displayed in Fig. (2).

Fig. (2). Simplified IR-MEMS pulse sequence.

The basic idea of MEMS is a preparatory initial 180° pulse which inverts all magnetisation into –z direction. After a period tIn during which longitudinal relaxation evolves, a 90° slice selective soft pulse is applied which is followed by a CPMG echo train for the T2 determination. This is repeated for several different inversion time periods. Due to the addi-tional switching of phase “ph” and read-out “ro” encoding gradients the complete inversion-recovery CPMG pulse se-quence becomes spatially resolved [3].

The signal intensity in each voxel obeys the following relation:

S(x, y, z,TIn

,TE

) = S0(x, y, z) 1 F exp(

tIn

T1

exp(nt

E

T2,app

(1)

with the inversion time tIn, the echo time tE, the number of echoes n, and the longitudinal and apparent transverse relaxation times, T1, and T2,app. The transverse relaxation time T2 is denoted as apparent since it comprises the true T2 plus surface and diffusional effects. The prefactor F is theoretic- cally 2, but in practice it should be fitted to the data in order to account for fast relaxation processes during the dead time of the receiver and effects of imperfect inversion pulses. Since soils possess quite fast apparent transverse relaxation times, T2,app, we used the shortest adjustable value of tE = 1.5 ms, which is paid by a relatively high slice thickness of 3mm. Further parameters were 0.009s < tIn < 1 s, tR = 1.8s, FOV = 64x32 (32 x 16mm), and 14 axial slices scanned in interleaved mode. Original data S(x,y,z,tIn,ntE) obtained for each voxel were fitted by Eq. 2, yielding maps (2D represen- tations for chosen slices) of the physical quantities S0, T1, and T2,app.

T1-T2 Correlation

The relation between longitudinal and transverse relaxa-tion times can also be combined in a T1-T2 single NMR re-laxation experiment. This is the non-spatially resolved ver-sion of IR-MEMS imaging and for the data analysis also Eq. 2 is valid. We have applied this method on the Kalden-kirchen soil sample described above in a 24 MHz Halbach low field scanner operated by a Stelar PC-NMR console (Stelar, Mede, Italy). tE = 0.3ms, n = 512, 0.001s < tIn < 1s. The data are analyzed by 2D inverse Laplace transformation using Eq. (1) as kernel for Eq. (2):

Mz( ) = P(T

1,T

2) 1 2exp(

tIn

T1

exp(nt

E

T2

dT1dT

2+ err

(2)

P(T1, T2) is the unknown distribution function, the terms in brackets constitute the kernel, tIn is the inversion time, tE is the echo time, n is the number of echoes, and err is an un-known error term [3, 11].

RESULTS

Fig. (3) shows T1 and T2 maps of four chosen axial slices of the saturated Kaldenkirchen soil sample. All regions from the soil give measurable and analyzable NMR signals. This means that this type of soil (sandy loam) is investigable by MRI which is not trivial since many soils show much shorter

Bpol

Brlx

Bacq

P90 FID Transm./Acquis.

var. var. Brlx

rf

nEcho

ro

sl

ph

nTi phase-steps

TIn TE

Relaxation in a Natural Soil The Open Magnetic Resonance Journal, 2010, Volume 3 59

transverse relaxation times due to small pore sizes and high content of paramagnetic ions [12]. The soil shows T1 in the range of about 50 ms and T2 of about 3 ms. Some spots with faster (green) and slower T1 relaxation times (violet-white) are also present in the sample, especially near the top. T2 appears more homogeneously. The long T1 values in the top layer are caused by the somewhat smaller local packing den-sity corresponding to larger pores.

Fig. (3). T1 and T2 maps for the Kaldenkirchen soil sample at 7 T

( H= 300 MHz).

From the T1 and T2 maps relaxation time distributions are extracted by calculation of histograms of the populations of relaxation times in the soil, excluding the porous plate layer (Fig. 4). T2 shows a comparably narrow distribution in the range between 1 and 8 ms, with an average of 2 ms. But also voxels with higher T2 (10 to 20 ms) exist, which are located near the edges. This is most probably due to packing faults at the soil-wall interface, where pores are larger. Also T1 in these regions is longer. These effects may be interpreted by the Brownstein-Tarr model [13], which relates 1/T1,2 to the inverse pore size parameter Surface/Volume S/V, see Eq. 3 [10].

Fig. (4). T1 (red) and T2 (blue) histograms of the maps in Fig. (3).

T1 shows a bimodal distribution, with a sharp peak at 7 ms and a broad mode between 10 and 300 ms. The sharp peak is near the shortest inversion time, and therefore must be interpreted with caution. It comprises all longitudinal re-laxation processes faster than 10 ms, which are present, but not further resolved. The main mode is approximately log-normal distributed with T1,mean = 80 ms and a half width of log (T1) = 0.3. T2 depends strongly on tE, so this is hard to compare. However, the reported relaxation times are in the range reported by Hall et al. [12] for several soils, e.g. Korkusova Hut sandy loam which has comparable texture with T2 = 0.9 ms for tE=1ms and t1 = 90 ms at 40% satura-tion. It should be noted that Votrubova et al. [5] found for saturated soil also by statistical analysis of T1 maps an aver-age T1 of 600 ms with a broad distribution over one order of magnitude. Issa et al. calculated T1 relaxation time distribu-tions for rocks in the same range [4]. Also a forest soil [14] with comparable texture shows similar relaxation times (5 and 45 ms at tE = 0.3ms). The T1 relaxation has been further investigated at lower Larmor frequencies between 5 kHz and 20 MHz by the FFC method.

Fig. (5) shows the measured relaxation curves, which are further analyzed by inverse Laplace transformation. The re-sults are presented in Fig. (6). Clearly distinguishable are three processes: A broad main mode with average T1 of 70 ms, which is accompanied by a faster relaxing component at about 10ms. Additionally, a small but very fast component at 2 ms is observed. Note that the limit of the shortest detect-able T1 component in the FFC experiments is about 1 ms. All modes shift simultaneously to faster times with decreasing Larmor frequency.

Fig. (5). T1 relaxation curves at different Larmor frequencies for the

saturated Kaldenkirchen soil sample, (modified, from [10]).

The data have already been analyzed further in terms of pore size distribution [10], where we found that the distribu-tion of T1 between 10 and 200 ms corresponds to pore sizes between one and some tens of microns using the Brown-stein-Tarr model, see below. Here we include results from the high-field measurement into the analysis. The mean re-laxation times for the main mode from FFC and MEMS measurements are plotted in Fig. (7). All data including the result at 300 MHz lie on one line. This indicates that the dis-

7 mm 0 s 0.1 0.2 0.3 0ms 10 20

a) T1 b) T2 32 mm

1 10 100 10000

100

200

0

50

100

Pop

ulat

ion

T2

T1,2

/ ms

Po

pu

latio

n T

1

0 100 200 3000

2

4

6

8

NM

R s

igna

l x10

4 (a.

u.)

(ms)

= 20 MHz

= 5 kHz


persive behaviour follows the same frequency dependence in the whole observed range. The comparably weak dispersion and the linear relation between 1/T1 and the logarithm of the Larmor frequency H indicates that the surface relaxation mechanism is controlled by local 2D diffusion processes at the pore walls [15].

Fig. (7). T1 relaxation at different Larmor frequencies from FFC

(Fig. 6) and MEMS T1 map (Fig. (4)) (modified, from [10]).

Finally, the results are compared to a T1-T2 correlation experiment at H = 24.3 MHz which is the non-spatially re-solved version of the IR-MEMS experiment. The result is shown in Fig. (8). On the right facing axis labelled “ntE” CPMG echoes are plotted, and the second, left facing axis, labelled “tIn”, contains the different tIn values. The diagram should be read as an array of CPMG curves as a function of increasing inversion times tIn, during which T1 relaxation evolves. The data are further analyzed by 2D inverse Laplace

transformation using the program of Y.-Q. Song [3]. Fig. (9) shows bimodal distributions for both T1 and T2 which indi-cates two distinct pore size classes. The low amplitude spots at the limits of the inversion range are due to noise in the experimental data. The maxima for T1 are at 20 and 90 ms and they correspond well to the FFC data for 20 MHz and the IR-MEMS data at 300 MHz (Fig. (7)). The main trans-verse relaxation modes at 9 and 50 ms are much slower than the T2 modes identified at 300 MHz. As the echo time is much shorter at low field and the internal gradients are higher at high field this may be explained by the effect of molecular diffusion at high field.

Fig. (8). T1-T2 correlation experiment on saturated Kaldenkirchen

soil at H = 24.3 MHz.

Fig. (9). T1-T2 correlation diagram of the data in Fig. (8) obtained

by inverse 2D Laplace transformation, Eq. 2.

With decreasing T1 (corresponding to decreasing pore sizes) the 2D distribution function shows an increasing ratio T1/T2 from two to three. These ratios agree well with several data obtained for porous media like rocks [16] and cements [17] and with theoretical considerations [17, 18]. An impor-tant observation is the lack of any detectable cross-peak con-tribution. A cross peak would indicate that for a give value

Fig. (6). T1 relaxation spectra for Kaldenkirchen soil at different

Larmor frequencies, normalized and offsetted (modified, from

[10]).

1 1 0 1 0 0 1 0 0 0

r lx

= 5 k H z

r lx

= 2 0 M H z

D(T

1),

a.u

., s

hift

ed

T1 / m s

0.01 0.1 1 10 1000.00

0.01

0.02

0.03

0.10

0.15

0.20

T1

-1(FFC) long

T1

-1(FFC) short

T1

-1, av.IR-MEMS

T1-1

/m

s-1

/ MHz

nTE (s) TIn (s)

Relaxation in a Natural Soil The Open Magnetic Resonance Journal, 2010, Volume 3 61

of T1 two or more values of T2 exist indicating multiple transverse relaxation mechanisms, so that T2 is affected also by parameters other than pore size.

The general way to transform T1 and T2 distribution func-tions into pore sizes is the scaling by means of the Brown-stein-Tarr equation:

V

S

TTbulk

2,1

,2,12,1

11+= , (3)

incorporating an average surface relaxivity parameter 1,

2, which on its side is obtained from the average T1, or T2. This is obtained from the average surface/volume ratio S/V = 8300 cm

-1 (from BET measurements, see [10]) as 1 =

0.0023 cm s-1

. 2 depends on tE, we obtained 2 = 0.0063 cm s

-1 for the conditions of this investigation. Using these data

the results from the 2D experiment in Fig. (9) can be recalcu-lated into pore size distributions, shown in Fig. (10). The curves agree very well, although that obtained from T2 is somehow broader.

Fig. (10). Pore diameter distributions obtained by projections on the

T1 and T2 axis from the 2D experiment in Fig. (9). The relaxation

times were rescaled to pore size distributions by means of the

Brownstein-Tarr equation, the average S/V ratio from BET meas-

urements [5], and assuming the capillary tube model.

The T2 distribution obtained from imaging experiments is restricted due to tE. A larger fraction of water in small pores relaxes during the first tE period; a rescaling of T2 distribu-tion is meaningless, since it is strongly affected by internal gradients. One only can estimate the pore size limit, below which water is not detectable by MRI with these settings. With T2,av = 2 ms, one obtains 2,300 MHz = 0.06 cm s

-1.

Choosing T2 = tE for the limiting value, one obtains dlimit = 4V/S = 4 2/(1/T2,limit-1/T2,bulk) 4 μm, assuming cylindrical pores. Please note also that this pore size limit corresponds to about the fast mode. A similar estimation for T1 yields

1,300 MHz = 0.0017 cm s-1

corresponding to a lower pore size limit of about 0.5 μm, if one uses a fastest detectable longi-tudinal relaxation time of 0.008s.

CONCLUSIONS

Kaldenkirchen sandy loam shows broad bimodal distri-bution functions of the longitudinal relaxation times T1 over a large range of Larmor frequencies. A further very fast

mode at around 2 ms is present but not well resolved. This short component probably corresponds to the clay compo-nents and represents strongly confined water in this phase. Similar short values have already been found in a different clayey soil [10]. The slow modes of T1 at low magnetic fields correspond to the slow mode observed spatially re-solved at high field (300 MHz). These modes are due to the relaxation in pores with a size in the range between 1 to some tens microns. The two slow modes (10 and 80 ms) are logarithmically dependent on the Larmor-frequency in the low field range and this continues for the slowest mode at high field of 300 MHz. According to Korb [15] this linear dependence on the logarithm of the Larmor frequency is interpreted as locally two-dimensional diffusion, where the basic relaxation mechanism is dipole-dipole interaction of water molecules with paramagnetic centres at the pore walls.

The spatially resolved relaxometric imaging method IR-MEMS generally yields additional information to the re-laxometric measurements, although with inferior signal/noise ratio. This is mainly due to the smaller volumes over which the signal is integrated, i.e. a single voxel rather than the whole sample. The validity of the data is proven by the agreement between the average values from T1 maps with T1 determined relaxometrically. But T1-mapping reveals struc-tures which possess locally higher T1 values in regions near the walls as well as near the top surface, i.e. the local pores are larger. This is due to packing faults near the walls and near the surface which are unavoidable for such natural sam-ples.

The texture of the soil is finer than 2mm, no large grains are present, but aggregates of course. Water may penetrate into these aggregates, and contribute to the relaxation. The usage of 9 mm cuvettes and repacking the soil will definitely change the size of interaggregate pores. However, such pores are mainly macropores. So the macroporous structure of the soil will be changed due to the preparation and sample size. The meso- and microporous structures will be affected much less, so the relaxation time and pore size distributions reflect mainly this meso- and microporous structure, which are most responsible for the water storage of the soil, since the macropores are emptied firstly.

This study shows that T1 mapping is a valuable tool for upcoming investigations of local pore size distributions in natural, i.e. undisturbed, soil cores for e.g. detection of pref-erential flow paths.

ACKNOWLEDGEMENTS

We thank N. Hermes, ICG-4, Research Centre Jülich for technical assistance and the DFG (SFB Transregional Col-laborative Research Centre 32, PO-746-2/1 and Sta 511-4/1) for financial support.

REFERENCES

[1] Dunn KJ, Bergmann DJ, Latorraca GA. Nuclear magnetic reso-

nance, petrophysical and logging applications. Pergamon: Amsterdam 2002.

[2] Kimmich R, Anoardo E. Field-cycling NMR relaxometry. Prog Nucl Magn Res Spectrosc 2004; 44: 257-320.

[3] Song Y-Q, Venkataramanan L, Hürlimann MD, Flaum M, Frulla P, Straley C. T1-T2 correlation spectra obtained using a fast two-

dimensional laplace inversion. J Magn Reson 2002; 154: 261-8.


[4] Issa B, Mansfield P. Permeability estimation from T1 mapping.

Magn Reson Imaging 1994; 12: 213-4. [5] Votrubova J, Cislerova M, Amin MHG, Hall LD. Recurrent

ponded infiltration into structured soil: a magnetic resonance imag-ing study. Water Resour Res 2003; 39: 1371.

[6] Edzes HT, van Dusschoten D, Van As H. Quantitative T-2 imaging of plant tissues by means of multi-echo MRI microscopy. Magn

Reson Imaging 1998; 16: 185-96. [7] Belliveau SM, Henselwood TL, Langford CH. Soil wetting proc-

esses studied by magnetic resonance imaging: correlated study of contaminant uptake. Environ Sci Technol 2000; 34: 2439-45.

[8] Cislerova M, Votrubova J, Vogel T, Amin MHG, Hall LD. Mag-netic resonance imaging and preferential flow in soils. In: Van

Genuchten MT, Leij FJ, Wu L, Eds. Characterization and meas-urement of the hydraulic properties of unsaturated porous media;

U.S. Salinity Laboratory: Riverside CA 1997; pp. 397-411. [9] Davies S, Hartwick H, Roberts D, Spowich K, Packer KJ. Quanti-

fication of oil and water in preserved rock by NMR spectroscopy and imaging. Magn Reson Imaging 1994; 12: 349-53.

[10] Pohlmeier A, Haber-Pohlmeier S, Stapf S. A fast field cycling NMR relaxometry study of natural soils. Vadose Zone J 2009; 8:

735-42. [11] Provencher SW. A constrained regularization method for inverting

data represented by linear algebraic or integral-equations. Comput Phys Commun 1982; 27: 213-27.

[12] Hall LD, Amin MHG, Dougherty E, et al. MR properties of water

in saturated soils and resulting loss of MRI signal in water content detection at 2 tesla. Geoderma 1997; 80: 431-48.

[13] Barrie PJ. Characterization of porous media using NMR methods. Ann Rep NMR Spectrosc 2000; 41: 265-316.

[14] Schaumann GE, Hobley E, Hurrass J, Rotard W. H-NMR re-laxometry to monitor wetting and swelling kinetics in high-organic

matter soils. Plant Soil 2005; 275: 1-20. [15] Korb JP, Bryant R. Magnetic relaxation dispersion in porous and

dynamically heterogeneous materials. Adv Inorg Chem 2005; 57: 293-326.

[16] Schoenfelder W, Glaeser HR, Mitreiter I, Stallmach F. Two-dimensional NMR relaxometry study of pore space characteristics

of carbonate rocks from a Permian aquifer. J Appl Geophys 2008; 65: 21-9.

[17] McDonald PJ, Korb JP, Mitchell J, Monteilhet L. Surface relaxa-tion an chemical exchange in hydrating cement pastes: a two di-

mensional NMR relaxation study. Phys Rev E Stat Nonlin Soft Matter Phys 2005; 72: 011409.

[18] Kleinberg RL, Farooqui SA, Horsefield MA. T1/T2 ration and frequency dependence of NMR relaxation in porous sedimentary

rocks. J Colloid Interface Sci 1994; 158: 195-8.

Received: August 17, 2009 Revised: December 11, 2009 Accepted: December 11, 2009

© Haber-Pohlmeier et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

Low-Field NMR of Water in Model Soils

Oscar Sucre*,1, Federico Casanova

1, Andreas Pohlmeier

2 and Bernhard Bluemich

1

1ITMC RWTH-Aachen University, Aachen, D-52056, Germany

2Forschungszentrum Jülich, Jülich, D-52425, Germany

Abstract: The evolution of water contained in soils is a physical phenomenon of importance in soil science and climatol-

ogy. This work presents preliminary results from the use of mobile NMR to measure moisture in soil columns. To demon-

strate the ability of the NMR technique to follow the drainage process of water in model soils, moisture measurements

were performed at a certain depth with a mobile NMR sensor during an one-step outflow experiment. The NMR sensor

exhibits a cylindrical geometry incorporating the principle of the u-shaped NMR-MOUSE. It could be raised and lowered

inside a plastic tube in the soil column similar to a wire-line logging tool. Working with a frequency of 8.8 MHz, the sen-

sitive volume lies 3 mm deep into the soil and 2 mm away from the tube walls, making the technique truly non-invasive.

For the purpose of quantitative analysis, the temporal evolution of moisture (described by the Richards Equation) was

modelled with the Hydrus1D Program. Out of these simulations, an assessment of the hydraulic parameters (Ks, , n of

the Van Genuchten model) of the soil was achieved.

Keywords: Low field NMR, moisture sensor, model soils, drainage, richards equation.

1. INTRODUCTION

The question of how water moves in soils is of impor-tance in agriculture and climatological sciences.

Being a partially saturated porous medium, soil in the vadose zone can either stop or permit the transport of water depending on gravity, surface tension and evaporation. Prob-ing of moisture on field plays a central role in addressing these issues, and several techniques have gained acceptance in the soil science community to perform that measurement: Time Domain Reflectometry (TDR) [1] and Neutron Absor-tion [2] among others.

The application of Nuclear Magnetic Resonance (NMR) in soils is not new, as it has been applied to gain information about the pore size distribution [3, 4], in the search of aqui-fers [5] or to study the chemistry of soil organic matter [6, 7]. However as far as we know, little has been done to de-velop an NMR instrument that can be deployed on field to measure local values of soil moisture, taking advantage of the non-destructive nature this technique possesses. In this sense, NMR at low fields has just begun to make its way among the above mentioned techniques. It can even show some comparative advantages over TDR: As the signal to probe water comes from the atomic nuclei (in case of water, simply protons), it is not affected by any ions in solution, so the NMR technique happens to be immune to any salt con-tent in water absorbed in soil. Furthermore, due to the reso-nance condition and the inhomogeneous magnetic fields generated the sensor senses water in a very well defined vol-ume [8], making the measure well resolved spatially. For those reasons we believe the NMR can become standard to

*Address correspondence to this author at the ITMC RWTH-Aachen

University, Aachen, D-52056, Germany; Tel: 00-49-241-8026423;


sense moisture among soil scientist. It has even reached a mature stage in its development as a logging tool for the oil industry [9, 10]. Undeniably NMR is no excepted from ex-perimental diffculties: low inherent signal-to-noise ratio con-stitutes its main drawback. Common way to improve this ratio is to average over repeatedly acquired data, however that can make the whole profiling of soil moisture costly in terms of time.

This article reports progress in probing soil moisture by low field NMR. The work was performed as a part of the interdisciplinary project Transregio 32 [11] (funded by the German Research Council DFG): the drainage processes in a first column filled with FH31 model soil and in a second column with W3 model soils under known boundary condi-tions were investigated with a newly developed low-field NMR sensor: the slim-line NMR logging tool. The moisture profile before and after drainage, and the moisture loss at certain depth during drainage were thus recorded with this instrument. To validate the results, the associated drainage problem is solved numerically, producing a theoretical curve that can be compared to the experimental results. By chang-ing soil hydraulical parameters in the numerical solution the better correspondence between theory and experiment is found and the so obtained values are compared to reported in the literature.

2. EXPERIMENT AND MATERIAL

The experimental setup is shown in Fig. (1) and consists of two concentrical tubes, being the space between them filled with the model soil to research. This arrangement or soil "column" stays in a water tank. A copper fabric was wound around the outer tube to shield it from electromag-netic noise. The central inner column has a diameter of 31 mm and walls 1 mm thick. The NMR sensor moves within the inner tube. Initially the model soil is completely saturated by raising the water level in the tank up to top of the column.

64 The Open Magnetic Resonance Journal, 2010, Volume 3 Sucre et al.

The top of the column is then closed with a rubber seal that allows the entrance of air (necessary for pressure balance during drainage) and limits evaporation as well. The bottom plate is connected to the water tank through perforations 4 mm wide and covered by a sheet of filter paper. This lets the water drain out from the column and gives hold to model soil lying thereupon. To obtain a complete profile of the satura-tion along the column the sensor is displaced with a stepper motor from the bottom to the top while NMR signal is ac-quired.

Fig. (1). Setup for soil drainage experiments.

Two experiments of drainage were performed, one on

coarse sand FH31 (distribution of grain size classes 2% (>0.72mm), 8% (0.71-0.5mm), 30% (0.5-0.355mm), 41% (0.36-0.25mm), 16% (0.25-0.18mm), 3% (<0.18mm)) and the other on fine sand W3 (distribution of grain size classes 7% (0.6-0.2 mm),60% (0.2-0.063 mm), 20% (0.063-0.02 mm), 6% (0.02-0.0063 mm), 3(0.0063-0.0002 mm), 2% (< 0.002mm)). Both obtained in Quartzwerke Frechen, Ger-many. Relevant parameters in these experiments are the cor-responding column height, measurement time and the sample porosity S. They are summarized in Table 1.

Table 1 Sample Porosities and Experimental Parameters

Sample Height (cm) Time (min) S

FH31 67 1372 0.36

W3 59 14400 0.36

After having reached complete saturation, the columns

were let to drain by allowing the water to flow out of the tank. In the special case of FH31, the initial fall of the water level in the tank was recorded in time (Fig. 2). This is be-cause the duration of the drainage process in FH31 (about 90 min) is comparable to time the tanks needs to release all wa-

ter (15 min) setting a boundary condition that must be taken into account for modelling. Afterwards the water is kept at the same level during the rest of the experimental time. Due to the fast drainage in FH31, moisture profiles were acquired before and after the drainage only.

Fig. (2). Pressure head at bottom outside the column.

Another important aspect in the experiment for FH31 is

the drainage dynamics. Since the drainage occurs so fast in comparison to the time needed to perform a sole moisture profile, we decided to follow it by maintaining the sensor at a depth of -22.1 cms during the first 150 min of experimental time. At this depth, following previous moisture profiles, we are certain of sensing a considerable drop in moisture. Al-though this corresponds to read the water content at a sole point, the data so obtained contains enough information to characterize the temporal behavior of the drainage com-pletely. Here the number of NMR acquisitions per moisture measurements (scans) plays a critical role: It must be high enough to guarantee a small uncertainty (say 7%) but it should be made in a time suficiently short to follow reliably the temporal change in water saturation. In our case, given the NMR longitudinal relaxation time of water in our sand (T1=800 ms) and a signal-to-noise ratio per echo and per scan of 0.17 (in FH31), 64 scans with 128 echoes were enough to achieve the aimed uncertainty (7%) at each point. In the case of W3 a moisture profile of the water content was measured each two days.

3. INSTRUMENT AND DATA PROCESSING

Initially perceived as a potential tool for medical research

[14], the NMR slim-line logging tool has been built with a

cylindrical shape. It is an inside-out NMR instrument similar

to some others found in the literature [12]. Fig. (3) shows a

picture of the sensor. Generally speaking, conditions to ob-

serve the nuclear resonance are a constant magnetic field B0

and a radiofrequency (rf) field B1 . The latter is generated by

a coil and the former by a magnet body, establishing together

a sensitive volume about 3 mm above the coil (Fig. 4). A

sequence of pulses is applied in the rf field, exciting there-

fore the spins in resonance. Since the inner tube where the

sensor is displaced is only 1 mm thick, the sensitive volume

is 2 mm well inside the soil, avoiding disturbance in the

Low-Field NMR of Water in Model Soils The Open Magnetic Resonance Journal, 2010, Volume 3 65

measurement due to any possible overlapping between the

walls and the sensitive volume. This depth of the sensitive

volume was experimentally checked by measuring the NMR

signal of rubber layers of different thickness. The main body

of the sensor consists of two cylindrical magnets 8 cm long

each (l in the Fig. 4) and 3 cm in diameter (D) and with a

magnetization M pointed as depicted, separated by a gap g

of 4 mm. Just above the gap lies the rf coil (c). The sensitive

volume is around 30 mm3, being its position and thickness

determined by profiles of the magnetic fields B0 and B1 ,

the operation frequency (8.8 MHz), the local value of the

magnetic gradient (30 T/m) and the pulse length used

(6 μsec) in the pulse sequence [8]. The coil has a dead time

of 20 μsec approximately. The NMR spectrometer used was

a Minispec Serie mq from Bruker.

In low field applications, the NMR signal is usually ac-

quired with a Carr-Purcell-Meilboom-Gill (CPMG) [12]

pulse sequence, with pulses separated by a echo time E. The

succession of NMR echoes, sequentially acquired, compose

altogether a decaying signal with a effective transverse re-

laxation time (here simply denoted by T2,eff ). Its decay under

highly inhomogeneous magnetic fields is governed by trans-

lational diffusion of molecules, interactions among nuclear

spins (the proper transversal relaxation) and interactions be-

tween the surface of the solid matrix and the nuclear spins

[13]. In spite of this variety of relaxation mechanisms, aver-

aging over a number of echoes (NE) still yields a figure di-

rectly proportional to water content, provided the time span

that corresponds to these echoes ( ENE) is small compared to

T2,eff . This can be easily shown by describing the succession

of echoes with these following formulae, being j an index

that labels the echo number (between the 1 and NE), n(j) the

noise at the echo j and A1 the amplitude of the initial echo:

Fig. (4). Main parts of the slim-line NMR logging tool. M labels the

magnetization vector.

S( j) = A1e( j 1) E

T 2, eff+ n(J )

If we average over the first NE echoes and apply well known mathematic formulae, we obtain < S >:

< S > =AO

NE

e(NE) E

T 2, eff1

eE

T 2, eff1

+ < n > NE

As long as NE E < 0*,*3T2,eff, we can approximate the ex-ponential term to a linear function (e

-x 1 - x) with an uncer-

tainty of 5%. Easy manipulations will lead to the following equation:

< S >= Ao+ < n >NE

So it is proven that averaging over the first NE echoes has no further effect on the obtained data than averaging the ran-dom noise. This way of data processing has the advantage that for a complete signal of NE echoes, we diminish the noise amplitude to that we had if the experiment were re-peated NE times. When acquiring noisy signals, this proce-dure is of not little advantage.

To give an example of this statement with real numbers, consider the Fig. (5) where data for water in saturated sand is depicted as it is supplied by the spectrometer. Experimental parameters in this decay are 256 scans, an echo time of 50 μs

Fig. (3). Photo of the slim-line NMR soil sensor.

Fig. (5). NMR signal acquired in saturated sand FH31. T2,eff = 28

ms.


and 3000 echoes. With a exponential function fitted to this whole data, an effective T2,eff of 28 ms can be easily ob-tained. The inset of this figure shows the first 128 echoes, corresponding to a time of 6.4 ms, a time lower than the boundary of 30% set by the demonstration given above. So we can confidently average this first 128 echoes to obtain a relative measure of the partial saturation in sand. With prior knowledge of the signal level for completely saturated sand, the partial saturation in soil can then be calculated by a linear interpolation. This was the way of processing data for all the moisture profiles presented here.

4. RESULTS

The evolution of water saturation as a function of depth is shown in the Figs. (6 and 8) for the sand and the silt col-umns, respectively. A fact easy to recognize is the big differ-ence in time evolution: while the FH31 has reached its equi-librium state in 1.5 hours, W3 does not show any important difference in saturation during 10 days.

Fig. (6). Moisture profle in a column of model soil FH31.

Height=67 cm.

Four profiles were measured in FH31, one prior to drain-

age (t = 0 min, Fig. 6) and three thereafter (t = 182, 891, 1372 min). Experimental parameters for these measurements were: Echo time: 50 μsec, 256 scans, 3 sec of recycling de-lay and 128 echoes. The similarity of these three last profiles reveals that the equilibrium state of the column had been reached already at the time of the first measurement (182 min). As is well known in soil physics, the profile in this state corresponds to the retention curve (h) [16]. We used therefore the profile at t = 891 min for fitting of the retention curve.

For the dynamical measurement, done by leaving the sensor measuring at certain depth during drainage as it was explained before, the results are depicted as open symbols the Fig. (7). Experimental parameters were the same as for the moisture profiles with the sole exception of the scans: 64 instead of 256, for reasons already explained. The continu-ous curve correspond to the numerical solution for perme-ability at complete saturation Ks.

The situation with the model soil W3 was completely dif-ferent. It retained water during 10 days with no measurable

change. Experimental parameters were the same than for FH31, although the used NE for averaging was 64 instead of 128, as the T2,eff for this soil was 20 ms. Given the height of the column the retention curve of this soil cannot be estab-lished (Fig. 8). However, a slight drop of 5% from the value at the bottom can be observed along the profile, correspond-ing probably to the water loss in the biggest pores.

Fig. (8). Moisture profile in a column of model soil W3. Height=59

cm.

5. MODELING WITH RICHARDS EQUATION FOR

THE FH31 COLUMN

To validate the measurements obtained with this new method to probe soil moisture, we use the well known Rich-ards equation (eq. 1) [15, 16]. It establishes the theoretical framework to model the evolution of moisture profile in the sand column, where h is the hydraulic head, the partial saturation, and Ks the permeability of the soil at complete saturation. After starting the drainage process the level of the water mirror that initially was at the top of the column re-cedes completely to the bottom in about 15 min, setting a variable pressure head (Fig. 2) at the bottom. This data was

Fig. (7). Moisture drop at depth -22.1 cm during 150 min for FH31.

In color, numerical solution for Ks=0.62 cm/sec.

Low-Field NMR of Water in Model Soils The Open Magnetic Resonance Journal, 2010, Volume 3 67

taken as lower boundary condition for a numerical solution of eq. 1.

t=

xK( )

h

x+1 (1)

To undertake the simulation of the drainage problem, we

use the Van Genuchten Model [17] for the retention curve

(eq. 3) (where and n are the Van Genuchten parameters)

and the permeability in dependency of the relative saturation

(4), having Ks the meaning explained in the previous section.

Parameter m equals 11

n and l=0.5 for most soils [15]. With

the relative saturation Se defined by eq. 2 in terms of the real

saturation , the porosity s and the residual saturation r; we

obtain a complete set of equations to solve eq. 1. The

HYDRUS code [18] supplies the computational tool to per-

form 1D simulations to model the illustrated drainage proc-

ess.

Several forward simulations were carried out to find the

values of the Van Genuchten parameters and n that would

reproduce the measured data. The best fit was corresponded

to = 0.0268 cm-1

and n = 9.0. The numerical profile and the

experimental data are depicted in Fig. (9).

Se =s r

(2)

Se =1

[1+ ( h)n ]m (3)

K(Se) = KsSel 1 1 Se

1

m

m 2

(4)

Fig. (9). Moisture profile in FH 31 at t=891 min and theoretical

profile using the Van Genuchten model. = 0.0268 /cm-1

and n =

9.0.

With these gained values for and n, further simulations

were performed with several values of the permeability at full saturation Ks, using the temporal change in partial satu-ration measured at a depth of -22.1 cm for comparison (Fig.

7). Best fit among the numerical and the experimental curve was found for Ks = 0.62 cm/min.

CONCLUSION AND OUTLOOK

The instrument has shown its capability of following the fast drainage process in model soil (FH31). The so obtained data can be analyzed to extract hydraulical parameters of the model soil. However, experimental parameter like repetition time (recycling delay), number of scans and number of ech-oes must be carefully selected to follow the drainage process successfully. One interesting remark can be made when ex-amining Fig. (7): There is some small delay in the moisture drop that the simulation does not reproduce, perhaps due to the hysteresis behavior in the retention curve of sand.

What information can be won of this information is still an open question, however we can see that small features in the temporal evolution of moisture can be followed with this instrument. About the Van Genuchten parameters obtained for FH31, that Kastelanek reports for a sand fraction with a grain sizes between 150 and 300 μm a value of n= 8.35 [19]. The obtained permeability corresponds to 80 % of the value previously reported [20], measured with Darcy's Law. The values agree satisfactorily with literature data for a fine sand of the textural composition like ours, and thus prove the va-lidity of the NMR sensor.

Work in progress concerns the construction of a sensor with a larger diameter of 48 mm that allows to measure 6 mm inside the soil at a frequency of 12 MHz. Our short term goal is to reach a signal to noise ratio per scan and per echo of 18, similar of the NMR-MOUSE [12] at the same depth. This would in principle enable us to measure moisture pro-files in shorter time, to avoid any surface effects due to wall proximity and to essay external noise suppression tech-niques, opening the way to a robust instrument which could be deployed outdoors.

ACKNOWLEDGEMENTS

One of the authors (Sucre) thanks the German Academic Exchange Service (DAAD) for his Ph. D. grant. Support by the Deutsche Forschungsgemeinschaft through research pro-ject TR32 is also grate fully acknowledged.

REFERENCES

[1] Robinson DA, Jones SB, Wraith JM, Or J, Friedman FP. A review

of advances in dielectric and electrical conductivity measurement

in soils using time domain reflectometry. Vadose Zone J 2003; 2:

444-75.

[2] Strebel DE, Landis DR, Huemmrich KF, Meeson BW, Eds. Col-

lected data of the first ISLSCP field experiment, NASA Goddard

Space Flight Center: Greenbelt, Maryland, USA 1988; Vol. 1.

[3] Hinedi ZR, Kabala ZJ, Skaggs TH, Borchardt DB, Lee RW and

Chang AC. Probing soil and aquifer material porosity with Nuclear

Magnetic Resonance. Wat. Res. Res 1993; 29: 3861-6.

[4] Pohlmeier A, Haber-Pohlmeier S, Stapf S. A fast field cycling

nuclear magnetic resonance relaxometry study of natural soils. Va-

dose Zone J 2009; 8: 735-42.

[5] Meju M, Denton P, Fenning P. Surface NMR sounding and inver-

sion to detect ground-water in key aquifers in England: compari-

sons with VESTEM methods. J Appl Geophys 2002; 50: 95-111.

[6] Conte P, Spaccini R, Piccolo A. O-alkylation of a lignite humic

acid by phase-transfer catalysis. Anal Bioanal Chem 2006; 386:

382-90.


[7] Jaeger F, Grohmann E, Schaumann GE. 1H NMR Relaxometry in

natural humous soil samples: insights in microbial effects on re-

laxation time distributions. Plant Soil 2006; 280(1-2): 209-22.

[8] Balibanu F, Hailu K, Eymael R, Demco D, Bluemich B. Nuclear

magnetic resonance in inhomogeneous magnetic fields. J Magn

Reson 2000; 145: 246-58.

[9] Sezginer A, Goswami J, Luong B. On the design of NMR sensor

for well-logging applications. IEEE Trans Antennas Propag 2000;

48: 1393-403.

[10] Kleinberg R. Well logging. In: Grant DM, Harris RK, Eds. Ency-

clopedia of nuclear magnetic resonance. Wiley: New York 1996;

Vol. 8: p. 4960.

[11] Interdisciplinary Collaborative Research Center TR32. Patterns in

the soil-vegetation-atmospheric system: monitoring, modeling, data

assimilation. Funded by the Deutsche Forschungsgemeinschaft,

DFG. 21, Jan, 2010, Available from: www.tr32.de

[12] Bluemich B, Perlo J, Casanova F. Mobile single-sided NMR. Prog

Nucl Magn Reson Spectrosc 2008; 52: 197-269.

[13] Huerlimann M, Grffin D. Spin dynamics of carr-purcell-meiboom-

gill-like sequences in grossly inhomogeneous Bo and B1 fields and

application to NMR well logging. J Magn Reson 2000; 143: 120-

35.

[14] Maule. J. Diploma Thesis. RWTH-Aachen. Aachen: Germany

2005.

[15] Bear J. Dynamics of fluids in porous media. New York: Dover

1988.

[16] Roth K. Soil Physics. Lecture notes. Institute of environmental

physics. University of Heidelberg: Heidelberg 2007.

[17] Van Genuchten M. A closed-form equation for predicting the hy-

draulic conductivity of unsaturated soils. Soil Sci Soc Am J 1980;

44: 892-8.

[18] Simunek J, Sejna M, van-Genuchten M. The HYDRUS-1D soft-

ware package for simulating the one-dimensional movement of wa-

ter, heat, and multiple solutes in variably saturated media, version

2.0, U.S. Salinity Laboratory, Agricultural Research Service, Riv-

erside, California USA 2008.

[19] Kastelanek F. Numerical simulation technique for vertical drain-

age. J Hydrol 1971; 14: 213-32.

[20] Pohlmeier A, van Dusschoten D, Weihermueller L, Schnurr U,

Vereecken H. Imaging water fuxes in porous media by magnetic

resonance imaging using D2O as a tracer. Magn Reson Imaging

2008; 27: 285-92.

Received: July 15, 2009 Revised: December 18, 2009 Accepted: December 19, 2009

© Sucre et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

MRI in Soils: Determination of Water Content Changes Due to Root Water Uptake by Means of a Multi-Slice-Multi-Echo Sequence (MSME)

A. Pohlmeier*,1, F. Vergeldt

2, E. Gerkema

2, H. Van As

2, D. Van Dusschoten

3 and H. Vereecken

1

1ICG-4, Research Centre Jülich, Germany

2Department Biophysics, Wageningen University, The Netherlands

3ICG-3, Research Centre Jülich, Germany

Abstract: Root water uptake by ricinus communis (castor bean) in fine sand was investigated using MRI with multiecho

sampling. Before starting the experiments the plants germinated and grew for 3 weeks in a cylindrical container with a di-

ameter of 9 cm. Immediately before the MRI experiments started, the containers were water-saturated and sealed, so water

content changes were only caused by root water uptake. In continuation of a preceding work, where we applied SPRITE

we tested a multi-echo multi-slice sequence (MSME). In this approach, the water content was imaged by setting TE = 6.76

ms and nE = 128 with an isotropic resolution of 3.1mm. We calculated the water content maps by biexponential fitting of

the multi-slice echo train data and normalisation on reference cuvettes filled with glass beads and 1 mM NiCl2 solution.

The water content determination was validated by comparing to mean gravimetric water content measurements. By co-

registration with the root architecture, visualised by a 3D fast spin echo sequence (RARE), we conclude that the largest

water content changes occurred in the neighbourhood of the roots and in the upper layers of the soil.

Keywords: MRI, root water uptake, root system, relaxometric imaging.

INTRODUCTION

The interplay between root activity and soil functions is not yet well understood on a local scale, although it belongs to the most important processes controlling plant growth and productivity. One reason is the limited observability under natural conditions, since roots are the “hidden half” of the entire plant [1]. Recently, progress has been made by the development of numerical modelling based on soil physical principles, where water motion is calculated in a coupled soil-plant network of water potential differences, hydraulic conductivities and root system architecture [2-4]. The suc-cess of such integrated modelling approaches requires ex-perimental information about the root system architecture as well as water content distributions under controlled bound-ary conditions in experimental scenarios. Here, non-invasive 3D tomographic methods are mandatory, since only these can give an undisturbed look into the soil and the processes taking place there. Today, three main techniques based on different physical principles are generally usable, X-ray [5], neutron [6] and nuclear magnetic resonance tomography (MRI) [7-10]. The potentials and limits of these methods for investigating root-soil interactions are a present challenging topic in soil and plant sciences.

The focus of X-ray CT is on the physical density with high resolution (some microns), and so this method is highly sensitive for imaging of the solid phase. On the other hand the differentiation between water and air turns out to be

*Address correspondence to this author at the Research Centre Jülich, ICG-

4 (agrosphere institute), D-52425 Jülich, Germany; Tel: (49)2461-612795;

Fax: (49)2461-6125; E-mail: [email protected]

tricky. In contrast, MRI is principally capable to image both

root system architecture and water content, since its sensitiv-

ity is based on the presence of water and on the properties of

the medium, in which it is located. Also MRI is capable to

image fluxes or water content changes on different time scales.

The present work is a continuation of our previous stud-

ies on the use of MRI for the investigation of root water up-

take [10, 11]. There we used a single point imaging tech-

nique, since this is especially convenient for matrices with

short transversal relaxation times as soils at low water con-

tent are [12]. However, single point imaging methods are

time consuming since the resolution scales with the cube of the repetition time.

In this work we want to elucidate further the potential of

relaxometric imaging for the water content determination

[13]. The major concern for such methods is that with de-

creasing water content the relaxation times decrease strongly

according to the Brownstein-Tarr model [14, 15]. So if one

just uses conventional spin-echo imaging the echo intensity

reflects not only the water content but also the local relaxiv-

ity, and it is not necessarily proportional to the water content.

As model system we have chosen ricinus communis carmen-

cita (castor bean) grown in medium sand. The scenario was

initial saturation and subsequent increasing drought stress

over two weeks by sealing the container and allowing tran-

spiration only via the leaves. Water contents were mapped

by relaxometric imaging using a multi-slice- multi-echo

method (MSME). Additionally, the root system architecture

was imaged by a 3D fast spin echo method (RARE) with high resolution.

70 The Open Magnetic Resonance Journal, 2010, Volume 3 Pohlmeier et al.

MATERIAL AND METHODS

Plant Setup

10cm high cylindrical Perspex containers with 8 cm in-ternal diameter were used for all experiments. Each was equipped with two 0.8 x 15 cm cylindrical marker tubes filled with glass beads and saturated with 5 mM NiCl2 solu-tion as internal water content standard ( =0.39) and for coregistration of the MRI images recorded at different dates. The containers were filled about 10 cm high with medium sand (0.2-0.7mm) and a 2cm top layer of peat. Ricinus seeds have been planted in the peat layer, see Fig. (1). After germi-nation they grew for three weeks. Then the container was saturated slowly from bottom and sealed on both sides so that transpiration proceeded only via the leaves. Three indi-vidual plants have been grown, of which one is chosen here for further evaluation of the data.

Fig. (1). Ricinus plant in the container immediately before measur-

ing.

Fig. (2) shows the gravimetrically determined water con-tent changes with time after saturation and sealing on day-0. The small vertical arrows indicate the dates of the MRI experiments.

Fig. (2). Gravimetric total water content in the container after satu-

ration and sealing. The arrows indicate the dates of the MRI ex-

periments.

MRI Hardware & Protocols

All MRI experiments have been performed at the Wageningen 3T scanner (127.9 MHz, Gmax=0.27 T/m), operated by a Bruker console. The root architecture was determined with high spatial resolution using a 3D-RARE pulse sequence with a turbo factor of 16. TE was 6.7 ms for the first echo and 4.3ms for the subsequent 15 echos, TR = 1.5s. The Voxel size was isotropic 0.78mm. Quantitative water content mapping was performed by a multislice- multiecho sequence (MSME)[13]. The simplified timing diagram is displayed in Fig. (3). In contrast to conventional single-echo pulse sequences it monitors a sequence of many echoes nE for each voxel, addressed by the slice selective initial 90° pulse, the phase encoding (pe) and the read-out (ro) gradients. Before further analysis the data, which have been recorded in the complex space, were phase corrected so that the information was only in the real part, and the imaginary part was ignored. This is especially important for biexponential fitting [16]. The signal intensity for biexponential decay due to transversal relaxation under the condition for sufficiently long repetition time is given by Eq. 1.

Fig. (3). simplified MSME imaging timing diagram. TE is echo

time, nE the number of echoes, rf, ro, sl, and pe refer to rf-channel,

read-out, phase-encoding, and slice selection gradients, respec-

tively. Further details see text.

+

s

EEs

f

EEfEE

T

TnS

T

TnSTnS

,2

0

,2

0 expexp)( (1)

S0 = S0,f + S0,s (2)

nE and TE are the number of echoes and echo time, respec-tively. S0,f , S0,s, T2,f, and T2,s are the amplitudes and transver-sal relaxation times of the fast and slow processes, respec-tively. So, by extrapolation the quantity S0 = S0,f + S0,s is ob-tained, which is proportional to the spin density and the wa-ter content in individual voxels. We used the following im-aging parameters: matrix size 32x32 and 3 mm slice thick-ness (10x10x9.6cm). Slices were addressed in interleaved

0 2 4 6 8 10 12 140.0

0.2

0.4

time (days)

MRI experiments

npe

rf

nE

ro

sl

pe

TE

MRI in Soils The Open Magnetic Resonance Journal, 2010, Volume 3 71

mode. TE = 6.76ms, nE = 128, TR = 30s. The water content maps are obtained by calibration of the 3D amplitude maps on S0 of the voxels of the calibration tubes. All data evalua-tion steps were performed by the IDL programming system.


Fig. (4) shows some exemplary horizontal and vertical slices through the ricinus container. Plotted are the intensity of the first echo and relaxation curves of selected voxels in-dicated by the yellow arrows. Although the resolution is lim-ited to 3 mm, we observe structural details in the container. In Fig. (4a) the yellow arrow points to the calibration tube. In Figs. (4b) to (4e) roots are clearly resolved, as well as the peat layer in the upper part of the container, which is much more heterogeneous than the sand packing below. Also the signal intensity is much higher. In Fig. (4d) a root is indi-cated and in Fig. (4e) the arrow points to the onset of the shoot inside the peat layer. In the lower parts of the subfig-ures relaxation curves in selected voxels are shown for a) in the marker, b) and c) in bulk soil on day-1 and day-6 after saturation, d) is from a root voxel and e) from a shoot voxel.

One should take into account that the voxel size is greater than the dimension of the roots, so that the intensity is com-posed of both signals from root and the immediately sur-rounding soil. This partial volume effect is responsible for the lower signal intensity in the root-voxel compared to the shoot voxel, since the shoot is much thicker than the roots.

In most voxels bimodal transverse relaxation was ob-served, so Eq. 1 was fitted to all voxels. From that we ob-tained amplitude, T2,f and T2,s maps. This is exemplarily shown for 30 vertical slices in Fig. (5) for the plant on day-1. The amplitudes are quite uniform over the sand packing, with exception of the slices 2 and 3 from bottom. Clearly distinguishable is the peat layer in the upper 2 cm. It is char-acterized by high amplitudes reflecting high water contents and by short fast transverse relaxation times T2,f, which are in the range between 3 and 8 ms, whereas the slow relaxation time T2,s is about 80ms and comparable to the range of the sand packing. The second feature is the inner and outer marker tubes in images 16 to 18 and 26 to 30, respectively, characterized by moderate amplitudes and long relaxation times (T2,f 20ms, T2,f 200ms). Also voxels including roots

Fig. (4). Exemplary MSME-MRI images of selected slices of the ricinus container. a) axial slice near bottom at day-1, b) and c) axial slices

near top on day-1 and day-6, resp. d) and e) vertical slices showing roots and shoots. The arrows indicate individual voxels for which the T2

decay curves are shown in the lower parts of the subfigures.

a) b) c) d) e)


are visible in the amplitude maps of the central vertical slices, whereas they are not clearly identifiable in the relaxa-tion time maps. However, it should be noted that in the areas with highest root densities, e.g. slices 17 and 18, both fast and slow relaxation times are faster (T2,f 8ms, T2,f 50ms), than in the outer regions of the container (T2,f 14ms, T2,f 80ms). The nature of the fast and slow relaxation time is not yet clear. Assuming that the Brownstein-Tarr model is valid, transverse relaxation is accelerated in small pores due to frequent wall contacts of diffusing water molecules in the neighbourhood of paramagnetic centres. Second, transverse relaxation is accelerated by diffusion in internal gradients [14]. This might explain the frequently observed bimodal relaxation in natural porous media, but this requires more detailed investigations, which exceed the frame of this work.

The central aim of this work is the determination of water content changes due to root system activity. As the next step we have calculated the water content (x,y,z) by calibration of the total amplitude maps on the marker tubes with = 0.39. This evaluation procedure is validated by comparing the average gravimetric water contents of the plant container with the MRI determined water content.

Fig. (6) shows that MRI-water content profiles, which have been obtained by integration over horizontal agree well with the average gravimetric water contents. This proves the validity of the entire measurement and evaluation procedure. Examples of the water content map for slice 19 on day-1 and day-6 are shown in Fig. (7). One can clearly recognize the roots and the peat layer on the top. The soil appears quite homogeneous with some enrichment of water near the bot-tom. On day-6 the whole system is much drier, the average water content in the sand is about = 0.05. The top peat layer is also much drier, so that the onset of the shoot be-comes visible. A striking point is that the voxels containing roots also appear much drier (blue colour vs. red colour on day-1), although one might expect that the roots do not loose water in significant amounts unless the plant is wilted. So they should appear white just like the shoot. The reason for this is that the voxel size is with 3 mm significantly larger

than the root diameter, so the lower intensity is due to a par-tial volume effect; one observes average water content com-posed of root and surrounding soil.

Fig. (6). Vertical water content profiles of the ricinus-root soil sys-

tem, calculated by integration over horizontal slices from the water

content maps.

For a comparison of the root system with the water con-tent maps we have also recorded 3D images of only the roots by MRI with a 3D RARE sequence. This is presented in Fig. (8), where voxels with intensities above a threshold of 25% of the maximum are enclosed by an isosurface. The peat layer is cut off.

Fig. (5). Total amplitude S0, T2,f and T2,s maps obtained for the ricinus container on day-1 by fitting of Eq. 1 to the relaxation curve in all

voxels. Shown are 30 from 32 vertical slices, FOV is 10x10cm, matrix size 32x32. The slices are enumerated 1 to 30 from top-left to bottom-

right.

S0f+s 0 400 800au

T2,f 0 25 50 ms T2,s 0 100 200 300 ms

0

20

40

60

80

100

0.0 0.1 0.2 0.3 0.4 0.5 0.6mean water content

heig

ht fr

om

bot

tom

/ m

m

day 6 day3 day1

MRI in Soils The Open Magnetic Resonance Journal, 2010, Volume 3 73

Fig. (7). Selected water content maps of the ricinus-root soil sys-

tem. Shown is a central vertical slice on day1 and day6 after satura-

tion and sealing.

Fig. (8). 3D MRI-RARE image of the ricinus root system on day-1.

The scales are voxels, voxel size is 0.078mm, matrix size

128x128x128, FOV 10x10x9.6cm.

Beneath the roots spots are also visible, which do not be-long to roots but are sites with locally high water content. For an easier comparison to the water content maps Fig. (9) presents an excerpt from the 3D image in Fig. (8), which corresponds to the vertical slice shown in Fig. (7).

The maximum width of the main roots is 3 voxels, i.e. 2.3mm which is much smaller than the voxel size for the water content maps. Going back to Fig. (6) one can see that for example the central root appears about 3 voxels wide, i.e. about 9mm. Since this is much larger than the root thickness itself this strongly indicates wetter zones around the roots. Such behaviour has also been observed by others [17, 18].

The final step of the evaluation is the calculation of water content difference maps in order to see where the root sys-tem has taken up water preferentially. In Fig. (10) the root

system architecture is overlaid with four axial slices of the water content differences between day-1 and day-6. The wa-ter content change was greatest in the upper central part of the soil core, whereas near bottom if remained wetter. Also in the neighbourhood of the roots greater changes of the wa-ter content is observable. This indicates that the hydraulic conductivity is decreased so far that the water uptake by the roots is not yet compensated instantaneously by flux from more distant regions.

Fig. (10). ricinus-root soil system. Difference of water content

between day-6 and day-1, overlayed by the root system, obtained

from the RARE MRI measurement. (The long cylinders in the left

half are the marker tubes).

CONCLUSIONS

MRI was used for both imaging of the root system archi-tecture and the water contents in a three week old ricinus

Fig. (9). Central vertical slice of the MRI-RARE image of the

ricinus root system.

: 0 0.5

day-1 day-6

8 cm


grown in natural sand. For the determination of the water content it is advantageous to use a relaxometric multi-echo method since this yields relaxation time maps as well as am-plitude maps, from which water contents can be obtained by calibration on internal reference samples. The results of these investigations agree with those we have obtained in a previ-ous work by the application of a single point imaging method (SPRITE) [10]. But in contrast to that MSME is faster, since the total measurement time single point imaging methods scales with the cube of the resolution, echo methods only with the square. The second advantage of MSME is the determination of T2 maps which reflect local soil properties, whereas SPI determines T2

*, which is mainly controlled by

magnetic field inhomogeneities. On the other hand multi-echo imaging requires T2 relaxation times greater than some milliseconds, since the minimum TE is about 1.5 ms. Since in many natural soils T2 can be faster, multi-echo imaging would show no signal, so that single point imaging must be applied.

By the analysis of the water content maps we found that in the neighbourhood of the roots zones with increased water content exist also when the total water content decreased to < 0.05. Anyway, the water content change in some these areas was greater than in the bulk soil, which can be ex-plained by initially increased water content. The next step of the analysis will be soil physical modelling of root water uptake processes, which requires a continuous representation of the root system (skeletisation). For this purpose the qual-ity of the root system imaging should be improved in terms of a better discrimination between soil and roots and the closing of apparent gaps along the roots strands. This could by performed by either other imaging methods or image processing techniques.

ACKNOWLEDGEMENTS

The authors are grateful to EU for funding the NMR ex-periments at the Wageningen NMR Centre (WNMRC07-001), to the DFG (SFB TR-32, PO-746-2/1) for financial support, and Beate Uhlig, ICG-3, Research Centre Jülich for growing the plants.

REFERENCES

[1] Waisel Y, Eshel A, Kafkafi U. Plant roots: the hidden half. Marcel

Dekker: New York 1991. [2] Javaux M, Schröder T, Vanderborght J, Vereecken H. Use of a

three-dimensional detailed modeling approach for predicting root water uptake. Vadose Zone J 2008; 7: 1079-88.

[3] Doussan C, Pierret A, Garrigues E, Pagès L. Water uptake by plant

roots: II – Modelling of water transfer in the soil root-system with explicit account of flow within the root system: comparison with

experiments. Plant Soil 2006; 283: 99-117. [4] Garrigues E, Doussan C, Pierret A. Water uptake by plant roots: I –

Formation and propagation of a water extraction front in mature root systems as evidenced by 2D light transmission imaging. Plant

Soil 2006; 283: 83-99. [5] Perret J, Al-Belushi ME, Deadman M. Non-destructive visualiza-

tion and quantification of roots using computed tomography. Soil Biol Biochem 2007; 39: 391-9.

[6] Tumlinson LG, Liu HY, Silk WK, Hopmans JW. Thermal neutron computed tomography of soil water and plant roots. Soil Sci Soc

Am J 2008; 72: 1234-42. [7] Moradi AB, Oswald S, Massner JA, Pruessmann KP, Robinson

BH, Schulin R. Dynamic of nickel in the rhizosphere of Berkheya coddii using magnetic resonance imaging. Geophys Res Abstr

2008a; 10: EGU2008-A-08398. [8] Nestle N, Baumann T, Niessner R. Magnetic resonance imaging in

environmental science. Environ Sci Technol 2002; 36: 154A-60A. [9] MacFall JS, Van As H. Magnetic Resonance Imaging of Plants. In:

Shachar-Hill Y, Pfeffer PE, Eds. Nuclear magneit resonance in Plant Biology. Rockville, MD: American Society of Plant-

Physiologists 1996. [10] Pohlmeier A, Oros-Peusquens AM, Javaux M, et al. Changes in

soil water content resulting from ricinus root uptake monitored by magnetic resonance imaging. Vadose Zone J 2008; 7: 1010-7.

[11] Pohlmeier A, Oros-Peusquens AM, Javaux M, Menzel MI, Vereecken H, Shah NJ. Investigation of water content and dynam-

ics of a Ricinus root system in unsaturated sand by means of SPRITE and CISS: correlation of root architecture and water con-

tent change. Magn Reson Imaging 2007; 25: 579-80. [12] Stingaciu LR, Pohlmeier A, Blümler P, et al. Characterization of

unsaturated porous media by high-field and low-field NMR re-laxometry. Water Resour Res 2009; 45: W08412.

[13] Edzes HT, van Dusschoten D, Van As H. Quantitative T-2 imaging of plant tissues by means of multi-echo MRI microscopy. Magn

Reson Imaging 1998; 16: 185-96. [14] Barrie PJ. Characterization of porous media using NMR methods.

Ann Rep NMR Spectrosc 2000; 41: 265-316. [15] Stingaciu LR, Pohlmeier A, Blümler P, Weihermüller L, Stapf S,

Vereecken H. Characterization of unsaturated porous media by high-field and low-field NMR relaxometry. Water Resour Res

2009; 45: W08412. [16] Van der Weerd L, Vergeldt F, de Jager PA, Van As H. Evaluation

of algorithms for analysis of NMR relaxation decay curves. Magn Reson Imaging 2000; 18: 1151-8.

[17] Carminati A, Moradi A, Vetterlein D, et al. Dynamics of soil water content next roots: the role of the rhizosphere. Plant Soil 2009: (in

print). [18] Haber-Pohlmeier S, Van Dusschoten D, Stapf S. Waterflow visual-

ized by tracer transport in root-soil-systems using MRI. Geophys Res Abstr 2009; 11: EGU2009-8096.

Received: July 15, 2009 Revised: November 17, 2009 Accepted: November 20, 2009

© Pohlmeier et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

Effect of RF Field Inhomogeneity and Sample Restriction on Spectral Resolution of CP/MAS-

13C NMR Spectra of Natural Organic Matter

Anne E. Berns*,1 and Pellegrino Conte

2

1Institute of Chemistry and Dynamics of the Geosphere, ICG-4: Agrosphere, Forschungszentrum Jülich GmbH, 52425

Jülich, Germany

2Dipartimento di Ingegneria e Tecnologie Agro-Forestali (DITAF), Università degli Studi di Palermo, v.le delle Scienze

13, edificio 4, 90128 Palermo, Italy

Abstract: It is well known that the induced B1 magnetic field in an NMR coil is inhomogeneously distributed. However,

this issue has so far received little attention in the field of environmental NMR. As this research field often aims at quanti-

tative results as well as relaxation phenomena, the repercussions of such inhomogeneity on peak integrals and relaxation

times need to be taken into account.

The objective of the present study was to test standard recording conditions on different sample positions in an NMR coil

in order to determine the effect of the RF field inhomogeneity on the spectrum of a molecularly complex humic material

and on some standard molecules of known structure and conformation. To this end, we measured the peak integral and

signal half-height width of constant sample amounts at different heights in the rotor. In addition, the effect of sample posi-

tion in the rotor on T1H and T2C relaxation times was determined.

We showed that the response profiles of different chemical groups are not necessarily comparable to each other and that

spectra of natural organic matter can change when confined to different regions of the coil. Furthermore, the relaxation

measurements revealed that T1H and T2C relaxation times are position-dependent. Finally, the application of sample res-

triction to the homogeneous region appeared very promising for enhancing the resolution of spectra of complex mixtures.

Keywords: CP/MAS 13

C-NMR, RF field inhomogeneity, physical sample restriction, natural organic matter, signal response profile.

INTRODUCTION

Cross-polarization magic-angle spinning (CP/MAS) 13

C-NMR spectroscopy has become a popular technique for studying natural organic matter (NOM) [1, 2] both in bulk soils and sediments [3] or when NOM is isolated from environmental matrices [4, 5]. However, this technique involves a number of pitfalls.

It is a well-known fact that the magnetic field induced by a radio frequency (RF) in a finite coil rapidly drops at both ends of the coil [6], resulting in an inhomogeneous distribu-tion of the B1 field in NMR spectroscopy [7]. This inho-mogeneity, inherent in all commercial standard solenoid coils, hampers, for example, quantitation of Bloch decay and CP experiments using spin counting [8, 9], where a carefully weighed quantity of an intensity standard [10] is added to a sample and peak integrals are compared. Restricting the sample to the homogeneous region of the solenoid coil is the easiest and most practicable way to avoid an inhomogeneous B1 field. However, the usable volume might be small and samples containing nuclei with low NMR sensitivity, be-cause of either low magnetogyric ratio values or low natural abundance, may quickly reach the limits of feasibility.

*Address correspondence to this author at the Institute of Chemistry and

Dynamics of the Geosphere, ICG-4: Agrosphere, Forschungszentrum Jülich

GmbH, 52425 Jülich, Germany; Tel: +49 2461 615656; Fax: +49 2461

612518; E-mail: [email protected]

Several optimized winding geometries such as coils with a variable pitch [7, 11, 12] or a variable wire width [13] have been proposed in order to enhance the homogeneous volume of solenoid coils. Furthermore, a number of selective pulse sequences and gradient techniques for spatial localization and sample restriction have been developed [14-20].

In addition to the inherent B1 inhomogeneity in solenoid coils, RF profiles of different frequencies, e.g.

1H and

13C,

are not necessarily symmetric about the coil centre as re-ported by Paulson et al. [21]. This phenomenon of wave-length effect, which results in a misalignment of the physical centre of the RF field, was first reported by Stuhlman & Githens in 1932 [22] and becomes experimentally important once the coil wire length reaches about 10% or more of the free space wavelength of the applied RF field [21]. This is especially of concern for the

1H field as this critical coil

length can be easily reached even at lower frequencies. A slight field centre misalignment of the

1H field is usually

counteracted by using a variable amplitude or ramped CP transfer [23], which sweeps through a range of (usually)

1H

RF field amplitudes. For severe misalignments of two or three RF fields in a double or triple resonance probe, Paulson et al. [21] developed a balanced series resonant circuit, which splits the capacitance across the inductor instead of using a resonant circuit with only one capacitance. Unfortu-nately, this hardware change is not practicable in every rou-tine NMR lab.

76 The Open Magnetic Resonance Journal, 2010, Volume 3 Berns and Conte

Magic-angle spinning in combination with cross-polarization adds to the problem of field inhomogeneity as dipolar interactions become time-dependent. In a non-spinning sample, the most efficient magnetization transfer is reached when the B1 field mismatch parameter = 1I – 1S between two spin systems I and S is close to zero. In a sam-ple which is spun at a rate r the most efficient transfer oc-curs when = ± n r (sideband matching condition) [21, 24]. At high spin rates, when the mismatch parameter be-comes a considerable fraction of the RF field nutation rate, i B1, the necessity of a homogeneous RF field becomes very important. This requirement is even more stringent at an op-erating frequency of 800 MHz or more when MAS rates reach or exceed dipolar couplings of

1H-

13C or

1H-

15N [21].

Spectral resolution depends upon the length (N) of the transform interval: N = F/ f, where f is the smallest fre-quency difference which can be distinguished at a given sampling frequency F. The frequency difference f depends on the signal width, which, in turn, is influenced by the re-laxation time of the sampled signal [25, 26]. Relaxation times are affected by the external magnetic field and the ap-plied RF field sent to the sample through the transmit-ter/receiver coil. As already described above the RF field homogeneity throughout this coil is a major factor affecting the NMR sensitivity and the resolution of the measurements.

The objective of the present study was to check standard recording conditions on different sample positions in an NMR coil in order to determine the effect of the RF field inhomogeneity on the spectrum of a molecularly complex humic material and on compounds of known structure and conformation. To this end, we performed CP/MAS-NMR experiments where constant sample amounts were moved through different heights in the rotor. At each height the peak integral and signal half-height width were measured and used as indicators of sensitivity and spectral resolution. In addition, the effect of sample position on T1H and T2C relaxation times was determined.


Standard Materials

Sodium dodecylsulfate (SDS) was purchased from Sigma-Aldrich

® (Steinheim, Germany) in purum quality and

glycine was purchased from Merck (Darmstadt, Germany) in pro analysi quality. The standard materials were used with-out further purification. Pure quartz was purchased from Merck (Darmstadt, Germany) in pro analysi quality and ground in a planetary mill (PM 400, Retsch, Haan, Germany) to fine powder.

Humic Acid

A humic acid from the A horizon of an andosol from the caldera of Vico (near Rome, Italy) was extracted, purified and characterized as reported in Cozzolino et al. [27]. The ash content was below 3 % and the Fe/C ratio « 1, hence the CPMAS

13C-NMR spectra were not affected by paramag-

netic impurities [28].

The HA was analysed to determine its elemental compo-sition (C, H, N) in a Fisons Interscience EA1108 elemental analyser. The elemental composition corrected for the ash content (< 3 %) was as follows: C 56.8 %, H 4.5 %, N 5.2 %.

The humic acid was also analysed by atomic-absorption spectrometry (AAS) in order to verify the presence of poten-tially paramagnetic species (Fe, Mn, Cu) in the ash content. The AAS analyses were performed on a Perkin–Elmer Ana-lyst 700 with an instrumental sensitivity of 0.1 mg kg

-1. Ali-

quots of the humic acid (50 mg) were boiled in a ni-tric/perchloric acid solution until complete mineralization of the organic matter and dissolution of the inorganic ashes was obtained. The solutions were transferred to 20-mL volumet-ric flasks and analysed by AAS. Negligible traces of paramagnetic metals were found.

General Solid-State NMR Conditions

A 7.05 T Varian UNITY INOVA™ (Varian Inc., Palo Alto, CA, USA), equipped with an Apex HX wide-bore probe operating at a

13C frequency of 75.4 MHz, was used to

acquire the 13

C-NMR spectra. The samples were packed in 6 mm zirconium rotors with Teflon

® bottom and top spacers

and Vespel®

drive tips. The temperature was kept constant at 25.0 ± 0.1 °C. Magic-angle spinning was carried out at 7500 ± 1 Hz. Repetititon times were 3s for glycine and 35 s for SDS. The spectra of glycine and SDS were recorded with a 1H RF field strength of 65.3 kHz with a ramp of 7 kHz.

CP/MAS spectra of Vico HA were acquired with a 1H RF

field strength of 40.4 kHZ and a ramp of 15.3 kHz to account for inhomogeneities of the Hartmann-Hahn condition [2, 21, 23, 24]. In both cases the matching

13C and

1H RF fields

were determined in separate (non-ramped) experiments by fixing the

13C RF field and arraying the (non-ramped)

1H RF

field. Decoupling was done using a TPPM sequence with a 1H field strength of 54 kHz, a phase of 8° and a pulse length

of 9.3 s.

VNMRJ software (Version 1.1 RevisionD, Varian Inc., Palo Alto, CA, USA) was used to acquire all the free induction decays (FID). Spectra elaboration was performed

Fig. (1). Pencil® rotor and dimensions of manufacturer-determined

sample volume.

Effect of RF Field Inhomogeneity The Open Magnetic Resonance Journal, 2010, Volume 3 77

by Mestre-C software (Version 4.9.9.9, Mestrelab Research, Santiago de Compostela, Spain). All the FIDs were transformed by applying first a 4 k zero filling and then an exponential filter function with a line broadening (LB) of 20 Hz. Fully automatic baseline correction using a Bernstein algorithm was applied for baseline corrections [29].

Sample Positioning in the Rotor

A constant sample amount was incrementally moved in a volume with 2 mm fill height through different heights in the rotor to record the spectra of the two standard substances at different positions in the rotor (Fig. 1). At every height the peak integral and the half-height width of the signals were determined as indicators of sensitivity and spectral resolu-tion. Furthermore, relaxation times of selected volumes were determined. The amount of sample needed to fill a volume of 2 mm fill height was 32 ± 1 mg for glycine (= 0.43 mmol) and 26 ± 1 mg for SDS (= 0.10 mmol). In a first set of ex-periments with SDS and glycine, the bottom and top parts were filled with fine quartz powder with a constant density of 26 ± 1 mg/mm. This was done to ensure a constant total weight of the rotor (3.6 ± 0.01 g) and avoid possible height differences due to the lifting of the rotor during spinning. The replacement of a defective preamplifier, although quickly discovered, during the first set of experiments, caused slightly different recording conditions for SDS and glycine so that we decided to verify the reproducibility of these experiments. The repetition of the experiments with SDS and glycine was done with purpose-built boron nitride inlets to facilitate the positioning of the sample and ensure an even better weight constancy of the rotor. The results re-ported are from this second set of experiments, which con-firmed the first set. The correct position of the sample in the rotor was determined with the help of a light table and a template. The total possible sample volume was given by the Teflon

® top and bottom spacers provided by the manufac-

turer matching the coil geometry. The bottom limit of this volume was defined as position 0 mm and the top as 14 mm (Fig. 1). The position of a sample is indicated through the

position of the centre of the sample in question. For example, a sample filling a volume from 2 to 4 mm height is indicated by the position at 3 mm.

After determination of the homogeneous region, experi-ments were conducted with centre, bottom & top and fully filled rotors. In this case “centre filled“ refers to samples where ground quartz powder (or BN inlets) was filled from 0 to 5 mm and from 11 to 14 mm and the sample occupied the volume from a height of 5 to 11 mm. The sample position was checked with the help of a light table and a template. In “bottom & top filled” rotors, the sample occupied the lower and top part of the rotor, while the centre was filled with ground quartz powder (or BN inlet).

CP/MAS and Relaxation Experiments

T1H relaxation was determined with the sequence displayed in Fig. (2A) with an array of delay times (d3) from 0 to 2 s for glycine and an array of 0 to 11 s for SDS. A spin echo sequence with an array of tau from 0 to 5333 s was performed to determine the T2C relaxation time constant (Fig. 2B). The fitting procedures for the relaxation curves were done with OriginPro 7.5 SR6 (Version 7.5885, OriginLab Corporation, Northampton, MA, USA). The errors reported are obtained from the fitting procedure.

The CP/MAS 13

C-spectra were evaluated for their peak integrals and widths at half-height. The determined integrals and widths at half-height were normalized to position 8.


In the 7.05 T instrument used for the present study, the 1H radio frequency of 300 MHz corresponds roughly to a

free space wavelength of 1 m. Hence, the resulting critical coil wire length, where wavelength effects are to be expected [21, 22], is 10 cm. The Apex Pencil

® probe utilized for the

present study is equipped with a 5-turn coil with an inner diameter of 6 mm and a wire thickness of 1 mm. Thus, the coil wire length is approximately 11 cm, excluding the lead lengths. The RF of 75 MHz for the

13C nucleus corresponds

Fig. (2). T1H pulse sequence (A) and T2C echo pulse sequence (B).


to approximately 4 m, i.e. a critical coil wire length of 40 cm. Hence, for the

1H RF field a slight centre misalignment

can be expected as the coil wire length amounts to a mini-mum of 11 % of the applied wavelength. A rapid measure-ment of

1H and

13C-integrals on a 2-mm-thick adamantane

sample confirmed that the centres of both RF profiles were not perfectly aligned (data not shown). Therefore, a ramped 1H field to counterbalance this misalignment must be ap-

plied.

To avoid RF field nutation rates, which match rates equivalent to the MAS rate, r, or twice as great, 2 r, (usu-ally conditions which destroy spin-locked magnetization during cross-polarization through recoupling of chemical shift anisotropy), an RF nutation rate exceeding 5 r is rec-ommended [21, 24]. A spin rate of 7.5 kHz and

1H RF fields

of 40.4 kHz and 65.3 kHz, as reported in Materials and Methods, were chosen for the present study. These condi-tions ensured that the RF field strengths exceeded the rec-ommended minimum RF field strength of 5 r (i.e. 37.5 kHz) so that no negative effects on the spin-locked magnetization were to be feared.

It should be noted that pulse widths and matching condi-tions were set with completely filled rotors and, as a conse-quence, sensitivity profiles exhibit less dependence than they would if the parameters were optimized on a centre-filled rotor [8].

The CP/MAS spectra of glycine and sodium dodecyl sul-fate recorded at different positions within the manufacturer-defined volume are displayed in Fig. (3). The detection-sensitivity profiles clearly show that the signal integrals are strongly dependent on the position of the sample in the rotor. Only in a small region in the centre of the rotor is the signal response reasonably consistent.

Figs. (4A) and (4B) show the normalized signal integrals of these CP/MAS spectra. The signal integrals are normal-ized to position 8, the centre of the determined homogeneous region. The sensitivity profile of the glycine methylene group displays a smooth curve, while the curve of the COOH group has two outliers in positions 8 and 9. In the profile of SDS, however, only the integrals of signal C2-10 are consis-tent. Considering the molar amounts present in the rotor, it becomes clear that signal C2-10 corresponds to 0.90 mmol

Fig. (3). Response profiles of glycine (A) and SDS (B).


of C, the two signals of glycine to 0.43 mmol C each and the remaining signals of SDS correspond to only 0.10 mmol C each. These reduced signals result in a larger error in the evaluation of the spectrum. Nonetheless, the general shape of the profile (as well as the profile of the first experiment, not shown) follows that of glycine with severely reduced signal integrals in the outer regions; the lower positions being worse than the upper regions.

As already stated by Campbell et al. [8], sensitivity pro-files in cross-polarization experiments (as used in the present study) should be expected to display a much less pronounced drop at the end of the coil. In fact, the inhomogeneous ampli-tudes of the

1H and

13C RF fields are counterbalanced

through the Hartmann-Hahn signal enhancement as long as the ratio of the

1H/

13C RF field levels is maintained (which is

given by the use of a ramped 1H RF field). Campbell et al.

[8] mention two factors which need to be considered in such position-dependent sensitivity profiles in Bloch decay and cross-polarization experiments. The first factor is the varia-tion of the RF field magnitude, which results in different pulse flip angles for different regions in the coil. The second factor is the variation of the curvature of the B1 field, which produces magnetization elements that are directed out of the detection plane. Due to the symmetry of the coil, these non-planar magnetization elements exist in opposing pairs which cancel each other out. Hence, not all of the generated mag-

Fig. (4). Sensitivity (A and B) and resolution profiles (C to F) of the Apex Pencil® probe for

13C-CPMAS spectra of glycine (A, C and E)

and SDS (B, D and F) (all normalized to position 8 or the central region, respectively).


netization is observed. According to Campbell et al. [8], these directional inhomogeneities are the main reason for the spatial dependence of the sensitivity profiles in cross-polarization experiments (as the inhomogeneous amplitudes of the

1H and

13C RF fields should be counterbalanced

through the Hartmann-Hahn signal enhancement).

In the case of the present study where the samples were high-rate-spun (Campbell at al. [8] studied non-spinning samples), a further factor needs to be considered. As the RF field amplitudes drop at the end of the coil, they may reach a value where the condition RF field strength > 5 r is no longer met, and they may in contrast approach a condition where the RF field nutation rates iB1 match the MAS rate ±

r or ± 2 r, thereby destroying the spin-locked magnetiza-tion during cross-polarization. Unfortunately, no equipment was at hand for a direct mapping of the RF field strengths. Furthermore, an exhaustive discussion of the effect of MAS on CP is already available in several high-quality papers [21, 24] and would go beyond the scope of the present paper.

For quantitative evaluations it is important to know whether the reduction in signal integral is distributed uni-formly all along the NMR signals of each chemical group. In the case of glycine at the lowest positions of the manufac-turer-defined region, the response decrease of the methylene group is stronger than that of the carboxyl group. The signal response profiles of the SDS signals are harder to interpret as the signal response is very scattered (even in the homogene-ous region determined), but in the lower region it can be seen that the signal of the methylene groups C2-10 tends to be disproportionately reduced as can be seen in the profile of glycine.

The line widths of the signals are also position-dependent as displayed in Figs. (4C) and (4D). The normalized display (to position 8 as in the signal response profiles) shows that the widths at half-height of the different groups broaden to a different extent outside the central homogeneous region. The line width of the glycine methylene group, already broader by nature through its faster relaxation, shows a stronger in-crease (factors 2.6 at positions 3 and 12) in its line width than the carboxyl group (factors 1.6 and 1.4 at positions 3 and 12). Also in the case of SDS, the methylene signals (C2-10 and C11) are the ones with a larger enhancement of the width at half-height outside the central region, while the line width of the methyl group (C12) is hardly perturbed at all. The width at half-height of the methylene signal closest to the sulfate group is already quite broad in the centre regions, and in most of the outer regions the signal width cannot be measured. The normalized widths at half-height determined on fully, centre and bottom & top filled rotors are shown in Figs. (4E and 4F). The signal widths in the bottom & top filled rotors were elevated similar to the widths in the incre-mented experiments, whereas the methylene signal of gly-cine was too broad to be reasonably measured. The widths in the fully filled rotors were in between the widths of the cen-tre and bottom & top filled rotors as expected.

The line width is determined by the overall decay rate of the transverse magnetization T2*. The latter, in turn, is af-fected by both homogeneous (the transverse relaxation rate T2) and inhomogeneous contributions (called inhomogene-ous broadening and often denoted T2

†). The transverse re-

laxation rate T2 is a fundamental property and is influenced by the type of molecule, its physical environment and its motion in this environment. Inhomogeneous broadening re-sults from non-uniform magnetic fields across the sample. A poorly shimmed probe head, for example, causes spins to experience different Larmor frequencies in different parts of the sample due to an inhomogeneous B0 field. The homoge-neous contribution can be distinguished from the inhomoge-neous part through a simple echo sequence. As the latter is due to spins precessing at different frequencies throughout the sample over time they get out of step and eventually can-cel each other out, resulting in a decay of the overall mag-netization. This simple spin dephasing can be reversed by applying a 180° pulse after an evolution time and recording the signal after another period, at which point the phases will have realigned. The homogeneous part cannot be re-versed as it is a true relaxation process, i.e. an approach to equilibrium [25, 26]. Measurement of the T2C relaxation times are shown in Figs. (5A and 5B). It can be seen that especially the longer relaxation times of the COOH group of glycine and the methyl group of SDS are strongly influenced by the position of the sample. The application of an echo sequence where decoupling started only with the recording of the signal revealed that dephasing during the echo periods is sensitive to decoupling. Without decoupling being turned on during dephasing the determined relaxation times showed no position dependence and were extremely short by reason of magnetization loss due to coupling (data not shown). It is hence likely that the higher T2C relaxation times determined in the centre of the rotor arise from more efficient decou-pling of the sample, due to the increased homogeneity of the RF pulses in this region. Calculation of the overall decay rate of the transverse magnetization T2* from the determined widths at half-height revealed a large discrepancy between the overall decay rate and the T2C relaxation, e.g. a T2C of 25 ms and a T2* of 6.4 ms for the COOH group of glycine. Hence most of the overall decay rate of the transverse mag-netization is due to inhomogeneous contributions. As the transverse relaxation takes place after the application of the B1 field the inhomogeneities influencing this factor must come from the B0 field. Inhomogeneities in the static mag-netic field are counterbalanced as effectively as possible by the shimming procedure. However, the simple presence of the coil in the magnetic field and the pulsed application of the B1 field lead to disturbances in the B0 field which are hard to counterbalance completely. Furthermore, the shim-ming is optimized to the line shape of adamantane, i.e. while the B1 field is active. The disturbances which the coil pro-duces in the B0 field are strongest at the edges of the coil and therefore the line widths increase in these regions. In the main, it must be noted that the determination of T2C times in the fully filled rotor obviously underestimates the relaxation times.

Measurement of the longitudinal relaxation T1H shows a longer T1H in the centre position than in the outer regions (Figs. 5C and 5D). This is due to the fact that as the RF field is not homogeneous, the 90° pulse is not perfect throughout the sample. Hence the outer regions of the sample experience a pulse which differs from 90° resulting in a lower magneti-zation from which the system needs to return to equilibrium. The centre position is closest to the true 90° pulse and hence


has the longest relaxation time. It can be seen that measure-ments on a fully filled rotor also lead to an underestimation of the T1H relaxation times.

In Fig. (6) the influence of physical sample restriction on the resolution of a spectrum of a humic acid is shown. All three spectra were recorded as having a similar signal-to-noise ratio and were graphically normalized to the carboxyl signal at 172 ppm. It can be seen that the relative integrals of some signals in the HA spectrum change, which leads to differences in the peak integrals ranging from 1 to 3 %. The main regions influenced are the aliphatic and O/N-alkyl re-gions (0-45 and 45-90 ppm). The signal in the aliphatic re-gion of the bottom & top filled rotor (blue spectrum) is shorter and broader than the signal resulting from the central region (red spectrum) and that in the spectrum of the fully filled rotor (black spectrum). Most interesting is the fact that in the O/N-alkyl region the resolution of the signals is en-hanced. Between 45 and 65 ppm the blue spectrum of the bottom & top filled rotor shows only one broad signal, whereas in the red spectrum of the centre filled rotor three peaks can be distinguished. The black spectrum of the fully filled rotor merges both features and is less resolved than the red spectrum, but better resolved than the blue spectrum. The more strongly enhanced resolution of the O/N-alkyl region can also be seen in the spectra of glycine (Fig. 4C), where the signal half-height width of the methylene group is re-duced by a factor of 2.6 when moving from the outer to the central region. As already stated for the T2C measurements,

the increase in resolution probably arises from a more effi-cient decoupling in the centre of the rotor.

Although the physical reduction of the sample to the cen-tral region requires longer measurement times to reach a good signal-to-noise ratio, the resolution enhancement achieved can be of importance when the interactions of hu-mic acids with labelled chemicals [30] or changes in a spe-cific region are monitored [31].

CONCLUSIONS

Although RF field inhomogeneity is a subject well studi-ed by NMR scientists in chemical and physical research groups, the issue has so far received little attention in the field of environmental NMR. As this research field often also aims at quantitative results as well as relaxation pheno-mena, the repercussions of the position dependence of the peak integral and of relaxation times on such NMR measu-rements need to be taken into account.

The present study shows that the response profiles are not necessarily similar for different chemical groups and hence spectra of complex organic matter can change when confined to different regions of the coil. It must be pointed out that the work presented in this paper was done on a 6 mm probe he-ad. The described effects might be smaller in smaller rotors or in coils with different geometry. We therefore recommend that the signal response profile in the coil should be checked, for example by the proposed procedure, and the sample should be restricted to the determined homogeneous region

Fig. (5). T2C (A and B) and T1H (C and D) relaxation rates of glycine (A and C) and SDS (B and D).


when quantitation is required as also Bloch decay experi-ments are sensitive to coil sample position. The same applies to relaxation measurements as relaxation times are also posi-tion-dependent. Furthermore, the application of sample res-triction to the homogeneous region appears very promising for enhancing the resolution of spectra of complex mixtures.

ACKNOWLEDGEMENTS

The authors thank Forschungszentrum Jülich GmbH (Germany) for financing PC as a visiting scientist at the NMR centre of the Institute of Chemistry and Dynamics of the Geosphere, Institute 4: Agrosphere. Many thanks are also due to Dr. Andreas Pohlmeier for helpful discussions on relaxation phenomena. The Language Services of Forschungszentrum Jülich are also gratefully acknowledged for revising the English of the manuscript.

REFERENCES

[1] Cardoza LA, Korir AK, Otto WH, Wurrey CJ, Larive CK. Applica-

tions of NMR spectroscopy in environmental science. Prog Nucl Magn Reson Spectrosc 2004; 45: 209-38.

[2] Conte P, Spaccini R, Piccolo A. State of the art of CPMAS C-13 NMR spectroscopy applied to natural organic matter. Prog Nucl

Magn Reson Spectrosc 2004; 44: 215-23. [3] Wilson MA. NMR techniques and applications in geochemistry

and soil chemistry. London: Pergamon Press1987. [4] Mao JD, Fang XW, Schmidt-Rohr K, Carmo AM, Hundal LS,

Thompson ML. Molecular-scale heterogeneity of humic acid in particle-size fractions of two Iowa soils. Geoderma 2007; 140: 17-

29. [5] Mopper K, Stubbins A, Ritchie JD, Bialk HM, Hatcher PG. Ad-

vanced instrumental approaches for characterization of marine dis-solved organic matter: Extraction techniques, mass spectrometry,

and nuclear magnetic resonance spectroscopy. Chem Rev 2007; 107: 419-42.

[6] Tipler PA. Physik. Spektrum Akademischer Verlag GmbH: Heidelberg 1998.

[7] Leifer MC. RF Solenoid with extended equiripple field profile. J Magn Reson Ser A 1993; 105: 1-6.

[8] Campbell GC, Galya LG, Beeler AJ, English AD. Effect of RF inhomogeneity upon quantitative solid-state NMR measurements. J

Magn Reson Ser A 1995; 112: 225-8. [9] Smernik RJ, Oades JM. Spin accounting and RESTORE - two new

methods to improve quantitation in solid-state C-13 NMR analysis of soil organic matter. Eur J Soil Sci 2003; 54: 103-16.

[10] Hall RA, Jurkiewicz A, Maciel GE. A bicyclic ketone as a solid-state C-13 NMR intensity reference. Anal Chem 1993; 65: 534-8.

[11] Idziak S, Haeberlen U. Design and construction of a high homoge-neity RF coil for solid-state multiple-pulse NMR. J Magn Reson

1982; 50: 281-8. [12] Horne D, Kendrick RD, Yannoni CS. Bond length measurements in

amorphous solids by nutation NMR spectroscopy - the role of RF field homogeneity. J Magn Reson 1983; 52: 299-304.

[13] Privalov AF, Dvinskikh SV, Vieth HM. Coil design for large-volume high-B1 homogeneity for solid-state NMR applications. J

Magn Reson Ser A 1996; 123: 157-60. [14] Charmont P, Lesage A, Steuernagel S, Engelke F, Emsley L. Sam-

ple restriction using magnetic field gradients in high-resolution solid-state NMR. J Magn Reson 2000; 145: 334-9.

[15] Charmont P, Sakellariou D, Emsley L. Sample restriction using radiofrequency field selective pulses in high-resolution solid-state

NMR. J Magn Reson 2002; 154: 136-41. [16] Tycko R, Pines A. Spatial localization of NMR signals by narrow-

band inversion. J Magn Reson 1984; 60: 156-60. [17] Kemp-Harper R, Styles P, Wimperis S. B1-selective pulses. J

Magn Reson Ser A 1996; 123: 230-6. [18] Shaka AJ, Keeler J, Smith MB, Freeman R. Spatial localization of

NMR signals in an inhomogeneous radiofrequency field. J Magn Reson 1985; 61: 175-80.

[19] Hoult DI. Solution of the Bloch equations in the presence of a varying B1 field - Approach to selective pulse analysis. J Magn

Reson 1979; 35: 69-86. [20] Malloy CR, Lange RA, Klein DL, et al. Spatial localization of

NMR signal with a passive surface gradient. J Magn Reson 1988; 80: 364-9.

[21] Paulson EK, Martin RW, Zilm KW. Cross polarization, radio fre-quency field homogeneity, and circuit balancing in high field solid

state NMR probes. J Magn Reson 2004; 171: 314-23. [22] Stuhlman O, Githens S. The magnetic field of a solenoid oscillating

at radio frequencies. Rev Sci Instrum 1932; 3: 561-71.

Fig. (6). 13

C-CP/MAS spectra of Vico humic acid in different positions of the manufacturer-defined sample region.


[23] Metz G, Wu XL, Smith SO. Ramped-amplitude cross-polarization

in magic-angle-spinning NMR. J Magn Reson Ser A 1994; 110: 219-27.

[24] Wu XL, Zilm KW. Cross-polarization with high-speed magic-angle-spinning. J Magn Reson Ser A 1993; 104: 154-65.

[25] Levitt MH. Spin dynamics. Chichester: John Wiley & Sons, Ltd. 2008.

[26] Keeler J. Understanding NMR spectroscopy. Chichester: John Wiley & Sons, Ltd. 2005.

[27] Cozzolino A, Conte P, Piccolo A. Conformational changes of hu-mic substances induced by some hydroxy-, keto-, and sulfonic ac-

ids. Soil Biol Biochem 2001; 33: 563-71. [28] Preston CM. Applications of NMR to soil organic matter analysis:

history and prospects. Soil Sci 1996; 161: 144-66.

[29] Brown DE. Fully automated base-line correction of 1D and 2D

NMR-spectra using Bernstein polynomials. J Magn Reson Ser A 1995; 114: 268-70.

[30] Berns AE, Conte P, Philipp H, Witte EG, Lewandowski H. Interac-tions between 2-aminobenzothiazole and natural organic matter as

evidenced by CPMAS Nitrogen-15 NMR spectroscopy. Vadose Zone J 2009; 8: 1-7.

[31] Golding CJ, Smernik RJ, Birch GF. Investigation of the role of structural domains identified in sedimentary organic matter in the

sorption of hydrophobic organic compounds. Environ Sci Technol 2005; 39: 3925-32.


© Berns and Conte; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

Interaction of a Recombinant Prion Protein with Organo-Mineral Complexes as Assessed by FT-IR and CPMAS

13C NMR Analysis

Fabio Russo1, Liliana Gianfreda

1, Pellegrino Conte

2 and Maria A. Rao*

,1

1Dipartimento di Scienze del Suolo, della Pianta, dell’Ambiente e delle Produzioni Animali. Università di Napoli

Federico II, via Università 100, 80055 Portici (NA), Italy

2Dipartimento di Ingegneria e Tecnologie Agro-Forestali. Università degli Studi di Palermo, v.le delle Scienze 13,

edificio 4, 90128 Palermo, Italy

Abstract: Prion proteins are considered as the main agents for the Transmissible Spongiform Encephalopathies (TSE).

The misfolded form, PrPSc

, which is also indicated as the etiological agent for TSE, exhibits high resistance to degradation

in environmental processes. Soil contamination by prion proteins is a real environmental issue since contaminated soils

can become potential reservoir and diffuser for TSE infectivity. In this work, the interaction of prion protein with organo-

mineral complexes was studied by using a recombinant non pathogenic prion protein and a model soil system. This latter

was represented by a soil manganese mineral coated with polymerized catechol. FT-IR spectra showed amide I and II sig-

nals which revealed protein involvement in catechol polymers coating manganese oxide surface. CPMAS 13

C-NMR was

applied to follow the complexation of the protein to the soil-like system. All the signals were attributed to C-N in peptidic

bonds, alkyl chains and methyl groups. The NMR spectrum of the prion protein interacting directly with birnessite re-

vealed disappearance of signals due to the paramagnetic nature of manganese oxide or protein abiotic degradation, while

the presence of organic matter strongly reduced the disappearance of prion protein signals.

Keywords: Prion protein, TSE diseases, Soil Organo-Mineral Complexes, FT-IR, CPMAS 13

C NMR.

INTRODUCTION

Transmissible Spongiform Encephalopathies are fatal, neurodegenerative diseases also known as prion diseases including bovine spongiform encephalopathy, human Creutzfeldt-Jakob disease, kuru, sheep scrapie, and chronic wasting disease of deer, elk and moose [1]. PrP

Sc is a mis-

folded isoform of the normal cellular prion protein (PrPC)

considered as the main, if not the sole, agent of TSE [2]. PrP

Sc causes pathogenesis in the central nervous system with

the formation of amyloidal aggregates and a consequent spongiosis in the brain of humans and animals. These dis-eases, nowadays considered incurable, lead slowly but in-exorably to death.

PrPSc

is protease resistant and exhibits a remarkable resis-tance to degradation and inactivation by standard decontami-nation procedures [3].

Dispersion of PrPSc

contaminated material in soil can oc-cur from meat and bone meal storage plants, use of fertilizers augmented with meat and bone, decomposition of animal carcasses buried in soil, liquid and solid waste fragments from abattoirs, and wastes from TSE-infected animals [4]. As firstly evidenced by Brown and Gajdusek [5], buried in-fectious prion protein can persists in soil even for years. Sei-del et al. [6] found that in soil, scrapie agent remained in an infectious form after 29 months. Moreover, infected soil

*Address correspondence to this author at the Dipartimento di Scienze del

Suolo della Pianta, dell’Ambiente e delle Produzioni Animali, Università

degli Studi di Napoli Federico II, via Università 100, Portici, Italy; Tel:

00390812529173; Fax: 00390812539186; E-mail: [email protected]

was supposed to preserve TSE capability to transmit the dis-ease to healthy grazing animals after 16 years from the con-tamination event [7].

Either the infective or the benign forms of prion proteins are able to strongly bind to mineral [8-12] and organic soil components [13-15]. PrP

Sc can be displaced from environ-

mental compartments to soil, but this does not necessarily decrease its infectivity [10,16]. Prion protein and other soil exogenous proteins can also easily interact with organo-mineral complexes that likely are in soil more abundant than the mineral or organic components alone. Soil has been sug-gested to act as a reservoir of TSE infectivity [17,18] and even to enhance the transmissibility of prion disease by oral uptake of soil complexed PrP

Sc [16]. While biotic PrP

Sc deg-

radation by some proteases can be reduced because of persis-tent PrP

Sc aggregates, its inactivation may arise by abiotic

oxidative reactions. In fact birnessite, a manganese oxide common in soil, is able to degrade the PrP

Sc in aqueous sus-

pensions [19]. The presence of organic matter coating reac-tive soil mineral surfaces could hinder abiotic degradation processes with different effects on prion stability.

Interaction of PrPSc

with organic matter can take place ei-ther with forming or pre-formed humic substances (HS). Interaction with already-formed HS could lead to superficial protein sorption, while entrapment phenomenon could arise if the protein is involved in the humification process [13]. In general, immobilized proteins may result more stable than free forms; for instance enzymes in humic complexes are more resistant to thermal denaturation and proteolysis than the respective free enzymes [20].

Interaction of a Recombinant Prion Protein The Open Magnetic Resonance Journal, 2010, Volume 3 85

In soil, prion proteins can interact with soil constituents and remain immobilized in soil colloids. Since in soil envi-ronment prion protein contamination can likely occur at the surface layers, the infectious agent can result more easily available for free-ranging animals promoting the disease diffusion in the environment. It is, therefore, important to understand how prion proteins interact with soil components present in the surface layers and how the soil organic matter and its transformation in humic substances could affect the stability of prions, possibly protecting them from environ-mental degradation.

The aim of the present work was to study through differ-ent spectroscopic techniques, such as FT-IR and CPMAS 13

C-NMR, the complexation of a recombinant ovine prion protein (recPrP) with soil organo-mineral complexes ob-tained by oxidative polymerization of catechol mediated by birnessite. The birnessite-catechol complexes resembling soil organo-mineral complexes were prepared in the presence of recPrP and by following two different sequences of addition of recPrP, before and after catechol polymerization, in order to obtain either entrapment or sorption, respectively.


Chemicals

Reagent grade catechol (Cat) (>99 % purity) and HPLC grade solvents were purchased from Sigma Aldrich (Ger-many). Birnessite (Bir) was synthesized according to McKenzie [21] using KMnO4 and HCl. X-ray spectra dif-fraction analyses performed on the synthesized MnO2 con-firmed the birnessite poorly crystalline mineral structure with characteristic peaks as reported in McKenzie [21]. Bir-nessite has a point of zero charge of 1.81 [22], and pH-dependent negative charge at typical soil pHs.

A purified recombinant ovine ARQ genetic variant prion protein (recPrP) with a MM of 23 kDa (residues 23-234) was prepared according to Rezaei et al. [23] and kindly furnished by Virologie et Immunologie Moléculaires lab of INRA (Jouy-en-Josas, France). The protein with an isoelectric point of 9.2 [23] is constituted by a well-folded C-terminal domain (residues 125-234) and a flexible N-terminal arm (residues 23-124) [24].

Preparation of recPrP-Organo-Mineral Complexes

Organo-mineral-recPrP complexes used in this study were prepared according to the methodology described by Rao et al. [13]. Briefly, either 3 or 5 mM catechol (Cat3 and Cat5, respectively), 0.5 mg ml

-1 recPrP and 5 mg ml

-1 birnes-

site in 0.1 M sodium acetate buffer at pH 5.5 were incubated in different systems. Two ternary samples (catechol-birnessite-recPrP) were produced: Cat-Bir-recPrP, where Cat, Bir and recPrP were incubated for 4 h at 25 °C, and Cat-Bir+recPrP, where Cat and Bir were incubated for 2 h before adding recPrP and incubating for further 2 h at 25 °C. Binary systems, Cat-Bir, Cat-recPrP and Bir-recPrP, and sample with only Cat, recPrP or Bir were also produced and served as controls.

All trials were incubated for a total of 4 h at 25 °C. Su-pernatants and insoluble precipitates were separated by cen-trifugation (30 min at 10,000 rev min

-1 and 4 °C). Residual

catechol and recPrP concentrations were evaluated by HPLC

analysis [13] with a Shimadzu instrument equipped with UV-Vis variable-wavelength absorbance detector, set at 280 and 222 nm respectively.

The pellets were washed twice with 0.02 M NaCl, twice with double-distilled water, freeze-dried and stored at 4 °C [13].

FT-IR Analyses

The Fourier transform infrared spectra of insoluble prod-ucts were recorded by the Universal Attenuated Total Re-flectance (UATR) method using a Perkin Elmer FT-IR spec-trometer. Each spectrum represents a collection of 24 scans recorded at a 4 cm

-1 resolution.

CPMAS 13

C-NMR Analyses

CPMAS 13

C NMR measurements were performed on a Bruker Avance-II 400 spectrometer (Bruker Biospin, Milan, Italy) operating at 100.6 MHz on carbon-13 and equipped with a 4 mm standard bore solid state probe. Samples were packed into 4 mm zirconia rotors with Kel-F caps and the rotor spin rate was set at 13,000 ± 1 Hz. In total, 4 k data points (TD) were collected over an acquisition time (AQ) of 50 ms, a recycle delay (RD) of 2.0 s, and 30,000 scans (NS). Contact time was 1 ms, while a

1H RAMP sequence was

used to account for possible inhomogeneity of the Hart-mann–Hahn condition at high rotor spin rates [25]. Precise 1H 90° pulse calibration for CP acquisition was obtained as

reported in Conte and Piccolo [26]. Bruker Topspin®

2.0 software was used to acquire all the spectra. Spectra elabora-tion was conducted by Mestre-C software (Version 4.9.9.9, Mestrelab Research, Santiago de Compostela, Spain). All the FIDs (free induction decays) were transformed by applying first a 4 k zero filling and then an exponential filter function with a line broadening (LB) of 100 Hz. Fully automatic baseline correction using a Bernstein algorithm was applied for baseline corrections [27].

RESULTS

FT-IR Analyses

Infrared spectra were recorded for polymeric products of catechol adsorbed on birnessite surface and their complexes with recPrP added under different sequences (Fig. 1).

Fig. (1). FT-IR spectra of A. Bir; B. Bir-Cat3; C. Bir-Cat3+recPrP;

D. Bir-Cat3-recPrP.

86 The Open Magnetic Resonance Journal, 2010, Volume 3 Russo et al.

The most characteristic bands and their assignments re-lated to samples obtained with 3 mM Cat as well as 5 mM Cat (data not shown) are summarized in Table 1.

All samples showed a broad band at 3333 cm-1

attribut-able to OH groups bound through intermolecular H bonds [28]. Spectral bands derived from vibrations of aromatic carboxylates (R-COO

-) and/or aromatic C=C structures

(1650-1620 cm-1

) were also present. FT-IR spectra revealed the presence of organic material attributed to the presence of Cat polymerization products, (1621 and 1255 cm

-1). Com-

plexes containing recPrP showed typical signals of amide I and amide II (1645 and 1520 cm

-1, respectively); a rein-

forcement of the weak signal at 1255 cm-1

was attributed to amide III. Similar FT-IR spectra were recorded for catechol–birnessite–PrP complexes obtained at 5 mM catechol [13].

CPMAS 13

C NMR Analyses

RecPrP was analysed by CPMAS 13

C NMR (Fig. 2A). The spectrum revealed a typical NMR signal pattern for pro-teins in the solid state [29]. In fact, a signal attributable to C=O groups was observed at 178 ppm, while signals at 161, 133, 120 ppm were assigned to aromatic moieties having different substitution degrees [30]. All the signals comprised between 90 and 0 ppm were attributed to C-N in peptidic bonds (80 ppm), alkyl chains (47 ppm) and methyl groups (34 ppm), respectively. The spectra of the birnessite with prion protein or catechol (Bir-recPrP, Bir-Cat5 and Bir-Cat3, respectively, Fig. 2B,C,D) appeared resolution-less, whereas those of samples obtained by interaction of birnessite, phenol and prion protein added before and after catechol polymeri-zation process (Bir-Cat3+recPrP and Bir-Cat3-recPrP, re-spectively, Fig. 2E,F) revealed the same signal pattern of recPrP alone as in Fig. (2A), but with a larger signal width.

DISCUSSION

The recPrP was completely removed from the solution by sorption or entrapment in catechol-birnessite soil-model sys-tems confirming the high affinity of prions to soil organo-

mineral colloids. The presence in the insoluble catechol-protein polymeric products of humic-like compounds and of the protein was confirmed by FT-IR signals typical of humic compounds and proteins (Table 1).

Fig. (2). CPMAS 13

C-NMR spectra of A. recPrP; B. Bir-recPrP; C.

Bir-Cat5; D. Bir-Cat3; E. Bir-Cat3+recPrP; F. Bir-Cat3-recPrP.

Table 1. Location of Relevant Indicator Bands in recPrP and Organo-Mineral Complexes and the Assignment to Functional

Groups

Wavenumber (cm-1

) Sample Assignment

3333 Cat3-Bir

Cat3-Bir-recPrP

Cat5-Bir+recPrP

OH, bonded through intermolecular H bonds

1621 Cat3-Bir Aromatic C=C, H-bonded C=O or alkenes in conjugation with C=O

1645 recPrP

Cat3-Bir-recPrP

Cat3-Bir+recPrP

C=O stretching (amide I)

1520 recPrP

Cat3-Bir-recPrP

Cat5-Bir+recPrP

N-H bending (amide II)

1255 recPrP

Cat3-Bir

Cat3-Bir-recPrP

Cat5-Bir+recPrP

C-O stretching and aromatic C=C; amide III

Interaction of a Recombinant Prion Protein The Open Magnetic Resonance Journal, 2010, Volume 3 87

Catechol transformation by birnessite started to produce soluble compounds as detected in the supernatants by UV-Vis analyses. They were more abundant in the samples with 3 mM catechol than with 5 mM catechol [13]. When the pro-tein was added before catechol polymerization (Cat-Bir-recPrP) as well as when catechol was preventively polymer-ized and partially adsorbed on birnessite surfaces (Cat-Bir+recPrP) recPrP interacted with birnessite-catechol com-plexes and was completely removed from solution, as dem-onstrated by HPLC analysis [13]. Conversely in Bir-recPrP sample, 33% of recPrP was detected as free in the super-natant. The stability of recPrP in the humic-like complexes was confirmed by the negative release of the protein after extractions with several, even strong, extractants [13].

As reported in literature, a partial recPrP abiotic degrada-tion mediated by MnO2 could be not excluded [19]. How-ever, in the presence of catechol the degradation of recPrP should have been reduced because birnessite active surfaces were covered by organic catechol polymers [31,32]. After catechol polymer deposition, a reduced birnessite activity is reasonable because the birnessite-mediated reaction is re-ported to be chemically surface-controlled and occurring by associative ligand substitution mechanism with a surface complex formation on Mn-oxide sites [22]. According to this mechanism, during the preparation of the complexes, catechol produced large polymers located at the surface of birnessite and capable of promoting the formation of large pores among the enclosed birnessite particles [15].

As also reported in Rao et al. [13], UV-Vis spectra and elemental analyses of the samples produced with 3 mM catechol solution showed a larger soluble polymeric content while samples produced with a 5 mM catechol solution had a dominant insoluble catechol polymeric fraction. The weaker bands of amide I and II observed in Cat3-Bir+recPrP than those of Cat5-Bir+recPrP confirmed their different behav-iour. In any case, the initial catechol concentration and the sequence of the addition of the protein had no effect on pro-tein immobilization that resulted always completely removed from the solution.

Acquisition of solid state (SS) NMR spectra is strongly affected by paramagnetism [33]. In fact, the presence of par-amagnetic systems (PS) alters NMR signal shape and inten-sity, thereby providing no signal at all when the amount of paramagnetic species is larger than that of the NMR investi-gated nucleus. It is reported, for example, that iron (III) in soil samples prevents reliable

13C NMR spectra acquisition

when the Fe/C mass ratio is »1 [25].

NMR signal disappearing is attributed to the interactions between the magnetic field generated by the paramagnetic centers and the applied magnetic field [34]. These interac-tions fasten both longitudinal (T1) and transversal (T2) re-laxation times so that NMR signals are both broadened and lowered [25]. In the solid state, an additional problem is the mismatching of the fundamental SS NMR condition, TCH«T1 (H), which usually guarantees obtainment of quanti-tative NMR spectra [25]. However, from a qualitative point of view, when a spectrum is acquired, the relative intensities of the various resonances do not reflect the real distribution present in the samples since some functional groups may be more affected by the presence of the paramagnetic centers

[35]. In fact, all the groups directly interacting with the par-amagnetic ions relax faster than those placed far away from PSs. As a consequence, the NMR signals of the quick relax-ing nuclei disappear more rapidly than the resonances of the remaining others.

Fig. (2A, B) reports the CPMAS 13

C NMR spectra of the recombinant prion protein before (Fig. 2A) and after (Fig. 2B) absorption on birnessite (Bir). Due to the presence of the paramagnetic manganese, all the signals of recPrP were broadened. However, a broad signal around 130 ppm due to aromatic structures and the resonance at 178 ppm due to car-boxyls were still visible. Only the signals between 90 and 0 ppm completely disappeared (Fig. 2B). A possible explana-tion for such behaviour can be related to the preferential in-teractions of recPrP with the surface area of birnessite through C-N and alkyl moieties.

The effect of natural organic matter on recPrP absorption on birnessite was estimated by analyzing the complexes ob-tained with catechol at the two different concentrations (3 and 5 mM, respectively) (see Materials and Methods). Fig. (2C, D) shows that the humic-like organic substances pro-duced by the incubation of catechol are adsorbed on birnes-site. In fact, the signals of the incubated catechol disappeared and only a very large band around 130 ppm became visible. When the incubation of catechol-birnessite mixture was con-ducted also in the presence of recPrP, added either after or before catechol polymerization, the spectra in Figs. (2E and 2F) were obtained, respectively. Both spectra resembled that in Fig. (2A), thereby indicating that only a humic-like medi-ated interaction between recPrP and birnessite was possible.

CONCLUSIONS

The interaction of recPrP with catechol-birnessite com-plexes occurred at all phenol concentrations used in the pre-sent study. Moreover, the interactions took place regardless of the sequence used to add the natural-organic-matter-like (NOM-like) system to the mineral phase. In fact, recPrP ap-peared to be bound to the birnessite through NOM-like me-diated bindings which were observed by comparing the CPMAS

13C NMR spectra of different birnessite-recPrP

complexes. FT-IR analysis contributed to confirm the pres-ence of recPrP in the insoluble polymeric products, already at the lowest catechol concentration. Both FT-IR and NMR studies were helpful to highlight the important role of humic-like substances formed by abiotic catalysis from phenolic compounds in adsorbing and/or entrapping recPrP, and the prevalent involvement of alkyl moieties rather than aromatic or carboxylic groups.

ACKNOWLEDGEMENTS

This work was founded by the European Union Project QLK4-CT-2002-02493 “TSE-Soil-Fate”. The NMR experi-ments were done at the Centro Grandi Apparecchiature (CGA) - UniNetLab – of Università degli Studi di Palermo (http://www.unipa.it/cga).

ABBREVIATIONS

recPrP = recombinant prion protein

Cat-Bir-recPrP = recPrP added before catechol polymerization

88 The Open Magnetic Resonance Journal, 2010, Volume 3 Russo et al.

Cat-Bir+recPrP = recPrP added to catechol preventively polymerized and partially adsorbed on birnessite surfaces

CPMAS 13

C NMR = cross polarization magic angle spin-ning carbon-13 nuclear magnetic resonance

FT-IR = Fourier transform infrared spectros-copy

PS = paramagnetic systems

SS NMR = solid state nuclear magnetic reso-nance

TCH = cross polarization time

T1 (H) = proton longitudinal relaxation time in the rotating frame

T1 = longitudinal relaxation time

T2 = transverse relaxation time

UV-Vis = ultraviolet visible spectrophotometry

REFERENCES

[1] Watts JC, Balachandran A, Westaway D. The expanding universe of prion diseases. PLoS Pathog 2006; 2: e26.

[2] Prusiner SB. Novel proteinaceous infectious particles cause scra-pie. Science 1982; 216: 136-44.

[3] Taylor DM. Inactivation of transmissible degenerative encephalo-pathy agents: a review. Veter J 2000; 159: 10-7.

[4] Andrèoletti O, Lacroux C, Chabert A, et al. PrP(sc) accumulation in placentas of ewes exposed to natural scrapie: influence of foetal PrP genotype and effect on ewe-to-lamb transmission. J Gen Virol 2002; 83, 2607-16.

[5] Brown P, Gajdusek DC. Survival of scrapie virus after 3 years' interment. Lancet 1991; 337: 269-70.

[6] Seidel B, Thomzig A, Buschmann A, et al. Scrapie agent (strain 263K) can transmit disease via the oral route after persistence in soil over years. PLoS One 2007; 2: 435-42.

[7] Georgsson G, Sigurdarson S, Brown P. Infectious agent of sheep scrapie may persist in the environment for at least 16 years. J Gen Virol 2006; 87: 3737-40.

[8] Vasina EN, Déjardin P, Rezaei H, Grosclaude J, Quiquampoix H. Fate of prions in soil: adsorption kinetics of recombinant unglyco-sylated ovine prion protein onto mica in laminar flow conditions and subsequent desorption. Biomacromolecules 2005; 6: 3425-32.

[9] Revault M, Quiquampoix H, BaronMH, Noinville S. Fate of prions in soil: trapped conformation of full-length ovine prion protein in-duced by adsorption on clays. Biochim Biophys Acta 2005; 1724: 367-74.

[10] Johnson CJ, Phillips KE, Schramm PT, McKenzie D, Aiken JM, Pedersen JA. Prions adhere to soil minerals and remain infectious. PLoS Pathog 2006; 2: 296-302.

[11] Rigou P, Rezaei H, Grosclaude J, Staunton S, Quiquampoix H. Fate of prions in soil: adsorption and extraction by electroelution of recombinant ovine prion protein from montmorillonite and natural soils. Environ Sci Technol 2006; 40: 1497-503.

[12] Ma X, Benson CH, McKenzie D, Aiken JM, Pedersen JA. Ad-sorption of pathogenic prion protein to quartz sand. Environ Sci Technol 2007; 41: 2324-30.

[13] Rao MA, Russo F, Granata V, Berisio R, Zagari A, Gianfreda L. Fate of prions in soil: Interaction of a recombinant ovine prion pro-

tein with synthetic humic-like mineral complexes. Soil Biol Bio-chem 2007; 39: 493-504.

[14] Polano M, Anselmi C, Leita L, Negro A, De Nobili M. Organic polyanions act as complexants of prion protein in soil. Biochem Biophys Res Commun 2008; 367: 323-9.

[15] Pucci A, D'Acqui LP, Calamai L. Fate of prions in soil: interac-tions of RecPrP with organic matter of soil aggregates as revealed by LTA-PAS. Environ Sci Technol 2008; 42: 728-33.

[16] Johnson CJ, Pedersen JA, Chappell RJ, McKenzie D, Aiken JM. Oral transmissibility of prion disease is enhanced by binding to soil particles. PLoS Pathog 2007; 3: 874-81.

[17] Miller MW, Williams ES. Prion disease: horizontal prion transmis-sion in mule deer. Nature 2003; 425: 35-6.

[18] Schramm PT, Johnson CJ, Mathews NE, McKenzie D, AikenJM, Pedersen JA. Potential role of soil in the transmission of prion dis-ease. Miner Geochem 2006; 64: 135-52.

[19] Russo F, Johnson CJ, Johnson CJ, McKenzie D, Aiken JM, Pedersen JA. Degradation of the pathogenic prion protein by a manganese mineral prevalent in soils. J Gen Virol 2009; 90: 275-80.

[20] Nannipieri P, Ceccati B, Grego S. Ecological significance of biological activity in soil. In: Bollag J-M, Stotzky G, Eds. Soil bio-chemistry. Marcel Dekker: New York 1990; Vol 6: p. 293.

[21] McKenzie RM. Manganese oxides and hydroxides. In: Dixon JB, Weed SB, Eds. Minerals in soil environments, 2nd ed. SSSA, Madison, WI 1989; p. 439.

[22] Matocha CJ, Sparks DL, Amonette JE, Kukkadapu RK. Kinetics and mechanism of birnessite reduction by catechol. Soil Sci Soc Am J 2001; 65: 58-66.

[23] Rezaei H, Marc D, Choiset Y, et al. High yield purification and physico-chemical properties of full-length recombinant allelic variants of sheep prion protein linked to scrapie susceptibility. Eur J Biochem 2000; 267: 2833-9.

[24] Eghiaian F, Grosclaude J, Lesceu S, et al. Insight into the PrPc - PrPsc conversion from the structures of antibody bound ovine prion scrapie-susceptibility variants. Proc Natl Acad Sci USA 2004; 101: 10254-9.

[25] Conte P, Spaccini R, Piccolo A. State of the art of CPMAS 13

C-NMR spectroscopy applied to natural organic matter. Prog Nucl Magn Reson Spectrosc 2004; 44: 215-23.

[26] Conte P, Piccolo A. Precise measurement of 1H 90° pulse in solid-

state NMR spectroscopy for complex and heterogeneous molecu-lar systems. Anal. Bioanal Chem 2007; 387: 2903-9.

[27] Brown DE, Fully automated baseline correction of 1D and 2D NMR spectra using Bernstein algorithm, J Magn Reson A 1995; 114: 268-70.

[28] Schnitzer, M. Characterization of humic constituents by spectros-copy. In McLaren AD, Skujins J, Eds. Soil biochemistry. Marcel Dekker: New York 1971; Vol. 2: p. 60.

[29] Watts A, Straus SK, Grage SL, Kamihira M, Lam YH, Zhao X. Membrane protein structure determination using solid-state NMR. Methods Mol Biol 2004; 278: 403-73.

[30] Wilson MA. NMR Techniques and applications in geochemistry and soil chemistry. 1st ed. London: Pergamon Press 1987.

[31] Naidja A, Huang PM, Bollag J-M. Comparison of the reaction products from the transformation of catechol catalysed by birnes-site or tyrosinase. Soil Sci Soc Am J 1998; 62: 188-95.

[32] Majcher EH, Chorover J, Bollag J-M, Huang PM. Evolution of CO2 during birnessite-induced of

14C-labeled catechol. Soil Sci

Soc Am J 2000; 64: 157-63. [33] Preston, CM. Application of NMR to soil organic matter analysis:

history and prospects. Soil Sci 1996; 161: 144-66. [34] Keeler, J. Understanding NMR spectroscopy. Chichester, Eng-

land: Wiley & Sons 2009. [35] Smernik RJ, Oades MJ. Paramagnetic effects on solid state car-

bon-13 nuclear magnetic resonance spectra of soil organic mat-ter. J Environ Qual 2002; 31: 414-20.


© Russo et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

CPMAS 13

C NMR Characterization of Leaves and Litters from the Reafforestated Area of Mustigarufi in Sicily (Italy)

Pellegrino Conte*,1, Claudio De Pasquale

1, Etelvino H. Novotny

2, Gianluca Caponetto

1,

Vito Armando Laudicina1, Maurizio Ciofalo

1, Michele Panno

1, Eristanna Palazzolo

1,

Luigi Badalucco1 and Giuseppe Alonzo

1

1Dipartimento di Ingegneria e Tecnologie Agro-Forestali (ITAF), Università degli Studi di Palermo, 90128 Palermo,

Italy

2Embrapa Solos, Rua Jardim Botânico, 1024, CEP 22460-000, Rio de Janeiro, RJ, Brazil

Abstract: Reafforestation is generally based on the planting of exotic fast growing tree species suitable for adapting to

even harsh environments. Once the introduced plants ameliorate soil conditions, they can be progressively replaced by au-

tochthonous plant species. Reafforestation is applied worldwide. However, only few studies on the effect of reafforesta-

tion on lands from Mediterranean regions are available. This paper reports the characterization by cross polarization 13

C

NMR spectroscopy of fresh leaves and superficial litters from a reafforestated area in central Sicily (Italy). NMR assign-

ment is attempted. A differentiation among the molecular systems within leaves and litters is also done on the basis of

NMR assessment. Results showed that the main differences among the leaves of four forest trees (two eucalyptus spp.,

one cypress sp. and one pine sp.) occur in the distribution of the aromatic and alkyl carbons. In particular, the alkyl moie-

ties in the eucalyptus spp. leaves were attributed to branched structures belonging to the eucalyptus oil, whereas linear

fatty acids were more representetive in the NMR spectra of pine and cypress leaves. In addition, the aromatic carbons of

the conifer leaves were assigned not only to lignin- and tannin-like structures, but also to common olefin carbons in un-

saturated fatty acids and abietic acid-like systems. The spectra of the litters resembled, as expected, those of the leaves.

However, the presence of very large carbohydrate NMR signals suggested that degradation processes were still ongoing in

litters. A comparative evaluation of CPMAS 13

C NMR spectra was done by applying principal component analysis. This

paper confirmed the suitability of CPMAS 13

C NMR spectroscopy in evaluating the differences among natural bio-masses

which are the major nutrient sources for soil micro-organisms and the main input for humification processes.

Keywords: NMR, leaves, litters, reafforestation, degraded lands, soils.

INTRODUCTION

Intensive land use and management as well as inappro-priate land practices have negative impacts on natural re-sources such as waters, soils, atmosphere, plants and animals due to nutrients decline, erosion and contamination [1, 2]. Therefore, land recovery and restoration are desirable efforts for the improvement of the ecosystem health status and sustainability [3]. Recovery and restoration are, however, complex and long processes which imply many tasks includ-ing environmental evaluations (such as pollution and/or ero-sion extents), strategic policies for managing degraded eco-systems, and technical accomplishments for rebuilding physical and biological ecosystem conditions [3].

One of the most used practices for restoration of the natu-ral ecosystems on abandoned lands (i.e. lands which ex-hausted their natural potential for human survival) is the re-afforestation [1, 4]. It consists in the planting of fast growing exotic tree species suitable to adapt to even harsh conditions,

*Address correspondence to this author at the Dipartimento di Ingegneria e

Tecnologie Agro-Forestali (DITAF), Università degli Studi di Palermo, v.le

delle Scienze 13, ed. 4, 90128, Palermo, Italy; Tel: 00390917028145; Fax:

0039091484035; E-mail: [email protected]

thus providing early substrates for the microbial re-colonization. Then, local species can be added by enrichment planting to improve microclimate and soil conditions and to create favorable circumstances for other indigenous species invasion [5].

Reafforestation compensates for CO2 emissions to the atmosphere through accumulation and transformation of natural organic matter into soils [2], promotes soil hydro-logical properties due to its impact on soil texture [4], favors mitigation of soil temperatures due to the vegetation cover [4] and re-establishes nutrient cycles thereby affecting de-velopment of soil microbial biomasses [6, 7]. Moreover, reafforestation reduces soil erosion and may have a great social impact since it increases economic potentials through enhancement of working possibilities [8].

Land restoration through reafforestation is achieved worldwide by establishing mainly eucalyptus and pine trees [1]. In fact, such plantations take deep roots and grow vigor-ously even in low fertility lands [4].

Only few papers up to now dealt with the effects of reaf-forestation on Mediterranean degraded lands [7, 9, 10]. For this reason, we have started the monitoring of the effects of reafforestation in soils of semi-arid Mediterranean areas of central Sicily (Italy) where desertification processes, related

90 The Open Magnetic Resonance Journal, 2010, Volume 3 Conte et al.

to intensive land uses, were ongoing. One of these areas is the regional forest property located in Mustigarufi, nearby Caltanissetta. Here eucalyptus, cypress and pine trees were planted since the late fifties and the beginning of the sixties.

In the present paper, we have concentrated our attention primarily on the molecular characterization of the fresh leaves and the superficial layers of litters of Mustigarufi for-est. This because, both of them represent the major nutrient sources for soil microbial biomass and the main input for humification processes, which are very important to restore soil quality in reafforestated areas.

To reach our goal we applied carbon-13 solid state nu-clear magnetic resonance spectroscopy with cross polariza-tion and magic angle spinning (CPMAS

13C NMR). CPMAS

13C NMR spectroscopy provides both qualitative fingerprint-

ing of carbonaceous materials and quantitative measure-ments on the relative content of the different molecular moieties in very complex organic mixtures [11]. Therefore, we report the assignment of the CPMAS

13C NMR spectra of

fresh leaves and litter superficial layers from Mustigarufi forest as well as a comparison among the molecular charac-teristics of these materials. This study represents an unavoid-able initial step for the integrated study of the C turnover within a reafforested semi-arid Mediterranean area.

EXPERIMENTAL

Leaf and Litter Samples

Eucalyptus camaldulensis Dehnh. (EC), Eucalyptus occidentalis Endl. (EO), Cupressus sempervirens L. (CI) and Pinus halepensis Mill. (PI) are tree species growing in Mustigarufi reafforested area which is located nearby San Cataldo (Cal- tanissetta, Sicily, Italy, 37°33’N, 13°55’E) on a hill placed at 470 m a.s.l.. In order to perform the sampling of fresh leaves and superficial litters, four distinct 20 x 20 m homogeneous squares, each including only one tree species, were selected. Composite samples of fresh leaves of each tree species were collected at random from three distinct canopies, whereas composite samples from superficial (0-3 cm) litters consisted of six sub-samples exploring the 20 x 20 m squares. It was not possible to collect the litter under Cu- pressus sempervirens L. since it appeared both heterogeneous and quantitatively insignificant. Leaves and litters, without any pre-treatment, were dried at 70°C in a vent-oven for 72 hours. Then, they were powdered in an ultra centrifu- gal rotor ZM200 Retsch

® mill equipped with a 1 mm sieve

in order to obtain solid samples for the NMR analyses.

CPMAS 13

C NMR Spectroscopy

CPMAS 13

C NMR measurements were performed on a Bruker Avance-II 400 spectrometer (Bruker Biospin, Milan, Italy) operating at 100.6 MHz on carbon-13 and equipped with a 4 mm standard bore solid state probe. Samples were packed into 4 mm zirconia rotors with Kel-F caps and the rotor spin rate was set at 13000 ± 2 Hz. A spectral width of 29761.90 Hz centered at 10061.78 Hz, an optimum contact time of 1 ms chosen after evaluation of variable contact time experiments [12], a recycle delay of 2 s, 2 k data points over an acquisition time of 35 ms and a RAMP sequence, to ac-count for inhomogeneities of the Hartmann-Hahn condition at high rotor spin rates [11], were used. 700 scans were ac-

cumulated to obtain all the spectra. The 1H 90° pulse was

calibrated on each leaf and litter sample with the pulse se-quence described in [13] at an attenuation level of -2.4 dB. 1H 90° pulse length varied within the interval 3.60-3.85 μs

depending on the sample under analysis. Spectra acquisition was achieved with Bruker Topspin

® 2.0. Data elaboration

was done with MestRe-C software (Version 4.9.9.9, Mestre-lab Research, Santiago de Compostela, Spain). The free in-duction decays (FIDs) were transformed by applying first a 2 k zero filling, then a line broadening of 50 Hz and finally an automatic baseline correction with a 3rd order polynomial and Bernstein algorithm [14]. All the spectra were divided in the following regions whose assignment is fully discussed later: 0-45 (alkyl C), 45-60 (O/N alkyl C), 60-90 (O alkyls in carbohydrates), 90-120 (acetal and lignin C), 120-160 (aro-matic C) and 160-180 (-COOH groups) ppm. The absolute areas (Ai) of each interval was measured and referred to the total absolute area of each spectrum (AT) as obtained by in-tegrating the -50 – 230 ppm interval. The -50 – 230 ppm region was accounted for to consider the effect of the spec-tral noise on the quantitative evaluation of the NMR spectra.

Principal Component Analysis

Principal Component Analysis (PCA) was carried out us-ing the CPMAS

13C NMR spectra obtained, after area nor-

malization and mean-centering of the data. To improve the interpretability of the loadings, a Varimax rotation was per-formed. The goal of this strategy was to obtain a clear pat-tern of loadings, i.e., factors marked by high loadings for some variables and low loadings for others. The model vali-dation was carried out using two different methods: full cross-validation, and segmented cross-validation with two samples per segment picked at random. The difference in the variance between the calibration and validation models was less than 5%. This analysis was performed using The Un-scrambler software (CAMO Software AS, Oslo, Norway).


Qualitative Interpretation of the Leaf NMR Spectra

Fig. (1) reports the CPMAS 13

C NMR spectra of the leaves sampled for the present study. Attribution of spectral regions is also indicated. According to literature [11, 16-19], six different intervals were recognized. The first one, be-tween 0 and 45 ppm, was attributed to alkyl systems. The most important signals in this interval were located at 26, 30 and 32 ppm in the spectra of Cupressus sempervirens L. (CI, Fig. 1A) and Pinus halepensis Mill. (PI, Fig. 1B) and at 17, 26, 30, 32, 39 and 42 ppm in the spectra of Eucalyptus camaldulensis Dehnh. (EC, Fig. 1C) and Eucalyptus occi-dentalis Endl. (EO, Fig. 1D).

The resonance at 17 ppm was assigned to methyl groups which terminate alkyl chains [16]. Among the other remaining alkyl signals (26, 30, 32, 39 and 42 ppm), those at 26, 30 and 32 ppm, visible in all the leaf spectra (Fig. 1), can be attributed to linear methylene (-CH2-) chains [16, 17] belonging to lipids, cutin-like structures and other aliphatic bio-moieties [20]. The last two, at 39 and 42 ppm can be assigned to secondary methyne carbons (-CH-, signal at 39 ppm) and to fully substituted quaternary carbons (CR4, signal at 42 ppm) [16, 19]. It is likely that the signals at 39 and 42 ppm can be generated by carbons in chlorophyll-like struc-

NMR Characterization of Biomasses from a Reafforestated Italian Area The Open Magnetic Resonance Journal, 2010, Volume 3 91

tures or in molecules belonging to eucalyptus-oil (a complex mixture of terpenoids) that is usually present into eucalyptus trees [21].

The second spectral interval between 45 and 60 ppm is traditionally attributed to nitrogenated and oxygenated alkyl systems (Fig. 1). Three signals positioned at 48, 53 and 56 ppm were evidenced in the spectra of Fig. (1). That at 48 ppm can be assigned to N-alkyl carbons in amino acids [18, 22]. The shoulder at 53 ppm in Figs. (1C and 1D) and the peak at 56 ppm in Figs. (1A to 1D), can be both attributed to O-alkyl groups such as methoxyls in lignin-like structures (56 ppm) and –CH2O– systems into branched molecules as those in the eucalyptus-oil [18].

The region comprised in the chemical shift interval 60-90 ppm (Fig. 1) is indicative of carbohydrates with the largest contribution due to celluloses and hemicelluloses [18]. In particular, the resonances at 63 and 65 ppm were due to car-bon 6 in amorphous and crystalline celluloses, respectively

[23], while the intensity at 72 ppm was assigned to the car-bons in the positions 2, 3 and 5 regardless of the cellulose nature (either crystalline or not) [24]. Carbon 4 appeared between 80 and 90 ppm.

Here, it generated a shoulder at around 83 ppm due to amorphous cellulose, hemicellulose and cellulose oligomers [25] and a signal at 88 ppm that was assigned to the ordered forms of celluloses on fibril surfaces and to “in core” para-crystalline celluloses, whose nature is still uncertain [26]. Among all the spectra (Fig. 1), the only one where the sig-nals at 63/65 and 83/88 ppm were clearly identifiable was generated by the pine leaves (Fig. 1B), thereby suggesting that they may contain a larger variety of cellulose forms.

Signals of carbohydrates are also observed in the range 90-120 ppm where the large resonance at 105 ppm was at-tributed to C1 of cellobiose units into cellulose I, and that at around 100 ppm was assigned to the acetal carbons in the xylane systems of hemicelluloses [26]. The 90-120 ppm in-terval contains also other signals such as a weak one due to common olefin carbons at 116 ppm in the spectra of CI (Fig. 1A) and PI (Fig. 1B), and a resonance at 109 ppm associable to acetal C in cellulose II.

The spectral region, that traditionally is assigned to aro-matic systems, is included between 120 and 160 ppm [16]. Three main peaks at 130, 144 and 154 ppm were revealed in the spectra of the leaves from CI (Fig. 1A) and PI (Fig. 1B), whereas only a signal at 137 ppm and a large resonance at 144 ppm appeared in the spectra of EC (Fig. 1C) and EO (Fig. 1D). All these signals are usually attributed to lignin systems [17]. In particular, p-hydroxyphenol derivative structures are assumed to give a signal at around 130 ppm, whereas O-aryl carbons from guaiacyl- and syringyl-units may give resonances at 137, 144 and 154 ppm [17]. Based on this simple attribution, we can conclude that the differ-ences between the leaves from the two coniferous trees and the two eucalyptus plants were due to a different concentra-tion of p-hydroxyphenol-, guaiacyl- and syringyl- structures. However, cell composition of the needles from pinus and cypress has been already well studied and it has been clari-fied that they contain large concentrations of acid resins made by long chain unsaturated acids and abietic acid de-rivatives [27, 28]. According to web data bases (such as http://riodb01.ibase.aist.go.jp/sdbs/) and to literature data [29], we attributed the resonance at 130 ppm in the spectra of CI (Fig. 1A) and PI (Fig. 1B) either to cis-mono-unsaturated fatty acids or to abietic acid-like structures (i.e. sp

2 carbons).

The signal at 144 ppm can be, consequently, also assigned to the C13 in dehydro-phenanthrene systems [29]. According to our interpretation, the differences in the 120-160 ppm inter-val in the leaf spectra of the coniferous plants (Fig. 1A and 1B) and of the two eucalyptus trees (Fig. 1C and 1D) cannot be due only to a different distribution of p-hydroxyphenol-, guaiacyl- and syringyl- structures, but also to the presence of acid resins which are less abundant in the leaf samples from EC and EO.

The last spectral region between 160 and 180 ppm was attributed to –COOH and amide groups.

Qualitative Interpretation of the Litter NMR Spectra

Signal attribution of the litter spectra (Fig. 2) is similar to that of the leaves. It has been already described in the para-

Fig. (1). CPMAS 13

C NMR spectra of the leaves of Cupressus

sempervirens L. (A), Pinus halepensis Mill. (B), Eucaliptus

camaldulensis Dehnh. (C) and Eucaliptus occidentalis Endl. (D).

Attribution of the spectral intervals is also reported.


graph above. As a general remark, however, it must be no-ticed that all the litter NMR spectra are dominated by the signals of carbohydrates. According to Lemma et al. [18] and Hopkins et al. [30], this feature reveals that the litters are still in the early stages of decomposition. The comparison of Figs. (1 and 2) shows some dissimilarities between fresh leaves and leaf litters. Signal at 65 ppm (which appears as a shoulder of the signal at 72 ppm in the leaf spectra) and a different distribution of the aromatic (120-160 ppm) carbons can be observed. The signal at 65 ppm was assigned to C6 in crystalline cellulose (see above). The higher resolution of such signal can be explained considering that the first steps of the litter decomposition mechanisms involve the degrada-tion of the amorphous celluloses [18]. As a consequence, a sharpening of the signals in the O-alkyl spectral region, which implies a better separation between the resonances at 72 and 65 ppm, is achieved. This hypothesis is further sup-ported taking into account the disappearance of the signal at 109 ppm (in EO and EC litters) assigned to acetal carbons in cellulose II. Litter decomposition is also related to a varia-tion of the nature of the aromatic carbons due to degradation of tannin- and lignin-like components and to a general reso-lution-loss of the signals ranging in the region of the alkyl moieties. Transformations of the aromatic structures turned in the coalescence of the signals at 144 and 154 ppm which was observed in the spectrum of the pinus litter (Fig. 2A) [18]. The same signals (at 144 and 154 ppm) were, con-versely, observed in the spectrum of EC litter (Fig. 2B). They were not present in the spectrum of the EC leaves (Fig. 1C). The presence of these two signals can be attributed to lignin-residues which were co-sampled with the leaf litter.

Statistical Comparison of NMR Spectra

In order to evaluate the capability of the CPMAS 13

C NMR spectroscopy in revealing differences among leaf and litter samples, the NMR spectra were used as input data for principal component analysis (PCA). PCA is already known to be a powerful tool for the recognition of similarities and dissimilarities in NMR spectra of very complex systems such as in foods [31-36], natural organic matter [25, 37-40], living organisms [41], and environmental compartments [42-45]. The basic idea of PCA is to reduce the number of variables into just few and to seek linear combinations of those vari-ables explaining most of the variability [15].

In the present study, PCA reduced the number of vari-ables to only 3 (PC1, PC2 and PC3) which accounted for 83% of the total variance (Fig. 3). Both PC1 and PC3 indi-cate decomposition of Eucaliptus leaves. The sole PC1, which accounted for the 35% of the variance, revealed that Eucaliptus leaves had the largest scores as compared to the Eucaliptus litters (Fig. 3A). Moreover, this component was also able to differentiate among tree species due to the score values which varied as EO>EC PI>CI.

Fig. (3). Scores of rotated (varimax) PCA from CPMAS 13

C NMR

spectra. CI=Cupressus sempervirens L.; PI=Pinus halepensis Mill.;

EO=Eucaliptus occidentalis Endl.; EC=Eucaliptus camaldulensis

Dehnh. A. PC1 vs PC2 plot; B. PC1 vs PC3 plot.

PC1 was characterised by high positive loadings in the region of methyl groups (< 30 ppm), with clear peaks at

Fig. (2). CPMAS 13

C NMR spectra of the litters from Pinus

halepensis Mill. (A), Eucaliptus camaldulensis Dehnh. (B) and

Eucaliptus occidentalis Endl. (C).


16.8, 20.7, 24.0, 28.3 ppm and shoulders at around 11.9 and 18.4 ppm (Fig. 4). According to Fig. (3A), methyl systems are more abundant in EO leaves (higher scores) which, in turn, are subjected to an easier decomposition than the leaves from the other tree species. PC1 also showed positive load-ings (Fig. 4) in the region of methyne and quaternary groups (39.5 and 41.8 ppm, respectively). Chlorophyll-like and oil-like structures (see above) appeared to resonate at those chemical shift values, thereby confirming that such struc-tures are peculiar in Eucaliptus trees.

Fig. (4). Loadings of first rotated (varimax) Principal Component

from PCA of CPMAS 13

C NMR spectra.

The negative loadings for the signals at 32.8 (weak peak, crystalline poly-methylene) and at 35.9 ppm are an indica-tion that during the initial stage of decomposition a relative accumulation of crystalline forms of poly-methylene may occur.

The chemical shift interval comprising the signals at 54.7, 52.4, 49.6, and 47.5 ppm (Fig. 4) and the region 154.7 and 152.4 ppm revealed opposite loading signs. In fact, the former region, generally attributable either to O-alkyl or N-alkyl carbons, showed positive loadings, whereas the second one, attributed to O-aryl groups, was negative (Fig. 4). Due to this inverse relationship, it can be concluded that the main organic systems which are subjected to decomposition are N-alkyls in peptides.

In addition, Fig. (4) reports negative loadings for the peaks due to crystalline cellulose (from 106 to 70.4 ppm) and positive loadings for the peak of amorphous cellulose (63.7 ppm), thereby revealing an accumulation of the crystal-line cellulose, while the amorphous one is decomposed.

Finally, the aryl region showed negative loadings, proba-bly due to a relative accumulation of lignin during the de-composition.

PC3 provide additional information concerning the al-terations of Eucalyptus leaves (Fig. 3B). In fact, Fig. (5) shows positive loadings from the chemical groups which are more abundant in Eucalyptus leaves and that sharply de-crease in litter samples (i.e. 70 and 32 ppm). Conversely, the negative signals at 104, 75 and 21 ppm (Fig. 5) belong to the groups which accumulate in the litter.

PC2 accounted for 29% of the total variance (Fig. 3A). It is characterized by positive loadings in the region 60-90 ppm (data not reported). Its scores clearly reveal decomposition of amorphous cellulose in PI-leaves while the crystalline one is accumulated.

Fig. (5). Loadings of third rotated (varimax) Principal Component

from PCA of CPMAS 13

C NMR spectra.

CONCLUSIONS

The molecular characterization by solid state 13

C NMR spectroscopy of leaves and litters from a reafforestated area in Sicily (Italy) is reported. Leaves from two different euca-lyptus plants and two conifers (pine and cypress) appeared significantly discriminated in the alkyl and aromatic NMR regions. In fact, the eucalyptus leaves were the richest in alkyl systems, whereas the pine and cypress leaves resulted to have the largest content of aromatic carbon. The aromatic C region in the leaves of the conifers was assigned not only to lignin- and tannin-like structures as reported in literature [17, 20], but also to common olefin carbons in unsaturated fatty acids (signal at 130 ppm) and abietic acid-like systems. This suggested that the differentiation between the leaves of the eucalyptus trees and the two conifers cannot be based only on p-hydroxyphenol, guaiacyl- and syringyl- deriva-tives, but also on the presence of acid resins which are well known to be very abundant in conifer leaves.

The spectra of the leaf litters were dominated by the car-bohydrate signals, thereby revealing that first stages of de-composition were ongoing [30].

Principal component analysis revealed that the initial stages of leaves decomposition leads to the preferential de-composition of methyl, methyne and quaternary C, N-alkyl and amorphous cellulose. On the other hand, this process leads to a relative accumulation of lignin and crystalline cel-lulose and crystalline poly-methylene.

CPMAS 13

C NMR spectroscopy turned out to be a very powerful technique for the molecular characterization of leaves and litters which, in turn, are very important for the development of microbial communities in reafforestated soils.


ACKNOWLEDGEMENTS

The NMR experiments were done at the Centro Grandi Apparecchiature (CGA) - UniNetLab – of Università degli Studi di Palermo (http://www.unipa.it/cga)

REFERENCES

[1] Kilian W. Forest site degradation – temporary deviation from the

natural site potential. Ecol Eng 1998; 10: 5-18.

[2] Cerli C, Celi L, Kaiser K, et al. Changes in humic substances along

an age sequence of Norway spruce stands planted on former agri-

cultural land. Org Geochem 2008; 39: 1269-80.

[3] Burger J. Environmental management: integrating ecological

evaluation, remediation, restoration, natural resource damage as-

sessment and long-term stewardship on contaminated lands. Sci

Total Environ 2008; 400: 6-19.

[4] Zhou GY, Morris JD, Yan JH, Yu ZY, Peng SL. Hydrological

impacts of Reafforestation with eucalyptus and indigenous species:

a case study in southern China. Forest Ecol Manag 2002; 167: 209-

22.

[5] Parham WE. Improving degraded land: promising experiences

from South China. Honolulu (USA): Bishop Museum Press 1993.

[6] Behera N, Sahani U. Soil microbial biomass and activity in re-

sponse to Eucalyptus plantation and natural regeneration on tropi-

cal soil. Forest Ecol Manag 2003; 174: 1-11.

[7] Alguacil MM, Caravara F, Roldàn A. Changes in rhizosphere mi-

crobial activity mediated by native or allochthonous AM fungi in

the reafforestation of a Mediterranean degraded environment. Biol

Fertil Soils 2005; 41: 59-68.

[8] Hunter IR, Hobley M, Smale P. Afforestation of degraded land –

pyrrhic victory over economic, social and ecological reality? Ecol

Eng 1998; 10: 97-106.

[9] Diaz E, Roldàn A. Effects of reafforestation techniques on the

nutrient content, photosynthetic rate and stomatal conductance of

Pinus halepensis seedings under semiarid conditions. Land Degrad

Dev 2000; 11: 475-86.

[10] Le Houreou HN. Restoration and rehabilitation of arid and semiarid

Mediterranean ecosystems in North Africa and West Asia: a re-

view. Arid Soil Res Rehabil 2000; 14: 3-14.

[11] Conte P, Spaccini R., Piccolo A. State of the art of CPMAS 13C

NMR spectroscopy applied to natural organic matter. Prog Nucl

Magn Res Spectrosc 2004; 44: 215-23.

[12] Conte P, Piccolo A, Van Lagen B, Buurman P, de Jager PA. Quan-

titative aspects of solid state 13C-NMR spectra of humic sub-

stances from soils of volcanic systems. Geoderma 1997; 80: 327-

38.

[13] Conte P, Piccolo A. Precise measurement of 1H 90° pulse in solid-

state NMR spectroscopy for complex and heterogeneous molecular

systems. Anal Bioanal Chem 2007; 387: 2903-09.

[14] Brown DE. Fully automated baseline correction of 1D and 2D

NMR spectra using Bernstein algorithm. J Magn Res A 1995; 114:

268-70.

[15] Brereton RG. Chemometrics. Data analysis for the laboratory and

chemical plant. John Willey & Sons: West Sussex (England) 2003.

[16] Wilson MA. NMR techniques and applications in geochemistry

and soil chemistry. Oxford (UK): Pergamon Press 1987.

[17] Roscoe R, Buurman P, Van Lagen B. Transformations in occluded

light fraction organic matter in a clayey oxisol; evidence from 13C-

CPMAS-NMR and 13C signature. R Bras Ci Sol 2004; 28: 811-

18.

[18] Lemma B, Nilsson I, Kleja DB, Olsson M, Knicker H. Decomposi-

tion and substrate quality of leaf litters and fine roots from three

exotic plantations and a native forest in the southwestern highlands

of Ethiopia. Soil Biol Biochem 2007; 39: 2317-28.

[19] Skjemstad JO, Frost RL, Barron PF. Structural units in humic acids

from south-eastern Queensland soils as determined by 13C NMR

spectroscopy. Aust J Soil Sci 1983; 21: 539-47.

[20] Nierop KGJ, van Lagen B, Buurman P. Composition of plant tis-

sues and soil organic matter in the first stages of a vegetation suc-

cession. Geoderma 2001; 100: 1-24.

[21] Taiz L, Zeiger E. Plant physiology. 4th ed. Sunderland (MA,

USA): Sinauer Associates Inc. 2006.

[22] Quideau SA, Chadwick OA, Benesi A, Graham RC, Anderson MA.

A direct link between forest vegetation type and soil organic matter

composition. Geoderma 2001; 104: 41-60.

[23] Maunu S, Liitiä T, Kauliomäki S, Hortling B, Sundquist J. 13C

CPMAS NMR investigations of cellulose polymorphs in different

pulps. Cellulose 2000; 7: 147-59.

[24] Taylor RE, French AD, Gamble GR, et al. 1H and 13C solid-state

NMR of Gossypium barbadense (Pima) cotton. J Mol Struct 2008;

878: 177-84.

[25] Wickholm K, Hult E-L, Larsson PT, Iversen T, Lennholm H. As-

signment of non-crystalline forms in cellulose I by CP/MAS 13C

NMR spectroscopy. Cellulose 8: 139-48.

[26] Larsson PT, Hult E-L, Wickholm K, Pettersson E, Iversen T.

CPMAS 13C-NMR spectroscopy applied to structure and interac-

tion studies on cellulose I. Solid State Nucl Magn Reson 1999; 15:

31-40.

[27] Zinkel DF, Magee TV. Resin acids of Pinus ponderosa needles.

Phytochemistry 1991; 30: 845-48.

[28] Arrabal C, Cortijo M. Fatty and resin acids of Spanish Pinus pinas-

ter Ait. Subspecies. J Am Oil Chem Soc 1994; 71: 1039-40.

[29] Smith WR. The 13C NMR spectrum of abietic acid and its methyl

ester. Org Magn Res 1978; 11: 427-28.

[30] Hopkins DW, Badalucco L, English LC, Meli SM, Chudek JA,

Ioppolo A. Plant litter decomposition and microbial characteristics

in volcanic soils (Mt Etna, Sicily) at different stages of develop-

ment. Biol Fertil Soils 2007; 43: 461-9.

[31] Andreotti G, Trivellone E, Lamanna R, Di Luccia A, Motta A.

Milk identification of different species: 13C-NMR spectroscopy of

triacylglycerols from cows and buffaloes’ milks. J Dairy Sci 2000;

83: 2432-37.

[32] Brescia MA, Caldarola V, De Giglio A, Benedetti D, Fanizzi FP,

Sacco A. Characterization of the geographical origin of Italian red

wines based on traditional and nuclear magnetic resonance spec-

trometric determinations. Anal Chim Acta 2002; 458, 177-86.

[33] Bertram HC, Wiking L, Nielsen JH, Andersen HJ. Direct meas-

urement of phase transitions in milk fat during cooling of cream —

a low-field NMR approach. Int Dairy J 2005; 15: 1056-63.

[34] Bertram HC, Straadt IK, Jensen JA, Aaslyng MD. Relationship

between water mobility and distribution and sensory attributes in

pork slaughtered at an age between 90 and 180 days. Meat Sci

2007; 77: 190-5.

[35] Fujiwara M, Ando I, Arifuku K. Multivariate analysis for 1H-NMR

spectra of two hundred kinds of tea in the world. Anal Sci 2006,

22: 1307-14.

[36] Conte P. 1H NMR spectroscopy with multivariate statistical analy-

sis as a tool for a rapid screening of the molecular changes occur-

ring during micro-oxygenation of an Italian red wine. Open Magn

Reson J 2008; 1: 77-80.

[37] Novotny EH, Knicker H, Martin-Neto L, Azeredo RBV, Hayes

MHB. Effect of residual vanadyl ions on the spectroscopic analysis

of humic acids: a multivariate approach. Eur J Soil Sci 2008; 59:

439–44.

[38] Thomsen M, Lassen P, Dobel S, Hansen PE, Carlsen L, Mogensen

BB. Characterisation of humic materials of different origin: a mul-

tivariate approach for quantifying the latent properties of dissolved

organic matter. Chemosphere 2002; 49: 1327-38.

[39] Peuravuori J. NMR Spectroscopy study of freshwater humic mate-

rial in light of supramolecular assembly. Environ Sci Technol

2005; 39: 5541-9.

[40] Sachse A, Henrion R, Gelbrecht J, Steinberg CEW. Classification

of dissolved organic carbon (DOC) in river systems: influence of

catchment characteristics and autochthonous processes. Org Geo-

chem 2005; 36: 923-35.


[41] Alcantara GB, Honda NK, Castro FMM, Gilberto FA. Chemome-

tric analysis applied in 1H HR-MAS NMR and FT-IR data for

chemotaxonomic distinction of intact lichen samples. Anal Chim

Acta 2007; 595: 3-8.

[42] Singh KP, Malik A, Mohan D, Sinha S, Singh VK. Chemometric

data analysis of pollutants in wastewater - a case study. Anal Chim

Acta 2005; 532: 15-25.

[43] Ohno T, Fernandez IJ, Hiradate IJ, Sherman JF. Effects of soil

acidification and forest type on water soluble soil organic matter

properties. Geoderma 2007; 140: 176-87.

[44] Albrecht R, Ziarelli F, Alarcón-Gutiérrez E, Le Petit J, Terroma G,

Perissol C. 13C solid-state NMR assessment of decomposition pat-

tern during co-composting of sewage sludge and green wastes. Eur

J Soil Sci 2008; 59: 486-96.

[45] Nordon A, McGill CA, Littlejohn D. Process NMR spectrometry.

Analyst 2001; 126; 260-72.

Received: June 15, 2009 Revised: December 04, 2009 Accepted: December 07, 2009

© Conte et al.; Licensee Bentham Open.




1874-7698/10 2010 Bentham Open

Open Access

13C-NMR Chemical Shift Databases as a Quick Tool to Evaluate Structural

Models of Humic Substances

Christian Nyrop Albers*,a,b

and Poul Erik Hansen*,a

aDepartment Science, Systems and Models, Roskilde University, Universitetsvej 1, DK-4000, Roskilde, Denmark

bDepartment Geochemistry, Geological Survey of Denmark & Greenland, Ø. Voldgade 10, DK-1350, Copenhagen,

Denmark

Abstract: Models for humic and fulvic acids are discussed based on 13

C liquid state NMR spectra combined with results

from elemental analysis and titration studies. The analysis of NMR spectra is based on a full reconstruction of the NMR

spectrum done with help of 13

C-NMR data bases by adding up chemical shifts of all substructures from the proposed

models. A full reconstruction makes sure that all carbons are accounted for and enables on the negative side to discuss

structural elements identified from recorded spectra of humic substances that cannot be observed in the simulated spec-

trum. On the positive side missing structural elements in the models can be suggested. A number of proposed structures

for humic and fulvic acids are discussed based on the above analysis.

Keywords: Humic acid, fulvic acid, 13

C NMR, spectral reconstruction.

INTRODUCTION

Humic substances (HS) are one of the most abundant or-ganic materials on earth. They represent 30-75% of soil or-ganic matter (SOM) [1, 2] and the majority of dissolved or-ganic matter (DOM) in both fresh and salt waters. Tradition-ally, HS from SOM has been subdivided into three fractions; Fulvic acids (FA) soluble at all pHs, Humic acids (HA) soluble at basic pHs and humin, not soluble in aqueous solu-tions. In DOM, only FA and HA, and often mainly FA, rep-resents HS [3]. Both HA and FA play a major role in binding and, hence, fate of both organic compounds, including pol-lutants like pesticides and polyaromatic hydrocarbons [4-7], and inorganic materials, especially cations [8]. The capacity for binding organic and inorganic pollutants as well as many other properties of HS is determined by the presence of vari-ous structural elements. Among those, aromatic structures have been shown to be important for binding of aromatic pollutants [5, 6] while carboxylic acid groups are important for the complex binding of cations [8]. The molecular struc-ture of HS is very variable and complex, and despite many years of research, no agreement for a common structure has been reached. There seems to be a general agreement that no single structure can be found. However, during the years very different representative models for HA and FA have been proposed (see e.g. Figs. (2 & 3) and references therein). These models have included very different structural subunits as well as similar structural elements in very differ-ent proportions. This variety of models and substructures can be highly confusing if one seeks inspiration in the literature e.g. to understand and predict interactions between a HS-fraction and some pollutant as well as understanding other

*Address correspondence to these authors at the Department Science,

Systems and Models, Roskilde University, Universitetsvej 1, DK-4000,

Roskilde, Denmark; Tel: +45 46742432; Fax: +45 46743011;

E-mails: [email protected], [email protected]

properties of HS. One possibility to explain the variety of structural models could be that they represent different envi-ronmental compartments and hence different parent material. Principal component analysis (PCA) and similar techniques have proven quite useful in distinguishing between origin based on measured functional groups and other parameters [9, 10], but in these studies it was found that while FA and HA fractions can easily be separated based on structural characteristics, the same fraction obtained from different soils are not so different. Soil and lignite HAs were clearly separated, though, so the origin does have an effect as well. The purpose of the present paper is to suggest ways of char-acterizing different fractions of humic substances and humic fractions from different environments using liquid state NMR and

13C NMR databases and to verify the validity of

the previously proposed models by such method.

In characterizing the complex humic substances it is ob-viously advantageous to use as many descriptors as possible. Some of these descriptors may be obtained using NMR. Both 1H- and

13C-NMR have been used, the latter also in the solid

state. Obvious 2D techniques are HETCOR, HMQC or HSQC [11-14]. The result is a correlation between

1H and

13C chemical shifts. A plot of these parameters shows that

the 1H and

13C chemical shifts are largely proportional as has

been found in general [15], but in some cases such spectra have proved useful, e.g. in identifying substructures of the aromatic parts of HA and FA [11]. HMBC spectra have also been included. The extra information of such spectra is to some extent counteracted by the many extra resonances and increased overlap (see e.g. Ref. [16]).

It is essential to distinguish between liquid and solid state spectra as different rules about intensities are found. In the present paper we concentrate on 1D

13C liquid state spectra,

which have been used extensively to identify the presence of functional groups such as ketones, carboxylic acids, amides, oxygen substituted aromatic carbons, methoxy groups and

Evaluation of Structures of Humic Substances The Open Magnetic Resonance Journal, 2010, Volume 3 97

aliphatic side chains as well as small molecular fragments such as carbohydrates and amino acids. Since

13C-NMR is a

chemical shift based method, even the carbon skeleton itself will give rise to widely different chemical shifts (Fig. 1). In a broad sense the number of terminal carbons in aliphatic chains, versus the central carbons and finally more branched situations can be estimated from the shape of the aliphatic signal. Other elements such as oxygen, nitrogen and sulphur are detected by means of their substituent effects. Except for HAs derived from lignite, very little sulphur is present so this is really not an issue. For single oxygens the substituent effects are so that both aliphatic and aromatic substituents can easily be detected, with e.g. mono-phenolic structural elements typically appearing around 160 ppm. For di- and tri-phenolic elements, the picture is less clear, and some or all of the O-substituted C-atoms will give rise to signals at much lower frequency (Fig. 1e).

A structural model of HS may be constructed based on identified structural elements, carboxylic acids, ketones aro-matic carbons etc. A more holistic approach is to identify structural elements, gather a possible model and reconstruct the entire NMR spectrum of the proposed model [10, 13]. This has been demonstrated based on data using chemical software like e.g. ChemDraw [13] or based on suggested structures and data from data bases [10]. 2D NMR data bases have also been constructed to identify structural elements and with the intent to reconstruct spectra of complex mix-tures like humic substances [16]. The entire 2D spectrum is, however, not yet possible to match, and the easy access to 1D NMR data bases and the ability to find suitable data for complex structures [17] have prompted this study of a series of suggested models of HA and FA.


Liquid State 13

C-NMR

HA and FA were purified from clayey agricultural soil (Orthic Luvisol, sampled from the A-horizon), sandy agricultural soil (Humic Podzol, sampled from the Bh-horizon), clayey grassland soil (Luvisol, sampled from the A-horizon) and sandy coniferous forest soil (aeolian sand in the beginning of a podzolization, sampled from organic H-horizon

and Bh-horizon), largely according to the IHSS standard procedure with slight modifications [10]. Briefly, the soil was air-dried and then extracted first with 0.1M HCl followed by extraction with 0.1M NaOH under a N2-atmposphere. HA was precipitated at pH=1, re-dissolved in KOH + KCl under a N2-atmosphere, re-precipitated at pH=1 and inorganic impurities were then removed with HCl/HF followed by dialy- sis. Aldrich HA was purchased from Sigma-Aldrich (Stein- heim, Germany) and purified to remove the large content of inorganic impurities and FA as described previously [10].

Liquid state 13

C-NMR of humic fractions was performed as follows: ~80 mg HS was dissolved in 680 μl H2O:D2O (4:1) and pH was adjusted with 10 M NaOH to pH=7 (FA) or pH=11 (HA). Spectra were recorded on a Varian 600 Inova Spectrometer (Varian, Palo Alto, California), working at 150 MHz on

13C, using a 5 mm BB-probe. Spectral width

was set to 40000 Hz, and 50000-100000 transients were re-corded with gated decoupling to suppress Nuclear Overhau-ser Effects with: Delay between experiments = 1000 ms, pulse width = 6 μs (corresponding to a flip angle of 52

o),

acquisition time = 438 ms. Spectra were also recorded with a delay of 2000 ms between experiments but this gave no dif-ferences in intensities. A line broadening (LB) of 50 Hz was applied to all spectra and 3-trimethylsilyl propionate (TSP) was used as an external reference. One spectrum of freshwa-ter FA and marine HA was taken from the literature in order to compare across a wider range of environmental compart-ments.

Simulated Spectra

The spectral data base Modgraph NMRPredict (Mestre- Lab Research) [17] was used to predict chemical shifts. The software predicts chemical shifts with protonated COOH and phenolic OH groups, while spectra of FA fractions were recorded at pH=7 where COOH groups are deprotonated, and HA fractions were recorded at pH=11 where most phenolic OH groups would also be deprotonated. In order to compare recorded and simulated spectra, the predicted COOH chemical shifts were changed by adding 9 ppm to the values obtained for the Aromatic COOH groups and 6 ppm to those calculated for the Alkyl COOH systems. 9 and 6 ppm are

Fig. (1). 13

C chemical shifts of model structures.

a) b)

c) d)

e) f)

98 The Open Magnetic Resonance Journal, 2010, Volume 3 Albers and Hansen

reported to be the average chemical shift changes occurring when deprotonation of carboxylic acids arises following the reactions: Aromatic COOH (Ar-COOH) Aromatic COO- (Ar-COO- ) and Alkyl-COOH Alkyl-COO-, respectively [18]. Chemical shifts of phenols are also affected by the deprotonation that occurs at pH=11, where the HA fractions were dissolved. The number of phenolic structures for which such a change in chemical shift has been reported, is scarce and more important, very different changes in chemical shifts have been observed, depending on numbers and types of substituents on the ring [18-20]. Therefore no attempts were made to correct this. For the predicted spectra of HA, this gives a slight error around 160-170 ppm, while for the FA fractions (analyzed at pH=7) this was not of concern. Spectra were simulated on the assumption of Lorentz-formed peaks with a line broadening (LB) of 400 Hz at half peak- height.


Models of HS

A large number of model structures have been suggested over the years. For the present study, we have chosen 11

structures, which present a wide range of the most recent models within different humic fractions and deriving from several environmental compartments (Figs. 2 & 3). The models were chosen on the basis that they have been cited often and/or are the newest representatives for their class of HS. It is obvious that no single structure as such can be con-structed for neither HS in general nor even for a specific fraction of HS from a specific environmental compartment. Most if not all of the authors who have proposed the struc-tural models in Figs. (2 & 3), have emphasized that their model is not the final solution to humic structures, but that the structures should rather be considered as assemblies of structural elements. Despite these reservations, the authors have presented such very different structures, that they obvi-ously cannot all be representative of some “average struc-ture” of HAs or FAs.

13C-NMR Spectra

Numerous 13

C-NMR spectra have been published during the years. Spectra have been recorded with the HS either in aqueous solutions or in the solid state. The intensities are clearly different in the liquid and solid state spectra as pointed out by Conte et al. [28] and as seen from Fig. (4).

Fig. (2). Structural models for humic acids (HAs), taken from the literature. a) Not further defined HA (Schulten & Schnitzer, 1993) [21]. b)

Not further defined HA (Stevenson, 1994) [1]. c) Agricultural soil HA (Albers et al., 2008) [10]. d) Agricultural soil HA. One of five sub-

structures (Diallo et al., 2003) [13] e) Terrestrial HA from commercial Acros Organics Humic Substance (Sein et al., 1999) [22]. f) Marine

HA (Harvey et al., 1983) [23].

30

35

40

CH3

N

OH

CH3

CH3 OH

O

O

OH

OH

OH O

O

CH3

OHOH

O

OH

O

O

OHO

CH3

OH

O

O

OHO

OH

OH OCH3

O

OH

OH O

N

CH3

O

OHO

O

OHN

CH3

O

CH3

OOH

O

OH

O

OH

CH3 OH

OH

OH

O

OH

OOH

OH

O

O

CH3

O

CH3

O

OH OH

OH

O

OH O

CH3

OH O

OH

O

O

OHOH O

OH

O

OH

OCH3

NH

CH3

OH

O

OH

OOHO

OH

NH

CH3

OOH

OH

CH3

OH

OH

O

OH

O

OH O

O

CH3

CH3

a)

O

O

O OH

OH

O

N

O

NH CH3

NH O

OH

O

OH

O

O

O

O

CH3

OH

O

OH

O

OH

OOH

O OH

OH

OH

O

O

OH

OH

OH

OH

O

OH

OH

O

OH

O

b)

O

O

O

OH

NH

OH

OH

O

OHOHOH

OH

O

NH

O

O OHO OH

O

OH

OOH

NH

CH3

O

OH

NH

OOH

CH3

CH3

CH3

OH

OH

O

NH

O

OH

CH3

CH3c)

CH3 OH

OOH

O

CH3

OOH

OH O OH

O

OH

OOOH

OHO

O

CH3

O

f)

OHO

O

NH2

O

O

OH OH

O

OH O OH

OH

O

OH

NH2

OH

OH e)

d)

OH

OO

CH3

O

OOH

OOH

OOH

O

OH

O

OH

OH

OH

O

S

CH3

OH

O

OH

OO

OH


Fig. (3). Structural models for fulvic acids (FAs) and whole soil HS from literature. a) Soil FA (Shin & Moon, 1996) [24]. b) Four structural

units representing Suwannee River FA (Leenheer & Rostad, 2004) [25]. c) Terrestrial FA from commercial Acros Organics Humic substance

(Alvarez-Puebla et al., 2006) [26]. d) Agricultural soil FA (Albers et al., 2008) [10]. e) Assembly of the partial structures included in a con-

ceptual model for whole soil SOM (Kleber et al., 2007) [27].

This could be a serious problem in an attempt to reconstruct spectra. Recent experiments have shown that a lignite humic acid could be separated into fractions with varying sizes and aromatic and aliphatic contents [29]. This is explained by the recently growing view that humic substances are su-pramolecular assemblies of compounds having relatively low molecular weights [30, 31]. This again could lead to loss of signal in the aliphatic region and therefore underestimation of the aliphatic region (0-45 ppm) if parts of the aliphatic chains are in a hydrophobic core of very large micelles or similar macro ensembles. However, as seen from our previ-ous experiments combining NMR, elemental analysis and titration studies a good balance was actually found between the different types of data [10]. This can be elaborated as follows for a typical agricultural field humic acid (SlA HA). From elemental analysis the formula was determined as

C65H65O32N5 [10]. This leads to 36 degrees of unsaturation (DoU). Eight carboxylic acid groups could be identified ei-ther from integration of liquid state

13C-NMR or a little less

(5-6) from titration studies and in addition six carbohydrate carbons and four amino acid C carbons could be deter-mined from the NMR experiments [10]. Subtracting the av-erage of the two estimates of carboxylic acids as well as car-bohydrate and amino acid carbons, the identified units we end up with ~48 carbons and ~28 DoU left for aromatic and aliphatic (0-45 ppm) structures. If we assume that all rings are benzene types this requires

2/3 DoU pr. carbon. However,

if also heterocyclic aromatic rings are present like furans, benzofurans etc. it is closer to DoU pr. carbon. 28 DoU corresponds in the case of

2/3 DoU pr. carbon to 42 carbons

and in the case of DoU pr. carbon to 37 carbons of aro-matic type leaving only 6 respectively 11 carbons for the

O

OH

O

OH

OO

OHO

O

OCH3

CH3

OH

O

OH

O ONH

O

CH3O

OH

a)

OH

O

O

CH3

OH

O

O

OH

O

O

OH

O

CH3

OHO

O

OH

CH3

O

OHOH

O

OH

CH3O

CH3OH

O

O

O

O

OH

O

OH

OHO

OCH3

O

OH

OH

O

O

O

OH

O

CH3

O

O

OH

O

OH O

OOH

O

OHO

b)

OH

SH

OH

OHOH

O

OH

OHO

OO

OOH

O

OH

O

O

NH2

OH

OH

O

OH

O

OH

OH

OH

OH

O

OH

OH

O

O

OH

OH O c)

O

O O

OH

OH O

OH

OH

OH OHO

O

OH

OH

O

O

OH

OH

O

O

OH

OHO

OOH

OOH

CH3

NHOH

OHO

OHO

OHOH

OH

OH

O

OOH

OH

O OH

O

OH

OOH O CH3

O

OH

O

d)

OH

O

CH3

CH3

OH

O

O OH

OH

O

OH

O

OH

CH3

NH2

O NH

NH

OSH

O NH

CH3

CH3

CH3

OH

OH

O

OH

O

OH

CH3

O

OH

CH3

OH

O

OH

O

OH

CH3

OH

POOH

O

OH

O

OHO

OH

OH

O

OH

O

NH

OH

OOH

OH

O

OHO

O

OH

OH

CH3

CH3

OHO

OHO

O

OH

OH

CH3

CH3

O

OH

OH

OH

OH

OH

e)


aliphatic region. Integration of liquid state 13

C-NMR gave 18% aliphatic (0-45 ppm) carbon [10] corresponding to 11.7 aliphatic carbons in the proposed structure with totally 65 carbon atoms. Since only 6-11 carbons were left to aliphatic structures in the above calculation, this even leaves room for a number of non-aromatic ring structures. Furthermore it shows that the aliphatic part in humic acids of field type is most likely not underestimated using liquid state

13C-NMR.

On that basis we have decided to use this NMR method for our reconstructions or in other words as our tool to evaluate the various models. Another good reason for this is the nar-rower line widths of liquid state spectra, and hence more insight into the presence of various structural elements (Fig. 4), as well as the possibility to distinguish amides/esters from carboxylic acids [10].

Spectra of various HAs and FAs are presented in Figs. (5 and 6). The presented spectra are typical for a range of HAs and FAs, and should largely cover those fractions and envi-ronmental compartments that the HS-models in Figs. (2 and 3) are suggested to represent. Spectra of HA from widely different soils are somewhat similar, reflecting that despite the differences in parent material, HS-fractions from differ-ent geographical and environmental compartments are most often surprisingly similar. This is the case regarding struc-tural elements as well as various physical behaviors. Also between soil horizons, the differences can be quite minimal, as seen for forest HA extracted from the organic horizon (Fig. 5d) and from a subhorizon (Fig. 5e). PCA analyses have previously demonstrated HAs extracted from lignite, like Aldrich HA, to be in a class of their own when com-pared to soil HA [9, 10]. This is due mainly to the lack of some specific groups like amino acids and carbohydrates (Fig. 5a), but despite this, Aldrich HA does share some simi-larities with the field and forest HAs. The marine HA (Fig. 5f), although less aromatic than the other HAs, also shares most of the structural elements of the soil HAs.

The FA-fractions (Fig. 6) are in general less aromatic than the HA-fractions from the same environments. Beside this, they do however carry most of the same structural ele-ments and show somewhat similar shapes of the

13C-NMR

spectra, also when comparing the terrestrial FAs (Fig. 6a & b) with the aquatic FA (Fig. 6c). Significant differences are lower amounts of amino acids in the FA-fraction (seen as a less pronounced peak ~55 ppm) and a shift away from long chain unbranched aliphatics (seen as a peak ~30-35 ppm)

towards peaks in the aliphatic area ~40 and/or 20-30 ppm. This does not really seem to be the case for the Suwannee River FA, though (Fig. 6c).

Simulated 13

C-NMR Spectra of HS-Models

Predicted 13

C-NMR spectra of all 11 HS-models in Figs. (2 and 3) are presented in Figs. (7 and 8).

Humic Acids

The authors who suggested the molecularly large soil HA polymer model in Fig. (2a) did not ascribe their model to HA from any specific environmental compartment [21]. Since the model lacks amino acids and carbohydrates, it is not im-mediately a very good model for soil HAs. This is clearly reflected in the simulated spectrum (Fig. 7a) and further-more, there seem to be too many aliphatic subunits and too few aromatic and carboxylic subunits. Also, the aromatic area is too narrow around the typical chemical shifts of un-substituted aromatic carbons, which indicates too few O-substituted aromatics compared to typical HAs. Neverthe-less, compared to any of the other model simulations, this model does show the best fit in the unsubstituted aromatic region (115-145 ppm), with a clear peak around 128 ppm, which is typically observed in

13C-NMR spectra of both HAs

and FAs. HAs from lignite do not contain significant amounts of amino acids and carbohydrates, and in compari-son with Aldrich HA (Fig. 5a), the simulated

13C-NMR spec-

trum of the model proposed by Schulten & Schnitzer [21] is rather good. Looking at other properties of the model, the fit of this HA is less perfect, with regards to e.g. the CHNO content, which is far from lignite HAs like Aldrich HA (Ta-ble 1), and also different from what is typically observed for soil HAs [1, 34], Table 1.

The model proposed by Stevenson [1], contains amino acid and carbohydrate structural elements and therefore could be a likely candidate for soil HA, but as quickly rec-ognized from the simulated spectra there are some deficien-cies in the aliphatic part of the spectrum. The rather recent models proposed by Diallo et al. [13] and Albers et al. [10] are both constructed to be models of soil HA. While the models are clearly better representatives of soil HA than many other published models, they also have some deficien-cies. In the Albers et al. HA, despite having the right propor-tions with regards to various structural elements and also the elemental content (Table 1) [10], some aliphatic signals in the region 30-35 ppm are missing. Furthermore, the aromatic

Fig. (4). Comparison of a) liquid and b) solid state 13

C-NMR spectra of a forest HA. Peaks “assigned” according to Albers et al., 2008 [10].

a)

Aliphatic

Amino acidαC

C1 C2-5 C6

ArCH

ArCO

Amide/ester

C=O

COO-Carbohydr.

200 180 160 140 120 100 80 60 40 20 0 ppm

Aliphatic

Amino acidαC

C1 C2-5 C6

ArCH

ArCO

Amide/ester

C=O

COO-Carbohydr.

200 180 160 140 120 100 80 60 40 20 0 ppm200 180 160 140 120 100 80 60 40 20 0 ppm b) 200 180 160 140 120 100 80 60 40 20 0 ppm200 180 160 140 120 100 80 60 40 20 0 ppm


Fig. (5). Liquid state 13

C-NMR spectra of humic acids (HAs). a) Aldrich HA, commercial HA extracted from lignite. b) Field HA, extracted

from clayey soil. c) Field HA, extracted from the Bh-horizon of an acidic sandy podzol. d) Forest HA, extracted from the organic H-horizon.

e) Forest HA, extracted from the Bh horizon of a forest podzol. f) Marine HA [32].

Fig. (6). Liquid state 13

C-NMR spectra of fulvic acids (FAs). a)

Field FA, extracted with acid from the Bh-horizon of an acidic

sandy podzol. b) Grassland FA, extracted from a clayey soil. c)

Aquatic FA, the IHSS standard FA from Suwannee River [33].

region is not shaped, with a clear peak around 128 ppm, as seen for recorded spectrum of soil HA. The same is the case for the model proposed by Diallo et al. [13] and furthermore this model clearly lacks aliphatic structures which will give rise to chemical shifts around 30 ppm. Likely candidates for this would be simple unsubstituted aliphatic chains, which might be incorporated in the proposed model. Furthermore, as previously pointed out [10], the N- and S-content should be changed in order to make this model in better agreement with typical soil HAs.

Sein et al. [22] proposed a model of terrestrial HA (Fig. 2e), but it was not clear if the HA used as a model HA, was derived from soil or lignite. The proposed model could pos-sibly be one structural element in HA, since, apart from a too intense carbonyl-signal, its simulated spectrum contains no wrong signals (Fig. 7e). It does however miss amino acids, which is reflected in the lack of signal ~55 ppm and further-more it contains no unsubstituted aliphatics, which is also reflected in the simulated spectrum as the lack of signal be-low ~32 ppm. It can therefore not stand alone as a represen-tative structure of HA.

We have included a model of a marine HA, since, com-pared to soil HAs, these HAs have very different parent ma-terial, which would also be expected to be reflected in their final structure and hence

13C-NMR spectra (Fig. 5f). One of

the characteristics of marine HAs is their low aromatic con-tents, and this feature is well captured by the model of Har-vey et al. [23]. Furthermore, the model contains rather long aliphatic chains, which give rise to the intense signal at ~30 ppm, which is also seen in the recorded

13C-NMR spectrum.

a)

Aldrich HA

220 200 180 160 140 120 100 80 60 40 20 ppm

Aldrich HA

220 200 180 160 140 120 100 80 60 40 20 ppm b)

Clay field HA

220 200 180 160 140 120 100 80 60 40 20 ppm

Clay field HA

220 200 180 160 140 120 100 80 60 40 20 ppm

c)

Bh field HA

220 200 180 160 140 120 100 80 60 40 20 ppm

Bh field HA

220 200 180 160 140 120 100 80 60 40 20 ppm d) 220 200 180 160 140 120 100 80 60 40 20 ppm

H-hor. forest HA

220 200 180 160 140 120 100 80 60 40 20 ppm

H-hor. forest HA

105

e) 220 200 180 160 140 120 100 80 60 40 20 ppm

Bh-hor. forest HA

220 200 180 160 140 120 100 80 60 40 20 ppm220 200 180 160 140 120 100 80 60 40 20 ppm

Bh-hor. forest HA

f)

Marine HA

220 200 180 160 140 120 100 80 60 40 20 ppm

Marine HA

220 200 180 160 140 120 100 80 60 40 20 ppm

a)

Bh field FA

220 200 180 160 140 120 100 80 60 40 20 ppm

Bh field FA

220 200 180 160 140 120 100 80 60 40 20 ppm

b)

Grassland FA

220 200 180 160 140 120 100 80 60 40 20 ppm

Grassland FA

220 200 180 160 140 120 100 80 60 40 20 ppm

c) 220 200 180 160 140 120 100 80 60 40 20 ppm

Suwanee River FA

220 200 180 160 140 120 100 80 60 40 20 ppm

Suwanee River FA


The two distinct peaks at ~55 and 63 ppm, attributed to amino acid C and C6 of carbohydrates, respectively, is

however not seen in the simulated spectra, and looking at the model structure (Fig. 2f), no carbohydrate or amino acid structural elements are seen. Since also a peak at ~105 ppm (anomeric carbohydrate signal) is seen in the recorded spec-trum of a marine HA, most of the signal around 75-80 ppm most likely derives from carbohydrate, and the model should probably contain less aliphatic alcohols and instead include carbohydrate structures. Furthermore, there are four ketonic carbons in the model, which gives rise to a rather intense signal around 205-210 ppm, while the ketonic signal in the recorded spectrum is low.

Fulvic Acids and Whole Soil HS

The simulated spectra of the four FA-models and the sin-gle HS-model are shown in Fig. (8). The soil FA model pro-posed by Shin & Moon [24], is a further development of models previously proposed for Suwanee River FA and the simulated NMR-spectra (Fig. 8a) is not too different from the Suwanee River FA model, proposed by Leenheer & Rostad [25] (Fig. 8b). No amino acids are included in neither of these models, which is immediately seen as the lack of signal ~55 ppm where you would usually see a signal from amino acid Cs. Aquatic FAs often do not contain signifi-cant amounts of amino acids but, although in lower amounts than the corresponding HA fractions, terrestrial FAs most often do and amino acids should probably be included in this model. The lower amounts of long-chain aliphatics is re-flected in a less pronounced peak ~30-35 ppm, in Fig. (8a & b). The shape of the aliphatic area in Fig. (8b) fits well with the recorded spectrum of some soil FA-fractions (Fig. 6b), but it fits the aquatic FA (Fig. 6c), which it was constructed for, less well.

Furthermore the intense peak at ~170 ppm (derived from esters in the model) is not really seen in any of the recorded spectra for FAs. The Suwanee River FA seems to be an illus-trative example, where an integration of the areas assigned to certain overall structural elements (e.g. aromatics or aliphat-ics) would lead to the conclusion that the model fits very good, while looking at the simulated spectra, it is obvious that the specific nature of the structural elements, cannot be more than partly correct. This is also the case for the model of terrestrial FA proposed by Alvarez-Puebla et al. [26]), in which the aliphatic fraction of C-atoms is so branched and substituted with O-atoms, that, besides being too small over-all, it shows most of the aliphatic signals >60 ppm and no aliphatic signals <40 ppm, which is obviously a deficiency. This shape in the aliphatic area is more pronounced but still somewhat similar to the HA proposed by Sein et al. [22] (Fig. 2e & 7e), which was actually the basis for developing this model of FA [26].

The FA-model proposed by Albers et al. [10] (Fig. 8d) reveals the best shape compared to the recorded spectra, es-pecially to the grassland FA in Fig. (6b). The shape of the aromatic peak is not perfect though, with signals missing especially ~130 ppm. This is probably caused by the fact that no unsubstituted aromatic C-atoms exist in the model (Fig. 3d). Furthermore, some fill-in between aromatic peaks is missing in general, which supports the concept that a puri-fied HS-fraction is a mixture of several more or less similar molecules.

Fig. (7). Simulated 13

C-NMR spectra of models of humic acids

(HAs). The units on the x-axes are all in ppm. a) Not further de-

fined HA (Schulten & Schnitzer, 1993) [21]. b) Not further defined

HA (Stevenson, 1994) [1]. c) Agricultural soil HA (Albers et al.,

2008) [10]. d) Agricultural soil HA taken from Diallo et al. (2003)

[13]. The spectrum was based on five proposed substructures of

which one is shown in Fig. (2d). e) Terrestrial HA from commer-

cial Acros Organics Humic Substance (Sein et al., 1999) [22]. f)

Marine HA (Harvey et al., 1983) [23].

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

a)

b)

c)

d)

e)

f)


Fig. (8). Simulated 13

C-NMR spectra of fulvic acids (FAs) and

whole soil HS. The units on the x-axes are all in ppm. a) Soil FA

(Shin & Moon, 1996) [24]. b) Four structural units representing

Suwannee River FA (Leenheer & Rostad, 2004) [25]. c) Terrestrial

FA from commercial Acros Organics Humic substance (Alvarez-

Puebla et al., 2006) [26]. d) Agricultural soil FA (Albers et al.,

2008) [10]. e) Assembly of the partial structures included in a con-

ceptual model for whole soil SOM (Kleber et al., 2007) [27].

The model proposed by Kleber et al. [27] is a model of SOM in general, that is both FA, HA, humin as well as non-

HS SOM, the latter typically making up 25-60% of SOM [1, 2]. The simulated spectrum of this model can therefore not be expected to make up more than ~50% of a mixed spec-trum of HA and FA. We have nevertheless included this model, since it is the first model constructed according to the supramolecular structural view of HS/SOM. Furthermore, since it is a model of SOM, which includes HA and FA, it should be possible to choose parts of the model, which would then fit the recorded spectra of HA and/or FA. Since the model contains no O-substituted aromatic C-atoms, this is however not possible, and the simulated spectrum is quickly recognized as very different from the recorded spec-tra of HA and FA. Since we use liquid state

13C-NMR as our

tool to evaluate models of HA and FA, it is not wise to make too many conclusions of this SOM-model. We can neverthe-less say, that it will be extremely difficult to choose struc-tural elements of the SOM-model that will fit either HA- or FA-fractions, which are, after all, important fractions of SOM, quantitatively as well as regarding several properties of SOM. This conclusion would be supported by an elemen-tal analysis, which for this model would give 64% C, 8% H, 1.7% N and 24% O (Table 1) and a titration of acid groups, which would also reveal the lack of phenolic structures, which are always part of the HA- and FA-fraction of SOM. Kleber et al. [27] have some very interesting discussions on organo-mineral interactions on soil, but the model, which they propose, seems to contain very little of the compounds, which would normally be extracted into the HA- and FA-fractions.

GENERAL REMARKS

The advantage of using a 13

C-NMR simulation is that all carbons within a structure are taken into account, and can be immediately compared to recorded NMR-spectra. Although several overlaps will occur in a 1D-NMR spectrum of humic substances and it therefore not always is possible to be con-clusive on the exact structural elements, the inclusion of all structural elements makes possible a better balance between the various parts of the spectrum. This balance should of course be checked in relation to an elemental analysis. A titration of acid groups to divide these into carboxylic acids and phenols likewise makes a good check when compared to the NMR results. Finally, both nitrogen and in the case of lignite HS also sulphur content play a role. The former can in some cases, when being part of amino acids, be correlated to the NMR findings [10].

The aromatic region in general plays an important role as many humic substances contain large proportions of aro-matic carbons. The very broad region assigned to aromatic structures, clearly shows that a vast combination of struc-tures are required not the least to explain the low frequency wing (100-115 ppm) generally found in

13C-NMRspectra of

HS-fractions. In HS-fractions containing significant amounts of carbohydrates, this region can to some extent, but not fully, be ascribed to anomeric carbons of carbohydrates (~105 ppm). The oxygen substituted aromatic carbons can to a good degree be determined as some of these signals will appear around 160 ppm. To help understand, which aromatic structures are present in a given HS-fraction, a check be-tween oxygen content, number of titrated phenolic and car-boxylic acid groups and NMR parameters, is necessary since

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

020406080100120140160180200220

a)

b)

c)

d)

e)


part of the oxygen substituted aromatic carbons are buried in the main signal due to structural elements as exemplified in Fig. (1f). Parameters such as bulk densities and solubility could also be taken into account [13]. Mass spectroscopic studies are clearly very useful. Diallo et al. [13] showed that most of the fragments of soil HA do not exceed 1200 Dalton. This fits well with most of the proposed structures evaluated in this paper, which are all below 1650 Dalton except the HA proposed by Schulten & Schnitzer (1993), which is signifi-cantly larger (~5800 Dalton). The simulated spectra have shown that even with very broad line widths, the resulting spectra of such small molecules do always miss some fill-in, especially in the aromatic region. This supports the idea that even a purified HS fraction is a mixture of several molecules and that one final structure of a HA or FA cannot be identi-fied.

As indicated here, there are numerous considerations, which have to be taken into account before it is possible to construct a reasonable model structure, which makes a good representation of either whole soil SOM or of selected frac-tions like HA and FA. In the present paper, apart from hav-ing evaluated some existing proposed models, we have shown that using simulated

13C-NMR spectra, in the process

of constructing such model structures, will help avoiding wrong structural elements and hopefully likewise be a help in choosing the right ones.

REFERENCES

[1] Stevenson FJ. Humus chemistry: genesis, composition, reactions. New York: John Wiley & Sons 1994.

[2] Swift RS. Organic matter characterization. In: Sparks DL, Ed.

Madison: Soil Science Society of America Inc. 1996; Vol. 3: pp. 1011-71.

[3] Reckhow DA, Makdissy G, Rees PS. Disinfection by-product precursor content of natural organic matter extracts. In: Karanfil, T.

Krasner SW, Westerhoff P, Xie Y. Eds. Washington: American Chemical Society 2008; Vol. 995: pp. 80-94

[4] Piccolo A, Celano G, Conte P. Adsorption of Glyphosate by Humic Substances. J Agric Food Chem 1996; 44: 2442-46.

[5] Albers CN, Banta GT, Hansen PE, Jacobsen OS. Effect of different humic substances on the fate of Diuron and its main metabolite 3,4-

dichloroaniline in soil. Environ Sci Technol 2008; 42: 8687-91. [6] Perminova IV, Grechishcheva NY, Petrosyan VS. Relationships

between structure and binding affinity of humic substances for polycyclic aromatic hydrocarbons: relevance of molecular descrip-

tors. Environ Sci Technol 1999; 33: 3781-7. [7] Senesi N. Binding mechanisms of pesticides to soil humic

substances. Sci Total Environ 1992; 123/124: 63-76. [8] Tipping E. Cation binding by humic substances. Cambridge:

Cambridge University Press 2002. [9] Thomsen M, Lassen P, Dobel S, Hansen PE, Carlsen L, Mogensen

BB. Characterisation of humic material of different origin: a multi-variate approach for quantifying the latent properties of dissolved

organic matter. Chemosphere 2002; 49: 1327-37. [10] Albers CN, Banta GT, Jacobsen OS, Hansen PE. Characterization

and structural modelling of humic substances in field soil display-ing significant differences from previously proposed structures. Eur

J Soil Sci 2008; 59: 693-705. [11] Haiber S, Herzog H, Buddrus J, Burba P, Lambert J. Quantification

of substructures of refractory organic substances by means of nu-clear magnetic resonance. In: Frimmel FH, Abbt-Braun G,

Heumann KG, Hock B, Lüdemann H-D, Spiteller M, Eds. Wein-heim: Wiley-VCH, 2002; pp. 115-28.

[12] Simpson AJ, Burdon J, Graham CL, Hayes MHB, Spencer N, Kingery WL. Interpretation of heteronuclear and multidimensional

NMR spectroscopy of humic substances, Eur J Soil Sci 2001; 52: 495-509.

Table 1. Elemental Analysis Data of HA and FA (wt%). The Top Three Lines are General Data on HA and FA. Then the Calcu-

lated CHNOS-Content for the 11 Structural Models of HS are Shown, and Just Below each Model, Elemental Analysis

Data are Included for the HA or FA for which the Model was Proposed, in the Cases where Such Data were Reported.

Type of HS C H N O S References

HA in general 54-59 3.2-6.2 0.8-4.3 33-40 <1 [1, 34]

FA in general 41-51 3.8-7.0 0.9-3.3 40-50 <1 [1, 34]

Lignite HA (Aldrich) 55 4.3 0.7 37 3.5 [10]

Schulten & Schnitzer HA model 67 6.0 1.5 26 0 [21]

Stevenson HA model 57 3.4 2.6 37 0 [1]

Albers et al. soil HA model 56 4.9 5.0 34 0 [10]

Albers et al. soil HA 55 4.6 4.9 36 <0.5 [10]

Diallo et al. Soil HA model 53 4.3 1.4 38 3.2 [13]

Diallo et al. Soil HA 51 4.0 4.0 40 0.9 [13]

Sein et al. HA model 57 5.2 3.8 34 0 [22]

Harvey et al. Marine HA model 63 8.1 0 29 0 [23]

Shin & Moon soil FA model 56 5.5 1.8 36 0 [24]

Alvarez-Puebla FA model 42 3.2 1.3 50 3 [26]

Albers et al. soil FA model 47 3.7 0.8 48 0 [10]

Albers et al. soil FA 47 3.8 1.0 48 <0.5 [10]

Kleber et al. SOM model 64 8 1.7 24 0 [27]


[13] Diallo MS, Simpson A, Gassman P, Johnson JH, Goddard WA,

Hatcher PG. 3-D structural modelling of humic acids through ex-perimental characterization, computer assisted structure elucidation

and atomistic simulations. 1. Chelsea soil humic acid. Environ Sci Technol 2003; 37: 1783-93.

[14] Mao J-D, Schmidt-Rohr K. Absence of mobile carbohydrate do-mains in dry humic substances proven by NMR, and implications

for organic-contaminant sorption models. Environ Sci Technol 2006; 40: 1751-56.

[15] Macomber RS. Proton-carbon chemical shift correlations. J Chem Educ 1991; 68: 284-5.

[16] Simpson AJ, Lefebre B, Moser A, et al. Identifying residues in natural organic matter through spectral prediction and pattern

matching of 2D NMR datasets. Magn Reson Chem 2004; 42: 14-22.

[17] NMR Predict, Mestrelab research. Available at: http://mestrelab. com

[18] Kalinowski H-O, Berger S, Braun S. Carbon-13 NMR spectros-copy. Chichester: John Wiley & Sons 1987.

[19] Hansen PE, Thiessen H, Brodersen R. Bilirubin acidity. Titrimetric and 13C NMR studies. Acta Chem Scand B 1979; 33: 281-93.

[20] Kato-Toma Y, Iwashita T, Masuda K, Oyama Y, Ishiguro M. pKa measurements from nuclear magnetic resonance of tyrosine-150 in

class C b-lactamase. Biochem J 2003; 375: 175-81. [21] Schulten H-R, Schnitzer M. A state of the art structural concept for

humic substances. Naturwissenschaften 1993; 80: 29-30. [22] Sein LT, Varnum JM Jansen SA. Conformational modeling of a

new building block of humic acid: approaches to the lowest energy conformer. Environ Sci Technol 1999; 33: 546-52.

[23] Harvey GR, Boran DA, Chesal LA, Tokar JM. The structure of marine fulvic and humic acids. Mar Chem 1983; 12: 119-32.

[24] Shin HS, Moon H. An “average” structure proposed for soil fulvic acid aided by depth/quat 13C NMR pulse techniques. Soil Sci 1996;

161: 250-6.

[25] Leenheer JA, Rostad C. Tannins and Terpenoids as Major Precur-

sors of Suwannee River Fulvic Acid. U.S. Geological Scientific In-vestigations Report 2004-5276, U.S. Geological Survey, Reston,

Virginia 2004. [26] Alvarez-Puebla RA, Valenzuela-Calahorro C, Garrido JJ. Theoreti-

cal study on fulvic acid structure, conformation and aggregation A molecular approach. Sci Total Environ 2006; 358: 243-54.

[27] Kleber M, Sollins P, Sutton R. A conceptual model of organo-mineral interactions in soils: self-assembly of organic molecular

fragments into zonal structures on mineral surfaces. Biogeochemis-try 2007; 85: 9-24.

[28] Conte P, Piccolo A, van Lagen B, Buurman P, de Jager PA. Quan-titative difference in evaluating soil humic substances by liquid-

and solid-state 13C-NMR spectroscopy. Geoderma 1997; 80: 339-52.

[29] Conte P, Spaccini R, Smejkalová D, Nebbioso A, Piccolo A. Spec-troscopic and conformational properties of size-fractions separated

from a lignite humic acid. Chemosphere 2007; 69: 1032-9. [30] Piccolo A. The supramolecular structure of humic substances: a

novel understanding of humus chemistry and implications in soil science. Adv Agron 2002; 75: 57-134.

[31] Sutton R, Sposito G. Molecular structure in soil humic substances: the new view. Environ Sci Technol 2005; 39: 9009-15.

[32] Saito Y, Haiano S. Characterization of humic and fulvic acids isolated from marine sediments of Sagami and Suruga Bays with

C-13 and proton nuclear magnetic resonance J Oceanogr Soc 1981; 36: 286-92.

[33] Leenheer JA, Brown GK, MAccarthy P, Cabaniss SE. Models of metal binding structures in fulvic acid from the Suwannee River,

Georgia. Environ Sci Technol 1998; 32: 2410-16. [34] Tan KH. Humic Matter in soil and the environment – principles

and controversies. New York: Marcel Dekker 2003.

Received: January 15, 2010 Revised: February 02, 2010 Accepted: February 02, 2010

© Albers and Hansen et al.; Licensee Bentham Open.



Hot Topic Tomrj Contents

Documents

Transcript of Hot Topic Tomrj Contents