analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic...

31
Journal of Analytical and Applied Pyrolysis 49 (1999) 385–415 Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter H.-R. Schulten * Institute of Soil Science, Uni6ersity of Rostock, Justus -6on -Liebig -Weg 6, 18059 Rostock, Germany Received 10 July 1998; accepted 12 December 1998 Abstract Humic acids (HA), fulvic acids (FA), non-humic substances (NHS) and dissolved organic matter (DOM) in a bog lake water are investigated by analytical pyrolysis. The applied thermal methods are direct, in-source pyrolysis-field ionization mass spectrometry in the high electric field (Py-FIMS), and Curie-point pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS) in combination with library searches. Based on the identified building blocks and together with complimentary analytical data, proposals for a general concept of the basic molecular structures of humic macromolecules in water are put forward. Computa- tional chemistry is utilized for structural modeling and geometry optimization of DOM. Molecular mechanics calculations are performed to evaluate the conformation of structural, three-dimensional models and to determine the total energy and the partial contributions from bond-, angle-, dihedral-, van der Waals-, stretch-bend-, and electrostatic energies. Quantitative structure – activity relationship (QSAR) properties are calculated and allow the correlation of molecular structures with properties such as mass, surface area, volume, partial charges (electronegativity), polarizability, refractivity, hydrophobicity, and hydration energy. The principal aim and long-term strategy are to develop step by step improvements of the presented model structures of organic matter in water which explain the molecular composition as well as their ecological meaning, dynamic character, and structure – property relationships in natural and contaminated aquatic and terrestrial systems. In a first integrated approach, the dissociation and association processes of humic substances are simulated at nanochemistry level and are proposed as concepts for future collaboration incorporating results of additional chemical, biological, spectroscopic and microscopic methods. © 1999 Elsevier Science B.V. All rights reserved. * Tel.: +49-381-4982137; fax: +49-381-4982159; e-mail: [email protected]. 0165-2370/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII:S0165-2370(98)00137-5

Upload: h-r-schulten

Post on 02-Jul-2016

217 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

Journal of Analytical and Applied Pyrolysis49 (1999) 385–415

Analytical pyrolysis and computationalchemistry of aquatic humic substances and

dissolved organic matter

H.-R. Schulten *Institute of Soil Science, Uni6ersity of Rostock, Justus-6on-Liebig-Weg 6, 18059 Rostock, Germany

Received 10 July 1998; accepted 12 December 1998

Abstract

Humic acids (HA), fulvic acids (FA), non-humic substances (NHS) and dissolved organicmatter (DOM) in a bog lake water are investigated by analytical pyrolysis. The appliedthermal methods are direct, in-source pyrolysis-field ionization mass spectrometry in the highelectric field (Py-FIMS), and Curie-point pyrolysis-gas chromatography/mass spectrometry(Py-GC/MS) in combination with library searches. Based on the identified building blocksand together with complimentary analytical data, proposals for a general concept of thebasic molecular structures of humic macromolecules in water are put forward. Computa-tional chemistry is utilized for structural modeling and geometry optimization of DOM.Molecular mechanics calculations are performed to evaluate the conformation of structural,three-dimensional models and to determine the total energy and the partial contributionsfrom bond-, angle-, dihedral-, van der Waals-, stretch-bend-, and electrostatic energies.Quantitative structure–activity relationship (QSAR) properties are calculated and allow thecorrelation of molecular structures with properties such as mass, surface area, volume,partial charges (electronegativity), polarizability, refractivity, hydrophobicity, and hydrationenergy. The principal aim and long-term strategy are to develop step by step improvementsof the presented model structures of organic matter in water which explain the molecularcomposition as well as their ecological meaning, dynamic character, and structure–propertyrelationships in natural and contaminated aquatic and terrestrial systems. In a first integratedapproach, the dissociation and association processes of humic substances are simulated atnanochemistry level and are proposed as concepts for future collaboration incorporatingresults of additional chemical, biological, spectroscopic and microscopic methods. © 1999Elsevier Science B.V. All rights reserved.

* Tel.: +49-381-4982137; fax: +49-381-4982159; e-mail: [email protected].

0165-2370/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved.PII: S0165 -2370 (98 )00137 -5

Page 2: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

386 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Keywords: Association/dissociation processes; Dissolved organic matter; Fulvic, humic acids; Modeling;Molecular mechanics; Nanochemistry; Pyrolysis-field ionization mass spectrometry; Pyrolysis-GC/MS;Surface-water; QSAR

1. Introduction

Humic substances and soil organic matter are essential bases for life on earth,and the study of their structures is the aim of a fast growing, interdisciplinary,scientific community. The soil organic matter (SOM) refers to the sum-total of allcarbon-containing substances in soils. It influences plant growth through its effectson the physical, chemical, and biological properties of soils. SOM consists of amixture of plant and animal residues in various stages of decomposition, ofsubstances synthesized microbiologically and/or chemically from breakdown prod-ucts, and of the bodies of live and dead microorganisms, small animals, and theirdecomposition remains. To simplify this chemically very complex and physicallyheterogeneous system, SOM is usually subdivided into non-humic and humicstructures. Non-humic substances include those with still recognizable chemicalcharacteristics (e.g. carbohydrates, proteins, fats, waxes, etc.). The major part ofSOM, however, consists of humic substances. These are amorphous, dark-colored,partly aromatic, polyelectrolyte-like materials which range in molecular weightsfrom a few hundred to several thousand [1,2]. The latter macromolecular structuresbetween 1 nm and 1 mm in size can be regarded as colloids and above 1 mm ashumic particles.

Similar humic substances can be isolated from aquatic systems by reverseosmosis, ultrafiltration and adsorption/desorption from resins. The water-solublepart obtained by filtration B0.45 mm can be determined as dissolved organiccarbon (DOC). Since the applied methods of analytical pyrolysis are well-suited forinvestigations of structural and molecular properties of humic substances, in thefollowing we refer to the sum-total of all carbon-containing substances in filtered,freeze-dried water as dissolved organic matter (DOM). Comprehensive surveys onthe isolation, characterization and properties of DOM and humic substances inwater and SOM in soils have been published and underline the structural complex-ity and environmental relevance of these materials [3–10].

Among the analytical methods of aquatic and terrestrial humic substances thecombination of pyrolysis and field ionization mass spectrometry (Py-FIMS) andCurie-point pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS) wasfound to be a powerful tool to produce structural information about the molecularbuilding blocks of SOM and DOM. The advantages of the methodology aresensitivity, specificity and speed. Moreover, future applications and basic researchin soil science utilizing results of pyrolysis-field ionization mass spectrometrytogether with other modern spectroscopic methods, particularly in an integratedapproach, appear promising [11].

The aim of this work is to clarify whether analytical pyrolysis, molecularmodeling, and computational chemistry can contribute to a better understanding of

Page 3: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

387H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

the structure of DOM and aquatic humic substances. Four crucial questions arise:Firstly, is it possible to identify the produced thermal fragments on the basis ofPy-FIMS data banks of well-described standards of biopolymers, humic substances,waters, and soils? Secondly, how far is support available by other analyticalmethods such as Py-GC/MS, W, FTIR, NMR, etc., for an unambigeous assign-ment of structural building blocks? Thirdly, what is the origin of these subunits inhumic substances and DOM and are there possibilities to understand and predictthe processes of DOM, FA, and HA formation? Finally, can 3D molecularstructural concepts for DOM be developed on the basis of accepted analytical datawhich allow the visualization and simulation of DOM? What are the results ofcomputational chemistry and nanochemistry?

2. Experimental

2.1. Materials

Standard reference samples of the priority research program ‘Refractory OrganicAcids in Water’ of the German Research Association (Deutsche Forschungsgemein-schaft; DFG) were investigated. Refractory organic acids in water were opera-tionally defined as organic acids of different origin prepared according to themethods recommended by the International Humic Substances Society (IHSS). Theisolation and purlfication by the XAD procedures [12,13] and characteristic data ofaquatic and terrestrial organic substances have been reported [14]. The brown waterof the Hohloh Lake (code no. HO10) was sampled in Feb. 1995 and the HA andFA fractions were prepared in the following month.

2.2. Curie-point pyrolysis-gas chromatography/mass spectrometry (Py-GC/MS)

The experimental set-up for Py-GC/MS has been described previously [8,15].Samples of freeze-dried water, humic substances, sediments and soils are pyrolyzedin a type 0316 Curie-point pyrolyzer (Fischer, 53340 Meckenheim, Germany). Thematerials were not pretreated except drying and, if necessary, milling. The finalpyrolysis temperatures employed were 300°C, 500°C and 700°C, respectively.However, the data reported for aquatic materials in this study were all obtained at500°C final temperature. The total heating time was varied between 3 and 9.9 s.Following split injection (split ratio 1:3; flow rate 1 ml 20 s−1) the pyrolysisproducts were separated on a gas chromatograph (Varian 3700, 64289 Darmstadt,Germany), equipped with a 30 m capillary column (DB5), coated with 0.25 mm filmthickness and an inner diameter of 0.32 mm. The starting temperature for the gaschromatographic temperature program was 40°C, and the end temperature 250°C,with a heating rate of 10 K min−1. The gas chromatograph was connected to athermionic nitrogen-specific detector and a double-focusing Finnigan MAT 212mass spectrometer. Conditions for mass spectrometric detection in the electronionization mode are+3 kV accelerating voltage, 70 eV electron energy, 2.2 kV

Page 4: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

388 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

multiplier voltage, 1.1 s (mass decade)−1 scan speed and a recorded mass rangebetween m/z 50 and 500. A detailed description of the principle, potential andlimitations of Py-GC/MS of humic fractions and soils has been presented [15].

2.3. Pyrolysis-field ionization mass spectrometry (Py-FIMS)

For temperature-resolved Py-FIMS, 1–5 mg of freeze-dried water and/or about100 mg of aquatic humic substances such as humic acids (HA), fulvic acids (FA),and non-humic substances (NHS) were thermally degraded in the ion source of aMAT 731 (Finnigan, 28127 Bremen, Germany) modified high performance (AMDIntectra GmbH, 27243 Harpstedt, Germany) mass spectrometer. The instrumentalset-up for the field ionization (FI) ion source has been reported [16].

The samples were weighed before and after Py-FIMS (error90.01 mg) todetermine the pyrolysis residue and the produced volatile matter and to comparethem with the corresponding results from off-line pyrolysis. A heatable/coolabledirect introduction system with electronic temperature-programming, adjusted atthe +8 kV potential of the ion source and a field ionization emitter, were used. Theslotted cathode plate serving as counter electrode was set at −6 kV potential.Thus, at 2 mm distance between the emitter tips and the cathode, a total potentialdifference of 14 kV is applied, resulting in an extremely high electric field strength,essential for soft ionization. All samples were heated in high vacuum (1.3×10−4

Pa) from 100 to 700°C at a heating rate of approximately 10 K min−1. Dependingon the volatility and thermal stability of the sample materials, 40–60 magneticscans of the gaseous, ionized pyrolyzate components were recorded in the massrange 15–1000 mass units.

In general, at least three replicates were run for each sample. The total ionintensities (TII) of the single spectra were normalized to 1 mg sample weight,averaged for replicate runs, and plotted against the pyrolysis temperature, produc-ing Py-FIMS thermograms. For the selection of biomarkers and quantitativeevaluations, in particular freeze-dried water and aquatic humic substances, detaileddescriptions of the method for SOM in soil particle-size fractions and whole soilshave been published [17]. It should be noted that the marker signals have beenselected mainly for humic fractions and topsoils and are valid only for investiga-tions using in-source field ionization. Moreover, careful examination of the data hasto consider: (a) only series of ions for one compounds class; (b) the appropriatetemperature interval of volatilization; (c) the thermograms of biomarker signalsshould display gaussian-shape convolutions; and finally, the crucial point is thecorroboration of the results by independent analytical methods. In our work mainlyPy-GC/MS results in connection with chemical, biological and physical data wereexamined.

2.4. Structural modeling and geometry optimization

Since practically all pyrolysis data of aquatic humic substances and DOM havebeen published as two-dimensional plots, it is of interest to illustrate the potential

Page 5: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

389H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

of available, relatively low cost software and personal computer (PC) equipmentwhich allow threedimensional (3D) displays and computer-aided design for chemi-cal structures and model reactions. Especially the possibilities for molecular model-ing and geometry optimizations of complex macromolecules, which are often thetarget of analytical pyrolysis, virtually open up a new dimension. This is demon-strated below for molecular visualization and simulation of aquatic humic sub-stances and DOM. One of the essential aspects of applying computationalchemistry to geometry optimization, determination of partial energies, associationand dissociation processes, trapping and binding of biological and anthropogenicmolecules, chemical properties, etc., is that the available spatial dimensions coverthe range of atomic-, functional group- and whole colloid-structures at nm level(Nanochemistry).

The previously proposed terrestrial HA [18,19], SOM [20,21] and soil [22,23]model structures were obtained using the HyperChem™ software [24] (release 5.02;Microsoft Windows 95™) for all model construction, chemical interaction studies,molecular mechanics and molecular dynamics calculations. The original Hyper-Chem outputs in Angstrom and kcal have been converted to in nm and kJ,respectively. Quantitative structure–activity relationship (QSAR) properties weredetermined using the ChemPlus™ software (Hypercube Inc., release 2.0). Theemployed IBM-compatible PC consisted of the processor Pentium™ Pro 200/256K,Intel-board Venus ATX P-Pro, in combination with 128 MB random accessmemory, Diamond Stealth Video 25 000 2MB, PCI graphic card, Iiyama 17%% colormonitor, 2.1 GB hard disk, and peripheral hardware.

The molecular mechanics calculations were performed using the HyperChemMM+ force field developed for organic molecules. Their atoms are treated asNewtonian particles interacting through a potential energy function. The potentialenergies depend on bond lengths and angles, torsion angles, and non-bondedinteractions which include van der Waals forces, electrostatic interactions, andhydrogen bonds. In these calculations, the forces on atoms are functions of atomicposition [24].

In this contribution we used molecular mechanics and an all atom force fielddeveloped for organic molecules to compute a potential energy surface. This surfacerepresents the energy of a molecular system with N atoms as a function of the 3NCartesian coordinates. In order to optimize the geometry (=minimize the totalenergy) of a chemical conformation, the number of computing cycles required fora gradient calculation is approximately proportional to the number of atoms N,and the time per cycle is proportional to N2. This gradient is the derivative of theenergy with respect to all Cartesian coordinates. Moreover, the size and quality ofthe initial structure, the computer capacity and the termination conditions of themodeling calculations are crucial. In general, the convergence limit is either thenumber of calculations cycles (the default number is 15 times the number of atoms)or a gradient for the molecular system (default of 0.419 kJ (0.1 nm)−1 mol−1). Forthe humic colloids and DOM a gradient \1 kJ (0.1 nm)−1 mol−1 was chosen andthus within reasonable calculation times (N\1000 g mol−1, B100 h) local minimaon the potential surface were obtained.

Page 6: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

390 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

3. Results and discussion

3.1. Bog lake water (Py-FIMS)

The sampling and characteristic chemical data for the brown water from theHohlohsee, Black Forest, Germany, have been described [14]. The summed andaveraged Py-FI mass spectrum and thermogram (upper right) of the freeze-dried

Fig. 1. Pyrolysis-field ionization mass spectra and thermograms of bog lake water (Hohloh Lake, BlackForest, Germany, sampling time April 1996): (a) freeze-dried original water (HO10 original); (b) fulvicacid (HO10 FA); and (c) humic acids (HO10 HA).

Page 7: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

391H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

water of the bog lake is shown in Fig. 1a. The maximum of ion intensity is observedat approx. 370°C. Volatile matter of 56.7% and total ion intensity (TII) of20.4×106 counts mg−1 sample are obtained. The mass range of recorded ions is inthe range from m/z 18 to 450. The spectrum displays good intensities between m/z18 and 394 with the base peak at m/z 110.

In comparison with this Py-FI mass spectrum of DOM, much higher intensitiesare obtained from the corresponding fulvic avid (FA; code no. HO10-FA, 3/1995),with recorded FI signals between m/z 18 and 500, base peak m/z 178, and themaximum of ion intensity at 460°C. The FA sample gave 65.7% volatile matter andTII 346×106 counts mg−1. In contrast, the humic acid (HA; HO10-HA, 02/95) inFig. 1c shows a relatively weak Py-FI spectrum (maximum of ion intensity at400°C; 54.7% volatile matter; TII 15.4×106 counts mg−1) and with only few veryintense signals in a narrow mass range. The weight-averaged (M( w) and number-av-eraged (M( n) molecular weights in the Py-FI mass spectra of DOM (247.1, 195.7),FA (300.2, 271.3), and HA (180.3, 112.1) and polydispersity values were 1.3, 1.1,and 1.6, respectively. These data reflect the thermal properties of the aquaticmaterials upon in-source volatilization with a heating rate of about 1°C s−1 andfinal effective temperature around 600°C. As expected the weight- and number-av-eraged molecular weights drop in the sequence FA\DOM\HA indicating struc-tural cross-linking and thus higher thermal stability. Polydispersity of the FApyrolysis products can be explained by high abundances in a relatively narrow andhigh temperature range, whereas the HA shows only few, widely different butintense thermal products.

This thermal behaviour is characteristic for a large number of DOM, FA, andHA samples taken at different times during almost 5 years. With some experiencethese three classes of aquatic materials are easily distinguished only by visualinspection. Distinction of large sets of aquatic and terrestrial humic fractions andoriginal water or soil specimen is readily available using fingerprinting and patternrecognition methods.

As shown in Fig. 2, one method is to establish series of marker signals which aretypical of different classes of compounds. The ten biomarkers are volatile fattyacids in the low temperature range of thermal volatilization up to 220°C (in-sourcevacuum, 56.7% volatile matter); suberin derived from plant bark and roots; signalsderived mainly from 22 essential amino acids, peptides, and aliphatic amides;thermostabile sterols such sitosterol, but also tocopherol, cholesterol etc.; Aromaticnitrogen (N) compounds, a wider range of alkylaromatic compounds and lipids,characteristic signals for lignin dimers and phenols/lignin monomers, and finallycarbohydrates. Fig. 2a shows the intensities in %TII which gives an estimate of theabsolute ion currents which obviously differ strongly in the direction FA\DOM\HA. It can be assumed that the volatilization and ionization efficienciesare similar for the marker substances in structurally related materials such asDOM, FA, and HA. Thus, these data are quantitatively applicable to comparisonswithin classes of environmental materials and also with different samples of similarorigin, e.g. aquatic and terrestrial humic substances, DOM and SOM. Definitivequantifications using internal standards, preferably stable isotope dilution appear

Page 8: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

392 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Fig. 2. Proportions of ten important structural classes of building blocks of the Hohloh Lake waterobtained by Py-FIMS: (a) absolute intensities, and (b) relative abundances of DOM and humic fractions.

difficult as homogenization of authentic sample and labelled standard poses prob-lems. The relative abundances given in Fig. 2b are easily used for preliminarycomparisons. Taking into account the strong differences in TII, it is clear that theerror in the HA abundances must be much greater compared with that infreeze-dried water and FA.

Page 9: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

393H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

3.2. Bog lake water (Py-GC/MS)

Under the experimental conditions described for Curie-point pyrolysis with afinal temperature of 500°C, the chromatogram shown in Fig. 3 of freeze-dried lakewater was obtained. Fourteen intense peaks in the chromatogram were numberedfor orientation and the following extensive step by step interpretation of thechromatogram by library searches gave the results in Table 1. In this Table on theleft-hand side the numbers for the magnetic scan Table 1 of the double-focusinginstrument are listed. In addition, the molecular weight (MW); purity (in% agree-ment with library data); Chemical Abstract Numbers (CAS); elemental composition(C, H, O, N, X=heteroatom); and CAS name of the identified compound aregiven. It is noteworthy that only tedious operator controlled evaluation is possibleas there is a plurality of overlapping peaks present. The cursor has to be usedcarefully for background corrections and different operators have checked theassignments at least twice. Different columns, Py-GC conditions (split, gas flowrate) and mass spectrometer conditions (sampling rate, mass window etc,) have tobe optimized. Moreover, repeating these measurements and identifications over 5years for specimen from the same Hohloh Lake (1995 to 1998) and differentseasons of the year, strongly confirmed the correct identifications. One of the mostimportant points is that the ‘Purity’ and ‘Fit’ given by the data system and the280 000 standards library (Wiley and in-house data) are only acceptable afterextensive experience and scepticism. On the other hand the Py-FIMS results,

Fig. 3. Curie-point pyrolysis-GC/MS of Hohloh Lake original, freeze-dried water.

Page 10: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

394H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 1Identification of constituents of the Hohloh Lake original tentatively assigned on the basis of Py-GC/MS and library searches

Scan MW Purity CAS no. C H O N X

39 94 64 74-83-9 1 3 – – Br Methane-bromo-82 86 534-22-5 551 6 1 – – Furan, 2-methyl-

61 78 82 71-43-2 6 6 – – – Benzene96 88 625-86-5 673 8 1 – – Furan, 2,5-dimethyl-96 79 – 676 8 1 – – 2,4-Dimethylfuran96 83 2758-18-1 684 8 1 – – 2-Cyclopenten-1-one, 3-methyl

80 94 89 108-95-2 6 6 1 – – Phenol67 94 109-97-7 496 5 – 1 – 1H-Pyrrole92 89 108-88-3 7102 8 – – – Benzene, methyl-92 89 544-25-2 7104 8 – – – 1,3,5-Cycloheptatriene60 88 64-19-7 2141 4 2 – – Acetic acid96 98 98-01-01 5153 4 2 – – 2-Furancarboxaldehyde82 89 930-30-3 5159 6 1 – – 2-Cyclopenten-1-one79 93 110-86-1 5163 5 – 1 – Pyridine

171 106 68 100-41-4 8 10 – – – Benzene, ethyl-106 93 106-42-3 8178 10 – – – Benzene, 1,4-dimethyl-98 79 591-11-7 5179 6 2 – – 2(5H)-Furanone, 5-methyl-96 68 – 5180 4 2 – – Cyclopent-2-en-1,4-dione84 18 – 4182 4 2 – – 1,4-Dioxadiene106 71 95-47-6 8193 10 – – – Benzene, 1,2-dimethyl-84 54 – 8197 14 1 – – 2-propyl-cyclopentanone96 81 1120-73-6 6205 8 1 – – 2-Cyclopenten-1-one, 2-methyl-110 79 764-13-6 8208 14 – – – 2,4-Hexadiene, 2,5-dimethyl-93 93 108-99-6 6212 7 – 1 – Pyridine, 3-methyl-84 63 – 2235 4 2 – – 2(5H)-Furanone120 86 98-82-8 9237 12 – – – Benzene, (1-methylethyl)-98 63 591-11-7 5239 6 2 – – 2(5H)-Furanone, 5-methyl-106 65 100-52-7 7239 6 1 – – Benzaldehyde110 93 620-02-0 6241 6 2 – – 2-Furancarboaldehyde, 5-methyl-120 65 611-14-3 9247 12 – – – Benzene, 1 -ethyl-3 -methyl-

Page 11: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

395H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 1 (Continued)

Scan MW Purity CAS no. C H O N X

256 120 94 526-73-8 9 12 – – – Benzene, 1,2,3-trimethyl-96 86 2758-18-1 6257 8 1 – – 2-Cyclopenten- 1-one, 3-methyl-

257 118 88 271-89-6 8 6 1 – – Benzofuran110 68 61892-54-4 6265 6 2 – – 3-Methyl-5-methyliden-2(5H) furanone94 981 98-67-9 6268 6 4 – S Benzenesulfonic acid, 4-hydroxy-120 65 108-67-8 9270 12 – – – Benzene, 1,3,5-trimethyl-134 50 99-87-6 10273 14 – – – Benzene, 1 -methyl-4-(1 -methylethyl)-112 88 80-71-7 6276 8 2 – – 2-Cyclopenten-1-one, 2-hydroxy-3-methyl-

276 128 83 95887 8 13 – – F 3-Fluoro-2,5-dimethyl-2,4-hexadiene291 110 81 1121-05-7 7 10 1 – – 2,3-Dimethyl-2-cyclopenten-1-one

120 85 98-86-2 8295 8 1 – – Ethano, 1-phenyl-297 98 83 – 5 6 2 – – 4-(methyl-2(5H)-furanone

108 95 95-48-7 7298 8 1 – – Phenol, 2-methyl-132 79 102835-72-3 10300 12 – – – 6,6-dimethyl-1-vinylfulven124 95 90-5-1 7306 8 2 – – Phenol, 2-methoxy-

308 108 95 106-44-5 7 8 1 – – Phenol, 4-methyl-132 81 4265-25-2 9314 8 1 – – Benzofuran, 2-methyl-

315 122 76 90-00-6 8 10 1 – – Phenol, 2-ethyl-322 98 54 98-00-0 5 6 2 – – 2-Furanmethanol

122 71 620-17-7 8333 10 1 – – Phenol, 3-ethyl-337 122 91 105-67-9 8 10 1 – – Phenol, 2,4-dimethyl-

134 64 527-53-7 10343 14 – – – Benzene, 1,2,3,5-tetramethyl-122 85 123-07-9 8344 10 1 – – Phenol, 4-ethyl-122 93 526-75-0 8346 10 1 – – Phenol, 2,3-dimethyl-

348 122 93 576-26-1 8 10 1 – – Phenol, 2,6-dimethyl-134 76 122-00-9 9351 10 1 – – Ethanone, 1-(4-methylphenyl)-138 81 93-51-6 8351 10 2 – – 2-Methoxy-4-methylphenol122 81 95-65-8 8355 10 1 – – Phenol, 3,4-dimethyl-146 91 28715-26-6 10360 10 1 – – Benzofuran, 4,7-dimethyl-135 24 95-16-9 7369 5 – 1 S Benzothiazole139 91 20189-42-8 7370 9 2 1 – 1H-Pyrrole- 1,5 -dione, 3-ethyl-4-methyl-134 93 – 8371 6 2 – – 2-Coumaranone

Page 12: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

396H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 1 (Continued)

MW Purity CAS no. C H OScan N X

144 88 18636-55-0 11378 12 – – – 1H-Indene, 1,1-dimethyl-384 152 63 2785-89-9 9 12 2 – – Phenol, 4-ethyl-2-methoxy-

132 92 83-33-0 9389 8 1 – – 1H-Inden-1-one, 2,3-dihydro-398 150 79 – 9 10 2 – – 4-Vinyl-2-methoxy-phenol

162 79 4792-30-3 9 6 3415 – – 1,3-Isobenzofurandione, 4-methyl-156 94 582-16-1 12438 12 – – – Naphthalene, 2,7-dimethyl-220 51 128-37-0 15461 24 1 – – Phenol, 2-6-bis(1,1-dimethylethyl)-4-methyl-204 81 35112-28-8 8467 6 2 – Cl2 Benzoic acid, 2,4-dichloro-, methyl ester

467 188 75 – 14 20 – – – 1,1,4,6,7-Pentamethyl-2,3-dihydroindene128 54 2051-78-7 7487 12 2 – – Butanoic acid, 2-propenyl ester

490 170 82 829-26-5 13 14 – – – Naphthalene, 2,3,6-trimethyl-

Page 13: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

397H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Fig. 4. Curie-point pyrolysis-GC/MS of Hohloh Lake fulvic acid.

particularly those based on high mass resolution and accurate mass measurementsdeliver the second independent proof whether the proposed identification is accept-able. Table 1 illustrates that compounds with heteroatoms such F, Cl, S, and Br areclearly detectable. As expected for DOM about 75% of the listed compoundscontain one or more oxygens. Also it should be noted that only a few aliphaticcompounds are found and by far the majority are substituted olefinic rings oraromatics.

As indicated in the chromatogram of aquatic FA of the Hohloh Lake 16 intensepeaks obtained by Py-GC/MS are labelled for orientation (Fig. 4). In Table 2 morethan 40 compounds have been identified which is clearly less than for freeze-driedwater. This is astonishing because the higher volatility and intense Py-FI massspectra of the aquatic FA would suggest higher intensities also for Py-GC/MS.

In comparison with the results of the aquatic FA the corresponding HA shows 17numbered intense peaks in the chromatogram obtained by Py-GC/MS (Fig. 5).Despite dramatically lower intensities in Py-FIMS, the chromatogram of aquaticHA is well-resolved, has a good baseline and consequently a very high degree ofagreement with the library spectra of standards. The evaluation of this chro-matogram shows the best average purity in the structural assignments of more than60 pyrolysate components (Table 3). This table shows practically exclusivelyaromatic compounds as can be expected for a higher degree of cross-linking in HAin comparison with DOM and FA.

Page 14: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

398H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 2Identification of constituents of the Hohloh Lake fulvic acid tentatively assigned on the basis of Py-GC/MS and library searches

Scan MW Purity CAS no. C H O N X

81 82 97 534-22-5 5 6 1 – – Furan, 2-methyl-96 88 625-86-5 6135 8 1 – – Furan, 2,5-dimethyl-

142 94 91 98-67-9 6 6 1 – – Phenol60 78 64-19-7 2147 4 2 – – Acetic acid92 98 544-25-2 7176 8 – – – 1,3,5-Cycloheptatriene72 90 79-10-7 3198 4 2 – – 2-Propenoic acid

234 96 90 3742-34-5 7 – – – – Cyclopentane, ethenyl-82 85 1453-58-3 4237 6 – 2 – 1H-Pyrazole, 3-methyl-82 67 616-47-7 4240 6 – 2 – 1H-Pyrazole, 1-methyl-98 87 1942-42-3 6261 – 1 – – Cyclobutanone, 2,3-dimethyl-, trans-86 81 1120-59-8 4265 6 – – S Thiophene, 2,3-dihydro-96 80 108-97-4 5272 4 2 – – 4H-Pyran-4-one106 71 95-47-6 8275 – – – – Benzene, 1,2-dimethyl-96 96 1120-73-6 6287 8 1 – – 2-Cyclopenten-1-one, 2-methyl-84 91 20825-71-2 4301 4 2 – – 2 (3H)-Furanone98 84 17108-52-7 6315 – 1 – – Furan, 2,3-dihydro-2,5-dimethyl-

320 120 85 98-82-8 9 – – – – Benzene, (1-methylethyl)-323 110 95 620-02-0 6 6 2 – – 2-Furancarboxaldehyde, 5-methyl-

96 88 2758-18-1 6329 8 1 – – 2-Cyclopenten-1-one, 3-methyl-339 174 97 98-67-9 6 6 4 – S Benzenesulfonic acid, 4-hydroxy-

120 83 622-96-8 9356 – – – – Benzene, 1-ethyl-4-methyl-134 89 25155-15-1 –357 – – – – Benzene, methyl(1-methylethyl)-110 80 13643-06-6 8361 – – – – 1,6-Heptadiene, 2-methyl-

369 110 82 1193-18-8 7 – 1 – – 2-Cyclohexen-1-one, 3-methyl-108 88 95-58-7 7376 8 1 – – Phenol, 2-methyl-108 92 106-44-5 7388 8 1 – – Phenol, 4-methyl-124 79 55683-21-1 8392 – 1 – – 2-Cyclopenten-1-one, 3,4,5-trimethyl-110 78 123-31-9 6399 6 2 – – 1,4-Benzenediol110 87 13327-27-0 5400 6 1 2 – 3(2H)-pyridazione, 6-methyl-132 72 4265-25-2 9401 8 1 – – Benzofuran, 2-methyl-112 75 624-29-3 8407 – – – – Cyclohexane, 1,4-dimethyl-, cis-

Page 15: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

399H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 2 (Continued)

MW Purity CAS no. C HScan O N X

147 83 1126-16-5 9421 9 1 1 – Cyanic acid, 2,4-dimethylphenyl ester122 89 123-07-9 8429 – 1 – – Phenol, 4-ethyl-138 84 2896-60-8 8442 – 2 – – 1,3-Benzenediol, 4-ethyl-110 91 120-80-9 6446 6 2 – – 1,2-Benzenediol120 86 494-16-2 8451 8 1 – – Benzofuran, 2,3-dihydro-

461 134 87 5905-00-0 8 6 2 – – 2,2%-Bifuran140 71 934-00-9 7472 8 3 – – 1,2-Benzenediol, 3-methoxy134 80 1746-11-8 9486 – 1 – – Benzofuran, 2,3-dihydro-2-methyl-150 85 1450-72-2 9493 – 2 – – Ethanone, 1-(2-hydroxy-5-mehylphenyl)-

507 148 71 119-84-6 9 8 2 – – 2H-1-Benzopyran-2-one, 3,4-dihydro-154 83 91-10-1 8508 – 3 – – Phenol, 2,6-dimethoxy-

532 162 66 54365 – – 2 – – 3(2H)-Benzofuranone, 2,7-dimethyl-

Page 16: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

400H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 3Identification of constituents of the Hohloh Lake humic acid tentatively assigned on the basis of Py-GC/MS and library searches

Scan MW Purity CAS no. C H O N X

80 82 92 534-22-5 5 6 1 – – Furan, 2-methyl-94 86 108-95-2 6143 6 1 – – Phenol

179 92 98 108-88-3 7 8 – – – Benzene, methyl-98 92 554-14-3 5183 6 – – S Thiophene, 2-methyl-98 92 616-44-4 5189 6 – – S Thiophene, 3-methyl-110 73 5771-32-4 7221 – 1 – – Spiro[2,4]heptan-4-one-

238 96 95 98-01-01 5 4 2 – – 2-Furancarboaldehyde79 99 110-86-1 5240 5 – 1 – Pyridine106 97 95-47-6 8263 – – – – Benzene, 1,2-dimethyl-96 97 1120-73-6 6292 8 1 – – 2-Cyclopenten-1-one, 2-methyl-110 94 1192-62-7 6295 6 2 – – Ethanone, 1-(2-furanyl)-84 93 20825-71-2 2320 4 2 – – 2(3H)-Furanone120 87 98-82-8 9326 – – – – Benzene, (1-methylethyl)-110 95 620-02-0 6329 6 2 – – 2-Furancarboaldehyde, 5-methyl-110 88 13327-27-0 5329 6 1 2 – 3(2H)-Pyridazinone, 6-methyl-20 92 108-67-8 9345 – – – – Benzene, 1,3,5-trimethyl-

348 174 98 98-67-9 6 6 4 – S Benzensulfonic acid, 4-hydroxy-362 120 92 95-63-6 9 – – – – Benzene, 1,2,4-trimethyl-

134 93 535-77-3 –363 – – – – Benzene, 1-methyl-3-(1-methylethyl)-383 108 96 95-48-7 7 8 1 – – Phenol, 2-methyl-

120 85 98-86-2 8387 8 1 – – Ethano, 1-phenyl-108 94 106-44-5 7395 8 1 – – Phenol, 4-methyl-124 89 90-5-1 7398 8 2 – – Phenol, 2-methoxy-

406 131 84 4265-25-2 9 8 1 – – Benzofuran, 2-methyl-131 94 17059-52-8 9409 8 1 – – Benzofuran, 7-methyl-122 94 108-68-9 8426 – 1 – – Phenol, 3,5-dimethyl-122 93 123-07-9 8434 – 1 – – Phenol, 4-ethyl-138 91 93-51-6 8447 – 2 – – 2-Methoxy-4-methylphenol110 92 120-80-9 6449 6 2 – – 1,2-Benzenediol120 87 496-16-2 8456 8 2 – – Benzofuran, 2,3-dihydro-166 87 22446-37-3 9466 – 3 – – Benzeneacetic acid, 2-hydroxy-, methyl ester

Page 17: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

401H

.-R.

Schulten

/J.

Anal.

Appl.

Pyrolysis

49(1999)

385–

415

Table 3 (Continued)

Scan MW Purity CAS no. C H O N X

134 83 5905-00-0 8469 6 2 – – 2,2%-Bifuran124 91 488-17-5 7474 8 2 – – 1,2-Benzenediol, 3-methyl-

476 240 43 13019-43-7 – – 3 – – Naphtho[2,3-B]furan-4,9-dione, 2-isopropyl-140 91 934-00-9 7 8477 3 – – 1,2-Benzenediol, 3-methoxy-152 85 2785-89-9 9483 – 2 – – Phenol, 4-ethyl-2-methoxy-

497 150 90 1450-72-2 9 – 2 – – Ethanone, 1-(2-hydroxy-5-methylphenyl)-154 91 91-10-1 8 –512 3 – – Phenol, 2,6-dimethoxy-152 92 121-33-5 8531 8 3 – – Benzaldehyde, 4-hydroxy-3-methoxy-

545 136 92 99-93-4 8 8 2 – – Ethanone, 1-(4-hydroxyphenyl)-

Page 18: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

402 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

In summarizing the results of about 5 years of Py-FIMS and Py-GC/MS offreeze-dried water and the derived aquatic FA and HA of the bog lake water(approx. 30 specimen in total and more than 200 single measurements, includingrepetitions), we assembled the following two schemes for a survey of the identifiedDOM building blocks:

As demonstrated in Scheme 1, the identified building blocks of DOM (O=orig-inal bog lake water), fulvic acid (F=FA), and humic acid (H=HA) are highlyabundant as shown in Figs. 3–5 and cover a wide variety of different chemicalstructures which are exclusively substituted Scheme 1 aromatics such as benzenes(1a, 42 structures), phenols (1b, 26 structures), and furans (1c, 35 structures).Interestingly, a general trend is observed that the majority of identified compoundsare present in the original freeze-dried water sample. The F and H structures arefound mostly together with O and partly complement each other to approximatethe DOM composition. There are, however, also unique F and H structuresdetected and it could expand the knowledge of artifact formation during humicfractionations. Since the building blocks are ordered clockwise according to increas-ing retention times during GC/MS, it is noteworthy that the (FA) and (HA)assignments in Scheme 1a–c either appear together with (O) or show a slightincrease towards the end of the GC runs which might indicate thermal stress of theoriginal DOM constituents. But here obviously more analytical work and support-ing data are necessary to verify these effects.

Fig. 5. Curie-point pyrolysis-GC/MS of Hohloh Lake humic acid.

Page 19: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

403H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Scheme 1.

In Scheme 2 the (d) aliphatic; (e) heterocylic; (f) carbocyclic; as well as (g)naphthalene-; (h) indene-; (i) sulfur-; and (j) halogen-derived compounds aredisplayed which were found by Py-GC/MS and library searches of DOM and thehumic fractions. The cyclic structures (e, f, g, h) are by far the majority of thecompounds identified in the pyrolysates. Considering in addition to the substitutedbenzenes, phenols, and furans heterocyclics in Scheme 1, the aliphatic compoundswith lower thermal stability and volatility, but higher ionization potentials appar-ently are suppressed. Of high interest are the detected sulfur-, halogen- andnitrogen-derived compounds [25] which are frequently involved in reactions ofwater with contaminants.

Page 20: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

404 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

With respect to the four questions raised above, the combination of Py-FIMSand Py-GC/MS allows a reliable identification of the thermal fragments of aquaticmaterials. Recent work by other groups partly confirms these findings using Py-MS(low-energy electron impact ionization; chemical ionization) and Py-GC/MS [26].Our knowledge of structural building blocks of aquatic materials such as DOM,FA, and HA, illustrates a wide variation of organic chemical structures with a coreof mostly substituted aromatics that are obviously ubiquitous. Investigations totrack the pathways and origin of the pyrolysis products are in progress using bulkisotopic analyses and GC/MS coupled with compound-specific isotope-ratio mea-surements [27]. One preliminary encouraging result is that the isotopic pattern of Cand N atoms is maintained during the pyrolysis processes, so that this isotopicmemory of the pyrolysate components can be compared, for instance, with plantprecursors. At present it is difficult to study DOM processes, mainly because noaccepted molecular structure exists, association/dissociation processes are onlypoorly understood and thus molecular weight determinations are still shaky.

Therefore an attempt was made to simulate a DOM particle on the basis of thedescribed results of analytical pyrolysis and further development of a proposedSOM model [22]. The idea is to initiate co-operation with other groups working inthis area of research and for improving the proposed hypothetic molecular structureby fitting and improving with hard analytical data of different methods.

3.3. Modeling of humic colloids and dissol6ed organic matter

For humic particles the vast number of different structural variations and thehigh capacity of trapping and binding of inorganics (gases, water, minerals) is

Fig. 1. (Continued)

Page 21: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

405H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Scheme 2.

characteristic. The interactions with biological (carbohydrates, peptides, lignins,etc.) and anthropogenic substances are even more pronounced. The operationaldefinition for humic colloids is that they are considered to range in diameterbetween 1 nm and 1 mm [9]. In order to demonstrate the capacity of the abovementioned hard- and software set-up and to indicate the present limits, aquatichumic colloids with an approximate diameter between 5 and 10 nm were con-structed. Basis for the structural design were the generally observed building blocks

Page 22: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

406 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

identified by analytical pyrolysis as described above. Another guideline was toapproximate an averaged elemental composition of aquatic materials, preferablyoriginal, freeze-dried water.

Fig. 2. (Continued)

Page 23: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

407H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Scheme 3.

3.3.1. DOM monomerAs starting material, the monomer SOM structure consisting of a humic structure

which binds a hexapeptide and trisaccharide via hydrogen bonds has been em-ployed [22] and was upgraded to higher polarity and oxygen content by inserting 20additional carboxylic functions. Most of the carboxyls were attached to aromaticrings by hydrogen substitution. Moreover the C6–C8 aliphatic chains with terminalcarboxyl funtions were connected to the carbon skeleton. As the oxygen contentwas still too low, the number of phenolic groups was increased in a stepwisemanner. Finally, in total 35 water molecules were placed in voids and close toreactive surface areas. In this manner the three boundary conditions of (a) size; (b)structural building blocks and (c) elemental composition resulted in a DOM modelwhich consisted of 1260 atoms and had a molecular weight of about 11 000 gmol−1 (Scheme 3). In Fig. 6 the molecular cluster of the constructed DOM modeland the 37 trapped molecules (hexapeptide, trisaccharide and 35 water molecules) isdisplayed in black sticks for the atomic bonds without multiple bonds, particalcharges, inertial axes, dipol moments, etc.

Page 24: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

408 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

One important aspect of humic, SOM and also DOM structures is apparently theformation of hydrogen bonds as postulated 20 years ago for soil humic substancesby Schnitzer [28]. As a result of molecular mechanics calculations (MM+ ), theH-bonds with progressing energy minimization were found mostly inside the DOMmodel structure with only a few exceptions on colloidal surfaces. For the modelstructure shown in Fig. 6 in total 22 H-bonds were observed and are marked bydashed lines.

Firstly, seven water molecules are H-donors for H-bonds as indicated by thedashed lines and labelling of the atom numbers 2 and 3. Five types of hydrogenbridges are observed with the following moieties: (1) aliphatic hydroxyl (×2); (2)aromatic hydroxyl; (3) aliphatic carboxyl; and (4) chinoid keto-group. In addition,at the described stage of energy minimization two H-bonds with formation of waterclusters are obtained. At no stage did water molecules leave the DOM. However,the mobility of the water molecules due to energetically favourable bonds inside theDOM structure is high and subsequently frequent changes in water—water bondsare shown during MM+ calculations. When the DOM colloid is submersed in awater box hydrogen [29] hydrogen bonding increases dramatically at the solvent-ac-cessible surface which was determined by the Chemplus QSAR software as animportant molecular property (Scheme 3: volume 52.5811 nm3). From the results of

Fig. 6. Molecular visualization and simulation of a DOM model (1262 atoms) derived from previouslypostulated HA and SOM models. This display shows the molecular skeleton by sticks for atomic bonds.During MM+ calculations and structure optimization in total 20 H-bonds are formed and are indicatedby dashed lines and the atom numbers of the participating H-donors.

Page 25: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

409H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Fig. 7. Color plot of the 3D model proposed DOM model. The elements colors are: carbon (cyan),hydrogen (white), oxygen (red), nitrogen (blue), and sulfur (yellow).

distance measurements in the voids and cleft of the DOM model it appears difficultor unlikely for water clusters or large anions such as OH− to penetrate into theDOM skeleton.

Secondly, 15 H-bonds in DOM are indicated by dashed lines for the hydrogenbonds and labelled by the donor atom number in Fig. 6. Abundant are the aliphatichydroxyls as donors of H-bonds to aromatic carboxyls (175; atom numbers italicsand in brackets); chinoid keto functions (610, 685); and aliphatic hydroxyls (833,985). Moreover, the aromatic hydroxyls generate H-bonds to aliphatic carboxyls(688, 778), aromatic carboxyls (764), aliphatic alcohols (781), and water (772). FourH-bonds are formed by aliphatic carboxyls to two aromatic carboxyls (571, 1046)and ester carbonyl (592) functions. Finally, the arginine residue in the trappedhexapeptide shows an intramolecular H-bond to the neighbouring imine moiety ofthe peptide bond to glutamic acid.

In Fig. 7 the color plot of the DOM model displayed as Balls and Cylinders givesa comprehensive look of the proposed structural details, the distribution ofheteroatoms such as N and S, but also an impression of the rigid and quitecondensed packing which leaves very small voids when displayed with full van derWaals radii. The corresponding chemical data such as molecular weight, elementalcomposition and analysis, total energy and bond-, angle-, dihedral-, van der Waals-,

Page 26: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

410 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

stretch-bend-, and electrostatic energy components calculated by computationalchemistry are listed in Scheme 3. The total energy of the model is determined as thesum of potential chemical energy of the DOM structure. The usually very highstarting energy is introduced by the preliminary drawing of the operator into theworkspace. During the process of energy minimization, controlled by the employedforce field and default parameters of the HyperChem software, step by step theconformation of the chemical model structure is improved. This leads to theobserved mobility of atoms, functional groups, residues, molecules and formationand rupture of van der Waals complexes and hydrogen bonds. The design of thesoftware does not allow to cleave or cross covalent bonds unless the operatorintervenes in this manner. It is this conservative feature which offers the chance toapproach a yet unknown structure by incorporating improvements derived fromhard data of other chemical, biological, and physical methods on the long marchtowards an accepted and working DOM or SOM realization.

Particularly in a macromolecular model structure of high aromaticity andpolarity, the non-bonded interactions (van der Waals energy) and hydrogen bonds(electrostatic energy) play an important role. Indeed, computational chemistry(Scheme 3) of the proposed DOM model gave a high van der Waals energy ofabout 517 kJ mol−1 and pronounced energy gain due to electrostatic energy ofabout -194 kJ mol−1, respectively. Thus, the observation of 24 H-bonds isconsistent with the molecular mechanics calculations.3.3.2. DOM tetramer

In a first approach to examine the association/dissociation reactions of DOMmolecules, four DOM monomers were moved into the x and y-axes at distanceswhich should allow to investigate the chemical interactions. As shown in Fig. 8, thetetramer cluster structure (5048 atoms) is assembled approximately rectangular tothe inertial axes of the monomers and forms a colloid with a thickness z of 2.25 nm.The shape of the thin sheet shows a central gap, four overlapping areas, and theouter atoms at radii of about 5 nm from the structural center. The colloid has amolecular weight of 44139.4919 g mol−1 and the elemental compositionC1788H1968O1224N60S8. Geometry optimization (energy minimization) gave a totalenergy of 11 164.22 kJ mol−1 at a convergence gradient of 2.11 kJ mol−1 nm−1. Atthis stage the different energy components were calculated (in kJ): bond energy(2278.3); angle (7165.57); dihedral (5384.78); van der Waals (2962.57); stretch-bend(134.46); and electrostatic (−761.44). Compared with the total energy of themonomer DOM (Scheme 3), the tetramer showed a considerably higher totalenergy of about 3500 kJ which was mainly due to strongly increased bond- and vander Waals energies due to narrow range interactions. Slightly raised or similarenergies were calculated for angle- and dihedral components whereas stretch-bendand electrostatic contributions dropped. The latter is particularly interesting aslower energies were observed with progressing MM+ calculations despite theformation of more hydrogen bonds inside the DOM structure. In comparison withmolecular properties of the monomer, the tetramer showed a slight relative decreasein solvent-accessible surface and volume whereas the van der Waals surface andvolume stayed constant as expected.

Page 27: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

411H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

As displayed in Fig. 9 the total energy dropped strongly between the initial andfinal stage of geometry optimization and might allow a relative interpretation ofstructural changes in terms of discrete energies. Apparently, the van der Waalsenergies show a very similar trend during increasing numbers of calculation cycles.In contrast, the electrostatic energy of the tetramer DOM cluster gradually de-creases. Thus, the capacity for investigations of association/dissociation reactions at

Fig. 8. DOM colloid assembled from four DOM model structures (5048 atoms) illustrates the covalentbonds, distances and angles of this conformation with optimized voids and clefts for trapping andbinding of biological and anthropogenic substances.

Page 28: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

412 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Fig. 9. For the colloid, the total energy (upper curve) and two energy components, such as van derWaals (middle curve) and electrostatic energies (lower curve) are displayed with increasing number ofMM+ calculation cycles.

atomic and molecular level were indicated and were characterized by bond typesand energy components; and selected sections of inter- and intramolecular H-bonds.

Summarizing these results the following DOM properties emerge: (a) Hydrogenbonds by hydroxyl and carboxyl groups increased in comparison with humicsubstances and soil organic matter; (b) The majority of the H-bonds are generatedin the central portions of the DOM model and only few in the surface area; (c)

Page 29: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

413H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

Intermolecular H-bonds are scarce and quite difficult to generate by the largecolloid surfaces and strong intramolecular forces; (d) Mostly shielded intramolecu-lar H-bonds are found which appear to be flexible and changing positions duringMM+ calculations; (e) In the aqueous phase, penetration by OH- and hydroniumwater clusters into the DOM colloid is hindered by the spacefilling and hydropho-bicity of the structural DOM network. Trapped biological and/or xenobioticsubstances may show high persistence as long as the protecting cover of DOMmolecules is stabilized. Nanochemistry and computational chemistry offer possibil-ities to investigate these phenomena and processes at the lowest chemical level ofsingle atoms and bonds; (f) Investigations by ChemPlus/QSAR software, allowcalculations of DOM models to determine molecular properties such as mass,surface area, volume, partial charges (electronegativity), polarizability, refractivity,log P (hydrophobicity), hydration energy and derived values for specific surface anddensity. At present the missing link (connectivity) between molecular properties andenvironmental activities is urgently needed to produce QSAR properties for controland prediction of in vivo environmental processes.

4. Critical evaluation and outlook

4.1. Analytical pyrolysis

Temperature-programmed, time-resolved pyrolysis-field ionization mass spec-trometry of aquatic humic fractions and colloids is a powerful tool for fingerprint-ing and characterization, but is limited by the low volatility of highly polarconstituents and cross-linked portions in the heterogenous structures of DOM.Curie-point pyrolysis-gas chromatography/mass spectrometry yields complemen-tary information which supports the identification of the pyrolysis products.However, the very rapid transfer of thermal energy (flash pyrolysis) in the order ofmilliseconds to the analytical specimen produces preferentially smaller productscompared with Py-FIMS. In addition, libraries available at present even with massspectral collections far larger than 280 000 standard spectra (including in-housedata), may not contain spectra of the materials mentioned above. More critical isthe situation in which the libraries offer a reasonable fit of the pyrolysis spectrawith stored electron impact mass spectra—but the assignments are completelywrong (as mainly standard spectra from synthetic, environmental, pharmaceutical,etc., areas have been collected over the last decades). In this case, it is theexperience of the spectroscopist and the wealth of in-house spectra collections ofbiopolymers, plants, humic fractions, terrestrial and aquatic materials, etc., thathave to take over in order to arrive at tentative structural assignments.

4.2. Structural modeling and geometry optimization

Computational chemistry which allows to draw, construct and optimize in 3Dspace biomacrolecules, such as aquatic and terrestrial humic substances, DOM, and

Page 30: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

414 H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

SOM with precise bond distances, bond angles, torsion angles, non-bonded dis-tances, hydrogen bonds, charges, and chirality is a powerful method in combinationwith analytical pyrolysis. Molecular visualization and simulation is hoped to shedmore light on the structure and dynamics of humic and dissolved organic matter asis expected from related methods. One limiting factor is of course that modeling inwater requires very large numbers of water molecules to produce the reactive waterbox and calculations make it mandatory to use large computers.

4.3. En6ironmental consequences

Preliminary studies showed the potential of Py-MS for investigating contami-nated water [30] and modeling of trapping and bonding of plastizisers in water andsoils [31]. By comparing spectra and thermograms of sewage water and soils ofsewage farms with non-contaminated sites nearby, the spectral markers for sewagecontaminations and their thermal properties were derived. Moreover, the firstresults on models for the interactions of dissolved humic-biocide complexes andwater have been put forward [32].

The principal aim and long-term strategy is to develop structural concepts ofDOM and SOM [33] which explain their properties as well as the ecologicalmeaning, dynamic character and structure–property relationships. Apparently,processes of binding and releasing anthropogenic substances, e.g. biocides, are notcompletely understood only by global parameters (e.g. elemental composition,labelling, extractions, conventional spetrocopic techniques) and require molecularand atomic understanding of the relevant processes.

Acknowledgements

This research was partially supported by the Deutsche Forschungsgemeinschaftin the priority program: ‘Refractory Organic Substances in the Environment’,(ROSIG, Schu 416/18-4), Bonn-Bad Godesberg, Germany. The author is verygrateful to Dr. G. Abbt-Braun and Prof. F. Frimmel (University Karlsruhe,Engler-Bunte-Institute, Germany) for kindly providing the ROSIG reference mate-rials and their co-operation.

References

[1] M. Schnitzer, S.U. Khan, Humic Substances in the Environment, Marcel Dekker, New York, 1972,pp. 1–327.

[2] M. Schnitzer, S.U. Khan (Eds.), Soil Organic Matter, Elsevier, Amsterdam, 1978.[3] F.H. Frimmel, R.F. Christman (Eds.), Humic Substances and Their Role in the Environment,

Dahlem Workshop Reports, Wiley, Chichester, 1988.[4] J.M. Bracewell, K. Haider, R. Larter, H.R. Schulten, in: M.H.B. Hayes, P. MacCarthy, R.L.

Malcolm, R.S. Swift (Eds.), Humic Substances II; Search of Structure, Wiley, New York, 1989, pp.181–222.

Page 31: Analytical pyrolysis and computational chemistry of aquatic humic substances and dissolved organic matter

415H.-R. Schulten / J. Anal. Appl. Pyrolysis 49 (1999) 385–415

[5] H.-R. Schulten, J. Anal. Appl. Pyrolysis 25 (1993) 97.[6] N. Senesi, T.M. Miano (Eds.), Humic Substances in the Global Environment and Implications on

Human Health, Elsevier, Amsterdam, 1994, pp. 3–1368.[7] D.L. Sparks, Environmental Soil Chemistry, Academic Press, San Diego, 1995.[8] H.-R. Schulten, in: S. Yamasaki, T.W. Boutton (Eds.), Mass Spectrometry of Soils, Marcel Dekker,

New York, 1996, pp. 373–436.[9] H.-R. Schulten, P. Leinweber, M. Schnitzer, in: P.M. Huang, N. Senesi, J. Buffle (Eds.), Structure

and Surface Reactions of Soil Particles, Wiley, Chichester, 1998, pp. 281–324.[10] H.-R. Schulten, P. Leinweber, J. Anal. Appl. Pyrolysis 38 (1996) 1.[11] M. Schnitzer, Soil Sci. 151 (1991) 41.[12] R.F.C. Mantoura, J.P. Riley, Anal. Chim. Acta 76 (1975) 97.[13] G.R. Aiken, in: G.R. Aiken, D.M. McKnight, R.L. Wershaw, P. MacCarthy (Eds.), Humic

Substances in Soil, Sediment and Water, Wiley, New York, 1985, pp. 363–385.[14] G. Abbt-Braun, F.H. Frimmel, P. Lipp, Z. Wasser-Abwasser-Forsch. 24 (1991) 285.[15] H.-R. Schulten, M. Schnitzer, Soil Sci. 153 (1992) 205.[16] H.-R. Schulten, J. Anal. Appl. Pyrolysis 12 (1987) 149.[17] M. Schnitzer, H.-R. Schulten, Adv. Agron. 55 (1995) 167.[18] H.-R. Schulten, Fresenius J. Anal. Chem. 351 (1995) 62.[19] H.-R. Schulten, B. Plage, M. Schnitzer, Naturwissenschaften 78 (1991) 311.[20] H.-R. Schulten, M. Schnitzer, Naturwissenschaften 80 (1993) 29.[21] H.-R. Schulten, M. Schnitzer, Naturwissenschaften 82 (1995) 487.[22] H.-R. Schulten, M. Schnitzer, Soil Sci. 162 (1997) 115.[23] P. Leinweber, H.-R. Schulten, J. Anal. Appl. Pyrolysis 47 (1998) 165.[24] HyperChem™, Hypercube, Inc., 1115 N.W. 4th Street, Gainesville, FL 32 601 USA.[25] H.-R. Schulten, M. Schnitzer, Biol. Fertil. Soils 26 (1998) 1.[26] Y. Huang, G. Eglinton, E.R.E. der Hage, J.J. van Boon, R. Bol, P. Ineson, Eur. J. Soil Sci. 49

(1998) 1.[27] H.-R. Schulten, G. Gleixner, Wat. Res., in press (1998).[28] M. Schnitzer, Humic substances: chemistry and reactions, in: M. Schnitzer, S.U. Khan (Eds.), Soil

Organic Matter, Elsevier, Amsterdam, 1978, pp. 1–64.[29] H.-R. Schulten, Int. J. Environ. Anal. Chem. 64 (1996) 147.[30] C. Sorge, H.-R. Schulten, R.G. Weyandt, N. Kamp, M. Brechtel, Int. J. Environ. Anal. Chem. 57

(1994) 1.[31] P. Leinweber, O. Blumenstein, H.-R. Schulten, Eur. J. Soil Sci. 47 (1996) 71.[32] H.-R. Schulten, Humic and Fulvic Acids; Isolation, Structure and Environmental Role, J.S.

Gaffney, N.A. Marley, S. Clark (Eds.), Am. Chem. Soc. Symp. Series 651, Washington, DC, 1996,pp. 42-56.

[33] H.-R. Schulten, J. Anal. Appl. Pyrolysis 32 (1995) 111.

.