ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2020

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1948

Characterisation of naturaldissolved organic matter withliquid chromatography and highresolution mass spectrometry

CLAUDIA PATRIARCA

ISSN 1651-6214ISBN 978-91-513-0970-5urn:nbn:se:uu:diva-410266

Dissertation presented at Uppsala University to be publicly examined in B21, BMC,Husargatan 3, Uppsala, Friday, 4 September 2020 at 09:15 for the degree of Doctor ofPhilosophy. The examination will be conducted in English. Faculty examiner: Dr NorbertHertkorn (Helmholtz Zentrum Munich, German Research Center for Environmental Health,Research Unit Analytical Biogeochemistry).

AbstractPatriarca, C. 2020. Characterisation of natural dissolved organic matter with liquidchromatography and high resolution mass spectrometry. Digital Comprehensive Summariesof Uppsala Dissertations from the Faculty of Science and Technology 1948. 67 pp. Uppsala:Acta Universitatis Upsaliensis. ISBN 978-91-513-0970-5.

Dissolved organic matter (DOM) is one of the most heterogeneous and complex mixtureon Earth. DOM plays a crucial role in biogeochemical processes on the global scale andit is essential to sustain and regulate the biological processes in aquatic ecosystems. DOMoriginates from a multitude of biological, physical and chemical transformations, leading to itsphenomenal chemical diversity. In order to understand and predict its effect on the global carboncycle, an intimate characterization at molecular level is necessary. The investigation of theextraordinary complexity of the DOM mixture represents a compelling challenge for analyticalchemistry. The focus of this thesis was the development of methods for the characterizationof DOM in natural waters. High resolution mass spectrometry (HRMS), was combined withhigh pressure liquid chromatography (HPLC) and electrospray ionization (ESI), to investigatethe chemical diversity of DOM. The first study demonstrated that cutting-edge techniques(such as the Fourier-transform ion cyclotron resonance mass spectrometer - FTICR-MS), arenot indispensable to disclose essential information on the DOM molecular composition, infact the Orbitrap mass analyser is a suitable alternative for the analysis of complex naturalmixtures. In the second study, the potential benefits offered by the online coupling of HPLCand HRMS instruments were explored, revealing significant advantages in terms of analysistime, achievable information and versatility of the method. The advantages of online separationwere further confirmed in the third study, focused on the characterization of autochthonouslabile DOM. Chromatographically resolved profiles emerged from the bulk-DOM, allowing themonitoring of labile autochthonous components in presence of heterotrophic bacteria. Despitethe advantages achieved by the application of online separation, a strong limiting factor in DOMcharacterization is the ESI source, suitable only for the analysis of the DOM fraction susceptibleto ionization. In the last study, the extent of the DOM material prone to ionization was estimated,revealing the occurrence of an extensive portion of the material resistant to routinely employedESI approach. The full characterization of DOM is still an open challenge and the combinationof multiple techniques is fundamental to unravel is extreme intricacy.

Keywords: aquatic carbon cycle, dissolved organic matter, DOM, high resolution massspectrometry, online liquid chromatography, organic matter characterization

Claudia Patriarca, Department of Chemistry - BMC, Analytical Chemistry, Box 599, UppsalaUniversity, SE-75124 Uppsala, Sweden.

© Claudia Patriarca 2020

ISSN 1651-6214ISBN 978-91-513-0970-5urn:nbn:se:uu:diva-410266 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-410266)

To my family

“I learned very early the difference between knowing

the name of something and knowing something”

Richard P. Feynman

DOM...

“Capital Letters Were Always The Best

Way Of Dealing With Things You Didn’t

Have A Good Answer To”

Douglas Adams

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Hawkes, J.A., Dittmar, T., Patriarca, C., Tranvik, L., Bergquist, J. (2016) Evaluation of the Orbitrap Mass Spectrom-eter for the Molecular Fingerprinting Analysis of Natural Dis-solved Organic Matter. Analytical Chemistry, 88 (15), 7698-7704

II Patriarca, C., Bergquist, J., Sjöberg, P.J.R., Tranvik, L., Hawkes, J.A., (2018) Online HPLC-ESI-HRMS Method for the Analysis and Comparison of Different Dissolved Organic Mat-ter Samples. Environmental Science & Technology, 2018, 52, 2091-2099

III Patriarca, C., Sedano, V., Garcia, S.L., Bergquist, J., Bertils-son, S., Sjöberg, P.J.R., Tranvik, L., Hawkes, J.A., Character and environmental lability of ionizable cyanobacteria-derived dissolved organic matter (Submitted)

IV Patriarca, C., Balderrama, A., Može, M., Sjöberg, P.J.R., Bergquist, J., Tranvik, L., Hawkes, J.A., Investigating the ioni-sation of dissolved organic matter by electrospray ionisation (Manuscript)

Reprints were made possible with permission from the respective publishers.

Contribution report

The author wishes to clarify her contribution to the research presented in Papers I-IV in this thesis.

I Took part in performing the experiments and data analysis. Par-took in the writing of the paper.

II Planned the research, performed the experiments and the data analysis. Wrote the paper with contributions from all co-authors.

III Took part in planning the experiment. Main responsible for data analysis and data evaluation. Wrote the paper with contribution from all co-authors.

IV Took part in planning the experiment and performed the analysis. Responsible for data evaluation. Wrote the paper with contribu-tion from all co-authors.

The author also wishes to note that parts of this thesis are based on her licen-tiate thesis from 2018. The content has been updated, expanded with current results and adapted to the form of a doctoral thesis.

Patriarca, C., Characterization of dissolved organic matter: An analytical challenge, Fil. Lic. thesis, Acta Universitatis Upsaliensis, 2018

Contents

Introduction ................................................................................................... 11 Dissolved organic matter in natural waters .............................................. 11 DOM: an analytical challenge .................................................................. 12

Aims .............................................................................................................. 15

DOM analysis: instrumentation and methods ............................................... 16 Sample preparation ................................................................................... 16 Chromatography ....................................................................................... 17 Mass spectrometry .................................................................................... 20 Electrospray ionization ............................................................................. 23 Complementary detector .......................................................................... 27

Charged aerosol detector ..................................................................... 27

Data analysis and visualization ..................................................................... 30 Detection limit .......................................................................................... 30 Spectral recalibration ................................................................................ 32 Formula assignment ................................................................................. 33 Chromatographic feature discrimination .................................................. 35 Visualization and statistics ....................................................................... 36

Results and discussion .................................................................................. 41 Is the Orbitrap suitable for DOM analysis? (Paper I) ............................... 41 Online chromatography and reduction of DOM isomeric complexity (Paper II) .................................................................................................. 43 Fate of labile DOM (Paper III) ................................................................. 45 Hide-and-seek: ESI-MS ‘invisible’ DOM (Paper IV) .............................. 47

Conclusions and future aspects ..................................................................... 50

Populärvetenskaplig sammanfattning ........................................................... 53

Acknowledgments......................................................................................... 55

References ..................................................................................................... 57

Abbreviations

CAD DI DBE DOM ESI FWHM FT FTICR H/C HPLC HRMS MS m/z O/C PcoA RP SEC SPE SRFA

Charged aerosol detector Direct infusion Double bond equivalence Dissolved organic matter Electrospray ionization Full width at half maximum Fourier transform Fourier transform ion cyclotron resonance Hydrogen to carbon ratio High pressure liquid chromatography High resolution mass spectrometry Mass spectrometry mass-to-charge ratio Oxygen to carbon ratio Principal coordinate analysis Reversed phase (chromatography) Size exclusion chromatography Solid phase extraction Suwannee river fulvic acid

11

Introduction

Dissolved organic matter in natural waters An intricate flow of carbon sustains life on Earth. The complexity and the heterogeneity of the carbon based entities in the environment can only be postulated, and the number of possible conceivable chemical structures ex-ceeds the number of atoms in the observable universe [1,2]. Dissolved or-ganic matter (DOM) in natural waters represents a fraction of this phenome-nal heterogeneity of molecules and constitutes one of the largest carbon res-ervoirs on the planet, similar in abundance to the carbon dioxide in the at-mosphere [3].

DOM generally constitutes the majority of the living and non-living or-ganic material in the water column of aquatic systems, and it is generally defined as the material permeating a filter having mesh size ranging from 0.2 and 0.7 µm. DOM plays an essential role in the biogeochemical processes at global level [4–6]: as nutrient source sustaining the microbial loop in the aquatic systems [7–10], acting as carbon-sink [11–13], operating as photo-regulator in the water column [14,15], providing binding sites for transport and bioavailability of metals [16,17]. DOM plays a fundamental role in the regulation of biogeochemical processes, its composition impacts the transport of pollutants in the environment [18–20] and the drinking water quality [21–23], affecting not only the organoleptic characteristics of the water, but also generating disinfection by-products (DBPs) during chlorina-tion processes [24–28]. This list illustrates the diversity of mechanisms in-volving DOM and it is far from being exhaustive. The ubiquity of DOM and its multiple roles imply that any variation in its size, proportion and dynam-ics could have a major impact at the global scale, including carbon storage and exchange of carbon dioxide with the atmosphere [13,29].

The intricacy of mechanisms generating the vast chemical diversity of DOM has not yet being elucidated. Its heterogeneity derives from a combi-nation of biotic and abiotic processes, regulated by thermodynamics and kinetics, generating a reactivity continuum between inorganic and organic molecules [30]. The complexity of the organic material and the currently limited knowledge on the chemical diversity of the mixture prevent a clear classification of the DOM. Researchers often label DOM according to its origin and/or turnover. DOM can be defined ‘autochthonous’, when original-ly produced in situ by primary producers organisms, or ‘allochthonous’

12

when produced elsewhere by decay of biomass, leachate and transport to aquatic ecosystems [31]. In different settings, the material deriving from one of these two sources (internal or external to the aquatic system) can prevail, as observed in marine and inland waters. Autochthonous material dominates marine-DOM, due to the prominence of primary producers compared to external input of organic material; similarly, allochthonous contributions are more prominent in inland waters, more exposed to terrestrial DOM [3]. A further classification of DOM defines the persistence of the material in the water column. Labile organic material identifies the most bioavailable frac-tion of DOM, supporting the energy demand of heterotrophic organisms, and it is rapidly consumed and respired. The DOM fraction resistant to fast deg-radation is defined as recalcitrant material and its mineralization requires longer time scales. Recalcitrant DOM can be further sub-classified as semi-labile, refractory or ultra-refractory in relation to its persistence in the aquat-ic system, which can vary from months to millennia [11,12].

Despite its ubiquity and ecological importance, DOM is still poorly char-acterized, mainly due to its enormous complexity [30]. In order to improve our understanding of the global carbon cycle and unveil the role of DOM in the biogeochemical processes, and thus predict the fate of this organic mate-rial in a changing environment (increasing global temperatures, anthropo-genic impact, etc.), it is essential to investigate and intimately understand the chemodiversity of this immensely complex mixture.

DOM: an analytical challenge DOM is operationally defined as the organic material fraction that passes through a 0.2 – 0.7 µm filter, with a concentration ranging from 0.1 to 30 mg C L-1 [12,32,33], up to exceptionally high values exceeding 300 mg C L-1 [34]. Due to its various origins and transformations, DOM consists of a mul-titude of organic compounds, from carbohydrates to proteins, from lignin to humic substances (humic and fulvic acids) [6]. This yields an incredibly complex mixture characterized by tens of thousands of unique molecular formulas, with broad molecular range, large diversity of functional groups, chemical properties and extremely high abundance of isomeric species (structural isomers) [11,35–38]. Due to these aspects DOM characterization represents a phenomenal analytical challenge.

To date, a massive effort has been made to characterize this gigantic pool of carbon, by applying numerous analytical techniques. DOM bulk proper-ties (such as aromatic content [39], molecular weight [40], relative abun-dance of humic against protein like fractions,[41] etc.) have been disclosed by optical characterization techniques, mainly absorbance and fluorescence spectroscopy. However, their application is limited to the presence of light-absorbing species.

13

Several filtration, extraction and desalting procedures have been applied and are under constant development in order to isolate, pre-concentrate, and re-move metals and abundant inorganic salts from the organic material [42–45]. Similarly, a wide range of chromatographic techniques have been and cur-rently are employed in order to fractionate and separate the extracted DOM into several classes of compounds, based on size, charge, volatility, polarity (ref. [46] and references therein). Despite the simplification of the organic mixture through chromatography, the extreme DOM complexity strongly limits the utility of the currently available separation techniques.

A better understanding at the molecular level of this “organic continuum” was possible only with the application of nuclear magnetic resonance (NMR) [30,47,48] and high resolution mass spectrometry (HRMS) [49–52].

HRMS represents the cutting edge of technique for the analysis of un-known complex mixtures [53–55]. HRMS techniques are able to discrimi-nate between species characterized by subtle differences in their elemental composition, allowing the identification of thousands of molecular formulas in the organic mixture [4,56,57]. Nowadays, the Fourier-transform ion cyclo-tron resonance (FTICR) mass spectrometry is the state-of-art technology for the DOM analysis [51,55,58–60], however its availability is strongly limited by high purchase and maintenance costs.

In recent years, the introduction of Orbitrap technology [61] enabled low-er cost and wider availability, allowing a larger group of researchers to per-form high resolution analysis [62–64].

There are few studies that have compared the Orbitrap performances to the cutting-edge FTICR-MS with regard to non-targeted DOM analysis [65,66]. To the author’s knowledge, the study presented in Paper I was the first to evaluate the suitability of this mass analyzer for the analysis of com-plex environmental mixtures, showing its limitation and advantages. We concluded that the Orbitrap represents a valid alternative to the FTICR-MS, effective for the investigation of relevant biogeochemical changes.

Multivariate statistical analysis and advanced data visualization are essen-tial tools for elaboration and interpretation of the enormous data set deriving from thousands of molecular formulas in the DOM bulk [67,68]. However, neither the most advanced HRMS nor the most innovative algo-rithm/statistical tools are able to resolve structural isomers in absence of a separation technique.

The ability to differentiate between different species characterized by the same molecular formula is the key to disclose DOM diversity and composi-tion. Liquid chromatography (LC) is a powerful tool in DOM analysis, demonstrating that the majority of DOM complexity is hidden behind iso-meric averaging [50,69]. Major advances in chemical characterization of DOM were made possible by the combination of chromatography with HRMS [52,70,71]. These methods employ chromatography in offline mode, thus introducing time-consuming fractionation-collection steps and further

14

sample manipulation, potentially introducing contamination, prior to mass spectrometric analysis. Paper II is among the first studies exploring the capabilities of the direct coupling of LC and HRMS for the characterization of DOM [71–74]. Online methods allow fast, easy and detailed comparison of DOM samples from a variety of sources with a single chromatographic run. Data evaluation based on in silico fractionation provides an advanta-geous flexibility for the data exploration. Even though isomeric averaging cannot be fully resolved by this technique [36], in Paper II it was demon-strated that a large amount of additional information can be achieved not greatly exceeding the time of a conventional direct infusion (DI) analysis. The application of online chromatography was revealed to be essential in the analysis of abundant autochthonous DOM as explored in Paper III, where chromatographically resolved peaks, emerging from the broad complexity of the bulk-material, could be observed and their selective removal monitored. Additionally, online analysis was indispensable for the discrimination be-tween the material susceptible and resistant to negative electrospray ioniza-tion (the most used ionization mechanism in DOM analysis), as investigated in Paper IV, where a remarkable underestimation of the DOM molecular profile emerged.

15

Aims

Characterization of dissolved organic matter represents a compelling chal-lenge for analytical chemistry. The work presented in this thesis adds a new chapter to the analysis of DOM. The Aim of this thesis was to show:

• The necessity to evaluate the suitability of the Orbitrap mass analys-er as alternative to the less accessible FTICRMS instrumentation for the analysis of DOM (Paper I).

• The reduction of isomeric complexity with consequential increment in the achievable information from a single analysis through coupled HPLC-HRMS (Paper II).

• That based on the promising results from the application of online chromatography to high resolution mass spectrometry, the applica-tion of this method could be further investigated through:

o The analysis of the ionisable labile DOM produced in axenic and

natural conditions by a common cyanobacterium (Paper III).

o The evaluation of the amount of ‘MS invisible’ DOM affecting the correct comparison of terrigenous and marine DOM (Paper IV).

16

DOM analysis: instrumentation and methods

Sample preparation In chemical analysis, a series of sample preparation procedures are generally employed to isolate the compounds of interest form the matrix. In DOM analysis the material of interest is part of the matrix, so the logical conclu-sion would be that no sample preparation is necessary. This statement is debatable, and depends on the type of method selected for the analysis of the DOM. When optical measurements aiming to study the bulk properties of the mixture are employed, sample pre-treatment processes can be avoided [75,76]; contrarily, techniques for the investigation of structural and molecu-lar properties of the DOM necessitate a rigid sample manipulation in order to pre-concentrate the sample and remove inorganic salts that are detrimental for the analysis [42,44]. The main techniques employed in DOM extraction are: reverse osmosis/electrodialysis (RO/ED), ultrafiltration, freeze drying and solid phase extraction (SPE). All these approaches can be discriminated on the base of their advantages and limitations, nevertheless they share the same inability to isolate and recover the totality of the DOM pool [77].

Among the previously mentioned techniques, SPE is one of the most rou-tinely employed sample pretreatment methods in DOM analyses, due to its simplicity, accessibility, versatility and robustness. Its mechanism is based on the retention of molecules on a substrate (a large diversity of sorbents exists), followed by elution with a suitable solvent. A sorbent largely em-ployed in DOM studies and selected in this thesis (Papers I, II and IV) is a non-polar stationary phase based on modified styrene divinylbenzene (PPL) [42,78,79]. Its retention mechanism requires the acidification of the sample (pH 2), necessary to neutralize the acids abundant in DOM mixture [48] and guarantee the retention on the substrate. SPE provides highly reproducible results for the extraction of the retained DOM [42,80,81]; PPL cartridges commonly retain 40 – 60% of aquatic DOM, thus limiting the characteriza-tion of the overall complex mixture. However, highly representative results can be achieved by the accurate selection of substrates and extractions condi-tions [82,83]. Studies have assessed that SPE does not substantially affect the overall information achieved by MS analysis [80,83]; this observation was confirmed in Paper IV, however it was also suggested that a large part of the retained material is not visible in routine MS analysis (negative ESI-HRMS) and that seems to affect terrestrial-DOM more than marine-DOM.

17

Chromatography Chromatographic separation has been largely employed in the analysis of DOM [46]. A wide range of chromatographic approaches has been used to investigate the chemical and physical properties of the complex mixture (such as polarity, charge, molecular weight). The majority of the studies using chromatographic methods combined separation to detectors able to describe bulk properties of the organic material, for example spectroscopic techniques (absorbance, fluorescence) and organic carbon detector (OCD) [70,84,85]. In comparison, fewer studies focused on the application of chro-matographic separation to techniques able to disclose the molecular infor-mation of DOM mixture, such as HRMS and NMR [71,86].

In mass spectrometric analysis of complex mixture, the most popular ap-proach is the direct infusion (DI), one of the easiest methods of sample in-troduction, where a liquid sample is continuously delivered into the ion source of the mass spectrometer using a syringe pump (Paper I). This ap-proach still dominates DOM analysis, in particular in connection to the FTICR-MS instrument [4,51,58,59]. This is mainly due to its reduced analy-sis time and instrumental simplicity. Despite these advantages, there are important drawbacks such as strong ion suppression, since many species populate the ion source at the same time, thus strongly competing for the charge. Other critical weaknesses are represented by the inability to distin-guish between isomers and the extensive spectral overlap achieved by this approach. Despite these disadvantages, a great deal of information about the bulk properties of DOM has been unveiled and fingerprints from different sources have been obtained [87–90].

DI approach requires (almost) always [90,91] the use of SPE or other ex-traction techniques able to pre-concentrate and desalt the sample. As previ-ously described, SPE allows the isolation of representative DOM extracts; however, harsh sample manipulation (such as acidification) might alter the original DOM composition and potentially crucial information associated to the un-retained material is lost. The use of complementary retention mecha-nisms can greatly enhance the analysis of complex mixtures, by adding an-other dimension to the mass spectrometric analysis. In particular, the appli-cation of liquid chromatography (LC) proved to be highly beneficial for a better understanding of DOM composition [69,70,86,92]. Many studies fo-cusing on the characterization of DOM by high resolution MS use chroma-tography in conjunction to the DI approach [69,70,86,93,94]; in these studies the separation system is used exclusively to fractionate the material, which is then analysed in DI mode.

This approach, defined as off-line chromatography, is advantageous since it allows retention of physical and chemical information of the mixture (such as polarity, charge, molecular weight), it also reduces the complexity of the sample, allowing a more detailed characterization of the DOM while main-

18

taining high signal to noise ratio as allowed by direct infusion. This approach is also characterized by several disadvantages, including the necessity of large amount of material (multiple fractionation), long analysis time, broad sample manipulation and limited method flexibility. The online coupling of an LC system to a HRMS is therefore advantageous compared to the off-line mode. However, in recent years only few studies involving the combination of chromatography online with HRMS emerged (Paper II and [71,72,74]). The limited availability of online techniques is plausibly due to the use of FTICR-MS instrument as state-of-art technology for DOM analysis [65,69,95–97]. The development of a method coupling a chromatographic system to the FTICR-MS is challenging as the continuous change in ion population from the column effluent produces a variable ion density in the FTICR cell resulting in a space charge effect affecting the mass accuracy of the spectra and requiring a complex data recalibration [98]. This effect is reduced in Orbitrap instruments due to the presence of the automatic gain control (AGC) that regulates the number of ions entering the mass analyser. Presumably, additional limiting factors to the development of LC-HRMS methods for FTICR-MS are the high costs and limited instrumental availa-bility. An additional reason might be the inability of chromatography to re-solve the broad isomeric complexity of DOM, which emerges as broad unre-solved ‘hump’ covering a large elution time (Paper II and [36,69]). General-ly, separation techniques are employed to reduce the sample complexity to observe singly eluting species; in DOM analysis this scenario is implausible as the mixture diversity is unresolvable by current analytical abilities, there-fore a change of perspective is necessary. Unrealistic goals of complete sep-aration should be postponed until a new technique able to resolve the intrin-sic DOM complexity arises, until then advantages offered by online chroma-tography should be employed to gain more insight on the DOM mixture.

Information about DOM complexity is accessible to every type of chro-matographic separation; in this thesis, two main chromatographic modes were selected: (i) reverse phase chromatography, based on hydrophobic in-teractions is excellent for the retention of neutralized carboxylic acids (Pa-per II-IV) and (ii) size-exclusion chromatography, which can discriminate material on the basis of hydrodynamic molecular size (Paper IV).

In reversed phase chromatography, hydrophilic material (oxygen rich) elutes first, while the more hydrophobic (low oxygen content) has a stronger interaction with the stationary phase, thus a longer retention time (Figure 1). Differential retention leads to decreased complexity in each mass spectrum. This means that the amount of species entering the ion source at the same time is decreased, leading to lower ion suppression and a better sensitivity (Figure 1).

19

Figure 1. (A) overlapped total ion current (TIC) from DI (black) and LC (grey) analysis of a lake sample; (B) comparison of the m/z distribution and intensities between DI (black) and LC (early eluting segment in blue, late eluting segment in orange) in a 0.25 Da window; oxygen rich species decrease along the chromato-graphic gradient.

Reversed phase chromatography further enhances the ionization efficacy by removing inorganic interferences and contaminants (e.g. plasticizers) from the organic mixture, which elute as sharp peaks only interfering in few mass spectra. This approach offers also the possibility to avoid severe sample ma-nipulation, for example salts can be removed with the dead volume of col-umn as they have no interactions with the apolar stationary phase. Similarly, a limited sample manipulation is necessary to pre-concentrate the mixture, the solvent is removed by freeze drying or simple solvent evaporation (as in Paper III and IV) followed by reconstitution in a mobile phase suitable for the chromatographic analysis. This way the majority of the sample is pre-served; however, minor sample alterations in the solvent removal step, or poor re-dissolution in the recovery solvent might occur.

Another advantage in the application of an online approach is the ability to separate isomers, as demonstrated by the spiked addition of three model compounds in a DOM sample (Figure 2). This type of isomeric isolation is not possible with an offline LC approach followed by DI-HRMS. Clearly, this scenario is unrealistic for bulk DOM, where due to the isomeric com-plexity of the mixture a single formula appears as an unresolved ‘hump’ and the natural abundance of individual isomers is too low to emerge as resolved chromatographic peak. This ability can be exploited in analysis of less com-plex and processed material (Paper III).

Online chromatography retains all the advantages of off-line separation, except transient averaging in DI and additionally provides a fast, versatile and broadly applicable approach for DOM analysis, and represents an im-portant advancement for the fast and easy monitoring of possible biomarkers in environmental metabolomics research.

20

Figure 2. Extracted ion current (XIC) of mass 369.08 related to the model com-pounds spiked into SRFA sample (10 ppb in 300 ppm C). Reprinted with permission from Paper II. Copyright (2018) American Chemical Society.

Mass spectrometry In mass spectrometry ionized chemical species are generated and then dis-criminated on the base of their mass-to-charge ratio (m/z) [99].

The basic components of any mass spectrometer are: ion source, mass an-alyser and detector. The function of the ion source is to convert molecules to gas phase ions, which are then transferred to the mass analyser. This is the core of the instrument, which generally employs a magnetic and/or electric field for the separation of the charged species based on their m/z. Finally, the detector generates a signal corresponding to the electrical current of each ion. With exception of some versions of the first component, a mass spec-trometer operates in high vacuum conditions in order to prevent unwanted collisions between neutral gas molecules and the ions.

Mass spectrometry represents one of the most powerful analytical tech-niques in virtue of its universality. In fact, it can provide information on any ionisable compound and therefore represents one of the best choices for the analysis of a complex mixture and other unknown compounds. Several types of mass spectrometers, characterized by different performances, are current-ly available. They can be classified by the ability to distinguish between two similar m/z values. The smaller the difference in m/z an instrument can dis-criminate, the better are the performances. Indexes of such ability are resolu-tion and resolving power. Resolution (R) is defined (IUPAC) as the minimum difference (∆m/z) that can be separated for a given m/z value (R = m/z/∆m/z = m/∆m). For an iso-

21

lated peak, ∆m commonly corresponds to the full width of the peak at half maximum (FWHM). The ability to distinguish between ions differing slight-ly in m/z is defined resolving power [99].

DOM consists of a multitude of organic compounds covering an extreme-ly broad mass range, where a multitude of isobaric compounds and structural isomers coexist. In this complex mixture, organic molecules characterized by the same nominal mass (isobars) but different chemical formula, can be dis-criminated by their mass defect (difference between exact and nominal mass). An example is represented by the ability to discriminate between CH4 and O, which exhibit the same nominal mass (16 Da) but different exact mass (∆m= 36 mDa). In order to achieve such separation, and resolve thou-sands of different formulas from this organic mixture, instruments character-ized by high resolution and resolving power are necessary.

Fourier-transform ion cyclotron resonance (FTICR) and Orbitrap mass ana-lysers used in this thesis, comply with this requirement [57,61,100,101]. They currently represent the most powerful technology in terms of high resolution (R ≥ 100.000 at m/z 400), high mass accuracy (difference between the theoret-ical and the measured m/z; currently < 1 ppm) and sensitivity (sub femtomol).

These two mass spectrometers share several similarities. Both analysers trap the charged species in ultrahigh vacuum in order to guarantee a long mean free path, all ions are detected simultaneously based on their image current recorded in the time domain, which is then processed by Fourier transform and converted to the frequency domain to generate the conventional mass spec-trum. In both instruments the resolution is proportional to the strength of the applied field (magnetic in the FTICR, electric in the Orbitrap), both devices are pulsed detector, which means that charged species are externally accumu-lated (multipole ion trap in FTICR, C-trap in Orbitrap) during the detection of ions from the previous accumulation period [102]. Despite the similarities FTICR instruments are still the most advanced mass analysers in terms of resolution (R > 1.000.000 at m/z 400) and mass accuracy (< 0.1 ppm) able to detect very subtle variation in the elemental composition [55]. The main limi-tations to the Orbitrap consist in collisions with residual gas as well as imper-fections of the electrode manufacturing, limiting the time a signal can be de-tected and therefore the resolving capacity [102,103].

Discrimination between isobaric compounds is one of the main challenges in DOM analysis; hence by exclusively considering resolution and resolving power, FTICR-MS is currently the most advanced technique for the ele-mental analysis of complex mixtures. However, a question arises: is ultra-high resolution always necessary?

If all chemically reasonable elemental combinations are considered, not even the state-of-art 21T FTICR-MS instrument [55] would be able to re-solve such phenomenal complexity [30,37,104]. The selection of the most suitable technique depends on the research question and considerations or previous knowledge on the sample, and both are essential to make a choice

22

of technique. Generally, biogeochemical studies rely on the use of FTICR-MS for the analysis of complex mixture [37,51,70,90,105–107]. However in Paper I it was demonstrated that ultra-high resolution is not necessary for the analysis of ionizable DOM in typical marine and lacustrine mixtures, and that Orbitrap resolution is suitable for the discrimination of the fundamental mass splits in samples dominated by CHO species. Obviously, in case more subtle discrimination is necessary, as in the investigation of phosphorous compounds or the analysis on species containing hetero-atomic combina-tions beyond the resolution provided by the Orbitrap, the application of ul-tra-high resolution instruments is unavoidable (Figure 3).

Figure 3. A 10 mDa window centred in 369.083 m/z is shown. 15T FTICR profile (black); Orbitrap high-resolution (blue) and low-resolution (orange) profiles. Com-pound mixtures A, B and C (black lines).

Figure 3 shows an example of the ability of Orbitrap to compete with FTICR instrument in the analysis of complex mixture. In high resolution mode the Orbitrap is suitable for the analysis and discrimination of the CHO species (C16H18O10; A) from the sulphur containing compounds (C13H22O10S; B) and showing an excellent mass accuracy compared to FTICR performances (mass error in ppm was 0.13 and 0.04 respectively). The same results cannot be achieved by the application of the Orbitrap in low resolution mode, suita-ble for the discrimination of CHO species but unable to confidently resolve sulphur containing formulas [108,109]. Compound mixture C represents an example of a signal that cannot be efficiently resolved by the Orbitrap and a crucial consideration is that the resolving power of the instrument needs only to be as good as the mixture is diverse. The FTICR data shows that neither

23

compound mixture B or C are present above the detection limit in this sam-ple.

DOM is not only rich in isobaric compounds, structural isomers are also abundant in the complex mixture [35,37,110]. Isomers are characterized by the same formulas, but different chemical conformation and are impossible to discriminate by the use of even the most powerful high-resolution mass spectrometer as they have the same exact mass. The only techniques able to provide structural information in a complex isomeric mixture are nuclear magnetic resonance (NMR) and Fourier-transform infrared spectroscopy (FTIR) [47–49,111]; however, even these powerful approaches are unable to unravel the structural complexity of DOM.

Separation techniques (such as solid phase extraction or chromatography) offer a partial isomeric discrimination of the mixture (Papers II-IV), reveal-ing the extreme isomeric complexity of the DOM and allowing the retention of structural information while providing additional information on the mix-ture in relation to the selected substrate [69,70,86].

Electrospray ionization Electrospray ionization (ESI) is one of the most widely applied atmospheric pressure ionization techniques. It is known to be one of the “soft” ionization sources, since it transfers the ions from solution to gaseous phase inducing no or very little fragmentation. This is crucial in non-targeted analysis, as minimal chemical change is made to the compounds analysed. ESI is popu-lar in the analysis of large biomolecules [112], but it also widely applied for the study of small polar organic molecules and is the most used ion source in DOM analysis [59,88,109,113,114].

The ESI source generates gas phase ions from analytes in an aqueous phase, thus allowing the infusion of aqueous solution and the coupling of liquid chromatography to mass spectrometry (Figure 4).

The sample solution is introduced into the ion source from a capillary. A high potential difference (generally 3-4 kV) is applied between the end of the capillary and the mass spectrometer inlet. The generated electrostatic field induces a charge accumulation at the liquid front on the capillary tip, forming a cone-jet called Taylor cone [115]. Under the action of electrostatic repulsion forces, the surface tension of the liquid is overcome (coulomb ex-plosion) and highly charged droplets are released. During the migration to-ward the mass spectrometer inlet (counter electrode), the size of the droplets decreases due to solvent evaporation (supported by a nebulizing gas, usually N2) thus increasing the charge density. This leads to a further coulomb ex-plosion, where smaller surface charged droplets are formed. The process is repeated until the ions are released into the gas phase.

24

Figure 4. Schematic of the electrospray ionization process

ESI can generate either positive or negative ions, depending on the applied electrical field in the ion source. In positive ion mode the generated ions are mostly protonated species, [M+H]+, while in negative-ESI the deprotonated molecules are the most abundant, [M-H]-. Acidic or basic conditions can assist in the ionization of molecules in positive or negative mode, respective-ly. In ESI is also possible to observe a phenomenon named ‘wrong way around’ ionization, where the ion formation is enhanced in conditions that should inhibit the ionization [116,117]. For example, high ionization effi-ciency was observed in negative mode for acidic species in an acidic envi-ronment, where they were expected to be neutral. This phenomenon is ex-tremely favorable for the combination of the ESI source operating in nega-tive mode to the chromatographic separation of acidic compounds, where acidic additives are employed to promote the retention of the stationary phase. It is also possible to form adducts with other ions present in the liquid effluent, for example sodium, chloride, ammonia, formate, etc.

The ionization efficiency of the analytes in the ESI source defines the ability of the molecules to generate gas phase ions from the liquid phase, and it strongly depends on the chemical and physical properties of the analytes [118–120]. However, the ionization process does not involve only the ana-lytes of interest, in particular when liquid chromatography is applied. The ionization can be influenced by a multitude of factors mainly related to sol-vent composition, additives, buffers, and operational parameters [121]. Some of the above mentioned parameters have been evaluated for the analysis of DOM [87,109,114].

25

The first study involving ESI together with low resolution MS, applied the negative ion mode [122]. After this, the first FTICR-MS analysis of natu-ral organic material (at considerably higher resolution) was conducted with positive-ESI [87]. The latter is characterized by a strong disadvantage related to the formation of sodium adducts, [M+Na]+, which can cause critical prob-lems in the spectral interpretation of DOM samples, as demonstrated by Koch et al. [58]. The mass difference between sodium adduct and non-sodiated species is only 2.4 mDa (exchanging NaH for C2), hence creating not completely resolved peaks even at very high resolving power (> 200.000 at 401 m/z). However, only few studies applied exclusively positive-ESI for DOM analysis [48,88].

Due to the interferences in positive mode and the anionic nature of car-boxylic acid functionality on DOM at neutral pH, the negative-ESI is gener-ally the chosen ionization mode for analysis of this complex mixture (ref. [123] and references therein). The same ionization mode was selected in all presented papers (Paper I-IV).

Although many studies have used and currently apply ESI for DOM in-vestigation, several concerns have arisen regarding the ionization of this complex material such as fragmentation, multiple charging, multimer for-mation and ion suppression.

ESI mass spectra of humic substances are usually dominated by low mo-lecular weight ions (< 2000 m/z), which diverge from other findings that extend the humic substances mass range to about 10 kDa. This discrepancy is often referred to as ESI low-molecular weight bias. Some authors suggest that this disaccord is due to the presence of aggregates [124] that are “bro-ken” in the ESI source [88]. Others hypothesize the presence of large mac-romolecules that are actually fragmented in the ion source, or to the occur-rence of multiply charged species which account for the lower masses as the mass spectrometer measures mass to charge ratio [125]. These aspects have been tested by several studies. Stenson et al. [109] tested the incidence of fragmentation in humic substances, no evidences of dissociation in the ESI source emerged, however they suggest that fragmentation cannot be com-pletely excluded.

The dominance of singly charged species over multiple charged ions has also been demonstrated [109,126]. Tandem MS experiments, where precur-sor ions are fragmented to obtain structural information, showed that DOM ions preferentially lose fragments corresponding to the neutral losses ex-pected from singly charged ions, namely CO2 (44 m/z) and H2O (18 m/z), conversely doubly charged ions would lose 22 and 9 m/z, respectively [35,36,126].

Further evidences of the dominance of singly charged ions emerge from the spacing between isotopologues [109]. By observing the mass difference between a monoisotopic peak (12Cn) and its heavy isotope (13C12Cn-1), the charge state can be determined. The average mass difference observed is

26

1.0034 Da, which corresponds to a singly charged species. Lower mass dif-ferences identify multiply charged ions, as observed in Paper IV, where low intensity m/z values with mass defect in the range 0.5-0.6 Da were detected suggesting the presence of doubly charged material.

The occurrence of unrepresentative ions (such as multimers) has been in-vestigated as well [114]. The conclusions suggest that these species are pre-sent but do not interfere significantly with the final DOM representativeness, and that possibly, the presence of numerous ions in the source increases the chance of collision, removing some dimers before they enter the mass spec-trometer. However, the authors make clear that further investigation is nec-essary.

Other important aspects influencing the ionization efficiency of DOM mixture are ion suppression and preferential ionization. The first plays an important role in the analysis of DOM, in particular if salts are not removed from the original sample before introduction to the ESI source. In fact salts ‘hide’ or ‘steal’ the charge from other species. The ion suppression is not only induced by the presence of salts, but it is also related to the intrinsic complexity of the material and it is a general and unavoidable problem in ESI-MS of complex samples; in fact, the multitude of species present at the same time in the ion source creates a strong effect of preferential ionization hiding many signals, thus strongly influencing the ionization efficiency of the single molecules in the mixture. This effect is more prominent when a direct infusion approach is applied without any previous sample pre-treatment or chromatographic separation [90]. The mass spectra resulting from ESI-MS analysis show a distribution of m/z values (peak) and intensi-ties that can never be truly representative of the original sample: in fact, each peak intensity is a combination of multiple factors, including the analyte concentration, the ionization efficiency, the stability of the generated spe-cies, the transmission of the ions and their detection. The data analysis is therefore usually ‘qualitative’, examining broad statistical differences be-tween samples.

In the analysis of DOM a uniform ionization is not achievable with a sin-gle ion source [37,42,114,118,119,127]. The intrinsic complexity of DOM mixtures implies that only a fraction of the material is susceptible to ESI ionization [53,80], especially when a single ionization mode is selected, meaning that in routine DOM analysis (typically negative ESI-MS), an un-known fraction of the DOM pool is undetected and therefore uncharacterized (Paper IV).

Molecules characterized by acidic protons and able to stabilize the nega-tive charge [119] are more suitable to ionization in negative ESI mode and are generally characterized by high ionization efficiencies. As previously mentioned, the large abundance of carboxylic acids in the DOM mixture [48,128,129] favors the selection of negative ESI as ionization source. How-ever, DOM is not exclusively composed by acidic compounds; a large part

27

of the organic pool (such as lignin phenols, polysaccharides, proteins) [130,131] is resistant to the ESI ionization mechanism, thus characterized by low or null ionization efficiency and under-represented in HRMS studies. The complexity of the ESI mechanism together with the heterogeneity of DOM complicate the assessment of the fraction of DOM susceptible to ioni-zation [92,132], therefore a combination of techniques able to discriminate between the ESI-active and inactive components in the organic pool is nec-essary for a more comprehensive DOM characterization (Paper IV).

Complementary detector Charged aerosol detector The charged aerosol detector (CAD) is considered a universal detector for all non-volatile substances [133–135]. Its response is proportional to the amount of injected material and independent from the physicochemical properties of the analytes [134–139].

An overview of the CAD operating principles is described in Figure 5. The eluent from the chromatographic column is nebulised in the spray chamber by a flow of pressurized nitrogen. Large droplets are removed by an impactor integrated in the spray chamber; such precaution is necessary in order to prevent the saturation of the instrument due to the inability of large droplets to fully desolvate in the given path. The remaining aerosol flows through the evaporation tube, where solvent and volatile species are re-moved. Resulting residue particles reach the mixing chamber and collide with a second flow of nitrogen gas positively charged by a corona needle; collisions between the charged gas and the residue particles allow the gener-ation of ions by diffusional charge transfer. The excess of high mobility par-ticles (mainly gas ions) is removed by an ion trap placed after the mixing chamber to prevent fluctuations in the background current. The residue par-ticles are then transferred to a collector where the charge is measured by a highly-sensitive electrometer [133–136]. The size and the flux of the charged particles reaching the detector generate the CAD signals, which correlates to the amount of injected material and it is unaffected by the chemical proper-ties of individual analytes.

The instrumental response is non-linear over a broad mass range, as showed in Equation 1:

= Equation 1

where A identifies the signal, M is the amount of substance, a indicates the response intensity and b describes the slope of the curve. This behavior is the

28

consequence of a non-linear relationship between the particle area and the amount of material. However, a linear behavior (b=1) is observed over a limited mass range [134–136]. The uniformity of the CAD response is re-markable; as the detection is based on the amount of material and independ-ent from the physicochemical properties of the analytes, similar responses are generated for identical amounts of different analytes [134–138]; there-fore this technique can be employed for the quantification of multiple ana-lytes with a single calibration model [134,140,141].

Figure 5. Schematic of the charged aerosol process

CAD uniform response is subject to few limitations, mainly related to the mobile phase composition [134,135]. An increase in the organic modifier enhances the nebulization efficiency, thus increasing the amount of analyte reaching the detector and resulting in a higher signal [142]. To solve this issue two strategies have been introduced: the first consists in the application of a 3D calibration routine, where the variation in the organic modifier per-centage is included [140,141]; alternatively, the composition of the eluent from the column can be altered in order to provide the detector with a con-stant mobile phase composition. This second approach offers a more rapid and universal application as it simply requires the use of a second pumping system delivering an opposite gradient [143]. Another aspect affecting the

29

CAD response is the presence of additives, often necessary to guarantee the desired chromatographic performances. Additives might induce higher background noise or act as counterions for the eluting material, thus increas-ing the associated signal [134]. A solution to this drawback consists in the selection of small volatile additives such as ammonium formate and formic acid. These additives are also the most employed in ESI-MS analysis, thus favoring the development of a common chromatographic method suitable for both techniques. The mentioned strategies preserve the detector’s uniform response, enabling the detection of a large variety of species independently from their physicochemical properties, making the CAD an excellent analyt-ical tool suitable for different applications [134,135].

DOM chemical complexity cannot be disclosed by a single technique: spectroscopic analyses (such as UV and fluorescence) imply the presence of chromophores; nuclear magnetic resonance (NMR) is able to observe every organic molecules, but small changes to the proton environment create unre-solved features; mass spectrometric response is limited by the ionization source selected for the generation of gas phase ions and by the individual ionization efficiency of the analytes. A combination of approaches is neces-sary to gain an insight into the organic material’s heterogeneity. In this re-gard, the charged aerosol based detector can act as complementary tech-nique, allowing the detection (but not characterization) of all non-volatile material, independently from its ability to ionize. Obviously, due to its work-ing principle, the CAD is unable to provide any spectral information on the species eluting from the column [138], but it can detect the amount of mate-rial unnoticed by other techniques, or quantify those that are detected. In Paper IV, the CAD was used in association with ESI-MS operating in nega-tive mode for the analysis of different DOM samples. The MS signal, gener-ated by the DOM susceptible to negative electrospray ionization, was com-pared to the CAD response, proportional to the amount of material eluting from the column; the resulting relative response factor was employed to discriminate between the ESI-active and inactive organic material in the mixture. The ability to isolate and identify the fraction of material undetect-ed by mass spectrometry is fundamental to understand transport and trans-formation rate of the organic material in different environmental settings.

30

Data analysis and visualization

The mass spectrometric analysis of DOM generates an impressive amount of signals (m/z values), increasing proportionally with the selected instrumental resolution. The combination with chromatographic separation enormously inflates the amount of information within a single analysis, allowing a more subtle analysis of the material.

Data processing is essential for a correct visualization of the results. Han-dling of the gigantic pool of generated signals is not trivial. Data processing requires a multistep approach, where mass spectrometric data are converted and analysed by statistical software (R or Matlab), often with user-defined scripts. Basic operations include: data signal clean up, aiming to retain only significant information; spectral recalibration, to ensure correct mass accura-cy; and formula assignments, to access the elemental composition and mo-lecular distribution of the complex mixture.,

Detection limit Due to the complexity of the sample they are generated from, the mass spec-tra contain several thousand peaks. These signals are generated both from the organic material we are interested in and to some extent from electronic noise. The very first step of the data analysis is therefore noise removal. In order to do that a detection limit threshold is applied.

The detection limit is generally set as the signal-to-noise ratio (S/N) used to distinguish between an analyte and the noise. However, this is not always simple to calculate, in particular in non-targeted analysis. In fact, when at low abundance an analyte can be considered as noise (false negative) as well as a noise signal above the cutoff levels can be included in the data set (false positive). In view of this consideration, how can detection limit be fixed in order to be able to distinguish between real peaks and noise in complex mix-tures for which automated routines generate mass lists?

Riedel and Dittmar [38] addressed this problem and proposed a reliable solution where the noise is directly estimated from each sample. In our stud-ies, the detection limit was calculated using this approach. The detection limit (in units of intensity) was calculated by considering a high percentile (e.g. 95th percentile in Paper II) of peaks with a mass defect in a range

31

where organic formulas cannot be found (e.g. 0.4 and 0.8 Da; Figure 6, grey shaded area).

The mass defect associated to each m/z value is the value after the deci-mal of the calculated mass (for 369.082, the mass defect is 0.082). At this point, the intensity of the high percentile of the peaks which fall in this noise interval is calculated and set to be the detection limit for the sample. Crucial-ly, all peaks detected at intensities below this value are removed, at all mass defects (including 0 – 0.4 Da, where the organic peaks fall). By considering only the first 95th percentile of peaks within the 0.4 – 0.8 Da window, for example – rather than the 100th percentile – we exclude some signals (de-rived from contaminants), which would otherwise be incorrectly assigned as noise. These wrong assignments would increase the intensity cutoff level, thus causing the removal of genuine organic peaks.

An example of this is showed in Figure 6. The peaks with mass defect 0.5 Da (labelled by the orange arrows in Figure 6) were incorrectly assigned as noise by the detection limit routine.

Figure 6. Mass spectrum window of SRFA sample injected by DI. The grey shaded area identifies the 0.4 – 0.8 Da region. The orange arrows show an example of the signals responsible for the too conservative noise cutoff (identified by the orange line); the blue line displays the final intensity level cutoff applied to the data set.

These signals probably originate from doubly charged species. The corrected detection limit was lower (horizontal blue line, Figure 6) thus allowing many more signals to be considered. However, it can be noticed that a conservative cutoff level was maintained, meaning the removal of false positives as well as the exclusion of many actual peaks (false negatives). Overall, we consider

32

it worse to have too low a detection limit and include noise peaks in the data analysis than to exclude some genuine peaks with too conservative a detec-tion limit value. The identification of the issue in Figure 6 (observed in one of the direct infused sample in Paper II) was possible only by comparing the cutoff level of this sample to those from the rest of the data set, in particular by comparing this result with the corresponding chromatography cutoff level from the same sample. This example illustrates clearly how chromatography is a powerful tool to avoid similar issue in the non-targeted analysis of com-plex mixture.

The purpose of this routine is to separate genuine peaks from the electri-cal noise. Some false positives and negatives are inevitable in a large dataset and in each paper the data were carefully considered to tune the applied de-tection limit.

Spectral recalibration Prior to formula assignments, spectral recalibration is generally performed. This step is necessary since, even after instrumental external recalibration and despite the precision and reliability of the mass analyser, fluctuations affecting the mass accuracy are always possible. The overall principle of recalibration is therefore to adjust the spectra according to one or more known m/z values present in each scan. For the recalibration of directly in-fused (DI) DOM samples, in reason of the spectral averaging, a single recal-ibration is sufficient (Paper I). When online chromatography is applied, every scan corresponds to the specific composition of the column effluent in each instant, thus every mass spectrum requires individual recalibration making the whole process more challenging. Additionally, the different abundances in the ion population in each transient generate mass accuracy fluctuations due to the change in space charge effect [57,67] in the analyser’s cell; this issue is reduced in the Orbitrap compared to FTICR-MS instrument due to the presence of an automatic gain control (AGC) limiting the number of ions in the mass analyser’s cell [144].

Recalibration on chromatographic data can be achieved in 'real-time' by the use of lock masses, known m/z values either mixed with the column effluent before the mass spectrometric analysis (post-column infusion) or already pre-sent in the sample. The main disadvantage of the post-column infusion is the possibility of mutual exclusion: (i) either the detection of the lock masses is inhibited by the sample, thus jeopardizing the recalibration, or (ii) the lock masses are responsible for the ion suppression of the material eluting form the column, especially when the analytes are present in low concentration and/or characterized by limited susceptibility, as in the complex mixtures.

Alternatively, known m/z values already present in the sample can be se-lected as lock masses in case they are present in every scan. Due to the DOM

33

isomeric complexity, it is not uncommon to observe a single m/z value over a broad chromatographic range, therefore the use of lock masses in the mix-ture is a valid alternative to the use of external compounds (Paper II).

However, this strategy cannot always be applied since it requires a pre-screen of the sample (not always possible) and the same degree of isomeric complexity in each analysed sample. Paper III is a perfect example of this scenario: the lock masses approach was excluded due to the absence of m/z values common to every sample and the impracticality to apply a post-column infusion. In absence of lock masses complex calibration routines, involving the use of custom built algorithms [67,98] are required and not always available (as in Paper III and IV). Therefore, after the evaluation of the stability of the Orbitrap analyser found to be within the manufacturer specifications (approximately ± 3 ppm), the spectra recalibration was by-passed and a conservative approach was employed, by increasing the toler-ance for formula assignments to ± 5 ppm and introducing a routine for the isotope ratio confirmation (see 'Formula assignment' section).

Formula assignment The experimental m/z values were assigned to formulas with an in-house Matlab routine, where the matches to a target mass list were evaluated in order to reliably assign the molecular formulas. In Paper I this target mass list was generated from the FTICR-MS data by the Dittmar laboratory [38], as this was the standard to which we were comparing the Orbitrap data. In Papers II-IV, a theoretical m/z list was employed. This list was generated on the basis of several criteria such as the elemental composition (e.g. phospho-rous containing compounds were removed in Paper II-IV), the mass-to-charge window and the possible molecular combination. This last parameter was subordinated to an additional set of rules aiming to exclude formulas beyond the resolution capabilities of the instrument and absurd combinations from further consideration (Figures 7).

The m/z list experimentally obtained was compared to the theoretical list and the closest match, within a certain tolerance dictated by the instrumental mass accuracy, was assigned. In case of no matches, the experimental m/z value was no longer considered. This strategy based on a simple mass differ-ence cutoff (ppm tolerance window), was found to be suitable for the analy-sis of bulk-DOM, as the theoretical list is generated upon the restrictions imposed by the instrumental resolution (see restrictions in each paper). However, incorrect assignments were observed when formulas, not generat-ed in the theoretical list, were included. This was observed in Paper III, where a set of formulas (metabolites) characterized by heteroatom combina-tions beyond the Orbitrap resolution was included in the target list.

34

Figure 7. Mass window of 0.3 Da at m/z 401. Possible formulas in the theoretical list before (A) and after (B) the screening according to the selected criteria

In Paper III, the cyanobacterial-derived DOM was investigated; therefore, the presence of certain nucleotides, toxins and so on was expected. Upon critical exploration of the data following the Matlab routine, the m/z value expected to be associated to adenosine monophosphate (AMP) was incor-rectly assigned to another formula in the list, representing a closer match to the experimental m/z value in terms of mass difference. To solve this issue, the tolerance window allowed for formula assignment was set to 5 ppm, to include multiple candidates from the theoretical list, and have the opportuni-ty to assign the correct formula. In order to discriminate among the different candidates within this larger mass window, an isotope ratio confirmation was introduced and applied to every formula assignment, according to the fol-lowing strategy [145].

For each formula, the theoretical ratio (P) between the monoisotopic and 13C abundances was calculated based on Equation 2:

= 1.1/98.9 Equation 2

where n is the number of carbon atoms, 1.1 and 98.9 are the abundances of 13C and 12C respectively.

For each detected signal, the corresponding 13C peak was determined, and their intensity ratio (R) was compared to the theoretical value (P). The clos-est observed intensity ratio to P was selected to discriminate among multiple formulas. The ratio between the experimental (R) and theoretical (P) values was then employed to evaluate the assignment’s likelihood, and only formu-las characterized by a ratio between 0.8 and 1.2 were considered confidently assigned (Figure 8).

The isotope ratio confirmation allowed the correct assignment of the AMP signal. In fact, the theoretical value (P) of the two possible candidates was 0.111 for AMP (C10H14N5PO7) and 0.178 for the alternative formula

35

(C16H13NO8); the experimental m/z values showed an intensity ratio (R) equal to 0.108. The AMP theoretical value resulted to be easily the closest to the experimental R, thus allowing the confident assignment of the formula.

Figure 8. Intensities of isolated chromatographic peaks observed in a sample of cya-nobacteria-derived DOM (blue dots) distributed upon the individual isotope ratio confirmation value (R/P). Confident assignments were confirmed for R/P values between 0.8 and 1.2. The black arrow at R/P 1.6 is an example of probable wrong assignment.

Chromatographic feature discrimination Chromatographic profiles of DOM mixtures appears as broad unresolved humps (Papers II and IV), or ‘humpograms’ as defined by Brown et al. 2016 [69]. The extreme isomeric complexity of the mixture prevents a com-plete chromatographic resolution in a single analysis. This heterogeneity manifests similarly on a smaller scale: when a single m/z value is isolated as extracted ion current (EIC) the same mass-to-charge ratio can be observed over a broad retention time (Figure 9B), highlighting the presence of a multi-tude of possible isomers sharing the same elemental composition but charac-terized by a different chemical structure.

In Paper III, the isomeric diversity of the cyanobacterial-derived DOM was not comparable to the complex isomeric distribution of typical DOM samples; cyanobacterial-derived DOM largely coincided with chromato-graphically resolved peaks, while the bulk-material generated featureless profiles (Figure 9). Manual evaluation of the EIC of each m/z values as-signed in the dataset was unreasonable, therefore the chromatographic peaks were identified and discriminated from the featureless DOM profiles by us-

36

ing the ‘findpeaks’ algorithm in Matlab (Figure 9). This function is used to locate the local maxima of a signal and sort the peaks by height, width, or prominence. A careful tune of the algorithm settings is necessary for correct peak discrimination. Figure 9 shows the profiles of two m/z values in two different samples. The discrimination between the chromatographic peaks and the broad profile was only possible when stringent settings were selected (Figure 9, A3 and B3). The algorithm settings were tuned according to the chromatographic profiles observed for resolved model compounds eluting throughout the polarity range.

Figure 9. Application of the ‘findpeaks’ algorithm for the discrimination between chromatographic peaks and broad profile. The identified peaks are indicated by the blue triangles. (A1-B1) Raw data from an artificial culture and a natural water sam-ples; (A2-B2) Loose settings (discrimination impossible); (A3-B3) Correct tuning (discrimination possible).

Visualization and statistics The amount of data generated by high resolution mass spectrometry is phe-nomenal. Visualization tools and statistical approaches are therefore neces-sary to present and interpret the results. A very common tool for inter-comparison of environmental data via reduction to a single number is the Bray-Curtis dissimilarity test, which is based on the abundances of the same species present in different samples (Equation 3):

Bray-Curtis dissimilarity = ∑ , ,∑ , , Equation 3,

37

where the signal intensity (I) is compared between sample p and q for each molecular mass k (from k1 to kn).

The Bray-Curtis dissimilarity index for each pair of samples can vary be-tween 0 and 1, with 1 representing the highest dissimilarity (completely dif-ferent) and 0 the highest similarity (identical). Results are collected in the dissimilarity matrix and are often expressed as percentage by multiplying the values by 100 (%Bray-Curtis dissimilarity).

There are different strategies to visualize the dissimilarity matrix, one is the heatmap, a graphical representation where values from a matrix are rep-resented as colours. An example of heatmap is showed in Figure 10, where the colour scale fades from blue (maximum similarity) to red (maximum dissimilarity).

Figure 10. Heatmap derived from the dissimilarity matrix generated by samples from three different sources. Letters followed by numbers identify source and seg-ments (S for SRFA, N for marine and P for lake). Letters followed by LC or DI corresponds to the averaged chromatographic run and direct infusion respectively. The colour scale is proportional to the dissimilarity values.

Another approach is the principal coordinate analysis (PCoA), a type of mul-tidimensional scaling that represents inter-object dissimilarity in a low-dimensional space. In this space, similar objects (small dissimilarity value) are grouped closer together, contrary to dissimilar objects which are ordinat-ed further apart.

The PCoA tries to preserve the original dissimilarity values between the objects in the limited space defined by the axes. An example is shown in Figure 11 (from Paper I) where 90% of the dataset is explained by the first two eigenvectors.

38

Other important parameters for data interpretation can be plotted in PCoA space. In Figure 11 it is possible to distinguish two external variables (per-centage of lake water and instrumental differences). Their position in the multidimensional space was evaluated afterwards by calculating the Pear-son´s correlation coefficients (ρ) between the environmental parameter val-ues and the scores of each sample on the two PCoA axes.

The Pearson´s correlation coefficient (ρ) is a measure of the linear corre-lation between two variables X and Y (Equation 4).

, = , Equation 4,

where cov(X,Y) represents the covariance between two variables (X and Y) and σ is the standard deviation (of X and Y respectively). It varies between +1 and −1, where 1 is positive linear correlation, 0 is no linear correlation, and −1 is total negative linear correlation.

Figure 11. PCoA of mixed water samples analysed with two different mass spec-trometer (FTICR in red; Orbitrap high and low resolution in blue and orange respec-tively). Labels from 0 to 100 represent the percentage of lake water in the samples. Reprinted with permission from Paper I. Copyright (2016) American Chemical Society.

A clear example is represented by the position of the percentage of lake wa-ter (% Lake) in the PCoA diagram. This environmental parameter shows a

39

strong correlation with the first PCoA axis (ρ ~ 1) but a weak relation with the second coordinate (ρ ~ - 0.25).

The tools described so far allow the visualization and comparison of many different samples and parameters, but in order to properly visualize thousands of assigned formulas in chemical terms, other graphical represen-tation are needed. As previously discussed, each formula is associated with a consistent number of parameters (elemental composition, H/C, O/C, molecu-lar mass, signal intensity, etc.) thus, any graphical representation requires an appropriate data selection.

The most common method for presenting the massive number of molecu-lar formulas in complex organic mixtures is the van Krevelen diagram (Fig-ure 12) [146–149].

Figure 12. van Krevelen diagram from SRFA sample. The different regions identified by the H/C and O/C values are marked according to the following criteria: lipids (O/C = 0−0.2, H/C = 1.7−2.2), proteins (O/C = 0.2 − 0.6, H/C = 1.5−2.2, N/C ≥ 0.05), lignin (O/C = 0.1− 0.6, H/ C = 0.6−1.7, AImod < 0.67), carbohydrates (O/C = 0.6− 1.2, H/ C = 1.5−2.2), tannins (O/C = 0.6−1.2, H/C = 0.5−1.5, AI mod < 0.67) and unsaturated hydrocarbons (O/C = 0−0.1, H/C = 0.7−1.5). Carboxylic-rich alicyclic molecules (CRAM) are identified by the red ellipse [48].

In this space, thousands of molecular formulas are displayed according to their O/C and H/C ratios. Each dot represents one peak that may be com-posed of numerous structural isomers, and dots are also superimposed on each other, obscuring molecules with different masses but the same O/C and H/C. Despite these limitations, the diagram is useful in visualizing overall

40

trends and properties in such complex datasets [149]. Different classes of compounds are in fact identifiable in the diagram [48,52,132,150] (Figure 12). Obviously, this partitioning of the van Krevelen space is not a sufficient criterion to ascribe a molecule to a certain class of compounds. In fact, if a molecule falls into one of these regions due to its O/C and H/C ratios, it does not necessarily belong to that compound class. Therefore, we refer to the different regions as ‘–like’ (e.g. protein-like).

The van Krevelen diagrams were used extensively in all four papers as a convenient way to visualize and discuss changes to thousands of molecular formulas in a single diagram

41

Results and discussion

Is the Orbitrap suitable for DOM analysis? (Paper I) The main goal of this study was to evaluate the suitability of the Orbitrap technology for the analysis of DOM by comparing its performances to state-of-art technology, FTICR-MS. The latter uses a cryogenic magnet for the separation of the accumulated ions, thus limiting the availability of this ul-trahigh-resolution technique due to the expensive running costs. The Or-bitrap, where the separation is performed by an electrostatic field, represents a competitive and broadly available alternative.

In this study, DOM samples (from marine and lake sources) and their mixtures (5-95% lake water) were analysed by DI-(-)ESI-MS employing ultrahigh-resolution FTICR-MS and Orbitrap (at two resolution settings: 100.000 and 30.000), and the resulting data were compared. Recorded broad-band mass spectra were then exported and processed as described in the “Data analysis” section.

Overall, FTICR was able to assign a larger number of formulas compared to the Orbitrap. This is due to several factors. First, the different apodization of the FT signal of the two instruments, which led to the detection of a larger number of small (low abundant) signals by the FTICR compared to the Or-bitrap (Figure 13).

The second and most important factor in formulas assignments is repre-sented by the resolution. In fact, even if all three settings showed a high mass accuracy (with respect to the common masses used in the calibration) this alone was not sufficient for formula assignment. Formulas were allowed only if the mass difference to the nearest large peak was high enough to be resolved at the instrumental resolution in question, in order to minimize false assignments. The low resolution data (Orbitrap-30.000) gave a clear exam-ple, where only a limited fraction of the formulas in the whole data set were reliably assigned. Low resolution was found to reliably identify the largest peaks, mainly CHO compounds, together with their heavy-isotope (CHO13C), and abundant, low mass nitrogen containing peaks. Conversely, low abundance N-containing signals could not be reliably assigned. In fact, at this resolution setting, these low-intensity signals could not be resolved from the heavy-isotope peaks (8.11 mDa). Obviously, more information was obtained at higher resolving power. Narrower peaks allow the detection of more signals at each nominal mass, thus increasing the number of assign-

42

ments. The Orbitrap in high-resolution mode was able to easily resolve the N-containing peaks, problematic for the low resolution setting (Figure 13), but showed limitation in the assignments of sulfur and phosphorus contain-ing species. Currently, the information provided by formulas containing different heteroatoms (N, S, P) and their combinations, can be obtained only at resolution above 100.000 at m/z 400. The lower resolution mode may be useful in chromatography applications were faster data acquisition speed is desirable.

Figure 13. Comparison of raw mass spectra over a mass window of 0.1 Da. As-signed molecular formulas at the exact (not measured) mass are indicated by four-pointed stars. The data are displayed for an even nominal mass, meaning that N-containing formulas need to be resolved from 13C-containing formulas. Note that the N- containing peak at ∼426.14 is not assigned in the Orbitrap 30.000 data set, as it was not resolvable from the 13C-containing formula at this resolution. Reprinted with permission from Paper I. Copyright (2016) American Chemical Society.

All settings exhibited the ability to identify slight changes in the organic mixture as demonstrated by the Bray-Curtis dissimilarity and as showed in Figure 11. The PCoA plot showed a clear trend, highly reproducible among different techniques and samples. The data variability was extensively ex-plained by the first principle coordinate, which strongly correlated with the biogeochemical gradient (Figure 11; % Lake). Conversely, the instrumental differences were associated with the second axis. We observed that differ-ences between these samples overcome those among the instruments.

The Orbitrap was also capable of excellent reproducibility Bray-Curtis dissimilarity among replicates < 3%), apparently better than FTICR-MS.

43

This suggests the suitability of this technology in recognizing subtle altera-tion within the complex mixture.

We showed that Orbitrap technology, even at low-resolution settings, showed excellent performance and can be a powerful tool for the investiga-tion of biogeochemical gradients. Despite the lower number of resolved mo-lecular formulas, this instrument represents a very promising alternative to high-field-strength FTICR-MS for the molecular fingerprinting of DOM in the environment. Moreover, its lower costs and broader availability allows a larger number of research groups to investigate on this complex organic mixture.

Online chromatography and reduction of DOM isomeric complexity (Paper II) The goal of this study was the development of an online method for the analysis of DOM, which could integrate the benefits of the chromatographic analysis with the advantages of high resolution mass spectrometry by main-taining the low analysis time of direct infusion approach.

The developed method showed excellent reproducibility (Bray-Curtis dis-similarity among replicates ≈ 2.5%) and ability to remove interferences re-sponsible for strong charge subtraction in DI analysis and consequent lower ionization efficacy of the organic material. The ability of chromatography to remove these contaminants from the bulk and the reduction of the sample complexity in each transient, led to an increased signal-to-noise ratio (S/N) in each transient and to the assignment of an overall larger number of formu-las in the mixture.

Samples from three different water sources (marine, lake and river) were analysed by two approaches: DI-Orbitrap and reversed phase chromatog-raphy-Orbitrap, by means of negative electrospray ionization as interface.

The organic material was retained on the column according to its affinity for the stationary phase, thus generating a broad ‘humpogram’ which could be easily divided in silico at the end of the analysis. The results obtained from the different methodologies (DI vs LC) were compared, in order to demonstrate the advantages of the separation approach over the DI. Overall, the total number of formulas assigned with chromatography was higher than the number of matches in DI analysis. This is due to (1) the reduction in charge competition at the ESI source and (2) to the changing ions composi-tion entering the Orbitrap in each transient, thus allowing the ionization and detection of a larger number of species. Even if DI analysis allows the de-tailed assessment of overall differences between different DOM samples, we observed that a large portion of the more apolar material is underestimated when direct infusion is applied.

44

Subsequently, the polarity segments within the same sample were com-pared, along with comparison of like polarity segments from different sam-ples (Figure 14).

Figure 14. (a) Chromatogram from SRFA sample: early eluting material (Segment 2; pink); late eluting material (Segment 6; light blue); (b) van Krevelen diagram of the molecular distribution for Segments 2 and 6: early eluting DOM (high O/C ratio; pink); late eluting DOM (low O/C ratio; light blue); (c) Overlapped chromatograms from SRFA sample (orange) and marine water (blue). Early eluting material (Seg-ment 2; grey); (d) van Krevelen diagram of the molecular distribution for the two homologous segments (Segment 2) from the two sources: SRFA-Segment 2 (high O/C, low H/C ratio; orange); marine DOM-Segment 2 (low O/C, high H/C ratio; blue). Partial reprint was possible with permission from Paper II. Copyright (2018) American Chemical Society.

We found that isomeric complexity led to broad retention of individual mo-lecular formulas. Moreover, important differences in molecular diversity were revealed across the chromatographic run, with oxygen-rich compounds eluting first (Figure 14).

When samples from different sources were compared, the ‘humpograms’ derived from the LC analysis showed similar retention times but different shapes, suggesting that homologous polarity fractions have variable abun-dance (Figure 14c). Despite experiencing the same retention time, these po-

45

larity fractions exhibited an extremely variable composition (Figure 14d); as further confirmed by multivariate statistical techniques.

A rather different distribution in terms of molecular saturation emerged from the comparison of terrestrial and marine DOM, especially in relation to the more hydrophilic material: a higher degree of saturation (H/C ≥ 1) char-acterized the marine DOM, whereas the terrestrial material showed an addi-tional contribution from unsaturated and oxygen rich components.

The online coupling of liquid chromatography to the high resolution Or-bitrap-MS enhanced enormously the amount of information achievable in a single run, maintaining high resolution data, reducing analysis time com-pared with fractionation and minimizing the influence of contaminations. Moreover, the introduction of in silico fractionation makes the method ex-tremely flexible. The developed method allows the easy and detailed com-parison of DOM samples from a variety of sources. Additionally, the wide availability of the employed instruments allows many research groups to perform similar analysis.

The method still presents significant limitations. We found that the extreme isomeric complexity of the natural mixture cannot be resolved into individual compounds with one-dimensional chromatography. Moreover, consistent overlap of structural isomers between adjacent segments was observed.

Fate of labile DOM (Paper III) In mass spectrometric studies focusing on the analysis of freshwater DOM, the characterization of the allochthonous components is favored compared to the autochthonous material, underrepresented in current research. This dis-crepancy is indubitably related to the molecular compositions of the two DOM sub-classes and their susceptibility to routinely employed mass spec-trometric approaches involving negative ESI (electrospray ionization). Al-lochthonous DOM is mainly of terrigenous origin, rich in carboxylic acids, easily extracted and favorable to negative ESI analysis. The autochthonous material is rich in biologically labile components (carbohydrates, proteins, amino acids, etc) difficult to isolate and less inclined to ionize in negative mode. The chemical characterization of this DOM fraction is crucial to un-derstand its fate in the water column and the study of its lability is important to evaluate the impact that a rapid increase in primary production might have in the biogeochemical processes, particularly during eutrophication.

In Paper III, the online method developed in Paper II was successfully applied for the characterization of ‘freshly’ produced DOM generated by cyanobacteria cultures (CyDOM). Two growing conditions were investigat-ed (axenic and natural environments); and finally the abundant CyDOM released in the axenic culture was incubated in unfiltered natural water in

46

presence of heterotrophic communities and the characterization of the ioniz-able original material was monitored over the incubation period.

As expected, DOM released from primary producers was far less complex than the recalcitrant material commonly observed in bulk-DOM samples. As previously introduced in Paper II, the isomeric complexity of DOM mix-tures cannot be resolved in a single chromatographic run; generally, the ex-tracted ion current (EIC) of single m/z values emerges as a large ‘hump’ eluting over a broad retention time interval. In the case of CyDOM produced in axenic conditions, despite the broad molecular distribution, the majority of the MS signal assigned to a formula was linked to chromatographically resolved peaks; even in presence of the bulk-DOM (incubation of cyanobac-teria in natural waters), fully resolved chromatographic profiles from CyDOM could be observed.

The ability to resolve chromatographic peaks allowed the monitoring of the ‘freshly’ produced material in presence of heterotrophic communities over the incubation (Figure 15). The results revealed a fast removal of the observed CyDOM profiles (Figure 15), preferential toward saturated (or labile, conventionally identified having H/C > 1.5) and hydrophilic compo-nents. The quick decrease in the abundance of these species in presence of heterotrophic bacteria, limits their presence in aquatic samples, thus explain-ing their poor characterization in the analysis of freshwater DOM.

Figure 15. Incubation of the cyanobacteria-derived DOM in presence of hetero-trophic bacteria: initial (start) and final (end) stages. van Krevelen diagrams: molec-ular distribution of chromatographically resolved peaks (points); peak intensity (size); non-confidently assigned peaks (grey-shaded points); confidently assigned peaks (coloured points); elution time in minutes (colour scale). Division of labile (H/C>1.5) [151] and stable regions (black line).

From the results emerged that primary producers, in the tested growth condi-tions, produced a large variety of material covering a broad polarity and saturation range and the dominance of the labile component was observed.

47

This study revealed that CyDOM susceptible to negative mode ESI is char-acterized by a high biological lability as it was rapidly processed in natural waters. However, the ionizable CyDOM probably represents only a fraction of the total material released by the primary producers, as a large part of the biologically produced material (such as polysaccharides and proteins) are not detected by the method. In this study, the online combination of chromatog-raphy with high resolution mass spectrometry was revealed to be highly useful in the analysis of ionizable autochthonous DOM. The application of a separation technique was indispensable to observe the fate of the labile com-ponents released by primary producers by monitoring the behavior of the chromatographic profiles emerging form the bulk-material.

Hide-and-seek: ESI-MS ‘invisible’ DOM (Paper IV) Not all DOM is ionizable by routine ESI-MS in negative mode because not all molecules ionize following the same mechanism. Therefore, due to the complexity of the mixture, only a partial ionization of the DOM pool is ex-pected in routine analysis. Despite the large number of analytical techniques applied to the analysis of DOM, a universal approach able to unravel the structural and elemental composition of the DOM is missing. Structural characterization and quantification of organic material can be achieved for example with the use of NMR or OCD, respectively, but information on the DOM elemental composition can be obtained only by the application of high resolution mass spectrometry. However, as previously mentioned HRMS results are indissolubly linked to the selected ion source. Different ionization sources are able to ionize different classes of compounds, but to date no sin-gle technique is able to fully ionize all the components in the mixture while preserving the analyte integrity.

Paper IV focused on the estimation of the DOM fraction ‘invisible’ to negative electrospray ionization. The same chromatographic approach used in Paper II and III was employed in this study; additionally to the ESI-HRMS operating in negative mode, a complementary detector was em-ployed. The charged aerosol detector (CAD) was used in combination with chromatography and mass spectrometry to enhance the detection capabilities of the methods

While HPLC-ESI-HRMS generated qualitative chemical information about the components in the mixture susceptible to the selected ionization source and conditions, the CAD response was proportional exclusively to the amount of material eluting from the chromatographic system. Mass spectro-metric signal and quantitative response from the CAD were merged to ex-press the ability of the material to ionize in negative ESI mode. This parame-ter, defined as relative response factor, was employed to monitor the suscep-

48

tibility of the material to the ionization during the chromatographic elution interval.

Suwannee River Fulvic Acid (SRFA) reference material was fractionated by size-exclusion chromatography (SEC) in four aliquots with decreasing molecular weight. The isolated fractions (Figure 16) revealed the presence of an abundant portion of material ‘invisible’ in negative ESI mode (ESI-inactive). Fractions characterized by higher molecular weight (Fractions 1 and 2) produced a signal in the CAD proportional to their abundance, but no meaningful signal was registered in ESI-HRMS (Figure 16C). Fractions 3 and 4 showed a high response in both detectors but some peculiar profiles emerged. The ESI-HRMS and the abundance profiles in Fraction 3 diverged: the early eluting component (more hydrophilic) was more susceptible to ionization than the late eluting material (more hydrophobic).

This discrepancy could be better appreciated by the use of the relative re-sponse factor (Figure 16 C). The relative response factor profiles emerging from the four fractions allowed further considerations. It was possible to confidently conclude that a third the material in SRFA (Fractions 1 and 2) was resistant to negative ESI ionization (ESI-inactive material) (Figure 16 B-C). The last two fractions (Fractions 3 and 4), were characterized by mate-rial susceptible to ionization (ESI-active), but some contribution from the presence of an ESI-inactive component could not be completely excluded.

The evaluation of the abundance of ESI-inactive component in SRFA could not be assessed exclusively by fractionation; therefore an additional strategy was employed. The relative response factor of the reference material was compared to the average relative response factor resulting from a mix-ture of purchased carboxylic acids and a simple two components model was proposed. The model was based on two extreme and simplistic assumptions: (i) only molecules with carboxylic functionalities generate a response in negative ESI-HRMS, therefore no signal means absence of carboxyl group and vice versa; (ii) the mixture of carboxylic acids was representative for the small acidic molecules in the complex mixture. The average relative re-sponse factor, representative of each sample (bulk and fractions), was em-ployed to predict the distribution of ESI-active and inactive components in the natural mixture. Eventually, the result emerging from the analysis of the bulk-SRFA revealed that about 70% of the material was ESI-inactive. How-ever, these results disagree with those emerging from the combination of the fractions, suggesting that only 17% of the material in SRFA is susceptible to ionization. This inconsistency could be related to a mechanism of ionization promotion that manifests in the bulk material and it is not replicated after the ‘simplification’ of the mixture induced by the fractionation by SEC; this aspect deserves further investigation, preferably by application of multidi-mensional chromatography.

Analogous analysis were performed on other reference materials (repre-sentative of terrestrial DOM) before and after solid phase extraction (SPE);

49

no significant change was observed before and after the sample treatment in the resulting relative response factors profiles; additionally, for all mixtures the model revealed a limited amount of ESI-active component, never ex-ceeding 35% of the total material. Ultimately, two samples isolated for the same aquatic system (fjord and river) were analysed. The model revealed a large ESI-active component in the fjord sample (69%) while the river was characterized largely by ESI-inactive material (66%), suggesting the removal of large part of the inactive component in the transport from land to sea. These findings can have important implications in the global carbon cycle, especially because this material eludes the most common routine analysis performed for DOM characterization. Further investigations on the ESI-inactive component in DOM samples are therefore essential.

Figure 16. (A) SRFA fractionation by size-exclusion chromatography (SEC) - charged aerosol detector (CAD); four isolated fraction delimited by dashed black lines; (B) SRFA fractions analyzed by reverse-phase chromatography coupled to ESI-MS (red) and CAD (blue); (C) Relative response factor (black) expressed as assigned MS signals per amount of material (from CAD).

50

Conclusions and future aspects

The aim of the presented thesis was to develop new methods able to enhance the analysis of the dissolved organic matter and enable a wider range of re-searchers to participate in the advancement of this field.

Initially, the applicability of the Orbitrap mass analyser for the analysis of complex mixtures was evaluated. Paper I demonstrated that cutting edge techniques (FTICR-MS) are not always necessary to achieve fundamental insights on the molecular composition of DOM. The suitability of Orbitrap technology offered the possibility to explore the potentialities of online chromatographic separation for the analysis of DOM samples. Paper II showed that online separation in combination with high resolution mass spectrometry and electrospray ionization is an excellent alternative to the commonly used direct infusion approach, and proved that the application of simple and broadly accessible methods, able to reduce the DOM isomeric complexity, enormously enhance the information achievable in a single analysis. Chromatographic separation also demonstrated the capability to resolve single components in the mixture if present in high abundance; this ability inspired the study performed in Paper III, where the poorly charac-terized autochthonous DOM produced by a common cyanobacterium was investigated. Online separation enabled the monitoring of chromatograph-ically resolved peaks, so the fate of the labile components, released by the primary producers and incubated in presence of heterotopic bacteria, could be observed.

The combination of online chromatography to high resolution mass spec-trometry is a powerful tool in DOM analysis, expanding the HRMS analysis capabilities by adding an extra dimension to the investigated chemical space. Standing alone HRMS techniques are able to discriminate among isobaric components and their efficiency is proportional to their resolution capabili-ties. Structural isomers are the ‘Achilles’ heel’ of these powerful instru-ments; in fact, in absence of a separation approach, components character-ized by the same molecular formula are indistinguishable. Enhanced isomer-ic discrimination is not the only advantage emerging from the application of chromatography in DOM analysis; the separation guarantees valuable quali-tative information on physical and chemical characteristics of the material in the mixture in relation to the selected retention mechanism. However, the potential of the chromatographic approach in DOM analysis is still limited by the employed ionization source. Well established routine analysis of natu-

51

ral complex mixtures involves the use of electrospray ionization (ESI) oper-ating in negative mode as interphase between the sample introduction meth-od and the mass spectrometer. The popularity of the ESI source is dictated by the abundance of acidic species in DOM samples, which are highly sus-ceptible to ionization in negative mode.

However, not all molecules ionize following the same mechanism; under fixed experimental conditions some molecules are more prone to ionization while others are resistant to forming ions. The intrinsic complexity of the DOM supports the idea that only a part of the material is susceptible to nega-tive ESI ionization in routine analysis, suggesting the presence of an unchar-acterized pool of organic material probably variable in abundance across different environmental settings. Paper IV explored the nature of DOM fractions susceptible and resistant to ESI operating in negative mode; this study demonstrated that routine analyses are not always suitable for a com-prehensive interpretation of the DOM molecular diversity (Figure 17). Therefore, awareness about the limitations of the electrospray ionization is essential to correctly interpret the results: to quote Dr Ryan P. Rodgers “The ESI source lies to you!”.

Figure 17. Conceptual diagram showing three main pools of DOM present in aquat-ic samples. Papers I-III focussed on the carboxylic acids and small metabolites, while paper IV also discussed the presence of lignin polymers, absent from ESI-MS data. Access to information about polarity (x axis) and size (y axis) is available with the use of chromatography.

52

The conceptual chemical space represented in Figure 17, organized in terms of polarity (logP) and molecular weight (MW) separation, gives a general overview on the distribution of the different compounds classes either acces-sible or ignored by negative ESI-HRMS in routine DOM analysis. The in-formation retained by the uncharacterized material can be disclosed by the combination of techniques complementary to ESI-HRMS analysis.

In summary, this thesis demonstrated that a broader understanding of the DOM complexity can be achieved by the application of simple and broadly accessible techniques. Upon the results presented in this work, I strongly recommend the inclusion of online chromatography coupled with HRMS among the routinely employed techniques for DOM characterization, togeth-er with the use of quantitative detection where possible. Further develop-ments require the use of complementary techniques and the elaboration of alternative data analysis paradigms able to combine the pool of information deriving from different approaches. Expand the toolbox of the available methods remains the key factors to unravel the intricacy of this Damned Obscure (but intriguing) Mess.

53

Populärvetenskaplig sammanfattning

Är det klara friska källvattnet i en fjällsjö verkligen kemiskt rent? För att besvara denna fråga är det nödvändigt att definiera vad "rent" betyder. I ke-mins värld består en ren substans uteslutande av en typ av atom eller mole-kyl. Rent vatten består uteslutande av en enkel molekyl (H2O) sammansatt av två väteatomer (H) och en syreatom (O). Så, tillbaka till frågan: om vi samlar lite vatten från en orörd fjällsjö, skulle provet uteslutande innehålla H2O? Det enkla svaret är nej. Så vad kan man då hitta i vattensamlingar förutom H2O? Jag antar att det mest korrekta svaret skulle vara: ett helt uni-versum av molekyler.

Naturliga vatten runt om i världen är rika på oorganiskt material (såsom natrium, klorid, kalcium, fosfat), men den verkliga kemiska komplexiteten representeras av den fenomenala mångfalden av organiska molekyler som finns närvarande, definierade som lösligt organiskt material (på engelska dissolved organic matter - DOM).

Vad är då DOM mer exakt? Och hur hamnar det i vattendragen? DOM är en allestädes närvarande och heterogen blandning av molekyler

(förmodligen miljontals olika) som härrör från ett stort antal biologiska, ke-miska och fysiska processer som är sammankopplade i ett evigt kretslopp mellan icke-levande och levande material känt som "kolcykeln eller kolets kretslopp". DOM är resultatet av omvandlingen och nedbrytningen av bio-massan och kan produceras antingen internt eller externt till vattendraget främst av primära producenter (fotosyntetiska organismer) eller transporteras till vattensystemet genom avrinning från omkringliggande mark, via perko-lationer och lakvatten. DOM spelar en avgörande roll i den globala koldi-oxidcykeln som reglerar och upprätthåller de biologiska processerna i vat-tensystemen. Trots sin betydelse för kretsloppet och DOMs globala påverkan på naturen så är den samlade kunskapen om dess kemiska mångfald fortfa-rande dåligt. Man kan verkligen undra varför det är så.

Som redan nämnts är DOM bland annat viktigt för det mikrobiella krets-loppet i vattenmiljöer på global nivå. Förändringar i regleringen av DOM leder till påverkad produktion och transformation, orsakad av exempelvis miljöförändringar (ökande globala temperaturer, antropogena effekt etc.). Dessa kan i sin tur ha en stor påverkan på biogeokemiska processer, vilket potentiellt ökar frisättningen av växthusgaser i atmosfären. Det enda sättet att förstå och förutsäga ödet för detta organiska material är att samla en

54

övergripande och detaljerad kemisk kunskap om denna oerhört komplexa blandning.

Fokus för denna avhandling var utvecklingen av metoder för karaktärise-ring av DOM i naturliga vatten ner på molekylär nivå. Undersökningen av den enorma komplexiteten hos DOM-blandningen representerar en verklig utmaning för analytisk kemi, den disciplin inom kemin som ägnas åt ut-veckling och tillämpning av metoder för analys och identifiering av kemiska ämnen. I detta arbete användes avancerade analytiska tekniker såsom kro-matografi och högupplösande masspektrometri i kombination med elektrosprayjonisering för att undersöka den kemiska mångfalden hos DOM. Dessa tekniker är de mest kraftfulla metoderna som finns att tillgå inom ke-misk analys. De verkar i synergi för att ge så rik information som möjligt om det undersökta provet: kromatografi ger separation av komponenterna i blandningen; masspektrometri är en universell detektor som möjliggör iden-tifiering och kvantifiering av komponenterna i provet; elektrosprayjonisering är gränssnittet mellan de två systemen som genererar och överför joniserade molekyler från kromatografin till masspektrometern.

Hypotetiskt skulle kombinationen av dessa tekniker kunna avslöja mång-falden av DOM, men i verkligheten har det visat sig att en fullständig förstå-else för den komplexa blandningens kemi ännu inte uppnåtts.

I denna avhandling demonstrerades det också att avancerade instrument inte alltid är nödvändiga för att uppnå grundläggande insikter om molekyl-kompositionen för DOM och att tillämpningen av enklare och mer allmänt tillgängliga metoder även kan bidraga med viktig information. Det demon-strerades också att även väletablerade rutinanalyser inte alltid är lämpliga för en kunna uppnå en korrekt tolkning av DOMs molekylär mångfald, eftersom de ibland leder fram till en del inte helt representativa resultat.

Den fullständiga karaktäriseringen av DOM är således för närvarande omöjlig baserat på en enda teknik; det är uppenbart att en kombination av ansträngningar och expertis är nödvändig och grundläggande för att bättre kunna förstå och följa komplexiteten i denna enorma pool av organiskt material.

55

Acknowledgments

How long does it take to realize that something is about to end? As I type I am thinking that this incredible Odyssey started more than four years ago, I remember exactly how it started. I received a phone call from an unusual number, on the display I read “+46”...in that moment I was having a small conversation with an English speaking friend, to practice my ‘communica-tion’ skills, she kindly invited me to answer the phone and Jonas was on the other side, offering me THE PhD position at the Uppsala University, …in Sweden!! I was so shocked that I had the most inefficient reply ever…I am not even sure I emitted any sound for few seconds. I remember that when we hung up, I asked my friend to confirm if what I understood was correct, and it was! That was a dream came true!! That was the moment that changed everything.

Jonas and Lars, you made this real!! I cannot thank you enough for that. I feel so lucky for the chance to meet you and work with you. Jonas, thank you for your constant support and for telling me that everything was going to be ok, you were right, every time! Lars, thank you for every time you point-ed out the positive aspects in a project, even when all I could see was unfin-ished results, your approval and guidance always boosted my confidence.

Differently from the majority of the PhD students, I was so lucky to be supported by an incredible number of supervisors. Per, you became officially my supervisor late in my studies, but you have been more than official for me from the very beginning. You are an excellent and patient teacher, a great researcher and a wonderful person, talking with you always inspires me, I am really happy I have met you.

Jeff, I do not know from where to start. I would not be writing these sec-tion of the thesis if it wasn´t for you. I could not hope for a better supervisor, everything I learned in these years it has been possible only because you were constantly available, always ready to listen to me and to help me through tough moments. You are not only a brilliant researcher but also an extraordinary friend. Everybody should know that you e-mailed me few hours before the submission of the thesis to confirm the last correction to the final document. You are truly great!

My special thanks to everyone at Analytical Chemistry! I missed you all already! Lately I missed so many fikas and lunches that it feels like I am not at the department anymore

56

Kumari, you always gave me so much affection and attention; I always felt I could count on you for everything. I will miss you a lot. Anna, I always en-joyed to share a late lunch with you, I am so sad that recently they became more and more rare. Marit and Jean, you are the column of the department, thank you for your endless patience and help in the teaching labs. Ingela, thank you for the boost of confidence you gave me during though moments. Kyle, your curiosity and inventiveness are contagious, talking to you about science make you think that everything is possible! I missed latest updated recently? How is going with your pronunciation of the italian ‘r’? Eszter, it was the time of my licentiate and I was already saying how much I adored you and with time that sentiment did just grow bigger and bigger. Thank you for continu-ously checking on me especially in these dark times of crazy writing :*

Catia, Johan, Leonidas, Philipp: guys I miss you! How long ago were we having a fika together? I promise homemade tiramisu as soon as possible!

To all the PhDs before me: Neil, Marcus, Hilde, Malin, Julia…I love you guys! My time here was special because of all of you. I know that we were never only colleagues, I know that in 30 years from now we’ll still be friends, maybe even scattered around the globe but never truly apart.

Alla mia famiglia voglio dedicare un ringraziamento speciale. Senza il vostro amore, supporto, sostegno, incoraggiamento e fiducia sarei persa. Vivere cosí distante da voi è spesso difficile. Le telefonate e le visite due volte l’anno non posso sostituire la quotidianità, vorrei potervi incontare ogni volta che sento la vostra mancanza, raccontarvi dal vivo della mia giornata, e sentire della vostra. Vorrei poter passare piú tempo con Stefano, non perché abbia paura che il tempo ci possa dividere, ma perché sento che sto perdendo momenti importanti che nessuno ci ridará mai. Mamma, Papá, mi dispiace di essere cosí lontana in un momento cosí difficile. Mamma, sei la persona piú forte e determinata che esista, e sono certa, non mi stancheró mai di ripertelo, che andrá tutto bene. La tua forza d’animo è immensa e la mia fiducia in te incrollabile. Sei il modello a cui aspiro e se avremo una bambina spero di avere con lei lo stesso meraviglioso rapporto che io ho con te. Vi amo immensamente.

Piero, amore mio non posso vivere la mia vita senza di te. Quando una delle mie assurde paure (tu sai quali) prende il sopravvento, sento che mi manca il respiro. Sei il sole della mia vita, ogni giorno mi stupisco di avere un uomo cosí meraviglioso al mio fianco. Con te mi sento amata e protetta, sei l’unico al mondo per me. Grazie di esistere amore mio bagolissimo e dolcissimo. Ti amo.

57

References

1. Economou EN, Economou EN. The Observable Universe. In: From Quarks to the Universe. 2016. p. 249–70.

2. Bohacek RS, McMartin C, Guida WC. The art and practice of structure-based drug design: A molecular modeling perspective. Vol. 16, Medicinal Research Reviews. 1996. p. 3–50.

3. Schlesinger WH, Bernhardt ES. Biogeochemistry: An Analysis of Global Change, Third Edition. Biogeochemistry: An Analysis of Global Change, Third Edition. 2013. 1-672 p.

4. Dittmar T, Stubbins A. Dissolved Organic Matter in Aquatic Systems [Internet]. 2nd ed. Vol. 12, Treatise on Geochemistry: Second Edition. Elsevier Ltd.; 2013. 125-156 p. Available from: http://dx.doi.org/10.1016/B978-0-08-095975-7.01010-X

5. Hansell DA, Carlson CA, Repeta DJ, Schlitzer R. Dissolved organic matter in the ocean a controversy stim ulates new insights. Oceanography. 2009;22(SPL.ISS. 4):202–11.

6. Mostofa KMG, Yoshioka T, Mottaleb MA, Vione D. Photobiogeochemistry of Organic Matter - Principles and Practices in Water Environments [Internet]. Vol. 3, Advances in Wind Energy Conversion Technology. Springer Berlin Heidelberg; 2013. 917 p. Available from: http://books.google.com/books?hl=en&amp;lr=&amp;id=LMjTXmTgXJoC&amp;oi=fnd&amp;pg=PR6&amp;dq=Information+Technologies+in+Environmental+Engineering:+New+Trends+and+Challenges&amp;ots=_0FtnJFN1U&amp;sig=WIAH21eAnJHempcNp5udXCVQePU

7. Kim C, Nishimura Y, Nagata T. Role of dissolved organic matter in hypolimnetic mineralization of carbon and nitrogen in a large, monomictic lake. Limnol Oceanogr. 2006;51(1):70–8.

8. Asmala E, Haraguchi L, Jakobsen HH, Massicotte P, Carstensen J. Nutrient availability as major driver of phytoplankton-derived dissolved organic matter transformation in coastal environment. Biogeochemistry. 2018;137(1–2):93–104.

9. Carlson CA, del Giorgio PA, Herndl GJ. Microbes and the dissipation of energy and respiration: From cells to ecosystems. Oceanography. 2007;20(SPL.ISS. 2):89–100.

10. Azam F, Fenchel T, Field J, Gray J, Meyer-Reil L, Thingstad F. The Ecological Role of Water-Column Microbes in the Sea. Mar Ecol Prog Ser [Internet]. 1983;10:257–63. Available from: http://www.int-res.com/articles/meps/10/ m010p257.pdf

11. Dittmar T. Reasons Behind the Long-Term Stability of Dissolved Organic Matter [Internet]. Second Edi. Biogeochemistry of Marine Dissolved Organic Matter: Second Edition. Elsevier Inc.; 2015. 369-388 p. Available from: http://dx.doi.org/10.1016/B978-0-12-405940-5.00007-8

58

12. Hansell DA. Recalcitrant Dissolved Organic Carbon Fractions. Ann Rev Mar Sci. 2013;5(1):421–45.

13. Tranvik LJ, Downing JA, Cotner JB, Loiselle SA, Striegl RG, Ballatore TJ, et al. Lakes and reservoirs as regulators of carbon cycling and climate. Limnol Oceanogr. 2009;54(6 PART 2):2298–314.

14. Morris DP, Zagarese H, Williamson CE, Balseiro EG, Hargreaves BR, Modenutti B, et al. The attenuation of solar UV radiation in lakes and the role of dissolved organic carbon. Limnol Oceanogr. 1995;40(8):1381–91.

15. Molot LA, Keller W, Leavitt PR, Robarts RD, Waiser MJ, Arts MT, et al. Risk analysis of dissolved organic matter-mediated ultraviolet B exposure in Canadian inland waters. Can J Fish Aquat Sci [Internet]. 2004;61(12):2511–21. Available from: http://www.nrcresearchpress.com/doi/10.1139/f04-165

16. Aiken GR, Hsu-Kim H, Ryan JN. Influence of dissolved organic matter on the environmental fate of metals, nanoparticles, and colloids. Environ Sci Technol. 2011;45(8):3196–201.

17. Benner R. Loose ligands and available iron in the ocean. Vol. 108, Proceedings of the National Academy of Sciences of the United States of America. 2011. p. 893–4.

18. Carter CW, Suffet IH. Binding of DDT to Dissolved Humic Materials. Environ Sci Technol. 1982;16(11):735–40.

19. McCarthy JF, Jimenez BD. Interactions between Polycyclic Aromatic Hydrocarbons and Dissolved Humic Material: Binding and Dissociation. Environ Sci Technol. 1985;19(11):1072–6.

20. Zhu FD, Choo KH, Chang HS, Lee B. Interaction of bisphenol A with dissolved organic matter in extractive and adsorptive removal processes. Chemosphere [Internet]. 2012;87(8):857–64. Available from: http://dx.doi.org/10.1016/j.chemosphere.2012.01.026

21. Herzsprung P, Von Tümpling W, Hertkorn N, Harir M, Büttner O, Bravidor J, et al. Variations of DOM quality in inflows of a drinking water reservoir: Linking of van krevelen diagrams with EEMF spectra by rank correlation. Environ Sci Technol. 2012;46(10):5511–8.

22. Zhang H, Zhang Y, Shi Q, Hu J, Chu M, Yu J, et al. Study on Transformation of Natural Organic Matter in Source Water during Chlorination and Its Chlorinated Products using Ultrahigh Resolution Mass Spectrometry. Environ Sci Technol [Internet]. 2012;46(8):4396–402. Available from: http://pubs.acs.org/doi/10.1021/es203587q

23. Zhang H, Zhang Y, Shi Q, Ren S, Yu J, Ji F, et al. Characterization of low molecular weight dissolved natural organic matter along the treatment trait of a waterworks using Fourier transform ion cyclotron resonance mass spectrometry. Water Res [Internet]. 2012;46(16):5197–204. Available from: http://dx.doi.org/10.1016/j.watres.2012.07.004

24. Richardson SD, Ternes TA. Water analysis: emerging contaminants and current issues. Anal Chem [Internet]. 2009;81(12):4645–77. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24502364

25. Hua G, Reckhow DA. Characterization of disinfection byproduct precursors based on hydrophobicity and molecular size. Environ Sci Technol. 2007;41(9):3309–15.

26. Richardson SD, Plewa MJ, Wagner ED, Schoeny R, DeMarini DM. Occurrence, genotoxicity, and carcinogenicity of regulated and emerging disinfection by-products in drinking water: A review and roadmap for research. Mutat Res - Rev Mutat Res. 2007;636(1–3):178–242.

59

27. Lavonen EE, Gonsior M, Tranvik LJ, Schmitt-Kopplin P, Köhler SJ. Selective chlorination of natural organic matter: Identification of previously unknown disinfection byproducts. Environ Sci Technol. 2013;47(5):2264–71.

28. Zhao Z, Gonsior M, Schmitt-Kopplin P, Zhan Y, Zhang R, Jiao N, et al. Microbial transformation of virus-induced dissolved organic matter from picocyanobacteria: coupling of bacterial diversity and DOM chemodiversity. ISME J [Internet]. 2019;13(10):2551–65. Available from: http://dx.doi.org/10.1038/s41396-019-0449-1

29. Cole JJ, Prairie YT, Caraco NF, McDowell WH, Tranvik LJ, Striegl RG, et al. Plumbing the global carbon cycle: Integrating inland waters into the terrestrial carbon budget. Ecosystems. 2007;10(1):171–84.

30. Hertkorn N, Ruecker C, Meringer M, Gugisch R, Frommberger M, Perdue EM, et al. High-precision frequency measurements: Indispensable tools at the core of the molecular-level analysis of complex systems. Anal Bioanal Chem. 2007;389(5):1311–27.

31. Wetzel RG. Limnology: Lake and River Ecosystems [Internet]. Vol. 37, Journal of Phycology. 2001. 1006 p. Available from: http://doi.wiley.com/10.1046/j.1529-8817.2001.37602.x

32. Giorgio PA del, Davis J. Patterns in Dissolved Organic Matter Lability and Consumption across Aquatic Ecosystems. Aquat Ecosyst. 2003;399–424.

33. Thurman EM. Organic Geochemistry of Natural Waters. Organic Geochemistry of Natural Waters. 1985.

34. Sobek S, Tranvik LJ, Prairie YT, Kortelainen P, Cole JJ. Patterns and regulations of dissolved organic carbon: an analysis of 7,500 widely distributed lakes. Limnol Oceanogr [Internet]. 2007;52(3):1208–19. Available from: doi.wiley.com/10.4319/lo.2007.52.3.1208

35. Zark M, Christoffers J, Dittmar T. Molecular properties of deep-sea dissolved organic matter are predictable by the central limit theorem: Evidence from tandem FT-ICR-MS. Mar Chem [Internet]. 2017;191:9–15. Available from: http://dx.doi.org/10.1016/j.marchem.2017.02.005

36. Hawkes JA, Patriarca C, Sjöberg PJR, Tranvik LJ, Bergquist J. Extreme isomeric complexity of dissolved organic matter found across aquatic environments. Limnol Oceanogr Lett [Internet]. 2018; Available from: http://doi.wiley.com/10.1002/lol2.10064

37. Hertkorn N, Frommberger M, Witt M, Koch BP, Schmitt-Kopplin P, Perdue EM. Natural organic matter and the event horizon of mass spectrometry. Anal Chem. 2008;80(23):8908–19.

38. Riedel T, Dittmar T. A method detection limit for the analysis of natural organic matter via fourier transform ion cyclotron resonance mass spectrometry. Anal Chem. 2014;86(16):8376–82.

39. Weishaar JL, Aiken GR, Bergamaschi BA, Fram MS, Fujii R, Mopper K. Evaluation of specific ultraviolet absorbance as an indicator of the chemical composition and reactivity of dissolved organic carbon. Environ Sci Technol. 2003;37(20):4702–8.

40. Helms JR, Stubbins A, Ritchie JD, Minor EC, Kieber DJ, Mopper K. Absorption Spectral Slopes and Slope Rations As Indicators of Molecular Weight,Source,and Photoleaching of Chromophoric Dissolved Organic Matter. Limnol Oceanogr. 2008;53(3):955–69.

41. Fellman JB, Hood E, Spencer RGM. Fluorescence spectroscopy opens new windows into dissolved organic matter dynamics in freshwater ecosystems: A review. Limnol Oceanogr. 2010;55(6):2452–62.

60

42. Dittmar T, Koch B, Hertkorn N, Kattner G. A simple and efficient method for the solid-phase extraction of dissolved organic matter (SPE-DOM) from seawater. Limnol Ocean Methods [Internet]. 2008;6:230–5. Available from: http://www.aslo.org/lomethods/free/2008/0230.html%5Cnhttp://m.aslo.info/lomethods/free/2008/0230.pdf

43. Kim S, Simpson AJ, Kujawinski EB, Freitas MA, Hatcher PG. High resolution electrospray ionization mass spectrometry and 2D solution NMR for the analysis of DOM extracted by C18 solid phase disk. Org Geochem. 2003;34(9):1325–35.

44. Koprivnjak JF, Pfromm PH, Ingall E, Vetter TA, Schmitt-Kopplin P, Hertkorn N, et al. Chemical and spectroscopic characterization of marine dissolved organic matter isolated using coupled reverse osmosis-electrodialysis. Geochim Cosmochim Acta [Internet]. 2009;73(14):4215–31. Available from: http://dx.doi.org/10.1016/j.gca.2009.04.010

45. Aiken GR, Mcknight IDM, Thorn IKA, Thurman EM. Isolation of hydrophilic organic acids from water using nonionic macroporous resins Hydrophobic acids XAD - 4 ~ I NaOH ). Hydrophilic acids. Org Geochem [Internet]. 1992;18(4):567–73. Available from: http://www.sciencedirect.com/science/article/pii/014663809290119I

46. Sandron S, Rojas A, Wilson R, Davies NW, Haddad PR, Shellie RA, et al. Chromatographic methods for the isolation, separation and characterisation of dissolved organic matter. Environ Sci Process Impacts [Internet]. 2015;17(9):1531–67. Available from: http://dx.doi.org/10.1039/C5EM00223K

47. Simpson AJ, Kingery WL, Hatcher PG. The identification of plant derived structures in humic materials using three-dimensional NMR spectroscopy. Environ Sci Technol. 2003;37(2):337–42.

48. Hertkorn N, Benner R, Frommberger M, Schmitt-Kopplin P, Witt M, Kaiser K, et al. Characterization of a major refractory component of marine dissolved organic matter. Geochim Cosmochim Acta. 2006;70(12):2990–3010.

49. Hertkorn N, Harir M, Koch BP, Michalke B, Schmitt-Kopplin P. High-field NMR spectroscopy and FTICR mass spectrometry: Powerful discovery tools for the molecular level characterization of marine dissolved organic matter. Biogeosciences. 2013;10(3):1583–624.

50. Capley EN, Tipton JD, Marshall AG, Stenson AC. Chromatographic reduction of isobaric and isomeric complexity of fulvic acids to enable multistage tandem mass spectral characterization. Anal Chem. 2010;82(19):8194–202.

51. D’Andrilli J, Dittmar T, Koch BP, Purcell JM, Marshall AG, Cooper WT. Comprehensive characterization of marine dissolved organic matter by Fourier transform ion cyclotron resonance mass spectrometry with electrospray and atmospheric pressure photoionization. Rapid Commun Mass Spectrom. 2010;24:643–50.

52. Stenson AC. Reversed-phase chromatography fractionation tailored to mass spectral characterization of humic substances. Environ Sci Technol. 2008;42(6):2060–5.

53. Hawkes JA, Sjöberg PJR, Bergquist J, Tranvik LJ. Complexity of dissolved organic matter in the molecular size dimension: insights from coupled size exclusion chromatography electrospray ionisation mass spectrometry. Faraday Discuss [Internet]. 2019;218:52–71. Available from: http://xlink.rsc.org/?DOI=C8FD00222C

61

54. Gonsior M, Powers LC, Williams E, Place A, Chen F, Ruf A, et al. The chemodiversity of algal dissolved organic matter from lysed Microcystis aeruginosa cells and its ability to form disinfection by-products during chlorination. Water Res [Internet]. 2019;155:300–9. Available from: https://doi.org/10.1016/j.watres.2019.02.030

55. Smith DF, Podgorski DC, Rodgers RP, Blakney GT, Hendrickson CL. 21 Tesla FT-ICR Mass Spectrometer for Ultrahigh-Resolution Analysis of Complex Organic Mixtures. Anal Chem. 2018;90(3):2041–7.

56. Hendrickson CL, Quinn JP, Kaiser NK, Smith DF, Blakney GT, Chen T, et al. 21 Tesla Fourier Transform Ion Cyclotron Resonance Mass Spectrometer: A National Resource for Ultrahigh Resolution Mass Analysis. J Am Soc Mass Spectrom. 2015;26(9):1626–32.

57. Marshall a G, Hendrickson CL, Jackson GS. Fourier transform ion cyclotron resonance mass spectrometry: a primer. Mass Spectrom Rev. 1998;17:1–35.

58. Koch BP, Witt M, Engbrodt R, Dittmar T, Kattner G. Molecular formulae of marine and terrigenous dissolved organic matter detected by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Geochim Cosmochim Acta. 2005;69(13):3299–308.

59. Kujawinski EB. Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry (ESI FT-ICR MS): Characterization of Complex Environmental Mixtures. Environ Forensics [Internet]. 2002;3:207–16. Available from: http://www.tandfonline.com/doi/abs/10.1080/713848382

60. Tolić N, Liu Y, Liyu A, Shen Y, Tfaily MM, Kujawinski EB, et al. Formularity: Software for Automated Formula Assignment of Natural and Other Organic Matter from Ultrahigh-Resolution Mass Spectra. Anal Chem. 2017;89(23): 12659–65.

61. Hu Q, Noll RJ, Li H, Makarov A, Hardman M, Cooks RG. The Orbitrap: A new mass spectrometer. J Mass Spectrom. 2005;40(4):430–43.

62. Cortés-Francisco N, Caixach J. Molecular Characterization of Dissolved Organic Matter Through a Desalination Process by High Resolution Mass Spectrometry. Environ Sci Technol [Internet]. 2013;47:9619–27. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23879562

63. Phungsai P, Kurisu F, Kasuga I, Furumai H. Molecular characterization of low molecular weight dissolved organic matter in water reclamation processes using Orbitrap mass spectrometry. Water Res [Internet]. 2016;100:526–36. Available from: http://dx.doi.org/10.1016/j.watres.2016.05.047

64. Mangal V, Stock NL, Guéguen C. Molecular characterization of phytoplankton dissolved organic matter (DOM) and sulfur components using high resolution Orbitrap mass spectrometry. Anal Bioanal Chem. 2016;408(7):1891–900.

65. Simon C, Roth V, Dittmar T, Gleixner G, Kaiser K. Molecular Signals of Heterogeneous Terrestrial Environments Identified in Dissolved Organic Matter : A Comparative Analysis of Orbitrap and Ion Cyclotron Resonance Mass Spectrometers. Front Earth Sci. 2018;6(September):1–16.

66. Pan Q, Zhuo X, He C, Zhang Y, Shi Q. Validation and Evaluation of High-Resolution Orbitrap Mass Spectrometry on Molecular Characterization of Dissolved Organic Matter. ACS Omega. 2020;5(10):5372–9.

67. Gavard R, Rossell D, Spencer SEF, Barrow MP. Themis: Batch Preprocessing for Ultrahigh-Resolution Mass Spectra of Complex Mixtures. Anal Chem [Internet]. 2017 Nov 7;89(21):11383–90. Available from: http://pubs.acs.org/doi/10.1021/acs.analchem.7b02345

68. Wünsch U, Hawkes J. Mathematical Chromatography Deciphers the Molecular Fingerprints of Dissolved Organic Matter. Analyst. 2020;

62

69. Brown TA, Jackson BA, Bythell BJ, Stenson AC. Benefits of multidimensional fractionation for the study and characterization of natural organic matter. J Chromatogr A [Internet]. 2016;1470:84–96. Available from: http://dx.doi.org/10.1016/j.chroma.2016.10.005

70. Koch BP, Ludwichowski K-UU, Kattner G, Dittmar T, Witt M. Advanced characterization of marine dissolved organic matter by combining reversed-phase liquid chromatography and FT-ICR-MS. Mar Chem [Internet]. 2008 Sep;111(3–4):233–41. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0304420308001059

71. Sandron S, Davies NW, Wilson R, Cardona AR, Haddad PR, Nesterenko PN, et al. Fractionation of Dissolved Organic Matter on Coupled Reversed-Phase Monolithic Columns and Characterisation Using Reversed-Phase Liquid Chromatography-High Resolution Mass Spectrometry. Chromatographia. 2017;80:1–11.

72. Petras D, Koester I, Da Silva R, Stephens BM, Haas AF, Nelson CE, et al. High-Resolution Liquid Chromatography Tandem Mass Spectrometry Enables Large Scale Molecular Characterization of Dissolved Organic Matter. Front Mar Sci [Internet]. 2017;4(December). Available from: http://journal.frontiersin.org/article/10.3389/fmars.2017.00405/full

73. Hawkes JA, Radoman N, Bergquist J, Wallin MB, Tranvik LJ, Löfgren S. Regional diversity of complex dissolved organic matter across forested hemiboreal headwater streams. Sci Rep. 2018;8(1).

74. Kim S, Kim D, Kim S, Son S, Jung MJ. Application of Online Liquid Chromatography 7 T FT-ICR Mass Spectrometer Equipped with Quadrupolar Detection for Analysis of Natural Organic Matter. Anal Chem. 2019;91(12):7690–7.

75. Vodacek A, Hogel FE, Swift RN, Yungel JK, Peltzer ET, Blough N V. The use of in situ and airborne fluorescence measurements to determine UV absorption coefficients and DOC concentrations in surface waters. Vol. 40, Limnology and Oceanography. 1995. p. 411–5.

76. Spencer RGM, Butler KD, Aiken GR. Dissolved organic carbon and chromophoric dissolved organic matter properties of rivers in the USA. J Geophys Res Biogeosciences. 2012;117(3).

77. Minor EC, Swenson MM, Mattson BM, Oyler AR. Structural characterization of dissolved organic matter: A review of current techniques for isolation and analysis. Environ Sci Process Impacts [Internet]. 2014;16:2064–79. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24668418

78. Li Y, Harir M, Lucio M, Kanawati B, Smirnov K, Flerus R, et al. Proposed Guidelines for Solid Phase Extraction of Suwannee River Dissolved Organic Matter. Anal Chem. 2016;88(13):6680–8.

79. Hertkorn N, Harir M, Cawley KM, Schmitt-Kopplin P, Jaff?? R. Molecular characterization of dissolved organic matter from subtropical wetlands: A comparative study through the analysis of optical properties, NMR and FTICR/MS. Biogeosciences. 2016;13(8):2257–77.

80. Raeke J, Lechtenfeld OJ, Wagner M, Herzsprung P, Reemtsma T. Selectivity of solid phase extraction of freshwater dissolved organic matter and its effect on ultrahigh resolution mass spectra. Environ Sci Process Impacts [Internet]. 2016;18(7):918–27. Available from: http://xlink.rsc.org/?DOI=C6EM00200E

81. Green NW, Perdue EM, Aiken GR, Butler KD, Chen H, Dittmar T, et al. An intercomparison of three methods for the large-scale isolation of oceanic dissolved organic matter. Mar Chem [Internet]. 2014;161:14–9. Available from: http://dx.doi.org/10.1016/j.marchem.2014.01.012

63

82. Li Y, Harir M, Lucio M, Gonsior M, Koch BP, Schmitt-Kopplin P, et al. Comprehensive structure-selective characterization of dissolved organic matter by reducing molecular complexity and increasing analytical dimensions. Water Res [Internet]. 2016;106:477–87. Available from: http://dx.doi.org/10.1016/j.watres.2016.10.034

83. Li Y, Harir M, Uhl J, Kanawati B, Lucio M, Smirnov KS, et al. How representative are dissolved organic matter (DOM) extracts? A comprehensive study of sorbent selectivity for DOM isolation. Water Res [Internet]. 2017;116:316–23. Available from: http://dx.doi.org/10.1016/j.watres.2017.03.038

84. Parlanti E, Morin B, Vacher L. Combined 3D-spectrofluorometry, high performance liquid chromatography and capillary electrophoresis for the characterization of dissolved organic matter in natural waters. Org Geochem. 2002;33(3):221–36.

85. Huber SA, Balz A, Abert M, Pronk W. Characterisation of aquatic humic and non-humic matter with size-exclusion chromatography - organic carbon detection - organic nitrogen detection (LC-OCD-OND). Water Res [Internet]. 2011;45(2):879–85. Available from: http://dx.doi.org/10.1016/j.watres.2010.09.023

86. Woods GC, Simpson MJ, Koerner PJ, Napoli A, Simpson AJ. HILIC-NMR: Toward the Identification of Individual Molecular Components in Dissolved Organic Matter. Environ Sci Technol [Internet]. 2011;45(9):3880–6. Available from: http://dx.doi.org/10.1021/es103425s

87. Brown TL, Rice JA. Effect of experimental parameters on the ESI FT-ICR mass spectrum of fulvic acid. Anal Chem. 2000;72(2):384–90.

88. Kujawinski EB, Hatcher PG, Freitas MA. High-resolution fourier transform ion cyclotron resonance mass spectrometry of humic and fulvic acids: Improvements and comparisons. Anal Chem. 2002;74(2):413–9.

89. Stenson AC, Marshall AG, Cooper WT. Exact masses and chemical formulas of individual Suwannee River fulvic acids from ultrahigh resolution electrospray ionization Fourier transform ion cyclotron resonance mass spectra. Anal Chem. 2003;75(6):1275–84.

90. Sleighter RL, Hatcher PG. Molecular characterization of dissolved organic matter (DOM) along a river to ocean transect of the lower Chesapeake Bay by ultrahigh resolution electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Mar Chem. 2008;110(3–4):140–52.

91. Sleighter RL, McKee GA, Hatcher PG. Direct Fourier transform mass spectral analysis of natural waters with low dissolved organic matter. Org Geochem. 2009;40(1):119–25.

92. Reemtsma T, These A. On-line coupling of size exclusion chromatography with electrospray ionization-tandem mass spectrometry for the analysis of aquatic fulvic and humic acids. Anal Chem. 2003;75(6):1500–7.

93. Dittmar T, Whitehead K, Minor EC, Koch BP. Tracing terrigenous dissolved organic matter and its photochemical decay in the ocean by using liquid chromatography/mass spectrometry. Mar Chem. 2007;107(3):378–87.

94. Reemtsma T, These a, Springer a, Linscheid M. Fulvic acids as transition state of organic matter: Indications from high resolution mass spectrometry. Environ Sci Technol [Internet]. 2006;40(19):5839–45. Available from: c:%5CDocuments and Settings%5CLeon%5CMy Documents%5CReferences%5CIn Library%5C#3200 Reemtsma.pdf

64

95. McDonough LK, O’Carroll DM, Meredith K, Andersen MS, Brügger C, Huang H, et al. Changes in groundwater dissolved organic matter character in a coastal sand aquifer due to rainfall recharge. Water Res. 2020;169.

96. Liu S, He Z, Tang Z, Liu L, Hou J, Li T, et al. Linking the molecular composition of autochthonous dissolved organic matter to source identification for freshwater lake ecosystems by combination of optical spectroscopy and FT-ICR-MS analysis. Sci Total Environ [Internet]. 2020;703:134764. Available from: https://doi.org/10.1016/j.scitotenv.2019.134764

97. Valle J, Harir M, Gonsior M, Enrich-Prast A, Schmitt-Kopplin P, Bastviken D, et al. Molecular differences between water column and sediment pore water SPE-DOM in ten Swedish boreal lakes. Water Res [Internet]. 2020;170:115320. Available from: https://doi.org/10.1016/j.watres.2019.115320

98. Gavard R, Jones HE, Palacio Lozano DC, Thomas MJ, Rossell D, Spencer SEF, et al. KairosMS: A New Solution for the Processing of Hyphenated Ultrahigh Resolution Mass Spectrometry Data. Anal Chem. 2020;92(5):3775–86.

99. Gross JH. Mass Spectrometry – A Textbook, 2nd ed.; Springer Heidelberg. Springer Heidelberg. 2011. Springer Heidelberg.

100. Comisarow MB, Marshall AG. Fourier transform ion cyclotron resonance spectroscopy. Chem Phys Lett. 1974;25(2):282–3.

101. Makarov A. Electrostatic axially harmonic orbital trapping: A high-performance technique of mass analysis. Anal Chem. 2000;72(6):1156–62.

102. Zubarev RA, Makarov A. Orbitrap mass spectrometry. Anal Chem. 2013;85(11):5288–96.

103. Scigelova M, Hornshaw M, Giannakopulos A, Makarov A. Fourier Transform Mass Spectrometry. Mol Cell Proteomics [Internet]. 2011;10(7):M111.009431. Available from: http://www.mcponline.org/lookup/doi/10.1074/mcp.M111.009431

104. Palacio Lozano DC, Gavard R, Arenas-Diaz JP, Thomas MJ, Stranz DD, Mejía-Ospino E, et al. Pushing the analytical limits: New insights into complex mixtures using mass spectra segments of constant ultrahigh resolving power. Chem Sci. 2019;10(29):6966–78.

105. Bae E, Yeo IJ, Jeong B, Shin Y, Shin KH, Kim S. Study of double bond equivalents and the numbers of carbon and oxygen atom distribution of dissolved organic matter with negative-mode FT-ICR MS. Anal Chem. 2011;83(11):4193–9.

106. Melendez-Perez JJ, Martínez-Mejía MJ, Awan AT, Fadini PS, Mozeto AA, Eberlin MN. Characterization and comparison of riverine, lacustrine, marine and estuarine dissolved organic matter by ultra-high resolution and accuracy Fourier transform mass spectrometry. Org Geochem [Internet]. 2016;101:99–107. Available from: http://dx.doi.org/10.1016/j.orggeochem.2016.08.005

107. Hertzog J, Carré V, Le Brech Y, Mackay CL, Dufour A, Mašek O, et al. Combination of electrospray ionization, atmospheric pressure photoionization and laser desorption ionization Fourier transform ion cyclotronic resonance mass spectrometry for the investigation of complex mixtures – Application to the petroleomic analysis. Anal Chim Acta [Internet]. 2017 [cited 2017 May 16];969:26–34. Available from: http://www.sciencedirect.com/science/article/pii/S0003267017303173

108. Boris P. Koch, * †, Thorsten Dittmar ‡, Matthias Witt § and, Kattner† G, Koch BP, Dittmar T, et al. Fundamentals of Molecular Formula Assignment to Ultrahigh Resolution Mass Data of Natural Organic Matter. Anal Chem. 2007;79(4):1758–63.

65

109. Stenson AC, Landing WM, Marshall AG, Cooper WT. Ionization and fragmentation of humic substances in electrospray ionization Fourier transform-ion cyclotron resonance mass spectrometry. Anal Chem. 2002;74(17):4397–409.

110. Zark M, Dittmar T. Universal molecular structures in natural dissolved organic matter. Nat Commun [Internet]. 2018;9(1):1–8. Available from: http://dx.doi.org/10.1038/s41467-018-05665-9

111. Artz RRE, Chapman SJ, Jean Robertson AH, Potts JM, Laggoun-Défarge F, Gogo S, et al. FTIR spectroscopy can be used as a screening tool for organic matter quality in regenerating cutover peatlands. Soil Biol Biochem. 2008;40(2):515–27.

112. Mann M, Meng CK, Fenn JB. Interpreting Mass Spectra of Multiply Charged Ions. Anal Chem. 1989;61(15):1702–8.

113. Kido Soule MC, Longnecker K, Giovannoni SJ, Kujawinski EB. Impact of instrument and experiment parameters on reproducibility of ultrahigh resolution ESI FT-ICR mass spectra of natural organic matter. Org Geochem [Internet]. 2010;41(8):725–33. Available from: http://dx.doi.org/10.1016/j.orggeochem.2010.05.017

114. Novotny NR, Capley EN, Stenson AC. Fact or artifact: the representativeness of ESI-MS for complex natural organic mixtures. J Mass Spectrom [Internet]. 2014;49(4):316–26. Available from: http://doi.wiley.com/10.1002/jms.3345

115. Taylor GFRS. Disintegration of water drops in an electric field. Proc R Soc L. 1964;280:383–97.

116. Mansoori BA, Volmer DA, Boyd RK. “Wrong-way-round” Electrospray Ionization of Amino Acids. Rapid Commun Mass Spectrom. 1997;11(10):1120–30.

117. Zhou S, Cook KD. Protonation in electrospray mass spectrometry: Wrong-way-round or right-way-round? J Am Soc Mass Spectrom. 2000;11(11):961–6.

118. Leito I, Herodes K, Huopolainen M, Virro K, Künnapas A, Kruve A, et al. Towards the electrospray ionization mass spectrometry ionization efficiency scale of organic compounds. Rapid Commun Mass Spectrom. 2008;22(3):379–84.

119. Kruve A, Kaupmees K, Liigand J, Leito I. Negative electrospray ionization via deprotonation: Predicting the ionization efficiency. Anal Chem. 2014;86(10):4822–30.

120. Oss M, Kruve A, Herodes K, Leito I. Electrospray ionization efficiency scale of organic compound. Anal Chem. 2010;82(7):2865–72.

121. Kostiainen R, Kauppila TJ. Effect of eluent on the ionization process in liquid chromatography-mass spectrometry. J Chromatogr A. 2009;1216(4):685–99.

122. McIntyre C. Electrospray mass spectrometry of groundwater organic acids. J mass Spectrom. 1997;32:328–30.

123. Reemtsma T. Determination of molecular formulas of natural organic matter molecules by (ultra-) high-resolution mass spectrometry. Status and needs. J Chromatogr A. 2009;1216(18):3687–701.

124. Piccolo A, Conte P. Molecular size of humic substance. Supramolecular associations versus macromolecular polymers. Adv Environ Res. 2000;3(4):508–21.

125. Leenheer JA, Rostad CE, Gates PM, Furlong ET, Ferrer I. Molecular resolution and fragmentation of fulvic acid by electrospray ionization/multistage tandem mass spectrometry. Anal Chem. 2001;73(7):1461–71.

126. Plancque G, Amekraz B, Moulin V, Toulhoat P, Moulin C. Molecular structure of fulvic acids by electrospray with quadrupole time-of-flight mass spectrometry. Rapid Commun Mass Spectrom. 2001;15(10):827–35.

66

127. These A, Reemtsma T. Limitations of Electrospray Ionization of Fulvic and Humic Acids as Visible from Size Exclusion Chromatography with Organic Carbon and Mass Spectrometric Detection. Anal Chem. 2003;75(22):6275–81.

128. Perdue EM, Ritchie JD. Dissolved Organic Matter in Freshwaters. In: Treatise on Geochemistry: Second Edition. 2013.

129. Leenheer JA, Wershaw RL, Brown GK, Reddy MM. Characterization and diagenesis of strong-acid carboxyl groups in humic substances. Appl Geochemistry. 2003;18(3):471–82.

130. Reuter H, Gensel J, Elvert M, Zak D. Direct Analysis of Lignin Phenols in Freshwater Dissolved Organic Matter. 2017;

131. Villacorte LO, Ekowati Y, Neu TR, Kleijn JM, Winters H, Amy G, et al. Characterisation of algal organic matter produced by bloom-forming marine and freshwater algae. Water Res [Internet]. 2015;73:216–30. Available from: http://dx.doi.org/10.1016/j.watres.2015.01.028

132. Hockaday WC, Purcell JM, Marshall AG, Baldock JA, Hatcher PG. Electrospray and photoionization mass spectrometry for the characterization of organic matter in natural waters: a qualitative assessment. Limnol Oceanogr Methods [Internet]. 2009;7(JAN.):81–95. Available from: https://www.scopus.com/inward/record.uri?eid=2-s2.0-62249171616&partnerID=40&md5=5508d90f96dc5eb923c9e1cc1889437a

133. Dixon RW, Peterson DS. Development and testing of a detection method for liquid chromatography based on aerosol charging. Anal Chem. 2002;74(13): 2930–7.

134. Gamache PH. Charged Aerosol Detection for Liquid Chromatography and Related Separation Techniques. Charged Aerosol Detection for Liquid Chromatography and Related Separation Techniques. 2017.

135. Vehovec T, Obreza A. Review of operating principle and applications of the charged aerosol detector. J Chromatogr A. 2010;1217(10):1549–56.

136. Gamache PH, McCarthy RS, Freeto SM, Asa DJ, Woodcock MJ, Laws K, et al. HPLC analysis of nonvolatile analytes using charged aerosol detection. LC-GC North Am. 2005;23(6 SUPPL.):101–6.

137. Fissan H, Neumann S, Trampe A, Pui DYH, Shin WG. Rationale and principle of an instrument measuring lung deposited nanoparticle surface area. J Nanoparticle Res. 2007;

138. Vervoort N, Daemen D, Török G. Performance evaluation of evaporative light scattering detection and charged aerosol detection in reversed phase liquid chromatography. J Chromatogr A. 2008;

139. Flagan RC. History of Electrical Aerosol Measurements. Aerosol Sci Technol. 1998;

140. Hutchinson JP, Li J, Farrell W, Groeber E, Szucs R, Dicinoski G, et al. Universal response model for a corona charged aerosol detector. J Chromatogr A [Internet]. 2010;1217(47):7418–27. Available from: http://dx.doi.org/10.1016/j.chroma.2010.09.056

141. Mathews BT, Higginson PD, Lyons R, Mitchell JC, Sach NW, Snowden MJ, et al. Improving quantitative measurements for the evaporative light scattering detector. Chromatographia. 2004;

142. Cobb Z, Shaw PN, Lloyd LL, Wrench N, Barrett DA. Evaporative light-scattering detection coupled to microcolumn liquid chromatography for the analysis of underivatized amino acids: Sensitivity, linearity of response and comparisons with UV absorbance detection. J Microcolumn Sep. 2001;

67

143. Górecki T, Lynen F, Szucs R, Sandra P. Universal response in liquid chromatography using charged aerosol detection. Anal Chem. 2006;78(9):3186–92.

144. Beynon JH, Brown L. Mass spectrometry [Internet]. Encyclopædia Britannica. 2017. Available from: https://www.britannica.com/science/mass-spectrometry

145. Stoll N, Schmidt E, Thurow K. Isotope Pattern Evaluation for the Reduction of Elemental Compositions Assigned to High-Resolution Mass Spectral Data from Electrospray Ionization Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. J Am Soc Mass Spectrom. 2006;17(12):1692–9.

146. van Krevelen DW. Graphical-statistical method for the study of structure and reaction processes of coal. Fuel. 1950;29:269–84.

147. Rice JA, MacCarthy P. Statistical evaluation of the elemental composition of humic substances. Org Geochem. 1991;17(5):635–48.

148. Wu Z, Rodgers RP, Marshall AG. Two- and Three-Dimensional van Krevelen Diagrams: A Graphical Analysis Complementary to the Kendrick Mass Plot for Sorting Elemental Compositions of Complex Organic Mixtures Based on Ultrahigh-Resolution Broadband Fourier Transform Ion Cyclotron Resonance. Anal Chem. 2004;76(9):2511–6.

149. Kim S, Kramer RW, Hatcher PG. Graphical Method for Analysis of Ultrahigh-Resolution Broadband Mass Spectra of Natural Organic Matter, the Van Krevelen Diagram. Anal Chem. 2003;75(20):5336–44.

150. Ohno T, He Z, Sleighter RL, Honeycutt CW, Hatcher PG. Ultrahigh Resolution Mass Spectrometry and Indicator Species Analysis to Identify Marker Components of Soil- and Plant Biomass-Derived Organic Matter Fractions. Environ Sci Technol [Internet]. 2010;44(22):8594–600. Available from: http://dx.doi.org/10.1021/es101089t

151. D’Andrilli J, Cooper WT, Foreman CM, Marshall AG. An ultrahigh-resolution mass spectrometry index to estimate natural organic matter lability. Rapid Commun Mass Spectrom. 2015;29(24):2385–401.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1948

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-410266

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2020