MSc Chemistry
Analytical Sciences
Master Thesis
The use of comprehensive two-dimensional liquid
chromatography to characterize complex mixtures in
crop protection formulations
by
J.D. Kruijswijk
11856864 (UvA)
2628251 (VU)
November 2019
48 EC
Supervisor/Examiner:
Examiner:
dr. Bob Pirok
prof. Peter Schoenmakers
I
Acknowledgements
This thesis serves as a requirement to fulfill the research project of the master’s degree for the study Chemistry, Analytical Sciences at the University of Amsterdam and VU University Amsterdam. The research was performed at the University of Amsterdam in the department of van ‘t Hoff Institute for Molecular Sciences, where I was supervised by dr. Bob Pirok and Prof. Peter Schoenmakers. I would like to thank both for their advice and for guidance. The latter part of the project was finalized at Life Scientific Ltd. located in Dublin, Ireland. I would like to express my gratitude to my supervisor Alan Ayling who supported me throughout my time abroad. Also, I would like to thank all my (international) colleagues for their patience and time, when I needed them; they were always ready to help. Thanks again for the informative and pleasant cooperation.
Lastly, I would like to acknowledge COAST for providing the opportunity to get in contact with Life Scientific.
Jordy Kruijswijk De Meern,
II
Samenvatting
De analyse van surfactant mengsels worden frequent uitgevoerd met een ion-pair reversed-phase vloeistofchromatografie methode. Veel gebruikte stoffen in gewasbeschermings-formuleringen zijn lignosulfonaten (Kraft lignine), naftaleensulfonzuren, en gesulfoneerde naftaleen-formaldehyde condensaten (SNFC). De mengsels zijn moeilijk te scheiden met eendimensionale vloeistofchromatografie vanwege de complexiteit van de mengsels en een limiterende piekcapaciteit van de techniek. Het gebruik van tweedimensionale vloeistof-chromatografie is om deze redenen onderzocht. Verschillende eendimensionale methoden zijn bestudeerd, namelijk RPLC, IPRPLC, HILIC, SEC en mixed-modes van IEX. Deze zijn beoordeeld op orthogonaliteit en compatibiliteit voor het gebruik in een tweedimensionale opstelling. Een ion-pair methode met tetrabutylammonium als ion-pair reagens is verkozen om te gebruiken voor de tweede dimensie, vanwege de flexibele compatibiliteit en snelle analysetijden. De volgende selectiviteiten en kolommen zijn geëvalueerd en succesvol gekoppeld met de ion-pair methode: Biphenyl, Surfactant Plus, Obelisc N, en aqueous SEC. Er vindt geen scheiding plaats op basis van size-exclusion mechanismen in de aqueous SEC methode. De koppeling van Biphenyl×IPRPLC produceerde een chromatogram met zichtbaar van elkaar gescheiden groepen, echter de methanolrijke eerste dimensie effluent zorgde voor grote pieken in tegendruk. In de Surfactant Plus×IPRPLC methode resulteerde de scheiding in groepen met de lignosulfonaten grotendeels gescheiden van de naftaleensulfonzuren en SNFC. De Obelisc N×IPRPLC methode kon de naftaleensulfonzuren van naftaleendisulfonzuren onderscheiden, met oplopende lengte van de alkylgroep. Een moeilijk interpreteerbare en onordelijke chromatogram werd verkregen met de aqueous SEC×IPRPLC methode. Een verplaatsende gradiënt werd gebruikt om de scheidingen te optimaliseren. Dit maakte het mogelijk om de scheidingsoppervlakte beter te benutten. Tevens waren de verschillende groepen behouden en was er meer scheiding binnen de groepen. Daarnaast was stationaire-fase geassisteerde modulatie kort onderzocht in combinatie met de Surfactant Plus×IPRPLC methode. De hoge intensiteit pieken werden gefocust, terwijl andere een lagere intensiteit kregen door verbreding van de pieken. Het is belangrijk om aan te geven dat volledig geoptimaliseerde verplaatsende gradiënten en actieve modulatietechnieken eventueel niet flexibel genoeg zijn om te gebruiken voor onbekende monsters.
III
Summary
The analysis of surfactant mixtures is frequently performed by ion-pair reversed-phase liquid chromatography. Regularly used components in crop protection formulations, and the compounds of interest, are lignosulfonates (Kraft Lignin), naphthalenesulfonic acids, and sulfonated naphthalene-formaldehyde condensates (SNFC). However, the mixtures are often insufficiently separated due to the complexity of the sample constituents and the limited peak capacity offered by one-dimensional liquid chromatography separations. Therefore, the use of two-dimensional chromatography was investigated. Several one-dimensional methods were studied, such as RPLC, IPRPLC, HILIC, SEC, and mixed-modes of IEX. These were assessed for their orthogonality and compatibility in a two-dimensional setup. An ion-pair method, using tetrabutylammonium as ion-pair reagent, became the second dimension of choice, due to its easy compatibility and fast analysis times. The following selectivities and columns have been successfully coupled to the ion-pair method: Biphenyl, Surfactant Plus, Obelisc N, and aqueous SEC. The aqueous SEC method was not fully governed by size-exclusion mechanisms. The Biphenyl×IPRPLC method was able to produce a chromatogram containing group-type information, although it suffered from pressure spikes induced by the methanol-rich first-dimension effluent. By employing Surfactant Plus×IPRPLC, group-type separation was achieved, with the lignosulfonates mostly separated from the naphthalenesulfonic acids and SNFC. The Obelisc N×IPRPLC method was able to distinguish the naphthalenesulfonic acids from the naphthalenedisulfonic acids, in series of increasing alkyl-chain length. A difficult to interpret and disordered chromatogram was obtained for the aqueous SEC×IPRPLC method. Moreover, shifting gradients were explored to optimize the separations. The shifting gradient was able to increase the coverage of the separation space, improve the separation within series, and maintain the group-type information. Additionally, stationary-phase-assisted modulation was applied to a Surfactant Plus×IPRPLC method. The high intensity peaks became more focused, while some diminished in intensity due to broadening of the peaks. It should be noted that fully optimized shifting gradients and active modulation techniques might not be flexible enough for unknown samples.
IV
Abbreviations and symbols
Abbreviations & Symbols Definition
1D-LC One-dimensional liquid chromatography
2D-LC Two-dimensional liquid chromatography
AEX Anion-exchange chromatography
ASM Active solvent modulation
C18 Octadecylsilane
C8 Octylsilane
CEX Cation-exchange chromatography
DMSO Dimethyl sulfoxide
GFC Gel filtration chromatography
GPC Gel permeation chromatography
HILIC Hydrophilic interaction chromatography
HPLC High-performance liquid chromatography
IEX Ion-exchange chromatography
IP Ion-pair reagent
IPRPLC Ion-pair reversed-phase chromatography
LC Liquid chromatography
LC×LC Comprehensive two-dimensional liquid chromatography
LC-LC Heart-cutting two-dimensional liquid chromatography
MM Mixed-mode chromatography
NPLC Normal-phase liquid chromatography
RPLC Reversed-phase liquid chromatography
SEC Size-exclusion chromatography
sLC-LC Selective comprehensive LC×LC
SNFC Sulfonated naphthalene-formaldehyde condensate
SPAM Stationary-phase-assisted modulation
TBAOH Tetrabutylammonium hydroxide
TBAHSO4 Tetrabutylammonium hydrogensulfate
TEA Triethylamine
THF Tetrahydrofuran
TMAOH Tetramethylammonium hydroxide
V
Table of Content
Acknowledgements ...I Samenvatting ... II Summary ... III Abbreviations and symbols ... IV
1. Introduction ... 1
2. Theoretical Framework ... 3
2.1. High-performance liquid chromatography ... 3
2.1.1. Reversed-phase liquid chromatography ... 3
2.1.2. Ion-pair reversed-phase liquid chromatography ... 4
2.1.3. Hydrophilic interaction chromatography ... 4
2.1.4. Size-exclusion chromatography ... 5
2.1.5. Ion-exchange chromatography ... 6
2.2. Two-dimensional liquid chromatography ... 6
2.2.1. (Non-) comprehensive two-dimensional liquid chromatography ... 7
2.2.2. Passive modulation ... 8 2.2.3. Active modulation ... 9 2.2.4. Shifting gradients in LC×LC ... 10 3. Experimental setup ... 12 3.1. Chemicals ... 12 3.2. Equipment ... 12 3.3. Sample preparation ... 14
3.4. Two-dimensional data visualization ... 14
4. Results and Discussion ... 15
4.1. One-dimensional separations ... 15
4.1.1. Reversed-phase liquid chromatography ... 15
4.1.2. Ion-pair reversed-phase liquid chromatography ... 21
4.1.3. Hydrophilic interaction chromatography ... 26
4.1.4. Size-exclusion chromatography ... 28
4.1.5. Mixed-mode reversed-phase ion-exchange chromatography ... 30
4.2. Comprehensive two-dimensional liquid chromatography ... 32
4.2.1. IPRPLC×RPLC ... 32
VI
4.2.3. Surfactant Plus×IPRPLC ... 36
4.2.4. Obelisc N×IPRPLC ... 38
4.2.5. SEC×IPRPLC ... 42
4.3. LC×LC with shifting gradient ... 44
4.3.1. Surfactant Plus×IPRPLC ... 44 4.3.2. Obelisc N×IPRPLC ... 45 4.3.3. SEC×IPRPLC ... 46 4.4. Active modulation in LC×LC ... 47 5. Conclusion ... 49 6. Recommendation ... 52 References ... 53 Appendix ... 57 Appendix 1 ... 57 Appendix 2 ... 59 Appendix 3 ... 61 Appendix 4 ... 65 Appendix 5 ... 68 Appendix 6 ... 70
1
1. Introduction
Agrochemical formulations are widely used for the protection of crops, i.e. functioning as herbicides, insecticides, and/or fungicides. The crop protection products contain complex mixtures of surfactants, dispersant and wetting agents, to achieve a stable water-based spray solution ready to be applied in the field. Often encountered additives are, naphthalenesulfonic acids, sulfonated naphthalene-formaldehyde condensates (SNFC) and Kraft lignin—a water-soluble lignosulfonate. Dispersing agents are also used in many other industrial applications, e.g. tanning agents [1] and as plasticizers in cement [2]. The molecular composition of the dispersing agents defines the physical properties of the end product. Analytical methods are required to understand their chemistry, manufacture, and control.
One common analysis technique for identifying the various components in dispersing agents is liquid chromatography (LC). A frequently cited method for SNFC analysis is described by Wolf et al., involving an ion-pair reversed-phase liquid chromatography (IPRPLC) analysis [3]. However, the complexity and structure of dispersing agents renders most eluting species insufficiently separated by the limited peak capacity offered by one-dimensional separations. To improve peak capacity, and in turn separation power, two-dimensional liquid chromatography (2D-LC) is a valid candidate for evaluation. However, for successful implementation of 2D-LC, both retention mechanisms must be orthogonal and compatible. Two main types exist, namely: heart-cutting two-dimensional chromatography (LC-LC), and comprehensive two-dimensional liquid chromatography (LC×LC). The key difference is the number of one-dimensional fractions that are sent to and separated by the second dimension. A more in-depth description of 2D-LC will be given in Chapter 2.2.
The aim of the project is to evaluate the feasibility of LC×LC for the characterization of dispersing agents. The samples of interest are commercial products available on the market. Initially, first-dimensional separations are explored for orthogonal retention mechanisms. Additionally, the compatibility of the two selected techniques will be assessed. The coupling of the two phases will be performed by passive modulation. Active modulation techniques are briefly explored for a more optimal separation. Overall, the research might demonstrate if the LC×LC equipment will have an added value in the portfolio of Life Scientific for the separation and characterization of competitor products.
The project was performed at two locations. The first period was carried out within the van ‘t Hoff Institute for Molecular Sciences at the University of Amsterdam, the Netherlands, in the group of Prof. Peter Schoenmakers. Next, the developed methods were transferred to Life Scientific Ltd. based in Dublin, Ireland. Further research was performed on-site to improve the separation the commercial mixtures.
This thesis summarizes the results obtained during the master’s research project. Initially, the theoretical framework will be introduced, providing an overview of the retention mechanism in various modes of HPLC, followed by the types of 2D-LC, passive and active modulation, and shifting gradients. Secondly, the experimental setup will define the utilized chemicals and equipment. Next, the results and discussion of the evaluated first-dimension
2 separations, the initial two-dimensional separations, the active modulated 2D-LC, and the shifting gradient 2D-LC will be presented. Finally, the work will be finished by summarizing the concluding remarks of, and recommendations for the research.
3
2. Theoretical Framework
2.1. High-performance liquid chromatography
Chromatography refers to the separation by means of a physical method, in which analytes of a mixture are separated by a distribution between a stationary and a moving phase. A range of categories are employed within chromatography, in which high-performance liquid chromatography (HPLC) is a subdivision of column chromatography [4]. By applying pressure, a solvent (mobile phase) is pushed through a packed column. The packing consists of porous stationary phase, commonly made of silica particles with specific, covalently attached groups. Separation involves injection of a mixture into the mobile phase, which is introduced into the column. As such, the components in the mixture are able to interact with the stationary phase, either chemically or physically. Each molecule interacts differently with the stationary phase, resulting in differences in flow rates for different molecules. Therefore, the molecules elute from the column at different retention times and the mixture becomes separated. To change the selectivity of the separation, the stationary phase can be changed, or more specifically, the attached groups can be tailored to the desired selectivity. As such, different modes of liquid chromatography exist, including: normal-phase liquid chromatography (NPLC), reversed-phase liquid chromatography (RPLC), hydrophilic interaction chromatography (HILIC), size-exclusion chromatography (SEC), ion-exchange chromatography (IEX) [5]. The modes are defined based on the nature of interactions, which may result from hydrogen bonding, van der Waals forces, electrostatic forces, hydrophobic forces, or exclusion by size. Recent advances in column and stationary phase technology have made ultrahigh-performance chromatography (UHPLC) possible, though the same principles of separation still apply [6,7].
Crop protection formulations often contain sulfonated naphthalene-formaldehyde condensates, and the following properties are present: a hydrophobic naphthyl-type backbone, charged sulfonic groups, and a size distribution. Additionally, the mixtures are water-soluble dispersants, meaning hydrophilic properties. The previously mentioned modes of chromatography can provide the appropriate selectivity for separation and will be described in the next sections.
2.1.1. Reversed-phase liquid chromatography
RPLC is accounted as one of the most versatile and widely used technique in HPLC [8,9]. The technique is often applied to analytes containing hydrophobic moieties, which are present in many molecules. In RPLC, a nonpolar stationary phase and a polar mobile phase is utilized. These are typically composed of a mixture of water and a water-miscible organic solvent. Separation is achieved by differences in hydrophobicity of the components in a mixture [10]. Retention occurs by the partitioning of the analytes between the two phases while progressing through the column [11]. Since the stationary phase is predominantly hydrophobic, the dissolved hydrophobic analytes present in the mobile phase have a high affinity for the stationary phase. In contrast, hydrophilic analytes, which have a lower affinity, will have fewer interactions and less retention. As such, the hydrophilic analytes will elute first, followed by the hydrophobic ones, leading to separation. Typical RPLC mobile
4 phases are composed of a mixture of water and acetonitrile, methanol or other organic solvents, also called the organic modifier. Characteristically in RPLC, apolar solvents have a high eluotropic strength, which is an indication of the ability of eluting strongly retained analytes from the column [12]. Separations can be performed by either isocratic or gradient elution. The mobile-phase composition is kept constant throughout the analysis for isocratic elution, while the composition is changing in gradient elution. In gradients, the eluotropic strength is initially weak and becomes stronger by increasing the modifier during the analysis. The previously described mechanism of retention applies for both types of elution. Although, for gradients the affinities change throughout the analysis, so that sufficient retention occurs of weakly retained analytes, while the strongly retained analytes elute more quickly. The selectivity of separation can be adjusted by changing the mobile phase composition or by different stationary phases [4,12]. The most commonly used stationary phases are alkyl-chain silica-bonded based, such as octadecylsilane (C18) and octylsilane (C8). Alternatively, other types of stationary phases are available, which provide a different selectivity in separation by allowing additional retention mechanisms besides hydrophobic interactions. For example, an (alkyl-)phenyl stationary phase has the possibility of π-π interactions. This makes it a suitable candidate for the separation of aromatic-moiety containing compounds, such as the ones found in the studied crop protection formulations. It is important to note that the choice of stationary phase can affect mobile phase selection. In the case of phenyl-type stationary phases the π-π interactions yield the selectivity based on aromaticity. As a result, these interactions must not be suppressed; acetonitrile is an unfavorable solvent, as the unsaturated triple bond interacts with the stationary phase, effectively shielding it from the analyte [13,14]. Instead, methanol is often employed in phenyl-type stationary phases, without the π-π interactions inhibiting properties.
2.1.2. Ion-pair reversed-phase liquid chromatography
Separation of ionic or ionizable compounds is challenging by conventional RPLC, as the charges cause a highly hydrophilic character, resulting in little or no retention on the hydrophobic stationary phase. However, by modifying the mobile phase, regular RPLC solvents and columns can be employed [15,16]. This modification involves the addition of an ion-pair reagent (IP) in the mobile phase. IPs are ionic compounds of the opposite charge as the analyte, while also containing hydrophobic moieties [4]. Typical IPs are alkyl ammonium salts for negatively charged ionic/ionizable compounds [9,17]. For positively charged compounds, alkyl sulfonate and alkyl sulfate salts are used. It is essential that the IP has both a hydrophobic moiety for hydrophobic interactions with the RPLC stationary phase, and an ionic group for the formation of analyte – ion-pair complexes, causing retention of the analyte. So, addition of IP to the mobile phase allows for interaction of the analytes with the hydrophobic stationary phase, thus increasing retention. Therefore, separation of charged, highly polar analytes is possible by using RPLC columns. The crop protection formulations contain permanently charged sulfonic groups. Accordingly, IPRPLC can provide the required selectivity for the separation.
2.1.3. Hydrophilic interaction chromatography
Historically, HILIC is seen as a variant of NPLC, as both use polar stationary phases. However, HILIC uses aqueous-organic mobile phases, compared to nonpolar ones in NPLC. Compared to RPLC without addition of ion-pair reagents and IEX (see Paragraph 2.1.5), it is
5 able to separate compounds that are too polar for RPLC and insufficiently charged for IEX [4,18]. The mechanism of retention is more complex than in NPLC and RPLC, depending on a combination of interactions. HILIC separations are often performed using gradient elution, where water is the solvent of high eluotropic strength [9]. HILIC initially employs a mobile phase consisting of organic solvent and a small percentage of water (minimum of 3%). After a lengthy and time-consuming equilibration, a water-rich, immobilized mobile phase layer on the surface of the polar stationary phase is formed. In combination with the water-deficient mobile phase, a type of liquid-liquid extraction interface is created. Separation is often described as the partitioning of the analytes between the layers. The partitioning equilibrium of hydrophilic compounds is shifted towards the immobilized water layer, hence retaining the compound. Consequently, less hydrophilic compounds elute first, followed by the more hydrophilic ones. The partitioning equilibrium changes throughout gradient elution, more water present in the mobile phase causes a shift towards the mobile phase. This shift makes the more hydrophilic compounds elute in an acceptable timeframe. At the end of the gradient, the surplus amount of water breaks down the immobilized water layer on the surface, which elutes most species. Prior to another analysis, the water layer must be reformed by equilibration. Interactions in HILIC are more complex than just the previously described hydrophilic partitioning, as surface adsorption, electrostatic interactions, and others also affect the retention mechanism [19]. Although the mechanism is not fully described, by empirical means it is still utilized as an effective and often used method of separation.
2.1.4. Size-exclusion chromatography
Two main types are regularly distinguished in SEC, these are gel filtration chromatography (GFC) and gel permeation chromatography (GPC) [8,20]. In general, both techniques are applied for separation based on size, or hydrodynamic volume, and are performed primarily on macromolecules, e.g. (bio-) polymers. The main difference between the two is the applied solvent [4,21]. GFC uses predominantly water, or aqueous solvents, while GPC uses organic solvents. Fundamentally, the separation mechanism is not dependent on the type of mobile phase, so the techniques are considered and grouped as one, namely SEC. The column packings in SEC are porous, with carefully controlled pore sizes, and separation is attained by the degree of access of the macromolecules into the pores. In SEC, the stationary phase is described as the stagnant mobile-phase solvent present in the pore volume of the column packing [22]. Additionally, the mobile phase flows through the interstitial volume of the column, acting as the mobile phase. For the macromolecules a distribution coefficient is achieved, which describes the time spent in the pores depending on solvated size of the macromolecules and pore distribution [23]. Small molecules, such as the solvent, can fully enter and access the available pore volume. Upon partitioning to the pore volume, or stationary phase, the small molecules become the latest to elute. Large molecules are excluded from the pore volume, as it is inaccessible, causing them to elute first. Macromolecules that are completely excluded from the pore volume will elute at the exclusion limit. Small molecules, typically the smallest entering the system is the solvent, are able to fully permeate the pores and elute last, corresponding to the permeation limit or void volume. Accordingly, all analytes are eluted in one column volume. The mechanism of separation assumes that the column is inert, indicating that no interactions (e.g. adsorption or partitioning) between the analytes and stationary phase occur that would cause retention.
6 If analytes elute after the permeation limit, other retention mechanisms have occurred. To prevent interactions, strong eluotropic solvents are employed. For example, often-used porous columns are modified with C18 and by using THF, most (hydrophobic) interactions with the stationary phase are severely weakened.
2.1.5. Ion-exchange chromatography
Another mode of chromatography is IEX; it allows for the separation of ionic and polar compounds based on charge [8,9]. The type of separation is complementary to RPLC and similar to IPRPLC, while using a different retention mechanism. IEX separation is caused by reversible electrostatic interactions between the compound and stationary phase. Therefore, it requires the use of stationary-phase moieties that are either permanently or temporarily ionized. IEX can be divided in two types: anion-exchange (AEX) or cation-exchange chromatography (CEX) [4]. Columns in IEX use ion-exchange stationary phases that contain covalently bonded basic, positively charged (cationic) groups for AEX, e.g. quaternary or tertiary amines. CEX stationary phases consist of acidic, negatively (anionic) charged groups, e.g. sulfonic or carboxylic groups. Present in the mobile phase, the counter-ions for the basic groups in AEX are most anions (OH-, Cl- and CO
3-) and for the acidic groups in CEX
it can be most cations (H+, Na+ and K+). The counter ions are reversibly retained by the
charged stationary phase. Retention of the compound is by ionic equilibrium between the compound and stationary phase, and separation is caused by the differences in the affinities to the stationary phase [24]. For retention to occur upon sample injection, the compounds must have a higher affinity for electrostatic interactions with the column than the counter-ions present on the stationary phase. Elution is performed using a gradient of either increasing ionic strength or changing the pH, causing a distinction in another two types of IEX, namely weak and strong IEX [25]. The two vary in the type of stationary phase; permanently charged phases are used for strong IEX, as mobile phase pH does not alter the charge of the column. However, temporarily ionized phases in weak IEX in which the charge does depend on pH of the mobile phase. A higher ionic strength, containing a high concentration of competing counter ions, shifts the ionic equilibrium of the analyte towards the mobile phase, causing the compounds to elute. By changing the pH, one can alter either the weak IEX column or the analyte charge. This causes a shift in the ionic equilibrium of the retained compounds, which elutes the compounds.
2.2. Two-dimensional liquid chromatography
The previously described separation mechanisms are typically performed in one-dimensional liquid chromatography (1D-LC). In 1D-LC, one chromatographic separation is achieved using a specific stationary phase selectivity [26,27]. In academic and industrial research, the objective was to separate the largest number of analytes from a complex mixture. Ideally, by using lower analysis times, having fewer solutes co-elution, and by increasing the separation power. Complex mixtures often originate from the field of life sciences, and environmental, food and medical industries. For these samples, full separation and characterization by one-dimensional separations alone is unlikely, as a high separation power is needed. One approach to quantify separation power is by peak capacity, i.e. the maximum theoretical number of single-analyte peaks that can be placed side by side in the available separation space, using defined column and analytical conditions. Peak capacity is
7 connected to the complexity of the mixture regarding to the number of analytes and the number of different chemical families that show distinctive selectivities to the selected stationary phase. This model was described as sample dimensionality, a concept introduced by Giddings [28]. Fundamentally, the concept states that for a specific sample, multiple sample dimensions can be described. The characteristics of the dimension can be highly defined allowing structure-retention relations, causing ordered separation. Orderly separation uses the separation space more effectively, resulting in a higher effective peak capacity.
As the sample complexity increases, due to a significant number of analytes distributed in a range of concentrations, the separation power of one-dimensional separations becomes insufficient to suitably separate the peaks in a tolerable analysis time [8,29]. The peak capacity can be significantly increased by employing two-dimensional separations (2D-LC). Here, the first dimension is coupled to a second dimension by means of modulation. The first-dimension effluent is sampled and injected onto the second dimension for another separation. After the first-dimension separation is completely sampled by the second dimension, a two-dimensional chromatogram can be created by rearranging each subsequent second-dimension chromatogram in a sequential manner using appropriate software. As a result, a three-dimensional separation space is formed, two dimensions of time and one dimension of signal intensity, generating the separation space. The separation space is often visualized as a two-dimensional surface, using contours or colors to distinguish signal intensity. Paragraph 2.2.1 describes the types of 2D-LC and the experimental setup in more detail. Recalling Giddings’ model, by maximizing the coverage of the two-dimensional separation space, larger numbers of analytes can be separated [30]. To fully utilize the separation space, the concept of orthogonality of the dimensions is important. A high degree of orthogonality indicates that the dimensions are independent; the separation selectivities are predominantly uncorrelated for the complete mixture. If complete orthogonality is attained, the complete separation space is utilized. Theoretically, a total peak capacity is reached that is the product of the maximum peak capacities in each dimension. For example, if the maximum peak capacity in both dimensions is a relatively low 50, the two-dimensional separation peak capacity is 50×50=2500, which is a drastic increase compared to one-dimensional separations. Nevertheless, complete orthogonality of two dimensions is practically unattainable, causing two-dimensional separations to be partially non-orthogonal. Accordingly, lower effective two-dimensional peak capacities are obtained. To quantify orthogonality, several methods are described that provide useful estimations: for example, by the bin counting method [31], %BIN [32], %FIT [33], and the asterisk equation [34].
2.2.1. (Non-) comprehensive two-dimensional liquid chromatography
A distinction is made in the approaches of 2D-LC, namely a heart-cutting or a comprehensive method [30]. In heart-cutting systems (LC-LC), the first-dimension separation is not sampled completely, merely one fraction, the convoluted peak of interest, is sent to the second dimension. In turn, the second dimension separates the fraction using an orthogonal retention mechanism. Therefore, additional resolution is obtained for the previously convoluted peak. Two chromatograms are acquired; the first one contains all the compounds and accounts for the full run. The second one is obtained upon injection of the
8 collected fraction and shows the separation of the previously convoluted peak present in the first dimension. LC-LC is often utilized for the quantitative determination of targets or for impurity analysis [35]. However, this approach is not employed for the significantly increased peak capacities needed for complex mixtures. In fully comprehensive two-dimensional methods (LC×LC), the entire first dimension is fractionated (off- or online) and injected onto the second dimension. Therefore, the entire mixture is separated by two different retention mechanisms. Instead of obtaining one or two chromatograms, a long continuous chromatogram is obtained, containing all analyzed fractions one after the other. For data visualization, the chromatogram is cut according to the fraction size; the chromatograms are laid side by side, creating a two-dimensional surface. To clearly define comprehensive 2D-LC, or LC×LC, from other two-dimensional separations, three requirements are determined [36]. First, the total sample is subjected to both separation mechanisms. The first dimension is divided in equal fractions; a set percentage (100% or lower) of each fraction is injected onto the second dimension. Lastly, the obtained first-dimension separation is maintained in the second first-dimension. Besides heart-cutting, other non-comprehensive approaches exist, such as selective comprehensive LC×LC (sLC×LC) [37] and multiple heart-cutting LC-LC [38,39].
2.2.2. Passive modulation
To successfully transfer the first-dimension effluent into the second dimension, some sort of interface is needed. Here, only the online mode of coupling the two dimensions will be discussed. Figure 1 displays a schematic/diagram of a modulation interface used in Agilent systems. The loops shown have a predefined volume, typically between 20 and 60 µL. The form of modulation described here is referred to as passive modulation. The loops define the first-dimension effluent volume that is fractionated, resulting in the second-dimension injection volume. Typically, due to the parabolic flow profile of the mobile phase, the recommendation is to use modulation volumes twice the size of the first-dimension effluent fraction to avoid sample losses [40]. In position 1, the effluent of the first dimension follows the blue lines and loads the left loop before going to waste. Meanwhile, a previously collected fraction in the right loop is injected onto the second dimension by the second-dimension eluent. The valve is switched to position 2 before anything from the loop connected to the first dimension is lost to waste. Now the first dimension fills the right loop and the other loop volume is injected onto the second dimension.
It is important to remember the following significant parameters for modulation: column dimensions, modulation time, and solvent compatibility [30,40]. Generally, the first-dimension separation is slow, so that the modulation volume can be transferred, and thus sampled, multiple times by a fast second-dimension separation. For the first-dimension separation to be sufficiently slow, the column is long and narrow. This results in high efficiencies at low flow rates and reasonable linear flow velocities. As such, low modulation volumes are obtained, diminishing the risk of overloading the second dimension. The second dimension is required to be fast, so that frequent sampling of the first dimension can occur. Consequently, the second-dimension column is short and wide, allowing fast separations and short modulation times. If the modulation time is too slow, the undersampling effect can occur [41]. Undersampling is a result from a low sampling rate of the first dimension, causing a loss of resolution in the first dimension. Previously resolved peaks enter the
9 second dimension as one peak. Therefore, it is important to avoid remixing of resolved components from the first-dimension separation, whereas using the best possible time available for the second-dimension separation.
Figure 1 Schematic of switching valve for coupling the first- and second-dimension column (Figure is adapted from Agilent OpenLAB CDS program).
Solvent incompatibility is another issue encountered in LC×LC, caused by either the immiscibility of solvents or the differences in mobile phase eluotropic strengths between the dimensions [40]. The difference in eluotropic strength of the solvents can distort the peak shape or separation in the second dimension, by a process called the breakthrough phenomenon [42]. When the first-dimension fraction has a higher solvent strength than the second-dimension mobile phase, the analytes will remain in the injected fraction throughout the second dimension. By means of diffusion/dispersion, a portion will be retained and separated according to the second-dimension retention mechanism. Effectively, two peaks are obtained, one near the dead volume and the other at a position under normal retention conditions. Breakthrough can occur in various LC×LC combinations, e.g. when SEC is combined with RPLC, the THF mobile phase is a highly eluotropic solvent in RPLC. Moreover, the acetonitrile in HILIC can cause the same phenomenon when coupled to RPLC. Breakthrough effects can be reduced by lowering the modulation volume or by active modulation.
2.2.3. Active modulation
In the previously described loop-based modulation interface of passive modulation, the full fractions of the first-dimension effluent are injected onto the second dimension. The concentration of the analytes or the volume of the effluent are unchanged. As such, it can be sensitive to breakthrough, due to solvent incompatibilities. Instead, an active modulation method can be employed that modifies the first-dimension effluent prior to injection. One developed approach is the so-called active solvent modulation (ASM) [43]. In ASM, a dilution solvent is added directly after the first-dimension separation and before the
10 modulation loop. The dilution weakens the strength of the first-dimension solvent, increasing the retention of the analytes in the second dimension. Therefore, a focusing occurs at the column head, improving peak shapes and total peak capacity. An alternative is stationary-phase-assisted modulation (SPAM), which replaces the empty loops in passive modulation with trap columns [44]. The trap columns serve to trap and concentrate the analytes in the first-dimension effluent. Therefore, the injection volume can be reduced, simultaneously affecting the solvent to more optimal conditions for the second dimension. Overall, by using SPAM reduced dilution of the analyte is observed in a lower analysis time [45]. Analytes that are weakly retained by the trap columns might be lost to the waste. Active modulation is effective for reducing the difficulties originating from solvent incompatibility. Moreover, a reduction in dilution—inherently present in HPLC—is achieved, enhancing the peak shape and total peak capacity.
2.2.4. Shifting gradients in LC×LC
In the development of an LC×LC separation, a gradient, or isocratic, elution method is optimized for both dimensions separately [46]. A preference for gradient elution exists as narrower, constant peak widths are obtained, improving the peak capacity. Moreover, the analysis time is reduced for highly retained analytes. When employing gradients in the second dimension, wrap-around effects are reduced or eliminated. In a standard LC×LC method, the second-dimension gradient is optimized separately [47]. Therefore, the gradient rapidly covers a wide, or the entire, range of mobile phase composition. In a LC×LC setup, the gradient is repeated continuously until the end of the full first dimension run, the defining parameter for the entire run time. The described type of second dimension employed is called a full gradient, or full-in-fraction gradient (visualized in Figure 2A). The steep gradient in a short modulation time causes narrow peak widths and high peak capacities. As the same gradient is repeated, the potential of wrap-around effects of strongly retained analytes is low. If two quite non-orthogonal selectivities are employed, compounds are likely to cluster around the diagonal of the separation space, minimizing the use of the separation space. Another second-dimension gradient type is the segment gradient, segmenting the gradient in 2 or more parts [48]. Moreover, a parallel gradient is described, which approximates isocratic conditions in each modulation. An alternative to a full gradient is the use of a shifting gradient (visualized in Figure 2B) [49]. Here, the second dimension uses a narrower gradient elution method, which changes the modifier concentration during the LC×LC analysis. As the second-dimension gradient method changes throughout the analysis, the potential of wrap-around effects is increased. Additionally, the modifier concentration is attuned more effectively to the retention of the fractions injected from the first dimension. The effective peak capacity and orthogonality are increased, as the peaks are located all over the entire separation space.
11
Figure 2 Gradient types employed in LC×LC. (A) Full gradient, or full-in-fraction gradient, and (B) shifting gradient. Figure adapted from Li, et al. [46].
12
3. Experimental setup
3.1. Chemicals
Methanol (ULC/MS-CC/SFC grade) and acetonitrile (LC-MS grade) were obtained from Biosolve B.V. (Valkenswaard, the Netherlands). Isopropanol (HPLC grade) and tetrahydrofuran (unstabilized, HPLC grade) were acquired from Biosolve B.V. (Valkenswaard, the Netherlands) and used for the cleaning procedure of a column. Methanol (215 SpS, HPLC grade) and acetonitrile (200 SpS, HPLC grade), obtained from ROMIL (Cambridge, United Kingdom), were employed within the Life Scientific laboratory. The ion-pair reagents for the mobile phase were prepared from tetramethylammonium hydroxide solution (TMAOH, 25 wt% in methanol), triethylamine (TEA, ≥99.5%), tetrabutylammonium hydroxide solution (TBAOH, 40 wt% in water), and tetrabutylammonium hydrogensulfate (TBAHSO4, ≥97%) all
obtained from Sigma-Aldrich (Steinheim, Germany). The 1- and 2-naphthalenesulfonic (>50% and technical grade 70%, respectively) standards were purchased from Sigma-Aldrich (Steinheim, Germany). The formate buffer used formic acid (ACS grade, 98-100%) and ammonium formate (99.9%) obtained from Merck (Darmstadt, Germany). The acetate buffer was prepared using acetic acid (ReagentPlus, ≥99%) and ammonium acetate (ACS reagent, ≥97%), both obtained from Sigma-Aldrich (Steinheim, Germany). All water was purified in-house using a Sartorius Arium 611VF (Göttingen, Germany) or Merck-Millipore Integral 5 HPLC water unit (Burlington, United States) at a resistivity of 18.2 MΩ∙cm.
3.2. Equipment
Several systems were employed for the method development for the two-dimensional separation of the dispersing agents, namely: a Shimadzu, and an Agilent system at the University of Amsterdam; and an Agilent system at Life Scientific.
The method development for the first dimension was mostly performed using a Shimadzu setup (Kyoto, Japan). The system was controlled by a Shimadzu system controller (model CBM-20A) and contained a LC-10ADvp quaternary pump, an FCV-10ALvp solenoid for mobile phase mixing, and an SPD-M20A photodiode array detector. A standard 6-port valve with a fixed 5 µL loop was utilized for manual injections. The software controlling the system was Shimadzu LCsolution (version 1.21 SP1).
The method development for the second dimension and 2D-LC analysis were performed using components from both an Agilent 1100 and a 1290 Infinity II series (Waldbronn, Germany). The system contained an Agilent 1100 autosampler (model G1313A), an Agilent degasser (model G1322A), an Agilent 1100 capillary pump (model G1376A), and an Agilent II 1290 binary pump (model G7120A). The temperature of the columns was controlled by Agilent 1290 II column compartment (G1316C), equipped with an Agilent 2-position/8-port valve (model 5067-4214). An Agilent 1290 II diode-array detector (DAD, G7117B), with an Agilent Max-Light cartridge cell (model G4214-60008, 10 mm, V(o) 1.0 µL) flow cell for detection. The system was controlled by Agilent OpenLAB CDS Chemstation Edition (rev.
13 C.01.09). The system had the possibility of performing a 2D-LC run using an active-solvent modulation (ASM) valve, within a comprehensive or heart-cutting setup.
In Life Scientific the system consisted of a combined Agilent 1260 Infinity and 1290 Infinity II series (Waldbronn, Germany). The system contained an Agilent 1260 autosampler (model G1329B). The pumps for the first and second dimension were an Agilent 1260 quaternary pump (G1311B) and an Agilent 1290 binary pump (G7120A), respectively. Temperature control of the columns was performed by an Agilent 1260 column compartment (G1316A). For detection the system employed an Agilent 1260 variable-wavelength detector (VWD, G1314F), with an Agilent standard (model G1314-60186, 14 mm, V(o) 14 µL) flow cell. The system was controlled by Agilent OpenLAB CDS Chemstation Edition (rev. C.01.10).
Various columns were investigated for the use in separating the crop protection formulations. Table 1 is a list of all the columns used in the research for scouting purposes, or 2D-LC separations.
Table 1 List of all used columns for each evaluated selectivity, including information about stationary phases and column dimensions.
Separation mode employed
Vendor Column name Stationary phase type Column length (mm) Internal diameter (mm) Particle size (µm) RPLC Agilent Zorbax SB Rapid
Resolution HD C18 100 2.1 1.8
RPLC Agilent Zorbax Eclipse Plus
Rapid Resolution HT Phenyl-hexyl 50 4.6 1.8
RPLC Restek Raptor Biphenyl Biphenyl 150 3.0 2.7
RPLC Agilent Zorbax SB Rapid
Resolution Cyano 150 4.6 3.5
RPLC
Pheno-menex Kinetex PFP PFP* 100 4.6 2.6
IPRPLC Agilent Zorbax Eclipse Plus
Rapid Resolution HT C18 50 4.6 1.8
IPRPLC Agilent Zorbax Eclipse Plus
Rapid Resolution HT C18 30 2.1 1.8
HILIC Waters Acquity UPLC BEH Amide 150 2.1 1.7
HILIC Agilent Zorbax HILIC Plus Rapid Resolution HD
Diol 150 2.1 1.8
SEC Agilent PLgel Mixed-E Mixed-E 250 10 3.0
SEC
(aqueous) Agilent PL Aquagel-OH Aquagel-OH 300 7.5 5.0
MM
Thermo-Fisher Scientific
Acclaim Surfactant
Plus Proprietary 150 2.1 3.0
MM Sielc Obelisc N Proprietary 250 4.6 5.0
14
3.3. Sample preparation
Four commercial dispersing agent samples are available for analysis: Morwet D425, Morwet EFW, Reax 88B, and Product X. Reax 88B is composed of Kraft lignin, a lignosulfonate or sulfonated lignin product. Morwet EFW contains primarily naphthalenesulfonic acids. Morwet D425 comprises of naphthalenesulfonic acids and sulfonated naphthalene-formaldehyde condensates, an oligomeric product of naphthalenesulfonic acids. Product X is a commercial product containing two bioactive ingredients and the previously described components. E698 was prepared by Life scientific as an in-house generic product that resembles Product X in composition. Two commercial products that contain propyl- & butylnaphthalenesulfonic acid are called A & B.
Stock solutions of 5,000 ppm (5 mg/mL) were prepared by adding 50 mg to 10 mL of MilliQ water. The first three products are fully soluble in water. However, Product X produces a milky white solution, likely caused by a solid support of the active ingredient. As such, the stock solution is run through a 0.45 µm PVDF filter, resulting in a white-yellowish transparent solution.
A mixture, named Surfmix3, of 1,000 ppm Morwet D425, 1000 ppm Morwet EFW, and 500 ppm Reax 88B was prepared from stock solution in water. The mixture is an effective sample for method development as it covers a more complete range of analytes and requires more separation power than the individual samples. To more easily detect Reax 88B, a second mixture was created named Surfmix3b containing: 1,000 ppm Morwet D425, 1,000 ppm Morwet EFW, and 1,000 ppm Reax 88B.
3.4. Two-dimensional data visualization
To evaluate the two-dimensional data, the program PIOTR (Program for Interpretive Optimization) is used. Later in the project the software to visualize and evaluate the two-dimensional data was changed to MOREPEAKS (Multivariate Optimization and Refinement Program for Efficient Analysis of Key Separations). Both software are developed in-house by the University of Amsterdam. The data in this report is visualized as it was processed in the original program. Thus, older data processed by PIOTR is not repeated in MOREPEAKS.
15
4. Results and Discussion
4.1. One-dimensional separations
For the development of an LC×LC method that demonstrates high peak capacity, one must use two separation mechanisms that are orthogonal, i.e. employing different separation selectivities. Here, the following principles of chromatography were explored: reversed-phase liquid chromatography (RPLC), ion-pair reversed-reversed-phase chromatography (IPRPLC), hydrophilic interaction chromatography (HILIC), size-exclusion chromatography (SEC), and mixed-mode ion-exchange chromatography (IEX).
4.1.1. Reversed-phase liquid chromatography
Figure 3 Chromatogram of a Morwet D425 separation on a C18 reversed-phase column (Agilent Zorbax SB-C18 Rapid Resolution HD, 2.1x100 mm i.d., 1.8 µm particles). Mobile phase A consisted of water/acetonitrile 95:5 [%v/v] and mobile phase B was 100% acetonitrile. The injection volume was 5 µL. The flow rate was set at 0.200 mL∙min-1. The capillary pump was programmed for gradient elution with a total run time of 40 min: 0 to 20 min,
linear gradient from 10 to 100% B, maintained at 100% B for 5 min; 25 to 26 min, linear gradient to 90% A. UV detection performed at 280 nm.
For initial method development, Morwet D425 was separated on a typical reversed-phase octadecylsilane (C18) column (Agilent Zorbax SB-C18, 2.1x100 mm i.d., 1.8 µm particles). In Figure 3, the chromatogram is displayed, showing a dead time of 1.5 min. Accordingly, the dead volume is approximately 0.3 mL. The dwell volume can be estimated by the volume between mixing the gradient and the detector. The rapid change of solvent in the gradient at 25 to 26 min produces a peak at 30 min. At a flow of 0.2 mL∙min-1, a gradient
delay time of 4 min and the 0.3 mL dead volume, a gradient dwell volume of 0.5 mL is approximated. The dwell volume describes the volume between mixing of the gradient and the inlet of the column, indicating that the gradient does not instantly reach the inlet of the column. In turn, in the first minutes of the analysis, an isocratic elution occurs. Separation of the sample occurs, as seen between 6 and 12 min. However, the peaks are not fully
16 separated and show ‘shoulders’, indicating overlapping elution profiles. Additionally, a high background can be observed while the gradient is at 100% acetonitrile, indicating that compounds are retained on the column. Do note that analytes could be permanently retained and clog the column; this would increase the backpressure significantly. The background is also observed in a blank injection after performing the Morwet D425 analysis. As the constituents of the sample are unknown, it is not possible to identify the eluting components by UV detection.
Figure 4 Chromatogram of a Morwet D425 separation on a C18 reversed-phase column (Agilent Zorbax SB-C18 Rapid Resolution HD, 2.1x100 mm i.d., 1.8 µm particles). Mobile phase A consisted of water/acetonitrile 95:5 [%v/v] and mobile phase B was 100% acetonitrile. The injection volume was 5 µL. The flow rate was set at 0.200 mL∙min-1. The capillary pump was programmed for gradient elution with a total run time of 30 min: 0 to 10 min,
linear gradient from 10 to 40% B; 10 to 12.5 min, linear gradient to 100% B, maintained at 100% B for 7.5 min; 20 to 21.5 min, linear gradient to 90% A. UV detection performed at 280 nm.
Figure 4 is a chromatogram of Morwet D425 after several injections and uses a different gradient than before, without changing the starting conditions. The linear gradient is shallower in the first 10 min than in the chromatogram before. At high organic concentrations, the background was reduced to the baseline. Peaks are observed before the gradient dwell volume and the separated peaks from before start to elute closely after, which is conflicting to the previous results. Components are eluting in the isocratic part and at a lower acetonitrile percentage. Notably, the backpressure increased considerably to a degree that a flow rate of 0.2 mL/min could no longer be maintained throughout the gradient without exceeding the backpressure limit of the column. The risk of permanently retained analytes and clogging the column, as mentioned before, likely occurred. Even after successive cleaning procedures (flushing with an isopropanol/water mixture and THF), the backpressure did not diminish. As a result, the column could no longer be used and analysis of the samples on a C18 reversed-phase column without additives, such as ion-pair reagents, was rejected.
17 Instead of using a C18 for a stationary phase, a phenyl-hexyl stationary phase can be used as an alternative. The selectivity is complementary to C18, as it allows for hydrophobic interactions. Moreover, due to π-π interactions, it may also be selective for compounds containing aromatic groups.
Figure 5 Chromatogram of a Morwet D425 separation on a phenyl-hexyl reversed-phase column (Agilent Zorbax Eclipse Plus phenyl-hexyl Rapid Resolution HT, 4.6x50 mm i.d., 1.8 µm particles). Mobile phase A consisted of water/methanol 95:5 [%v/v] and mobile phase B was 100% methanol. The injection volume was 5 µL and column compartment was set to 60 °C. The flow rate was set at 2 mL∙min-1. The binary pump was programmed for gradient
elution with a total run time of 12 min: 0 to 5 min, linear gradient from 0 to 100% B, maintained at 100% B for 5 min; 10 to 11 min, linear gradient to 100% A. UV detection performed at 280 nm.
As a starting point for evaluation, a relatively long chromatographic run, including a long time at high organic concentration, was performed on Morwet D425. Figure 5 shows that the sample is visually more separated than on a C18 stationary phase. At a concentration of 100% methanol, all the peaks have appeared. Additionally, the background at high methanol concentration is minimal, indicating that all components likely eluted.
In 2D-LC, a fast second dimension is required to sufficiently sample the first dimension. RPLC is a suitable candidate, as the total analysis and equilibration time can be performed quickly, minimizing the total second-dimension run time. Additionally, RPLC methods are often easily compatible with a variety of first-dimension selectivities. A potential fast gradient on a moderately short phenyl-hexyl column for a second-dimension separation is visualized in Figure 6. The total run time is 1.5 min, which would also be the modulation time when implemented in a 2D setup. The separation is mostly limited from 0.2 to 0.8 min (maximum of 1.0 min), wasting around half of the separation space for dead time and equilibration. However, reasonable separation is still achieved. As such, the method could be applied as a second dimension for a comprehensive 2D separation.
18
Figure 6 Chromatogram of a Morwet D425 separation on a phenyl-hexyl reversed-phase column (Agilent Zorbax Eclipse Plus phenyl-hexyl Rapid Resolution HT, 4.6x50 mm i.d., 1.8 µm particles). Mobile phase A consisted of water and mobile phase B was 100% methanol. The injection volume was 5 µL and column compartment was set to 60 °C. The flow rate was set at 2 mL∙min-1. The binary pump was programmed for gradient elution with a total run
time of 1.5 min: 0 to 0.5 min, linear gradient from 25 to 100% B, maintained at 100% B for 0.5 min; 1.0 to 1.01 min, linear gradient to 75% A. UV detection performed at 280 nm.
A reversed-phase biphenyl stationary phase is complementary to both C18 and hexyl-phenyl. As a biphenyl column contains no aliphatic-alkyl structure, it is less hydrophobic in nature than C18 and hexyl-phenyl. Additionally, it exhibits increased π-π interactions than hexyl-phenyl, rendering the biphenyl more selective for separation based on aromaticity. The selectivity is suitable for the aromatic naphthalenes present in the samples. Although, the lack of hydrophobic interactions can cause the Kraft lignin to be poorly resolved.
Figure 7 displays the separation of Surfmix3 on a biphenyl column using a linear gradient. The separation of the most intense signals is confined within the first 10 minutes. Low intensity peaks are detected till 25 minutes, which elute at 100% methanol content. Less definition of the peaks is observed within the first region of the chromatogram, in comparison to the Morwet D425 hexyl-phenyl separation. The possible cause for the lack of resolution is the decrease in hydrophobic interactions due to the absence of the alkyl chain in the biphenyl column. In contrast, the later eluting peaks are better resolved than the hexyl-phenyl, possibly due to the increased π-π interactions. Consequently, an alkyl-biphenyl might employ both selectivities: the hydrophobicity and aromaticity for the first and second region of the chromatogram, respectively. The background signal increases as the methanol content increases, with a fronting peak at roughly 20 min. However, the fronting peak elutes before the mobile phase consists of 100% methanol, making an impurity in the methanol an unlikely cause. The biphenyl present in the stationary phase is UV-active, which could indicate a column bleeding effect.
In a 2D-LC setup, the reversed-phase biphenyl stationary phase can be used as both a first or second dimension. Though, the column dimensions in Figure 7 are not favorable as a second dimension. When applied as a first dimension, the solvents are compatible to various second-dimension separations. A drawback of the method is the requirement of a
water-19 methanol gradient, which exhibits significant pressure differences due to the varying viscosities of a water-methanol mixture. Therefore, the first-dimension effluent injected onto the second dimension can create a spike in backpressure, possibly causing an overpressure error. In a second-dimension separation, the backpressure differences could restrict the flow rate, limiting the run time of the second-dimension gradient and thus the modulation time.
Figure 7 Chromatogram of a Surfmix3 separation on a biphenyl reversed-phase column (Restek Raptor biphenyl, 3.0x150 mm i.d., 2.7 µm particles). Mobile phase A consisted of water and mobile phase B was 100% methanol. The injection volume was 2 µL and column compartment was set to 40 °C. The flow rate was set at 0.510 mL∙min-1.
The quaternary pump was programmed for gradient elution with a total run time of 35 min: 0 to 20 min, linear gradient from 5 to 100% B, maintained at 100% B for 5 min; 25 to 26 min, linear gradient to 95% A. UV detection performed at 280 nm.
Perfluorophenyl columns are similar in chemistry to a phenyl-type column, in that all hydrogens on the aromatic ring are replaced by fluorine. However, the fluorination offers a combination of aromaticity and polarizability, causing a different selectivity than phenyl and conventional C18 columns [50,51]. A run was performed using methanol as the organic modifier, to minimize suppression of π-π interactions. In Figure 8, a chromatogram of Surfmix3 is displayed that ran an identical gradient to the biphenyl separation shown before. The mixture almost entirely elutes at the dead time, with little separation right after. Santasania and Bell [52] describe that PFP columns exhibit ion-exchange properties. These properties arise due to the electronegativity of the fluorine atoms, which generates a partial negative and positive charge on the outer edge and inner region of the aromatic ring, respectively. The ion-exchange interactions dominate when analyzing cationic species. Although the authors note that such properties may arise from the silanol groups inherently present on the surface of the silica particles. As the pH was uncontrolled in the analysis, the retention might have been affected. If the column would be further investigated, the water needs to be replaced by an appropriate buffer, which suppresses or enhances these ion-exchange interactions.
20
Figure 8 Chromatogram of a Surfmix3 separation on a perfluorophenyl reversed-phase column (Phenomenex Kinetex PFP, 4.6x100 mm i.d., 2.6 µm particles). Mobile phase A consisted of water and mobile phase B was 100% methanol. The injection volume was 2 µL and column compartment was set to 40 °C. The flow rate was set at 2.000 mL∙min-1. The quaternary pump was programmed for gradient elution with a total run time of 30 min: 0 to 20
min, linear gradient from 5 to 100% B, maintained at 100% B for 5 min; 25 to 26 min, linear gradient to 95% A. UV detection performed at 280 nm.
Figure 9 Chromatogram of a Surfmix3 separation on a cyano column in reversed-phase mode (Agilent Zorbax SB Rapid Resolution Cyano, 4.6x150 mm i.d., 3.5 µm particles). Mobile phase A consisted of water and mobile phase B was 100% methanol. The injection volume was 2 µL and column compartment was set to 40 °C. The flow rate was set at 1.200 mL∙min-1. The quaternary pump was programmed for gradient elution with a total run time of 30 min:
0 to 20 min, linear gradient from 5 to 100% B, maintained at 100% B for 5 min; 25 to 26 min, linear gradient to 95% A. UV detection performed at 280 nm.
The selectivity of a cyano column can be regarded as orthogonal to a C18 separation [51]. Cyano columns are capable of π-π interaction, as well as dipole-dipole interactions. Although the samples of interest are not highly polar, in combination with the column selectivity based on aromaticity it might provide a practical separation. Figure 9 displays a
21 chromatogram of Surfmix3 separated on a cyano column. The separation is positioned within the first 12 minutes. By lowering the slope of the gradient and stopping the gradient before 100% organic modifier, a more spread-out chromatogram could be obtained. When comparing the chromatogram to Figure 4 and Figure 5 of a Morwet D425 separation, similar features are obtained. Additionally, in repeated injections of Surfmix3, peaks started to shift and broaden. As a result, the cyano column was disregarded as either a first or second dimension for a 2D-LC setup.
4.1.2. Ion-pair reversed-phase liquid chromatography
The permanently charged sulfonates in the sample are expected to affect the retention mechanism of the analyte. Ion-pair reagents are added to the mobile phase, to counteract the anionic charge and enhance retention by increasing hydrophobic interactions of the more hydrophobic analyte – ion-pair complex. The type of ion-pair influences the selectivity of the analysis. For the negatively charged sulfonates, several cationic alkyl-ammonium ion-pair reagents have been evaluated: tetramethylammonium hydroxide (TMAOH), triethylamine (TEA), tetrabutylammonium hydroxide (TBAOH), and tetrabutylammonium hydrogensulfate (TBAHSO4).
Analyzing Morwet D425 on a C18 column using TBAOH as an ion-pair reagent, separation of Morwet D425 is achieved, as shown in Figure 10. It is notable that the separation is limited to the narrow range starting in the middle of the gradient, at around 20 min until 30 min. Again, an increased background, without any defined separation, is seen at higher organic concentrations. However, it levels back to the baseline when the gradient is finished. Consecutive runs did not raise the backpressure as observed previously, so clogging and overpressure are not expected to occur over time. As such, TBAOH could be used as an ion-pair reagent for a first-dimension separation. Additionally, it is likely that a fast second-dimension separation can be achieved.
Figure 10 Chromatogram of Morwet D425 separation on a C18 reversed-phase column (Agilent Zorbax Eclipse Plus C18 Rapid Resolution HT, 4.6x50 mm i.d., 1.8 µm particles). Mobile phase A consisted of water + 5 mM TBAOH and mobile phase B was methanol + 5 mM TBAOH. The injection volume was 5 µL. The flow rate was set at 0.250
22
mL∙min-1. The quaternary pump was programmed for gradient elution with a total run time of 70 min: 0 to 40 min,
linear gradient from 25 to 100% B, maintained at 100% B for 15 min; 55 to 55.01 min, linear gradient to 75% A. UV detection performed at 280 nm.
Figure 11 Chromatogram of Morwet D425 separation on a C18 reversed-phase column (Agilent Zorbax Eclipse Plus C18 Rapid Resolution HT, 4.6x50 mm i.d., 1.8 µm particles). Mobile phase A consisted of water/methanol 95:5 [%v/v] + 5 mM TEA and mobile phase B was methanol + 5 mM TEA. The injection volume was 5 µL. The flow rate was set at 0.250 mL∙min-1. The quaternary pump was programmed for gradient elution with a total run time of 70
min: 0 to 40 min, linear gradient from 0 to 100% B, maintained at 100% B for 15 min; 55 to 55.01 min, linear gradient to 100% A. UV detection performed at 280 nm.
Next, the ion-pair reagent TEA was added to the mobile phase. Figure 11 shows the separation of Morwet D425, using a similar method as before, the gradient starting conditions are lowered. The sample appears to be separated more efficiently and components elute at lower methanol concentrations. A larger background signal can be observed that emerges earlier in the chromatogram. Additionally, an increase backpressure resulted that the previously achievable 0.250 ml∙min-1 could only just be reached; in
consecutive runs, the backpressure might exceed the pressure limit in the middle of the run. As a result, TEA was discarded as a potential ion-pair reagent for second-dimension separations.
Lastly, TMAOH was tested as a potential ion-pair reagent. An identical method as before is performed on Morwet D425. Figure 12 displays the chromatogram, a similar result is obtained compared to TBAOH. However, the peaks are more spread out and again eluted at lower methanol concentrations. Initially, the backpressure slightly increased, then stabilized throughout consecutive runs. TMAOH, like TBAOH, could be used for the separation in either a first- or second-dimension reversed-phase run.
23
Figure 12 Chromatogram of Morwet D425 separation on a C18 reversed-phase column (Agilent Zorbax Eclipse Plus C18 Rapid Resolution HT, 4.6x50 mm i.d., 1.8 µm particles). Mobile phase A consisted of water/methanol 95:5 [%v/v] + 5 mM TMAOH and mobile phase B was methanol + 5 mM TMAOH. The injection volume was 5 µL. The flow rate was set at 0.250 mL∙min-1. The quaternary pump was programmed for gradient elution with a total run
time of 70 min: 0 to 40 min, linear gradient from 0 to 100% B, maintained at 100% B for 15 min; 55 to 55.01 min, linear gradient to 100% A. UV detection performed at 280 nm.
Both TMAOH and TBAOH are viable options as ion-pair reagents in the separation of the samples. The experiments previously described contained ion-pair reagent in both mobile phase A and B. Potentially, one could add the reagent only to mobile phase A, resulting in that the reagent decreases in concentration throughout the gradient. Meanwhile, the organic mobile phase increases. Additionally, by substituting methanol for acetonitrile, faster flows can be attained due to the lower viscosities when mixed with water. Although, changing the mobile phase and flow rate, the separation selectivity and resolution can be affected. The expectation is that the components elute more quickly, making a second-dimension separation achievable.
Figure 13 displays the results of separating Surfmix3 using a method that decreases ion-pair concentration as acetonitrile concentration increases. The separation achieved is relatively poor, as broad, unresolved peaks are observed. The majority of the separation occurs between 0.3 and 1.4 min, so half of the separation space is wasted for the dead time and reequilibration. However, for a second-dimension separation, less convoluted peaks are expected due to the separation in the first dimension, so an optimal separation efficiency is not essential. Moreover, the method is relatively fast and can be implemented as a second dimension for comprehensive LC×LC analysis, with a modulation time of 2 minutes.
